Lithium Batteries Science and Technology

LITHIUM
BATTERIES
Science and Technology
Gholam-Abbas Nazri
Gianfranco Pistoia
[Eds.)
Stabilization of Layered Oxide Cathode
LiNi02 Structure
LITHIUM BATTERIES
LITHIUM BATTERIES
Edited by
Gholam-Abbas Nazri
General Motors Research & Development Center
Warren, Michigan, U.S.A.
Gianfranco Pistoia
Consultant Rome,
Italy
Springer
Editors
Gholam-Abbas Nazri Gianfranco Pistoia
General Motors R&D and Planning Center Via G. Scalia, 10
MC 480-102-RCEL 00136 Rome
30500 Mound Road Italy
Warren, MI 48090-9055
USA
ISBN: 978-1-4020-7628-2 (hardcover)

ISBN: 978-0-387-92674-2 (softcover) e-ISBN: 978-0-387-92675-9
Library of Congress Control Number: 2008941279
© Springer Science+Business Media, LLC 2003, First softcover printing 2009

All rights reserved. This work may not be translated or copied in whole or in part without the written
permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York,
NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in
connection with any form of information storage and retrieval, electronic adaptation, computer
software, or by similar or dissimilar methodology now known or hereafter developed is forbidden.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if
they are not identified as such, is not to be taken as an expression of opinion as to whether or not
they are subject to proprietary rights.
Printed on acid-free paper
springer.com
ents
Contributors XV
xvii
damentals
erials Aspects: An Overview

anthiram
l.l. Introduction 3
1.2. Basic Concepts of Rechargeable Lithium Batteries 3
1.3. Cells with Insertion Compounds as Both Anode and Cathode 8
1.4. Layered Oxide Cathodes 11
1.4.1. Layered LiCo02 12
1.4.2. Layered LiNi02 19
1.4.3. Other Layered Oxides 24
1.5. Spinel Oxide Cathodes 25
1.5.1. Spinel Manganese Oxides 26
1.5.2. Other Spinel Oxides 30
1.6. Oxide Cathodes with Polyanions 32
1.7. Other Oxide Cathodes 33
1.8. Anode hosts 35
1.9. Conclusions 36
References 37
Role of Electronic Properties in the Electrochemical Behavior of

ercalation Compounds from a First Principles Vantage Point
Vander Ven and G. Ceder
2.1. Introduction 42
2.2. General Electronic Structure of Octahedral Transition Metal Ions 45
2.3. First Principles Method 48
2.3.1. The Local Density Approximation (LDA) 54
2.3.2. The Generalized Gradient Approximation (GGA) 54
2.3.3. Spin Polarization 55
2.3.4. Caveas About Approximations to DFf 55
2.3.5. Different Numerical Schemes 57
2.4. Electronic Structure of LixCo02 58
2.4.1. Covalent Bonding in LixCo02 58
2.4.2. Experimental Evidence for Shift in Covalency 62
2.4.3. The Role of Covalency in Determining Thermodynamic,
Kinetic and Structural Properties of LixCo02 64
2.5. Electronic Properties in LiNi02: the 180° Interaction 67
2.6. Layered to Spinel Transformation in LixMn02 71
2.6.1.The Crystallography and Thermodynamics of the Layered
to Spinel Phase Transformation 71
2.6.2. A Two Stage Transformation Mechanism 73
2.6.3. Suppressing the Layered to Spinel Transformation 78
2.7.Conclusions 79
References 80
Contents
thesis of Battery Materials

. Whittingham 85
3.2. The Synthesis and Manufacture of Titanium Disulfide
3.3. Synthesis of Alkali Ion Intercalates by Insertion into the Host 88
Lattice (and the Reverse) 90
3.4. High Temperature Synthesis of Layered Oxides 93
3.5. Hydrothermal and Solvothermal Synthesis of Cathode Materials 95
3.6. Ion Exchange Methodes for Lithium Manganese Oxides Synthesis
3.7. Stabilization of Layered LiMn02 and the Mixed Metal Oxides 97
Li(Ni,Mn,Co)02
3.8. Addition of Carbon Conductive Diluents Post-Synthesis and 99
in-situ: LiFeP04 102
3.9. Synthesis of Anode Materials 103
3.10. Mechanochemistry and Mechanical Grinding 105
References
ode Materials
es and Composite Anodes: An Overview

Nazar and 0. Crosnier
4.1. Introduction: Lithium Metal and Carbon Based Materials 112
4. I.I. Issue of Safety, A Brief History 113
4.1.2. Carbonaceous Materials 114
4.1.3. Looking to the Future 115
4.2. Compounds and Composites of Sn, Sb and AI 116
4.2.1. Intermetallics Composites and Binary Phases Comprised of 117
Sn, Sb and AI
4.2.2. Silicides 127
4.3. Metal Oxides 129
4.3.1. Transition Metal Oxides Where Insertion Governs the
Process 130
4.3.2. Transition Metal Oxides Where Reduction of Metal Oxide
to Metallic or Semi-metal Occurs 130
4.3.3. Mixed Transition Metal Oxides- Where Both Reduction and 132
and alloying are Responsible for Li Capacity
4.4. Nitrides and Phosphides 132
4.4. I. Nitrides 132
4.4.2. Phosphides 134
4.5. Conclusions 138
References 139
onaceous and Graphitic Anodes

inter, K.-C. Moeller and J. 0. Besenhard
5.1. Carbons in Ion Transfer Batteries 144
5.2. History of Lithiated Carbons 146
5.3. Electrochemical Li Storage in Carbon 148
5.3.1. Carbonaceous Host Materials and Their Properties 148
5.3.2. Electrochemical Li Storage in Graphitic Carbons 153
5.3.3. Electrochemical Li Storage in Non-Graphitic Carbons 160
5.4. Reactions of Lithiated Carbons with the Electrolyte 167
ts 5.4.1.
Interph
ase
Formation- Metallic Li Anode vs. LixC 6 Anode and Oxide vii
Cathode
5.4.2. Unsolvated and Solvated Lithium Intercalation
5.5. What is Special About Lithiated Carbons?
References 168
171
hite-Electrolyte Interface in Lithium-Ion Batteries 179
azri, B.Yebka, and G.-A. Nazri 180
6.1. Introduction
6.2. Experimental Set Up to Evaluate Surface Reactions of Graphite
Anodes
6.2.1. Graphite Types 195
6.2.2. A General Procedure to Fabricate Carbon Anodes and
their Evaluation 198
6.3. Graphite Cycling and Film Formation 198
6.3.1. Gas Generation on Graphites
6.3.2. Infrared Spectroscopy of SEI Layer on Graphite Anodes 200
6.3.3. Thermal Analysis of SEI Layer on Graphite Anodes 203
References 208
216
Key Role of Nanoparticles in Reactivity of 3d Metal Oxides Towards Li 217
Tarascon, S. Grugeon, S. Laruelie, D. Larcher and P. Poizot 218
7.2. Experimental 221
7.3. Results and Discussion 222
7.3.1. Reactivity of 3d Metal Oxides Towards Li 222
7.3.2. Rationalization and Potential Applications 226
7.3.3. Practical Application Considerations 234
7.3.4. Acting on Reaction Paths Through Particle Size 236
References 245
de and Silicide Negative Electrodes

Amatucci and N. Pereira
8.2. The Nitrides 248
8.2.1. The Ternary Nitrides 248
8.2.2. The Binary Nitrides 254
8.2.3. Summary- Nitrides 258
8.3. Silicides 259
8.3.1. Introduction 259
8.3.2. Ternary Li-Si-M Systems at Elevated Temperature 260
8.3.3. Silicides at Ambient Temperature 262
8.3.4. Summary- Silicides 267
References 268
s and Intermetallic Anodes

Huggins
9.I. Introduction 270
9.2. Alloys as an Alternative to Pure Lithium 271
Contents
9.2.1. Lithium- Carbon Alloys 271

9.3. Principles Determining the Potentials and Capacities of Alloy
Electrodes Under Equilibrium Conditions 272
9.3.1. Relations Between Phase Diagrams, the Composition
Dependence of the Potential, and the Capacity of
Electrodes 273
9.3.2. An Example, The Lithium-Antimony System 275
9.3.3. Relation Between the Potentials and Thermodynamic
Data 278
9.3.4. Temperature Dependence of the Equilibrium Potential 279
9.3.5. Experiments at Ambient Temperature 283
9.4. Liquid Binary Alloys 283
9.5. Mixed-Conductor Matrix Electrodes 286
9.6. Decrepitation 290
9.7. Formation of Amorphous Products at Ambient Temperature 293
9.8. Summary 294
References 295
rent Issues of Metallic Lithium Anode

shikawa and M. Morita
I 0.1. Intrinsic Advantages and Drawbacks of Metallic Lithium Anode 297
10.2. Historical Aspects in Rechargeable Metallic Lithium Batteries 298
10.3. Recent Trends in the Development of Batteries with aLi Anode 300
10.4. Optimization of the Lithium Anode Interface 300
I 0.4.1. Fundamentals 300
10.4.2. Effect of Additives 303
10.5. Physical Factors Affecting the Lithium Interface 303
10.6 Recent Advances in the Research on Metallic Lithium Anode 304
References 311
athode Materials
nds in Cathode Materials for Rechargeable Batteries
Pasquali, S. Passerini and G. Pistoia
11.2. Methods of Synthesis 317
11.3. Effect of Particle Size and Morphology on Cathode Behavior 321
11.4. Natural vs. Artificial Cathodic Films 324
11.5. Specific Cathode Materials 327
11.5.1. LiNi02 and LiCo02 327
11.5.2. Manganese Spinels 335
11.5.3. Layered LixMn02 and Similar Cathodes 341
11.5.4. 3 V Cathodes 344
11.5.5. A Special Case: LiFeP04 347
11.5.6. Sulphur and Organic or Inorganic Sulphides 350
11.5.7. High Voltage Materials 351
11.6. Concluding Remarks 352
References 353
el Cathode Materials for Lithium-Ion Batteries

ia and M. Yoshio
ts ix
12.2. LiMn204 and Metal-Ion Doped Spinels LiMxMn2.x04 362

12.3. Optimizing the Capacity and Rechargeability by Using
a Li-Mn-0 Phase Diagram 366
12.4. Capacity Fading on Cycling 369
12.5. Oxygen-Deficient Spinel LiMn204.y 373
12.5.1. Charge/Discharge Profile of Spinels with Different Li/Mn
Ratios and Oxygen Contents 373
12.5.2. Phase Transitions During Charge/Discharge 374
12.5.3. Phase Transition at Low Temperature 375
12.6. Summary 377
References 377
yered Manganese Oxides as Cathodes

mmundsen
13.2. Overview of the Structure and Stability of Layered Manganese 383
Oxides
13.2.1. Layered vs. Orthorhombic Structure of LiMn02 383
13.2.2. Instability with Respect to the Spinel Structure 384
13.2.3. Alternative Layered Structures 387
13.2.4. Layered Lithium Manganese Oxides with Tetravalent
Manganese 388
13.3. Layered Manganese Oxides Based on LiMn02 389
13.3.1. Orthorhombic LiMn02 389
13.3.2. Lithium Manganese Oxide with the 03 Layered Structure
Prepared by "Soft Chemistry" Routes 390
13.3.3. Substituted LiMn02 with the 03 Layered Structure
Prepared by High Temperature Processing 392
13.3.4. 02 Layered Structures 395
13.4. Layered Solid Solution Cathode Materials Based on Tetravalent
Manganese 397
13.4.1. Overview of Solid Solution Approaches 397
13.4.2. Solid Solutions with LiCr02 399
13.4.3. Lithium-Rich Solid Solutions with LiCr02 402
13.4.4. Solid Solutions with LiNi02 and LiCoxNi 1.x02 402
13.4.5. Further Work 406
13.5. Summary and Outlook 406
References 407
hodes Based on LiCo02 and LiNi02

ark, Y. J. Kim and J. Cho
14.1. LiCo02 Cathode Materials 410
14.1.2. Capacity Fading Mechanisms in LiCo02 Cathode Materials 414
14.1.3. Electrochemical Properties ofLiCo02 Cathodes 416
14.1.4. Doping the LiCo02 Cathode 420
14.2. LiNi02 Cathode Materials 422
14.2.2. Structure of LiNi02 422
Contents
14.2.3. Electrochemical Properties and Capacity Fading

Mechanisms of LiNi02 423
14.3. Metal-Oxide Coating on LiCo02 Powders 426
14.3.2. Preparation of Metal-Oxide Coated LiCo02 Powders 426
14.3.3. Cycle-Life Performance of Metal-Oxide Coated LiCo02 427
14.4. Thin-Film LiCo02 Cathodes 43 I
14.4.2. Electrochemical Properties of AhOrCoated LiCo02 Thin
Films 431
14.4.3. Future Work 441
References 442
yanion-Based Positive Electrode Materials

Masquelier, S. Patoux, C. Wurm, M. Morcrette
15.1. Introduction: from Oxides to lonically Conducting
Polyanionic Frameworks 445
15.2. Tuning the M"+/M<n-1)+ Redox Couple in the NASICON structure
by Changing the Chemical Nature of the XO/. Groups: Inductive
Effect 448
15.2.1. Relative Position of Various M"+tM<n-1)+ Redox Couples
(M =Fe, Ti, V, Nb) in NASICON Phosphates 450
15.2.2. Position of the Ti4+/Ti3+ Couple vs. Li+/Li and Na+/Na 451
15.2.3. Relative Positions of the Ti4+/Ti 3+, Fe3+/Fe2+, Nb5+/Nb4+,
Nb4+/Nb3+, V4+tv 3+ and V 3+/V2+ Couples vs. Na+/Na and Li+/Li
in the NASICON Structure 453
15.3. Extension of the Concept to Other Polyanionic Structures 456
15.3.1. The Monoclinic Forms ofFe2 (S04h and Li 3M2 (X04h
(M=Fe, V,X=P,As) 457
15.3.2. The Rhombohedral Form of Li3Fe2 (As04h 460
15.3.3. The Diphosphates and Diarsenates Li,MX207 (M =Fe,
V, Ti; X= P, As) 461
15.3.4. Amorphous or Crystalline Iron Phosphates: FeP04 .nH20
and Fe4(P207h.4H20 463
15.3.5. V5+/V4+ and V4+/V 3+ Couples in LixVOX04 (X= P, As) 466
5 4
15.3.6. Nb +/Nb + Couple in LixNbP05 467
15.3.7. The Olivine LiMP04 (M =Fe, Mn, Co) 469
References 474
erstanding Phase Transformations in Lithium Battery Materials by

nsmission Electron Microscopy
Shao-Horn 478
16.2. Transmission Electron Microscopy 481
16.2.1. Electron Diffraction of Crystals 483
16.2.2. TEM Imaging 484
16.2.3. TEM Sample Preparation 484
16.2.4. Caveat toTEM Techniques
16.3. Lithium and Vacancy Ordering 484

16.3.1. Lithium and Vacancy Ordering in 03 Li,Ni02 485
ts xi
16.3.2. Lithium and Vacancy Ordering in 03 L,Co02 487

16.3.3. Lithium Ordering in 02 derived Li,Co02 489
16.4. Spatially Localized Phase Transitions 492
16.4.1. Jahn-Teller Induced Phase Transition in Li,[Mn2 ]04 493
16.5. Phase Transformation Between 2D Layered and 3D Spinel Related
Configurations 494
16.5.1. Intermediate Layered-Spinel Cation Configurations in LT-LiCo02 497
16.5.2. Layered to Spinel Transformation in Cycled Li,Mn02 499
16.6. Perspectives 500
16.6.1. High-Resolution TEM Imaging and Simulation 500
16.6.2. EELS Analysis and Imaging 502
16.6.3. In-Situ TEM Imaging 502
References 503
lectrolytes
uid Electrolytes: Some Theoretical and Practical Aspects
Nazri 509
17.1. Introduction to Solvents 511
17.2. Coulombic Forces Between Ions and Dipolar Solvents 511
17.2.1. Ion-Dipole Forces 513
17.2.2. Dipole-Dipole Interactions 513
17.2.3. Dipole-Induced Dipole Interactions 514
17.2.4. Instantaneous Dipole-Induced Dipole Interaction Forces 515
17.2.5. Hydrogen Bonding
17.3. Electron Pair Donor-Electron Pair Acceptor Interactions 516
(EPD/EPA Interactions) 517
17.4. Solvation 518
17.4.1. Solvation Number 518
17.4.2. Selective or Preferred Solvation 519
17.5. The Basic Requirements of Electrolytes for Lithium Batteries 520
17.5.1. Electrolyte Conductivity 525
17.5.2. The Electrochemical Stability of Organic Electrolytes 527
References
vanced Liquid Electrolyte Solutions

Aurbach and A. Schechter

18.1.1. Types of Advanced Liquid Electrolyte Solutions 530
18.1.2. Scope of Subjects of Interest 532
18.2. General Properties of Liquid Electrolyte Solutions 532
18.2.1. On the Physical Properties of Polar Aprotic Electrolyte
Solutions 532
18.2.2. How to Evaluate Solvents for Electrolyte Solutions 535
18.2.3. On the Conductivity of Polar Aprotic Solutions and
Diffusion of Ions in Them 539
18.3. Electrochemical Windows of Nonaqueous Electrolyte Solutions 541
18.3.1. On Cathodic Reactions of Polar Aprotic Electrolyte
Solutions 545
Contents
18.3.2. Oxidation Reactions in Polar Aprotic Electrolyte 547

Solutions
18.3.3. A Short Summary of the Anodic Limit of Nonaqueous 552
Electrolyte Solutions
18.4. Liquid Electrolytes Solutions for Rechargeable Li (Metal) 553
Batteries 553
18.4.1. Problems with the Use ofLi Metal Anodes
18.4.2. On the Choice of Electrolyte Solutions for Rechargeable Li 556
(Metal) Batteries 560
18.5. Liquid Electrolyte Solutions for Rechargeable Li-Ion Batteries
18.5.1. The Requirements of Suitable Solutions for Li-Ion 560
Batteries 561
18.5.2. Standard Electrolyte Solutions for Li-Ion Batteries 564
18.5.3. Advanced Electrolyte Solutions for Li-Ion Batteries 567
18.6. Molten Salts 567
18.6.1. Introduction
18.6.2. Electrochemical Properties of Molten Salt Solutions and 568
Battery Applications 568
References
ymeric Electrolytes: An Overview

. Kerr
19.2. Lithium Transport in Lithium Batteries 575
19.2.1. Solvent-free, Ion-Coupled Systems 577
19.2.2. Effect of Polymer Molecular Structure (Architecture) 579
19.2.3. Effect of Polymer Solvation Structure 582
19.2.4. Effect of Surfaces on Polymer Electrolyte Behavior 587
19.2.5. Gel Polymer Systems 591
19.3. Polymer Electrolytes in Lithium Batteries 592
19.3.1. The Effect of Transport Properties on Cell Performance 595
19.3.2. Polyelectrolyte Single-Ion Conductors 602
19.3.3. Chemical and Electrochemical Stability of Polymer Electrolytes 604
19.3.4. Dendrite Growth 609
19.3.5. Mechanical Properties 613
References 617
s and Ceramic Electrolytes for Lithium and Lithium-Ion Batteries

. Dudney
20.2. Glass Electrolytes 625
20.2.1. Oxide Glasses 626
20.2.2. Oxynitride Glasses and Lipon 626
20.2.3. Sulfide and Oxysulfide Glasses 628
20.2.4. Thionitride Glasses 629
20.2.5. Lithium Halide Doped Glasses 629
20.3. Crystalline Ceramic Electrolytes 630
20.3.1. Perovskite Electrolytes 631
20.3.2. Nasicon-Type Phosphates 632
20.3.3. Lisicon-Type Materials 633
ts xiii
20.3.4. Lithium Metal Halides, Nitrides, and Phosphides 633

20.3.5. Other Compositions 634
20.4. Models for Ion Conduction 634
20.5. Conclusion 635
References 637
tery Systems and Applications

on Batteries for EV, HEV and Other Industrial Applications
Broussely
21.1.1. Specific Properties Required of Industrial Li-Ion Batteries 646
21.2. Which Li-Ion Chemistry for a Large Battery? 646
21.2.1. Positive Materials 646
21.2.2. Negative Materials 651
21.3. General Technology and Manufacturing Processes 651
21.3.1. General Considerations for Battery Assembly 652
21.4. EV Applications 652
21.4.1. EV Battery Requirements and Technology Comparison 653
21.4.2. Battery Monitoring and Management 654
21.4.3. Li-Ion Technology for EV 655
21.5. HEY Applications 663
21.5.1. Application Requirements 663
21.5.2. Competing Battery Technologies 664
21.5.3. High Power Li-Ion Technology Challenges 664
21.6. Stationary Applications 671
21.6.1. Standby (Float) Operations 671
21.6.2. Intermittent Charging (Cycling) Applications 673
21.6.3. Buffer Operation 673
21.6.4. Stationary Battery Trials 674
21.7. Space Applications 674
21.7.1. Battery Requirements for Satellites 675
21.7.2. Competing Technologies 676
21.7.3. Saft Cell and Battery Design 677
21.7.4. Other Battery Designs 680
21.7.5. Launchers 683
References 684
ium Batteries for Medical Applications

. Takeuchi, R. A Leising, D. M. Spillman,
ubino, H. Gan, K. J. Takeuchi and A C. Marschilok
22.1 Introduction 686
22.2 Primary Batteries 687
22.2.1. The Use of Lithium Metal as an Anode 687
22.2.2. Lithium/Iodine Systems and their Medical Applications 688
22.2.3. Lithium/Thionyl Chloride Systems and their Medical Applications 689
22.2.4. Lithium/CFx Systems and their Medical Applications 691
22.2.5. Lithium/Silver-Vanadium-Oxide Systems and their Medical
Applications 691
22.2.6. Lithium/Mn02 Systems and their Medical Applications 694
Contents
22.3. Secondary Batteries 694

22.3. I. Lithium-Ion Systems - General Considerations 694
22.3.2. Lithium-Ion Systems- Current and future Medical Applications 697
22.4. Summary 699
eferences 699
rent Issues and Market Trends of Li-Ion Batteries for Consumer Applications
acArthur
23. I. Introduction 70 I
23.2. Lithium-Ion batteries 702
23.2.1. The Market 702
23.2.2. Different Technologies 704
23.2.3. Targets: Small and Large Batteries 705
References 706
Index 707
f Contributors
Chapter Name Chapter
Amatucci
mundsen
rbach
Besenhard
oussely
der
osnier
Dudney
ugeon
Huggins
ikawa
err
Kim
cher
uelle
Leising
cArthur
nthiram
Marschilok
squelier
Moeller
orcrette
ce:
The science and technology of lithium batteries have dominated the

f advanced power sources and replaced many other batteries in the
, particularly in the areas of communications, computers, electronics,
n more power demanding devices such as power tools and
rtation. The significant progress of lithium batteries is mainly due to
merous innovations in materials, design and safety aspects of the
s. Solid-state chemists, material physicists, materials scientists and
chemists are among the main contributors to the field of advance
batteries. Potential improvements in performance and application of
ost yet safer materials in lithium batteries are in near horizon.
This book provides the most current knowledge and recent concepts
anced lithium batteries. The book is designed to serve the researchers
ield also as a text for academic teaching. The emphasis is on materials
of advanced lithium batteries. Applications of the lithium batteries in
l field and transportation also have been reviewed. The general
ts of the device are discussed first and detailed electrode and
lyte materials are covered in the subsequent chapters.
Chapter one is providing a general background for the advanced
batteries and an over view of electrode materials, electrode
mances, and materials engineering. This chapter is highly
mended for students and practitioners in the field. Chapter two is a
overview of advanced theoretical approaches currently in use for
understanding of lithium battery electrodes and provides a guide for
ng new electrode materials. The pros and cons of each theoretical
, and the mechanism of charge transfer and structural modification of
lation electrodes and contribution of anions are discussed. Chapter
provides a general overview of advance synthesis for electrode
ls. Both traditional methods such as ceramic and solution techniques
ntraditional methods such as hydrothermal and mechanochemistry
s are covered in this chapter.
Part two of this book is a comprehensive discussion on alternative
materials for lithium batteries. Chapter four provides a general
ound on alternative anodes with a detailed analysis of intermetallics,
sites, metal oxides, nitrides and phosphides anodes. Chapter five
chemistry and physics of lithium intercalation into graphitic anode.
ation of carbonaceous materials and various graphitic materials are
scussed. Chapter six is an overview of carbonaceous materials with
sis on reactivity of lithiated graphite with organic electrolytes.
lyte decomposition on reactive anodes and formation of SEI layer and
s species are also discussed. Chapter seven is covering the role of
articles in reactivity of transition metal oxides anodes. This chapter
s a comprehensive discussion on mechanism, thermodynamics and
ii
s of the reversible redox process in transition metal oxide with lithium.

uilibrium voltage values and overpotentials for various metal oxides
apped in this chapter. The reaction pathways and mechanism of
on metal oxides reaction with lithium are clearly illustrated by in-situ
iffraction, magnetic and electrochemical measurements. Chapter eight
with most recent advances in metal nitrides and metal silicides anodes.
nism of reversible lithium insertion - extraction process in these
of materials is discussed. The role of nano-particle in enhanced
ity of transition metal nitrides are also presented. Chapter nine is
ng the fundamentals of lithium alloys and intermetallic anodes. The
diagrams of alloys also complement the voltage profiles of this class of
anodes. Chapter ten provides a general background on metallic lithium
and reviewed recent progresses made in understanding the nature of
anode - electrolyte interface. The role of various additives, stack
re and temperature are also discussed in this chapter.
In part three the cathode materials are reviewed. Chapter eleven
es a general background and trends in rechargeable cathode materials. r
twelve deals with synthesis, characterization and performance of
nese oxide spinel as cathode materials. Potentials and limitations of
cathodes, particularly their performances at elevated temperatures are
sed in this chapter. Chapter thirteen is a comprehensive review of the
pment of advanced layered transition metal oxide based on manganese
. Chapter fourteen is covering the structure, and performance of the
d lithium cobalt and lithium nickel oxide cathodes. Various synthesis
for preparation of layered transition metal oxides are also presented.
r sixteen is covering the electrochemical performances of layered
denum oxides. The role of hydrated water molecules on
chemical performance of the hydrated molybdenum oxide are
ted. Thermodynamic parameters are extracted from temperature
ent voltage profiles of the molybdenum oxide-based cathodes.
r seventeen is covering the microstructure of various cathode
als and providing a general background on application of transmission
n microscopy for analysis of battery electrodes.
Part three deals with various aspects of non-aqueous, polymeric and
c electrolytes. Chapter eighteen is a general overview of fundamentals
-aqueous electrolytes. The conductivity of various multi-blend organic
ates is presented. The dynamic nature of solvent-solvent and lithium
olvent molecule interactions is discussed. Numerous tables with
al properties of organic solvents are presented. This chapter also serves
ference for physical properties of aprotic solvents. Chapter nineteen
es a comprehensive analysis of liquid electrolyte and fundamentals of
de-electrolyte interface. A general over view of polymeric electrolytes
sented in chapter twenty. Recent advances are discussed and older ts
are challenged. Chapter twenty-one deals with synthesis, fabrication
xix
erformances of ceramic electrolytes. Physical methods for preparing

lm electrodes and micro-batteries are also presented in this chapter.
Part five is devoted to the application of lithium batteries in
ortation and medical field. The market analysis of lithium batteries and
uture prospects are also presented in part five. Chapter twenty-two is a
l review of lithium battery for transportation with an emphasis on high
lithium ion technology for hybrid electric vehicles. The lithium ion
logy based on advanced nickel oxide based cathode and graphitic
are presented. Application of lithium batteries in medical fields and
iaturization are discussed in chapter twenty-three. Chapter twenty-four
es current market trends and future prospects for the lithium ion
es.
Authors would like to thank all contributors to this book and many
to the Kluwer Publishing Company, particularly Mr. Alex Greene for
dance and patience during the course of preparing this book. We also
thank our families for their continuous support while we were
g on this manuscript.
Gholam-Abbas Nazri Gianfranco Pistoia

GM Res. & Dev. Center Battery Consultant
Warren, Michigan, USA Rome, Italy
pter 1
MATERIALS ASPECTS: AN OVERVIEW

A. Manthiram
aterials Science and Engineering Program, The University of Texas at Austin,

Austin, TX 78712, USA
INTRODUCTION
The exponential growth in portable electronic devices such as cellular

s and laptop computers during the past decade has created enormous t
in compact, light-weight batteries offering high energy densities.
growing environmental concerns around the globe are driving the
pment of advanced batteries for electric vehicles. Lithium-ion
es are appealing for these applications as they provide higher energy
y compared to the other rechargeable battery systems such as lead
ickel-cadmium, and nickel-metal hydride batteries as shown in Figure
he higher volumetric and gravimetric energy densities of the lithium lls
in Figure 1.1 are due to the higher cell voltages (- 4 V) achievable e
use of non-aqueous electrolytes, which also allow a wider
ature of operation. Lithium-ion cells have become a commercial
after the initial announcement by Sony in the early 1990s because of
ense world-wide activity on lithium insertion compounds (electrode
als) during the past three decades. This chapter provides an overview
electrode materials aspects of rechargeable lithium batteries. A more d
discussion of the specific cathode and anode materials systems as
s the electrolytes is presented by various authors in the subsequent
rs.
BASIC CONCEPTS OF RECHARGEABLE

LITHIUM BATTERIES
Rechargeable lithium batteries involve a reversible insertion/
ion of lithium ions (guest species) into/from a host matrix
Chapter 1
ode material), called lithium insertion compound, during the

rge/charge process. The lithium insertion/extraction process occurring
a flow of ions through the electrolyte is accompanied by a
ion/oxidation (redox) reaction of the host matrix assisted with a flow
ctrons through the external circuit. This concept of rechargeable
m batteries was first illustrated with a transition metal sulfide, TiS2, as
2
thode, metallic lithium as the anode, and a non-aqueous electrolyte.
iS2 cathode has a layer structure as shown in Figure 1.2. During
rge, the lithium ions are inserted into the van der Waals gap between
fide layers and the charge balance is maintained by a reduction of the
3
ns to Ti +. During charge, exactly the reverse process involving the
tion of lithium from the van der Waals gap and an oxidation of Te+ to
ccurs. During the lithium insertion/extraction (discharge/charge)
s, the layer structure is maintained, resulting in good reversibility.
50
Smaller
00
50 1-o
..Q...).
Lithium-ion
00
C)
oO
50
Ni/Cd Ni/MH
Lead-acid
0
0 100 200 300 400
Volumetric Energy Density (Wh/L)
1.1. Comparison of the gravimetric and volumetric energy densities of various

able battery systems.
l Aspects: An Overview 5
s
Ti ............................
............................
s
Li 00
s
............................ TI
1.2. Illustration of lithium insertion/extraction into/from the layered TiS 2 during the
ge/charge process.
The open-circuit voltage Voc of such a lithium cell is given by the

nce in the lithium chemical potential between the cathode (f.lu<cl) and
ode (f.lu(a)) as:
V = f.lu(c) - f.lu(a)
(1.1)
oc F
F is the Faraday constant. Figure 1.3 shows a schematic energy

m of a cell at open circuit. The cell voltage Voc is determined by the
es involved in both the electron transfer and the Lt transfer. While
ergy involved in electron transfer is related to the work functions of
thode and anode, that involved in Li+ transfer is determined by the
structure and the coordination geometry of the site into/from which ns
3
are insertedlextracted. Thermodynamic stability considerations
e the redox energies of the cathode (Ec) and anode (Ea) to lie within
nd gap Eg of the electrolyte, as shown in Figure 1.3, so that no
ted reduction or oxidation of the electrolyte occurs during the charge
rge process. Thus the electrochemical stability requirement imposes a
ion on the cell voltage as:
(1.2)
Chapter 1
Although the concept is simple, the lithium insertion compound

satisfy several criteria in order for it to be a successful cathode
al in a rechargeable lithium cell:
e insertion compound LixMyXz (X = anion) should have a high lithium

mical potential (fJuccJ) to maximize the cell voltage. This implies that
transition metal ion Mn+ in LixMyXz should have a high oxidation
e.
LUMO JA.a(Li)
Jlc(Li)
eVoc
Cathode Anode
Electrolyte/separator
1.3. Schematic energy diagram of a lithium cell at open circuit. HOMO and LUMO
spectively, to the highest occupied molecular orbital and lowest unoccupied molecular
n the electrolyte.
insertion compound LixMyXz should allow an insertion/extraction of

rge amount of lithium, x, to maximize the cell capacity. This depends
the number of available lithium sites and the accessibility of multiple
ences for M in the insertion host. A combination of high capacity and
voltage can maximize the energy density, which is given by a
duct of the capacity and voltage.
lithium insertion/extraction process should be reversible with no or

imal changes in the host structure over the entire range x of lithium
ertion/extraction in order to provide a good cycle life for the cell. This
plies that the insertion compound LixMyXz should have good structural
bility without breaking any M-X bonds.
e insertion compound should support mixed conduction. It should have
d electronic conductivity cre and lithium-ion conductivity cru to
imize polarization losses during the discharge/charge process and
reby to support a high current density and power density. This
ends on the crystal structure, arrangement of the MXn polyhedra,
metry and interconnection of the lithium sites, nature and electronic
figuration of the Mn+ ion, and the relative positions of the Mn+ and xn
rgies.
e insertion compound should be chemically stable without undergoing

reaction with the electrolyte over the entire range, x, of lithium
ertion/extraction.
e redox energy of the cathode in the entire range, x, of lithium

ertion/extraction should lie within the band gap of the electrolyte as
wn in Figure 1.3 to prevent any unwanted oxidation or reduction of
electrolyte.
m a commercial point of view, the insertion compound should be

xpensive, environmentally benign, and lightweight. This implies that
Mn+ ion should preferably be from the 3d transition series.
Following the initial demonstration of the concept of rechargeable

batteries with the layered TiS2, 2 several other sulfides and
4
genides were pursued during the 1970s and 1980s as cathodes.
ver, most of them exhibit a lower cell voltage of <2.5 V versus
ic lithium anode. This limitation in cell voltage is due to an overlap of
gher valent Mn+:d band with the top of the nonmetal:p band - for
le, with the top of the S:3p band as shown in Figure 1.4. Such an
p results in an introduction of holes into (or removal of electrons
2
the S -:3p band and the formation of molecular ions such as s;-'
in tum leads to an inaccessibility of higher oxidation states for Mn+ in
de LixMySz. The stabilization of higher oxidation state is essential to
ize the cell voltage. Recognizing this difficulty with chalcogenides,
nough's group at the University of Oxford focused on oxide cathodes
2
the 1980s.5 7- The location of the top of the 0 -:2p band much below
2
p of the S -:3p band and a larger raising of the Mn+:d energies in an
compared to that in a sulfide due to a larger Madelung energy (Figure
ake the higher valent states accessible in oxides. For example, while
an be readily stabilized in an oxide, it is difficult to stabilize Co3+ in a
Chapter 1
2 13
since the Co + + redox couple lies within the S2-:3p band as seen in
1.4. Accordingly, several transition metal oxide hosts crystallizing in
ty of structures (2-dimensional layered and 3-dimensional framework
res) have been pursued during the past two decades and Figure 1.5
res the electrochemical potential ranges of some selected lithium
on compounds versus metallic lithium. Cell voltages of as high as 5 V
metallic lithium have been achieved with oxide insertion compounds
n in Fifiure 1.5. The readers may refer to some of the review articles
6
ooks s- in the field to get a more detailed account of the various
insertion compounds that have been investigated over the past three
s.
E E
Co3+/2+
Co3+12+
Co4+/3+
Density of State N(E) Density of State N(E)
1.4. Relative energies of metal:d (for example, Co:3d) and nonmetal:p in (a) a sulfide
an oxide.
CELLS WITH INSERTION COMPOUNDS AS

BOTH ANODE AND CATHODE
Despite the development of several lithium insertion compounds

the 1970s and 1980s, the commercialization of rechargeable lithium
es was hindered for many years until the announcement by Sony in
ly 1990s. The delay was mainly because of the difficulties posed by
etallic lithium anodes. As an alkali metal, lithium is chemically
e with the non-aqueous electrolyte, which results in the formation of
ivating film on the metallic lithium anode. Although the passivating

n the surface prevents the bulk from further corrosion, it leads to a
niform plating of lithium during charging. This results not only in a
ell failure due to dendritic short circuiting, but also in serious safety
ms due to local over heating. The difficulties associated with the
ic lithium anode prompted to pursue a strategy in which a lithium
on compound instead of metallic lithium could be used as an anode.
>
-s
g-
Q)
C)
ctS
LixGriaphite ILixCoke
0- - Li Metal
-LiAl
Insertion Compound
1.5. Electrochemical potential ranges of some lithium insertion compounds in

e to metallic lithium.
Such cells consisting of lithium insertion compounds as both cathodes

nodes are called lithium-ion cells or rocking-chair cells since the
ion shuttles or rocks between the cathode and anode hosts during the
/discharge process as illustrated in Figure 1.6. This strategy, however,
s a careful selection of cathode and anode pairs in order to maintain
eptable cell voltage of at least 3 V and to realize a reasonable energy
without unduly increasing the weight or volume of the cell through
ertion compound anode.
With this strategy, the anode and cathode insertion hosts should
espectively, the lowest and highest voltages versus metallic lithium in
to maximize the cell voltage while satisfying the other required
. Accordingly, among the various known lithium insertion
Chapter I
unds, LiCo02, LiNi02 and LiMn204 oxides having a high electrode

ial of 4 V ··versus metallic lithium, have become attractive cathodes
hium-ion cells. Although one might tend to prefer a cathode like
Co04 offering 5 V versus lithium (Figure 1.5), the long-term stability
h the electrolyte and the cathodes under such a high voltage need to be
assessed before they could be considered for commercial cells (see
On the other hand, graphite and coke having a lower electrode
ial of <1 V versus metallic lithium and lightweight have become
ive anodes. The lithium-ion cells are presently made with LiCo02
es and carbon anodes as shown in Figure 1.6. In such a cell, the
m ions migrate from LiCo02 cathode to the LixC6 anode through the
lyte and the electron flows through the external circuit from the
e to the anode during the charge process. An exactly reverse reaction
during the discharge process. Although the replacement of metallic
m having a huge capacity of 3860 rnAhlg by carbon having a capacity
Cathode
Discharge
u+
Charge
Electrolyte
1.6. Schematic illustration of the charge/discharge process in a lithium-ion cell

ng of lithium insertion compounds as both anode and cathode.
rnAh!g causes an important sacrifice in energy density, the lithium

lls involving the shuttling of lithium ions between the two insertion
without any metallic lithium offer significant advantages in terms of
and cycle life.
From a materials design point of view, the anode insertion
und LixMyXz should also satisfy several important criteria like the
e insertion host: a large degree x of lithium insertion/extraction
capacity), high electronic and ionic conductivity, good chemical and
ural stabilities (reversibility), and affordable cost. The main difference
en the anode and cathode insertion hosts is that the former should have
lithium chemical potential (f.!u(a)) in order to maximize the cell
e, which implies that the Mn+ ion in the LixMyXz anode should have
xidation state.
Apart from the various intrinsic characteristics of the insertion
ode materials discussed, several other criteria are important in
ing a good lithium-ion cell that can offer high performance with long
life. The electrolyte should have high lithium-ion conductivity, but
be an electronic insulator in order to avoid internal short-circuiting.
h ionic conductivity in the electrolyte is essential to minimize IR drop
mic polarization and achieve good rate capability. With a given
lyte, the IR drop due to electrolyte resistance can be reduced and the
pability can be improved by having a higher electrode interfacial area
in separators. The electrolyte should also have good chemical stability
ould not undergo any direct reaction with the electrodes. Additionally,
gineering involved in cell design and fabrication plays a critical role in
erall cell performance. 12 For example, high electronic conductivity
hium-ion diffusion rate in the electrodes are essential to minimize cell
zations and the electronic conductivity of the electrodes can be
ved by adding electronically conducting additives such as carbon.
ver, the amount of additive should be minimized to avoid any undue
ce in gravimetric or volumetric capacity. Finally, cell safety,
nmental factors, and raw material and fabrication costs are additional
tant considerations in both materials selection and cell design.
LAYERED OXIDE CATHODES

Several oxides with a general formula LiM02 (M = V, Cr, Co, and Ni)
llize in a layered structure in which the u+ and M3+ ions occupy the
ate (111) planes of the rock salt structure to give a layer sequence of -
0-M-0- along the c axis as shown in Figure 1.7 for LiCo02• The
ure has an oxygen stacking sequence of ...ABCABC... along the c
nd the u+ and M3+ ions occupy the octahedral interstitial sites of the
close-packed oxygen array. This structure is designated as the 03
tructure since the u+ ions occupy the octahedral sites (0 referring to
dral) and there are three M02 sheets per unit cell. The structure with a
ly (covalently) bonded M02 layers allows a reversible
tion/insertion of lithium ions from/into the lithium planes. The
nnected lithium-ion sites through the edge-shared Li06 octahedral
Chapter 1
ement between the M02 layers provide fast two-dimensional lithium

ffusion leading to high cru. On the other hand, the edge-shared M06
dral arrangement with a direct M-M interaction can provide good
nic conductivity O'e depending on the electronic configuration of the
n. As a result, the LiM02 oxides crystallizing in the 03 structure have
e attractive candidates as cathodes and they are discussed below.
o Li
•co
c/2 Go
. 7. Crystal structure of LiCo0 2 having the 03 layered structure.
Layered LiCo02
LiCo02 synthesized by conventional high temperature procedures at T

C adopts the 03 layer structure (Figure 1.7) with an excellent ordering
u+ and Co3+ ions on the alternate (111) planes of the rock salt lattice.
ntrast, synthesis at lower temperatures (-400°C) results in a
3
erable disordering of the u+ and Co + ions leading to the formation of
iated spinel-like LiCo02 phase with a cation distribution of
c[Co2h6d04 as indicated by X-ray diffraction and single crystal
17 20
n diffraction studies. - This low-temperature form is designated as
Co02 and it exhibits poor electrochemical properties, making it an
21
ctive cathode. Thus heating the LiCo02 sample at high temperatures
00°C) is essential to have a good mobility for the cations and a
3
uent ordering of the Lt and Co + ions on cooling to give the desired
ered structure.
The 03-type LiCo02 obtained by the high temperature procedure
ts good electrochemical properties and most of the commercial
-ion cells are currently made with the LiCo02 cathodes. A high
3 14
m chemical potential )..lu(c) associated with the highly oxidized Co + +
5
provides a high cell voltage of around 4 V, and the discharge
e does not change significantly with the degree of lithium
partially filled t;;x
tion/insertion x in Li1_xCo02 (Figure 1.8). The direct Co-Co interaction
band associated with the Co 3+14+ couple leads to
lectronic conductivity for Li 1_xCo0 2• Additionally, a strong preference
3
low-spin Co + (t;ge) and Co 4+ (tgge) ions for the octahedral sites
es good structural stability for Li 1_xCo0 2 without encountering a
3
ion of the Co +14+ ions from the octahedral sites of the cobalt plane to
tahedral sites of the lithium plane via the neighboring tetrahedral sites
ter). These features have made LiCo02 an excellent cathode for the
-ion cells.
c a
b-
3
0 50 100 150 200
Capacity (mAh/g)
1.8. Typical discharge curves of (a) layered LiCo02, (b) layered LiNi0.85Co0.850z, and
el LiMn204.
However, only 0.5 lithium ions per LiCo02 formula could be

bly extracted/inserted and cycling to (l-x)<0.5 in Li 1_xCo0 2 results in
ty fade. This translates into a practical capacity of around 140 mAh/g,
is 50% of the theoretical capacity. The limitation in practical capacity
en attributed in the literature to an ordering of lithium ions and
uent structural distortions (hexagonal to monoclinic transformation)
2
x=0.5 in Li 1_xCo0/ and side reactions at >4.2 V (versus metallic
Chapter 1
23
m). However, it has been found recently that the cyclability of the
o02 cathodes could be improved significantly down to (l-x)::::0.3,
g reversible capacities of close to 200 mAh/g, 24by29 modifying its
e with other oxides such as Ah03, Zr02, and Ti02. - For example,
1.9 compares the cyclability of Ah03-modified LiCo02 cathode
29 with
unmodified LiCo02 cathode to various cut-off voltages. As can be
he surface modified samples exhibit good capacity retention to higher
f voltages with practical capacities of close to 200 mAh/g. Cho et
attributed the improved cyclability from ex-situ X-ray diffraction data
uppression of both the changes in the c axis with (1-x) and the
onal to monoclinic phase transition arising from an ordering of the
m ions around (l-x)=0.5. However, in-situ X-ray diffraction data
te that the surface modified samples also exhibit changes in c axis
1-x) and the hexagonal to monoclinic 30 phase transition around (1-
just like the uncoated Li 1_xCo02. Also, Transmission electron
copic (TEM) studies show that the coating oxides such as29Ah03 30 and
re present as loose nanoparticles on the surface of LiCo02. ' These
gs suggest that the capacity fade commonly encountered by the
2 cathodes (uncoated) on cycling to (l-x)<0.5 may not be simply due
hexagonal to monoclinic transition at (l-x)=0.5 and some other factors
ay a role.
250
- 200
(f)
<E
-u
(e)
150 (d)
(b)
(a)
Bu 100
l
J
50
0
0 20 40 60 80 100
Cycle number
1.9. Comparison of the cyclability data of unmodified and Al 2 0rmodified LiCo02

s in different voltage ranges at C/5 rate: (a) unmodified LiCo02 (4.3 to 3.2 V), (b)
ied LiCo02 (4.5 to 3.2 V), (c) unmodified LiCo02 (4.7 to 3.2 V), (d) Al20rmodified
(4.3 to 3.2 V), (e) Al203-modified LiCo02 (4.5 to 3.2 V), (f) Al203-modified LiCo02
.2 V), and (g) Al20rmodified LiCo02 (4.7 to 3.2 V).
The lack of a clear understanding of the factors that limit the practical
ty of the LiCo02 (uncoated) cathode to 140 mAhlg is partly due to the
22 31 34
hat most of the investigations in the literature • - have focused
on the structural characterization (both in-situ and ex-situ) of the
chemically charged Lh-xCo02 samples. No chemical characterization
uch as the variations of the oxidation state of Co and the oxygen
t with lithium content were available since it is difficult to chemically
terize the electrochemically charged samples as they are contaminated
arbon, binder, and electrolyte. To overcome these difficulties and to
p a better understanding of the factors that control the practical
ties, recent investigation by our group35 -38 has focused on both the
cal and structural characterizations of the Li 1_xCo02 samples that were
ed by chemical delithiation. The chemical delithiation reactions were
d out by stirring the LiCo02 powder under argon atmosphere with an
itrile solution of an oxidizing agent such as N02PF6 (or N02BF4) that
39
reduction potential of around 5 V versus metallic lithium and is
enough to extract all the lithium from LiCo02:
10 20 30 40 50 60 70 80
Cu Ka 2e (degree)
JO. X-ray diffraction patterns of the Li 1.xCo02 phases obtained by chemical

ion.
Chapter 1
Figure 1.10 shows the evolution of the X-ray diffraction patterns of

38
hemically delithiated Li1.xCo02 samples. The Li1.xCo02 system
ins the initial 03-type structure for 0.5 (1-x) l.O and a new phase
to form around (1-x)=0.45 as indicated by the appearance of a
er to the right of the (003) reflection. The new phase continues to
with decreasing lithium content and the end member Co02 consists of
ions corresponding to only the new phase, which could be indexed on
sis of a P3-type structure. In the intermediate region 0<(1-x) 0.45,
the 03-type and the P3-type phases coexist (Figure 1.11). The
ation of a P3-type structure for Co02 is in contrast to the 01-type
32 34
re reported in the literature - for the electrochemically synthesized
es. However, the P3-type structure observed for the chemically
sized Co02 sample is metastable and it transforms slowly to the 01-
tructure with increasing reaction time of >2 days with the oxidizer
37
F6• The formation of the metastable P3-type phase during chemical
ation is due to the rapid extraction of lithium in <15 min, which
tes into an extremely high charging rate of >4C, compared to the
slower electrochemical charging (<C/10 rate) process normally
2 34
• The P3- and 01-type phases have an oxygen stacking sequence of,
tively, ...AABBCC... and ...ABABAB... along the c axis compared
tacking sequence of ...ABCABC... in the initial 03-type LiCo02
(Figure 1.12). The P3- and 01-type phases are formed from the initial
e structure by a sliding
14.4
'$
CJ
14.8
14.0
• 000
--.
0
.. • •
0 0
13.6 0
IT1I] 3
....... •
2.84 0 0 0 oo
OJ
2.82 •
ca
2.80 •
0.0 0.2 0.4 0.6 0.8 1.0
Lithium content, (1-x)
1.11. Variations of the a and c lattice parameters of the chemically delithiated Li 1_

ith (1-x).
e Co02 sheets as shown in Figure 1.13 without involving the breaking

Co-O bonds. The 03-, P3- and 01-type structures have, respectively,
nd 1 Co02 sheets per unit cell. While the alkali metal ions have an
dral coordination in both the 03- and 01- type structures (0 referring
ahedral), they have prismatic coordination in the P3-type phase (P
ng to prismatic).
•u
•
0
OCo
c c eo
0•
a a
03 P3
.12. Crystal structures of 03-type LiCo02, P3-type Co02, and 01-type Co02 viewed
e (100) plane.
A A
c BA AC B
c
. BA
AC
ce c
c B
ce
B A
A
B
A :!c
P3 01
.13. Schematics of the sliding of some Co0 2 sheets leading to the transformation of
tructure to P3 and 01 structures. The letters refer to oxygen stacking sequence along
s.
Chapter I
1.98
-
c ••• o•
0 o • •o
Q)
§ • (a)
u 1.92 • 0
ffi 1.86 00
0>
1.80 0
0 0
1.74
0
3.8 A (b)
AAA.a
Q)
iii 3.6 At:,.

iii
c 6.. r:,.t:.. t:..r:,.t:..l:i. A
0 3.4 6..
" " C
·x 3.2 A
0 A
3.0 6..
0.0 0.2 0.4 0.6 0.8 1.0

1.14. Variations of the (a) oxidation state of Co and (Ni 0_ 85Co0.15) and (b) oxygen
with lithium content (1-x) in Li 1.xCo02 and in Li1.xNio.8sCoo. 150z.
Figure 1.14 shows the variations of the oxidation state of Co and the
n content with lithium content for the chemically delithiated Li 1.xCo02
38
es. The data in Figure 1.14 were obtained by a wet-chemical analysis
40
ing iodometric titration and the application of charge neutrality
ple. The oxidation state of cobalt increases linearly with (1-x) and the
n content remains around 2 for 0.5s(1-x)sl.O in Lh.xCo02.s- For Os(l
, the oxidation state of cobalt remains nearly constant and the charge
nsation is achieved by a loss of oxygen from the lattice with a 8=0.28
e end-member Co02. 8. The loss of oxygen from the lattice signals
a cal instability for (l-x)<0.5 in Li 1.xCo02.0, which may be the
factor
g its reversible capacity. The increased reversible capacity for the
e modified samples in Figure 1.9 is due to a more stable electrode
lyte interface and the consequent improvement in the chemical
ty. Interestingly, the lithium content (1-x) at which the oxygen loss
to occur coincides with the lithium content at which the new phase
gins to form (Figures 1.10 and 1.11). The results indicate that the P3
is oxygen deficient with a shorter 0-0 distance across the van der
gap between the Co02 sheets, signaling an increased 0-0 interaction.
. Layered LiNi02
LiNi02 crystallizes in the 03 layered structure shown in Figure 1.7

3 14
iCo02• The Ni + + couple with a high lithium chemical potential J..lu(c)
6
es a high cell voltage of around 4 V like LiCo02 as seen in Figure
iNi02 provides an important advantage compared to LiCo02 since Ni
expensive and less toxic than Co, but it suffers from a few problems:
3
ficulty to synthesize LiNi02 with all Ni + and as a perfectly ordered
3 41 43
without a mixing of u+ and Ni + ions in the lithium plane, - (ii)
eller distortion (tetragonal structural distortion) associated with the
3 7 44
in Ni +:d (t;ge) ion, '45 (iii) irreversible phase transitions occurring
32 46
the charge-discharge process, ' and (iv) exothermic release of
47
n at elevated temperatures and safety concerns in the charged state. '
a result, LiNi02 is not a promising material for lithium-ion cells.
ver, some of these difficulties have been overcome by a partial
ution of Co for Ni. For example, the composition LiNi0.8sCoo.IsOz has
ound to show a reversible capacity of around 180 mAh/g (Figure 1.8)
49
xcellent cyclability. This capacity is 30% higher than that of LiCo02
corresponds to a reversible extraction/insertion of 0.65 lithium per
ion metal ion (65% of the theoretical capacity). In situ X-ray
tion fine structure studies have shown that the substitution of Co for
44
ppresses the cation disorder and Jahn-Teller distortion. .4 5 The
ution also suppresses the phase transitions commonly encountered
ithium extraction from LiNi02 and LiNh-yCoy02 (0.15::;;y::;;0.2) is
vely being pursued particularly for electric vehicles due to its lower
mpared to LiCo02•
However, the LiNi 1_yCoy02 (0.15::;;y::;;0.2) cathodes have been found to
50
p impedance during cycling particularly at elevated temperatures.
ason for the impedance development is not fully understood, but it
be related to the migration of the nickel ions from the nickel plane to
hium plane during cycling. The nickel-rich Li 1_xNi 1_yCoy0 2 samples
51 52
ed by both chemical delithiation and electrochemical charging
een found to exhibit a decrease in cia ratio on heating at T >50°C
the Li 1_xCo02 system does not suffer from such a difficulty.53 Rietveld
is has shown that the decrease in the cia ratio is due to a migration of
kel ions from the octahedral sites of the nickel plane to the octahedral
f the lithium plane via the neighboring tetrahedral sites as illustrated
ure 1.15. The tendency for such a migration in Lh-xNi 1_yCoy0 2 and its
e in Li 1_xCo02 is due to a lower octahedral-site stabilization energy
3 7
) associated with the low-spin Ni +:d (t;ge) ions compared to that
3 6
e low-spin Co +:d (t g) ion (Table 1.1).53• 54 While a moderate
Chapter 1
•• •m 0 Layer
M Layer
0 Layer
OJ 0
-
Li Layer
oo
b_L
cP
•• •
1.15. Schematic representation of the paths for the migration of M 3+ ions in layered
2 (M = Mn, Co and Ni). The open squares and the dotted squares refer, respectively,
m ion vacancies and tetrahedral sites. T 1 and T2 refer to tetrahedral sites at (0, 0,
nd (0, 0, 0.375) respectively.
allows the Ne+ ions to migrate through the tetrahedral sites

mild heat, the stronger OSSE of Co 3+ hinders such a migration.
Nevertheless, the LiNi0.85Co0. 1502 cathodes exhibit a higher reversible
ty of 180 mAhlg compared to that of the analogous LiCo02 (140
) cathodes. However, the reason for the difference in capacity is not
understood in the literature. With an objective to develop a better
tanding, our group has focused on the structural and chemical
terizations of the chemically delithiated Li1 _xNi0.85Co0. 1502
samples comparison of the results with those of Li1_xCo02• Figures
1.16 and how the X-ray diffraction patterns of the chemically
delithiated Li1_ Coo.1502 samples and the lattice parameter variations
with the lithium t (1-x). The Li,_xNi0.85Co0.,50z system maintains the
initial 03-type re over a wide lithium content of 0.3 (1-x)1 and a new
phase begins
m at a much lower lithium content of around (1-x)=0.25 compared to
und around (1-x)=0.45 in the Li 1_xCo02 system (Figure 1.10) as
ed by the small shoulder to the right of the (003) reflection. The new
grows with decreasing lithium content and the end member
o0.,502 consists of only the new phase. However, the new phase could
indexed on the basis of an 03-type structure in contrast to the P3-
ructure found for Co02. The new 03-type phase of Ni0.85Co0.1502 has
1.1. Crystal field stabilization energies (CFSE) and octahedral site stabilization
s (OSSE) of some 3d transition metal ions.
Octahedral coordination Tetrahedral coordination OSSEd
Configuration" CFSEb Configuration• CFSEb.c
tz eo -8 Dq ezto -5.33 Dq -2.67 Dq

2g g 2
3 0
t ge
2 g
(HS)
t3 el
2g g
(HS)
3 2
t2 geg
(LS)
t6 g
2g
eo (LS)
6 I
t2 geg
d HS refer, respectively, to low-spin and high-spin configurations.

g energies are neglected for simplicity.
ned by assuming t = 0.444 L\,; t and L\, refer, respectively, to tetrahedral and
ral splittings.
ed by taking the difference between the CFSE values for octahedral and tetrahedral
ations.
ler c parameter than the initial 03 phase as evident

38 in Figure 1.17, and
ve designated this new 03 phase as 03' phase. In the intermediate
0<(1-x) 0.25, both the 03 and 03' phases coexist. The lower c
eter of the 03' phase may again signal a stronger 0-0 interaction
the van der Waals gap between the (Co,Ni)02 sheets like in the P3-
o02. The overall trends in the c parameter variations of both the Li 1_
and the Lit-xNio.s5Coo.t502 systems are in general agreement with that d
for the electrochemically charged samples, despite the slight nces
in the type of phases observed for the two delithiation processes.
Figure 1.14 compares the variations with lithium content of the

e oxidation state of Ni0.85Co0. 15 and the oxygen content in Li1_
Coo.t502 with those of Li1-xCo02. The Lit-xNio.s5Co0.t502_0 system loses
Chapter 1
10 20 30 40 50 60 70 80
Cu Ka 29 (degree)
l.l6. X-ray diffraction patterns of the Li 1_xNi 0. 85Co0.1502 phases obtained by

chemical tion.
n at a much lower lithium content of around (1-x)=0.25, and the

ember Ni0.85Coo.Is0z-li has a much smaller 8=0.1 compared to that in
_xCo02_0 system. However, similar to that observed in the Lh-xCo02_0
, the lithium content (1-x) at which the oxygen loss begins to occur
des with the lithium content at which the new phase 03' begins to
indicating that the 03' phase is oxygen deficient with a shorter 0-0
ce across the van der Waals gap and an increased 0-0 interaction.
importantly, the realization of a higher capacity for the
5Co0. 1502_0 system (180 mAhlg) compared to the LiCo02 system (140
) could be due to the better chemical stability of the former without
g to lose oxygen from the lattice down to a lower lithium content of
0.3. Although the data presented in Figure 1.14 are for
cally delithiated samples, the data collected
with the
chemically charged samples also show a similar trend.38 However, it
be noted that neutral oxygen may not be evolved during an over
of lithium-ion cells, but rather the cathode might undergo a slow
ion reaction with the electrolyte during cycling.
The differences in the chemical stability (oxygen loss) between the
oOz and the Lit-xNio.ssCoo.1sOz systems can be understood
14.8
.--,14.4 A A&
·::su 14.0 AA
Ae,.
A A
13.6 6. 4.
A 6. [I;] '
A
2.86 A
· 2.84 6.
ro 6. &A
2.82 J!ii.. A
6.
0.0 0.2 0.4 0.6 0.8 1.0

1.17. Variations of the a and c lattice parameters of the chemically delithiated Li 1•

o0.1502 with (1-x).
sidering the qualitative band diagrams for the Li 1_xCo02 and Li 1_xNi02
s (Figure 1.18). With a low-spin Co3+:3d6 configuration, the t2g band
pletely filled and the eg band is empty (t;
e ) in LiCo02• As
8
is extracted from LiCo02, the Co + ions are oxidized to Co4+ by a
3
al of electrons from the t 2g band. However, at deeper lithium

2
tion with (1-x) <0.5, removal of electrons may occur from the 0 -:2p
creation of holes) as well since the t2g band overlaps with the top of
-:2p band. The removal of significant amount of electron density from
2
-:2p band will result in an oxidation of the 0 - ions and an ultimate
3
t;
oxygen from the lattice. In contrast, the LiNi02 system with a low-
i +: econfiguration involves the removal of electrons only from
8
2
band. Since the eg band barely touches the top of the 0 -:2p band, the
i02 system loses a small amount of oxygen only towards the end of
extraction with (1-x)<0.2.35 In the case of LiNi0.85Co0.1502 with a
amount of Co, it loses oxygen slightly earlier for (l-x)<0.3. For a
rison, the band diagram of the Li 1_xMn 204 spinel system is also shown
3
t;
ure 1.18. With a high-spin Mn +: 8 econfiguration, the removal of
ns occurs only from the eg band in Li 1_xMn 204• Since the eg band lies
bove the top of the 0:2p band, the Li 1_xMn 204 system has been found
56
mical analysis data not to lose any oxygen. Also, while the partially
zg band leads to metallic conductivity for (l-x)<0.77 in Li 1_xCo02, the
etely filled tzg band leads to semiconducting behavior for 0::;(1-x)::; 1
Li,_xNiOzand Li,_xNio.ssCoo.Is02. 35
Chapter 1
E E
M n3+1 +:e8
Co3+14+:c8
Mn +14+:tz8
3
Ni 3+14+:e
Co3+14+:t z8
8
Ni 3+14+:tz8
o2-:2p o2-:2p o2-:2p
N(E) N(E) N(E)
Lio.sMnz04 Lio.sCo02 Lio.s NiOz
1.18. Comparison of the energy diagrams ofLi 05Co02, Li0_ 5Ni02, and Li0_5Mn 204•
The qualitative band diagrams shown in Figure 1.18 and the observed
n loss behavior are consistent with (i) the spectroscopic data that holes
roduced into the 02":2p band rather than into the Co:3d band in Li1.
, but into the Ni:eg band in Li1-xNiOz and Lii-xNio.ssCoo.IsOz during
57 59
chemical charging, - and (ii) the theoretical calculations that the
e charge on oxygen decreases with decreasing lithium content in Li1 _
60 61 Furthermore, the introduction of holes into the 0 2-:2p band
. '
ates the 0-0 interaction across the van der Waals gap between the
heets, which could in tum be the driving force for the formation of the
03' phases with smaller c parameters than the 03 phase (Figures 1.11
17). In fact, the P3 phase is generally believed to be inaccessible since
ide ions lie one above the other along the c axis (Figure 1.12), and the
ion of the P3 phase only in the case of Co02.8 could very well be due
2
presence of significant amount of holes in the 0 -:2p band and
erable amount of oxygen vacancies.
Other Layered Oxides

Both LiV02 and LiCr02 crystallize in the 03 structure (Figure 1.7),
e vanadium ions migrate to the lithium planes for (1-x)<0.67 in Li1.
due to the low OSSE (Table 1.1), while it is difficult to extract
from LiCr02• So, both LiV02 and LiCr02 are not promising cathode
als. LiTi02 is difficult to synthesize and both LiMn02 and LiFe0 2
sized by high temperature procedures do not adopt the 03 structure.
heless, the sodium analogs NaM0 2 (M = Mn and Fe) adopt the 03
re and the 03-type LiMn02 and LiFe02 have been obtained by ion-
63 65
nge reactions of NaM02• - Also, a partial substitution of Cr orAl
n has been found to access the 03-type LiMn02 by direct ceramic
53 66 68
ures. • - Unfortunately, the 03-type LiMn02 transforms to spinel
hases during cycling due to the low OSSE of Mn 3+ (Table 1.1) and the
3
quent migration of the Mn + ions from the Mn plane to the Li plane.
rly, the orthorhombic LiMn02 that is obtained by conventional
69
ures also transforms to spinel-like phase, leading to poor cyclability.
3-type LiFe02 also exhibits poor electrochemical cycling properties
3 5
structural instabilities since the high-spin Fe +:3d with an OSSE
of zero (Table 1.1) can readily migrate from the octahedral to
dral sites.
SPINEL OXIDE CATHODES

A few oxides with the general formula LiM 204 (M = Ti, V, and Mn)
lize in the normal spinel structure (Figure 1.19) in which the u+ and
3 14
+ + ions occupy, respectively, the 8a tetrahedral and 16d octahedral
f the cubic close-packed oxygen array to give a cation distribution of
M2h 6ct04• A strong edge-shared octahedral [M2]04 array permits
0 Li (Sa site)
0 Mn (16d site)
• 0 (32e site)
.19. Crystal structure of spinel LiMn 2 04.
ble extraction of the u+ ions from the tetrahedral sites without

ing the 3-dimensional [M2]04 spinel framework. An additional
Chapter 1
m can also be inserted in to the empty 16c octahedral sites of the spinel
work to give the lithiated spinel Li2 [M2]04. However, an electrostatic
ion between the u+ ions in the 8a tetrahedral and 16c octahedral sites,
share common faces, causes a displacement of the tetrahedral Lt ions
e neighboring empty 16c sites to give an ordered rock salt structure
a cation distribution of {Lizh 6c[Mzh 6ct04. Thus theoretically two
m ions per LiM204 formula unit could be reversibly inserted/extracted.
the edge shared M06 octahedral arrangement with direct M-M
ction, as in the layered LiM02 oxides, provides good electrical
ctivity O'e, the interconnected interstitial (lithium) sites in the three
sional spinel framework provide good lithium-ion conductivity cru. As
lt, the spinel LiM204 oxides have also become attractive candidates
ey are discussed below.
. Spinel Manganese Oxides

Mn is inexpensive and environmentally benign compared to Co and
, therefore, the spinel LiMn 204 cathodes have become appealing for
-ion cells. The extraction/insertion of two lithium ions from/into the
7
2]04 spinel framework occurs in two distinct steps. While the lithium
tion/insertion from/into the 8a tetrahedral sites occurs at around 4 V
e 1.8) with the maintenance of the initial cubic spinel symmetry, that
nto the 16c octahedral sites occurs at around 3 V by a two-phase
nism involving the cubic spinel Li[Mn 2]04 and the tetragonallithiated
Liz[Mn2]04. Although both involve the same Mn 3+ 14+ couple, the 1 V
nce between the two processes is a reflection of the differences in the
ergies.3 A deep energy well for the 8a tetrahedral u+ ions and a high
ion energy required for the u+ ions to move from one 8a tetrahedral
another via an energetically unfavorable neighboring 16c site lead to
er voltage of 4 V. The cubic to tetragonal transition on going from
2]04 to Liz[Mnz]04 is due to the Jahn-Teller distortion associated with
3 4
gle electron in the eg orbitals of a high spin Mn +:3d (t;ge) ion
e 1.20). A cooperative distortion of the Mn06 octahedra with long
bonds along the c axis and short Mn-0 bonds along the a and b axes
wn in Figure 1.20 results in a macroscopic tetragonal symmetry for
nz)04.
The cubic to tetragonal transition is accompanied by a 16% increase
cia ratio and 6.5% increase in the unit cell volume. This change is too
for the electrodes to maintain structural integrity during the
rge-charge cycle and so LiMn204 exhibits rapid capacity fade in the 3
on.
Therefore, LiMn204 could be used only in the 4 V region with a

d practical capacity of around 120 mAhlg (Figure 1.8), which
ponds to an extraction/insertion of 0.4 lithium per Mn. Unfortunately,
with a limited capacity, LiMn204 tends to exhibit capacity fade in the
Free Mnn+ ion Cubic Tetragonal
1.20. Illustration of the Jahn-Teller distortion in manganese oxides: (left) Mn4+:3d 3

bic symmetry (no Jahn-Teller distortion) and (right) Mn3+:3d4 with tetragonal
ry (Jahn-Teller distortion).
egion as well particularly at elevated temperatures (50°C). Several

s such as Jahn-Teller distortion occurring on the surface of the
70 71
es under conditions of nonequilibrium cycling • manganese
72 74
ution into the electrolyte, - formation of two cubic phases in the 4 V
76 77 78
,75 loss of crystallinity, • and development of micro-strain during
g have been suggested to be the source of capacity fade.
Several strategies have been pursued to overcome the capacity fade of
04. For example, increasing the oxidation state of Mn via (i) an
se in the oxygen content to give the defect spinel LiMn204+o (or
409_ 15)79 -81 which is actually a cation-deficient spinel Lh-yMn2_2y 3y04,
i) cationic substitutions to give LiMn2_yMy04 (M = Li, Cr, Co, Ni and
9 have been pursued with an aim to suppress the difficulties of Jahn-
4 3
distortion since Mn +:3d (t 8 e) does not undergo Jahn-Teller
ion. However, such a strategy leads to a decrease in capacity in the 4
on. For example, the end member (Li)[Mnl.67Lio.33]04 (or Li4Mn50n)
Chapter 1
M = Li has an average oxidation state of 4+ for Mn and does not t

any capacity in the 4 V region corresponding to the extraction of m
82 85
from the 8a tetrahedral sites. . It exhibits capacity only in the 3 V
corresponding to an insertion of an additional lithium into the 16c
dral sites to give {Liz}16c[Li0.33Mn ,67] 16d04• On the other hand, the
.yMy04 spinel oxides with M = Cr, Co, Ni, and Cu exhibit two
us for the extraction/insertion of lithium from/into the 8a tetrahedral
- : one around 4 V corresponding to the oxidation of Mn + to Mn +
89 3 4
he other around 5 V corresponding to the oxidation of the other

ion metal ions.
It is interesting to note that while the M = Co3+14+ and Ni3+14+ couples
around 4 V corresponding to the extraction/insertion of lithium
nto the octahedral sites of the layered LiM02, they offer 5 V
ponding to the extraction/insertion of lithium from/into the tetrahedral f
the spinel LiMn2.yMy04• The 1 V difference is due to the differences
3
site energies as discussed earlier. However, the 5 V capacity has
lly been observed only when the other transition metal ions are t
90 93
along with Mn in the spinel lattice. " For example, both
0 1
Ge150/ • 91 and LiCo20/ do not show any capacity above 4.5 V.
observations together with an examination in the 5 V region of spinel
such as LiMn2.yLiy04, LizMn409.1;, and Li4Mn5012 that contain no
ion metal ions other than Mn suggest that the participation of 02-11-
ossibly Mn4+tS+ couples could be involved in imparting the 5 V
94
ty.93• Also, it has been found that good crystallinity for the LiMn2.
samples by synthesizing at higher temperatures T>500°C is essential
94
ize capacities in the 5 V region. Finally, although cathodes with 5 V
pealing from an energy and power density points of view, the LiMn 2.
oxides may be prone to suffer from oxygen loss from the lattice
r to that observed for the layered Li 1.xCo02 for (1-x)<0.5 (Figure 1.14)
nsequent safety concerns in the 5 V region. However, the oxygen loss
ms may be minimized for M = Ni since the Ni + +:eg band barely
3 14
2
s the top of the 0 -:2p band (Figure 1.18).
Another strategy that has been pursued to improve the cyclability of
Mn204 spinel is the modification or coating of its surface with other
77
such as LiCo02, V205, Alz03, or Mg0 '95 -97 with an aim to suppress
solution of manganese in contact with the electrolyte. The modified
es exhibit improved cyclability at both ambient and elevated
ratures. However, firing at higher temperatures (-800°C) after coating
to much better capacity retention compared to firin§. at lower
9
ratures (-400°C) in contrast to the coated LiCo02 samples discussed
that requires firing only at low temperatures (-300°C). This suggests
a diffusion of the coating or modifying oxides MyOz into LiMn 204 is
al to achieve good cyclability and (ii) the improved cyclability need
simply due to the protection of the cathode surface from the
olyte and the consequent suppression of manganese dissolution.

y, additional factors other than manganese dissolution could be
g a role in improving the cyclability on coating/modifying the
04 cathodes.
With an aim to fully understand the factors that cause capacity fading
Mn204, our group has recently focused on a number of LiMn2.yMy04
Mn2.2yLiyMy04 (M = Co, Ni, and Ti) compositions. Our data reveal a
98
relationship between the lattice parameter difference a between the

bic phases formed in the 4 V region and the % capacity loss (Figure
The % capacity loss decreases with decreasing lattice parameter
nce a. For example, compositions such as
LiMn18. 5Li0m 5Ni 0.07504
t excellent capacity retention at elevated temperatures (Figure 1.22)
a small sacrifice in reversible capacity (100 mAhlg). The
.85Li0.075Ni0.07504 sample also exhibits superior rate capability (Figure
and excellent capacity retention after storage at 60°C at different
of discharge (DOD) compared to LiMn204.98' 99 The compositions that
t good capacity retention and rate capability are found to maintain
98 99
crystallinity with sharp diffraction peaks after extended cycling, •
sting a suppression of the development of micro-strain that is normally
ntered during the cycling of LiMn204.77'78 The study reveals that the
e changes and lattice strain arising from the difference in lattice
eters between the two cubic phases formed in the 4 V region play a
120 •
·
0)
\
-
110
..c 100
•\ (b)
--
<(
E 90
(c)
(d)
'(3 >. 80
<tl
c.. 70
<tl
()
60
(a)
50
0 10 20 30 40 50
Number of Cycles
1.21. Cyclability data of LiMn2.2yLiyNiy04 at 60°C: (a) LiMn204, (b)

Lio.osNio.os04, (c) LiMni.ssLio.D7sNio.o?s04, and (d) LiMnl.8Li0.1Ni 0.104.
Chapter 1
l role in the capacity fading of LiMn204, which could be overcome

ptimum cationic substitutions such as LiMnl.85Li 0.o75Nio.o7504. These
gs are encouraging and with optimum cationic substitutions, it could
e possible to employ the spinel manganese oxides for electric vehicle
ations.
4.4
::.:: : : :=: , c - -<-

4C 2C IC C/2 C/5 CliO
-- > 4.4 C/2
-
C/5
C/10
Q)
C)
ctS 3.6
0
>
4.4
4.0
3.6
0 20 40 60 80 100 120
Capacity (mAh/g)
1.22. Discharge profiles of LiMn 2.2yLiyNiy04 at various Crates (C/10, C/5, C/2, IC,
4C).
Other Spinel Oxides

Both LiTiz04 and LiV 204 crystallize in the normal spinel structure
M2] 16d04 (M = Ti and V) and are metallic due to the direct M-M
tions with a partially filled t2g band. LiTiz04 inserts an additional
into the empty 16c octahedral sites to give the lithiated spinel
6c[Tiz]16d04, which occurs with a flat discharge profile (two-phase
100
n) at a much lower voltage of around 1.5 V. Accordingly,
il.67Lio.33h6d04 (or Li4Tis0tz), which is much easier to synthesize
all Ti4+ than LiTiz04, has become appealing as an anode [101].
it.67Lio.33h6d04 and the lithiated spinel {Lizh6c[Tit.67Lio.33h6d04 differ
ir unit cell volume by 0.1%, which is attractive to maintain good

de integrity and capacity retention unlike in the case of LiMn204•
ver, the higher voltage (1.5 V) and lower capacity (175 mAh!g) of
012 make it difficult to compete with the currently used carbon anodes.
4 also inserts an additional lithium into the 16c sites. The lithium
0.08 •
•
0.06 • •
o<(
'-"
0.04
co
<l
•
•
0.02
0.00 ....
••r=--.---,.--.---,-----,.---r---.--.--
0 10 20 30 40
Capacity Loss (%)
1.23. Relationship between % capacity loss and lattice parameter difference (Lla)
the two cubic phases in Li 1.xMn2.yMy04 and Li 1.xMn2•2yLiyMy04 (M = Li, Ti, Co and
102 103
ould also be extracted from the 8a tetrahedral sites of LiV204. •
er, LiV204 suffers from a migration of the vanadium ions during
rocesses, which leads to poor capacity retention.
Both LiCr204 and LiFe204 are not known. LiCo204 could not be
y conventional high temperature methods, but it could be accessed as
al spinel (Li)8a[Co2h6ct04 by chemically extracting 50% of lithium
20 21 104
queous acid or N02PF6 in acetonitrile medium ' from the low
17 19
ature form of LiCo02 that has a lithiated spinel structure. ' LiCo204
is metallic due to the direct Co-Co interaction and the partially filled
3 14 104
d associated with the low-spin Co + +. While the extraction of
from the 8a sites occurs around 3.9 V, the insertion of an additional
21
into the 16c sites occurs around 3.5 V. However, the system
from a huge polarization loss as indicated by a large separation
n the discharge and charge profiles, which could possibly be related
low-temperature synthesis. Attempts to make LiNiz0 4 spinel by
ally extracting 50% of lithium from LiNi02 followed by heating at
°C result in a spinel-like cubic phase, but with the Ni3+14+ ions in both
and 16d sites. Heating at T>200°C results in a disproportionation to
iNi02 and Ni0.53
Chapter 1
OXIDE CATHODES WITH POLYANIONS

Iron oxides offer significant advantages compared to other 3d
ion metal oxides from both cost and toxicity points of view.
3 5
unately, the high spin Fe +:3d ion with a zero OSSE value (Table
nds to migrate readily from octahedral to tetrahedral sites as discussed
with respect to the layered LiFe02, causing structural instabilities.
problem could, however, be overcome by designing comfolex iron
2 2 05 106
consisting of poly ions such as (S04) - and (Mo04) -. • For
le, Fe2(S04)3 crystallizing in NASICON-related 3-dimensional
work structures was shown in the 1980s to exhibit a capacity of
105
110 mAh/g with a flat discharge voltage of 3.6 V. In these
res, the Fe06
1.24. Crystal structure of olivine (a) LiFeP04 and (b) FeP04 consisting of Fe06
ra and P04 tetrahedra (Ref. 62). The circles refer to lithium ions.
dra share comers with the S04 tetrahedra with a Fe-0-S-0-Fe linkage
thium ions could be inserted into the interstitial voids of the
2 13
work. The higher voltage of 3.6 V observed with the Fe + + couple
2(S04)3 compared to that found in simple iron oxide is due to the
6
ive effect caused by the counter cation S +. A stronger S-0 covalent
g weakens the 7t-bond Fe-0 covalence through inductive effect,
2 13
results in a lowering of the Fe + + redox couple and an increase in the
ltage. However, a poor electronic conductivity associated with the Fe-
-Fe (X= S or P) linkages leads to poor rate capability.
Following this initial concept of using poly anions, several phosphates
been investigated in recent years. 107- 109 Among them, LiFeP04
lizing in the olivine structure with Fe0 6 octahedra and P04 tetrahedra
e 1.24) has been shown to be a promising material exhibiting a flat

109
rge voltage of around 3.3 V. However, LiFeP04 also suffers from
d rate capability due to poor electronic (cre) and lithium-ion (au)
ctivity compared to the layered and spinel oxide cathodes discussed
. While only around 0.6 lithium per iron could be reversible extracted
he samples synthesized by conventional solid-state synthesis, nearly
thium ion per iron could be reversibly extracted with materials
sized by low temperature procedures and having a good atomic scale
110 112
of carbon. - Unlike in the case of layered LiM02 (M =Co, Ni or
xides, the presence of covalently bonded P04 units as well as the
3
ion of Fe2 13 couple rather than M + + couples leads to good structural
14
+ +
emical stabilities resulting in good safety features. However, the low
apability may limit its use to low power applications. Recent
ments involving a doping of LiFeP04 with Nb or Zr have been
d to improve the electronic conductivity by several orders of
113
tude, but further studies are needed to confirm the finding and to
tand the mechanism of improvement in conductivity.
OTHER OXIDE CATHODES

The cation migration and structural instability encountered by ions
ow OSSE values such as Fe 3+ and Mn3+ in close-packed structures
d and spinel structures) could also be overcome by having non-close
structures. For example, Na 0.5Mn02 - designated as Nao.44Mn0z in
erature - adopting a non-close-packed tunnel structure does not
rm to spinel-like phases and shows extraordinary structural stability
114 115
peratures as high as 300°C. ' Although only a small amount of
could be extracted from the ion-exchanged sample Na0.5_xLixMnOz,
nal lithium could be inserted into Na0.5_xLixMn0 2, exhibiting good
ility.
Although both layered LiV02 and spinel LiV204 suffer from
ral instability and poor cyclability, a number of other vanadium
116 117
such as V02(B)- a metastable form of V02 - and exhibit
V6013
apacity with good cyclability. They can be considered to have shear
res derived from the hypothetical V03 having the Re03
118 119
re. ' The structure consists of distorted V06 octahedra sharing
s and edges. For example, the structure of V6013 (Figure 1.25)
ts of edge shared octahedra forming single zig-zag strings (denoted as
double zig-zag ribbons (denoted as II), which are then joined together
ring corners to give a 3-dimensional lattice. The resulting structure
ns tricapped cavities joined through shared square faces with the three
aces of the cavity permitting lithium-ion diffusion along (010). On the
hand, V02(B) contains one dimensional tunnel. Additionally, both
Chapter 1
20 121 122
· and LiV308 having layered structures have been shown
to t high capacity. The vanadium oxides generally exhibit good
chemical ty with better safety characteristics without releasing oxygen
from the
3 14
compared to the cobaIt or m.e keI ox1'des as th e V +ts+ and y + +
4
2
s lie well above the top of the 0 -:2p band.
1.25. Crystal structure of V 6013 consisting of single strings (I) and double ribbons (II).
Although it is difficult to charge the layered LiCr02, a few chromium

such as Cr205, Cr6015 and Cr308 having oxidation states of 2:5+ for
123
w high capacities. However, the synthesis of these oxides generally
es decomposition of Cr03 in an autoclave or under high oxygen
re and the products are often contaminated with undecomposed Cr03.
In addition to the various crystalline cathode hosts discussed so far, a
r of amorphous and nanocrystalline oxide compositions LixMyOz with
u6,124,125 Cr126,127 Mn128-133 Fe134 and Fe Cu 135 have been shown to
' ' ' ' 1-T] T]
t good cyclability. Among them, the manganese and vanadium oxides
t remarkably high capacities of around 300 mAh/g. However, the
ty occurs over a wide voltage range of 4.3 to 1.5 V with a sloping
rge profile as the lithium insertion is not limited to specific
lographic sites and grain boundaries are involved in amorphous and
ystalline oxides. Nevertheless, a moderate capacity of around 200
could be readily utilized with these oxides. Also, the amorphous
nese oxides have been found (i) to accommodate the lattice (Jahn
distortions more smoothly compared to the crystalline manganese
136
as indicated by spectroscopic studies and (ii) not to transform to
like phases during electrochemical cycling. However, the large
e area associated with the smaller particle size may aggravate
ted electrode-electrolyte reactions, which could become an issue for
rm cyclability and charge efficiency particularly at elevated
atures. Surface modifications that can provide a more robust
de-electrolyte interface could help in this regard. Also, more in-depth

terizations by spectroscopic and other techniques are needed to fully
tand the mechanism of lithium insertion/extraction and the origin of
h capacities in the amorphous and nanocrystalline hosts.
Although some of the oxides discussed in this section are appealing as
xhibit high capacities with good safety characteristics compared to
2, most of them do not contain extractable lithium. The as-synthesized
are in the charged state and they are not suitable for fabricating
-ion cells with carbon anodes. However, development of new
-containing anodes may make these cathodes viable for lithium-ion
Also, these cathodes may be attractive for lithium polymer batteries
ying metallic lithium anode as the lower discharge voltage associated
many of them is advantageous to achieve good stability with the
er electrolytes.
ANODE HOSTS
Carbon has become the material of choice for anode in the present
tion of lithium-ion cells. 137' 138 The lightweight and low
chemical potential lying close to that of metallic lithium (Figure 1.5)
made carbon an attractive anode. It has a theoretical capacity of 372
, which corresponds to an insertion of one lithium per 6 carbon atoms
in LixC6). One of the drawbacks with the carbon anodes is the
ence of significant amount of irreversible capacity during the first
rge-charge cycle due to unwanted, irreversible side reactions with
lyte. Also, with electrolytes consisting of propylene carbonate (PC),
l graphite cannot be charged as it leads to an evolution of gas at
1 V. However, with electrolytes consisting of other solvents such as
ne carbonate (EC) and diethyl carbonate (DEC), the side reactions are
ssed and it can be cycled without much difficulty.
Carbon materials can be classified into soft carbon (graphitic carbon)
rd carbon (glassy carbon). The hard carbons are typically obtained by
mal decomposition of phenolic and epoxy resins and products from
um pitch. They have significant amount of open rnicropores, which
become closed on heating at higher temperatures, and some of them
t of single graphene sheets. Generally, hard carbons show higher
ties than graphite, which is thought to be due to the adsorption of
on both sides of the single graphene sheets, accommodation of extra
into nanometer size cavities, and storage of additional lithium at the
and surfaces. Also, hard carbons can be used with PC-based
lytes unlike graphite. However, the hard carbons show a sloping
rge profile between 0 and 1 V unlike graphite, which shows a nearly
scharge profile between 0 and 0.3 V. With an aim to improve the
Chapter 1
mance of carbon anodes, structural modifications such as texture

l, surface modification by mild oxidation and coating, and
oration of other elements such as B, 0, Si, and P have been studied.
A few oxides, nitrides, phosphides, and intermetallic compounds have
een investigated as anode hosts. Metal nitrides and phosphides offer
voltages versus lithium due to the covalent character and the
zation of lower oxidation states. As discussed earlier in section 5.2,
101
Li4Ti5012 offers 175 mAh/g at around 1.5 V, but it is less attractive
red to carbon due to its lower capacity and higher voltage. Sn02
a reversible capacity of as high as 600 mAh/g at 0 to 2 V, but it
139
ts a high irreversible capacity loss during the first cycle. Rutile-type
140 141 14 143 144
and WO2 ' '23' Fe 0 MnP4 and intermetallic compounds such
6Sn5 having NiAs-type structure have also been investigated. These
s, including carbon, do not contain Li and they can be coupled only
athodes such as LiCo02 and LiMn204 that already contain lithium.
respect to finding anodes that could be coupled with cathodes free
ithium, some lithiated transition metal nitrides 137' 145 and intermetallic
unds 146 have become appealing. For example, Li7.xMnN4 and Lh.
exhibit capacities of around 200 mAh/g with a flat discharge voltage
und 1.2 V. Lh6-xCoo.4N, Lh6-xCUo.4N, and Lh7-xFeo.3N that have
res similar to that of LhN exhibit much higher capacities of around
Ah/g at 0.2 to 1.3 V due to the formation of an amorphous phase
the initial stages. Two lithium per formula unit could be reversibly
146
ted from LhCuSn. However, further work is necessary to assess the
otential of these anodes. If successful, they have the possibility of
coupled with some of the lithium-free cathodes discussed in Section
CONCLUSIONS
After providing a brief introduction to the basic concepts of
geable lithium batteries, this chapter presented an overview of the
s cathode and anode materials. In view of maximizing the cell voltage
nergy density, transition metal oxides offering around 4 V versus
have emerged as the choice for cathodes while the light-weight
with an electrochemical potential close to that of metallic lithium has
e the anode. With carbon anode that does not contain lithium, the
d LiM02 (M = Co and Ni), spinel LiMn204, and olivine LiFeP04
ning lithium have emerged as the leading cathode candidates.
er, only 40 to 65% of the theoretical capacity of the layered LiM02
inel LiMn204 could be practically utilized due to the chemical and
ral instabilities encountered by these cathodes at deep lithium
ion. Although recent studies indicate that 100% of the theoretical
ty of LiFeP04 can be utilized, the added weight of the P04 units and
oor electronic conductivity lower the energy and power densities
cantly.
The future challenge is to develop simple oxide cathodes without
elements such as P in which at least one lithium ion per transition
ion could be reversibly extracted/inserted to give close to 300 mAhlg
keeping the materials cost and toxicity low. Such cathodes can double
ergy density compared to the present level. There are also possibilities
rease the capacity of anodes perhaps by focusing on amorphous
als and metal nitrides, pnictides, borides, and carbides with significant
nt character. An alternative approach is to develop cells with lithium
ning anodes and lithium-free cathodes. This strategy will allow the use
e of the already known high capacity cathodes such as the vanadium
and nanocrystalline manganese oxides that offer higher capacity with
chemical stability and safety characteristics compared to the currently
iCo02 cathode. From a safety, and cycle and shelf life points of view,
es with a lower voltage (3 to 4 V), but with an increased capacity are
ble for future applications. Such cathodes would also be attractive for
er batteries from a stability point of view.
NOWLEDGMENT
Financial support by the Welch Foundation Grant F-1254 is gratefully

wledged.
ERENCES
ium Battery Energy Storage (LIBES) Publication, Technological Research
ociation, Tokyo (1994).
. Whittingham, Science 192 (1976) 1126.
K. Aydinol, G. J. Ceder, J. Electrochem. Soc. 144 (1997) 3832.
S. Whittingham, A. J. Jacobson, Intercalation Chemistry, Academic Press, New York
2).
Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough, Mat. Res. Bull. 15 (1980)
.
Goodenough, K. Mizushima, T. Takeda, Jap. J. Appl. Phys. 19 (1983) 305.
. Thackeray, W.I.F. David, P.G. Bruce, J.B. Goodenough, Mat. Res. Bull. 18 (1983)
Gabano, (Ed.) Lithium Batteries, Academic Press, London (1983).

. Venkatasetty, (Ed.) Lithium Battery Technology, John Wiley, New York (1984).
Pistoia, (Ed.) Lithium Batteries: New Materials, Developments and Perspectives,
strial Chemistry Library, Vol. 5, Elsevier, Amsterdam (1994).
ulien, G.A. Nazri, Solid State Batteries: Materials Design and Optimization, Kluwer,
on (1994).
inden, (Ed.) Handbook of Batteries, second edition, McGraw Hill, New York (1995).
Chapter 1
Wakihara, 0. Yamamoto, (Eds.) Lithium Jon Batteries: Fundamentals and

formance, Wiley-VCH, Weinheim (1998).
M. Thackeray, Prog. Solid State Chern. 25 (1997) 1.
Manthiram, J. Kim, Chern. Mat. 10 (1998) 2895.
Manthiram, J. Kim, Recent Res. Dev. Electrochem. 2 (1999) 31.
. Gummow, M.M. Thackeray, W.I.F. David, S. Hull, Mat. Res. Bull. 27 (1992) 327.
. Gummow, D.C. Liles, M.M. Thackeray, Mat. Res. Bull. 28 (1993) 235.
. Gummow, D.C. Liles, M.M. Thackeray, Mat. Res. Bull. 28 (1993) 1177.
hao-Hom, S.A. Hackney, C.S. Johnson, A.J. Kahaian, M.M. Thackeray, J. Solid State
rn. 140 (1998) 116.
hoi, A. Manthiram, J. Electrochem. Soc. 149 (2002) A162.
Reimers, J.R. Dahn, J. Electrochem. Soc. 139 (1992) 2091.
. Amatucci, J.M. Tarascon, L.C. Klein, Solid State Ionics 83 (1996) 167.
ho, C. Kim, S.l. Yoo, Electrochem. Solid-State Lett. 3 (2000) 362.
ho, Y.l. Kim, B. Park, Chern. Mater. 12 (2000) 3788.
ho, Y.l. Kim, B. Park, J. Electrochem. Soc. 148 (2001) A1110.
ho, Y.I. Kim, T.J. Kim, B. Park, Angew. Chern. Int. Ed. Engl. 40 (2001) 3367.
iu, H. Wang, L. Fang, J.Y. Lee, L.M. Gan, J. Power Sources 104 (2002) 101.
. Kannan, L. Rabenberg, A. Manthiram, Electrochem. Solid-State Lett. 6 (2003) A16.
hen, J.R. Dahn, Electrochem. Solid-State Lett. 5 (2002) A213.
hzuku, A. Ueda, M. Nagayama, J. Electrochem. Soc. 140 (1993) 1862.
. Amatucci, J.M. Tarascon, L.C. Klein, J. Electrochem. Soc. 143 (1996) 1114.
. Tarascon, G. Vaughan, Y. Chabre, L. Seguin, M. Anne, P. Strobel, G.G. Amatucci,
olid State Chern. 147 (1999) 410.
. Yang, X. Sun, J. McBreen, Electrochem. Comm. 2 (2000) 100.
. Chebiam, F. Prado, A. Manthiram, Chern. Mater. 13 (2001) 2951.
. Chebiam, F. Prado, A. Manthiram, J. Solid State Chern. 163 (2002) 5.
enkatraman, A. Manthiram, Chern. Mater. 14 (2002) 3907.
enkatraman, Y. Shin, A. Manthiram, Electrochem. Solid-State Lett. 6 (2003) A9.
. Wizansky, P.E. Rauch, F.J. DiSalvo, J. Solid State Chern. 81 (1989) 203.
Manthiram, J.S. Swinnea, Z.T. Sui, H. Steinfink, J.B. Goodenough, J. Amer.Chem.
109 (1987) 6667.
Dutta, A. Manthiram, J.B. Goodenough, J. Solid State Chern. 96 (1992) 123.
Hirano, R. Kanno, Y. Kawamoto, Y. Takeda, K. Yamamura, M. Takano, K. Ohyama,
Ohashi, Y. Yamaguchi, Solid State Ionics 78 (1995) 123.
Kanno, H. Kubo, Y. Kawamoto, T. Kamiyama, F. Izumi, Y. Takeda, M. Takano, J.
d State Chern. 110 (1994) 216.
akai, T. Nakagome, Electrochem. Solid-State Lett. 1 (1998) 259.
akai, K. Takahashi, Y. Shiraishi, T. Nakagome, F. Nishikawa, J. Solid State Chern.
(1998) 145.
Ohzuku, A. Ueda, M. Nagayama, Y. Iwakashi, H. Komori, Electrochim. Acta 38
93) 1159.
Dahn, E.W. Fuller, M. Obrovac, U. von Sacken, Solid State Ionics 69 (1994) 265.
hang, D. Fouchard, J.R. Rea, J. Power Sources 70 (1998) 16.
Li, J. Curie, J. Electrochem. Soc. 144 (1997) 2773.
Nagasubramanian, Proceedings of the 40th Power Sources Conference, Cherry Hill,
June 10-13, 2002, p. 434.
. Chebiam, F. Prado, A. Manthiram, J. Electrochem. Soc. 148 (2001) A49.
. Kannan, A. Manthiram, J. Electrochem. Soc. 150 (2003) A349.
hoi, A. Manthiram, J. Electrochem. Soc. 149 (2002) A1157.
Huheey, Inorganic Chemistry, second edition, Harper and Row, New York, 1978,
48.
onci, B. Scrosati, V.R. Albertini, P. Perfetti, Electrochem. Solid-State Lett. 3 (2000)
.
. Chebiam, A.M. Kannan, F. Prado, A. Manthiram, Electrochern. Cornrnun. 3 (2001)

.
. Montoro, M. Abbate, J.M. Rosolen, Electrochern. Solid-State Lett. 3 (2000) 410.
. Mansour, X.Q. Yang, X. Sun, J. McBreen, L. Croguennec, C. Delmas, J.
ctrochern. Soc. 147 (2000) 2104.
Balasubramanian, X. Sun, X.Q. Yang, J. McBreen, J. Electrochern. Soc. 147 (2000)
3.
Ceder, Y.M. Chiang, D.R. Sadoway, M.K. Aydinol, Y.l. Jang, B. Runny, Nature 392
98) 695.
Vander Yen, M.K. Aydinol, G. Ceder, G. Kresse, J. Hafner, Phys. Rev. B58 (1998)
5.
. de Picciotto, M.M. Thackeray, W.I.F. David, P.G. Bruce, J.B. Goodenough, Mat.
. Bull. 19 (1984) 1497.
. Armstrong, P.G. Bruce, Nature 381 (1996) 499.
apitaine, P. Gravereau, C. Delmas, Solid State Ionics 89 (1996) 197.
uchs, S. Kemmler-Sack, Solid State lonics 68 (1994) 279.
Davidson, R.S. McMillan, J.J. Murray, J. Power Sources 54 (1995) 205.
avidson, R.S. McMillan, H. Slegr, B. Luan, I. Kargina, J.J. Murray, J.P. Swainson, J.
er Sources 81-82 (1999) 406.
Wang, Y.l. Jang, Y.M. Chiang, Electrochern. Solid-State Lett. 2 (1999) 490.
Gummow, D.C. Liles,.M.M. Thackeray, Mat. Res. Bull. 28 (1993) 1249.
Ohzuku, M. Kitagawa, T. Hirai, J. Electrochern. Soc. 137 (1990) 769.
M. Thackeray, Y. Shao-Hom, A.J. Kahaian, K.D. Kepler, E. Skinner, J.T. Vaughey,
. Hackney, Electrochern. Solid-State Lett. 1 (1998) 7.
. Jang, Y.J. Shin, S.M. Oh, J. Electrochern. Soc. 143 (1996) 2204.
noue, M. Sano, J. Electrochern. Soc. 145 (1998) 3704.
Du Pasquier, A. Blyr, P. Courjal, D. Larcher, G. Amatucci, B. Gerand, J.M. Tarascon,
lectrochern. Soc. 146 (1999) 428.
Xia, M. Yoshio, J. Electrochern. Soc. 143 (1996) 825.
Huang, C.A. Vincent, P.G. Bruce, J. Electrochern. Soc. 146 (1999) 3649.
. Kannan, A. Manthiram, Electrochern. Solid-State Lett. 5 (2002) A167.
hin, A. Manthiram, Electrochern. Solid-State Lett. 5 (2002) ASS.
M. Thackeray, A. de Kock, M.H. Rossouw, D.C. Liles, D. Hoge, R. Bittihn, J.
trochern. Soc. 139 (1992) 363.
Gummow, A. de Kock, M.M. Thackeray, Solid State lonics 69 (1994) 59.
hoi, A. Manthiram, J. Electrochern. Soc. 147 (2000) 1623.
M. Thackeray, M.F. Mansuetto, D.W. Dees, D.R. Vissers, Mater. Res. Bull. 31 (1996)
.
Gao, J.R. Dahn, J. Electrochern. Soc. 143 (1996) 1783.
M. Thackeray, M.F. Mansuetto, C.S. Johnson, J. Solid State Chern. 125 (1996) 274.
im, A. Manthiram, J. Electrochern. Soc. 145 (1998), L53.
Sigala, D. Guyomard, A. Verbaere, Y. Piffard, M. Toumoux, Solid State Ionics 81
95) 167.
Kawai, M. Nagata, H. Takamoto, A.R. West, Electrochern. Solid-State Lett. I (1998)
.
Zhong, A. Bonakdarpour, M. Zhang, Y. Gao, J.R. Dahn, J. Electrochern. Soc. 144
97) 205.
Ein-Eli, W.F. Howard, Jr., S.H. Lu, S. Mukerjee, J. McBreen, J.T. Vaughey, M.M.
ckeray, J. Electrochern. Soc. 145 (1998) 1238.
Kawai, M. Tabuchi, M. Nagata, H. Tukamoto, A.R. West, J. Mater. Chern. 8 (1998)
3.
awai, M. Nagata, H. Tukamoto, A.R. West, J. Power Sources 81-82 (1999) 67.
hao-Hom, R. L. Middaugh, Solid State Ionics, 139 (2001) 13.
Chapter 1
. West, H. Kawai, H. Kageyama, M. Tabuchi, M. Nagata, H. Tukamoto, J. Mater.

rn. 11 (2001) 1662.
hin, A. Manthiram, Electrochirn. Acta (submitted).
ark, Y. Han, Y. Kang, P.S. Lee, S. Ahn, H. Lee, J. Lee, J. Electrochern. Soc. 148
01) A680.
ho, G.B. Kim, H.S. Lim, C. Kim, S. Yoo, Electrochern. Solid State Lett. 2 (1999) 607.
weon, G. Kim, D. Park, US Patent No. 6,183,911 (2001).
hin, A. Manthiram, Electrochern. Solid-State Lett. 6 (2002) A34.
hin, A. Manthiram, J. Electrochern. Soc. (submitted).
.W. Murphy, R.J. Cava, S.M. Zahurak, A. Santoro, Solid State Ionics 9 & 10 (1983)
3.
Ferg, R.J. Gummow, A. de Kock, M.M. Thackeray, J. Electrochern. Soc. 141 (1994)
147.
A. de Picciotto, M.M. Thackeray, Mat. Res. Bull. 20 (1985) 1409.
. Manthiram, J.B. Goodenough, Can. J. Phys. 65 (1987) 1309.
Choi, A. Manthiram, J. Solid State Chern. 164 (2002) 332.
. Manthiram, J.B. Goodenough, J. Power Sources 26 (1989) 403.
. Manthiram, J.B. Goodenough, J. Solid State Chern. 71 (1987) 349.
.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, S. Okada, J.B. Goodenough, J.
ectrochern. Soc. 144 (1997) 1609.
.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, J.B. Goodenough, J. Electrochern.
c. 144 (1997) 2581.
.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochern. Soc. 144 (1997)
188.
. Yamada, S.C. Chung, K. Hinokuma, J. Electrochern. Soc. 148 (2001) A224.
. Huang, S-C. Yin, L.F. Nazar, Electrochern. Solid-State Lett. 4 (2001) A170.
P. Prosini, M. Carewska, S. Scaccia, P. Wisniewski, S. Passerini, M. Pasquali, J.
ectrochern. Soc. 149 (2002) A886.
Y. Chung, J.T. Bloking, Y.M. Chiang, Nature Materials 1 (2002) 123.
.M. Doeff, T.J. Richardson, L. Kepley, J. Electrochern. Soc. 143 (1996) 2507.
.U. Jeong, A. Manthiram, Electrochern. Solid-State Lett. 2 (1999) 421.
Tsang, A. Manthiram, J. Electrochern. Soc. 144 (1997) 520.
. Lampe-Onnerud, J.O. Thomas, M. Hardgrave, S. Yde-Andersen J. Electrochern.
c. 142 (1995) 3648.
Wilhelmi, K. Waltersson, L. Kihlborg,Acta Chern. Scand. 25 (1971) 2675.
Theobald, R. Cabala, J. Bernard, J. Solid State Chern. 17 (1976) 431.
.L. Tipton, S. Passerini, B.B. Owens, W.H. Smyrl, J. Electrochern. Soc. 143 (1996)
73.
M. Kannan, A. Manthiram, J. Electrochern. Soc. (in press).
. Kumagai, A. Yu, J. Electrochern. Soc. 144 (1997) 830.
Takeda, R. Kanno, Y. Tsuji, 0. Yamamoto, J. Power Sources 9 (1983) 325.
.B. Owens, S. Passerini, W.H. Smyrl, Electrochirn. Acta 45 (1999) 215.
M. Kannan, A. Manthiram, Solid State lonics (in press).
. Yamamoto, Y. Takeda, R. Kanno, Y. Oyabe, Y. Shinya, J. Power Sources 20 (1987)
1.
Kim, A. Manthiram, J. Electrochern. Soc. 144 (1997) 3077.
Kim, A. Manthiram, Nature 390 (1997) 265.
J. Xu, A.J. Kinser, B.B. Owens, W.H. Smyrl, Electrochern. Solid-State Lett. 1 (1998)
Kim, A. Manthiram, Electrochern. Solid-State Lett. 2 (1999) 55.

. Im, A. Manthiram, J. Electrochern. Soc. 149 (2002) A1001.
. lm, A. Manthiram, J. Electrochern. Soc. (in press).
. Im, A. Manthiram, Solid State lonics (in press).
Kim, A. Manthiram, J. Electrochern. Soc. 146 (1999) 4371.
. Choi, A. Manthiram, J. Electrochem. Soc. 149 (2002) A570.

.R. Home, U. Bergmann, J. Kim, K.A. Streibel, A. Manthiram, S.F. Cramer, E.J.
airns, J. Electrochem. Soc. 147 (2000) 395.
. lmanishi, Y. Takeda, 0. Yamamoto, in Lithium Ion Batteries: Fundamentals and
erformance (M. Wakihara and 0. Yamamoto, Eds.), 1998, p. 98, Wiley-VCH,
einheim.
. Winter, J.O. Besenhard, in Lithium Ion Batteries: Fundamentals and Performance
M. Wakihara and 0. Yamamoto, Eds.), 1998, p. 127, Wiley-VCH, Weinheim.
. Liu, X. Huang, Z. Wang, H. Li, L. Chen, J. Electrochem. Soc. 145 (1998) 59.
J. Aubom, Y.L. Barbeiro, J. Electrochem. Soc. 134 (1987) 638.
.M. Abraham, D.M. Pasquariello, E.B. Willstaedt, J. Electrochem. Soc. 137 (1990)
43.
. Morzilli, B. Scrosati, F. Sgarlatta, Electrochim. Acta 30 (1985) 1271.
.C.S. Souza, V. Pralong, A.J. Jacobson, L.F. Nazar, Science 296 (2002) 2012.
.D. Kepler, J.T. Vaughey, M.M. Thackeray, Electrochem. Solid-State Lett. 2 (1999)
07.
L.C. Rowsell, V. Pralong, L.F. Nazar, J. Amer. Chern. Soc. 123 (2001) 8598.
T. Vaughey, K.D. Kepler, R. Benedek, M.M. Thackeray, Electrochem. Comm.
999) 517.
pter 2
THE ROLE OF ELECTRONIC PROPERTIES IN

THE ELECTROCHEMICAL BEHAVIOR OF
INTERCALATION COMPOUNDS FROM A
FIRST PRINCIPLES VANTAGE POINT
A.Vander Ven and G. Ceder

Department of Materials Science and Engineering,
Massachusetts Institute ofTechnology, Cambridge, MA, U.S.A.
INTRODUCTION
Lithium transition metal oxides are remarkable intercalation

ounds, exhibiting an array of intriguing electronic and phase
ormation phenomena. Their ability to undergo large variations in lithium
ntration, often without suffering irreversible changes, makes these
ials ideal electrodes for rechargeable lithium batteries. Lithium
ion metal oxides consist of a metal oxide host with a crystal structure in h
lithium ions occupy a relatively open network of interstitial sites. e
some lithium transition metal oxides can serve as an anode in
rgeable lithium batteries, most are more suited for the role of cathode as
typically exhibit a high voltage with respect to a metallic lithium anode,
deal theoretical reference electrode. In a rechargeable lithium battery,
m ions are shuttled between an anode and a cathode, whereby lithium
are removed from and inserted into the electrodes. During deintercalation
ithium transition metal oxide, vacancies are created on the lithium sites
e host. During intercalation, these sites are refilled by lithium. Removal
insertion of lithium ions can lead to several phenomena that can
icantly affect the electrochemical properties of the compound. For one,
ions in lithium concentration alter the electronic properties of the
ole of Electronic Properties ... 43
ition metal oxide host. The valence electron of each lithium ion is
rally donated to the host where it can either shift the valence state of the
ition metal ion and/or alter the nature of the bonds between the transition
l and the oxygen ions. Simultaneously, lithium removal from the host
structurally destabilize the metal oxide structure or may lead to order
der phase transitions between lithium and vacancies once a critical
ncy concentration is reached. These phenomena often affect the voltage
cteristic or the lattice parameters of the compound in important ways.
Within the last decade, much attention has been devoted to
1
rstanding the properties of LixCo0 2, LixNi02, LixMn02, LixMn204 and
ePO/, currently among the most important candidate cathode materials
echargeable lithium batteries. Other compounds receiving attention are
ally doped variants of the Co, Ni and Mn compounds. While many
m transition metal oxides have similar crystal structures, either a layered
or one derived from the spinel structure (LixFeP04 has the olivine crystal
ture), they are often characterized by very different electrochemical
rties owing to the unique electronic structure of each transition metal.
o02, for example undergoes a concentration driven metal-insulator phase
ormation whereby the electrons over much of the lithium concentration
(x<0.75) are delocalized and exhibit metallic properties? In contrast, the
n204 compound in important lithium concentration regions is
cterized by more localized electrons with large magnetic moments
46
ng to phenomena such as charge ordering and Jahn-Teller distortions. -
, phenomena such as these can be rationalized with crude crystal-field

molecular orbital models. More subtle properties such as the relative
ity of similar crystal structures, stable ordered lithium-vacancy
gements and charge-ordered and magnetic-ordered electronic states are
difficult to predict or rationalize with intuition based on simple
onic structure models. It is here that first-principles numerical-schemes
olving the Schrodinger equation of solids are proving invaluable. The
d of the last decade of first principles studies of lithium transition metal
Chapter 2
es has demonstrated that these methods are an important new tool in

40
cting and understanding the properties of these materials. 7-
The state of the art in first principles electronic structure methods for
s is based on density functional theory (DFf).41- 43 Density functional
y casts the problem of solving the many-body Schrodinger equation of
olid into an equivalent problem of solving for the electronic charge
ty of the solid. While density functional theory is exact in principle, its
ical implementation for real solids requires approximations, the most
mon of which are referred to as the local density approximation (LDA)
he generalized gradient approximation (GGA). The properties that can be
tly derived with density functional theory are the charge (or spin) density
he total energy of the solid. Other properties can only be calculated if
cal models relating them to energies or charge densities exist. By using
7 8 10 12 15 17
odynamic and statistical mechanical models, voltages • • • • • ,
ve stabilities 11'14' 18' 27 and kinetic properties such as atomic migration

33 36
anisms and rates31' ' can be predicted. The first principles study of
m transition metal oxides has also received an important boost with
opments in alloy theory during the last twenty years, most notably the
opment of the cluster expansion27' 44.4 5 which enables the statistical
anical treatment of solids with significant configurational disorder. In
alation compounds this capability is critical due to the many degrees of
om of distributing lithium ions and vacancies over the interstitial sites.
arly, the disorder in mixed compounds with multiple transition metal
28
ents can only be described properly with cluster expansion techniques.
combination of accurate DFf calculations with sophisticated statistical
anics techniques have enabled accurate calculations of phase diagrams
'
39
tercalation compounds, 18 28' voltage intercalation curves 12' 15' 17 and
m diffusion coefficients.31•36
In this chapter, we focus on a description of important electronic
rties of LixCoOz, LixNi02 and LixMn02 • Most of our information will be
ed from first principles calculations. The chapter starts with an overview
and-field and molecular orbital theory, and continues with a description

rst principles density functional theory. We next illustrate how each
y can be used to understand and, in the case of density functional theory,
ct the electronic properties of LixCo0 2, LixNi02 and LixMn02. We
asize the electronic properties that are predicted by density functional
y but would not be a priori expected from ligand field or molecular
al models. In the case of LixCo02, density functional theory has shown
important role of covalency in influencing the structural and
odynamic properties of the host. In LixNi02, DFT calculations have
sed the existence of an intriguing long-range attraction between lithium
of different lithium planes mediated through a Jahn-Teller distorted Ni+3
edron. This long-range attraction, termed a 180° interaction, leads to
ochemical properties in LixNi02 that differ significantly from those
ved in LixCo02 even though both compounds have the same crystal
ure. Finally, in LixMn02, DFT calculations have predicted a charge
oportionation reaction between the Mn ions, creating Mn+2 that favor
edral sites. This mechanism is responsible for the rapid transformation
inel that occurs during cycling of layered LixMn02, an irreversible
ormation that has prevented commercialization of layered LixMn02 for
m battery applications.
In all three examples, the first principles predictions expose
anisms that play a central role in determining the electrochemical
rties or stability of these materials as cathodes. Understanding the link
en electrochemical properties and their electronic origins is invaluable
imizing these materials.
GENERAL ELECTRONIC STRUCTURE OF

OCTAHEDRAL TRANSITION METAL IONS
The ability of lithium transition metal oxides to undergo large

ions in lithium concentration is attributable to the flexibility of the
Chapter 2
ce electronic structure formed by the transition metal and oxygen ions.

7
um is generally completely ionized within the oxide, having donated its
ce electron to the host. The electronic properties of the transition metal
e are dictated to a large extent by the number of valence electrons and by
nteraction of the d-orbitals of the transition metal with the p-orbitals of
xygen ions.
In both the layered and spinel crystal structures, the transition metal is
edrally coordinated by oxygen. Figure 2.1 illustrates the electronic e
densities of hydrogenic d-orbitals within an octahedral oxygen
onment. Although the five d levels when occupied by a single electron
de of a closed shell core are degenerate in a free transition metal ion, in
tahedral environment, this degeneracy is broken. Crystal field theory 46-49 ,
xample, shows that the interaction of the d orbitals with the electrostatic
tial due to the negatively charged oxygen ions of the octahedron
ces an increase in the energy of the d 3z2-r2 and dx2-y2 orbitals with
ct to the energy of the dxy• dxz and dyz orbitals. This occurs because the
2 and dx2-y2 orbitals have lobes pointing toward the negatively charged
en ions while the lobes of the dxy• dxz and dyz orbitals point between
en ions.
Only considering the effect of the electrostatic field of the oxygen ions
cts the important role of covalency that can exist between the transition
and oxygen orbitals. A more accurate picture of the interaction between
ansition metal ion with the surrounding oxygen ions arises from
4649
cular orbital theory. The d3z2-r2 and dx2-y2 orbitals directly overlap
he Px• Py and Pz orbitals of oxygen forming bonding and antibonding
evels referred to as egb and eg*· These levels are illustrated
matically in Figure 2.2. Since the energy of the p orbitals are below the
y of the d orbitals, the bonding egb levels are predominantly of oxygen p
cter while the antibonding eg* levels (frequently referred to as simply eg)
st mainly of metal d states. This dichotomy in character between the
2.1. Hydrogenic d-orbitals within an octahedral oxygen environment. (a) The d3z2-,2
) the dx2-y2 orbitals have lobes pointing toward the oxygen ions. (c) The dxy orbital,
is symmetrically equivalent to the dxz and dyz orbitals, has lobes pointing between
n ions.
ng and antibonding levels becomes more pronounced as the difference

energy of the metal d-levels and the oxygen p-levels widens. The dxy•
nd dyz orbitals which do not directly overlap with oxygen p orbitals to
cr bonds, are frequently said to form a set of nonbonding levels denoted
46 49
. - A band gap llo separates the t 2g levels from the antibonding eg levels.
octahedral environment, the t2g levels are below the eg levels. Although
y• dxz and dyz orbitals do not form cr bonds with the oxygen ions, they do
n bonds with the oxygen p orbitals.46
Instead of being non-bonding levels, t2g energy levels of Figure 2.2,
accurately correspond to the bonding levels of the 1t overlap between dxy•
d dyz and the oxygen p levels. Increased covalency of this n bond will
in a lowering of the energy of the t2g levels and hence an increase in the
gap L10 •46
Other levels arise from the overlap of oxygen p-levels with the
tion metal 4s and 4p levels. These are referred to as t 1u and a 1g. Since the
y of the oxygen p levels are again below the energy of the metal 4s and
els, the bonding t 1u and a 1g levels are predominantly of oxygen character.
Chapter 2
it: \
\
4p \
\
\ \ .... ---..;a;:;_g...., \
\
\
4s ---...--';-- e*
\
\
\
\
\ \
g \ \
.
\ \ \ \
\ \ \ \
\ \ ' \ \ \
\ \ ,' \
\
\
\
\
\
\
>' \ \ \ \
-:.' g
\ _,,\ =
\ \\
3d \\\
\ ,,
,,,
,,
_ \ \ \
\
\
\
\
\
\
b
,,
\ \ \
\ \ ':==l:i§u= ---,",',
\ \ \ 2p
:-
\ \
\ \
'
,','
\ \
\ \ b
alg '
''
\ \
\ ''--"'-- '
''
\
\
''
\
' Oxygen
Metal '·===='
2.2. Schematic illustration of the bonding and antibonding levels that arise for a
ion metal ion in an oxygen octahedral environment due to the hybridization between the
e electronic states of the transition metal ion with those of the oxygen ions.
In a crystal, the levels of Figure 2.2 become bands. Furthermore,

ugh the transition metal ions reside in octahedral interstitial sites, the
inder of the crystal outside of the octahedral site generally does not
it the symmetry of a perfect octahedron. Hence the degeneracy of the
s is further broken. Nevertheless, the schematic picture of Figure 2.2 is
ly still identifiable in the band structure.
While general trends can be rationalized with simple crystal field and
cular orbit models, more accurate numerical methods based on density
ional theory are necessary to understand and predict the relationship
en the electronic structure of a compound and specific electrochemical
rties.
FIRST PRINCIPLES METHODS
Calculating properties of solids from first principles without

imental input involves solving the many body Schrodinger equation for
50 • 51
olid. The time-independent Schrodinger equation (in this section, all
ions are expressed in atomic units) is an eigenvalue equation
(2.1)
e H is the Hamiltonian operator for the solid, '¥ is the many body wave
ion describing Ne electrons and E is the total energy of the solid. Within
Born-Oppenheimer approximation, the electrons are assumed to
ntaneously adjust their state to any change in the positions of the nuclei.
coordinates of the nuclei Rn then serve only as parameters in the
dinger equation and the Hamiltonian can be written as
H = T +Vee+ L v(;:;)+ L L
-ZnZn:_ I
j n m<n
Rm -Rn
he electronic kinetic energy operator
(2.3)
ee describes the Coulomb interactions between the different electrons
(2.4)
the rj refer to the positions of the electrons. The sums in Eqs. (2.2),
Chapter 2
and (2.4) are over all the electrons in the solid. The third term in Eq.
describes the coulomb interaction between the electrons and the nuclei
e solid and the last term is the Ewald energy arising from the coulomb
action between the nuclei having charge Zm· The Ewald energy is simply
ditive term and does not affect the electronic wave function \f. It can be
ped when solving the eigenvalue equation for the electronic ground state
nly needs to be added to the electronic energy to obtain the total energy
e crystal.
For solids of practical interest, solving the many-body Schrodinger
ion is intractable. One approach to solving Eq. (2.1) is with density
ional theory (DFT).41-43 As was shown by Hohenberg and Kohn (and
41 52
ded by Levy), ' the ground state properties of a crystal are uniquely
mined by the electron density p(r) (which is determined by the many-
wave function \f). The fundamental theorems of density functional
y state that the ground state energy of a solid is a functional of the
on density41
E[p] = F[p]+ fp(r)v(r)dr (2.5)
(2.6)
versal functional and v(r) the coulomb potential due to the nuclei of the
The lJ' in Eq. (2.6) is the electronic wave function that minimizes
+ Vee I' ') subject to the constraint that lJ' produces the electron density
ccording to DFT, if the functional F[p] were known, the ground state
y of the solid with external potential v(r) is obtained by variationally
mizing the functional E[p] with respect to p41•42 The minimization
eds over the set of all densities p that can be obtained with
mmetric wave functions.5 2
F[p] is not known and approximations to it are therefore necessary.

ally, F[p] can also be written as
(2.7)
e the kinetic energy and the electron-electron interaction energy are

idually expressed as functionals of the electron density. Of the two terms
. (2.7), the kinetic energy T[p] is the most elusive to approximate as a
42 43
ional of p. 53 introduced a
' To side step this difficulty, Kohn and Sham
rent separation of F[p] by writing it as:
F[p] = Ts [p ]+ J[p]+ Exc [p] (2.8)
e T8 [p] is the kinetic energy of a system of non-interacting electrons with

ame density p. J[p] is often referred to as the Hartree term and is a
ical Coulomb energy given by
J[p] = _!_ ffp(r)p(r')drdr'. (2.9)

2 lr -r'l
last term Exc[p], called the exchange-correlation energy, which can be
n as
(2.10)
cludes the difference between the kinetic energy of the actual system of
cting electrons with density p and the kinetic energy of a system of
endent electrons with the same density. This difference, though, is
42
ally expected to be small and Exc primarily accounts for a correction to
arising from the correlations between electrons. Since the electron
ty corresponds to a probability distribution and not a charge density in
Chapter 2
lassical sense, J[p] is a mean-field approximation to the electron-electron

actions. Embodied in J[p] is the assumption that the probability of having
electrons at r and r' respectively is equal to the probability of having
lectron at r times the probability of having another electron at r' .
ected are the conditional probabilities resulting from the fact that if an
ron is already present at r , the probability of having another electron at
ill be different than the average probability. These correlations between
of electrons are implicitly accounted for by Exc[p].
Correlations between electrons, which become more important as r
r' approach each other, arise from two effects. The first type of
lation results from electrostatic repulsions which ensure that electrons
the vicinity of other electrons. These are called direct correlations. The
nd type of correlation has its origin in the Pauli exclusion principle which
ds electrons with parallel spins from having the same spatial position
inates. This effect does not arise from the many body Hamiltonian, but
forced by the fact that the wave function is antisymmetric. Since the
function is continuous, the Pauli exclusion principle implies that the
bility of having electrons with the same spin approach each other is
. This correlation is referred to as an exchange correlation. The energy
ibution of exchange correlation is typically an order of magnitude larger
that of direct correlation.43
The advantage of introducing the non-interacting kinetic energy
ional T.[p] is that it can be calculated exactly with a Slater determinant
f independent electron orbitals If/ 1(r). The independent electron kinetic
y functional then becomes42
r,fp] = (' 'niTI'Yn) = t Jvj(r{- v 2

) (2.11)
vj(r)dr
1
e '¥0 and the If/ 1 are such that they produce the electron density p.
e it is possible that more than one 'I'n could produce the same density, the
cular 'I'n appearing in Eq. (2.11) is the one that minimizes ('PviTI'Pv)
xed p.
With the 'I'n and p expressed in terms of the independent orbitals
Kohn and Sham53 variationally minimized the energy functional
(2.12)
obtained what are now referred to as the self-consistent Kahn-Sham

ions
(2.13)
Veff
(- )r -V
Jipr (_rr')'l d-'
+
Vxc .
(-) (2.14)
( -)
r
+ T T
q. (2.14), v xc (r) is the exchange correlation potential and is the

ional derivative of the exchange correlation energy functional
-)- &'xc[p] (2.15)

Vxc (r - Jp(r)
An equation of the form of Eq. (2.13), exists for each electron in the
The Kohn-Sham procedure replaces the many body eigenvalue equation,
(2.1) and (2.2), with a set of independent-electron-like eigenvalue
ions. Despite their appearance, however, the Kohn-Sham equations are
dependent due to the dependence of Veff on the density p (Eq. (2.14))
Chapter 2
h itself is determined by all the orbitals 1f1 j. This means that the Kohn
equations are to be solved self-consistently.
1. The Local Density Approximation (LDA)
The Kohn-Sham equations as represented in Eq. (2.13) are exact

ded the universal functional Exc[P] (and hence its functional derivative
r)) is known. As with F[p] of Eqs. (2.5) and (2.6), Exc[P] is not known.
most common approximation to Exc is the local density approximation
53
A). Within LDA, Exc is written as
(2.16)
e Exc is the exchange correlation energy per electron at r. exc(p(r)) is

qual to the exchange correlation energy per electron of a homogeneous
on gas with the same density p(r) . LDA therefore assumes that Exc is
and neglects the effects of non-uniform charge densities around r.
e are different parameterizations of Exc as a function of the homogeneous
ron gas density p. The local density approximation has proven
rkably successful considering its simplicity. Typically though,
ibrium lattice parameters, volumes and band-gaps of solids are under
cted within LDA.
. The Generalized Gradient Approximation (GGA)
The generalized gradient approximation (GGA) includes in the

ximation to Exc[P] a term that depends on the gradient of the electron
ty at each point. While the inclusion of gradient corrections to the local
ty approximation often improves the accuracy of predictions, this is not
ally guaranteed and for some solids, properties predicted within LDA
more with experiment than those predicted within GGA. Typically,
predicts lattice parameters and volumes that are slightly larger than
observed experimentally.
. Spin Polarization
Many solids, especially transition metal oxides, are characterized by

magnetic moment produced by an unequal electron population of +112
1/2 spin states. This is true of many atoms as well, as filling of the
ic orbitals according to Hund's rules often results in a net electronic spin.
of Hund's rules arises from exchange and correlation interactions
en electrons. Due to the Pauli exclusion principle, two electrons with the
electronic spin cannot occupy the same position in space. Hence,
ons with parallel spin do not occupy atomic orbitals with the same l
distribution, minimizing direct electrostatic repulsion. No such
um-mechanical restriction exists on a pair of electrons with opposite
Electrons with opposite spin can therefore occupy atomic orbitals with
cal spatial distribution but are then energetically penalized by enhanced
ostatic interactions.
To account for the possibility of spin polarization within practical
mentations of density functional theory, the total electron density is
ated into an electron density pa(i=) for the spin +112 electrons and an
on density p/3 (r) for the spin -112 electrons. While such a separation is
rictly necessary within exact density functional theory, it enables more
ate approximations to the exchange-correlation functional whereby a
ent exchange-correlation potential is introduced between electrons with
el spins and electrons with opposite spin.
. Caveats About Approximations to DFT

Chapter 2
At this point, several caveats of DFf and LDA and GGA in

cular deserve attention. The orbitals If/ j and orbital energies c j of the
n-Sham equations do not correspond to real electronic states and
ronic energy levels. It is only the total energy and electron density that
any physical meaning. Nevertheless, the Kohn-Sham energy levels often
ive a good characterization of the band structure of crystalline materials
are frequently compared to quasi-particle energies measured in photo
sion experiments. The kinetic energy as expressed in terms of the If/ j by
2.11), is not the real kinetic energy of the solid, though it is a close
42
oximation of it. The local density approximation is exact for a uniform
ron density and a good approximation for a slowly varying one.
ermore, for many solids exhibiting rapidly varying electron densities,
has proven surprisingly accurate. Yet, for solids in which the electronic
s are highly localized in space, we can expect LDA to break down since it
ot be expected to capture the strong correlations between the localized
rons.
For systems with well localized electrons, specific inadequacies of
cal density approximation can be identified. One inadequacy is that the
ximation to the exchange-correlation potential Vxc depends only on the
density and not on the specific orbital that a particular localized electron
pies. In the Hartree-Fock approximation for example, the exchange
y, which is exact within the approximation of the anzats wave function
er determinant consisting of single electron orbitals), is a non-local
ral between different electron orbitals. These integrals not only depend
e local electron density, but also on the shape of the different orbitals.
ar orbital dependencies can be expected to arise from direct correlations.
54
ods such as LDA+U attempt to remedy this deficiency of LDA.
her major inadequacy of LDA is that it has self-interaction (see, e.g., Ref.
In reality, electrons interact with other electrons. They do not interact
ostatically with themselves. In the Hartree term J[p] defined by Eq. (2.9)
e energy functional Eq. (2.12), a coulomb interaction of each electron

itself is included. For the exact Exc[P] functional, this self-interaction
onent is canceled out. (In the Hartree-Fock approximation, the self
action in the Hartree term is explicitly canceled out by an identical term
e exchange term). When an approximation for Exc is used, however, the
nteraction does not cancel. This error becomes severe for well localized
onic states. LDA calculations with self-interaction corrections were
ered by Perdew and Zunger 55 and have been used to investigate the
onic properties of several transition metal oxides56 and
57
conductors. Incorporating self-interaction corrections leads to orbital
58
ndent effective potentials which complicates the numerical solution of
nergy and charge density, and as a result it is rarely done.
. Different Numerical Schemes
Many different numerical techniques exist for solving the Kahn

equations. For oxides, the most reliable methods have proven to be the
r Augmented Plane Wave (LAPW) method59 and the pseudopotential
60 61
od. • The LAPW method is currently considered the most accurate and
the standard, yet it is computationally the most costly. In the LAPW
od, all electrons of the solid are treated and their states are expanded in a
consisting of both local atomic-like orbitals and plane waves. In the
opotential method, the effect of the core electrons around the ions that
ot participate in bonding are replaced by a pseudopotential. The
opotentials are designed such that the valence single electron pseudo
functions have the same scattering properties as the actual valence
ons would have with the core electrons. The pseudopotential
ximation is valid as long as the core electrons do not participate in the
ng of the solid - that is the changes in their wave functions and energy
are negligible when the atom is placed in different environments. For
solids this approximation is valid. Comparisons of energy differences for
Chapter 2
ixCo02 system has shown that the pseudopotential method and LAPW
62
lations agree to within 10 meV per LixCo02 formula unit. Modem
dopotentials are determined from all electron calculations of atoms,
63 68
ng the pseudopotential method a first principles approach. -
The general procedure of solving the Kahn-Sham equations is to

nd the orbitals 1f1 j (r) in terms of a set of basis functions. In
dopotential calculations, these basis functions are usually plane waves.
ELECTRONIC STRUCTURE OF LixCoOz
LixCo02 is currently the most important transition metal oxide

ound for battery applications and is ubiquitous in commercial lithium
atteries. In this section, we describe important electronic properties of
o02 as predicted with DFf-LDA calculations. Striking is the important
hat shifts in covalency play in determining relative stability and lattice
meter variations with lithium concentration.
. Covalent bonding in LixCo02
Figure 2.3 illustrates the LDA-calculated electronic band structures

o02 and LiCo02.8 "69 The lowest six bands correspond to the egb• t 1u and
vels of Figure 2.2 which we will refer to as the oxygen p-levels. The next
levels are the t2g levels and the highest two levels correspond to the anti
ng eg bands. In LiCo0 2, the Fermi level lies between the t2g and the eg
s, making the compound a semiconductor. This is consistent with the
70
imental observations of van Elp et al. When all the lithium ions are
ved, the Fermi level resides in the t 2g bands as one electron per Co02
ula unit is removed.
The band structures of LiCo02 and Co02 are qualitatively similar.
mportant difference between the two band structures is the distance
en the oxygen p like levels and the t 2g levels which is larger in LiCo02
in Co02. The electron donated by Li when inserted in the Co02 host

es in a t2g level. Through coulombic interaction, the addition of the extra
on to the t2g levels raises the energy of the other occupied t 2g levels thus
asing the energy difference between the metal d-states and the oxygen p
. This results in a reduction in the hybridization between the Co d-levels
the oxygen p-levels which translates into a change in the nature of
ing from a more covalent character in Co02 to a more ionic character in
02•
To better understand this shift in bonding characteristics with Li
18
ion, it is useful to inspect a charge difference plot. Figure 2.4 shows a e
difference plot between LixCo02 at x=l/4 and Co02 both in the layered
tructure of Figure 2.5. Electronic charge densities were calculated with
seudopotential method in the local density approximation for both
18
o02 and Co02 • The lattice parameters and the positions of Co and 0
the same in the two structures to enable a point by point subtraction of
harge density of Co02 from Li 114Co02• The resulting difference shows
the electron distribution changes when Li is added to the host, and in
ular, where the electron of the added lithium resides. The plane
ated in Figure 2.4 corresponds to the shaded plane shown in Figure 2.5.
plane cuts through Li, Co and 0 ions.
The charge difference plot illustrates that lithium intercalation
17' 18
es a significant redistribution of charge within the Co02 host. Figure
hows that there is an accumulation of charge density around all the Co
n lobes pointing in directions between oxygen ions. This increase in e
results from an electron addition to the partially filled t2g bands as m
is added to the host. The charge accumulation occurs around all the
t ions indicating that these states are delocalized. Figure 2.4 also shows
here is a significant depletion in electron density around the Co ions
st to the lithium ions. The charge depletion occurs in regions that
ble the charge density of a dx2-y2 atomic orbital (see Figure 2.1). The
atomic orbital together with the oxygen p orbitals form the cr bonding
Chapter 2
-> 5
>.
0
...... l
O
(])
c
UJ
-5
z r F L r
(a)
> 0
-
>.
Ol
Cii
c
UJ
-5
p
r--....
-- " =--
"
--------- - - -:::::::
-10
z r F L r
(b)
2.3. The partial band structures of (a) Co0 2 and (b) LiCo02 as calculated within the
ensity approximation. The dashed line shows the Fermi level.
nd anti-bonding eg * bands. The depletion around the Co ions coincides

a significant increase in charge density in atomic p-like orbitals on
en ions which point toward the lithium ions and also overlap with the
ted cobalt dx2-y2 like lobes.
The simultaneous depletion of charge in dx2-y2 like orbitals around t
and accumulation of charge around oxygen indicates a change in the
ization of the cr eg bonds between oxygen and cobalt. Since the
onding eg * bands are unoccupied in LixCo02, the polarization of the cr
reflects changes in the nature of the bonding egb bands. As lithium is
, the egb bands obtain more of an oxygen character and less of a Co
2.4. Charge difference plot between Li 114Co02 and Co02• Dark shade signifies regions
rge accumulation and light areas correspond to regions of charge depletion.
--c o cobalt
•Lithiu
--B
Qoxygen
--A
c
--c
--B
--A
2.5. The 03 crystal of LiCo02. The crystallographic plane on which the electronic
density difference is viewed in Figure 2.4 corresponds to the shaded plane in this figure.
Chapter 2
cter. This causes the cr bond between Co and oxygen to become less
ent and more ionic with increasing x. The result is a significant increase
e electron density at the oxygen sites immediately surrounding the
m ions. The net effect is that the electron transfer from lithium to the
s very local and occurs predominantly to the oxygen ionsY
The conventional picture of lithium transition metal oxides is that the
on donated to the host by lithium is transferred to the transition metal
7 8 17 18
e it shifts its valence state. First principles work within LDA ' ' • show
he electronic changes of the host with x are in fact more subtle. The
ion of lithium to the Co02 host causes a shift in the nature of the bond
een oxygen and cobalt from a covalent character at low lithium
entration to a more ionic character at high x. While the electron donated
hium is added to the t2g band, which is predominantly of Co d character,
ffect of this addition is to polarize the Co-O bond such that the charge
d the oxygen ions actually increases upon lithium insertion (it is
rtant to realize that oxygen is not in its fully ionized state of -2 in Co02).
e, while it is reasonable to talk about formal valence changes of Co upon
ertion/removal, one has to keep in mind that the rehybridization around
obalt ions occurs in a manner that ensures that the local net charge d
17
the cobalt ions remains fixed as the lithium concentration changes.
. Experimental Evidence for Shift in Covalency
Both X-ray absorption spectroscopy (XAS) and electron energy loss

roscopy (EELS) have been used to study the local electronic properties
xCo02 as a function of x and have provided evidence consistent with a
71 73
in covalency with varying lithium concentration. - Both XAS and
measurements rely on transitions of core electrons to unoccupied states
e the Fermi level. Of particular importance are the 0 K-edge, resulting
the excitation of an 0 ls cores electron into unoccupied 0 2p states, and
o Lz.3 edge, arising from the excitation of Co 2p electrons to unoccupied
d states. As the lowest unoccupied states in LixCo02 are those made up of

ybridization of the Co 3d and 0 2p levels, the 0 K-edge and the Co
Lz.r supply useful information about the Co-O bond. Variation of
the rum with lithium concentration sheds light on the nature of bonding
and ates where the charge of the electron from the intercalating lithium
goes. xcitations of core electrons with XAS and EELS are local
phenomena election rules derived from electric dipole allowed
transitions isolate defined excitations. Hence the 0 K-edge spectra,
for example, offer mation about the unoccupied 2p states projected on
the oxygen sites as ansition from 0 ls to Co 3d is forbidden within
the electric-dipole ximation.
Measurements of the Co Lz.redge in LixCo02 as a function of lithium
ntration show only minor variation of the spectra with lithium
ntration.71 -73 The Co Lz.3 spectrum consists of two major peaks, Lz, and,
rising from excitations of Co 2p312 and 2p112 core electrons to unoccupied
d states. The splitting of Co core 2p levels is induced by spin-orbit
ing. While Graetz et al.72 were unable to discern a variation with x in the
73
ity of the Lz.3 peaks, Yoon et al. were able to quantify a very slight
se to higher energy in the peak positions as LixCo02 is deintercalated.
The variation of the 0 K-edge spectra with lithium concentration is
more pronounced 71- 73 than that of the Co Lz.redge. Even in fully
ed LiCo02, a well defined transition from the core 0 ls state to the
upied 2p state is observed, a feature that is absent in completely ionic s
where the oxygen 2p states are locally fully occupied. This result tes
some degree of covalency between the Co-O bond in LiCo02• As
02 is deintercalated, the 0 K-edge spectra change dramatically.71 -73
ercalation is accompanied by an increased broadening of the 0 K near
pre-peak as well as an increase in the integrated intensity of the pre
2 In fact Yoon et al.73 identified the growth of three new shoulder peaks
deintercalation that emerge around the original peak corresponding to
ls to 2p transition of fully lithiated LiCo02•
Chapter 2
The pronounced variation of the 0 K-edge spectra compared to the

r variation of the Co .redge spectra indicates the important role that
en plays in the states near the Fermi level. The increase in integrated
sity of the 0 K near edge pre-peak upon deintercalation implies electron
tion of the 0 2p states and hence a reduction in the local charge density
71 72
nd the oxygen ions. •
The increase in covalence with decreasing lithium concentration plays

portant role in determining the structural properties and the sequence of
e phases with x observed experimentally and predicted from first
iples for layered LixCo02• It is also of importance in understanding the
negligible concentration dependence of the activation barrier for lithium
sion.
3. The Role of Covalency in Determining Thermo

dynamic, Kinetic and Structural Properties of
LixCoOz
Delithiation of LixCo02 leads to phase transformations and lattice

meter changes that are closely linked to the accompanying changes in
70
ronic structure. LiCo02 is a semiconductor but within a delocalized
structure picture should become metallic upon lithium removal due to
reation of holes in the t2g valence bands. Experimentally, though the
val of lithium for x ranging between 0.93 and 1.0 is found to result in the
on of electronic-hole states that are localized in space and exhibit an
ated conduction mechanism? The actual transition from the semi
ucting state to a metallic state is obscured by a first order structural phase
74
formation between x=0.93 and 0.75. Below x=0.75, LixCo02 exhibits
3 3 18 27
lic properties. It has been argued • • that the semi-conducting to metal
tion induces the structural phase transformation between x=0.93 and
as the electronic transition is likely to break the convexity of the free
y in part as a result of additional configurational entropy from localized
18 27
on-hole states in the semi-conducting state. ·
Further delithiation of LixCo02 leads to a series of ordering and

ural phase transformations along with a dramatic contraction of the e
parameter perpendicular to the oxygen planes below .x=0.5. 75•76 Reimers
74
ahn were the first to observe an ordering phase transformation at .x=l/2
eby rows of lithium ions are alternated by rows of vacancies. Amatucci et
demonstrated the stability of fully delithiated Co02 but observed that
is stable in the 01 crystal structure which is characterized by an
77
B oxygen stacking sequence. Comprehensive first principles
2,
tigations 15" 17" 18 of phase stability in LixCo0 confirmed the
odynamic stability of the ordered phase at x=l/2 observed by Reimers
74
Dahn as well as the 01 phase observed by Amatucci et al.76
ermore, a staged phase in the vicinity of .x=0.15 was also predicted to be
18
odynamically stable from first principles. 15 • The predicted staged phase,
d Hl-3, 15 18· is characterized by a stacking sequence that is a hybrid
en 03 and 01. Alternating lithium planes in Hl-3 have an 03
onment (i.e. lithium sites are edge sharing with Co octahedra) and are
lly occupied by lithium while the remaining lithium planes have an 01
nment (lithium sites share faces with Co octahedra) and are vacant. The
ity of the H1-3 phase has recently been confirmed experimentally. 78•79
The progression in stability in layered LixCo02 from 03 to H1-3 to
ring delithiation can be attributed to the increased covalency of the Co
d with decreasing x. In the delithiated state, 03 Co02 is isomorphic to
, a relatively ionic compound, while 01 Co02 is isomorphic to Cdl2, a
covalent compound. The H1-3 phase at low lithium concentrations
es accommodation of the increased covalency of the Co-O bonds in the
ty of the 01 like stacking while simultaneously accommodating the
m ions within the alternating planes locally having the 03 stacking. H1-
be thought of as a local phase separation between a delithiated 01 block
Li-richer 03 block.
Part of the significant contraction of the lattice parameter c of layered
Chapter 2
o02 upon deintercalation can be attributed to a reduction in the number of

m ions which electrostatically pull the 0-Co-0 sheets together. Another
ficant factor contributing to this contraction, though, is the change with x
e electronic properties of the compound. The increased covalency of the
bonds with delithiation results in a contraction of the cobalt-oxygen
nces and, therefore, a reduction in width of the 0-Co-0 slabs as x is
ased. Furthermore, the increased covalency is accompanied by a
tion in the polarization of negative charge toward the oxygen ions.
e the electrostatic repulsion between negatively charged oxygen ions of
ent 0-Co-0 slabs diminishes as lithium is removed from the host.
Kinetic properties such as the migration barrier for Li diffusion are
ted by changes in the electronic properties with varying lithium
entration. First principles calculations31'36 predict that lithium diffusion in
ed LixCo02 occurs with a divacancy mechanism whereby the lowest
0- ------------:0
o::·:::::.....>:o-.-.-::::: : ------ :0
0 ...- - ·· · ··o
···o:::·:.-_·_·_·_· . ·_--:::o·········
- --- - . . .. - . ...... -.-- --
2.6. The dominant migration mechanism in layered LixCo0 2 is by means of a divacancy.

resence of a divacancy enables lithium to pass through an adjacent tetrahedral site as
ted above (filled circles are lithium, empty squares are lithium vacancies, large empty
are oxygen and small empty circles are cobalt).
y migration path between neighboring octahedral sites passes through an

ent tetrahedral site as illustrated in Figure 2.6. First principles
lations also predict an increase in the activation barrier upon delithiation.
The tetrahedral site along the lithium migration path shares a face with
jacent octahedral site containing positively charged cobalt. The resulting
rostatic repulsion energetically penalizes the tetrahedral site relative to
ctahedral site. This electrostatic repulsion is screened to some extent by
egative electronic charge on the oxygen ions. Nevertheless, the screening
nishes with decreasing lithium concentration because of the
mpanying reduction of the charge on oxygen. As the Li concentration is
ed, the accompanying reduction of screening, along with the contraction
e c axis, which results in a reduction of volume of the tetrahedral site,
to a dramatic increase of the activation barrier for lithium migration.
ELECTRONIC PROPERTIES IN LiNi02: THE 180°

INTERACTION
As is the case for LiCo02, LiNi02 is also stable in the 03 crystal

ure. Nevertheless, nickel has one extra valence electron which
atically alters the properties of layered LixNi02 as compared to those of
o02• LiCo02 is a semiconductor with occupied t 2g bands and unoccupied
nds. Upon delithiation, holes are introduced in the t2g bands. In LiNi02,
dditional valence electron resides in one of the eg bands. Within crystal
theory, the two eg states in an octahedron are degenerate, and single
ancy of the eg states can potentially lead to a Jahn-Teller distortion of
ctahedron. The distortion lifts the degeneracy of the eg levels lowering
nergy of the occupied level while raising the energy of the unoccupied
As described above, molecular orbital models characterize the eg states
ti-bonding levels which are typically more localized electron states.
these simple models we can anticipate distortions of Ni octahedra in
Chapter 2
02 and local charge ordering on the Ni ions upon delithiation. While

ctive distortion of the crystal due to Jahn-Teller activity has not been
ved experimentally,80 83
- evidence exists for the presence of
ordinated locally distorted octahedra.80'81 In fact, accurate first principles
lations within the generalized gradient approximation have predicted a
energetic preference for a collective Jahn-Teller distortion of LiNi0 2
al.35 Furthermore, at intermediate lithium concentrations charge ordering
een Nt 3 and Nt 4 ions accompanied by local Jahn-Teller distortions has
predicted from first principles. 34 •39 40
•
While accurate first principles calculations confirm the conclusions

emerge from crystal field and molecular orbital models, detailed first
iples calculations also predict an unexpected electronically induced long
attraction between lithium ions in different lithium planes.34 39• 40• A
4
matic stud/ of the energies and local distortions of Ni octahedra in
ures of LixNi02 with different lithium-vacancy arrangements showed a
rkable attraction between lithium ions at opposite ends of a distorted Ni
Q Oxygen
e Lithium
• Nickel
2.7. In LixNi0 2 a long range attraction between a pair of lithium ions of different lithium
is mediated by a Jahn-Teller distortion of a Ni+ 3 octahedron.
edron as illustrated in Figure 2.7. 180° Li-0-Ni-0-Li configurations

ole of Electronic Properties... 69
3
nd Ni+ ions along the elongated axis of the Jahn-Teller distorted
edra were consistently present in low energy configurations of LixNi02•
A similar long range interaction between four lithium ions was also
rned in which four lithium 2s orbitals hybridize with the eg molecular
34
al made of Ni d,2.y2 and oxygen p orbitals as illustrated in Figure 2.8.
3
ctahedron containing Nt and coordinated by four lithium ions as in
e 2.8 are characterized by a negative Jahn-Teller distortion in which the
Ni-0 bonds pointing to Li ions are elongated and two Ni-0 bonds
ing to Li vacancies are contracted.
3
The stability of the 180° Li-O-Ni+ -0-Li configurations around
34
rted octahedra can be rationalized as follows. The presence of lithium
at opposite corners of an octahedron lowers the energy of an occupied eg
due to hybridization between the Li-2s, 0-2p and aNi eg orbital (either
2.8. Attraction between four lithium ions over long distances in Li,Ni0 2 is mediated by
tive Jahn-Teller distortion of a Nt 3 octahedron (legend as in Figure 2.7).
2 or d3z2.,2). Although a Jahn-Teller distortion in any environment

dy leads to a lowering of the energy of an eg level, the presence of Li ions
posite corners of the oxygen octahedra further stabilizes this eg orbital
gh hybridization. Hence the occurrence of the long-range 180° Li-Li
tion and the Jahn-Teller distortions are coupled, one reinforcing the other.
Chapter 2
energy gain from a Jahn-Teller distortion is maximal when the occupied

bital is localized. This favors localization of electronic charge around the
ns, which upon deintercalation leads to charge ordered states consisting
ell defined Nt3 and Ni+4 ions. Due to the 180° Li-O-Nt3-0-Li attraction,
4
ithium-vacancy arrangements and the Nt 3 and Nt arrangements are
gly coupled.
The 180° coupling between Jahn-Teller active ions and the lithium
of different lithium planes is in part responsible for the dramatically
ent electrochemical behavior between LixNi02 and LixCo02• 39
actions between different lithium ions in a transition metal oxide are
ally short-ranged due to extensive screening by the oxygen ions. The
coupling in LixNi02, which is absent in LixCo02, leads to a long range
ling between lithium ions of different lithium planes that are more than 8
art. The added long range interactions favor different lithium-vacancy
ing reactions than is observed and predicted in LixCo02• In fact, first
iples calculations predict a very stable ordered phase at x=2/5 in LixNi0 2,
suspected type of ordering on a triangular lattice.39
In reality, synthesis of ideal LiNi02 has proven elusive, with actual
mens invariably having an excess of Ni ions.80 The excess Ni ions
py sites within the lithium layer. So while it is possible to make
ctions from first principles about ideal LixNi02, the presence of excess
the lithium planes in experiment is likely to alter these predictions to
extent. Excess Ni ions in the lithium planes will certainly alter the
ogy of the phase diagram of ideal LixNi02 and may even change the
of ordered phases that occur. Nevertheless, the existence of the 180°
range lithium attraction should persist in some form, even in the Ni
s samples.
LAYERED TO SPINEL TRANSFORMATION IN

LixMnOz
Most forms of LixCo02 and LixNi02 have the layered 03 crystal

ture. Lithium manganese oxide, however, is generally synthesized in the
l crystal structure with stoichiometry LixMn 204. The voltage profile of
pinel crystal structure is characterized by two plateaus, one around 4.2 V
<l and one around 3 V between .x=l and .x=2. The second plateau occurs
esult of a first order phase transformation. For many years, much effort
evoted to synthesizing layered LixMn02• 84' 85 Nevertheless, successfully
esized layered LiMn02 was found to transform to a phase resembling
l upon lithium removal.86- 89
. The Crystallography and Thermodynamics of the

Layered to Spinel Phase Transformation
The fact that the layered form of LiMn02 is observed to readily

orm to the spinel variant upon delithiation is not surprising from a
allographic point of view. Both the layered 03 crystal structure and the
l crystal structure have the same close-packed oxygen framework with
BCABC stacking sequence. The two phases only differ in the
gement of the lithium and transition metal ions over the interstitial sites
e oxygen skeleton. Transformation from layered to spinel therefore only
res a reordering of the transition metal and lithium ions.
In the layered phase, the transition metal ions occupy octahedral sites
ernating layers between close-packed oxygen planes, giving the phase a
bohedral symmetry. The lithium ions occupy the remaining octahedral
that also form alternating layers between close-packed oxygen planes. In
pinel crystal structure, the transition metal ions are distributed more
rmly among the octahedral interstitial sites: 3/4 of the metal ions reside
Chapter 2
lternating layers between close-packed oxygen planes, while the

ining 1/4 of the metal ions reside in the other alternating planes. This
cular metal ordering produces a transition metal oxide host structure with
symmetry. While the lithium ions can occupy the remaining vacant
edral sites in the host, energetically favorable tetrahedral sites are also
able; the metal ordering in spinel results in a set of tetrahedral sites that
ot share faces with metal-oxygen octahedra. Upon deintercalation of the
ed 03 crystal structure, channels within the oxygen framework open up
nable transition metal ions to migrate to the vacated lithium layers. For
yered phase to transform to spinel, exactly one in four transition metal
must migrate to the lithium layer. This transformation thus can occur
ut reconstruction of the oxygen skeletal structure.
First principles calculations have exposed that for many transition
oxides LixM02 (M=Ti, V, Mn, Fe, Co, Ni) 16' 27 the spinel crystal
ture is energetically more stable than the layered crystal structure at
. Figure 2.9, which illustrates the calculated free energy curves of 03
pinel LixCo02 as a function of lithium concentration/0 clearly shows the
odynamic stability of spinel around x=l/2 over the 03 phase. This
ve stability means that for layered lithium transition metal oxides a
odynamic driving force exists for transformation to the spinel crystal
ure upon deintercalation. Nevertheless, only LixMn02 is observed to
go this transformation at room temperature.86- 89 Layered LixCo02 for
ple, maintains its layered structure upon continued cycling. The high
lity of Mn ions which makes the transformation of LixMn02 to a spinel
ure at room temperature possible, has its origin in the particular
onic structure of the Mn ion in octahedral and tetrahedral environments.
-> Q)
E........
100
0
...
>.
0 >
.... . -100
Q)
••
c
Q) -200 ••••
Q) ••••
Q )
.... .
LL
-300 ••••
••••••••••
-400
0 0.2 0.4 0.6 0.8 1

Li concentration
2.9. Free energy of layered 03 (solid line) and spinel (dashed line) LixCo02 as a
on of lithium concentration.
2. A Two Stage Transformation Mechanism
First principles calculations have shown that the layered to spinel

transformation in LixMn0 2 occurs in two stages?3 This is consistent
experimental observations.90- 93 The kinetics of the layered to spinel phase
formation relies crucially on the first step which is driven by a unique
e-disproportionation reaction predicted from first principles to occur
in LixMn02 and not in the Co and Ni layered compounds.33
A first step in the transformation from layered LixM02 to the spinel
nt is the migration of transition metal ions to adjacent tetrahedral sites.
e tetrahedral sites share faces with an oxygen-octahedra in the transition
layer and three octahedra in the lithium layer (Figure 2.10).
Chapter 2
Li layer
Mn layer
lli, 0 Oxygen
'
' o Lithium s:
: : : : : : e Manganese
'
'
2.10. The layered 03 host and spinel have the same oxygen skeletal structure.
ormation from layered to spinel can occur by a redistribution of the transition metal ions
this structure. The first step of such a transformation involves migration of 114 of the
ons to tetrahedral sites (thick dashed lines) in the lithium layer.
The migrating transition metal ion must then migrate further to an

edral site within the lithium layer. If one in four transition metal ions
te to the lithium layer in a coordinated manner (see Figure 2.11)
mpanied by lithium rearrangement to the non-face sharing tetrahedral
that result from the metal rearrangement, the transformation would yield
ect free spinel phase. Due to the thermal nature of such a reordering
ormation, however, we can expect the transition metal rearrangement to
in a more random way, creating a disordered spinel phase. Over time,
h, the intermediate disordered phase should gradually evolve to the
ct spinel ordering as a result of continued ion rearrangement within the
en skeletal structure.
2.11. Projection of a metal layer (transition metals are the filled circles) and the oxygen
oxygen ions are the empty circles) above it in the 03 crystal structure. Transformation to
involves migration of 114 of the metal ions to the lithium layer. This leaves a vacant
site (empty squares) within the metal layer. The migrated metal ions must migrate to a
site above a triangle of metal ions (shaded circle). In order for this rearrangement to
o spinel, the arrangement of vacancies within each metal layer must be ordered as
ated in the above figure.
.1. Stage I of the Layered to Spinel Transformation in LixMn02
A tetrahedral site as in Figure 2.10 is only a viable path for transition

migration to the lithium layer if the three adjacent octahedral sites in the m
layer are vacant. Hence a precondition for the layered to spinel
ormation is some degree of lithium deintercalation such that the
taneous existence of three lithium vacancies around a tetrahedral site has
sonable probability of occurring. As the lithium concentration is reduced,
oncentration of such tri-vacancies increases. Typically the transition
s considered for cathode materials with stoichiometry LixM0 2 prefer
edral sites over tetrahedral sites. Migrating to a tetrahedral site is then
3
mpanied by an increase in the energy of the crystae and therefore
odynamically unfavorable. But first principles calculations predict that
Chapter 2
ons with a valence state of +2 behave differently, actually preferring

hedral sites over octahedral sites.
The formal valence state of transition metal ions in LiM02 is +3
h on average increases towards +4 as lithium is removed. Nevertheless,
33
e Li concentration is reduced, first principles calculations predict that
ions can form through a charge disproportionation reaction involving
3
ron transfer from a neighboring Mn+ ion according to
+3 +2 =4
2 Mnoct --7 Mntet + Mnoct
)
The above reaction equation should be interpreted as follows:

3
ation of a Mn+ ion from an octahedral site in the metal layer to an
ent tetrahedral site surrounded by lithium vacancies becomes
3
etically favorable provided that it is adjacent to another Mn+ ion from
2
h it can borrow an electron. The result of the reaction is a Mn+ ion in a
4
edral site within the lithium layer and a Mn+ ion in the metal layer. First
2
iples calculations predict that the valence electrons of the Mn+ ion in the
3 2
edral site have parallel spin? A high spin Mn+ ion has a half filled d
that chemically behaves similar to a closed shell ion. Hybridization with
oordinating oxygen ions then occurs predominantly with the empty 4s
3
p levels, which in a tetrahedral environment, hybridize into sp orbitals.
he ability of the Mn ions to form a closed d shell electron configuration
makes migration to adjacent tetrahedral sites energetically favorable. This
e electronic property of Mn is absent in other transition metal ions in
02 (M=Co or Ni) and is crucial in precipitating the first step of the
ed to spinel phase transformation. In the other layered transition metal
s, migration of the transition metal to a tetrahedral site is met with an y
33
barrier; in the Mn compound, it leads to a reduction in energy, and
will occur spontaneously as long as sufficient vacancies exist in the
m layer and sufficient Mn+3 ions are available in the metal layer to feed
harge-disproportionation reaction of Eq. (2.17).
The availability of vacancies in the lithium layer and Mn+3 ions in the
layer are requirements that have an opposite dependence on lithium
entration. Decreasing the lithium concentration increases the number of
m vacancies but it decreases the number of Mn+3 ions. A maximum in
ctahedral to tetrahedral migration rate can therefore be expected at
mediate lithium concentration, the concentration which also happens to
ide with the maximal thermodynamic driving force for the layered to
l transformation.
.2. Stage II of the Layered to Spinel Transformation in LixMn02
In layered LixMn02, the migration of Mn to tetrahedral sites between

etal and lithium layer is predicted to occur spontaneously (without
y barrier) at intermediate lithium concentration.33 In fact, at .x=l/2,
ient Mn+3 ions exist for 1/4 of the Mn ions to migrate from the metal
to tetrahedral sites by means of the charge-disproportionation reaction of
2.17). The migration of Mn ions from octahedral sites in the metal layer
rahedral sites leads to other tetrahedral sites in the lithium layers to no
r share a face with Mn octahedra. These tetrahedral sites then become
etically favorable for lithium occupation. Spontaneous Mn migration to
edral sites is predicted to be accompanied by lithium migration to
33
edral sites. While migration of Mn ions to tetrahedral sites at .x=l/2
s spontaneously, this phase with 1/4 tetrahedral Mn+2 and some
edral lithium is still intermediate in energy between the 03 phase and
erfectly ordered spinel phase (all lithium in tetrahedral sites, cubic
ing of Mn ions over octahedral sites).33 We refer to this intermediate
at .x=l/2 as the "splayered" phase and emphasize that it has 1/4 of the
ons in tetrahedral sites distributed uniformly but in a more or less
dered arrangement. The existence of a partially transformed layered
Chapter 2
90 93
e with tetrahedral Mn has been observed experimentally. - Further
formation of the splayered phase to spinel occurs by thermally activated
hops within the lithium layers and therefore proceeds slower than the
ed to splayered transformation, which has no barrier.
In a battery, charging rarely stops at .x=l/2 but continues to lower
entrations. Further lithium removal from the spontaneously formed
2
ered phase would require oxidation of Mn+ ions, destabilizing the Mn
in the tetrahedral sites. Lithium removal below .x=l/2 therefore leads to
migration from tetrahedral sites back to octahedral sites. Nevertheless, a
on in a tetrahedral site can migrate to four neighboring octahedral sites,
which resides in the metal layer and three which reside in the lithium
. The higher availability of octahedral sites in the lithium layer leads to a
igration of tetrahedral Mn ions to octahedral sites in the lithium layer g
further deintercalation. Repeated cycling will enhance the migration of
ons to the lithium layers and this is observed experimentally where
ng of the material leads to an increased spinel formation.
3. Suppressing the Layered to Spinel Transformation
The electronic origin of the layered to spinel phase transformation in

n02 can be exploited to suppress this transformation by appropriately
g the compound. The crucial step in the layered to spinel phase
formation is the charge-disproportionation reaction fed by the presence
3
n+ ions. Doping of the compound to prevent the layered to spinel
3
formation should hence deprive the host of Mn+ ions. This can be done
elements that are more electronegative than Mn, therefore oxidizing
4
to Mn+ , even in the fully lithiated state. This has been successfully
94 0 0 2
ved by Amrnundsen et al. who found that layered LixMn .4Cr .4Li
0 .2 0
s without transformation to spinel. Substituting half the Mn with Ni has

95 96
been shown to suppress the layered to spinel phase transformation. • In
ed LixMn0. 5Nio. 502, Ni is predicted to have a +2 valence while Mn has a
8
alence? Charging of the compound leads to the oxidation of Ni from the
ate to the +4 state. The absence of Mn+ 3 ions at all lithium concentrations
nts the detrimental layered to spinel phase transformation in this
ound.
CONCLUSIONS
We have reviewed theoretical models and tools that enable us to

rstand and predict the electronic properties of lithium transition metal
s. We have focussed in particular on a detailed description of density
ional theory (DFf), currently the most reliable and accurate approach to
ing the electronic properties of complex oxides from first principles. We
also reviewed results and insights derived from DFf studies of three
rtant lithium transition metal oxides, namely LixCo02, LixNi02 and
n02• In our discussion of these three compounds, we have focused on
rtant electronic properties exposed by DFf calculations which have a
ular bearing on the electrochemical properties of these compounds. In
o02, DFf calculations have shown the importance of covalent bonding
en oxygen and cobalt and the role of increased covalency with
asing x in affecting the thermodynamic and structural properties of this
ound. In LixNi02, a DFf study has predicted a long range attraction
en lithium ions of different lithium planes. This long-range attraction is
ated by a Jahn-Teller distorted oxygen octahedra containing a Nt 3 ion.
lithium-lithium attraction, termed the 180° interaction, is in part
nsible for the very different electrochemical properties of LixNi02 as
ared to those of LixCo02• For LiMn02, a DFf investigation has led to
ediction of a remarkable charge-disproportionation reaction between Mn
which leads to spontaneous migration of Mn ions to tetrahedral sites
n the lithium layer of layered LixMn02• It is this charge
oportionation reaction, identified from first principles, which is
nsible for the undesirable layered to spinel phase transformation that has
Chapter 2
nted the successful implementation of layered LixMn02 in lithium

ries.
KNOWLEDGMENTS
We are grateful to Elena Arroyo, John Reed, Chris Marianetti, Claude

as and Bing Joe Hwang for insightful discussions. We thank the
rtment of Energy (award no. DE-FG0296ER45571) andthe National
ce Foundation (award no. DMR 98-08941 and NPACI supercomputer
ort).
ERENCES
M. Thackeray, J.O. Thomas, M.S. Whittingham, MRS Bulletin 25 (2000) 39.
. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Electrochem. Soc. 144 (1997)
8.
Menetrier, I. Saadoune, S. Levasseur, C. Delmas, J. Mater. Chern. 9 (1999) 1135.
reedan, N.P. Raju, A.S. Will, C. Morin, S.M. Shaw, J.N. Reimers, Chern. Mater. 10
98) 3058.
odriguez-Carvajal, G Rousse, C. Masquelier, M. Hervieu, Phys. Rev. Lett. 81 (1998)
0.
ousse, C Masquelier, J. Rodriguez-Carvajal, M. Hervieu, Electrochem. Solid -State Lett.
1999) 1.
eder, M.K. Aydinol, A.F. Kohan, Camp. Mat. Sci. 8 (1997) 161.
K. Aydinol, A.F. Kohan, G Ceder, G, K. Cho, J. Joannopoulos, Phys. Rev. B 56
7) 1354.
. Aydinol, G Ceder, J. Electrochem. Soc. 144 (1997) 3832.
Deiss, A. Wokaun, J.L. Barras, C. Daul, P. Dufek, J. Electrochem. Soc. 144 (1997)
877.
Benedek, M.M. Thackeray, L.H. Yang, Phys. Rev. B 56 (1997) 10707.
Ceder, A.F. Kohan, M.K. Aydinol, P.D. Tepesch, A. Vander Ven, J. Amer. Cer. Soc. 81
1998) 517.
Ceder, Y.M. Chiang, D.R. Sadoway, M.K. Aydinol, Y.l. Jang, B. Huang, Nature 392
1998) 694.
. Wolverton, A. Zunger, Phys. Rev. B 57 (1998) 2242.
. Vander Yen, M.K. Aydinol, G Ceder, J. Electrochern. Soc. 145 (1998) 2149.
. Wolverton, A. Zunger, J. Electrochern. Soc. 145 (1998) 2424.
. Wolverton, A. Zunger, Phys. Rev. Lett. 81 (1998) 606.
. Vander Yen, M.K. Aydinol, G Ceder, G Kresse, J. Hafner, Phys. Rev. B 58 (1998)
2975.
Benco, J.L. Barras, M. Atanasov, C.A. Daul, E. Deiss, Solid State lonics 112 (1998) 225.
. Vander Yen, G Ceder, Phys. Rev. B 59 (1999) 742.
Benco, J. L. Barras, C. A. Daul, E. Deiss, Inorg. Chern. 38 (1999) 20.
.S. Mishra, G Ceder, Phys. Rev. B 59 (1999) 6120.
Benco, J. L. Barras, M. Atanasov, C. Daul, E. Deiss, J. Solid State Chern. 145 (1999)
503.
S. Braithwaite, C. R. A. Catlow, J.D. Gale, J. H. Harding, Chern. Mat. 11 (1999) 1990.
Benedek, M. M. Thackeray, L. H. Yang, Phys. Rev. B 60 (1999) 6335.
Ceder, M.S. Mishra, Electrochern. Solid-State Lett. 2 (1999) 550.
Ceder, A. Vander Yen, Electrochirn. Acta, 45 (1999) 131.
Buta, D. Morgan, A. Vander Yen, M. K. Aydinol, G Ceder, 1 Electrochern. Soc. 146
1999) 4335.
S. Braithwaite, C. R. A. Catlow, J.D. Gale, J. H. Harding, P. E. Ngoepe, J. Mat. Chern.
0 (2000) 239.
S. Braithwaite, C. R. A. Catlow, J. H. Harding, J.D. Gale, Phys. Chern. Chern. Phys. 2
2000) 3841.
Vander Yen, G Ceder, Electrochern. Solid-State Lett. 3 (2000) 301.
Vander Yen, C. Marianetti, D. Morgan, G Ceder, Solid State lonics 135 (2000) 21.
Reed, G Ceder, A. Vander Yen, Electrochern. Solid-State Lett. 4 (2001) 78.
. E. Arroyo y de Dompablo, C. Marianetti, A. Van der Yen, G Ceder, Phys. Rev. B 63
2001) 144107.
Marianetti, D. Morgan, G Ceder, Phys. Rev. B 63 (2001) 224304.

Chapter 2
Vander Yen, G Ceder, M. Asta, P. D. Tepesch, Phys. Rev. B 64 184307 (2001).
E. Grechnev, R. Ahuja, B. Johansson, 0. Eriksson, Phys. Rev. B 65 (2002) 174408.
Reed, G Ceder, Electrochem. Solid-State Lett. 5 (2002) A145.
. E. Arroyo y de Dompablo, A. Vander Yen, G Ceder, Phys. Rev. B 66 (2002) 64112.
. E. Arroyo y de Dompablo, G Ceder, Chern. Mater. 15 (2003) 63.
Hohenberg, W. Kohn, Phys. Rev. B 136 (1964) 864.
G Parr, W. Yang, Density Functional Theory of Atoms and Molecules, Oxford
niversity Press (1989).
0. Jones, 0. Gunnarsson, Rev. Mod. Phys. 61 (1989) 689.
M. Sanchez, F. Ducastelle, D. Gratias, Physica 128A (1984) 334.
de Fontaine, Solid State Phys. 47 (1994) 33.
. C. Day, J. Selbin, Theoretical Inorganic Chemistry, Reinhold Pub. (1962).
J. Ballhausen, Introduction to Ligand Field Theory, McGraw-Hill (1962).
M. Dunn, D. S. McClure, R. G Pearson, Crystal Field Theory, Harper and Row (1965).
H. Hall, Group Theory and Symmetry in Chemistry, McGraw-Hill (1969).
W. Ashcroft, N.D. Merrnin, Solid State Physics, Saunders College Publishing (1976).
Madelung, Introduction to Solid State Theory, Springer Verlag (1978).
Levy, Proc. Natl. Acad. Sci. 76 (1979) 6062.
Kohn, L. J. Sham, Phys. Rev. 140 (1965) A1133.
I. Anisimov, F. Aryasetiawan, A. I. Lichtenstein, J. Phys.Condens. Matter 9 (1997) 767.
. Perdew, A. Zunger, Phys. Rev. B 23 (1981) 5048.
Szotek, W. M. Temmerman, H. Winter, Phys. Rev. B 47 (1993) 4029.
M. Rieger, P. Yogi, Phys. Rev. B 52 (1995) 16567.
Goedecker, C. J. Umriga, Phys. Rev. A 55 (1997) 1765.
J. Singh, Planewaves, Pseudopotentials and the IAPW Method, Kluwer Academic Pub.
1994).
C. Phillips, L. Kleinman, Phys. Rev. 116 (1959) 287.
C. Payne, M.P. Teter, D. C. Allan, T. A. Arias, J.D. Joannopoulos, Rev. Mod. Phys. 64
1992) 1045.
Wolverton, private communication (1997).
R. Hamann, M. Schluter, C. Chiang, Phys. Rev. Lett. 43 (1979) 1494.

B. Bachelet, D. R. Hamann, M. Schluter, Phys. Rev. B 26 (1982) 4199.
Kleinman, D. M. Bylander, Phys. Rev. Lett. 48 (1982) 1425.
M. Rappe, K. M. Rabe, E. Kaxiras, J.D. Joannopoulos, Phys. Rev. B 41 (1990) 1227.
Troullier, J. L. Martins, Phys. Rev. B 43 (1991) 1993.
Vanderbilt, Phys. Rev. B 41 (1990) 7892.
. T. Czyzyk, R. Potze, GA. Sawatzky, Phys. Rev. B 46 (1992) 3729.
van Elp, J. L. Wieland, H. Eskes, P. Kuiper, GA. Sawatzky, F. M. F. de Groot, T. S.
urner, Phys. Rev. B 44 (1991) 6090.
A. Montara, M. Abbate, J. M. Rosolen, Electrochem. Solid-State Lett. 3 (2000) 410.
Graetz, A. Hightower, C. C. Ahn, R. Yazami, P. Rez, B. Fultz, J. Phys. Chem. B 106
2002) 1286.
.-S. Yoon, K.-B. Kim, M.-G Kim, M.-K. Lee, H.-J. Shin, J.-M. Lee, J.-S. Lee, C.-H. Yo,
. Phys. Chem. B 106 (2002) 2526.
N. Reimers, J. R. Dahn, J. Electrochem. Soc. 139 (1992) 2091.
Ohzuku, A. Ueda, J. Electrochem. Soc. 141 (1994) 2972.
G Amatucci, J. M. Tarascon, L. C. Klein, J. Electrochem. Soc. 143 (1996) 1114.
Delmas, C. Fouassier, P. Hagenmuller, Physica B 99 (1980) 81.
H. Chen, Z. H. Lu, J. R. Dahn, J. Electrochem. Soc. 149 (2002) 1604.
Croguennec, S. Levasseur, M. Morcrette, J. M. Tarascon, C. Delmas, Materials
esearch Society (MRS) Fall meeting, Boston (MA), Dec. 2-6, 2002.
Rougier, C. Delmas, A. V. Chadwick, Solid State Commun. 94 (1995) 123.
Delmas, M. Menetrier, L. Croguennec, S. Levasseur, J.P. Peres, C. Pouillere, G Prado,
. Fournes, F. Weill, Int. J. lnorg. Mater. 1 (1999) 11.
N. Mansour, X. Q. Yang, X. Sun, J. McBreen, L. Croguennec, C. Delmas, J.
lectrochem. Soc. 174 (2000) 2104.
. Peres, A. Demourgues, C. Delmas, Solid State lonics 111 (1998) 135.
R. Armstrong, P. G Bruce, Nature 381 (1996) 499.
Capitaine, P. Gravereau, C. Delmas, Solid State Ionics 89 (1996) 197.
Blyr, C. Sigala, G Amatucci, D. Guyomard, Y Chabre, J. Electrochem. Soc. 145
998) 194.
-M. Chiang, D. R. Sadoway, Y. I. Jang, B. Huang, H. Wang, Electrochem. Solid-State

Chapter 2
ett. 2 (1999) 107.
Shao-Horn, S. A. Hackney, A. R. Armstrong, P. G Bruce, R. Gitsendanner, C. S.
ohnson, M. M. Thackeray, J. Electrochern. Soc. 146 (1999) 2404.
Vitins, K. West, J. Electrochern. Soc. 144 (1997) 2587.
Bruce, A. Armstrong, R. Gitzendanner, J. Mater. Chern. 9 (1998) 193.
Ammundsen, J. Desilvestro, T. Groutso, D. Hassell, J. B. Metson, E. Regan, R. Steiner,
. J. Pickering, J. Electrochern. Soc. 147 (2000) 4078.
Wang, Y. I. Jang, Y. -M. Chiang, Electrochern. Solid-State Lett. 2 (1999) 10.
J. Hwang, H. S. Park, J. H. Choy, Chern. Mater. 12 (2000) 1818.
Ammundsen, J. Paulsen, I. Davidson, R. S. Liu, C. H. Shen, J. M. Chen, L. Y. Jang, J. F.
ee, J. Electrochern. Soc. 149 (2002) A431.
S. Kim, C. S. Johnson, M. M. Thackeray, Electrochern. Cornrn. 4 (2002) 205.
H. Kang, J. Kim, M. E. Stoll, D. Abraham, Y. K. Sun, K. Amine, J. Power Sources 112
2002) 41.
pter 3
SYNTHESIS OF BATTERY MATERIALS
M. S. Whittingham
Chemistry and Materials Science

State University of New York at Binghamton, Binghamton, NY 13902, U.S.A.
INTRODUCTION
Synthesis is critical not only to the generation of new materials for
ation of new structures and fundamental properties, but also for the
ion of materials with the optimum electrochemical behavior for
ercial devices. The technique used is quite often very different for the
two cases. The initial synthesis of a material should provide it in a
enough state that its intrinsic behavior and properties can be
ined; it should for example be possible to accurately determine
er its structure is cubic spinel, layered or some other form. There is
tly much confusion in the literature because the synthesized material
sufficiently well-characterized. Once well characterized, the material
to be synthesized in a form suitable to be used in a battery, for
le it may be doped or coated to enhance the conductivity, its particle
nd morphology will be optimized for maximum reactivity but
um corrosivity and side-reactions.
Many battery materials are metastable phases, and therefore non
nal synthesis methods must be devised to allow kinetics to over-ride
dynamics. Hence, many soft chemistry techniques have come to the
uch as hydrothermal, ion-exchange, intercalation, etc. In addition, to
ze the formation of the desired material frequently the reactants are
mixed prior to final reaction by for example sol-gel formation or co
tation as hydroxides.
A number of critical synthetic parameters will be discussed in this
r. The temperature of formation of the electrode material determines
ect structure, for example in all the layered structures strict ordering
cations is desired. Thus, in TiS2 no titanium must reside in the lithium r
diffusion is limited. This restricts the final firing temperature to a
Chapter 3
um of 600°C, otherwise there is significant titanium occupancy in the

er Waals layer. A similar behavior is observed in the layered metal
, such as LiNi02, LiNi 1_yMny02 and LiMn02. In the nickel oxides it is
l to find about 7% nickel occupancy on the lithium sites. In the
tion of layered LixMn02, great care must be taken to avoid
ratures over 100°C otherwise manganese migrates converting the
al to the spinel phase. For the mixed metal oxides, the temperature of
sis has a marked effect on the reversible capacity of the electrode. n
materials such as LiFeP04, the temperature of formation is key; too
temperature and some of the iron migrates to the lithium sites.
The synthesis and manufacture of battery materials will be
ssed by looking at a series of examples of materials that have moved
curiosity to research to development. Amongst these the concept of
lation reactions from the 1970's will be considered, high temperature
ns, followed by hydrothermal and ion-exchange reactions, then a
of synthetic methods for conductively coating the lithium iron
hate class of material. Finally, the synthesis of anode materials will be
sed.
Although cost is not critical in the research phase of a project, it is
l for the cost-effectiveness of batteries, that not only the electrode
als be themselves low cost, but that the process costs for synthesizing
12
terials also be low in cost. •
THE SYNTHESIS AND MANUFACTURE OF

TITANIUM DISULFIDE
When a battery is being commercialized, the synthesis process used in
oratory often cannot be used because of the processing costs or cost of
rting materials. As an example, titanium disulfide will be considered
in the laboratory bulk titanium was used to provide electronic grade
m disulfide, and sponge titanium provided research battery grade
al. The latter could be produced a pound at a time in silica tubes, had a
2 2
e area of 5 m /g, and allowed current densities of 10 mA/cm to be
ed. However, both involved reaction with sulfur, which took from
for the sponge to days for the bulk metal. Such material would cost
than $100/lb. An inspection of the commercial process, shown in
3.1, for sponge manufacture revealed that the precursor was titanium
loride, a liquid at room temperature. This tetrachloride is available in
e quantities as it is used in the manufacture of titanium dioxide paint
nt. Therefore, a manufacturing process was devised by two european
nies that involved the formation of stoichiometric TiS2 by deposition
is of Battery Materials 87
he gas phase reaction of TiC14 with H2S. This produced a sulfide with
phology with many plates growing in three dimensions from a single l
point, which shows excellent electrochemical behavior.
The stoichiometry and ordering of the titanium is critical to the
chemical behavior of TiS2• Stoichiometric and ordered TiS2 has been
3
to exist if the temperature is kept below 600°C, and to have metallic
4
ctivity. The titanium disorder can be readily measured by attempting
ercalate weakly bonding species such as ammonia or pyridine. In
e a slight excess, :::;1%, is beneficial in that it reduces the
iveness of the sulfur without significantly impacting the cell potential
lithium diffusion coefficient. It is preferable to add this extra metal to
tial reaction medium.
ILMENITE
(+rutile)
CHLORINATION
850-950°C
PETROLEUM COKE
I
2
ENT ILE) H.,S 500-600°C
I
KROLL PROCESS
Mg metal
I
TITANIUM
SPONGE
s, l;jQQ-l;jl;jQOQ TiS2
24 Hours
I
ARC FURNACE
I
TITANIUM s gQQOQ TiS2
INGOT 1 Week
.1. Synthesis and manufacture of titanium and titanium disulfide.

Chapter3
SYNTHESIS OF ALKALI ION INTERCALATES

BY INSERTION INTO THE HOST LATTICE
(AND THE REVERSE)
Although the heavier alkali intercalates of the transition metal oxides
56
halcogenides had been extensively studied, ' the lack of a suitable
ng agent delayed the formation and characterization of the pure
compounds until the early 1970's. Although lithium had been
78
d by dissolution in liquid ammonia, • this also resulted in the insertion
ammonia, resulting in degradation of the host lattice as the sulfide or
9
layers must all shift relative to one another as indicated in Figure 3.2.
10 11
trast when lithium is inserted by electro-intercalation, ' from butyl
12 14
in hexane, - or from mercury amalgams no solvent is incorporated
e lattice simply expands. Just as in the case of ammonia solutions,
can also be incorporated from aqueous solutions either chemically or
chemically, 10' 11' 15' 16 then the water must be removed. The many
tic procedures possible for the layered dichalcogenides are shown
9
atically in Figure 3.3.
Li
...
n-Butyllithium
Electrointercalation
0
Ti Li
•
.2. 110 projections of the structural changes occurring during the lithiation of disulfide
in the presence and absence of a co-intercalating solvent such as ammonia.9
Many materials of interest do not exist as the thermodynamically

phase in the desired structure, and so must be prepared at low
ratures where kinetics control the path of the reaction rather than
odynamics. For example, the stable phase of titanium disulfide is the
d structure for LixTiS2 for 0$x:S1 and batteries can be built in either the
d state Li/ffiS2 or in the discharged state C/LiTiS2• 17 In contrast,
gh the layered sulfides LiVS2 and LiCrS2 were well-known their
free analogs were not, and so researchers at Bell Labs synthesized
18
y de-intercalation of the fully lithiated compounds. They developed
19
range of chemical lithiating and delithiating agents. As noted above,
ered hexagonal close-packed form of TiS2 is the thermodynamically
20 22
phase; however a number of researchers - successfully synthesized
bic form by de-intercalation of, for example, the spinel CuThS4. This
structure can also be reversibly intercalated with lithium, although the
on coefficient is not as high as in the layered form.
The more ionic transition metal dioxides are not readily formed as
structures in the absence of cations between the M02 layers, other
being more stable such as in Ti02 where a range of phases such as
anatase and brookite are known but layered Ti0 2 is not. Goodenough
o-workers, whilst investigating the magnetic behavior of LiCo02
ized that it might be possible to also form a layered compound,
23
ous to the sulfides by lithium removal. They successfully formed a
d LixCo02 by lithium de-intercalation and showed that this material
reversibly incorporate lithium over at least a range of 8x of around 0.6.
ed to the first commercially successful lithium secondary battery by
4
The lithium cobalt oxides can be readily prepared by the high
ature firing of materials containing lithium and cobalt in the
riate ratio in an oxidizing environment. Amatucci et al. 25 more
y showed that all the lithium can be removed giving Co02, which has
me 1T structure as TiS 2•
A number of the layered dichalcogenides and oxides can incorporate a
lithium ion forming the phase LizMX2. This was first observed for
14 26
um diselenide, ' and can be accomplished either electrochemically
27
mically, for example by using butyl-lithium. LizNi02 was formed
lectrochemically and chemically using lithium benzophenone in
drofuran; it's structure switches from the 3R-LiNi02 phase to one
al to those of LizTiS2 and LhVSe2, where the lithium ions sit in all the
dral sites between the Ni02 sheets forming a 1T structure. In a similar
r LizMn0.5Ni 0.50z has also been synthesized in electrochemical cells,
th lithium ions can be cycled when part of the Mn is substituted by
28
m.
Another innovative synthetic approach was developed by Haering29
owed that 2H-MoS2, where the molybdenum is in trigonal prismatic
Chapter 3
Oxidation-MX 2 - Reduction
r .JC.-1---n- Butyl, lithium -----t,.r.....;:,.-1

1----A +Heat Ax MX 2
In NH3 i 00 H g -. --.-....--'
BD in Y
t
Polar solvent
z
A, , Y 11 _z., M X 2
.3. Schematic of the reactions of the dichalcogenides.9
nation, could be induced to change structure to the 1T TiS2 form by

oration and then removal of lithium. This structural form showed
improved cyclability over the 2H form which is poorly reversible, and
e basis of the MoliEnergy MoS 2 lithium battery. 30
HIGH TEMPERATURE SYNTHESIS OF LAYERED

OXIDES
The layered compound LiCo02 is made by the high temperature

23
n in air of the carbonates at 900°C. A number of other layered
metal dioxides, such as LiNi02 can also be formed by high
ature synthesis, but in this case there is significant nickel disorder so
31
e lithium layer always contains some nickel, seven to ten % being
ypical. In contrast there is no measurable cobalt in the lithium layer
LiCo02 is formed at the optimum temperature. However, there is the
ility of forming spinel-like (some Co in the Li layer) or the spinel
32
sometimes known as LT-LiCo02, 33 when lower temperatures such
°C are used in the synthesis. It is not easy to distinguish between the
d and spinel forms of these layered oxides, as the X-ray diffraction s
look almost identical at a first look. A clue to spinel formation is the io,
which is around 4.90 for the spinel, while it is closer to 5.00 for the
ohedral layer structure; both have cubic close-packed oxygen lattices
hich the ideal cia ratio is 1.633 (= 4.90/3). A second clue to pure
d formation, which results from the non-ideal close packing in the
d structure, is the existence of a doublet in the X-ray powder pattern at
66° 28 associated with the 108 and 110 reflections; the spinel has a
reflection, the 440, in this region for CuKa radiation. The LT-LixCo1_
phases, also formed at 400°C, were found to have spinel-like
res where there is partial occupancy of the lithium sites by the
34
ion metal ions.
The synthesis of LiCo02 requires firing at temperatures from 850 -
C for one to two days. However, such high temperatures may not be
ary. For example, ion exchange between CoOOH and excess LiOH
in pure LiCo02 after firing to 280°C to remove residual carbonate and
35
ide. This ion-exchange was carried out in the solid state at 100°C
st a trace of water added as a transport agent. HT-LiCo02 was formed
denced by the doublet at 66° 28, the cia ratio of 5.00 and the
chemical fingerprint. The success of the synthesis in this case can be
ated with the structure of the starting reagent, CoOOH, which is
d and does not change under the low-temperature ion-exchange
ions. The electrochemical behavior was quite good, but the capacity
ss than that of the high temperature formed LiCo02; it was suggested
ere is some cobalt disorder, some of the lithium sites being occupied.
35
2 was similarly synthesized but no electrochemical or nickel ordering
ere presented, but the rather broad X-ray lines suggest significant
disorder. LiCo02 formed hydrothermally at 230°C from cobalt nitrate,
hydroxide and hydrogen peroxide, had rather poor cycling
teristics, the capacity dropping by more than 70% over 20 cycles at a
36
e.
37
The spinel LiNiz04 is also known, being formed when Li0.5Ni02
by electrochemical removal of lithium from LiNi02 is annealed at
; this structural conversion begins at 150°C just as observed for
38
d Li05Mn0z. The coalescence of the 108 and 110 X-ray reflections is
observed. This spinel is only stable until 300°C, when it decomposes
mixture of products.37 As noted above, neither Ni02 nor Co02 has
Chapter 3
ynthesized directly because of their thermodynamic instability. In the f

Ni02, the oxygen partial pressure is well over one atmosphere at so
it is inherently unsafe even under oxidizing conditions. In an
t to reduce the possibility of explosive interactions with the
lyte, additional elements can be used to partially replace the nickel.
ample, addition of a trivalent ion such as boron or aluminum prevents
mplete removal of all the lithium thus limiting the overall average
ion state of the nickel. Alternatively, part of the nickel may be
ed by for example cobalt as in LiNi1_yCoy02, which forms a continuous
olution. A combination of these two approaches has been extensively
9
d, as in LiNi0.85Coo.IAlo.osOl 40
' and LiNio.soCoo.1sAlo.os02.41 The LiNii
42
phases have been prepared '43 by the direct reaction of LhC03, NiO
o304. A finely ground mixture of the reagents is first fired at 500°C for
hours, followed by 48 hours in oxygen at 800-1000°C. A solid solution
d over the entire range of y from 0 to 1.
Manganese has also been substituted in the cobalt oxide, LiCo1 _
2 for O<y 0.2 using a combinatorial approach starting from an
us solution of lithium hydroxide, cobalt acetate and manganese acetate,
44
ed by firing at 730°C; the lattice parameter perpendicular to the
was found to expand with increasing manganese content. At higher
nese contents, second phases begin to be observed.
45
Spahr et al. showed that the replacement of some of the nickel in
2 by manganese also served the same purpose as cobalt in improving
clability and capacity of the nickel oxide electrode. They found that
mposition LiNi0.5Mno.502 was optimum. The synthesis of the logical
tep, LiNi1_y-zMnyCoz02 will be discussed below under manganese
.
46
Grenier et al. have shown that oxygen can be electrochemically
lated from strong basic electrolytes into some superconducting
als, thus increasing the oxidation state of the transition metal. If this
que can be extended to organic electrolytes at room temperature then it
be very useful in increasing the cell potential and possibly the
ty of cathode materials; the importance of such an oxidation was
when the capacity of N(CH3)4V307 was tripled when it was oxidized
47
CH3)4V308. Such oxidations must be performed below the
position temperature of the compound. A similar behavior has been
ed in the formation of hexagonal Mo03 by the thermal decomposition
hydrate, Mo03•0.6H20, formed by ion-exchange of the sodium form
6M nitric acid;48 this molybdenum oxide reacts reversibly with about
/Mo0349 similar to that of Mo03 itself.
HYDROTHERMAL AND SOLVOTHERMAL

SYNTHESIS OF CATHODE MATERIALS
In this synthetic method, the reactants are dissolved in water or
r solvent and then heated above the boiling point of the solvent for the d
length of time. The heating may be performed in either a
ntional oven or in a microwave oven. This process is used
ercially for a number of materials, including zeolites in tonnage
ties daily.
Although the technique has been used for several decades, it is only in
st ten years that much effort has been focused on transition metal
50 51
unds for battery electrode use. The early work has been reviewed '
mphasized many tungsten, molybdenum and vanadium oxides. Thin
of tungsten oxides for electrochromic or sensor applications can be
52
d in-situ in the hydrothermal reactor. Vanadium oxides in a wide
47
of morphologies have been formed hydrothermally, including the
chemically active nanotubes.53. 56 The studies of vanadium oxides
d the critical role that pH plays in determining the structure. At high
lues, the structural building blocks contain predominantly V04
dra, at intermediate values vo5 square pyramids, and under the most
conditions V06 octahedra.57 As V04 tetrahedra are only stable in the
dation state, materials containing these are unsuitable candidates for
es. In contrast the V06 octahedra are stable over many oxidation
+5 to +2, and so are much more likely to provide rechargeable
e materials. Thus, the electrochemically active o-
V4010 phases and
ls are all made usually at pH values less than 3. The electrochemically
47
H2V 308 phase has also been synthesized solvothermally, and shows
lectrochemically behavior.
A number of layered structure manganese oxides, MxMn02, where
Na or K have been synthesized by hydrothermal and other soft
try approaches. When KMn04 aqueous solutions are hydrothermally
58
at 170°C, well-crystalline layered K0.27Mn02.nH20 is formed. The
may be readily removed by heating. In this reaction, KOH is also
ed and oxygen gas evolved according to the equation:
The corresponding lithium and sodium manganese oxides have also

38
ynthesized from the permanganate. ' 59 The alkali content of these
nese oxide compounds is very dependent on the alkali metal and the
n medium; thus, in water x in MxMn02 is 0.47, 0.35 and 0.27 for Li,
K respectively. In these compounds the manganese site occupancy is
o unity, and about 0.6 Li/Mn can be readily cycled in lithium cells.
Chapter 3
on of cobalt or nickel to the reaction medium, allows the partial

tution of the manganese. This substitution even at the 1 to 5% level,
60
ses the electronic conductivity by two orders of magnitude, and
61
cantly improves the recyclability. Heating of the lithium compound
rts it to the spinel phase; this conversion reaction begins even at
38
, and is rapidly completed on heating to 400°C.
The hydrothermal treatment of A.-Mn02 with 0.5 M LiOH or 4 M
aqueous solutions at 170°C for 24 days leads to the formation of the
l" Li0.63Mn .7304•0.4H20 with a=8.18 A and the layered
62
Mn02.15•0.6H20 respectively. If the LiOH solution is stronger than 1
n LhMn03 was formed.
Manganese dioxides can also be prepared 63 using a sol-gel reduction
64 of
nganate using a reductant such as sugar or fumaric acid. The
m compound, Na0.45Mn02•0.6H20, has been formed by the reduction
of
04 by oxalic acid formed in-situ followed by a 600°C anneal and a
65
oxidation. The sodium ions can then be ion-exchanged by lithium
65
ransition metals. After annealing at 300-600°C the material shows
-like characteristics and cycles well on the 3V plateau; the X-ray
ns are also best fitted by a cubic spinel structure with a=8.129A,
tent with a 600°C anneal of a Li0.45Mn02 composition.
66 layered ranceite-type manganic acid,67HyMn1.002•nH20, has been
The
d following the method of Tsuji et al. by the acid hydrolysis of
04 at 60°C. Ion exchange of the manganic acid for two weeks in a large
of an equimolar solutions of 1M MCl and IM MOH, where M=Li,
K, leads to the alkali compounds. The alkali metal contents of these
als are very similar to those formed in the one-step hydrothermal
s described above, namely 0.42 Li, 0.33 Na and 0.26 K per formula
Although the acid is reported to have the IT structure with a=2.83A
67
c=7.55A, these compounds almost certainly have the same
58
ohedral structure as found for the hydrothermal compounds, but their
rystallinity precludes an effective structural analysis.
The acid leaching of layered manganates formed at elevated
ratures, such as of NaMn02, results in both ion exchange and partial
68
ion. The degree of reduction and ion-exchange is surprisingly
ely dependent on the acid strength, for example IM HCl leads to
1
Ho. 13Mnmo.JMn vo.s6•0.6H20 whereas O.OIM HCl leads to
1
Ho.27Mnmo.27Mn vo.73•0.6H20.
ION EXCHANGE METHODES FOR LITHIUM

MANGANESE OXIDE SYNTHESIS
Many sodium and potassium transition metal oxides had been
sized using standard high temperature methods in the 1970s and early
69
° 7
. These included KxMnOz, NaxMnOz, KxC00z, NaxCoOz,
73
71 72
72
2 and NaxCr02. The complete phase diagrams of these systems are
ell understood. Much less was known about the corresponding lithium
unds, and indeed the corresponding LiMn02 compound cannot be
sized at elevated temperatures.
Although ion-exchange methods had been used extensively in the
74 76
sis of solid electrolytes, in particular of beta-alumina - and beta" a
much less effort had been made to prepare lithium based cathodes
on-exchange until the 1990s. As noted above, layered LiMn02 cannot
nthesized at elevated temperatures because of its thermodynamic
lity. Thus, a number of studies have been made on ion-exchanging the
77 78
m analogs to the lithium form. Groups at Bordeaux • and St.
79
ws showed simultaneously that layered NaMn02 could be readily
ged to the layered form of LiMn02• In these experiments, the sodium
und is immersed in hot lithium salts dissolved in water-alcohol
es for periods of hours to days. Care has to be taken that the
ature is not too high, otherwise the spinel structure can be formed.
doped material has also been synthesized by the same approach of
80
reparing the sodium material followed by ion-exchange; • 81 these
81
containing materials show much enhanced cycling even though they
ally convert to the spinel phase even for LiMn0.5Co0.502•
The above layered LiMn02 phases have the same rhombohedral
re as the lithium, and other alkali phases formed by the hydrothermal
38
position of LiMn04, and other permanganates.58 However, these
hermal compounds have a lower alkali content, for example Li0.5Mn02
their initially being hydrates. A comparison for the potassium
nese oxides is given in Figure 3.4, which also distinguishes between o
high temperature regimes. Higher alkali contents can be obtained by g
out the hydrothermal reactions under very high alkali hydroxide
trations. The range of synthesis reactions and redox/intercalation
ns for the layered manganates are shown schematically in Figure 3.5.
2 85
- ion exchanged the Na 0.44Mn02 tunnel structure 86 giving
inantly the lithium analog, which shows excellent cycling with no
of structure unlike the layered manganese dioxides. These results
d convincingly that a stable manganese dioxide can be formed. The
can be improved by replacing part of the manganese by titanium, but
e capability is negatively impacted.87• 88 Pure Li0.44Mn02 was prepared
exchange,89 and also showed excellent cycling.
Chapter3
Lithium iron oxides, such as the layered LiFe02, have also been
90
sized by ion exchange of the sodium compounds. The sodium
und a-NaFe02 was formed by hydrothermal treatment of FeOOH
large excess of NaOH (Na:Fe = 250) at 220°C; this sodium compound
hen partially converted to the lithium compound by ion exchange
rothermal
Mn02 KxMn02 •nH20
3 block
C
Mn02 Ko.2sMn02 KxMn02-8 Ko.7Mn02 KMn0 2
• •
tunnel 3 block
• •
C
Mn02 Ko.2sMn0 2 KxMn02-8 Ko.7Mn02 KMn0 2
• •
tunnel 3 block 2 block • •
3.4. Composition of the potassium manganese oxides, both hydrotherma158 and high
69
ture forms.
90
hydrothermal conditions, with 80% conversion in four hours at
. Reaction between NaFe02 and molten lithium halides results in
complete ion-exchange results in the formation of LiFe02, but with
ic lattice constant of 4.162 A indicating that the lithium and iron have
91
e disordered giving a NaCl lattice. Reducing the temperature by
a eutectic mixture of LiCl and LiN03 resulted in the formation of a
phase. An extensive study of the formation of lithium-iron-oxygen
92
failed to synthesize the layered LiFe02 compound. However,
93
e et al. reported the formation of "layered" LiMn0 2 which consisted
layered a-NaFe02 phase and a spinel phase; they synthesized this by
change of a-NaFe02 in a LiCl/KCl molten salt at 400°C.
The electrochemically active vanadium oxide nanotubes are also
sized by ion exchange from the initially formed alkylammonium
55 56
giving the sodium or manganese compound.
STABILIZATION OF LAYERED LiMn02 AND

THE MIXED METAL OXIDES Li(Ni,Mn,Co)02
Although layered LiMn02 would be an ideal cathode material, it is
le relative to spinel formation and so the structure must be stabilized.
yered form and the spinel form have essentially the same structure as
ther with cubic close packing of the oxygen as discussed above for
2. The difference lies in the atomic positions the manganese ions
y. Thus, a means must be found to prevent the manganese from
Heat Aqueous
phous Solid Solution at 170°C
AxMn1 2 •y"H20
wherey" >1
.5. Schematic showing the synthesis, redox and intercalation reactions for Mn02 .
Chapter 3
ng. There are two distinct ways of doing this: geometric and electronic
zation of Mn02 layers. In the former, the structure is prevented from
to ccp, for example by placing pillars between the Mn02 sheets, or by
a none ccp structure, such as a two-block structure. Here, one is trying
feat thermodynamics by making the transformation difficult -
ally controlled reaction. In the latter, the manganese is made to be
cobalt-like by doping the lattice with a transition metal to the right of
such as nickel. Whichever way is chosen presents a synthetic
nge, as in many cases a metastable structure must be formed.
In the geometric approach, cations larger than lithium must be inserted
en the sheets. The potassium manganese oxides, described above,
the potassium ions are in trigonal prismatic sites thus breaking-up the
n ccp are readily formed at elevated temperatures or by hydrothermal
ques (see Figure 3.4). Though potassium enhances the cyclability,
ions eventually diffuse out of the lattice and it reverts to a spinel
re. Thus, immobile ions must be inserted into the lattice, between the
. Vanadium oxide pillars, probably V-0-V groups, were successfully
d using a combination of ion-exchange resins to form the starting
94
al followed by hydrothermal reaction; the structure is believed to be r
to that of the pillared Zn3(0H)2V207•2H20.95 This vanadium pillared
re was stable and did not convert to spinel even at 5 volts; however,
e characteristics were poor relative to LiMn02 itself. Leroux et al.96
sized a similar Ranceite based compound, which also was stable but
had poor rate capabilities.
The second geometric approach again aims to prevent the formation of
97
lattice, by forming a LiMn0 2 in which the oxygen layer stacking is
ger the three-block IABICAIBCIABI structure, but for example is the
ock IABICBIABI structure or even the hexagonal close-packing found
iS2 or in Co02• As none of these structures are stable for lithium at
emperatures, the sodium or potassium compounds must first be
sized by standard high temperature techniques; these may then be ion
ged to give the corresponding lithium compound as discussed above.
itial composition of the sodium or potassium compound may be varied
a wide range of stacking opportunities (see Figure 3.4); moreover, in
potassium compounds and in some of the sodium compounds the
cation is in trigonal prismatic coordination; as the lithium ion must be
hedral coordination, there will be sliding of the Mn02 layers leading
further stacking opportunities. All these structures are inherently
dynamically metastable, as the Mn02 sheets can slide relative to one
r and the manganese ions are quite mobile.
In the electronic stabilization approach, the goal is to make the
nese look electronically more like cobalt where the layer structure is
rmodynamically stable phase, for example by doping with nickel. 98
, the mixed phase LiNi1.yMny02. can be readily synthesized for y:S

The composition LiNi0.5Ni0.502, which is isoelectronic with LiCo02,
the optimum electrochemical behavior in the mixed nickel-manganese
45
, and its electrochemical profile is very similar to that of LiNi02.
compounds were synthesized by firing the monoxides of nickel and
99
nese with LiOH•H20 after mechanical grinding. As discussed above,
Ni02-LiCo02 system is completely miscible, and 20% of the cobalt
44
substituted by manganese giving LiCo0. 8Mn0.202. The logical next
100 101
was the synthesis of the system, LiNh.y-zMnyCoz02. ' These
unds are best formed by mixing the metals in aqueous solution, from
the hydroxides are precipitated before being fired at between 700 and
C in air. Just as with most of the other layered compounds, the
sis conditions are important in determining the metal ordering and in
bsequent electrochemistry. Figure 3.6 shows the effect of firing
ature on the 102 electrochemical capacity of the composition,
Mno.4Coo.zOz.
The synthesis of the spinel LiMn204 will not be specifically discussed
t is readily synthesized by the solid state reaction of stoichiometric
103 104
ies of Li2C03 and Mn02• ' A number of anneals at around 800°C
essary to produce the optimum electrochemical material.
ADDITION OF CARBON CONDUCTIVE

DILUENTS POST SYNTHESIS AND IN-SITU:
LiFeP04
Many cathode materials, unlike the metallic conducting TiS2, must be
d with a conductive diluent for use as a cathode. The challenge in
ases is to obtain an intimate mixture and contact between the active
e, the conductive diluent, and the electrolyte as it is only at this three
contact that reaction can occur. Thus, it is invariably best to try and
he cathode material itself electronically conductive so that the reaction
cur at the two phase contact between cathode and electrolyte. This
can either be added post the electrode material synthesis or during the
synthesis. Carbon black has been used in the synthesis of a number of
mm where 1.t was 1-1'0un to s1. gm'f1i cant1y reduce t he
'd
ox1 es, ·105 106 d
ation in cycling cells; the vanadium oxide is described as being coated
he carbon with an 8 wt% carbon loading. Carbons have been
vely studied in the case of lithium iron phosphate where a reducing
here is necessary during the synthesis process to prevent the oxidation
ous to ferric. This material, LiFeP04, will be discussed in some detail
Chapter 3
s it exemplifies well a number of synthesis procedures, some of which

nd others that are less successful.
107
Lithium iron phosphate, LiFeP04, has been reported to be an
ent cathode but with high a resistivity that limits the capacity to
0.6 lithium. This material is normally synthesized by the high
rature heating of lithium, iron and phosphate materials for example
acetate, iron oxalate and ammonium phosphate. Such salts can be
ssolved making an aqueous solution that can be dehydrated forming a
th the reactants in atomic contact with one another. This material is
red under an inert gas, to prevent oxidation of ferrous to ferric, at
5
4
3
2
1
0
,-.....
.>._.., 4
3
00
ro
....... 2
........
0
> 1
0
4
3
2
1
0
0 50 100 150 200
Capacity (mAh/g)
.6. Effect of synthesis firing temperature on the electrochemical capacity of
102
n0.4Co0.202.
s of Battery Materials 101
n 600 and 900°C forming LiFeP04. Such material has been well
erized,108 and has lattice parameters of a=10.329 A, b=6.007 A,
1 A and volume= 291.0 N. LiFeP04 can also 109 110been synthesized by a
of other techniques, including hydrothermal ' and microwave.
In the hydrothermal process, lithium acetate, ferrous oxalate and
horic acid dissolved in water are heated at around l20°C for up to 5
110 The reaction was found to be complete in much less than an hour. It
so be carried out under microwave heating. However, X-ray analysis
d that there is some iron on 111the lithium sites which restricts the rate of
insertion and removal, making the hydrothermal approach
able for LiFeP04 synthesis. Such disordering can be recognized from
tice parameters, a=10.403 A, b=6.023 A, c=4.726 A and volume =
N, which are significantly larger than the ordered material. A high
ature anneal appears to be necessary to cause the ordering of 112 the
and iron atoms. A recent report on the microwave processing of
04 would seem to have the same problem, as indicated by the rather
ectrochemical capacity even at elevated temperatures.
The original publication on the LiFeP04 electrode suggested that the
resistivity of the material was the cause of the rather low
chemical capacity, around 0.6 Li!Fe. Several efforts were directed at
ing the effective electronic conductivity, including modifying the
sis process so that a conductive coating is formed at the time of
sis - in-situ coating. Such approaches have been used in the case of
113 114
um oxides. Ravet et al. • proposed firing the initial reactants with
aceous materials such as sugar, so as to form a carbon coating thus
ing the overall conductivity of the composite so formed. These
sites allowed almost complete utilization of the lithium. Even more
sive results were obtained when a carbon gel was fired with the
leading to complete utilization of the LiFeP04 at 5 mg/cm2
115
ts,
116
gs. Masquelier showed the effectiveness of ball-milling carbon with
04 in giving complete capacity at elevated temperatures. A
rative study of the effect of synthesis/coating procedure, including in-
r bon ge I, sugar and aqueous ge1 versus post synthes1.s coatm. g
117
usm· g e range of techniques suggested that the coating procedure

is not
118
larly important at high cathode loadings, 20 - 70 mg/cm2.
er, the addition of carbon at high levels is very deleterious
119 to the
low volumetric energy density of the LiFeP04 cell. What may be
critical as the coating process, is the nature of the material formed,
ng its particle size and absence of ferric material and unlithiated
ate so as to enhance the lithium ion diffusivity.
An alternative to the in-situ coating of the lithium iron phosphate
the synthesis of the cathode material is to actually modify the intrinsic
nic conductivity of the phosphate itself by doping the lattice as
Chapter 3
bed above for the layered manganese oxides. ° Chiang et al. have
6 120
the lattice with for example 1 wt% niobium and showed that the
9
nce is reduced from around 10 ohm-em in very pure LiFeP04 to
2
10 ohm-em. The very low doping levels have a surprisingly large
t; the material is black, but there are apparently some carbonaceous
2
es present. The electrochemical capacity is around 80% at 0.4 mA/cm
2
2.5 mg/cm loading, significantly lower than that from the carbon gel
115
gs. The resistance of regular battery grade LiFeP04 formed from
5 6 118
t grade chemicals is around 10 -10 ohm-em, not significantly
60
nt than that of undoped manganese oxide. The impact of synthesis
ure on the lithium ion conductivity is still not known and this might
as critical if not more critical than the electronic conductivity.
SYNTHESIS OF ANODE MATERIALS

Lithium when used in the pure state is very prone to dendrite
ion on electrolytic deposition, thus metal alloys have been
gated for more than 30 years in organic electrolytes. Dey first
bed 121 the electrochemical formation in organic electrolytes of lithium
with a wide range of metals, including aluminum and tin. In these
ments, the cell was simply short-circuited externally or a 1 kohm
al resistor was used so that the voltage profile with time could be
ed. The compounds formed were shown to be same as those formed h
temperature metallurgical reaction. These intermetallic compounds
rmally very brittle, so it is not easy to form an electrode out of them;
xxon in its LiAl/TiS2 cells constructed the alloy anode in-situ by
g a sheet of lithium on top of the aluminum on cell construction. No
n occurs until the electrolyte solution is added to the cell, immediately
o cell closure. The formation and cycling behavior of the lithium
um alloys synthesized both electrochemically and thermally have
122 123
tudied. Wang et al. synthesized at high temperatures the lithium
of tin, antimony and bismuth in both a porous and dense state by
g the components together in molybdenum crucibles; they used these
als to determine the lithium diffusion coefficients and phase equilibria
nic electrolyte based cells.
Herold 124 first formed the alkali metal intercalation compounds of
te, but the electrochemical formation of the lithium compound LiC 6
mplicated by the co-intercalation of the electrolyte solvent. 125 126
• • Use
solid electrolyte, polyethylene oxide, allowed the formation at 80°C of
vent free sodium graphite compound NaC64; higher sodium content
127
als were not formed. This shows the importance of choice of solvent
synthetic process, or in the battery, where it may participate in the
n even if only as a solvating specie. This can be equally true for the
e as for the anode. Thus, when propylene carbonate was used as the
lyte solvent in the LiffiS2 cell, if a trace of water was present then
ene carbonate was co-intercalated with the lithium expanding the
hree fold. The ether dioxolane was found to be a particularly ve
electrolyte solvent for lithium batteries, not being co-intercalated eing
128 129
very good for lithium electroplating. " Such solvents were
ed for use with the lithium-graphite anode. °13
Carbon based anodes
131
w used in essentially all commercially lithium-ion batteries, and
l to their success is choice of solvent with ethylene carbonate being
successful. Hard carbons rather than pure graphite are preferred for
132
cycle life and capacity. The batteries are constructed in the
rged state, and the lithium is electro-intercalated into the graphite on
st charge. Thus, the cathode material provides all the lithium for the
which means that excess anode cells are not constructed. Thus, any
side-reactions have an immediate negative impact on cell capacity.
133
et al. showed that the graphite surface is coated with a protective
n the first synthesis of C6Li which irreversibly consumes some of the
. This film is stable and so need not be reconstructed on each
chemical cycle but it does permanently reduce the capacity of the
. Hence, there is much interest in synthesizing other potential anode
als that have a higher capacity, and safer operation than graphite based
unds. The outstanding stability and rechargeability of graphite after
t cycle will be hard to surpass.
MECHANOCHEMISTRY AND MECHANICAL

GRINDING
Much emphasis has therefore been placed on investigating other
materials that might store lithium by a simple reaction around room
ature. A novel approach to the synthesis of new materials developed
inantly in Russia at the Institute for Solid State Chemistry and
134 135
nochemistry ' is that of mechanochemistry or mechanical
g. In this approach physico-chemical transformations and chemical
ns of solids are induced by mechanical action. Although the grinding
cally performed at room temperature, the actual temperature at the
e contact is probably substantial. Although most scientists were
al initially about mechanochemistry the technique has now gained
136 137
ete credibility. ' A number of anode materials have been
sized by this technique, and carbon has been effectively coated onto
116
04• The precursor
138
for LiFeP04 has also been formed using
nical grinding, which allowed the formation of LiFeP0 4 even at
Chapter 3
; the quality of the material in electrochemical cells has not been

bed.
A number of anode materials have been synthesized by
nochemical reactions. Many intermetallic compounds have been
sed for anode use using different concepts, one where the structure of
6ermetallic
5• 139 140is at least in part retained during reaction with lithium such
Sn ' A second approach is where the structure is destroyed but
omponent acts as a mechanical buffer and the other reacts with the
m, an example might be Sn2Mn. A third option is where a composite
e of intermetallics is used.
The phase Cu6Sn5 phase is very well characterized, being formed at
erface when copper is soldered using tin based materials. It has been
igated as a potential lithium anode, and has been synthesized in a
of ways. In a mechanochemical process, powders of tin and copper
and-grinding were mixed and about 3-4 wt % of carbon was added as
lubricant prior to milling for 20 hours; purity of phase was shown by
140
analysis. On cycling lithium is inserted into this compound forming
nCu like phase. Cu6Sn5 has also been prepared by reaction of copper
141
under argon at 400°C. An alternative synthesis approach has been
142
d by Tamura et al. They electrodeposited tin onto a copper foil,
nnealed the foil at 200°C for 24 hours in vacuum to form Cu 6Sn5. The
chemical behavior of some of these tin containing materials is
143
red in Figure 3.7, where it can be seen that the pure tin foil retains
142
me capacity as the annealed Cu 6Sn5 phase for the first ten cycles.
800 Tin foil 0.8 mA/cm

2
OJ)
600
-400
u
200
10
Cycle
.7. The electrochemical capacity of several tin based anodes, synthesized by different
es.
144
The intermetallic InSb has also been prepared by ballmilling. In this
dium and antimony metals in a 1:1 ratio were ballmilled in air at room
ature for 20 hours; after which the resulting powder was annealed
flowing argon at 400°C for 12 hours. This material had a capacity of
220 mAh/g for the 2nd to 20th cycle.
Mechanical alloying is particularly amenable to producing
145 146
etallic alloys or mixtures of them. Mao et al. ' prepared
sites of SnFe2 and SnFe3C. Powders of tin, iron and carbon were
in hardened steel vials together with hardened steel balls; these were
in an argon environment, and then shaken for 20 hours. X-ray
tion showed that the resulting grain size was about 10 nm. In this case
e SnFe2 phase is electrochemically active, the carbide phase being the
matrix; the cell capacity which exceeds 1500 mAh/cc (around 200
) for 24% SnFe2 can be enhanced by increasing the ratio of active
nent but it is at the expense of capacity retention. The same
147
nical alloying process has also been used to reduce the particle size
e2 formed by high temperature processes. A recent review of tin-based
148
etallics details some of the other synthetic processes used.
NOWLEDGEMENTS
This work was supported in part by the National Science Foundation

US Department of Energy.
ERENCES
Gaines, R. W. Francis, G. H. Newman, B. M. L. Rao, "Ambient Temperature Electric
le Batteries Based on Lithium and Titanium Disulfide", in lith lntersociety Energy
ersion Conference (1976) 418. Stateline, Nevada.
ines, R. Cuenca, "Costs of Lithium-Ion Batteries for Vehicles", in Argonne National
atory Report ANUESD-42. 2000, Argonne National Laboratory, p. 57.
Thompson, F. R. Gamble, C. R. Symon, Mater. Res. Bull. 10 (1975) 915.
Thompson, Phys. Rev. Lett. 35 (1975) 1786.
not, A. L. Blanc, J. Rouxel, Bull. Soc. Chirn. France (1969) 2670.
xel, M. Danot, J. Bichon, Bull. Soc. Chirn. France (1971) 3930.
dorff, Chirnia 19 (1965) 489.
not, J. Rouxel, Cornptes Rendus Acad. Sc. Paris C276 (1973) 1283.
Whittingham, Prog. Solid State Chern. 12 (1978) 41.
. Whittingham, U.S. Patent 4040017 and U.K. Patent 1468244 (1973)
. Whittingham, J. Chern. Soc. Chern. Cornrn. (1974) 328.
. Dines, Mater. Res. Bull. 10 (1975) 287.
. Whittingham, M. B. Dines, J. Electrochern. Soc. 124 (1977) 1387.
. Murphy, F. J. DiSalvo, G. W. Hull, J. V. Waszczak, lnorg. Chern. 15 (1976) 17.
chollhom, H. Meyer, Mater. Res. Bull. 9 (1974) 1237.
Chapter 3
V. SubbaRao, J. C. Tsang, Mater. Res. Bull. 9 (1974) 921.

S. Whittingham, U.S. Patent 4009052 and U.K. Patent 1468416 (1973)
W. Murphy, J. N. Carides, F. J. DiSalvo, C. Cros, J. V. Waszczak, Mat. Res. Bull. 12
77) 825.
W. Murphy, P. A. Christian, Science 205 (1979) 651.
chollhorn, A. Payer, Angew. Chern. Int. Ed. 24 (1985) 67.
inha, D. W. Murphy, Solid State Ionics 20 (1986) 81.
. Goodenough, A. Manthiram, B. Wnetrzewski, J. Power Sources 43-44 (1993) 269.
Mitzushima, P. C. Jones, P. J. Wiseman, J. B. Goodenough, Mat. Res. Bull. 15 (1980)
.
zawa, Solid State Ionics 69 (1994) 212.
G. Amatucci, J. M. Tarascon, L. C. Klein, J Electrochem Soc. 143 (1996) 1114.
S. Whittingham, Mater. Res. Bull. 13 (1978) 959.
. Dahn, U. v. Sacken, C. A. Michal, Solid State Ionics 44 (1990) 87.
. Johnson, J.-S. Kim, A. J. Kropf, A. J. Kahaian, J. T. Vaughey, M. M. Thackeray,
trochem. Commun. 4 (2002) 492.
A. Py, R. R. Haering, Can. J. Phys. 61 (1983) 76.
. Haering, J. A. R. Stiles, K. Brandt, U.S. Patent 4,224,390 (1980)
. Reimers, J. R. Dahn, Phys. Rev. B 46 (1992) 3236.
ossen, J. N. Reimers, J. R. Dahn, Solid State Ionics 62 (1993) 53.
. Gumrnow, M. M. Thackeray, W. I. F. David, S. Hull, Mater. Res. Bull. 27 (1992)
.
. Gumrnow, M. M. Thackeray, J. Electrochem. Soc. 140 (1993) 3365.
. Amatucci, J. M. Tarascon, D. Larcher, L. C. Klein, Solid State Ionics 84 (1996) 169.
urukhin, 0. Brylev, P. Hany, B. R. Churagulov, Solid State Ionics 151 (2002) 259.
G. S. R. Thomas, W. I. F. David, J. B. Goodenough, Mater. Res. Bull. 20 (1985) 1137.
hen, M.S. Whittingham, J. Electrochem. Soc. 144 (1997) L64.
udhavi, G. V. SubbaRao, B. V. R. Chowdari, S. F. Y. Li, J. Power Sources 93 (2001)
.
udhavi, G. V. SubbaRao, B. V. R. Chowdari, S. F. Y. Li, Solid State Ionics 152
2) 199.
u, K. Xu, T. R. Jow, K. Amine, Electrochemical Society Fall Meeting, Salt Lake City
2)
adoune, C. Delmas, J. Materials Chemistry 6 (1996) 193.
elmas, I. Saadonne, Solid State Ionics 53-56 (1992) 370.
anase, T. Ohtaki, M. Watanabe, Solid State Ionics 151 (2002) 189.
E. Spahr, P. Novak, B. Schnyder, 0. Haas, R. Nesper, J. Electrochem. Soc. 145 (1998)
3.
. Grenier, M. Pouchard, A. Wettiaux, Current Opinion in Solid State & Materials
nce, 1 (1996) 233.
. Chirayil, P. Y. Zavalij, M.S. Whittingham, Chern. Mater. 10 (1998) 2629.
uo, P. Zavalij, M.S. Whittingham, Eur. J. Solid State Chern 31 (1994) 833.
. Guo, P. Zavalij, M.S. Whittingham, J. Solid State Chern 117 (1995) 323.
. Whittingham, Current Opinion in Solid State & Materials Science, 1 (1996) 227.
S. Whittingham, J. Guo, R. Chen, T. Chirayil, G. Janauer, P. Zavalij, Solid State Ionics
1995) 257.
uo, Y. J. Li, M. S. Whittingham, J. Power Sources 54 (1995) 461.
E. Spahr, P. Stoschitzki-Bitterli, R. Nesper, M. Miiller, F. Krumeich, H. U. Nissen,
ew. Chern. Int. Ed. Engl. 37 (1998) 1263.
. Spahr, P. Stoschitzki-Bitterli, R. Nesper, 0. Haas, P. Novak, J. Electrochem. Soc.
(1999) 2780.
dstrom, T. Gustafsson, S. Nordliner, Electrochem. Solid-State Lett. 4 (2001) A129.
obley, K. Ngala, S. Yang, P. Y. Zavalij, M.S. Whittingham, Chern. Mater. 13 (2001)
2.
Y. Zavalij, M.S. Whittingham, Acta Cryst. B55 (1999) 627.
Chen, P. Zavalij, M.S. Whittingham, Chern Mater 8 (1996) 1275.
Chen, T. Chirayil, M.S. Whittingham, Proceedings of the lOth International
posium on Solid State lonics, Singapore, December 1995. Solid State Jonics 86-88
96) 1.
harma, G. Moore, F. Zhang, P. Y. Zavalij, M. S. Whittingham, Electrochem. Solid-
te Lett. 2 (1999) 494.
Zhang, M.S. Whittingham, Electrochem. Solid-State Letters 3 (2000) 309.
Feng, H. Kanoh, Y. Miyai, K. Ooi, Chern. Mater. 7 (1995) 1226.
Ching, J. A. Landrigan, M. L. Jorgenson, N. Duan, S. L. Suib, C.-L. O'Young, Chern.
er. 7 (1995) 1604.
Bach, M. Henry, N. Baffler, J. Livage, J. Solid State Chern. 88 (1990) 325.
ranger, S. Bach, J.P. Pereira-Ramos, N. Baffler, J. Electrochem. Soc. 147 (2000)
6.
eroux, D. Guyomard, Y. Piffard, Solid State lonics 80 (1995) 299.
Tsuji, S. Komarneni, Y. Tamaura, M. Abe, Mat. Res. Bull. 27 (1992) 741.
Omomo, T. Sasaki, M. Watanabe, Solid State lonics 151 (2002) 243.
Delmas, C. Fouassier, Z. Anorg. Allg. Chern. 420 (1976) 184.
Mendiboure, C. Delmas, P. Hagenmuller, J. Solid State Chemistry 57 (1985) 323.
Delmas, C. Fouassier, P. Hagenmuller, J. Solid State Chemistry 13 (1975) 165.
Delmas, J.-J. Braconnier, C. Fouassier, P. Hagenmuller, Solid State lonics 3/4 (1981)
.
Delmas, M. Devalette, C. Fouassier, P. Hagenmuller, Mat. Res. Bull. 10 (1975) 393.
Y. Yao, J. T. Kummer, J. lnorg. Nucl. Chern. 29 (1967) 2453.
S. Whittingham, R. A. Huggins, J. Electrochem. Soc. 118 (1971) 1.
S. Whittingham, R. A. Huggins, NBS Special Publications 364 (1972) 139.
Delmas, F. Capitaine, 8th IMLB 8 (1996) 470.
apitaine, P. Gravereau, C. Delmas, Solid State lonics 89 (1996) 197.
R. Armstrong, P. G. Bruce, Nature 381 (1996) 499.
D. Robertson, A. R. Armstrong, A. J. Fowles, P. G. Bruce, J. Mater. Chern. 11 (2001)
.
R. Armstrong, R. Gitzendanner, A. D. Robertson, P. G. Bruce, Chern. Commun. (1998)
3.
M. Doeff, M. Y. Peng, Y. Ma, L. C. DeJonghe, J. Electrochem. Soc. 141 (1994) L145.
M. Doeff, T. J. Richardson, L. Kepley, J. Electrochem. Soc. 143 (1996) 2507.
M. Doeff, M. Y. Peng, Y. Ma, S. J. Visco, L. C. DeJonghe, US Patent 5,558,961
96)
M. Doeff, K.-T. Hwang, T. J. Richardson, L. C. DeJongue, Electrochem. Soc. Proc.
2 (1999) 184.
G. Mumme, Acta Cryst. B24 (1968) 1114.
M. Doeff, T. J. Richardson, K. T. Hwang, A. Anapolsky, Electrochem. Soc. Proc. 99-
2000) 48.
M. Doeff, T. J. Richardson, K.-T. Hwang, A. Anapolsky, ITE Battery Lett. 2 (2001) B.
R. Armstrong, H. Huang, R. A. Jennings, P. G. Bruce, J. Mater. Chern. 8 (1998) 255.
Tabuchi, C. Masquelier, T. Takeuchi, K. Ado, I. Matsubara, T. Shirane, R. Kanno, S.
u, H. Sakaebe, H. Okuyama, 0. Nakamura, Abstracts of the lOth Int. Conf on Solid
e Ionics, Singapore, December 3-8 (1995) 197.
uchs, S. Kemmlersack, Solid State lonics 68 (1994) 279.
Tabuchi, K. Ado, H. Sakaebe, C. Masquelier, H. Kageyama, 0. Nakamura, Solid State
cs 79 (1995) 220.
hirane, R. Kanno, Y. Kawamoto, Y. Takeda, M. Takano, T. Kamiyama, F. Izumi,
d State lonics 79 (1995) 227.
Chapter 3
hang, K. Ngala, M.S. Whittingham, Electrochem. Commun. 2 (2000) 445.

. Zavalij, F. Zhang, M.S. Whittingham, Acta Cryst. C53 (1997) 1738.
eroux, A. L. G. L. Salle, D. Guyomard, Y. Piffard, J. Materials Chern. 11 (2001) 652.
S. Whittingham, 1996 USDOE BAIT Annual Report (1997)
hen and M. S. Whittingham, Proc. Annual Automotive Technology Development
tomers Coordination Meeting, DOE-OTT (Dearborn, MI, Oct. 27-30, 1997), III
97) 301.
ossen, C. D. W. Jones, J. R. Dahn, Solid State lonics 57 (1992) 311.
Ohzuku, Y. Makimura, Chern. Lett. (2001) 642.
Lu, D. D. MacNeil, J. R. Dahn, Electrochem. Solid-State Lett. 4 (2001) A200.
K. Ngala, N. A. Chernova, M. Ma, P. Y. Zavalij, M.S. Whittingham, Chern. Mater. 15
03) in press.
Guyomard, J. M. Tarascon, J. Electrochem. Soc. 139 (1992) 937.
M. Tarascon, W. R. McKinnon, F. Coowar, T. N. Bowmer, G. Amatucci, D.
yomard, J. Electrochem. Soc. 141 (1994) 1421.
C. Torardi, C. R. Miao, M. E. Lewittes, Z. Li, Electrochemical Society Proceedings,
00-21 (2000) 68.
C. Torardi, C. R. Miao, M. E. Lewittes, Z. Li, J. Solid State Chern. 163 (2002) 93.
K. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Electrochem. Soc. 144 (1997)
88.
S. Andersson, B. Kalska, L. Haggstrom, J. 0. Thomas, Solid State lonics 130 (2000)
B. Armand, Personal Communication (2001)

Yang, P. Y. Zavalij, M.S. Whittingham, Electrochem. Commun. 3 (2001) 505. Yang,
Y. Song, P. Y. Zavalij, M.S. Whittingham, Electrochem. Commun. 4 (2002)
9.
Higuchi, K. Katayama, Y. Azuma, M. Yukawa, M. Suhara, J. Power Sources (2003)
press.
Ravet, J. B. Goodenough, S. Besner, M. Simoneau, P. Hovington, M. Armand,
ctrochem. Soc. Abstracts, 99-2 (1999) 127.
Ravet, S. Besner, M. Simoneau, A. Vallee, M. Armand, J.-F. Magnan, European
ent 1049182A2 (2000).
Huang, S.-C. Yin, L. F. Nazar, Electrochem. Solid-State Lett. 4 (2001) A170.
Masquelier, C. Wurm, M. Morcrette, J. Gaubicher, Int. Meeting on Solid State lonics,
irns, Australia, July 9-13, paper A-IN-06, (2001)
Dominko, M. Gaberscek, J. Drofenik, M. Bele, S. Pejovnik, Electrochem. Solid-State
t. 4 (2001) A187.
Yang, Y. Song, K. Ngala, P. Y. Zavalij, M. S. Whittingham, J. Power Sources (2003)
press.
Chen, J. R. Dahn, J. Electrochem. Soc. 149 (2002) A1184.
Y. Chung, J. T. Bloking, Y.-M. Chiang, Nature Materials 1 (2002) 123.
N. Dey, J. Electrochem. Soc. 118 (1971) 1547.
M. L. Rao, R. W. Francis, H. A. Christopher, J. Electrochem. Soc. 124 (1977) 1490.
Wang, I. D. Raistrick, R. A. Huggins, J. Electrochem. Soc. 133 (1986) 457.
Herold, Bull. Soc. Chim. France (1955) 999.
0. Besenhard, H. P. Fritz, Electroanal. Chern. lnterfac. Electrochem. 53 (1974) 329.
0. Besenhard, Carbon 14 (1976) 111.
Ge, M. Fouletier, Solid State lonics 28-30 (1988) 1172.
S. Whittingham, G. H. Newman, US Patent 4,086,403 (1976).
S. Whittingham, G. H. Newman, J. Electrochem. Soc. 128 (1981) 706.
Basu, US Patent 4,423,125 (1982).
Mohri, N. Yanagisawa, Y. Tajima, H. Tanaka, T. Mitate, S. Nakajima, M. Yoshida,
Yoshimoto, T. Suzuki, H. Wada, J. Power Sources 26 (1989) 545.
. Nishi, in: M. Wakihara and 0. Yamamoto (Eds.), Lithium Ion Batteries, Kodansha,
kyo, 1998,p. 181.
. Fong, U. v. Sacken, J. R. Dahn, J. Electrochem. Soc. 137 (1990) 2009.
. V. Boldyrev, Experimental Methods in Mechanochemistry of Inorganic Solids, 1986,
ovosibirsk. Nauka (in Russian).
. V. Boldyrev, Mechanochemistry and Mechanical Activation of Solids, 2004, World
ientific.
. Senna, Solid State Chemistry 63-65 3.
Gutman, Mechanochemistry of Materials, 1997, Cambridge International Science
b.
. V. Kosova, E. T. Devyatkina, Electrochem. Society Abstracts, 2003-1 (2003) 1009.
. D. Kepler, J. T. Vaughey, M. M. Thackeray, Electrochem. Solid-State Lett. 2 (1999)
7.
T. Vaughey, K. D. Kepler, C. S. Johnson, T. Sarakonsri, R. Benedek, J. O'Hara, S.
ackney, M. M. Thackeray, Electrochem. Soc. Proc. 99-2 (1999) 280.
. M. Thackeray, J. T. Vaughey, A. J. Kahaian, K. D. Kepler, R. Benedek, Electrochem.
omm.1(1999)111.
. Tamura, R. Ohshita, M. Fujimoto, S. Fujitani, M. Kamino, I. Yonezu, J. Power
urces 107 (2002) 48.
Yang, P. Y. Zavalij, M. S. Whittingham, Mater. Res. Soc. Proc. 756 (2003) 295.
T. Vaughey, J. O'Hara, M. M. Thackeray, Electrochem. Solid-State Lett. 3 (2000) 13.
Mao, R. L. Turner, I. A. Courtney, B. D. Fredericksen, M. I. Buckett, L. J. Krause, J.
Dahn, Electrochem. Solid-State Lett. 2 (1999) 3.
Mao, J. R. Dahn, J. Electrochem. Soc. 146 (1999) 414.
Mao, R. A. Dunlap, I. A. Courtney, J. R. Dahn, J. Electrochem. Soc. 145 (1998) 4195.
. Winter, J. 0. Besenhard, Electrochim. Acta 45 (1999) 31.
pter 4
ANODES AND COMPOSITE ANODES:

AN OVERVIEW
L. F. Nazar and 0. Crosnier
Department of Chemistry, University of Waterloo,

Waterloo, Ontario, Canada N2L 3Gl
INTRODUCTION: LITHIUM METAL AND

CARBON-BASED MATERIALS
The past years have witnessed significant improvements in the nature

pability of energy storage systems. A combination of fundamental and
logically driven studies were responsible for these advances, which
een reviewed in Chapter 1. Motivation for the advancement has been
d by the ever-increasing demands that a multitude of applications are
g on the energy storage battery. In addition to the ever-popular needs
rtable energy electronics, future demands lie in rechargeable batteries
brid electric vehicles, miniaturized electronics; space exploration,
1 23
rupted power supplies, and medical devices. • • These various
ations differ in their requirements of the energy storage cell: some
g high power, and others needing high capacity. All demand safety,
er, and most require excellent stability of the cell over long-term usage. l
cells are still many years away from meeting the needs in most of
areas, lithium-ion rechargeable batteries offer the only technological
n at present, and the best long-term solution for the foreseeable future
ny of the areas. Developments in the positive electrode arena have
ed materials capable of gravimetric energy densities in the regime of
Ahlg; therefore anode materials are sought to match these high-capacity
es. This chapter is devoted to a review of the new technologies in the
at may address these issues.
nd Composite Anodes: An Overview 113
Issue of Safety: a Brief History
he problems of the lithium battery, and inception of the lithium-ion

as an alternative arose from well-known problems with the anode. In
ade before 1990, lithium batteries were introduced as novel, high
density cells whose promise lay in the low atomic weight of lithium
nd the high degree of reversibility of the systems. The first materials to
commercialization, TiS2 4, and MoS2,5 were used as positive
es versus lithium foil in several commercial products. These layered
n metal dichalcogenides benefited from relatively fast lithium ion
n, along with appealing cell voltages vs. Li (about 1.8V). The use of
metal as anode material in rechargeable solution-electrolyte batteries
carded in the late 1980's, however, after a few well-documented
ular failures of the Li/MoS2 cells in operating devices. The problem is
derstood to arise from reaction of the solution electrolyte with the
lithium surface. The electrode surface, passivated by the formation
um alkyl carbonates, sustains repeated non-epitaxial deposition of
after each cycle as a result of the complex surface chemistry. This in
sults in continuously increasing electrode surface area, and more e
surface reactions. The insulating, complex deposits that build up on
face of the electrode lead to the growth of uneven "fingers" or
s. Ultimately cell failure occurs when the dendrites reach the positive
e surface and/or the insulating layer causes a large rise in cell
nce, leading to an internal short-circuit.
The "rocking-chair" battery, pioneered in the mid 80's, solved the
problem by replacing the metallic lithium anode with a lithium
n compound. These "rocking-chair" or "shuttle-cock" batteries
ed on the principle of simultaneous lithium ion de-insertion and
n in both electrodes. Lithium ions shuttle back and forth across the
yte, in concert with electron transport through the external circuit.
ese cells contained no metallic lithium, they were much safer than the
battery. The choice of the anode is based on properties that include
ertion kinetics, and a redox potential vs. Li which is less than that of
hode to provide a sufficiently large cell voltage. In the earliest
ion, the anode employed was a low potential lithiated insertion
nd such as LixW02.6 Lower density carbon-based insertion materials
the oxide in batteries commercialized by Sony Energytech in 1991,7
ise to a series of patents and a product called a "lithium-ion" battery. l
capacities and cell voltages were obtained by marrying the
ceous material with a lithium-containing high-voltage cathode that
ed as the lithium reservoir, i.e., LiCo02• These cells have since
hed dominance in the lithium-ion battery market. Much effort has been
d in the last decade to improve these batteries, with research devoted
Chapter4
8 9 0 11 12
hree components of the cell. ' ,t , ' These have resulted in significant
vements in the performance (increase in energy density and rate
lity by a factor of two) and cost (more than a factor of two). Aspects of
sitive electrode are discussed in Part III and the electrolyte in Part IV of
ook. From the anode point of view, these directions have concentrated e
areas: improvement of carbonaceous materials; utilization of metallic m
in combination with stable electrolytes; and a search for new als.
The most suitable direction for new anode materials is not
ndent of the other components in the cell - for example, successful
mentation of a 5V cathode and electrolyte system would allow for
s that have working potentials above 1 V to be utilized. Similarly,
ons at improving the stability of metallic lithium anodes are directed
ds implementation of lower voltage cathodes that have the advantage of
lyte stability.
Carbonaceous Materials
Most of the improvements in anodes actually implemented over the last

have involved carbonaceous materials, in concert with LiCo02 or
layered oxides. This is partly based on the fact that graphite and
tic carbons, well known as intercalation compounds, have been
13
ghly studied in the past 30 years. The mechanism of lithium
lation that gives rise to a maximum uptake corresponding to LiC6 is
nderstood. The reasons that lie behind the commercial success of
-based negative electrodes include the relatively low inherent cost of
, its excellent reversibility for lithium insertion, and its formation of a
tive surface film with many electrolyte solutions. When a graphite
de is polarized to negative potentials during initial Li intercalation in an
ne carbonate based solution, the organic solvent reductively
14 15
poses on the graphite surface to form a stable surface film. • This
16
which is often called the solid electrolyte interface (SEI) effectively
ates the graphite surface and prevents further co-interaction and
position of solvent molecules, allowing only Li ion migration.17 As
ne carbonate is quite viscous, it is often used in combination with other
arbonate solvents such as dimethyl carbonate.
One form of carbon is the soft or poorly ordered carbons ("petroleum
products). For soft carbons heat-treated below graphitization
atures, turbostratic disordered carbons are obtained. Lithium cannot be
accommodated in the interlamellar graphene galleries and low
ties are observed. ' ' °
18 19 2
For graphitized soft carbons, the level of
er has a direct effect on the mechanism of the reaction with lithium.
isplay complex behavior due to surface effects, internal porosity and
19 21
g surface bonds. ' Capacities well above the theoretical limit of 372
are observed, but are inevitably associated with a large hysteresis.
pecific new forms of carbon which have replaced petroleum coke and
rise to high capacity/rate capability anodes include MCMB
arbon microbeads or spherical graphite), MCF (microcarbon fiber),
ith natural and synthetic graphite flakes. MCMB's in particular, are
utilized. They exhibit large capacities after heat treatments at low
22 23 24
tures, as well as after graphitization at higher temperatures. • •
materials are derived from petroleum industry byproducts or coal
s, but are expensive owing to low yields. They, along with other
ceous and graphitic anodes are discussed in detail in Chapters Sand 6.
Looking to the Future

espite the advances made in carbons, it has been recognized for some
ow that new anode materials are needed. First, graphitic carbons can
rom solvent co-intercalation resulting in large interlayer expansions
subsequent degradation of the graphite structure. This, however, can
ressed by the adroit choice of solvent (i.e., alkyl carbonate mixture)
25
additives. More importantly, carbons suffer from a fundamental
on in gravimetric capacity, and particularly in volumetric capacity
ical limit of 1000 mAh/cc). Developments in cathode materials that
fectively doubled the capacity within the last few years demand an
o match. An increase in the operating voltage above the current values
bonaceous materials is highly desirable to inhibit lithium metal
on that can occur at fast rates. To date, no new non-carbon containing
have been commercialized however, as they currently are not
tive with carbon in terms of cost or stability on cycling.
onetheless, a large number of alternative possibilities for negative
es have been recently reported in the literature that may lead to new
ns in meeting the anode challenge. One important area is in the
ment of composite materials. It is apparent that promising
ristics can result from development of nanostructured domains within
s during the initial cycles of the cell. The partial disorder inherent to
aterials can have a substantial impact, affecting the kinetics and/or
ynamics of the materials in highly beneficial ways. In fact, the
ed development of materials over the past few years suggests that one
ent alone will not suffice. These composites fall into two categories:
eated in-situ as mentioned above, or those deliberately constructed
arious synthesis-by-design approaches. What follows is a brief w
of some developments that highlights key points in anode design n
these concepts.
Chapter 4
COMPOUNDS AND COMPOSITES OF Sn, Sb AND

AI
Amongst the first materials to be examined as electrodes for Li uptake
potential, well before carbon gained prominence, were other members
oup IV, and the neighboring Group III and Group V elements. These
e AI, Si, Ge, Sn, Pb, Sb and Bi, which all form alloys with Li at
ely low voltages vs. Li. Much of the early work concentrated on alloys
AI (Li9Al4 being the end member, although LiAl is observed in
26
e), and Group IV elements which display stoichiometries as high as
/M. Studies on the Li-Sn system, carried out at elevated temperatures
27
ggins and others, '28 show that seven line phases exist within the phase
m: LizSn5, LiSn, LhSn3, Li5Snz, Li 13Sns, Li7Snz and LizzSns. Based on
nown structures, Courtney et al. recently were able to calculate the
tical electrochemical voltage profile in the lithium-tin system using a
29
o-potential plane-wave approach. Their results were in good
30
ment with experiment. Silicon, being the lightest member of Group IV
from carbon, shows the highest gravimetric capacities (up to 4000
for the end member Liz 1Si5) although achieving this at ambient
rature is difficult. 31 A detailed re-examination of the crystal structure of
d members of the Li/Group IV compounds has shown, in fact, that
gh the end member composition for Si is correct, the precise
ometry for Ge, Sn and Pb is Li339Ge80, Li341Sn80, and Li340Pb80. The
composition can therefore be approximated as Li 17M4, not LizzM5. 32
Regardless of the element, the formation of alloys within Groups III, IV
is associated with a large volume expansion that results in pulverization
electrode material and loss of electrical contact between the material
on prolonged cycling. The degree of volume expansion depends on the
nt, as shown in Table 4.1. In all cases the resulting rapid capacity fade
s their practical use as electrodes. To overcome this problem, efforts
urned to the encapsulation of the main group metal center in a matrix
is conducive to u+ transport (and/or ideally, to electron transport) in
n to possessing mechanical flexibility; and also to the formation of
systems based on intermetallics.
. Capacities and volume changes of different elements.
material c AI Si Sn Bi
d phase LiC6 Li9Al4 Li21Si5 Li 17Sn4 Li3Bi
ical specific
372 2235 4010 959 385
y (Ah/kg)
ical volumetric
y (Ah/1) 833 6035 9340 7000 3773
changes(%) 12 238 297 257 115
Intermetallic Composites and Binary Phases

Comprised of Sn, Sb and AI
ultiphase electrode materials composed of MM'compounds, in which
ne component forms an alloy with lithium, exhibit better cycling
ance than a single phase host. This has been attributed to the
on of an "inactive" M' matrix that surrounds the "active alloying M
nd helps maintain the integrity of the microstructure of the composite
e during the (de-)alloying process. Much of this work has been
out using a transition metal as the inactive component and a main
metal as the active alloying centre: these are termed "intermetallics"
gh strictly, the name applies to Al compounds and other true metals.
ncept was based on the first "matrix" approach involving main group
xides, in which the LhO that is formed on reduction acts as the
matrix component. Another approach is to use intermetallics MM'
w strong structural relationships to their lithiated products.
hese are discussed in the following sections.
Tin Oxide Composites
TCO, a new anode material comprised of an amorphous, multi

composite oxide glass (SnxAlyBzPpOn), containing "active" Snn
inspired much interest in the area of main-group metal oxide
33
s. Studies on related compounds such as SnO, Sn02, and Sn2BP06
hed that the initial Li uptake results in an irreversible reaction to form
Chapter4
nd metallic Sn. The subsequent Li uptake that results in the reversible

y is based on alloying of the Li with the Sn centres, to ultimately form n
34 35
nanoparticles embedded in the LhO matrix. ' A similar mechanism
ought to be responsible for Li uptake in the tin-oxide composite glasses.
ed studies which followed, based on a variety of spectroscopic and
tion methods, revealed that the "Li-Sn" alloy nanoparticles are highly
38
36 37 The proximity of oxygen in
ered at the limit of deep discharge. '
'
trix (either on the bulk of the particles or within them) may give rise to
sorder - and in any case, certainly affects the thermodynamics of the
.
The reactions occurring within the tin composite glasses take place in a
miting regime, where the thermodynamics are strongly affected by the
36
urface energy of the particles." The reaction between Li and Sn, which
crystalline Li4.4Sn under high-temperature conditions can be much
omplex on a nanoscopic level. First, the kinetics of the system limit the
ion of ordered "bulk" phases at ambient temperature, even in tin oxide
but especially in the dilute tin-based glasses. Secondly, the tin clusters
d on Li uptake are in close proximity with both lithium and oxygen as a
of these particles having a very high surface area: volume ratio. We
suggested that partial oxygen incorporation within the bulk (as an
itial element) may be responsible for some of the disorder observed in
ray diffraction patterns of the fully discharged material, as well as
ally low frequency 7Li NMR shifts indicative of Li in the proximity of
36
duced Sn and oxygen. Oxygen is present as an interstitial component
n group compounds such as Zr5Sn30 and La5Sn30 and is implicated in
39
e of intermetallics. The possible interstitial, and surface oxygen can
te the reversibility of Li uptake by the Sn. The surface energy of these
es allows the "back-reaction" of lithium with oxygen to take place at a
potential than predicted from simple considerations that exclude surface
contributions.
37
This has been corroborated by Mossbauer studies, and EXAFS
36
, which give evidence of oxygen incorporation/disorder at low
al. The availability of matrix bound oxygen to participate (surface or
itial) will be determined by the surface energy of the electrochemically
particles, hence the smaller the aggregates, the more reactive they will
Other tin oxide composites (and negative electrodes in general) are

40
arized in a comprehensive review article by Tirado et al.
Inactive/Active Composites. Reduction-Decomposition/

Alloying
1. Tin compounds
order to reduce the irreversible capacity observed in the case of

it was concluded that oxygen bonded to the electrochemically active
41
should be avoided. This approach, first suggested by Besenhard and
42
s , has been applied to several systems based on tin intermetallics,
43 44
where M is an electrochemically inactive transition metal (M=Fe '
47 44 44
Mn • Co, ). The fundamental concept is simple. Li uptake/reduction
dissociation of the starting material into an active component that will
ith lithium to form Li-M alloys, and an inactive part, as described
ion 4.1:
(4.1)
he role of M is mainly to provide a matrix that will absorb the massive

changes that occur within the electrode upon the lithiation
ion)/delithiation (contraction) processes. It therefore maintains the
ical integrity between the particles and also with the current collector.
contact is a major source of capacity fade. The role of the inactive
also to improve the electronic conductivity of the composite electrode
fectively than the insulating LizO-based matrix formed in the case of
However, this is at the expense of inhibiting lithium transport, which
ke place upon grain boundaries. Upon charge (lithium removal), two
es can occur based on the degree of back-reaction (see Figure 4.1):
he delithiation of the LixSn alloys leads to metallic tin dispersed in a
ork composed of theM atoms (equation 4.2); or the starting material is
d (according to equation 4.3):
LixSn + MSn + M +xLi+ +xe (4.2)

LixSn + MM-Sn + xLi+ +x e- (4.3)
he first example of the application of this concept was in iron-tin

43
nds where iron forms the inactive matrix surrounding the tin atoms.
57
s of Sn2Fe prepared by high energy ball-milling were shown by Fe
uer studies to decompose upon lithiation to nanosized iron and
tin alloys, as described in Figure 4.1. Upon charge (delithiation), the
s aid lithium removal by reacting with Sn to re-form small grains of
However, due to the large volume expansion during reaction with
some tin regions gradually lose connection, leading to an incomplete
Chapter4
4.1. Dissociation reaction of lithium with tin-based intermetallic compounds M-Sn.
ation of the starting material. Such compounds exhibit high specific

ties of 800 mAhlg and 650 mAh/g during the first discharge and first
, respectively, but capacity fading on long-term cycling. As noted for
35
ide-based materials, an improved cycle life (associated with lower
ties) of such electrodes can be obtained by reducing the voltage window
en 0 and 0.55V vs. Lt/Li.
The effects of increasing the Fe:Sn ratio were subsequently investigated
same group. They found that increasing the Fe:Sn ratio from SnFe to
5 also resulted in dissociation. However, owing to the higher iron
t, the Fe atoms form an impenetrable "skin" on the surface of the grains
hibits complete reaction of lithium with the tin and does not allow
ete dissociation of the Sn-Fe compound. The more iron in the alloy, the
active is the material towards lithium.
A compromise between the higher reactivity/greater fading of the Sn
aterials and lower reactivity/lesser fading of the iron-rich materials was
by generating composites of Sn2Fe (active component) and SnFe3C
ve component). These materials, prepared by mechanical alloying,
better cycling stability than the pure Sn2Fe compound, and higher
ties than SnFe3C. Thus the initial goal of tin atoms embedded in an
chemically inactive iron matrix was not completely achieved, but the
obtained with the active/inactive composite electrode on the Fe/Sn/C
suggest that this concept is promising. Similar results were obtained
47
same group in the Sn/Mn/C system. Again, increasing the ratio of the
ve' atoms in the alloys prevents the reaction of lithium with tin and
the capacity of these materials.
Development of nanostructured SnMn 3C led to the concept of grain
ary storage of lithium. SnMn 3C has a perovskite structure that is
e toward lithium. However, nanosized SnMn 3C particles (12 nm)
sized by high energy ball milling provide a 'host' for the reversible
on of lithium that can "intercalate" in the spaces between the grains.
thors showed that lithium can reversibly react at and within the grain
ries, delivering 150 mAhlg with a good cycling stability. The

ck is the limited capacity obtainable by this approach. The concept,
r, may be also applicable to nanostructured anode materials as a
of additional Li capacity.
anocrystalline composite Mo 1_xSnx has been prepared as a thin film by
ring method. The films show solid solution behavior below x=0.45;
xistence of the above phase and nanocrystalline tin when x>0.45; and
of composition-modulated Mo1_xSnx and nanocrystalline tin when the
48
tin content is >40%. Relatively stable electrochemical cycling is
49
d with promising electrochemical properties.
.2. Antimonides
ater work continued with antimony-based intermetallic compounds.

to tin, antimony itself exhibits high capacities associated with alloying
50
s to form LhSb (theoretical capacity of 660 mAhlg). It also exhibits
e expansion/contraction problems on Li cycling (see Table 4.1 above),
refore poor capacity retention is improved by the use of an inactive
provided by a transition metal element to improve the cyclability.
most cases, the reaction of the MxSb compounds can be summarized
ariation of equation 4.1 above: namely simple dissociation of the
material into M and Sb to form LhSb alloys and metal particles. The
tion reaction seems independent of the stoichiometry or structure of
ting phase, as it is observed for all transition metal antimonides.
for example, crystallizes in the skutterudite structure, comprised of
hared CoSb6 octahedra (similar to the Re03 structure except with f
51
the octahedra; see Figure 4.2). During the first discharge the uptake
ium takes place at ca. 500 mV vs. Li+/Li, corresponding to a specific
52 53
of 800 mAhlg. ' The process leads to destruction of the crystalline
e and formation of Li3Sb alloys dispersed in a Co matrix, as confirmed
54
b Mossbauer spectroscopy. On charge, a large irreversibility of
hlg is partially caused by the formation of a SEI layer. Upon
ent cycles, the lithium reactions occur at higher voltages (-800 mV
2 V vs. Li+/Li for discharge and charge processes respectively) and are
to the formation/dissociation of lithium antimony alloys. Despite the
e of highly dispersed cobalt clusters within the electrode, the capacity
fades to 200 mAhlg after 20 cycles.
44
Similar dissociation behavior is also observed for MSb2 (M = Ti, V)
stallize in the 8-AhCu structure, and MSb2 (M = C , Fe ) that
5 56
57 58
ze in a marcasite-type structure. • The latter is comprised of comer
e joined MSb6 octahedra that form small channels (see Figure 4.2).
Chapter4
.2. CoP 3 skutterudite structure with comer-joined CoP6 octahedra (left) and FeP2
structure with edge-joined FeP6 octahedra (right).
imilar to CoSb3, the first reaction step (which takes place at 0.6 V and
vs. Li+/Li for CrSb2 and FeSb2 respectively), leads to a complete
ization of the structure and the formation of LhSb alloys as shown by
tterns (Figure 4.3).
3500
3000
2500
2000
c::1"' 1500
0
(.)
1000
500
0
50 60 70
10 20 30 40
20( 0
)
.3. XRD patterns of FeSbz (a): reference, (b): pnstme material synthesized by
g, (c): after 5 cycles stopped at the end of discharge and (d): after 5 cycles stopped at
charge. (*) and ( )represent Li3Sb and Sb peaks respectively.
ore stable cycling behavior was noted for FeSb2 than in CoSb3,
r as a capacity of 300 mAhlg can still be obtained after 30 cycles at a
e.56
eversing (increasing) the M:Sb ratio provides similar outcomes,
ified by Li uptake in Mn 2Sb which also gives simple dissociation of
59
ting material into Mn and Sb to form LhSb alloys (equation 4.1).
n be contrasted with the behavior of copper antimonides, such as the
tural Cu2Sb (see below), where the apparent limited region of Li-Sn
Cu solubility initially leads to ternary phases.
Intermetallic "Insertion" Hosts: Reduction/Insertion- Alloying
1. Cu-Sn and Cu-Sb Intermetallics
n contrast to the active/inactive concept, the copper/tin system was

proposed by Thackeray et al. as a potential intermetallic Li-insertion
60 61
., a host capable of topotactic Li intercalation. ' Such a compound
principle, exhibit high specific and volumetric capacities without the
ck of the massive volume expansion problems consecutive to an alloy
on. The mechanism of lithium insertion into TJ-Cu6Sn5, is now thought
r in two distinct steps which are a function of the voltage; in the first
u6Sn5 initially undergoes a phase transition involving Li insertion to
known LhCuSn-type structure. This transition is observed at 400 mV
i, corresponding to an idealized reaction that can be written as:
n fact, distinction between LhCuSn and "Liz.17CuSn0.83" is not made,

wo are probably isostructural. This process has been confirmed by in
62
RD studies, recent theoretical calculations using the FP-LAPW
63 119 64
s and by Sn Mossbauer spectroscopy. To explain the insertion
ism, Figure 4.4a describes the NiAs-type hexagonal structure of the
material TJ-Cu6Sn5 in which layers of tin atoms are sandwiched
sheets of copper atoms. During the lithiation, half of the tin atoms
laced to form columns of tin, while the copper and the other half of tin
emain spatially intact, creating small hexagonal channels in which
can be inserted to form Li2CuSn (Figure 4.4b).
his "topotactic" reaction is associated with a 61 % volume change.
he voltage is curtailed at 0 V during discharge, the decomposition of
ary LhCuSn phase occurs according to the equation:
Chapter4
b.
e• Copper 0 • Tin e• Lithium
4.4. Phase transformation of'YJ-Cu 6Sn5 during lithiation (a) to form "Li2CuSn" (b).
At deep discharge, LixSn alloys (i.e., Li4.4Sn) are ultimately formed, as

62
by in-situ XRD studies. The specific capacity corresponds to
Ahlg for a fully lithiated electrode. As observed for the Fe/Sn/C system,
partial reformation of the starting material occurs, and both copper
on and a large volume expansion lead to a poor cycling life of such
des.
The behavior of copper antimonide, Cu2Sb, is related to the above
s. Thackeray et al. demonstrated that it reacts to form LizCuSb
uctural with LizCuSn) before transforming to Li3Sb.65 The transition of
to LizCuSb takes place with extrusion of 50% of the Cu atoms, and a
structural change to yield the zinc blende phase of CuSb, and then
Sb. Further lithiation results in a Liz+xCuJ_xSb solid solution that
ates in the end member, LhSb at x=l. The reaction is highly reversible,
ing a capacity of 290 mAh/g that is 90% of the theoretical. The
ce of the extruded copper particles is thought to impart good electronic
ctivity to the electrode. Most important in this work, and that of InSb
Sb that follow, is the principle of the reversible lithium insertion/metal
ement reactions within a face-centered cubic (fcc) Sb host structure.
ore in the case of Cu2Sb, the absence of significant internal
ement of the fcc Sb during the transformation aids the reversibility.
hium and copper, of comparable sizes, can be considered to move in
t of the lattice with little disruption to the framework. This interesting
66
has been recently reviewed, and details are given elsewhere.
2. IhrSb (Where IM is an lntermetallic)
he same principle of strong structural relationships between parent and

compounds governs Li insertion into both InSb and SnSb to some as
66
noted by Thackeray et al. Of note is that both components in ases
can partake in lithium alloying reactions, but they occur ially.
he zinc blende structure of InSb has a favourable three-dimensional
ial space for lithium diffusion, and two crystallographic sites for Li
n, as described by the first authors who reported its uptake of
67
' The reaction of Li with InSb occurs in several steps which can be
classified as 1) a reaction of Li to form Liyin 1.ySb, accompanied by
n of In from the framework, and 2) a second step that involves
of the extruded In with Li to form Lixin. The precise details of the
hemical reaction during the initial stages are subject to controversy,
68
r. The original authors have concluded, on the basis of in-situ XRD
69
AFS studies, that Li is inserted into, and In extruded from, a fixed
framework to give Lix+yin 1.ySb (0=x<2, O<y<l). They propose that,
is inserted into the interstitial sites of the zinc blende structure to give
. Second, further reaction of lithium induces extrusion of In and
ment by Li to give Lix+y+zin 1.ySb + y In. The extent to which Li is
before In extrusion occurs is not yet accurately known, and is the
of the controversy. EXAFS measurements do not support early In
n, but other author's detailed XRD measurements suggest that In
70
n occurs at x=0.27. Irrespective of the details, both agree that
ly LhSb is formed at the end of the first stage, and that the last step
s formation of Lixin.
he good performance of the electrode (300 mAhlg after 25 cycles) has
ributed in part to the reversible lithium insertion/indium extrusion n
68
the fcc Sb host, as described above for the Cu-Sn compounds; i.e.
uctural similarity of InSb and Li3Sb. The small increase in
ographic volume of the electrode after full discharge (taking into
the extruded indium) is only 46.5%, suggesting that appropriate
ring of the materials might yield a system with good stability.
nately the extensive crystal growth of the extruded indium and
on of Lixln alloys compromises the reversibility of the electrochemical
Li3Sb transformation.
he reaction of Li with SnSb can be described in similar terms to that
g InSb. The reaction can be summarized as Li insertion into, and tin
ment from, an fcc Sb host structure.66 LhSb is formed after the first
Chapter4
nd the second step involves reaction of the extruded tin with Lito form
71 119 7
alloys. A recent study using Sn Mossbauer combined with Li MAS
spectroscopy confirmed the dissociation of the starting material during
rst discharge and also demonstrated that a partial recovery of the
ure occurs upon charge for SnSb single phase electrodes.The good
lity of this system can be attributed to the fact that 1) both components
equentially: LhSb formation at 700 mV vs. Li+!Li is followed by LixSn
41
ormation at lower voltages as shown in the first report on this system.
fore at each step of the reaction, the lithiated phase is embedded in an e
72
matrix; 2) the extruded tin and subsequent LixSn alloys are more ble
for reaction than in the case of InSb, since bulk tin crystallites are ntly
not formed on the surface. However, low voltage reaction ing
formation of tin alloys result in further electrode expansion and loss
erparticle contact between LhSb and Sn, causing irreversibility and
ty fading.
The best results were obtained for a Sn/SnSb mixture with a reported
73
-limited specific capacity of 360 mAh/g during 200 cycles. However,
materials exhibit a large irreversibility during the first cycle due to the
ion of a SEI layer and also because of residual impurities inherent in
nthesis process.74 '75
Synopsis. In summary, both tin and antimonide intermetallic

unds show high specific and volumetric capacities, but problems with
sibility and capacity fading. Different reaction mechanisms can be
ed depending on the composition. Where the materials are of type M
M-Sb (M =transition metal), inactive/active composites can be formed
by Li uptake. The transition metal that is extruded during reduction
s the inactive component, and the Sn or Sb participate in alloying
ns with the Li. These composites can also be prepared "by design" by
illing selected components. Alternatively, the binary intermetallic host
sert Li concomrnitant with extrusion of the inactive component, which
followed by alloying of Li with the "active" component that generates
nal capacity. When both components are "active", a combination of
on, extrusion and alloying occurs.
Despite experimental specific capacities often lower than for graphite,
lumetric capacities obtained with intermetallic compounds are usually
higher than the 830 Ah/1 observed for graphite (i.e., 4500 Ah/1 for
1650 Ah/1 for lj-Cu6Sn5) due to their high density (greater than
3 3
, 2.24 g/cm for graphite). Moreover, since 21 elements can form
with lithium, and up to 70 metals can form intermetallic compounds,
ncept offers some future prospects. However, composite electrodes
ed in situ during cycling are still insufficient in terms of cycle life of the
de. Such compounds need to be improved to replace graphitic materials
mercial batteries.
. Aluminides
lthough ternary intermetallic phases are known for Al-Cu-Li, such as

u, LhCuA16, and LiAlCu, AhCu is completely inactive towards Li
n. The reasons for this were addressed in a systematic study of A2B
unds that all have the 8-AhCu structure (Sb2Ti, Sb2V, Sn2Co Sn2Mn,
and AhCu). The authors concluded that even though thermodynamics
that the alloying reaction should proceed, the large activation energy to
break anAl-transition metal bond hinders reactivity.
Silicides
everal efforts have also been undertaken on silicon-based electrodes.
76 77 79
on has been given to Mg2Si, ' 78 ' ' but despite much effort its
ies have proven disappointing. More promising are silicon-based
ites prepared/designed by a variety of techniques.
MgzSi
any authors have investigated Mg2Si due to its anti-fluorite structure

was reported to be amenable to Li insertion. The reaction mechanism
ever, still unclear. Moriga et al. proposed a lithium insertion into the
sites, leading to the formation of a ternary phase LiMg2Si, followed by
76
i , along with extrusion of Mg atoms that will react to form Li/Mg
This process is in78 conflict
79 with the simple dissociation reaction
d by other groups, • followed by the formation of lithium alloys
g and Si. Despite a large initial specific capacity of 1370 mAhlg at an
voltage of 350 mV vs. Li+/Li, the mechanical instability of the
de causes a drastic fade of 80
the capacity upon cycling, even by reducing
age window of the study. Extended studies were also performed on
81 82
tural compounds, such as Mg2Sn and Mg2Ge , showing better
behavior (in the case of the tin compound) than Mg2Si with capacities
mAhlg after 60 cycles in an appropriate voltage window.
Si-Based Composites
s noted above in Section 4.2.1., Si displays the highest gravimetric

Chapter4
olumetric capacity of all of the Group IVN elements. Much attention

erefore been devoted to preparing composites of Si with various
unds/elements, by different methods that allow good dispersion of the
hin the matrix. Early work in this area devoted to C/Si composites d
chemical vapor deposition to produce highly dispersed Si clusters a
carbon matrix; these materials showed promising electrochemical
83
ior.
Ball-milling has since been used with varying success to prepare
sites of Si with both hard and soft materials. Mechanical milling of Si
itanium nitride, known for its conductivity, mechanical strength and
84
ss, has recently been reported. TiN plays the role of the inactive
nent, and also aids in preserving the morphology of the electrode. The
site obtained by high energy mechanical milling (HEMM) that exhibits st
electrochemical properties (Si:TiN ratio of 1:2), is comprised of
rystallites of TiN (5-7 nm) agglomerated with amorphous Si to generate
es a few hundred nanometers in diameter. It exhibits good cycling
ty over 20 cycles associated with a specific capacity of 300 mAhlg. This
ty is still lower than the theoretical of 776 mAh/g (assuming reaction to
Li21Si5), apparently owing to a large fraction of inaccessible Si buried
the TiN matrix. However, the electrode structure is relatively stable
cycling, as little cracking/crumbling was observed after 30 cycles.
-energy ball-milling has been employed to prepare composite
des with silicon and "soft" materials such as graphite. The Si/C
sites recently reported, namely C 1_xSix powders with Si contents
g from 10-20% made by milling crystalline Si and graphite, were
d to contain Si microencapsulated by a thin layer of amorphous carbon.
display reversible capacities of up to 1040 mAhlg which fade to 800
85
after 20 cycles. The increase in Si content in those composite
als is accompanied by the coexistence of inactive SiC during their
sis. Better materials were obtained using nanocrystalline Si, prepared
86
er-induced silane decomposition, ball-milled with carbon black.
were shown to exhibit high initial reversible capacities up to 2000
, but extensive fading over 20 cycles was still observed.
Composite electrodes can also be prepared by coating or sputtering
ds. The use of thermal vapour decomposition (TVD) has been recently o
87
obtain carbon coated graphite. The resultant materials display sed
rate capacities and decreased irreversibility by comparison to
aceous materials, where graphite layer exfoliation often limits
bility. Thermal vapor decomposition has been more recently extended
88
formation of Si/C composites. Composites containing 20% carbon
g exhibit excellent performance. Up to 800 mAh/g can be obtained
n excellent cyclability. This method also inhibits the decomposition of
lytes, which can be a problem on the surface of unprotected Si-
es, and furthermore provides a conductive network.

owever, in general (as noted above) obtaining stable cycling for alloy
e electrodes is an unattainable goal, unless the capacity is constrained
s near that of graphite. The reason for cycling instability is believed to
to inhomogeneous volume expansion in coexisting regions of phases
ferent lithium concentrations. Another approach to anode composites
ed in a patent application,89 is based on using amorphous metals and
ids. The advantage of amorphous materials lies in the potential
tion of the two-phase regions between phases of different lithium
ration, which can lead to homogeneous volume expansions and
ed charge-discharge cycling behavior. Such materials, based on
ous Si/Sn alloys have been prepared using a magnetron sputtering
. 90 The authors showed through electrochemial and in-situ XRD
ents that the materials remain amorphous at all times during cycling,
t the two local environments for Si and Sn are distributed evenly
out the bulk. These are believed to be the reasons that the materials
stain large volume expansions without drastic cracking and
ation. These materials were found to readily uptake up to 4 lithium
xSnx formula when cycled to 0 V vs. Li+/Li, leading to a specific
of 3000 mAh/g for the composition .x=0.33.
hese materials, and the new Si-carbon composites seem amongst the
romising as negative electrodes in lithium-ion batteries, although
ion of large-scale amounts of materials by sputtering technique may
hallenging.
METAL OXIDES
etal oxide anode materials can be divided into two classes. One is the
ertion compounds: metal oxides capable of intercalation chemistry at
ential. These materials are characterized by relatively high potentials
between 1.4-1.8 V), low lithium capacities and often highly
le insertion reactions. The other class of metal oxides display Li
es derived from reaction with the lithium to form another material,
s usually the reduced metal oxide encapsulated in a matrix of LhO,
to the case of the tin oxides described above. These materials also
a relatively high average charge potential, but also a large hysteresis
nt with the necessity of oxygen transport within the bulk during
n. Critical to the functioning of these materials, therefore, is the
mixture at the nanoscale regime, of metal sub-oxide (or pure metal,
ng on the degree of Li uptake) and oxygen from the lithium oxide
hat facilitates the oxygen recovery process.
Chapter4
. Transition Metal Oxides: Where Insertion Governs the

Process
As these are not strictly "composite" anode materials they will be
oned only briefly.
The most widely considered metal oxide for low potential applications
Ti5012, described as a zero-strain insertion material owing to the almost
ible changes in the unit cell on insertion of up to one lithium into the
91
ure. The formula can be written as Li[Li 113Ti513]04, illustrating the
g of Li and Ti on the octahedral 16d sites, and occupation of the
ning Li on the tetrahedral 8a sites. To account for the virtually zero
es in volume on Li insertion, Ohzuku et al. have suggested that Li is
d into the unoccupied 16c sites, which are also additionally occupied
to migration of Li from the tetrahedral sites. The combination of little
ral hysteresis and rapid lithium diffusion result in good electrochemical
ties for this material, despite an average insertion potential of 1.5 V.
ally, capacity values close to 150 mAh/g are observed. Materials
ting such capacities at very high rates can be obtained from in high
e area modifications, such as in nanocrystalline films made by sol-gel
92
ds or in commercial nanomaterials prepared by spray drying or other
ques. Li[CrTi]04, with a related structure, also shows similar
93
teristics.
The other oxide that has received attention is Ti02 as the anatase
94
orph, and particularly as a nanocrystalline material. Many reports
been devoted to preparing nanostructured anatase, with techniques
ng the gamut of spray drying, sol-gel chemistry, and the polyol
95
d. A synthetic polymorph of Ti02 produced by "chimie douce"
96
ds, known as Ti02(B), shows slightly better behavior than Ti02
e. All these material suffer, however, from a higher insertion potential
97
e spinel Li4Ti5012, and less efficient lithium mobility.
Transition Metal Oxides: Where Reduction of Metal

Oxide to Metallic or Semi-metal Occurs
The first reports of a metal oxide being subjected to deep discharge

ed lithium insertion in a-Fe203 or Co304, which were reported to lead
extrusion of the transition metals and formation of Liz0.98 For the iron
this occurs in a series of stages involving progressive formation of
d spinel structures, with reversibility implied in some of the steps.
uent studies on metal oxides concentrated on "complex" oxides,
ng LiMV04 (M = Zn, Cd, Ni),99 LixMO, and LixMyV 1_pz (M =
100 101
on metal) and amorphous or semi-crystalline RV04 (R =In, Fe),
102 103
(M =Fe, Mn, Co), and AxMo03•
hen these materials were first reported, the mechanism responsible
Li uptake and reversibility was not understood; however great strides
nce been made. A beneficial mix of metal and metal-oxide on the
le appears to be responsible for promising properties exhibited by a
g number of transition metal oxides. The initial work on Li uptake in
y reduced molybdenum
7 oxides such as Li0.25Mo03 showed by means
ombination of Li NMR, XRD and XAS probes, that a complex
ous nanocomposite is formed at low potential. It consists of a
red, highly oxygen deficient lithium/molybdenum-oxide (a "lithium/
enum sub-oxide") in intimate association with lithium oxide. 103 The
ccounts for the irreversible capacity observed by electrochemical
Subsequent EXAFS and XANES studies revealed that although
Mo is formed on deep discharge, a molybdenum oxide is re
104 105
ed on oxidation. ' The depth of discharge is important in
ing the metal-oxygen environment. At potentials deep enough to result
plete reduction of the metal, the molybdenum retains most of its
coordination environment. The latter is only lost at repeated discharge
low potential. Thus, if the voltage window is restricted above this
the Mo essentially shuttles between oxygen poor/lithium rich
ge) and oxygen rich/lithium poor (charge) amorphous phases with
ed cyclability being the result. Complementary findings in Fe and Co
e systems on Li uptake have also been revealed by in-situ X-ray,
106
, Mossbauer and microscopy studies. The recently reported
r of iron borates such as Fe3B06 and Fe3B03 as low-potential Li
107
materials is also based on similar principles.
hese findings led to investigation of simple nano-sized metal oxides
ed on the rock-salt structure (M = Co, Fe, Ni). High-resolution TEM
of the material at deep discharge demonstrated the presence of metal
ticles about 20 A in diameter. Again, the highly divided, high-surface
nature of the nanoparticles facilitates the back-reaction with oxygen
e lithium oxide matrix
108 to reform the metal oxide on charge, this time
i-crystalline form. The reaction, particularly reversible for Co, but
for Fe and Ni, can be simplified as reaction between Li and MO to
O + M, which is reversible if the materials are on the nanoscale.
Chapter4
Mixed Transition Metal Oxides - Where Both

Reduction and Alloying are Responsible for Li
Capacity
The Sn-oxide, transition metal oxide, and Sn/Fe intermetallic studies

above opened avenues to investigate oxide materials containing both
ion metal and main-group metal active centers. An interesting class of
als based on the inherently open architecture of CaFe204 are suitable
109
ates. This structure exhibits partial solid solution based on Sn
ution for Fe with accompanying removal of Ca for Li, the end member
the stoichiometry Li0.6Ca0.4Sn0.6Fe .404, and relatively high ionic
ctivity. On introducing Sn into the structure the reversible capacity is
ntially increased compared with the parent material, owing to the
nal reaction of the Sn. Although irreversible capacity is observed that
from the partial formation of LhO, reversible Li uptake is observed
ood stability. The materials sustain a stable reversible capacity of >600
105
for the optimized composition Li0.5Cao.5Sn0.5Fe1.504• The profile of
ferential capacity plots suggest there is no complete phase separation to
alloy phases on reduction, but that a lithium-rich, oxygen deficient
oxide matrix is formed. These results are also supported by XAS
rements which suggest that Sn-Fe alloy nanophases formed at
ediate potentials lie in close proximity with oxygen.
NITRIDES AND PHOSPHIDES
Nitrides
By contrast to metal oxides, where the high degree of electronegativity

nce between the metal and the anion tends to drive the reduction to the
ion of the metal on Li uptake, nitrides display more covalent bonding n
the metal and anion. Different behavior is expected - and observed.
cases, the mechanism of Li uptake/extraction is still subject to scrutiny.
ous binary, and ternary (Li-containing) nitrides are summarized in an
110
ehensive review by Gregory et a/. but most have not been found to be
chemically active.
The first examination of the electrochemical behavior of metal nitrides
111
ased on layered LiMoN2, isostructural to LiCo02• Despite this
ing similarity in structure to a known intercalation compound, LiMoN2
to show any electrochemical activity although chemical extraction of
was achievable using butyl-lithium. Subsequent studies involved
s of the first row transition metals. These Li-M-N structures follow a
ble trend. For Ti through Fe, an antifluorite structure is favored with

112
M in the tetrahedral sites, M being present in a high oxidation state.
13 114
nitrides in this class such as LhMnN/ and LhFeN2 display
hemical behavior at about 1.2 V vs. lithium, with reversible capacities
ange of 300-350 mAhlg. For Co, Ni, and Cu, a structure based on a
formed, with M + substituting for u+ between the hexagonal LizN
1
These Liz(Lh.xMx)N compounds (M = Mn---+ Cu) are more promising.

ered morphology is advantageous, supporting fast Li-ion conduction
transition metal oxidation states. The first studies on Liz(Lh-xMx)N
nds (M =Co, Ni or Cu) suggested that the Co member showed the
capacity (480 mAhlg) and most reversible behavior. The significantly
performance of the Ni and Cu members has not been explained. Other
by Shodai reported that Liz.6Co0.4N can exhibit a capacity up to 760
in the voltage window 0-1.4 V. 115
wing to the high cost and toxicity of cobalt compounds, a better
l would be based on Fe or Mn. Both are unique in their behavior
ies between the two structural extremes described above. For example,
s both the antifluorite phase with iron in a high formal +3 oxidation
nd layered phases with iron in lower oxidation states. Of the latter,
ports were of single crystals isolated as minor phases from mixtures,
y indicated the possibilities of intriguing chemistry. A "metastable"
2
phase Li eN2 contains layers of LizN with ordered Fe + and
116 1
es between the layers. Crystals of Liz(Li0.37Fe0.67)N (Fe +) have been
from a mixture formed from decomposition of the antifluorite phase
117
ium flux. We devised a synthesis route to a novel lower iron-content
layered Liz(Li0.7Fe0.3)N, that provides virtually gram-scale quantities
yield within a day, by simple reaction of Fe with LhN at 800°C. 118
aterial is isostructural with the layered Liz.6Co0.4N analogue.
hemical studies showed that Liz(Li0.7Fe0.3)N displays a capacity of up
mAhlg in a potential window of 0 ---+ 1.3 V, with relatively good
on cycling.
oth the antifluorite and higher capacity layered nitrides, however, are
to three concerns: 1) They are highly hygroscopic and moisture
e. 2) They become amorphous on lithium extraction, making it
to ascertain the mechanism responsible for reversible Li
on/uptake. Studies to determine the mechanism responsible for
ization and Li uptake suggest that reversible migration of the N atoms
119
uring charge and discharge. 3) They need to be "charged" first by
oval of lithium from the structure, and hence they act as a reservoir for
not a sink. This results in problems in finding a suitable cathode
, as most of the preferred materials exist as lithiated air-stable phases.
these concerns, Lh.6Co0.4N has been recently re-examined by groups
120
sonic/Matsushita and proposed as an anode versus LiMn204 in
Chapter4
121
-ion cells. In recent work, the material has also been coupled with
122
ntaining anode materials to compensate for irreversibility in the latter.
The electrochemical behavior of main group nitrides capable of acting
123
i "sink" have been recently described by the Telcordia group. They
that uptake of Li by Zn 3N 2 results in formation of LiZn in a matrix of
. On charge, complex reactions result in the oxidation of LiZn to
ic Zn, and reaction of -LhN with LiZn to form LiZnN which is the end
er phase for all subsequent cycles. This first-reported reversible LhN
rsion gives a reversible capacity of 555 mAh/g but the cycle life was
d.
Phosphides
Like nitrides in some respects and antimonides in others, phosphides

recently been shown to exhibit interesting chemistry that is a
nation of the other Group V members. In common is the fact that the
ying mixed anion-metal bands, and a high degree of electron
lization lead to a low formal oxidation state of the metal, and strong
nt character of the M-pnictogen bond. 124' 53 The materials therefore
an overall lower intercalation potential compared to the respective
. Differences within Group V compounds arise from the fact that
ides display more covalent bonding than the corresponding nitrides, s
than the predominantly semi-metallic antimonides. These bonding
nces are reflected in the band structure and chemistry, as shown in
4.5.
Oxide
Density of states Density of states Density of states
.5. Comparison of DOS for oxides, nitrides and phosphides.
n nitrides, the uppermost band is thought to expose nitrogen-based

s and hence result in holes appearing in the nitrogen-dominant band
xidation (and evolution of N 2 on overcharge). 120 In antimonides by
t the redox reactions involve the metal-Sb bond, and hence Li
reduction results in phase separation to form metal and LixSb alloys.

sphides, the chemistry seems to lie midway between the two.
nces in the strength of the M-pnictogen (PN) bond, and relative
ution of the PwPN bond results in a paucity of transition metal-poor
nitrides (but plethora of transition metal antimonides). For example,
analogous to MnP4, does not exist. As no binary nitrides are known
n uptake u+te·, the lithium ions must first be removed from the
e to give reversible cycling, a drawback of these materials that also
in irreversible collapse to an amorphous structure. More importantly,
ative electrode in a cell should act as a reservoir for Lias MnP4 does,
urce. In this respect, phosphides have advantages over nitrides.
he electrochemistry that is responsible for reversible Li uptake varies
ng to element, and as to whether a lithiated ternary phase exists in the
iagram with the same M:P stoichiometry as the binary phase. Unlike
monides, the mechanism is dependent on the stoichiometry/structure
starting phase.The behavior exhibited can be separated into four
those in which the binary and ternary phases exist i.e.,MP <->LiMP;
which only either the binary or ternary phase exist, i.e., MP (FeP2) or
Li2CuP); and those in which Li uptake results in phase decomposition
etallic state, as in CoP3. These are discussed below.
Binary (MP <·> LiMP) Combination
here a fully lithiated end member exists, conversion between the

nd ternary crystalline phases can occur on lithium insertion. Such a
is exhibited by MnP4, which undergoes a first-order phase transition
125
crystalline LhMnP4• The mechanism by which this occurs is more
x than a typical intercalation reaction, however, as substantial
al differences exist between the two phases. The layered 8-MnP4
e is composed of infinite sheets of MnP6 octahedra linked by P-P
oth within the sheets, and between the sheets. LhMnP4 has the
rite structure. On u+te· uptake, P-P bonds between the layers are
ely cleaved (P"
1
p·\ and Li ions are inserted in the resulting
ellar voids. Translation of Mn from an octahedral to a tetrahedral site
ach coordination rhomboid provides space for additional tetrahedral
ial sites for u+, to give the antifluorite structure. A scheme illustrating
posed mechanism is shown below. Importantly, the process can be
reversed by reversing the voltage, resulting in "electrochemical
llization" of the original MnP 4 phase as a crystalline compound. The
ds in the MnP4 structure that are cleaved on Li insertion (reduction) to
MnP4, are reformed on re-oxidation to MnP4. Thereby, they act as an
storage reservoir.
Chapter4
The reaction is a quasi-topotactic intercalation process that represents

st unequivocal example of reversible lithium insertion of lithium at low
ial, aside from graphite. There are obvious differences, however. Li
on in graphite cleaves only the very weak van der Waals bonding
en the graphene layers, and is hence a facile process, whereas the
required to cleave and re-form the covalent P-P bonds is significantly
. Contrary to the pure intercalation process in graphite that can
ore proceed with little hysteresis, the phosphide reaction is kinetically
d by the structural rearrangement involved in the phase transition. The
ntial volume/density change also seen in other high-capacity anodes
esults in some cracking and loss of contact with the current collector.
reversible capacities (700 mAhlg) than theoretical (1050 mAhlg) are
ed on the first cycle, that fade on cycling.
Nonetheless, reversible capacities of 550 mAh/g, stable over >25
, can be sustained at C/4 rates in nano-sized particles formed by ball
126
g. Interestingly, Li can also be de-inserted from crystalline Li7MnP4
e similar electrochemical results. However, the MnP4 formed on
ion is very poorly crystalline (almost amorphous), as is the Li7MnP4
uently formed on re-insertion. This is presumably due to the fact that
nt ordering of Li and Mn is implied on insertion of Li in MnP4. This
be duplicated by starting in the opposite direction, due to the perfectly
m distribution of Li and Mn in LhMnP4.
Li de-insertion from Li7VP4 has also been recently reported to give
127
ible capacities of 550 mAh/g, on the first cycle at a C/34 rate. They
unusual stability of the VP4 framework to different lithium
sitions, based on the fact that de-insertion of Li appeared to result in no
ral change to the lattice. The authors, however, allowed for the
ility that the oxidized phase could be amorphous.
. Binary MP; No Stable LiMP Ternary
Where a binary MP exists but no stable LiMP ternary phase of the same
oichiometry, two possibilities for Li uptake can be predicted. The
n can either yield a metastable ternary phase, such has been proposed
hium insertion in FeP2; or decomposition to yield Li 3P and metal can
akin to the reaction in antimonides. The latter is exhibited by CoP3• In
ases, Li uptake (and subsequent extraction) results in an amorphous
al, making the nature of the Li uptake process difficult to discern.
1. FeP2
he reaction of FeP2 with Li occurs with a remarkably low degree of

128
ibility on the charge cycle as shown in Figure 4.6.
2.5
+ 2.0
_J
rn
> 1.5
Q)
Cl
.l!! 1.0
g 0.5
0.0+- .- .--.----,r---"----r ---. ----r -i

0 2 3 4 5 6 7
x in Li.FeP2
6. Voltage vs composition for FeP2 cycled at a C/4 rate.
he process of Li insertion has been investigated using electrochemical

in concert with X-ray diffraction, XAS and magnetic measurements.
take of six Li during the first discharge leads to a specific capacity of
Ah/g, and an amorphous phase retaining substantial Fe-P bonding. It
s superparamagnetic behavior and an extremely low transition
ture below lOK. Upon charge, 5.5 lithium can be extracted from this
ble ternary "Li-Fe-P" phase, leading to a reversible capacity of 1250
56
To account for the presence of Fe-P bonding observed in the XAS -
experiments, the magnetic behavior of the particles on discharge, and
identifiable LhP, it was proposed that Li uptake occurs by expansion
hosphorus lattice, and incorporation of the additional lithium to form a
red, stuffed metastable "Li 3Fe05P" lattice that still maintains Fe within
cture. The low (ca. 10-15%) irreversibility exhibited by FeP2 makes it
esting material in the search for new negative electrode materials.
herefore, like in MnP4, the high degree of covalent bonding and low
of the metal induces a transformation more akin to "insertion", in to
the iron antimonide, FeSb2, where complete phase separation into um
antimonide and the metal occurs as described in Section 4.2.1.2.2
other metal phosphides, the possibility of a transition between a
LiMP phase and a metastable MP phase could be envisioned. Such a
mation appears responsible for the reversible de-insertion from
from Li2CuP to give "CuP" recently discovered in our lab.
Chapter4
2.2. CoP3
Phase separation of metal phosphides into metal and Li3P is observed

e phosphide, CoP3, however, as recently described by two different
129 130
ch groups. · Their conclusions are similar, although they differ on
arge process. The presence of Co particles is not apparent in the XRD
n of reduced CoP3 due to their small dimensions (well below the
ening limit of 500 A in XRD) but their presence was indirectly
sted by the clear presence of LhP in the diffraction pattern. TEM
rements revealed the presence of 10-20 A nanoparticles embedded
the original crystallite. As the SAED pattern showed only broad, weak
with d-spacings corresponding to LhP, and no interplanar spacings
able to cobalt crystallites, it was concluded the metal must be very
dispersed; perhaps
129 weakly bonded to the matrix as highly metal-rich
hide clusters. The metallic state of Co was revealed in XAS
rement of Alcantara et al.,130 however, who also showed the presence of
hrough MAS-NMR studies.
The process on charge appears to involve redox reactions of LhP, but
ticipation of the cobalt. The same mechanims as in the antimonides
ggested by Alcantara et al., involving a reaction based on formation/re
tion of LhP from its elements, Li + P. A reversible capacity of 487
was observed in this case. A slightly different mechanism was
129
ed by Pralong et al., in which cycling (400 mAhlg) was attributed to
nsformation between LhP and LiP on the basis of XPS measurements.
h cases the observed capacities are well below the theoretical.
p sis. The behavior of the phosphides represents a departure from

that ted by ionic negative electrode materials such as oxides (where the
high of the metal-localized bands drives the reduction to the metallic
state uptake); nitrides; and intermetallic negative electrode materials
where
ake relies on formation of alloys LixM (M = Sb, Sn). Furthermore,
covalent bond rearrangment and atom migration that lead to
chemical "recrystallization" (MnP4) or facile introconversion of
hous phases (FeP2) are a signature of these materials. New
unities are hence afforded by some phosphides, if difficulties with
lyte reactivity/SEI formation can be overcome which currently limit g
stability.
CONCLUSIONS
The task of finding alternate materials for anodes in lithium-ion cells to
ceous materials has proven to be a daunting and challenging task that

nned more than a decade of work amongst many research groups.
erview has attempted to summarize some of the major directions that
ominated the field, both in anode materials in general and in
ites in particular. It is clear that the goal of developing materials
of insertion/uptake of large proportions of lithium is incompatible
e difficulties of the large volume changes necessitated by such
s, and resultant capacity fading. However, clever approaches can be
ameliorate the problem, many of which lie in the composite approach.
sites-by-design (such as carbon coated Si materials) and the concept of
nactive materials provide good possibilities for the future, along with
synthesis of intermetallic compounds, nitrides, phosphides and
that may demand a composite (and/or combinatorial) approach to
e their electrochemical performance. Alternatively, designed
hes to nanostructured alternatives of these materials are needed.
anding of the mechanism of lithium uptake is critical to advancing s
in the area. Finally, attention must be given to reactions of these
s with electrolytes at low potential which can compromise
hemical performance. SEI formation with materials aside from carbon
ell understood, and the liquid electrolytes today that are optimized for
may prove less than ideal for other materials.
RENCES
roussely, J. Power Sources 82 (1999)140.

alkanski, Sol. Energy Mater. Sol. Cells 62 (2000) 21.
erada, T. Yanagi, S. Arai, M. Yoshikawa, K. Ohta, N. Nakajima, A. Yanai, N. Arai, J.
er Sources 100 (2001) 80.
. Abraham, D. M. Pasquariello, F. J. Martin, J. Electrochem. Soc. 133 (1986) 661.
Jacobson, R. R. Chianelli, M.S. Whittingham, J. Electrochem. Soc. 126 (1979) 2277.
azzari, B. Scrosati, J. Electrochem. Soc. 127 (1980) 773.
agaura, K. Tozawa, Prog. Batt. Solar Cells 9 (1990) 209.
stoia, (Ed.), Lithium Batteries, New York: Elsevier,1994.
inter, J. 0. Besenhard, M. E. Spahr, P. Novak, Adv. Mater. 10 (1998) 725.
rosati, Electrochim. Acta 45 (2000) 2461.
akihara, Mater. Sci. Eng. R33 (2001) 109.
Tarascon, M. Armand, Nature 414 (2001) 359.
chollhom, A. Payer, Angew. Chern. Int. Ed. 24 (1985) 67.
gumi, M. Inaba, Bull. Chern. Soc. Jpn 71 (1998) 521.
inter, J. 0. Besenhard, M. E. Spahr, P. Novak, Adv. Mater.lO (1998) 725.
led, J. Electrochem. Soc. 126 (1979) 2047.
led, in Handbook of Battery Materials J. 0 Besenhard Ed., Wiley-VCH Weinheim,
.
Dahn, T. Zheng, Y. Liu, J. S. Xue, Science 270 (1995) 590.
lcantara, J. M. Jimenez, P. Lavela, J. Morales, J. L. Tirado, Mater. Sci. Eng. B 39
Chapter4
996) 216.
Alcantara, J. M. Jimenez-Mateos, J. L. Tirado, J. ElectrochemSoc. 149 (2002) 201.
R. Dahn, W. Xing, Y. Gao, Carbon 35 (1997) 825
Tatsumi, N. !washita, H. Sakaebe, H. Shioyama, S. Higuchi, A. Mabuchi, H. Fujimoto,
Electrochem. Soc. 142 (1995) 716.
lnaba, H. Yoshida, Z. Ogumi, J. Electrochem. Soc. 143 (1996) 2572.
Alcantara, F. J. Femandez-Madrigal, P. Lavela, J. L. Tirado, J. M. Jimenez-Mateos, C.
mez De Salazar, R. Stoyanova, E. Zhecheva, Carbon 38 (2000) 1031.
lnaba, Z. Siroma, A. Funabiki, Z. Ogumi, T. Abe, Y. Mizutani, M. Asano, Langmuir
(1996) 1535; M. Inaba, Z. Siroma, Y. Kawatate, A. Funabiki, Z. Ogumi, J. Power
urces 68 (1997) 221; S. K. Jeong, M. Inaba, T. Abe, Z. Ogumi, J. Electrochem. Soc.
8 (2001)A989.
J. Wen, B. A. Boukamp, R. A. Huggins, W. Weppner, J. Electrochem. Soc. 126 (1979)
58; Y. Hamon, T. Brousse, F. Jousse, P. Topart, P. Buvat, D. M. Schleich, J. Power
urces 97 (2001) 185.
.S. Foster, C. E. Crouthamel, S. E. Wood, J. Phys. Chern. 70 (1966) 3042.
J. Wen, R. A. Huggins, J. Electrochem. Soc. 128 (1981)1181.
. Courtney, J.S. Tse, 0. Mao, J. Hafner, J. R. Dahn, Phys. Rev. B 58 (1998) 15583.
Chouvin, J. Olivier-Fourcade, J.C. Jumas, B. Simon, 0. Godiveau, Chern. Phys. Lett.
8 (1999) 413; R. A. Dunlap, D. A. Small, D. D. MacNeil, M. N. Obrovac, J. R. Dahn,
lloy Compd. 289 (1999) 135.
J. Weydanz, M. Wohlfahrt-Mehrens, R. A. Huggins, J. Power Sources 81-82, (1999)
7.
R. Goward, N.J. Taylor, D. C. S. Souza, L. F. Nazar J. Alloy Compd 329 (2001) 82.
Iodata, T. Kubota, A. Matsufuji, Y. Maekawa, T. Miyasaka, Science 276 (1997) 1395.
A. Courtney, J. R. Dahn, J. Electrochem. Soc. 144 (1997) 2045.
A. Courtney, J. R. Dahn, J. Electrochem. Soc. 144 (1997) 2943.
R. Goward, F. Leroux, W. P. Power, G. Ouvrard, T. Egami, W. Dmowski, L. F. Nazar,
ctrochem. Solid-State Lett. 2 (1999) 367; G. R. Goward, L. F. Nazar, W. P. Power, J. ter.
Chern. 10 (2000) 1.
Chouvin, C. Branci, J. Sarradin, J. Olivier-Fourcade, J. C. Jumas, B. Simon, Ph.
nsan, J. Power Sources 81 (1999) 277; J. Chouvin, J. Olivier-Fourcade, J. C. Jumas, B.
mon, 0. Godiveau, Chern. Phys. Lett. 308 (1999) 413; A. Hightower, P. Delcroix, G. Le
er, C-KK. Huang, B. V. Ratnakumar, C. C. Ahn, B. Fultz, J. Electrochem. Soc. 147
00) 1.
A. Courtney, J. S. Tse, 0. Mao, J. Hafner, J. R. Dahn, Phys Rev. B. 58 (1998) 15583.
. Corbett, E. Garcia, A. M. Guloy,W-M. Humg, Y-U. Kwon E. A. Leon-Escarnilla,
ern. Mater. 10 (1998) 2824.
. Tirado, Mat. Sci. Eng. R40 (2003) 103.
Yang, M. Winter, J. 0. Besenhard, Solid State lonics 90 (1996) 281.
A. Huggins, B. A. Boukamp, 1984 U.S. Patent 4,436,796.
Mao, R. A. Dunlap, J. R. Dahn, J. Electrochem. Soc. 146 (1999) 405; J. R. Dahn, ibid
6 (1999) 414; 0. Mao, J. R. Dahn, ibid., 146 (1999) 423
Larcher, L. Y. Beaulieu, 0. Mao, A. E. George, J. R. Dahn, J. Electrochem. Soc. 147
00) 1703.
Crosnier, T. Brousse, X. Devaux, P. Fragnaud, D. M. Schleich, J.Power Sources
(2001) 169.
M. Ehrlich, C. Durand, X. Chen, T. A. Hugener, F. Spiess, S. L. Suib, J. Electrochem.
. 2000 147 (2000) 886.
Y. Beaulieu, J. R. Dahn, J. Electrochem. Soc. 147 (2000)3237; L. Beaulieu, D.
rcher, R. A. Dunlap, J. R. Dahn, J. Alloys Compd. 297 (2000) 122.
. Dahn, R. L. Turner, 0. Mao, R. A. Dunlap, A. E. George, M.M. Buckett, D. J.
Clure, L. J. Krause, Thin Solid Films 408 (2002) 111.
. Turner, D. J. McClure, L. J. Krause, M. M. Buckett, J. R. Dahn, 0. Mao, Patent

perative Treaty Application 1999 W099/49532.
ang, I. D. Raistrick, R. A. Huggins, J. Electrochem. Soc.133 (1986) 457.
jekshus, D. G. Nicholson, T. Rakke, Acta. Chern. Scand. 27 (1973) 1307.
Alcantara, F. J. Fernadez-Madrigal, P. Lavela, J. L.Tirado, J.C. Jumas, J. Olivier
rcade, J. Mater. Chern. 9 (1999) 2517.
efebvre-Devos, M. Lassalle, X. Wallart, J. Olivier-Fourcade, L. Monconduit, J. C.
as, Phys. Rev. B.63 (2001) 125110
ldon, A. Garcia, J. Olivier-Fourcade, J. C. Jumas, F. Femindez-Madrigal, P. Lavela,
erez Vicente, J. L. Tirado, IMLB 2002, Monterrey, USA, Abstract #255.
Fernandez-Madrigal, P. Lavela, C. Perez-Vicente, J. L.Tirado, J. Electroanal. Chern.
2001) 205.
rosnier, D. C. S. Souza, J. E. Greedan, L. F. Nazar, Materials Research Society
osium Dec 2-6, 2002, Abstract EE8.3. p 683
jekshus, P. G. Peterzens, T. Rakke, A. F. Andresen, Acta. Chem.Scand. 33 (1979)
olseth, A. Kjekshus, Acta. Chern. Scand. 23 (1969) 3043.

. L. Fransson, J. T. Vaughey, K. Edstrom, M. M. Thackeray, J.Electrochem. Soc. 150
3) A86.
. Kepler, J. T. Vaughey, M. M. Thackeray, Electrochem. Solid-State Lett. 2 (1999)
. Wang, L. Sun, D. H. Bradhurst, S. H. Dou, H. K. Liu, J. Alloys and Compounds

(2000) Ll2.
archer, L. Y. Beaulieu, D. D. MacNeil, J. R. Dahn, J. Electrochem. Soc.147 (2000)
8.
harma, L. Fransson, E. Sjostedt, L. Nordstrom, B. Johansson, K.Edstrom, J.
rochem. Soc. 150 (2003) A330.
ransson, E. Nordstrom, K. Edstrom, L. Haggstrom, M. M. Thackeray, J. Electrochem.
149 (2002) A736.
. L. Fransson, J. T. Vaughey, R. Benedek, K. Edstrom, J. 0. Thomas, M. M.
keray, Electrochem. Comm. 3 (2001) 317.
M. Thackeray, J. T. Vaughey, C. S. Johnson, A. J. Kropf, R. Benedek, L. M. L.
sson, K. Edstrom, Journal of Power Sources 113 (2003) 124.
Vaughey, J. O'Hara, M. M. Thackeray, Electrochem. Sol. St. Lett. 3 (2000) 13.
. Johnson, J. T. Vaughey, M. M. Thackeray, T.Sarakonsri, S. A.Hackeney, L.
sson, K. Edstrom, J. 0. Thomas, Electrochem. Comm. 2 (2000) 595.
Kropf, H. Tostman, C. S. Johnson, J. T. Vaughey, M. M. Thackeray, Electrochem.
m. 3 (2001) 244.
. Hewitt, L. Y. Beaulieu, J. R. Dahn, J. Electrochem. Soc.148 (2001) A402.
Fernandez-Madrigal, P. Lavela, C. Perez-Vicente, J. L.Tirado, J. C. Jumas, J. Olivier
cade, Chern. Mater. 14 (2002) 2962.
Wachtler, M. Winter, J. 0. Besenhard, J. Power Sources 105 (2002) 160.
ng, M. Wachtler, M. Winter, J. 0. Besenhard, Electrochem. Solid-State Lett. 2
9)161.
Wachtler, J. 0. Besenhard, M. Winter, J. Power Sources 94 (2001) 189.
i, L. Shi, W. Lu, X. Huang, L. Chen, J. Electrochem. Soc. 148 (2001) A915.
nani, R. A. Huggins, J. Power Sources 38 (1992) 351.
T. Moriga, K. Watanabe, D. Tsuji, S.Massaki, I. Nakabayiasho, J. Sol. St. Chern, 153
0) 386
ntos-Pena, T. Brousse, D. M. Schleich, lonics 6 (2000)133.
im, J. Choi, H.-J. Sohn, T. Kang, J. Electrochem. Soc. 146 (1999) 4401.
. Oberts, E. J. Cairns, J.A. Reimer, J. Power Sources 110 (2002) 424; S-W. Song, K.
riebel, R. P. Reade, G. A. Roberts, E. J. Cairns, J. Electrochem. Soc. 150 (2003) A121
Chapter4
Kim, Y-J. Kim, D.G. Kim, H-J. Sohn, T. Kang, Solid State Ionics 144 (2001) 41
Sakaguchi, H. Honda, T. Esaka, J. Power Sources 81 (1999) 229
M. Wilson, B. M. Way, J. R. Dahn, T. van Burren, J. Appl. Phys. 77 (1995) 2363.
Kim, P. N. Kumta, G. E. Blomgren, Electrochem. Solid-State Lett. 3 (2000) 493.
S. Wang, G. T. Wu, X. B. Zhang, Z. F. Qi, W. Z. Li, J. Electrochem. Soc. 145 (1998)
51.
Li, X. Huang, L. Chen, Z. Wu, Y. Liang, Electrochem. Solid-StateLett. 2 (1999) 547.
Wang, M. Yoshio, A. Takeshi, Z.Ogumi, Electrochem. Solid-State Lett. 4 (2002) A499.
Yoshio, H. Wang, K. Fukuda, T. Umeno, N. Dimov, Z. Ogumi, J. Electrochem. Soc.
9 (2002) A1598.
L. Turner, World Intellectual Property Organization PatentApplication 2000
00/03444.
Y. Beaulieu, K. C. Hewitt, R. L. Turner, A. Bonakdarpour, A.A. Abdo, L. Christensen,
W. Eberman, L. J. Krause, J. R. Dahn, J. Electrochem. Soc. 150 (2003) A149.
Ohzuku, A. Ueda, N. Yamamoto, J. Electrochem. Soc.142 (1995)1431.
Kavan, M. Gratzel, Electrochem. Solid State Lett. 5 (2002) A39.
Ohzuku, K. Tatsumi, N. Matoba, K. Sawai, J. Electrochem. Soc. 147 (2000) 3592.
Huang, L. Kavan, I. Exnar, M. Gratzel, J. Electrochem. Soc.l42 (1995) L142.
O'Regan, M. Gratzel, Nature 533 (1991) 737
P. Feist, P. K. Davies, J. Solid State Chern. 101 (1992) 275; G. Nuspl, K. Yoshizawa,
Yamabe, J. Mater. Chern. 7 (1997) 2529.
Wagemaker, A. P.M. Kentgenst, F. M. Mulder, Nature 418 (2002) 397.
M. Thackeray, W. I. F. David, J. B. Goodenough, J. Solid State Chern. 55 (1984) 280;
A. Picciotto, M. M. Thackeray, Mat. Res. Bull. 21 (1986) 583.
Sigala, D. Guyomard, Y. Piffard, M. Tournoux, C.R. Acad.Sci.Paris. 320 (1995) 523;
Guyomard, C. Sigala, A. LeGal LaSalle, Y. Piffard, J. Power Sources 68 (1997) 692.
Kumagai, Jap. Patent08-241101 (1996).
Denis, E. Baudrin, M. Touboul, J-M. Tarascon, J. Electrochem. Soc. 144 (1997) 4099.
Leroux, G. Piffard, Ouvrard, J-L. Mansot, D. M. Guyomard, Chern. Mater. II (1999)
48.
Leroux, G. R. Goward, W. P. Power, L. F. Nazar, Electrochem. Solid-State Lett. 1
998) 201; F. Leroux, L. F. Nazar, Solid State Ionics 133 (2000) 37.
Leroux, L. F. Nazar, Mat. Res. Soc. Symp. Proc, New Materials for Batteries and Fuel
lls, II, 575 (1999) 173.
Nazar, G. Goward, F. Leroux, M. Duncan, H. Huang, T. Kerr, J. Gaubicher, J. Int. J.
rg. Mat. 3 (2001) 191.
Denis, E. Baudrin, F. Orsini, G. Ouvrard, M, Touboul. J-M. Tarascon, J. Power Sources
(1999) 79; S. Denis, E. Baudrin, F.Orsini, J-M. Tarascon, Chem.Mater. 12 (2000)
33.
Rowsell, J. Gaubicher, L. F. Nazar, J. Power Sources in press.
Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon, Nature 496 (2000) 407.
Duncan, L. F. Nazar, Mat. Res. Soc. Symp. Proc, Solid State Ionics 548 (1999) 71; M.
ncan, G. Ouvrard, L. F. Nazar, J. Gaubicher, Abstract# 235, The Electrochemical
ciety Meeting Abstracts, Hawaii,l999.
H. Gregory, J. Chern. Soc., Dalton Trans. (1999) 259.
H. Elder, L. H. Doerrer, F. J. DiSalvo, J. B. Parise, D. Guyomard, J. M. Tarascon,
rn. Mater. 4 (1992) 928.
Juza, K. Langer, K. von Benda, Angew. Chern. Int. Ed. Engl. 7 (1968) 360.
Nishijima, N. Tadokoro, Y. Takeda, N. Imanishi, 0. Yamamoto, J. Electrochem. Soc.
1 (1994) 2966; S. Suzuki, T. Shodai, Solid State Ionics 116 (1999) 1.
Nishijima, Y. Takeda, N. Imanishi, 0. Yamamoto, J. Solid State Chem.l13 (1994)
5.
Shodai, S. Okada, S. Tobishima, J. Yamaki, Solid State Ionics 86-88 (1996) 785
Gudat, R. Kniep, A. Rabenau, Angew. Chern. Int. Ed. Engl. 30 (1991) 199.
latyk, R. Z. Kniep, Kristallogr. NCS 214 (1999) 447.
. C. Rowsell, V. Pralong, L. F. Nazar, J. Am. Chern. Soc. 123 (2001) 8598.
. Kim, M.G. Kim, J. M. Lee, T. Kang, H-J. Sohn, Electrochem. Solid-State Lett. 5
2) A103.
hodai, Y. Sakurai, T. Suzuki, Solid State Ionics 122 (1999) 85.
akeda, M. Nishijima, M. Yamahata, K. Takeda, N. Imanishi, 0. Yamamoto, Solid e
Ionics 130 (2000) 61
ang, Y. Takeda, N. Imanishi, 0. Yamamoto, J. Electrochem. Soc. 147 (2000) 1671.
ereira, L.C. Klein, G. G. Amatucci, J. Electrochem. Soc., 149 (2002) A262
onconduit, M. Tillard-Charbonnel, C. Belin, J. Solid State Chern. 156 (2001) 37.
ouza, V. Pralong, A. J. Jacobson, L. F. Nazar, Science 296 (2002) 2012.
ouza, V. Pralong, A. J. Jacobson, L. F. Nazar, 11th IMLB, Abstract #60, Monterey
fornia 2002.
L. Doublet, F. Lemoigno, F. Gillot, L. Monconduit, Chern. Mater. 14 (2002) 4126.
azar, D. Souza, V. Pralong, Patent Cooperative Treaty International Application 2002
/CA02/00016
ralong, D. C. Souza, L. F. Nazar, Electrochem. Comm. 4 (2002) 516
lcantara, J. L. Tirado, J. C. Jumas, L. Monconduit, J. Olivier-Fourcade, J. Power
ces 109 (2002) 308.
apter 5
ARBONACEOUS AND GRAPHITIC ANODES
M. Winter, K.-C. Moeller and J. 0. Besenhard
titute for Chemical Technology of Inorganic Materials, Graz University of Technology,

Stremayrgasse 16, A-8010 Graz, Austria, e-mail: martin.winter@tugraz.at
CARBONS IN ION TRANSFER BATTERIES

In rechargeable lithium ion batteries the main1 application of carbon is
use as a host anode accessible by lithium cations via electrochemical
tion, the lithium being provided by another insertion material used as
cathode (Figure 5.1A). This "electro-insertion" reaction is basically a
/guest solid state redox reaction involving electrochemical charge
fer coupled with insertion of mobile guest ions (in lithium ion batteries,
e are lithium ions) into the structure of a solid host, which is a mixed
ronic and ionic conductor. The major structural features of the host are
after the insertion of the guests. A common convention in the literature
at "intercalation" is regarded as a special case of "insertion". The term
rcalation" implies the restricting condition that a layered host takes up
ts within its interlayer gaps ("galleries"), which may result in a volume
ge perpendicular to the layers, but which causes no other structural
ges. However, according to this strict definition, even graphite would be
a pure intercalation host, as during Li accomodation the stacking
ence changes by sliding of the graphene layers. Therefore, in this review
erm intercalation will be used where the historical conventions or the
ent practice make its use appropriate, i.e., we will use the term lithium
hite intercalation compounds for lithiated graphites (cf. Refs. 1-16).
hould be noted here, that cation and anion intercalation can also occur into carbon
ials used as conductive electrode additives. At the anode, this cation intercalation is a r
reaction occurring in parallel to intercalation in the active carbonaceous anode ial.
At the cathode, anion intercalation can happen during overcharge of the electrode.
onaceous and Graphitic Anodes 145
Graphite is a redox-amphoteric intercalation host, i.e., it can be

calated by cations and anions yielding so-called donor- and acceptor-
graphite intercalation compounds. 11' 15- 17 All-graphite-based
E ctrolyte
c
2
0 0 0
0 0 0
Electrolyte
0 0 0
Dts.charg•
tr
0 0
0 0
- 0
H 0
0
0
0
0 0
000 0
Discharge
lnMrtion Ho•t
( Or l!lph ile Anodto)
ln1ortlon Ho•l
(Cothodo)
c,.,x c,,x
oo
n
000
000
000
o g Elec trolyte
ooo
0 0 0 + . 980
000
0 0 92 2
M C,
0 "'"""..
c,x c.
+---
Charge g oo
c,x
c, c,
Charge
e 5.1. Rechargeable cells with active graphite intercalation electrodes: (A) graphite as
m cation intercalation host anode (negative electrode) in lithium ion cells, (B) graphite as
(accepting cations M) and cathode (positive electrode, accepting anions X) in dual ite
cells, and (C) hydrogen sulfate ion transfer cell based on graphite anode and cathode
accepting anions X=HS04.), CnX denotes a stage-1, C2nX a stage-11
gensulfate/graphite intercalation compound (cf. Section 5.3.2.2).
argeable batteries have been investigated since decades. In these so d
"dual-graphite" cells, 18 22 - the charge/discharge reaction is the
calation/release of electrolyte cations and anions into graphite (Figure
). As the ion concentration of the electrolyte changes during cycling,
gh electrolyte salt (cations +anions X) must be incorporated in the cell
uarantee the electrolyte conductivity, and enough solvent must be
Chapter 5
ided to leave the electrolyte salt dissolved at any state of charge/

harge.
In contrast with lithium ion cells, where the electrolyte is basically an
transfer medium, this charge/discharge mechanism requires excess of
trolyte salt and solvent at the expense of energy density. Furthermore, as
cell operates at up to 5 V, the electrolyte suffers from strong electrolyte
ction and oxidation.
A very interesting example of an all-graphite-based rechargeable
23
ry has been proposed by Rlidorff and Hofmann as early as 1938. They
constructed an electrochemical cell by combining a graphite cathode,
intercalated with HS04-(H2S04)y (this is a so-called stage I graphite
calation compound, the HS04- guest ion being solvated by sulfuric acid
ecules; cf. Sections 5.3.2.2 and 5.4.2 for explanations) with a plain (not
calated) graphite anode. During charge, the anions are transferred from
cathode to the anode, until both electrodes possess the same guest
entration. During discharge, the guest concentration in the cathode is re
blished by the backwards ion transfer. Using today's terminology, this
trochemical set-up would be named a hydrogen sulfate ion cell. This cell
e first "ancestor" of the lithium ion cell.
HISTORY OF LITHIATED CARBONS
The first occurrence of lithiated carbons formed by reaction of lithium

carbonaceous materials is said to go back to the development of the
monuclear hydrogen bomb in the 1940s, whose enormous energy release
2.4 MeV is based on the strongly exothermal fusion reaction of the
2 4 4
pes 63Li and 1H yielding two molecules of 2He.Z
Nevertheless, the first intended and official synthesis of lithiated
25 26
ons was reported by Herold • in 1955, who prepared a lithium-graphite
calation compound (Li-GIC) from the elements via a chemical synthesis
e. The formation of Li-GICs is in competition with the formation of Li
7 29
des.Z - Therefore, the chemical synthesis has to be carried out at
erate temperatures and lithium vapor pressures. The chemical synthesis e
30 31
GICs of the heavy alkali metals K, Rb, and Cs was reported before, ·
32 33
reas the first Na-GIC's emerged afterwards. • Fully intercalated highly
talline graphites have the stoichiometric formulae of MC 8 (M=K, Rb,
Cs) and LiC6.26 Na hardly intercalates into graphite yielding diluted
11 26
's of the formulae :::NaC24 - NaC 64• ' A higher Na uptake was obtained
34 38
disordered carbons. - An overview about the early works on
mically prepared GICs is given in Ref. 26.
The electrochemical formation of Li-GICs was performed more than
years after the chemical synthesis. As in the case of metals which alloy
lithium,39-42 electrochemically prepared lithiated graphites 43-45 emerged

in room temperature lithium cells, when they were used as substrates for
lectrodeposition of metallic lithium from Lt-containing electrolytes. It
a while until the occurring phenomenon was identified to be a
rsible lithium intercalation into graphite.46.47 Not much later, graphite
proposed as reversible anode for nonaqueous electrolyte based
argable lithium batteries48 by adapting the already known concepts of
ro-insertion and electro-intercalation. 1-3
Though the safety problems with the dendritic lithium deposition in
argeable nonaqueous organic electrolyte lithium batteries (cf. reviews
4,49-61) urgently demanded a replacement by an insertion anode, which
ded the use of Li metal, the attempts to introduce a graphitic carbon
e in a rechargeable lithium battery resulted in no practical realization for
46 47
than a decade. These attempts included the use (i) of electrolytes •
h co-intercalate into graphite (cf. Section 5.4.2), (ii) of solid polymer
62
rolytes, '63 and (iii) of a lithiated (charged) graphite anode in molten
64 65 64 66
• or organic electrolytes, ' which is sensitive to air and moisture and
complicates cell assembly.
The first two research reports on carbon anodes with practical
ance for rechargeable lithium batteries by Kanno et al.67 and Mohri et
appeared in 1989. Mohri et al.68 even constructed a complete lithium ion
consisting of a pyrolytic carbon anode and a metal oxide cathode, which
able to deliver several hundreds of cycles at satisfactory efficiencies. In
several japanese companies have investigated and patented the usage of
69
ific carbonaceous anodes before that time. The first announcement of a
mercial application of a carbon insertion host anode in a so-called
m ion cell was done by Sony in 1991.70• 71 Their cell was based on a
dered carbon anode and a LiCo02 cathode, which means the electrode
rials are assembled in the discharged state, the electrode materials being
ery sensitive to moisture and air.
72
The underlying ion transfer concept was proposed and applied with
73 77
native electrode materials - before. However, though lithium ion cells
lithium-aluminum alloy anode73 or an insertion anode based on oxidic
rials74-77 were feasible in principle, they suffered from poor practical
gy densities (oxides) or cycling stabilities (alloys). The superiority of
m ion cells is chiefly connected with the use of carbonaceous anode
rials.
The announcement of Sony induced a run on the research topic lithium
atteries, which in tum produced tens of thousands of papers, patents,
other publications (cf. books and compilations78-82 and reviews
42,54,55,57-59,61,83,84). With regard to carbonaceous anodes, this
publication output is mainly due to the variety of carbon materials
tigated for lithium insertion and the various resulting lithium storage
Chapter 5
hanisms. In addition, the manifold interactions of the carbon material

other battery components, in particular with the electrolyte, have to be
idered. The related issues will be discussed in the following sections.
ELECTROCHEMICAL Li STORAGE IN CARBON

1. Carbonaceous Host Materials and Their Properties
The quality and quantity of sites which are capable of reversible
um accommodation and the type and extent of interaction with the
rolyte depend in a complex manner on the crystallinity, texture, (micro-)
ture, and (micro-) morphology of the carbonaceous host material. The
of carbon determines the current/voltage characteristics of the
rochemical intercalation reaction and also potential side reactions.
onaceous materials suitable for lithium insertion or intercalation are
mercially available in manifold types and qualities (cf. reviews
4,53,57,60,85-94). Many exotic carbons have been specially synthesized
14
laboratory scale by pyrolysis of various precursors. Even fullerene and
95
on nanotubes have been evaluated, however with little practical
ess.
Heteroatoms, i.e., atoms in the carbonaceous material different from
on, can play an important role in the bulk (Section 5.3) and surface
istry (Section 5.4) of carbon and furthermore have strong influence on
haracteristics in the lithium ion cell. The term "carbon alloys" has been
96
ed for structurally and chemically heterogeneous carbons.
It has to be emphasized that the assumed suitability of a carbonaceous
rial for a lithium intercalation host depends strongly on the method of
valuation. Quite a few carbons may have been rejected as anode
rials due to an inadequate evaluation. As a consequence, the
ification of a carbon as "good" or "poor" anode material can be
etimes only preliminary.
Due to the variety of available carbons a classification is inevitable.
t carbonaceous materials which are capable of reversible lithium storage
oughly be classified as graphitic and non-graphitic (disordered).
.1. Graphitic Carbons
Graphitic carbons basically comprise of sp2-hybridized carbon atoms

h are arranged in a planar "honeycomb-like" network, i.e., a "graphene"
is formed97 (Figure 5.2). Van der Waals forces provide a weak
sion of the graphene layers leading to the layered graphite structure.
a strictly crystallographic point of view, the term "graphite" is only

icable to carbons having a layered lattice structure with a perfect
king order of graphene layers, either the prevalent AB (hexagonal
hite, 2H-, or a-phase ) or the less common ABC (rhombohedral
hite, 3R-, or -phase). Due to the small energy for transformation of AB
ABC stacking (and viceversa), perfectly stacked graphite crystals are
eadily available. Therefore, the term "graphite" is often used regardless
acking order.
The ratio of hexagonal to rhombohedral contents can be varied by
in processes. Mechanical treatments, especially milling, shearing, or
85 98 99
sonic impact result in a larger extent of the -phase, • • whereas
mal annealing procedures at elevated temperatures convert the material
he thermodynamically more stable a-form. By heat treatment, it is
ible to create a graphite which contains 100% a-phase, whereas this
d not be achieved by mechanical measures for the -form. So far, e
contents up to 30% have been reported. The rhombohedral and
85 98 100
gonal contents of graphites can be estimated from XRD spectra. •
• -
t should be noted here, that in addition to a change of the stacking
ence, the mechanical milling is able to destroy carbon-carbon bonds
104 108
thus to increase the number of structural defects, - the kind and
109 111
nt of defects being dependent on the milling procedure. - On the
rary, heat-treatment does "heal" the graphite structure by reducing the
85 101
ber of defects. ' After long term milling, graphite can develop a fully
110 112
rphous structure. ' Shorter milling times can confine the
101 109 111
rphization to the surface. ' ' Milling is also able to change the
ical (surface) composition of the carbon material, when done under a
113
in atmosphere, as the defects created by milling are highly reactive. -
basal plane surface
outline of
unit cell
e 5.2. Crystal structure of hexagonal graphite showing the AB layer stacking sequence,
it cell, and the subdivision in basal plane and prismatic surfaces.
Chapter 5
The structure of layered graphite gives rise to basically two kinds of

aces, prismatic (edge) surfaces and basal plane surfaces (Figure 5.2). l
(defect- and contaminant-free) basal plane surfaces are homogeneous
"smooth" and consist of carbon atoms. In contrast, the prismatic surfaces
heterogeneous and "rough" and apart from carbon may contain various,
tly (but not only) oxygen-containing, surface groups. The quantity and
ity of surface groups for graphitic and non-graphitic carbonaceous
rials can be changed by mechanical (e.g., milling), thermal (e.g., heat
ment), and chemical measures (e.g., exposure to reagents creating a
in surface chemistry), or a combination of these methods. For further
ils on carbon surface chemistry, cf. 11,117-122. The problems with the
rmination of the surface chemisty of carbons are discussed in Refs.
124.
It is well known that the prismatic and basal plane surfaces of graphite
117 118
a different electrochemical behavior in many respects. •
calation, for example, proceeds via the prismatic and not via the basal e
surfaces. The interaction of graphitic materials with the electrolyte is
dependent on the respective surface and its properties. The information
t the relative and absolute extents of prismatic and basal plane surface
s in a graphite material is essential for understanding and controlling the
ions of the graphite anode in lithium ion cells.
The adsorption of nitrogen at 77 K is a commonly used measurement
2
he determination of the total specific surface area (as m /g) of graphites
other materials through application of the well known Brunauer
ett-Teller (BET) equation. The determination of the BET surface area is
d on the premise that the surface of the solid adsorbent is homogeneous
thus, the whole surface adsorbs the same extent of nitrogen per area,
that the calculation of the nitrogen volume forming a mono-layer on the
rbent allows to determine the surface area. Any surface heterogeneities
ng to different degrees of (multi-layer) nitrogen adsorption at different
ce sites are not accounted for in the BET theory. A more sophisticated
ry than the BET theory has to be applied in order to estimate the total
125 126
ce area of graphitic carbons. ' This theory also allows to
idually estimate the extents of prismatic and basal plane surfaces of a
hitic material, and thus for example may disclose which part of the
hite surface is active for lithium intercalation and which not l 85• 1271•
The actual structure and surface of practical carbonaceous materials
ates more or less from the ideal graphite structure and surface. Even
ly ordered graphites typically have a number of structural defects, such
128 129
slocations, steps, boundaries, point defects, and cracks. • Moreover,
onaceous materials consisting of aggregates of graphite crystallites are
ed graphites as well. For instance, the terms natural, artificial, synthetic,
pyrolytic graphite are commonly used, although the meant materials are
130
crystalline. The crystallites may vary considerably in size, ranging
om to (lm. In some carbons, the aggregates are large and relatively free
efects, e.g., in highly oriented pyrolytic graphite (HOPG). Furthermore,
re effects can be observed as the crystallites may be differently oriented
ch other. In addition to essentially graphitic crystallites, carbons may
include crystallites containing carbon layers (or packages of stacked
on layers) with randomly distributed misfits and misorientation angles of
131
stacked segments to each other (turbostratic disorder ) . The latter
der can be identified from a non-uniform and on average increased
130 132
layer spacing compared to graphite; ' the corresponding XRD
rns show relatively broad and quite asymmetric (001) diffraction
133
s.
.2. Non-Graphitic Carbons
hen the structure disorder becomes more dominating among the

allites, the carbonaceous material can be no longer considered graphitic
must be regarded as non-graphitic (disordered). For carbons that contain
characteristic graphitic and non-graphitic structure units, the
ification in graphitic and non-graphitic types can be somehow arbitrary
n many cases only made for the sake of convenience.
In case of non-graphitic carbons most of the carbon atoms are arranged
planar hexagonal network too. The basic structural unit (BSU) contains
ar aromatic structures consisting of several tens of rings, which have
134
rent relative arrangements. Though layered structure segments made
BSUs are probable, there is actually no far-reaching crystallographic
r inc direction. This is typical of the BSUs form carbon structures which
haracterized by amorphous areas embedding and partially crosslinking
135 137
graphitic-type (layered) structure segments. - The number and size
ese areas vary and depend both on the precursor material and on the
ufacturing process, e.g., manufacturing temperature and pressure. The
of the BSUs and of the structure segments made from the BSUs can be
l, so that many internal edges with internal heteroatoms or unpaired
117
rons are possible. Most non-graphitic carbons are prepared by
lysis of organic polymer or hydrocarbon precursors at temperatures
w -1500°C. Further heat treatment at temperatures from -1500 to
0°C allows to distinguish between two different types of carbons.
Graphitizing carbons develop the graphite structure continuously
g the heating process. The structure segments are mobile enough to
graphite-like crystallites as crosslinking between them is weak.
hitization of these carbons is simply the disorder order transformation
all but disorderedly stacked carbon layers and graphite crystallites into
ordered (more graphitic) arrangements.
Chapter 5
Non-graphitizing carbons exhibit no true development of the graphite

ture, even at high temperatures (2500-3000°C), since the carbon layers
mmobilized by strong crosslinking. In order to increase the number of
slinks, organic precursors which tend to crosslink, or crosslinking agents
138
as oxygen can be applied. Since non-graphitizing carbons are
hanically harder than graphitizing ones, it is common to divide the non
hitic carbons into soft and hard carbons.135 The precursors, the
aration, and assumed structures of hard carbons resemble those of glassy
139 140
on. • The inter-layer and intra-layer disorder in hard carbons is so
ng that the carbon layer fragments building up the carbonaceous material
sizes of only z1 nm. 141' 142
Franklin 135 reported that, compared to graphitizing carbons, non
hitizing carbons exhibit a considerably larger fine-structure micro- and
143
-porosity. The size of the pores is in the range of 1 nm. 141•
The different structural features (graphitic order, disorder, and
pores) of graphite, soft carbons, and hard carbons result in different
53 85
um storage mechanisms (see below). Dahn and colleagues • and
umi and colleagues144 '145 have proposed theories based on the Franklin
el, which allows to correlate the complex XRD patterns of non-graphitic
ons with the relative amounts of unorganized (randomly oriented and
rphous) and organized (layered) areas. As a result, the lithium storage
city of a specific carbon material can be approximately predicted.
146 147
Azuma and colleagues ' have developed a novel model of the
ture of soft and hard carbons. The model is based on the analysis of the
layer distances, true densities, and nearest-neighbor-interactions of
le carbon layers in a carbonaceous material and allowed them to predict
tural characteristics beneficial for high lithium storage capabilities. The
difference to the Franklin model is that the model considers each single
on layer, whereas the Franklin model considers small crystallites, each
aining two or three carbon layers, with equidistant interlayer distances
equal orientation to each other.
For graphitizing carbons, Mering and colleagues 148 has proposed a
el which assumes that each single carbon layer is disordered.
hitization is primarily considered as an ordering of this intra-layer
rder. The development of inter-layer ordered graphite structure is a
ndary effect, resulting from the intra-layer ordering process. The
ication of this model to the lithium intercalation behavior is reported in
. 85,149.
The mobility of the carbon structure units, i.e., depending on the
el, the mobility (i) of small graphite crystallites, (ii) of single carbon
rs (or BSUs), or (iii) within the single layers, determines the degree of
ostructural ordering as well as the texture of the carbonaceous material.
pends on the state of aggregation of the intermediate phase during
137
lysis, which can be solid, liquid, or gaseous. Non-graphitizing carbons
usually products of solid-phase pyrolysis, whereas graphitizing carbons

commonly produced by liquid- or gas-phase pyrolysis. Examples of
ucts of solid-phase pyrolysis are chars and glassy (vitreous) carbon.
use of the small crystal size and high structural disorder of the precursor
mers and the presence of crosslinkings, the ability of these carbons to
hitize is low. Pyrolysis of thermally stabilized polyacrylonitrile or pitch,
h are the precursors for carbon fibers, yields also solid intermediate
es, but stretching of the fibrous material during the manufacturing
137
ess produces an ordered microstructure. The synthesis of petroleum
, which is the most important precursor for the manufacture of carbons
graphites, is an example of liquid-phase pyrolysis. Petroleum coke is
uced by the pyrolysis of petroleum pitch, which is the residue from the
llation of petroleum fractions. Cokes are also products from pyrolysis of
tar pitch and aromatic hydrocarbons at 300-500°C. Carbon black,
carbon, and carbon films are examples of gas-phase pyrolysis products,
products of thermal cracking of gaseous hydrocarbon compounds which
130 137
eposited as carbon on a substrate. •
The ability to graphitize also depends on the pre-ordering and pre
re of the respective precursor. For example, the graphitization ability
in the carbon layer should be higher (i) when the precursor material
prises highly condensed aromatic hydrocarbons, which can be
idered to have a graphene-like structure and (ii) when neighboring
s, graphene layers or graphitic crystallites are suitably orientated to each
r.
Apart from manifold structures, carbons can have various shapes,
s, and textures including powders with different particle size
91 150152
ibutions, foams, whiskers, foils, felts, papers, fibers, • spherical
150 153 87 154160
cles ' such as mesophase carbon microbeads (MCMB), ' etc.
erent shapes, forms, and textures do not only influence the lithium
calation behavior and the interactions with the electrolyte, but also may
91
rucial for electrode construction. 90 ' Comprehensive overviews on
on properties are given e.g., in Refs. 130,136,137. Further information
arbonaceous materials can be found in Refs. 130,135,137,161-164.
2. Electrochemical Li Storage in Graphitic Carbons

The electro-insertion of Lt cations into carbon proceeds according to:
discharge
charge
Chapter 5
During electrochemical reduction (charge) of the carbon host, Li

ons penetrate into the carbon and form a lithiated carbon LixCn. Li cation
ase during carbon oxidation is the anode discharge reaction.
2.1. In-Plane Ordering
In general, a maximum lithium content of one Li guest atom per 6

on host atoms can be reached for lithium intercalation into highly
red, highly crystalline graphite (n6 in LiCn or1 in LixC6) at ambient
sure. Higher lithium storage capacities into carbon can be reached with
in non-graphitic carbons and specifically treated graphites (see below).
The intercalation reaction proceeds via the prismatic surfaces (Figure
. Through the basal planes, intercalation is possible at defect sites only.
ng intercalation the stacking order of the graphene layers shifts to AA.
s, two neighboring graphene layers in LiC6 directly face each other
ure 5.3A). The energetically favored AA stacking sequence of LiC6 has
1165 1661
proven by quantum mechanical calculations. · Due to the hosted
um the interlayer distance between the 16graphene 167 168 layers increases
erately (10.3% has been calculated for LiC6 · · ). The stacking order
he lithium interlayers is aa (a Li-C6-Li-C6-Li chain exists along the c
16'169). In LiC6 the lithium is distributed in-plane in such a manner that
r:
37 0
·l!::;;:
$
ccupation of nearest neighbor sites is avoided (Figure 5.3B).
A Li layer B C
A
-1§ ij@@ . 0 0
cl
"""'
0· "<?'t:>
0.430 nm grap hene Iayer I} "<'fao
e 5.3. AA layer stacking sequence and the aa interlayer ordering of the intercalated
m in LiC6 (A). In-plane structures of LiC6 (B) and LiC2 (C).
A higher lithium in-plane density by occupation of nearest neighbor
is obtained in the phases LiC2 - LiC4, i.e., x in LixC6 = 2-3, which are
ared chemically from graphitic carbon under high pressure (-60 kbar)
89 173
high temperature (-300°C) conditions. • 170 - The close Li-Li distance
C2 (Figure 5.3C) results in a higher chemical activity 174
of lithium than that
thium metal (Li-Li bond length (20°C) = 0.304 nm ). Under ambient
itions LiC2 decomposes slowly via various metastable intermediate Li/C
89 172
es to LiC6 and metallic lithium. ' A study of the electrochemical
vior of LiC 2 can be found in Ref. 89. For more details on the chemical
hesis of LixC6, see Refs. 26,88,89,175,176.
2.2. Stage Formation
A general feature of intercalation into graphite is the formation of a

dic array of unoccupied layer gaps (galleries) at low concentrations of
25 26 177 180
11 15 16 23
t species, called stage formation. ' • • - This stepwise process
' ' '
be described by the stage index "s" which is equal to the number of
hene layers between two nearest guest layers. Staging is a
modynamic phenomenon related with the energy required to "open" the
der Waals gap between two graphene layers for the guests entering the
. The repulsive coulombic interactions between the guest ions are less
tive. As a consequence, few but highly occupied van der Waals gaps are
getically favored over a random distribution of guests.
Staging phenomena as well as the degree of intercalation can be easily
A
-
- -" Li
0.3
B
·m
3
S
== tag e II
la yer
Is> IV (1' Li)
1
IV]
stage I
I
I
.2 S> IV 0.2 liV III

(1" Li)
+
IV > III IIL
II L ll
(ij
0.1
IV II L
+ +
I
+ & (Reduction)
Ill :II L: II
I I
II+ I
t
0 0 '-'------+
t (i = const.)
I
-0.2 0.34 0.5 1 xin Li,C.
-30 18 12 6 n in LiC,
Ill II L II I Stage
e 5.4. Stage formation during electrochemical lithiation of graphite. A: Schematic
nt current charge curve. B: Schematic voltarnmetric curve.
Chapter 5
rved during the electrochemical reduction of carbons in Lt containing

rolytes. Figure 5.4A shows a schematic potential/composition curve for
onstant current reduction (charge) of graphite to LiC6 corresponding to a
um storage capacity of 372 Ahlkg with respect to the graphite mass. The
ntial plateaus in the curve arising during reduction indicate two-phase
ons (coexistence of two phases). Under potentiodynamic control (linear
ntial sweep voltammetry) the two-phase regions are indicated by current
s (Figure 5.4B).
Only within the small one-phase regions, the Li+ transport is a
sion process. Within the two phase regions, the stage transformation
eeds by nucleation of lithium and following phase boundary movement, a
181 183
reaction front propagates through the graphite. - The reaction front a
narrow width of <5 p,m and a rate of 5-10 mm/min. 183
Apart from the stage s = I, other binary phases corresponding to the
es s =IV, III, II L, and II (which can be also obtained by chemical
hesis12• 14• 25•26 ' 177' 179' 184) were identified and confirmed by X-ray
action 12' 14' 184- 188 and Raman spectroscopy. 189• 190 The splitting of stage II
two, s = II (x=0.5 in LixC6) and s = II L (x=0.33 in LixC6), is due to
rent lithium packing densities. The term II L indicates a liquid like
ture, which has no in-plane ordering [ 1911. A phase diagram of LixC 6 was
tructed from chemical184 and electrochemical188 data. There are different
rts in the literature about what happens at the beginning of intercalation,
how the "181 lithium" is accommodated in graphite, before the stage IV
pound is formed. It is reported, that the 1st lithium forms stages higher
189 192
s = IV. ' In contrast, it is said that the 1st lithium is randomly
ibuted in the graphite.188'193' 194 The discrepancies about staging are
ussed in Ref. 90.
.3. Nature of the Bond in Li-Graphite Intercalation Compounds
In 1934-1938 Hofmann and co-workers23'195 have proposed that the

in graphite intercalation compounds is of electrostatic nature, which led
to the term "graphite salts". Based on this proposal, Hennig and
34 196
agues ' and Ubbelohde and colleagues 197' 198 have developed the
ic model", e.g., lithium is stored in cationic form in graphite, whereas the
hite host takes over the correponding negative charge. Today, this model
ill in use and may be the basis for the term "lithium ion battery". In
ral, GICs show a larger electronic conductivity than pure graphite
efore GICs are called "synthetic metals"), which is an indicator for the
up of electrons by graphite. Moreover, the electrons are not
ogeneously, but locally distributed over the graphene planes, the
ron density being particularly high near the intercalated cationic
7 143 146 199 202
tsY Li NMR spectroscopy ' ' - as well as quantum mechanical
ulations203 confirm the mostly ionic state of lithium in Li-GICs.
However, there is considerable criticism on the purely ionic model.
cially the Li-graphite bond in LiC 6 seems to be partially metallic (or
lent). Due to the aa stacking of the lithium guest layers, the Li-Li
nce along the c axis is smaller than the in-plane Li-Li distance (Figure
). On the other hand, the small increase in volume due to Li
calation can be better explained by assuming that the lithium is stored as
204
all size guest, i.e., in ionic form. Recently, Hightower et al. attributed
utral character to the Li in stage-!LiC 6 by using transmission electron
gy loss spectroscopy. Indeed, the potential of the stage-! compound
ure 5.4A) is very close (but is not the same!) to metallic lithium, whereas
higher stage compounds and the diluted stage-!compound show more
tive potentials. It is reasonable to presume that at the beginning of
calation the lithium is accommodated into graphite in a more ionic form.
is also in agreement with the possibility of formation of solvated Li
s (cf. Section 5.4.2). During further reduction, the intercalated lithium
ually becomes more neutral. However, as there is still a small potential
between the potential of metallic Li and that of LiC6, the Li in LiC6 still
ld be partly ionic. Even the lithium which is stored in hard carbons at
ntials of only 5 mV vs. LilLi+ is not in a fully metallic state (cf. Section
). Finally, also the regular in-plane order of guests in LiC6 (Figure 5.3B)
be better explained if the Li contains still some ionic charge which is
iding repulsive coulombic interactions with its near lithium neighbors.
.4. Reversible and Irreversible Capacity
Constant current charge/discharge curves for u+ intercalation/de

calation into/out of graphite clearly prove the staging phenomenon
ure 5.5A). Nevertheless, there are no sharp discontinuities between the
phase regions because (i) the packing density of LixC 6 slightly varies (a
e width exists), 12• 14 (ii) various types of overpotentials cause plateau
ing in galvanostatic measurements (and peak broadening in
12
mmetric measurements, respectively), '14 and (iii) in practical cases
um intercalation does not proceed homogeneously over the electrode.
retically, the Li+ intercalation into carbons is fully reversible. In the
tical charge/discharge curve, however, the charge consumed in the first e
significantly exceeds the theoretical specific charge for the first stage of
372 Ah/kg. The subsequent discharge recovers only -80-95% of this ge.
In the second and subsequent cycles the charge capacity is lower and
harge recovery is close to 100%. The excess charge consumed in the
cycle is generally ascribed to irreversible reactions (i) of Lt with
hite and (ii) of the (lithiated) graphite with the electrolyte.
Chapter 5
The first reasoning is only to a minor part responsible for the excess
ge. It is believed that reduction of impurities like H20 or 02 present on
arbon surface, reduction of surface groups at the prismatic surfaces, and
ersible lithium incorporation into the carbon matrix ("formation of
205 207
ue compounds" - ), e.g., by reduction of heteroatom-groups located
rbon layer edges in polycrystalline carbons, take place.
The main part of the irreversible charge loss, however, is due to
tions with the electrolyte, which typically consists of a mixture of
nic solvents and a highly soluble lithium salt. The strong reducing
er of graphite (and other carbon anodes), at potentials where lithium
53 208 211
calation proceeds, leads to reductive electrolyte decomposition. • -
suitable nonaqueous electrolytes, however, "passivating" films of
2.00
0
8 A
:::. 1.60
c
6 _J
4 .; 1.20
Discharge > Discharge
2
>
0 0.80
"iii
8 'E
Q)
6 0.40
0
4 a..
2 0.00
0
Charge 0 200 400 600 800
200 400 Capacity I Ah kg·'
Capacity I Ah kg·' 2.00
8
0 ·- +
::J
:::. D
6
B _J 1.60
ui
.4- > 1.20
2 >
0 Iii 0.80
8 .. 'E
6 ..
4 ..
a
a.. 0.40
2
0 0.00
100 200 0 200 400 600 800
Capacity I Ah kg·' Capacity I Ah kg·'
e 5.5. Typical schematic I" cycle charge/discharge curve of (A) graphite, (B) soft
n (coke), (C) soft carbon (low temperature, hydrogen containing), (D) hard carbon.
containing electrolyte decomposition products protect the carbon

ces and the electrolyte against further reactions with each other. These
s are considered to act as a "sieve" being selectively permeable to the
rochemically active charge carrier, the Lt cation, but impermeable to
other electrolyte component that would react with lithium, i.e., they are
idered to behave as a electronically insulating solid electrolyte

212 213
phase (SEI). '
Since the reactions of lithium with carbon and surface impurities as
as the film formation on LixC6 are associated with the irreversible
umption of material (lithium and electrolyte), the corresponding charge
is frequently called "irreversible specific charge" or "irreversible
city". The reversible lithium intercalation, on the other hand, is called
ersible specific charge" or "reversible capacity". The losses have to be
mized because the charge and lithium losses are detrimental to the
ific energy of the whole cell and, moreover, increase the material
nses due to the necessary excess of costly cathode material, which is the
um source in a lithium ion cell after cell assembly.
The extent of the irreversible charge losses due to SEI formation
nds to a first approximation on the surface area of the lithiated carbon
109 208
h is wetted by the electrolyte. ' '211'214-222 However, detailed
223
stigations proved that the particle-size, ·224 -shape (fiber, flake, bead, or
225 226
to shapes, etc.), ' -thickness 211 '227 '228 and -morphologl29 -232 also
ence the performance.
A third contribution to the irreversible capacity is rather related with
roperties of the electrode than the properties of the active material. The
rate capability of a carbon electrode may be achieved by thin layer
posite electrodes that are made up from small particle size carbons, with
in the Jl.m range, that are mechanically bound together by an usually
58
nic binder material '215 '218 '233 -236 which supports the compactness of the
rode. Usually, the binder amount is in the range of a few wt%. The
er/carbon composite is attached to a thin copper foil functioning as
nt collector.
2
The presence of these "inactive" components makes the relation of
rial and electrode parameters with reversible and irreversible capacities
complex. For example, the binder in the electrode on the one hand
y coats the electrochemically active surface area, so that SEI formation
218 237
solvent co-intercalation reactions are reduced. • ' 238 On the other
, the rate capability is decreased because of the covered surface areas. In
ion, it has to be considered that the binder reduces also the capacity
use of the extra inactive mass.
Another influence of the inactive electrode components is due to the
that they experience the same potential as the active material. On all
ronically conductive electrode components (active material, conductive
ive and current collector) in contact with the electrolyte, SEI formation
place. In some cases, the binder also reacts in an irreversible reaction.
ductive additives like carbon blacks, which reversibly take up lithium, can not be
ered as inactive.
Chapter 5
rominent example for the latter effect are fluorinated binders like
(tetrafluoro ethylene) (PTFE; "teflon", (-CF2-CF2-)n) and several
239
(vinylidene fluoride) (PVdF, (CHrCFr)n)· Kavan has disclosed that
rinated polymers can be electrochemically reduced ("electrochemical
onization") at the conditions a graphite anode experiences in contact
an organic electrolyte in a lithium ion cell during charge. In general,
alogenated polymers react with lithium very easily. The dehalogenation
ds LiHal (LiF, LiCl...) and elemental carbon. The reduction of PTFE by
um is well investigated as PTFE was suggested by Gabano and
240
eagues as cathode for lithium primary cells. PTFE can be completely
241
erted into carbon by electrochemical reduction, therefore increases
42
irreversible capacitl and consequently should be avoided as binder
carbon anodes. The reaction of Li with halogenated polymers
aining vicinal halogen atoms is facilitated due to the formation of
getically favored products, i.e., conjugated C=C double bonds and LiF.
ially halogenated polymers which contain no vicinal halogen atoms, such
VdF and poly(vinyl chloride) (PVC) usually do not react, because the C
ond is more resistive against reaction with Li, which blocks the
239
ation of a conjugated carbon-carbon bond system. In contrast,
2431
san et al. [ have observed a highly exothermal reaction of PVdF with
ated graphite at elevated temperatures. It is likely that the presence of
cts or irregularities in the monomer sequence, which locally lead to
al halogen atoms in the partially halogenated polymer, may alleviate the
rochemical reduction by lithium.
3. Electrochemical Li Storage in Non-Graphitic

Carbons
Compared to graphite, non-graphitic (disordered) carbons offer a

r variety of structures and thus of sites available to lithium storage (cf.
ion 5.3.1.2). This leads to different constant current charge/discharge
ntial profiles [Figure 5.5(B-D)]. In analogy to graphite, however, there is
reversible part in the charge/discharge potential profile of these carbons, h
again can be attributed to reactions with the electrolyte and SEI
ation and to irreversible reactions of lithium with the carbon host.
It should be noted again that the use of the term intercalation is
icted to processes which involve the lithium accommodation within the
layer gaps of a layered host like graphite. Any other Li storage process
arbons without significant layered structure is considered as insertion
ess.
Depending on the precursor and the heat-treatment temperature the
onaceous materials discussed in this chapter contain heteroatoms in
tion to the prevailing carbon atoms. In particular, carbons prepared at

eratures below 1000°C do not consist of pure carbon and can contain
tantial amounts of heteroatoms. Even highly crystalline graphite is
ated with heteroatoms at dislocations in the crystallites and at edges of
graphene layers (prismatic surfaces), which are supposed to affect the
ation of residue compounds and the (electro-)chemistry at the
rode/electrolyte interface. From this point of view, the influence of
-carbonaceous" elements which have been particularly added to the
on precursor material is of great interest. If the non-carbon atoms are
duced in the carbon lattice by intention, the term "doped carbons" is
. This definition is in contrast with carbons made from "carbon
usors" which contain heteroatoms by chance.
.1. Non-Graphitic Carbons Made from "Carbon Precursors"
Non-graphitic carbons can offer additional or simply more sites for

m accommodation than graphites. As a result, they show a higher
bility of reversible lithium storage than graphites. These so-called "high
ific charge" or "high capacity" carbons are usually synthesized at rather
temperatures of -500 to -1000°C and can exhibit reversible capacities
400 to -2000 Ah/kg, depending on the heat treatment, the organic
ursor, and the electrolyte used. When lithium is stored in the bulk of
graphitic (soft and hard) carbons, one can suppose that a higher specific
ge (in Ah/kg) requires a correspondingly higher volume (lower density)
e carbonaceous matrix to accommodate the lithium. The calculated X
ensities and average carbon layer distances of soft and hard carbons
te from that of graphite as follows: graphite: 2.26 g/cm 3 & 335.4 pm;
3 3
carbon: ""2.0 g/cm & 375 pm; hard carbon: 1.4-1.8 g/cm & 380-
85 43 146
pm. .1 • As a consequence, the charge densities (in Ah/1, these
metric values are important for small portable applications) of these
ons should be comparable to those of graphite.
The variety of precursor materials and of manufacture processes leads
rbonaceous host materials with various structures and compositions.
might explain the numerous models which have been suggested to
88 244
in the high capacity. Yazami et al. ' - 248 proposed the formation of
dendritic metallic lithium multilayers on external graphene sheets and
249 250
ces. Sato et al. " suggested that Lh covalent molecules occupy
st neighbor sites in intercalated carbons, i.e., they assume a higher in
density of lithium guests. Yata's and Yamabe's groups discussed the
bility of lithium bilayer accommodation within the graphitic galleries of
ns with high interlayer spacing of -0.400 nm.251- 256 Peled et al.213• 257•258
ested that the extra charge gained by mild oxidation of graphite is
uted to the accommodation of lithium at the prismatic faces between
Chapter 5
adjacent crystallites and in the vicinity of defects (cf. Section 5.4.2.3). In

. 259-264 it is assumed that small particle size carbons can store a
iderable amount of lithium at graphite edges and surfaces in addition to
lithium located between the graphene layers. The existence of different
265
torage sites is discussed by many other authors, -271 mostly however on
basis of different models. Other groups proposed that additional lithium
be accommodated in nano-cavities87'154 '156 -158 or pores,272• 273 which are
ent in the carbon at temperatures below -800°C and which the
272
trolyte can not enter. Alternative comparisons of the models can be
d in the reviews 12,14,60,85,87,90,91. The significance of the models is
er controversially discussed.267 -278
For a systematic correlation of the lithium storage capacity with the
cture of the parent carbonaceous material, we will use the considerations
Dahn's group,277-281 which are supported by work of Fischer and
eagues 141' 142' 282' 283(cf. also their recent reviews 282' 284).
Their understanding is based on the structural model of graphitizing
t) and non-graphitizing (hard) carbons proposed by Franklin (cf. Section
1.2 for the Franklin model and other related information). If soft and hard
ons are prepared at temperatures of "'650 to 900°C, they show very high
cities, which are especially high at very low temrceratures.274• 285 In
278 2 0 281
252 277
tion, they exhibit a potential hysteresis: • • • the lithium
'
rtion occurs close to 0 V vs. LilLi+ whereas the lithium de-insertion
rs at much more positive potentials (Figure 5.5C). The hysteresis is due
netic hindrances,281 as the insertion/release of Li into/from these carbons
sociated with large exothermal heats.286
These low-temperature carbons can be considered as an intermediate
rial between the organic precursor compounds and carbon, as
287
iderable hydrogen is still contained in their structure. The extent of
um storage capacity and of potential hysteresis is assumed to be
ortional to the hydrogen content in the carbon.274' 278' 288 The hydrogen is
ed at the periphery of aromatic molecules constituting the basic
ture units. The ratio of hydrogen to carbon (H/C ratio) in the material is
when a substantial amount of hydrogen is present during manufacture,
r because hydrogen is already incorporated in the precursor material
or by manufacture under H2 atmosphere. It has been suggested that
um is bound near single proton terminated edge carbons which induces a
3
al bond change at the carbon from sp2- to sp -hybridization.281• 282 The
change requires activation energy for both the lithium insertion and
oval process leading to the observed potential hysteresis. Alternatively, a
kage-recovery model of weak C-H bonds has been suggested in Refs.
290. Other ideas for the lithium storage mechanism are discussed in
. 291-293.
With increasing temperature the electrical conductivity of the material
294
ases by several orders of magnitude and hydrogen is removed.
295
lly, all hydrogen is lost at temperatures of 1000°C, whereas oxygen
296 297
nitrogen (1500°C) and sulfur (2000°C) are lost at higher
eratures and thus may still have an effect on the lithium storage
vior. The specific charges achieved after the removal of hydrogen
nd on the structure of the carbonaceous material, whether it is a soft
phitizing) or a hard (non-graphitizing) carbon.J57,159,202,278,281,298-30J
At temperatures -1000°C the graphitizing (soft) carbons develop a
rdered structure, which offers a lower number of sites for lithium
12 14 53 85
calation as compared to graphite. ' ' " In addition, cross-linking of
on layers in disordered carbons hampers the shift to AA stacking, which
cessary for the accommodation of a higher lithium amount into graphitic
02 304
? - Correspondingly, rather low capacities of ::::150-200 Ah/kg are
rved in soft carbons like turbostratic carbons, and more disordered
ons like cokes and certain carbon blacks. The potential profile of the 1st
ntercalation/de-intercalation cycle of a coke electrode differs from that
raphite, in the sense that the reversible intercalation of Lt begins at a
ntial >1 V vs. Li!Lt, and the curve slopes without distinguishable
aus (Figure 5.4B). This behavior is a consequence of the disordered
ture providing electronically and geometrically non-equivalent sites,
eas for a particular intercalation
3 stage in highly crystalline graphite the
are basically equivalent.5
With increasing treatment temperature soft carbons develop more
hitic structure segments. The sites for lithium storage which were
erly determined by the disordered structure (see above) change to
hitic sites, where lithium resides in the Van der Waals gaps between
red graphene layers. Finally, at -3000°C the structure 12of14graphite with
ready reported electrochemical performance is achieved. '
In contrast to soft carbons, many non-graphitizing (hard) carbons
ned at temperatures of -1000°C show a high specific charge with little
resis. This high specific charge of several hundreds Ah/kg can be
ed only at a very low potential of a few mV vs. Li!Lt (Figure 5.5D).
paring with graphitic carbons and cokes which show volume expansions
contractions in the range of -10% during cycling, the hard carbons are
ed not to be subject to dimensional changes during lithium uptake and
val due to the high distance of >-0.380 nm between two neighboring
143
on layers. In order to explain the high capability of lithium storage in
277 279 305 308
graphitizing carbons, Dahn et al. ' · - suggested that lithium is
orbed" on both sides 305 of single graphene la1ter
06 sheets which are arranged
a "house of cards" or like "falling cards" (Figure 5.6A). The "falling
" model is the advanced form of the "house of cards" model and takes
into account the storage of lithium in micropores. Both the pore size and
ore openings should be small to avoid the reaction of stored lithium
the electrolyte. '279" 305' 309 Recent studies on a hard carbon prepared at
278
°C reveal that the size of the single layers as well as the size of the
Chapter 5
141 142 282

s are in the range of 1 nm. ' • In conclusion, a large fraction of
le layer sheets and nanopores is beneficial for the lithium storage
bilities of hard carbons.
On the other hand, heat treatment higher than 1oooac leads to
159 310
tically reduced capacities of hard carbons. ' This may be related to
279 284
bum-off of material or to a slight mobility of single carbon layers,
ch causes the opening of the nanopores. Electrolyte can penetrate in, and
ites in the pore are no longer available for lithium storage (Figure 5.6B).
sensitiveness of nanoporous hard carbons might be a difficulty for
ace treatment processes involving higher temperatures.
The existence of different lithium storage sites in these carbons have
7 143 146 311
also monitored by Li NMR spectroscopy. In conclusion, the
• •
um stored in soft carbons is considered as almost fully ionic, whereas the
um in hard carbons is present in two different states, (i) a pseudo
llic (but not metallic!) state of the lithium, which is stored at potentials
e to the lithium metal potential, and (ii) an ionic state of the lithium
ed at higher carbon electrode potentials. The pseudo-metallic character
thus the- compared to ionic lithium- reduced repulsive coulombic
e 5.6. A) Mechanism of lithium storage in hard (non-graphitizing) carbons;

ectrolyte penetration in the pores after pore opening.
action between neighboring lithium is an explanation for the higher

um packing density in pores and at the single carbon layers.
Finally, it should be emphasized that the above proposed models for
um storage in non-graphitic carbons are based on a limited number of
riments made with a limited number of carbonaceous materials (for the
precursors, cf. Refs. 85,94). This means that carbons may be
hesized which do not belong to any of the above categories. For
ple, it cannot be excluded that lithium storage in nanopores and near
ogen terminated edge carbons proceeds in parallel. For another example,
s reported that milling of hard carbons alters the charge/discharge curve
3121
pletely, as hysteresis is observed. This may be explained with a
um storage mechanism near oxygen surface groups, as carbon-carbon

cleavages due to milling create unpaired electrons which react with
en from air thus creating the surface groups.
Tarascon and colleagues found that also milling of highly graphitic
ons resulted in graphites with very high reversible and irreversible
109
cities. - 111 As the milling creates structural disorder, an
rochemistry-structure relation may be indeed valid also in this case.
ever, the milling creates also a large number of broken bonds and
fore unpaired electrons, which react with hydrogen or oxygen present in
tmosphere during the milling process; some contribution of heteroatoms
e capacity may be taken into account, too.- 113• 116
282
Despite the fact that the high specific charge carbons show a multiple
e specific charge of graphite, there are still some problems to solve.
In many cases extremely high irreversible specific charges were
rved,157• 253• 265'277- 279• 284• 313- 315 occasionally also at higher cycle
bers.265•278•313 •314•316 •317 The' irreversible capacities can be correlated with
formation of the SEI. 157 278279 '
313 317
' ' However, an "irreversible lithium
265 317 320
rporation" into the carbon is discussed, too. • - Apart from an
ersible trapping of lithium in the pore volume of disordered carbons, 320
arge fraction of (internal and external) heteroatorns in the non-graphitic
ons, such as hydrogen and oxygen, can irreversibly bind a considerable
unt of lithium during reduction of the carbon.Z65• 284·317 -319 In addition, the
amount of polar surface groups adsorbs large amounts of impurities
air, which can irreversibly react with lithium and thus have to be
321
ved.93•312• All of the above charge losses have to be compensated by
xcess of cathode material as lithium ion cells are assembled in the
arged state. As, in contrast to hydrogen-containing carbons, hard carbon
electrochemical properties suitable for application in lithium ion
143
ries, there are attempts to reduce the irreversible capacity of hard
312 315 318
ons by changing the surface properties.284 • • •
3221
With one exception r , carbons exhibiting hysteresis show a low
iency and poor cycling performance. Furthermore, they can be
251
arged only over a broad potential region of about 1-2 V.87• 157 158 -
• •
7-279'305'308'313 As a result the practical energy efficiency of a lithium ion
s reduced.
The end of charge potential of non-graphitic carbons must be chosen
close to 0 V vs. LilLi+ in order to obtain the available capacity. The
w "safety gap" separating from the potential where metallic lithium is
sited on the carbon surface might give rise to some safety problems of
carbons. In particular, if fast charging is applied, there is a high risk of
m plating at the end of charge. In some cases 278 •279305
•
308
• the hard carbon
rode was indeed charged below 0 V vs. Li/Lt and dendritic323 lithium
sition on such carbons was observed. In contrast, there is a difference of
Chapter 5
oximately 0.1 V between the potential of LiC6 and the potential of

um deposition (Figures 5.3A and 5.5A).
At a very low temperature, -30°C, a high polarization of the carbon
trode is observed, which is equivalent to a high overpotential. As the
potential shifts the lithium insertion potential downwards beyond the
um deposition potential, parts of the lithium storage capacity are not
lable anymore. Even the lithium intercalation capacity of graphite is very h
24
reduced in this case? Cokes might benefit from their smooth change
324
thium storage capacity as a function of the potential (Figure 5.5C).
effects of low temperature on the lithium storage capacity of hard
ons at 5 mV vs. LilLi+ (Figure 5.5D), however, must be dramatic.
The reported limitations of non-graphitic carbons seem to favor the
of graphitic anode materials at present. On the other hand, the
ge/discharge profile of hard carbons gradually changes with time, which
vantageous for estimation of the capacity of a lithium ion cell. Hybrids
raphitic and non-graphitic carbons may be a reasonable compromise (cf.
ion 5.4.2.3).
.2. Non-Graphitic Carbons with Heteroatoms ("Doped Carbons")
The reaction of precursor materials (i) containing carbon atoms

ther with (various) heteroatoms or (ii) combining a "carbon precursor"
"heteroatom precursors" can result in a variety of doped carbons
rials with interesting physical and (electro-)chemical properties. The
bination" of carbon and heteroatoms is done by the following methods:
hanical milling, chemical treatment of carbons with the dopant atom
aining precursor at elevated temperatures, chemical vapor co-deposition,
co-pyrolysis. Doping of carbons has been originally applied in order (i)
hange the chemical composition in the bulk and/or the surface, (ii) to
ge the electron distribution, and (iii) to enhance (catalyze) or slow the
325 327
hitization process. - The dopant can be substituted for carbon or is
tionally incorporated in the carbon. For further details, in particular
rding the synthesis and preparation of the materials, see Refs.
5,328,329. For the application in lithium ion cells, B, N, S, Si, and P
been preferably used as dopants.
Boron is assumed to act as an electron acceptor in a carbon host and
to cause a shift of the intercalation potential to more positive values and
53 85 330 335
crease of the reversible capacity due to more electronic sites.
• • -
can enable a more rapid charging process. For HOPG it has been found
336
B substitution results in a more disordered structure. This might be the
n for the compatibility of B-doped graphites with electrolytes which are
compatible with undoped graphites. B-doped graphitized carbon
327 334 337 338 339 340
s ' ' ' are used by Toshiba. •
Nitrogen can be incorporated substitutionally into carbon, and a co

ng with B is also possible. N presumably functions as an electron donor
should therefore have the opposite effect on the reversible capacity and
nsertion potential.53 "341 Contrarily, others342 "343 report that N-containing
ons can provide high specific charges of -500 Ah/kg. Whereas the
er lithium storage capability of the above B- and N-containing anode
rials is mainly explained by different electronic properties of the
53 331
• the high specific charge of phosphorous-doped carbons, 344.349
dy observed in 1991,350 is discussed on the basis of steric effects.344
r substituted carbons, such as layered B/C/N materials351 -357 and
42 358
• have been evaluated, too.
High capacity lithium storage materials derived by pyrolysis from
on-containing precursors like silicon-containing polymers or epoxy e
composites have been the subject of investigations, too.359 -368 In
tion to silicon and carbon, these materials can contain a substantial
unt of hydrogen and oxygen, which can lead to a large variety of
361
positions?60 • Apart from high reversible capacities of up to 770 Ah/kg,
h seem to depend on the Si content in the matrix, 361" 362 these materials
lly exhibit strong potential hysteresis and large irreversible capacities
0 Ah/kg). The interactions of the various heteroatorns with each other
with the carbon atom neighbors and their influence on reversible and
ersible lithium storage seem to be quite complex and have not been
ely clarified yet. Also, the nature of the lithium insertion mechanism is
366 367
uncertain. In some reports, • it was proposed that carbon/silicon
posite materials were prepared by above synthesis procedures. The high
sible capacities can be explained by the high Li alloying capacity of
on in addition to the lithium incorporation in disordered carbon regions
eeding independently?68 "369 If true, the insertion mechanism of a Si
l/carbon composite would be different from that of Si-doped carbon.
For all doped carbons based on non-graphitic low temperature
ons, there is some uncertainty about the effect of doping. In particular, in
e cases it is not clear whether the structure, the chemical composition, or
are responsible for the observed lithium storage behavior.
REACTIONS OF LITHIATED CARBONS WITH

THE ELECTROLYTE
The solid electrolyte interphase (SEI) concept (cf. Section 5.3.2.4)

ribes the filming behavior observed on the LixC6 anode in a very basic
general way. In fact, SEI formation is a complex process and the formed
do obviously not have the properties of a true solid electrolyte for
m cations with a transference number tu+ = 1. Nevertheless, the SEI
Chapter 5
ept has been quite generally accepted due to its simplicity.

The first note on the solid electrolyte character of the film formed on
370 213 371 372
llic Li in organic electrolytes was made by Dey. Peled • • has
ussed the reaction of Li and other reactive metals with organic and
ganic electrolytes in detail and has proposed the term SEI for the unique
the term being in use until today. It should be emphasized that the
nal work of Peled considers the SEI as a new phase formed between
e phase and electrolyte phase at the electrode/electrolyte interface, i.e.,
an "interphase", not an "interface". Depending on the electrolyte, other
mposition products may be liquid and dissolve or accumulate in the
43
rolyte or are gaseous, ' - the latter increasing the cell pressure.
373 378
1. Interphase Formation- Metallic Li Anode vs. LixC 6

Anode and Oxide Cathode
Though
• the SEI concept has been basically adapted for carbonaceous
08 210 212 213
es/ • ' there are significant differences between film formation
esses on metallic lithium and lithiated carbon. Film formation on
llic Li takes place upon contact with the electrolyte. The electrolyte,
h consists of various electrolyte components (solvent(s), electrolyte
ives, electrolyte salt, impurities) decomposes spontaneously with low
14
tivity, and some of the decomposition products form the film. On the
ary, film formation on the surface of carbonaceous hosts takes place as
arge consuming side reaction in the first few Lt intercalation/de
calation cycles, especially during the first reduction of the carbon host
rial, leading to the so-called "irreversible capacity". In this case, the
rolyte components which are least stable towards reduction selectively
first. The chemical composition of the SEI on lithiated carbon varies
14
gh the depth of the film. In addition, the film composition and
hology is also laterally non-homogeneously distributed over the carbon e
79
surface? As a consequence of the difference in the reduction
vior to metallic lithium, dissimilar intermediate reduction products and
rent decomposition mechanisms of the electrolyte may occur in the
nce of lithiated carbon.
Finally, the behavior of metallic lithium and of lithiated carbon,
ctively, during battery charge and its influence on the SEI has to be
ssed. Upon recharge of the Li anode, lithium plating occurs
ltaneously to lithium corrosion and "passivation" (formation of SEI).
, lithium deposits as highly disperse, highly reactive metal particles.
e dendrites are covered with SEI films, which in some cases
ronically insulate them from the rest of the Li. Thus, the dendrites are
ally electrochemically inactive. This reduces the efficiency of the

um dissolution/deposition process. Moreover, the dendrites grow to
ents upon cycling, which may short-circuit and locally overheat the cell
cause a disastrous thermal runaway due to the low melting point of
ly reactive Li (-180°C) and reaction with the organic electrolyte. The
e reactions imply that the surface of a metallic lithium electrode is
dically renewed during cycling, i.e., a "new" SEI is formed in each e.
Unless the Li electrode becomes inactive due to full passivation, this
ess can be repeated 12 49 59until the Li anode and/or the electrolyte are
61
pletely consumed. ' - '
At difference, the surface of the lithiated carbon electrode is all the
passivated by the SEI during cycling and no dendrites are formed as
um is inserted into carbon. The film formation on lithiated carbons
eeds in several subsequent steps. These are of course influenced by the
ions taking place in parallel at the carbon anode, e.g., reduction of
on surface groups and surface impurities, unsolvated and solvated
m insertion and self-discharge reactions. For ester-based electrolytes
ollowing reaction sequence can be assumed: (i) electron transfer from
380 382
cathodically polarized electrode in the electrolyte solution - and
381 382
ation of a solvated radical anion · (and possible intermediate
213 383
ence of solvated electrons ' ), (ii) chain reaction of the radical with
hboring electrolyte species (e.g. further electron transfer) leading to
381 382
uble products " (iii) precipitation of these products on the carbon
384
e surface by nucleation and crystal growth, (iv) further precipitation
385 387
a dense, but also thicker film is formed, - as well as further
ction of the film products deposited near the carbon anode which is
ming gradually more reductive. In contact with the highly reductive
lithiated carbon anode only fully reduced (inorganic) species, such as
LiCl, LizO, or LizC03 can exist. Partially reduced (organic) species can
"survive" in SEI regions far 12 14away
213 from the reductive carbon anode
ce near the electrolyte side. ' ' The inorganic part may be as thin as
388
389 390
The film thickness and resistance increase during charge · and
390
arge. The increase seems to be a discontinuous process proceeding in
385 386 389391
al steps. · · · The film formed in the first cycle has a thickness of
08 378 385 387
al tens of nm.Z ' ' - The thickness of the SEI increases to several
·14·388 durm. g eye1"mg or storage
h . .
t e rest stance creases m
and .
m
392 399 400
lel. - This process contributes to "cell aging". The increase of the
resistance and thus the higher overpotentials or even the electronic
ation of carbon particles by the SEI are considered as main reason for
96 398 401
city fade of carbon electrodes during cycling? · · .4°2 In parallel, SEI
ucts may dissolve in the electrolyte, which is a major factor for the self
403 405
arge of lithiated carbons. - SEI formation and dissolution are in a
Chapter 5
amic equilibrium.405 Peled406 accounts the breaking of the SEI due to

hite expansion/contraction during lithium uptake/release for the self
harge phenomenon. A list of factors contributing to self-discharge is n
in Ref. 405. Especially in the first cycles, self-discharge is a serious
lem as the SEI is not fully developed and thus sufficiently protective.
surements of self-discharge of sufficiently filmed lithium carbon
trodes show a linear lithium loss with storage time during -40 d of
age. Thereafter, the reaction rate decreases, but self-discharge is still
407 408
tent after even -170 d. ' The self-discharge rate of the carbon anode
409 410
eases with temperature, ' which may be associated with the
easing rate for SEI dissolution or breakdown.409•4 u.412 Broussely et
0·410 concluded that the electronic conductivity of the SEI is the main
ributor to the self-discharge of a graphite electrode. This assumption
ies that the SEI is not a true solid electrolyte.
Also at the cathode an interphase consisting of reaction products of
ode material and electrolyte is formed, 12.413.415 which shows "SEI-like"
erties. The electrolyte decomposition products at the cathode are,
ever, formed by oxidation. Due to the fact that various cathodes, e.g.,
o02, LiNi02, LiMn204, LiFeP04 and derivatives are used, the interphase
positions and formation mechanisms may vary more than for
onaceous anodes, which all basically consist of carbon. It has been
419
rted,392·416- that the cathode interphase contributes more to the overall
resistance than the anode interphase (SEI).
It is furthermore assumed that SEI components can migrate between
anode and cathode, which would correspond to a shuttle mechanism
420
ting self-discharge. As the transition metal oxides used for the cathode
thium ion cells partially dissolve in the electrolyte,409·421-423 and then are
sited in metallic form on the carbon 424 surface or are accumulated in the
e-SEI, Komaba and co-authors have studied the influence of
ganese, cobalt, and nickel ions on the SEI formation process. They
d detrimental effects both on capacity and on rechargeability. A general
conclusion to be drawn is that reactions at the cathode and anode and
interactive behavior should be considered in parallel.
The chemical composition and the properties of the films formed on
hite and the mechanisms involved with their formation have been
yzed by various sophisticated in situ and ex situ analytical tools. For
e reasons, in this chapter we may only mention the works of Aurbach's
p,413-415.425·426 Novak's group 216'217'375-378 and Peled's and Golodnitzky's
_212,213,379
The obtained information, however, has to be carefully considered.
ral reports had to be withdrawn because later they were disproved by a
sophisticated method. A general limitation is also that a certain
ytical method is only providing a piece of information, but does not give
omplete set of facts. Therefore, a certain property or property change of

ilm, which has been detected by a particular method may, but not
tably, be related to a corresponding (change of) electrochemical
ormance. Furthermore, the electrolyte decomposition products are highly
tive to reaction with moisture and other atmospheric species, which s
the meaning of ex situ results. On the other hand, using in situ or on
methods, the relevant information may be so complex that it is not easy
xtract the relevant information. The situation is getting even more
plex, due to the large number of electrolyte components and their
ures as well as due to the fact that the formed film and the other
rolyte decomposition products contain various species. Moreover,
rent charge/discharge regimes under various conditions, such as
erature, result in SEI films with different properties. For instance,
er charging currents result in low irreversible capacity and low SEI
406 409
tance. • Finally, also the effect of impurities in/on the electrode and
427
e electrolyte, .428 such as water,378 •409 .429 -432 always has to be considered.
large number of effects may be the reason for the quite often
nsistent reports on the SEI film.
2. Unsolvated and Solvated Lithium Intercalation

Apart from reactions with the electrolyte at the carbon surface, the
ersible capacity is furthermore strongly affected by the possible co
calation of polar solvent molecules between the graphene layers of
y graphitic matrices. This so-called solvated intercalation reaction
nds (i) on the composition of the electrolyte and (ii) on the crystallinity
morphology of the parent carbonaceous material.
.1. Effects of the Electrolyte on Co-Intercalation and SEI

Formation
In general, the intercalation of Li+ from electrolytes with organic

r solvents into fairly crystalline graphitic carbons quite often yields
ted (ternary) lithium-graphite
• intercalation compounds, Lix(solv)yCn
433 12 14 6 7 59 60 208
re 5.7B). ' .4 .4 • • • The co-intercalation of the large solvent
cules is associated with extreme expansion of the graphite matrix
cally >100%), frequently leading to graphite exfoliation (shedding of
e graphene layers or packages of graphene layers), and mechanical
tegration of the electrode. This large expansion is possible as the
tive interaction of the donor solvent shell with the Li+ cation is much
ger than the Van der Waals forces between the graphene layers of
Chapter 5
hite. In the "best" case reduced charge storage capabilities, in the worst
complete electrode destruction are the results of this reaction.
As long as the content of lithium in the graphitic carbon is low (
in LixC6), the ternary lithium-graphite intercalation compounds are
modynamically favored over the unsolvated binary compounds, LixC 6
ure 5.7A). Hence, the potentials of their electrochemical formation are
e positive than those for the formation of the corresponding compound
6. At this stage of Lt intercalation the coulombic interaction between
lithium guest layer (Ln and the balancing negative charge distributed
the graphene layers (Cn) is weak, and space to accommodate large
12 14 59 60 34 435
ent molecules is still available. • • • .4 •
The solvated GICs are thermodynamically unstable with respect to the
12 14 59 60
ction of the co-intercalated solvent molecules. The kinetically
• · •
rolled reduction depends on the type of co-intercalated solvent. It is slow
e.g., dimethyl sulfoxide, where even staging of solvated GICs can be
.4 6.4 46
where the
' -
11 3 37 43 438 441
rved, but very much faster for, e.g., PC, '
trochemical intercalation followed by fast decomposition of the
calated Lt(solv)y can be misunderstood as simple electrolyte
mposition. The reduction of Li\solv)y inside the graphite is associated
extra irreversible capacity.
Both effects resulting from solvent co-intercalation, the mechanical
ruction and the higher irreversible capacities, seriously complicate the
ation of graphitic anode materials. Due to the fact that ternary lithiated
ons are thermodynamically favored at low lithium concentrations in the
hite, i.e., at the beginning of intercalation, kinetic measures have to be
ied to diminish or even to completely suppress solvent co-intercalation.
Binary (unsolvated)
A lithium-graphite
intercalation compound
00000000
00000000
Ternary (solvated)
B lithium-graphite
intercalation compound
-graphene layer o
lithium
0 solvent
e 5.7. Schematic drawing (A) of binary and (B) of ternary lithium-graphite intercalation
ounds, as well as (C) of the SEI formed on/in graphite via electrochemically formed
ted lithium/graphite intercalation compounds.
This can be accomplished by using electrolyte components that form

tively protecting SEI films, which are only permeable for unsolvated
208
um cations. The first effective solvent in this respect was ethylene
433 42 446
onate (EC). .4 - Since the viscosity of electrolytes using pure EC is
r high, mixtures of EC with low viscosity solvents such as dimethyl
onate (DMC), diethyl carbonate (DEC), ethyl methyl carbonate (EMC)
widely used.447-453
Although the formation of binary lithium-graphite intercalation
pounds prevails in EC-based electrolytes, many examinations point at a
formation mechanism, where also solvated lithiated graphites
cipate. Film formation in the first cycle during the first reduction of the
material is not a simple surface reaction but a rather complex three
nsional process, taking basically place at a potential of -1 - -0.6 V vs.
445 454 455 189
t (Figure 5.5A). In situ dilatometric, STM, • and Raman
ods indicate a (partially) large expansion of the graphite host
sponding to the (intermediate) formation of Lix(solv)yCn at those
ntials. The reduction of Lix(solv)y on parts of the internal surfaces
een the graphene layers results in an "extra" film, which penetrates into
bulk of the graphite host (Figure 5.7C). Correspondingly "extra"
445
ersible charge losses are observed. The penetration depth of the film
the graphite depends on how deep the reaction front of solvated lithium
can propagate into the graphite, before the solvated lithium is
ctively decomposed inside the graphite. Several investigations indicate
the film formation and solvent co-intercalation
192 385 386 389 454 processes
456 on/into
hite can be even more complex. · · · ' - Solvent co-
intercalation
C-based electrolytes depends on the kind of the low viscosity co
458 459
nt457 and the kind and concentration of the electrolyte salt.
The structures of EC and PC are very similar 217 and both solvents evolve
216 217
s, ethylene (EC) ' .460 or propylene (PC)43 ' during reduction, but
tendency to exfoliation of graphite is dramatically different. Therefore,
is controversy about the reactions occurring in EC- and PC-based
rolytes.
461
Ross and colleagues have observed a significant kinetic hindrance
462
C reduction. Chung et al. .463 have proposed a strong dependence of
olvent co-intercalation behavior on structure of the solvent, by using
rently methyl-substituted PC molecules: cis- (c-BC) and trans-butylene
onate (t-BC). c-BC does, t-BC does not co-intercalate into graphite,
h is explained by steric hindrances.
425 464
Aurbach and colleagues ' attribute the exfoliation of graphite in PC
rolytes to the penetration of graphite openings such as pores and fissures
lvent molecules. Due to the fact that the graphite is only insufficiently
vated by the SEI in PC electrolytes, PC decomposes inside these
ings to propylene gas, which pressure induces cracking of the graphite.
formed cracks allow massive penetration of PC molecules inside the
Chapter 5
hite matrix, where they are reduced. Again gas is produced, which leads
rther cracking and so on. This reaction mechanism involves no PC co
calation step. This behavior seems to be only typical for PC electrolytes,
or electrochemical reduction or oxidation of other nonaqueous and
ous electrolytes, which results into strong gas evolution; a similar and
rapid mechanical destruction mechanism has not been reported so far.
465
Takamura and colleagues investigated highly graphitized carbon
rs in PC based electrolytes. They found a dependence of the extent of
liation on the number of used carbon fibers and their relative orientation
ach other. With a single fiber no exfoliation and very low irreversible
city could be observed. With a bundle of five fibers in which the fibers
been irregularly oriented to each other, strong exfoliation takes place;
irreversible capacity is high. However, if the fibers in the bundle are
nged in parallel to each other, the extent of the exfoliation reaction is
much reduced. They attributed the exfoliation to an inhomogeneous
ent distribution in the bundle, which does not occur with a single carbon
r. One might, however, also speculate that the irreversible capacity in
case is also reduced, when the current density is high and furthermore is
orm for the total surface area of the anode material.
466
The authors have investigated the reactions of graphite vs. PC by in
electrochemical dilatometry. Strong macroscopic expansions of graphite
r, which can be partially reversible under certain conditions. The results
est that PC co-intercalation into graphite is the initial step, which opens
ssures and pores in the graphite. The second step is physical penetration
C electrolyte into these openings of graphite, then gas evolution inside
hite, which causes strong mechanical stress and graphite destruction.
462 63 67
subsequent reaction process is also supported by Chung et al. .4 .4
2.2. Film Forming Electrolyte Additives and Electrolyte

Components
The choice of solvents or solvent mixtures for a certain lithium ion

ry application is usually a compromise between the desired physical
, electrolyte conductivity, volatility, flammability, wetting ability, etc.)
electrochemical properties (reduction and oxidation at the respective
rode/electrolyte interfaces, solid electrolyte interphase (SEI) formation
v1. 0r, etc.).468·469 A n elegant way to overcome the m· evt·tabl e
1t·mt·tatt·ons is compromise is the use of electrolyte components, which
even in small unts ("additive amounts") improve the electrolyte
properties in the
468
ed direction. The technical realization of this concept is quite simple,
e electrolyte additive can be just added to the base electrolyte. Among
y additive applications, such as safety improvement and overcharge
ction, electrolyte components for improved film formation processes at
e and cathode have found particular interest. Beside EC, significant

ovements have been achieved at the anode with a large number of film
ing components.
The first
ponents such as reports deal with gaseous
co2 ,390.403.433,446,470-474 electrolyte
Nz0,390,474 additives and
and 502_445,474-476
Some other prominent examples are derivatives or analogous
pounds of the cyclic carbonates ethylene carbonate (EC) and propylene
477 481
onate (PC), e.g., chloro ethylene carbonate - (which evolves C02
460 482 483
ng reduction ' ), 4,5-dichloroethylene carbonate, fluoro ethylene
484 477 481 485
onate, trifluoro propylene carbonate, ' ' vinylene carbonate
486 491 492 468 493 494
), - catechol carbonate, ethylene sulfite ' ' and propylene
495
te.
390 474 96 497 498
Other sulfur additives are: S/, ' .4 ' sulfites ( and cf. above),
499 500 501 493
ane sultone, ' S02 (above), sulfates, etc.
Also derivatives or analogous compounds of the linear carbonates,
yl carbonate (DEC) and dimethyl carbonate (DMC), e.g. dimethyl
493 502 503
te and diethyl sulfite, • S,S-dialkyl dithiocarbonates, 2-
504 505 506
oxyethyl (methyl) carbonate, dimethyl pyrocarbonate (DMPC), ·
506 182 507
tyl pyrocarbonate, ethyl propyl carbonate, ' ethyl methyl
508 509 510 511
onate · and other asymmetric alkyl methyl carbonates, · 1-
512
oroethylmethyl carbonate, and other partially fluorinated linear
513 406 514 515
onates have been proven to be useful. Even LizC03 ' ' is an
tive electrolyte additive.
A third family of compounds include partially halogenated organic
pounds such as bromo butyrolactone,516 chloro and fluoro ethylene
onate (see above), trifluoro propylene carbonate (see above), fluorinated
517 518 519
yclic compounds as glycol ethers and other ethers, urethanes,
520 521 522
ol esters, N,N-dimethylamino trifluoro acetamide, • trifluoro
methyl carbonate and other partially fluorinated linear carbonates (see
516
e), and methyl chloroformate. An overview on fluorinated solvents,
beneficial low-temperature and safety behavior, how they operate in the
etc., is given in Refs. 523,524.
A novel class of additives incorporates polymerizable unsaturated
s. These include vinylene compounds, such as vinylene carbonate (VC,
522 525 527 528
above), acrylic acid nitrile, ' - ethyl cinnamate, and vinylene
529 526 527
te. The reaction ' is proceeding according to:
=/X
+e· .. =/X
=--'X
•e
" -X" being an electron-withdrawing group (see also below).

In general, the films
390 505formed in the presence
489 of electrolyte additives
it a low resistance • and are very thin.
Chapter 5
Many of the above film forming compounds take advantage from the
that they are usually reduced at quite positive graphite electrode
ntials, i.e., at the early stages of the first reduction of the graphite
trode, before detrimental side reactions such as solvent co-intercalation
eed. The formed films have SEI properties. Solvent co-intercalation n
as a side reaction) is suppressed. This sensitiveness towards reduction
be explained (i) by the electron-withdrawing effect of chlorine- or
rine-groups introducing partial positive charge on the molecule
orinated and fluorinated solvents) or (ii) by the presence of atoms or
ps in the molecule which have a high electron affinity, e.g., sulfur in the
ation state +IV (sulfites) or double bonds (vinylene compounds). The
ction behavior and the reduction potentials of film forming components
461
be also calculated, '486 though the dependence on the chosen
rimental conditions (type of electrode and co-electrolyte components,
ge regime, kinetics, etc.) will inevitably make a practical test necessary.
In many cases, electrolyte additives not only improve typical lithium
battery electrolytes, but also allow the use of PC-based electrolytes and
r electrolytes not compatible with graphitic anodes. Especially, the use
C instead of EC in a lithium ion battery electrolyte is highly desired, as
as a large liquid range from -49°C to +240°C, whereas EC is a solid at
temperature. Thus, PC-based electrolytes usually show a low
erature performance desired for use in lithium ion cells. The low
osity of the PC at lower temperatures is counteracted by the still high
uctivity of the SEI film formed in the presence of PC.530
There is no universally applicable electrolyte additive or electrolyte
ent available and therefore additive/solvent mixtures are used at the
ent. However, the cross-interaction of the various electrolyte
ponents contributing to SEI formation is an issue of concern. For
ple, in electrolytes designed for good low temperature performance, not
still comparatively high electrolyte conductivity, but the low ionic
uctivity • of lithium within the SEI formed in these
rolytes,324 506' 531' 532and especially the low conductivity within the carbon
24
riae •506• 533 at these temperatures, limit the low temperature
ormance. Similar observations were made with flame retardant or other
y-relevant electrolyte additives which may deteriorate,534• 535 but also
improve536 '537 the SEI formation process. Finally, electrolyte additives
esired, which are in parallel beneficial for the interphase formation at
e and cathode.
.3. Effects of the Carbon on Co-Intercalation and SEI formation
Disordered soft and hard carbons are much less sensitive to solvent co-
calation than highly crystalline graphites. The crosslinking between the

3
hene layers and graphite crystallites by sp -hybridized carbon atoms
hanically suppresses the formation of solvated lithiated
12 14 53 59 60 85 208
hites. • • • • • • As a result the gap between the layers cannot
nd very much and thus there is not enough space for the solvent to co
calate. Furthermore, due to the presence of defects (pores, etc.) in the
ture, the transport of solvated lithium ions inside the graphite particle by
nsertion mechanism is hindered. Therefore, these carbons have the
ntage that they can operate in electrolytes which are free of typical film
ing electrolyte components.
Using thermally treated graphites with a low number of defects (and a
hexagonal phase content), i.e., graphites with extremely high
allinity, even the use of EC-based electrolytes results in a solvent co
85 538
calation, exfoliation and corresponding irreversible capacities. ' On
ther hand, strongly milled or otherwise mechanically treated graphites
a low number of defects (and a higher rhombohedral phase content) do
85 538
how exfoliation at all. ' -541 The tendency to solvent co-intercalation is
arily dependent on the defect concentration and not on the relative ratio
541
xagonal to rhombohedral phase.
Composite carbonaceous materials, comprising a "core" of graphite
a thin protective "shell" of non-graphitic carbon, combine the beneficial
erties of graphite (stable lithium storage potential plateau, larger gap to
metallic lithium deposition potential, high reliability of manufacturing,
cost, etc.), with the high tolerance of non-graphitic carbons to electrolyte
201
ms not compatible with graphite. ' 542- 551 These carbons also show a
r extent of irreversible capacity with solvents having low tendency to
tercalation, obviously due to an improved SEI formation process.
Graphitic materials with a crystalline core, but with more disordered
-surface structure" (i.e., thin part of graphite bulk material next to the
101 467
ce) yield similar results as the above composites. • •552 -555 Such
hite surface and sub-surface structures can be created by mechanical
101 554 555
ng and by high temperature gas treatment. • •
556 557 545 558 562
Polymer-coated, ' and metal-coated ' - carbons exhibit several
ovements compared to the uncoated material. In addition to impacts on
formation and solvent co-intercalation behavior, the metal-coated
hites benefit from better electronic conductivities and higher Li storage
cities. The latter are particularly observed when the coating consists of
ls which alloy with lithium, as lithiated metals and alloys show much
42
er lithium storage capacities than graphite (cf. Refs. 544,563-572 and
on 5.3.4.2 for C/Si composites).
In general, surface modification has been proven to be a tool for
er improvements of the carbonaceous anode materials, in particular in
of improved SEI formation. Coatings and surface modifications of hard
ns have already been discussed in Section 5.3.4.1. Takamura and his
Chapter 5
gues93,32I,465,558,562,573-575 started work on the topic "surface

ification" very early. The basic intention is to control irreversible
tions with the electrolyte, such as solvent co-intercalation and SEI
osition, as well as to remove or introduce carbon surface properties (e.g.,
ain morphologies and chemical compositions) detrimental or beneficial
performance. The interaction of anode surface properties, surface
ings and surface modifications with the electrolyte can be easily
rstood, when SEI formation on the carbon surface is compared with a
576 578
sition process (Figure 5.8). -
Apart from surface coatings, chemical treatments such as surface
524 579 580
rination, surface oxidation, surface reduction ' or heat treatment
573
r inert atmosphere ("outgassing"), as well as chemical bonding of
558 576
in substances to carbon surface groups and surface impurities, · and
3 6
ng under certain gas atmosphereli -II have been investigated. Oxidation
e most commonly applied method of surface modification. It has been
ormed by gas (air, 02, C02, 03) treatments2I3,257,258,384,554,555,577-58I and by
582 584
chemical measures. - A most interesting phenomenon is that in
e, but not all of the cases, the reversible capacity of oxidized graphites
exceed the theoretical value of the first stage compound LiC6 by more
213 257 258 582
10%. • · · A problem remains the interpretation of the results, as
basically equal treatment procedures the results vary considerably. The
A B
SEI Products
SEI Products Nuc
Substrate Substrate
(Carbon Surface) (Carbon Surface)
e 5.8. Deposition (plating") of SEI film products. Composition and morphology of the
eposit are controlled by the "plating conditions", i.e., by (i) the electrolyte composition, e
pretreatment (coating or modification) of the carbon surface ("substrate"), and (iii) the g
conditions (constant current or pulse plating, etc.): A) Plating conditions favoring
ation result in a compact and dense SEI. B) Plating conditions which facilitate the crystal
h of the initially formed nuclei induce the formation of a rough SEI deposit, which
es more SEI products to get pin-hole-free, i.e., consumes more irreversible capacity.
considerations can be applied to other anode and cathode materials, too.
ty of graphites and their various properties again makes the matter

plex and obviously not reproducible. Generally speaking, morphology
ges213' 257' 258' 384' 577 as well as surface chemistry changes 213·257·258· 384· 582
treatment are accounted for the obtained results. In addition, the

-off of reactive carbon sites or complete reactive particles, which
d contribute to the irreversible capacity to a large proportion, may also
nsidered.
Though a comprehensive interpretation of the above effects of surface
ifications is still missing, the reports have elucidated the importance of
ce properties for the performance of carbon anodes.
A most critical issue is still the influence of carbon surface
ogeneities on performance. Both unsolvated and solvated intercalation
place via the prismatic surfaces. In addition, the prismatic surface and
plane surface are different both from the viewpoint of morphology and
istry, i.e., they are very different foundations for SEI deposition (cf.
re 5.8). Thus, not the total external surface area but rather the ratio of
prismatic and basal plane surfaces will affect the irreversible
120 127
city • as well as the rate capabilities of graphites. The reports of
d and colleagues213'379 ·585 are in line with this idea, as they have detected
pletely different SEI compositions at the edge and basal plane surfaces
OPG. The existence of different surface sites and their influence on
ersible capacities is discussed also in Refs. 224,515,586,587. The
rage and/or removal of detrimental or sensitive surface sites may be the
motivation for surface coatings and modifications.
WHAT IS SPECIAL ABOUT LITHIATED

CARBONS?
Carbons have been and still are considered to be the anode material for
rgeable high energy density lithium batteries, due to a synergetic
ination of properties beneficial for performance and production. These
erties include: high lithium storage capacities, negative redox potentials
he cathode, long term cycling performance, low costs and large
dance. Any other anode material discussed in this book may possess
al, but not all of these properties. Nevertheless, as (surface)
fications or the use in composites seem to be the perspective for future
n anode materials, the same might be true for other anode materials,
e in a composite together with carbon.
Two "wonders" are associated with the story of success of carbon
es. The first wonder is strongly related with the properties of metallic
m. Apart from having the highest capacity, among the light metals M =
a, and Mg, solely metallic Li shows a chemical and electrochemical
vior which favors its use in high energy density batteries, because
ce films with the unique SEI properties are formed. In contrast,
cts of electrolyte decomposition of alkali metals other than Li are
Chapter 5
cally partially soluble and therefore do not form similarly protective

12 588
s. Films formed on Mg are typically highly passivating. Mg can be
used in combination with electrolytes which do not form films on the
589
l surface. The resulting cell, however, by far can not reach the energy
590
ities of lithium ion batteries. The excellent protective properties of the
s formed on lithiated carbons can be directly related to the extraordinary
s formed on metallic lithium.
The second wonder is the obviously "perfect accommodation" of
um cations inside carbon. There is no other cationic guest species, that
be inserted into carbons at such high concentrations, with such high rates
at a such low potential vs. a cathode material. The alkali metals M = K,
Cs form 1st stage GICs MC8 (cf. LiC6) and the increase in interlayer
ing due to intercalation of the large M cations is large. As a result, the
ge storage capacity is poor. Sodium can hardly be intercalated into
hitic materials, but in disordered soft and hard carbons (cf. Section 5.2).
ever, also in this case the sodium storage capacities are significantly
r than the Li storage capacities in the same material, and the
rsibility is poor, which may be related with insufficiently protective
ace films.
It is a most lucky case in battery science and technology, that lithiated
ons combine all the properties necessary for an outstanding anode
rial. The final questions are: What comes after lithiated carbon? Will
be a third wonder in the future?
KNOWLEDGEMENT
We acknowledge the long term support of the Austrian Science Funds

F) for our work on carbon anodes and other lithium battery materials.
FERENCES
M. B. Armand, in: Fast Ion Transport in Solids, W. van Gool, Ed., Elsevier, Amsterdam,
1973,665.
B. C. H. Steele, in: Fast Ion Transport in Solids, W. van Gool, Ed., Elsevier, Amsterdam,
1973, 103.
M.S. Whittingham, Science 1976, 192, 1226.
M.S. Dresselhaus (Ed.), Intercalation in Layered Materials, NATO ASI Series, Plenum,
New York, 1986, Vol. 148.
A. P. Legrand, S. Flandrois (Eds.), Chemical Physics of Intercalation, NATO ASI Series,
Plenum, New York, 1987, Vol. 172.
M.S. Whittingham, A. J. Jacobsen (Eds.) Intercalation Chemistry, Academic Press, New
York, 1982.
W. Miiller-Warmuth, R. SchOllhom (Eds.), Progress in Intercalation Research, Kluwer,
Dordrecht, 1994.
S. S. Prasad, in: Handbook of Solid State Batteries and Capacitors, M. Z. A. Munshi,
Ed., World Scientific, Singapore, 1995,467.
R. SchOllhorn, Angew. Chern. Int. Ed. Engl. 19 (1980) 983.

. B. Goodenough, Solid State Ionics 69 (1994) 184.
. 0. Besenhard, H. P. Fritz, Angew. Chern. Int. Ed. Engl. 95 (1983) 950.
M. Winter, J. 0. Besenhard, M. E. Spahr, P. Novak, Adv. Mat. 10 (1998) 725.
W. R. McKinnon, in: Solid State Electrochemistry, P. G. Bruce, Ed., Cambridge
University Press, 1995, 163.
M.Winter, J. 0. Besenhard: in: Handbook of Battery Materials, J. 0. Besenhard, Ed.
Wiley-VCH, Weinheim, 1999,383.
R. Schlogl, in: Progress in Intercalation Research, W. Mtiller-Warmuth, R. SchOllhorn,
Eds., Kluwer, Dordrecht, 1994, 83.
R. Moret, in: Intercalation in Layered Materials, NATO ASI Series, M.S. Dresselhaus,
d., Plenum, New York, 1986, Vol. Bl48, 185.
M.S. Dresselhaus, G. Dresselhaus, Advances in Physics 30 (1981) 139.
. R. Dahn, J. A. Steel, J. Electrochem. Soc. 147 (2000) 899.
. P. McCullough, U.S. Pat. No. 5532083, 1996.
R. T. Carlin, H. C. DeLong, J. Fuller, P. C. Trulove, J. Electrochem. Soc. 141 (1994)
73.
R. Santhanam, M. Noel, J. Power Sources 76 (1998) 147.
. Beck, in: Advances in Electrochemical Science and Engineering, Wiley-VCH,
Weinheim, 1997, Vol. 5, 303.
W. Rtidorff, U. Hofmann, Z. Anorg. Allg. Chern. 238 (1938) 1.
ee: Textbooks of Inorganic Chemistry.
A. Herold, Bull. Soc. Chim. France 187 (1955) 999.
A. Herold, in: Chemical Physics of Intercalation, NATO ASI Series, A. P. Legrand, S.
landrois, Eds., Plenum, New York, 1987, Vol. Bl72, 3.
. Guerard, A. Herold, Carbon 13 (1975) 337.
. B. Ebert, Ann. Rev. Mater. Sci. 6 (1976) 181.
. V. Avdeev, A. P. Savchenko, L.A. Monyakina, I. V. Nikol'skaya, A. V. Khvostov, J.
hys. Chern. Solid 51 (1996) 947.
. Weintraub, US Pat. No. 922645, 1909, & French Pat. No. 585185, 1924.
. Fredenhagen, G. Cadenbach, Z. Anorg. Allg. Chern. 158 (1926) 249.
. C. Asher, J. lnorg. Nucl. Chern. 10 (1959) 238.
. Metrot, D. Guerard, D. Billaud, A. Herold, Synth. Met. 1(1979/89) 363.
. Hennig, J. Chern. Phys. 19 (1951) 922.
M. M. Doeff, M. Y. Pengg, Y. Ma, L. C. de Jonghe, J. Electrochem. Soc. 141 (1994)
145.
. A. Stevens, J. R. Dahn, J. Electrochem. Soc. 148 (2001) A803.
. Thomas, D. Billaud, Electrochim. Acta 47 (2002) 3303.
. Alcantara, J. M. Jiminez Mateos, J. L. Tirado, J. Electrochem. Soc. 149 (2002) A201.
W. S. Harris, Thesis No. UCRL-8381, University of California, 1958.
40. M. L. B. Rao, J. Electrochem. Soc. 114 (1967) 665.
. P. Yao, E. D'Orsay, D. N. Bennion, J. Electrochem. Soc. 115 (1968) 999.
M. Winter, J. 0. Besenhard, Electrochim. Acta 45 (1999) 31.
. N. Dey, B. P. Sullivan, J. Electrochem. Soc. 117 (1970) 222.
. Brenner, J. Electrochem. Soc. 118 (1971) 461.
. S. Dunning, W. H. Tiedemann, L. Hsueh, D. N. Bennion, J. Electrochem. Soc. 118
1971) 1886.
. 0. Besenhard, H. P. Fritz, J. Electroanal. Chern. 53 (1974) 329.
. 0. Besenhard, Carbon 14 (1976) 111.
. Armand, French Pat. No. 7832977, 1978.
.-P. Gabano (Ed.), Lithium Batteries, Academic Press, London, 1983.
. 0. Besenhard, G. Eichinger, J. Electroanal. Chern. 68 (1976) 1.
. Eichinger, J. 0. Besenhard, J. Electroanal. Chern. 72 (1976) 1.
.C. Levy, P. Bro, Battery Hazards and Accident Prevention, Plenum, New York 1994.
. R. Dahn, A. K. Sleigh, H. Shi, B. M. Way, W. J. Weydanz, J. N. Reimers, Q. Zhong,
. von Sacken, in: Lithium Batteries, New Materials, Developments and Perspectives,
. Pistoia, Ed., Elsevier, Amsterdam 1994, 1.
.-M. Tarascon, M. Armand, Nature 414 (2001) 359.
. Broussely, P. Biensan, B. Simon, Electrochim. Acta 45 (1999) 3.
. Aurbach, E. Zinigrad, H. Teller, P. Dan, J. Electrochem. Soc. 147 (2000) 1274.
. Fauteux, R. Koksbang, J. Appl. Chern. 23 (1993) 1.
. Megahed, B. Scrosati, J. Power Sources 51 (1994) 79.
. 0. Besenhard, M. Winter, in: 2. Ulmer Elektrochemische Tage, Ladungsspeicherung in
er Doppelschicht, W. Schrnickler, Ed., Universitatsverlag Ulm 1995,47.
Chapter 5
M. Winter, J. 0. Besenhard, in: Lithium Ion Batteries, M. Wakihara, 0. Yamamoto, Eds.,

Kodansha!Wiley-VCH, Tokyo/Weinheim, 1998, Chapter 6.
K. Brandt, Solid State Ionics 69 (1994) 173.
R. Yazami, P. Touzain, J. Power Sources 9 (1983) 365.
R. Yazami, J. Power Sources 97-98 (2001) 33.
S. Basu, J. Power Sources 81-82 (1999) 200.
S. Basu, US Pat. No. 4304825, 1981
S. Basu, US Pat. No. 4423125, 1983.
R. Kanno, Y. Takeda, T. Ichikawa, K. Nakanishi, 0. Yamamoto, J. Power Sources 26
(1989) 535.
M. Mohri, N. Yanagisawa, Y. Tajima, H. Tanaka, T. Mitate, S. Nakajima, M. Yoshida,
Y. Yoshimoto, T. Suzuki, H. Wada, J. Power Sources 26 (1989) 545.
A. Yoshino, K. Sanechika, T. Nakajima, US Pat. No 4668595, 1987 (Asahi Kasei), T.
Suzuki, H. Wada, Y. Yoshimoto, M. Yoshida, S. Nakajima, EU Pat. Appl. No. 0346 088,
1989 (Sharp), Y. Nishi, H. Azuma, A. Omaru, EU Pat. Appl. No. 0357001, 1989 (Sony),
T. Nagaura, in: Progress in Batteries and Solar Cells (Eds: JEC Press Inc. and IBA Inc.),
JEC Press Inc., Brunswick OH, 1991, Vol. 10, 218.
T. Nagaura, K. Tozawa, in: Progress in Batteries and Solar Cells (Eds.: JEC Press Inc.
and IBA Inc.), JEC Press Inc., Brunswick, OH 1990, Vol. 9, 209.
M. B. Armand, in: Materials for Advanced Batteries, D. W. Murphy, J. Broadhead, B. C.
H. Steele, Eds., Plenum, New York 1980, 145.
J. 0. Besenhard, in: Euchem Conference Proceedings, "Solid State Chemistry and
Electrochemistry, Especially with Respect to its Application in Battery Research",
Rosenheim, Germany (April30- May 3, 1979).
D. W. Murphy, F. J. DiSalvo, J. N. Cardides, J. V. Waszak Mat. Res. Bull 13 (1978)
1395.
M. Lazzari, B. Scrosati, J. Electrochem. Soc. 127 (1980) 773.
B. DiPietro, M. Patriarca, B. Scrosati, J. Power Sources 8 (1982) 289.
B. DiPietro, M. Patriarca, B. Scrosati, Synth. Met. 5 (1982)1.
G. Pistoia (Ed.), Lithium Batteries, New Materials, Developments and Perspectives,
Elsevier, Amsterdam, 1994.
M. Wakihara, 0. Yamamoto (Eds.), Lithium Ion Batteries, Kodansha/Wiley-VCH,
Tokyo/Weinheim, 1998.
J. 0. Besenhard (Ed.), Handbook of Battery Materials, Wiley-VCH, Weinheim, 1999.
D. Aurbach (Ed.): Electrochim. Acta, Special Ed. "Li Insertion into Host Materials", Vol.
45, 1999.
W. A. van Schalkwijk, B. Scrosati (Eds.), Advances in Lithium Ion Batteries,
Kluwer/Plenum, New York, 2002.
J. 0. Besenhard, M. Winter, J. Pure Appl. Chern. 70 (1998) 603.
M. Wakihara, Mat. Sci. Eng. R33 (2001) 109.
J. R. Dahn, A. K. Sleigh, H. Shi, J. N. Reimers, Q. Zhong, B. M. Way, Electrochim. Acta
38 (1993) 1179.
S. Flandrois, B. Simon, Carbon 37 (1999) 165.
A. Mabuchi, Tanso 165 (1994) 298.
R. Yazami, M. Z. A. Munshi, in: Handbook of Solid State Batteries and Capacitors, M.
Z. A. Munshi, Ed., World Scientific, Singapore, 1995,425.
R. Yazami, in: Lithium Batteries, New Materials, Developments and Perspectives, G.
Pistoia, Ed., Elsevier, Amsterdam, 1994,49.
N. Imanishi, Y. Takeda, 0. Yamamoto, in: Lithium Ion Batteries, M. Wakihara, 0.
Yamamoto, Eds., Kodansha/Wiley-VCH, Tokyo/Weinheim 1998, Chapter 5.
M. Endo, C. Kim, K. Nishimura, T. Fujino, M. Miyashita, Carbon 38 (2000) 183.
T. Zheng, J. R. Dahn, in: Carbon Materials for Advanced Technologies, T. D. Burchell,
Ed., Pergamon, Elsevier, Amsterdam, 1999, 341.
T. Takamura, NPC'98 Proceedings, Annual Nagoya International Battery and Power
Sources Conference (Aug. 26-28, 1998), 49.
M. Noel, V. Suryanarayanan, J. Power Sources Ill (2002) 193.
E. Frackowiack, S. Gautier, H. Gaucher, S. Bonnamy, F. Beguin, Carbon 37 (1999) 61.
Y. Tanabe, E. Yasuda, Carbon 38 (2000) 329.
B. McEnaney, in: Carbon Materials for Advanced Technologies, T. D. Burchell, Ed.,
Pergamon, Elesevier, Amsterdam 1999, 1.
H. Gasparoux, Carbon 5 (1967) 441.
S. Flandrois, A. Fevrier, P. Biensan, B. Simon, US Pat. No. 5554462, 1996.
M. Inagaki, Tanso 94 (1978) 106.
W. Kohs, H. J. Santner, F. Hofer, H. Schrottner, J. Doninger, I. Barsukov, J. H. Albering,
K.-C. Mollea, J. 0. Besenhard and M. Winter., J. Power Sources, submitted.

H. Shi, J. Barker, M. Y. Saiidi, R. Koksbang, L. Morris, J. Power Sources 68 (1997)
91.
H. Shi, J. Barker, M. Y. Saidi, R. Koksbang, J. Electrochem. Soc. 143 (1996) 3466.
T. D. Shen, W. q. Ge, K. Y. Wang, M. X. Quan, J. T. Wang, W. D. Wie, C. C. Koch,
Nanostruct. Mater. 7 (1996) 393.
W. L. Zhou, Y. Ikuhara, W. Zhao, J. Tang, Carbon 33 (1995) 1177.
J. Tang, W. Zhao, L. Li, A. U. Falster, W. B. Simmons, W. L. Zhou, Y. Ikuhara, J. H.
Zhang, J. Mater. Res. 11 (1996) 733.
J. B. Aladekomo, R. H. Bragg, Carbon 28(1990)897.
M. Nakarnizo, H. Honda, M. lnagaki, Carbon 16 (1978) 281.
F. Disma, L. Aymard, L. Dupont, J.M. Tarascon, J. Electrochem. Soc. 143 (1996) 3959.
F. Salver-Disma, J.-M. Tarascon, C. Clinard, J.-N. Rouzaud, Carbon 1941 (1999) 1959.
F. Salver-Disma, A. du Pasquier, J.-M. Tarascon, J.-C. Lassegues, J.-N. Rouzaud, J.
ower Sources 81-82 (1999) 291.
N.J. Whelman, J. S. Williams, Carbon 36 (1998) 1309.
T. S. Ong, H. Yang, Carbon 38 (2000) 2077.
F. Chevallier, L. Aymard, J.-M. Tarascon, J. Electrochem. Soc. 148 (2001) A1216.
T. S. Ong, H. Yang, J. Electrochem. Soc. 149 (2002) AI.
D. Aurbach, B. Markovsky, A. Nimberger, E. Levi, Y. Gofer, J. Electrochem. Soc. 149
2002) Al52.
B. R. Puri, in: Chemistry and Physics of Carbon, P. Walker, P. Thrower, Eds., Marcel
Dekker, New York 1970, Vol. 6, 191.
.P. Randin, in: Encyclopedia of Electrochemistry of the Elements, A.J. Bard, Ed.,
Volume VII (Carbon, Vanadium), Marcel Dekker, New York, 1976, p. 2
P. M.A. Sherwood, in: Practical Surface Analysis, 2nd Ed., D. Briggs, M.P. Seah, Eds.,
Wiley, Chichester, 1990, p 384.
P.M. A. Sherwood, J. Elect. Spect. Relat. Phenom. 81 (1996) 391.
H.-P. Boehm, E. Diehl, W. Heck, R. Sappok, Angew. Chern. 17 (1964) 742.
. Atamny, J. Blocker, A. Diibotzky, H. Kurt, 0. Timpe, G. Loose, W. Mahidi, R.
chl5gl, Mol. Phys. 76 (1992) 851.
R. I. R. Blyth, H. Buqa, F. P. Netzer, M. G. Ramsey, J. 0. Besenhard, P. Golob,
M. Winter,Appl. Surf Sci. 167 (2001) 99.
R. I. R. Blyth, H. Buqa, F. P. Netzer, M. G. Ramsey, J. 0. Besenhard, M. Winter, J.
ower Sources 97-98 (2001) 171.
. Ross, J.P. Olivier, J. Phys. Chern. 65 (1961) 608.
.P. Olivier, S. Ross, Proc. Roy. Soc. (Land.) 265A (1962) 447.
.P. Olivier, M. Winter, J. Power Sources 97-98 (2001) 151.
. Amelinck, P. Delavignette, M. Heerschap, in: Chemistry and Physics of Carbon, P.L
Walker Jr., Ed., Marcel Dekker, New York, 1965, Vol. 1, Chapter 1
H. Chang, A. J. Bard, Langmuir 7 (1991) 1143.
H. 0. Pierson, Handbook of Carbon, Graphite, Diamond and Fullerenes, Noyes
ublications, Mill Road, Park Ridge NJ 1993, 43.
B. E. Warren, Phys. Rev. 59 (1941) 693.
. E. Franklin, Acta Cryst. 4 (1951) 253.
B. E. Warren, J. Chern. Phys. 2 (1934) 551.
A. Oberlin, in: Chemistry and Physics of Carbon, P. A. Thrower, Ed., Marcel Dekker,
ew York 1989, Vol. 22, 1.
R. E. Franklin, Proc. Roy. Soc. (London) A209 (1951) 196.
K. Kinoshita, Carbon, Electrochemical and Physicochemical Properties, Wiley, New
ork, 1987.
. Vohler, F. von Sturm, E. Wege, H. von Kienle, M. Voll, P. Kleinschmitt, in:
Ullmann's Encyclopedia of Industrial Chemistry (Executive Ed.: W. Gerhartz), 5th
dition, VCH, Weinheim 1986, Volume SA, 95.
. B. Barr, I. C. Lewis, Carbon 16 (1978) 439.
G. M. Jenkins, K. Kawamura, Polymeric Carbons, Carbon Fibre, Glass and Char,
ambridge University Press, Cambridge, 1976.
G. Diibgen, Z. Popp, Werkstofftech. 15 (1984) 331.
. Zhou, P. Papanek, R. Lee, J. E. Fisher, W. A. Kamitakahara, J. Electrochem. Soc. 144
1997) 1744.
. Zhou, R. Lee, AS. Claye, J. E. Fisher, Carbon 36 (1998) 1777.
Y. Nishi, in: Lithium Ion Batteries, M. Wakihara, 0. Yamamoto, Eds., Kodansha!Wiley
CH, Tokyo/Weinheim, 1998, Chapter 8.
. TatsuiDI, N. !washita, H. Sakaebe, H. Shiyoma, S. Higuchi, A. Mabuchi, H. Fujimoto,
Chapter 5
J. Electrochem. Soc., 142 (1995) 716.

N. Imanishi, Y. Takeda, 0. Yamamoto, in: Lithium Ion Batteries, M. Wakihara, 0.
Yamamoto, Eds., Kodansha!Wiley-VCH, Tokyo/Weinheim, 1998, Chapter 5. H.
Azuma, H. Imoto, S. Yamada, K. Sekai, J. Power Sources 81-82 (1999) 1. H.
Azuma, J. Appl. Cryst. 31 (1998) 910.
J. Maire, J. Mering, in: Chemistry and Physics of Carbon, P. L. Walker, Ed., Marcel
Dekker, New York, 1971, Vol. 6, 193.
S. Flandrois, A. Fevrier-Bouvier, K. Guerin, B. Simon, P. Biensan, Mol. Cryst. Liq.
Cryst. 244 (1994) 89.
M. lnagaki, Solid State Jonics 86-88 (1996) 833.
W. Ruland, Adv. Mater. 2 (1990) 528.
M. Endo, Y. A. Kim, T. Hayashi, K. Nishimura, T. Matusita, K. Miyashita, M. S.
Dresselhaus, Carbon 39 (2001) 1287.
Z. Ma, X. Yuan, D. Li, X. Liao, H. Hu, J. Ma, J. Wang, Electrochem. Comm. 4 (2002)
188.
K. Tokumitsu, A. Mabuchi, H. Fujimoto, T. Kasuh, J. Electrochem. Soc. 143 (1996)
2235.
H. Fujimoto, A. Mabuchi, K. Tokumitsu, T. Kasuh, Carbon 38 (2000) 871.
K. Tokumitsu, H. Fujimoto, A. Mabuchi, T. Kasuh, Carbon 37 (1999) 1599.
H. Fujimoto, A. Mabuchi, K. Tokumitsu, T. Kasuh, J. Power Sources 54 (1995) 440.
K. Tokumitsu, A. Mabuchi, H. Fujimoto, T. Kasuh, J. Power Sources 54 (1995) 444.
A. Mabuchi, H. Fujimoto, K. Tokumitsu, T. Kasuh, J. Electrochem. Soc. 142 (1995)
3049.
T. Kasuh, A. Mabuchi, K. Tokumitsu, H. Fujimoto, J. Power Sources 68 (1997) 99.
B. T. Kelly, Physics of Graphite, Applied Science Publishers, Englewood NJ, 1981.
E. Fitzer, K. Mueller, W. Schaefer, in: Chemistry and Physics of Carbon, P. Walker, Ed.,
Marcel Dekker, New York, 1971, Vol. 7, 193.
H. Marsh, Introduction to Carbon Science, Butterworth & Co. Ltd., 1989.
Book Series: Chemistry and Physics of Carbon, Marcel Dekker, New York.
C. Hartwigsen, W. Witsche1, E. Spohr, Ber. Bunsenges. Phys. Chem. 101 (1997) 859.
M. K. Song, S.D. Hong, K. T. No, J. Electrochem. Soc. 148 (2001) A1159.
D. Billaud, E. McRae, A. Herold, Mat. Res. Bull. 14 (1979) 857.
X. Y. Song, K. Kinoshita, T. D. Tran, J. Electrochem. Soc. 143 (1996) L120.
J. Rossat-Mignod, D. Fruchart, M. J. Moran, J. W. Milliken, J. E. Fisher, Synth. Met. 2
1980) 143.
V. V. Avdev, V. A. Nalimova, K. N. Semenenko, High Pressure Research 6 (1990) 11. D.
Guerard, V. A. Nalimova, in: Solid State Ionics IV, G.-A. Nazri, J.-M. Tarascon, M.
Schreiber, Eds., Mat. Res. Soc. Symp. Proc., 1995, 369, 155.
V. A. Nalimova, G. Guerard, M. Lelaurain, 0. V. Fateev, Carbon 33 (1995) 177.
V. A. Nalimova, C. Bindra, J. E. Fisher, S lid State Comm. 97 (1996) 583.
Handbook of Chemistry and Physics, 68' Edition, R. C. Weast, Ed., CRC Press, Boca
Raton FL, 1987.
R. Yazami, A. Cherigui, V. A. Nalimova, D. Guerard, in: Lithium Batteries, S.
Surampudi, V. R. Koch, Eds., The Electrochemical Society, Pennington NJ, 1993, PV
3-24,379.
Y. Mizutani, T. Abe, K. Ikeda, E. ihara, M. Asano, T. Harada, M. Inaba, Z. Ogumi,
Carbon 1997 35 (1997) 61.
N. Daumas, A. Herold, C. R. Acad. Sci. Paris 268C (1969) 373.
W. Riidorff, in: Advances in Inorganic Chemistry and Radiochemistry, H. J. Emeleus, A.
G. Sharpe, Eds., Academic Press, New York, 1959, Vol. 1, 223.
N. Daumas, A. Herold, Bull. Soc. Chim. 5 (1971) 1598.
S. A Safran, in: Chemical Physics of Intercalation, NATO ASI Series, A. P. Legrand, S.
landrois, Eds., Plenum Press, New York, 1987, B172, 47.
A. Funabiki, M. Inaba, T. Abe, Z. Ogumi, Electrochim. Acta 45 (1999) 865.
A. Funabiki, M. Inaba, T. Abe, Z. Ogumi, J. Electrochem. Soc. 146 (1999) 2443.
A. Funabiki, M. Inaba, T. Abe, Z. Ogumi, Carbon 37 (1999) 1591.
. E. Fischer, in: Chemical Physics of Intercalation, NATO ASI Series, A. P. Legrand, S.
landrois, Eds., Plenum Press, New York, 1987, Vol. B172, 59.
. R. Dahn, R. Fang, M. J. Spoon, Phys. Rev. B 42 (1990) 6424.
D. Billaud, F. X. Henry, P. Willmann, Mat. Res. Bull. 28 (1993) 477.
A. H. Whitehead, K. Edstrom, N. Rao, J. R. Owen, J. Power Sources 63 (1996) 41.
. R. Dahn, Phys. Rev. B 44 (1991) 9170.
W. Huang, R. Frech, J. Electrochem. Soc. 145 (1998) 765.
M. Inaba, H. Yoshida, Z. Ogumi, T. Abe, Y. Mizutani, M. Asano, J. Electrochem. Soc.
142 (1995) 20.

K. C. Woo, W. A. Kamitakahara, D.P. Divincenzo, D. S. Robinson, H. Mertwoy, J. W.
MiJliken, J. E. Fisher, Phys. Rev. Lett. 50 (1983) 182.
S. Mori, H. Asahina, H. Suzuki, A. Yonei, E. Yasukawa, J. Power Sources 68 (1997) 59.
0. Tanaiki, M. Inagaki, Carbon 35(1997) 831.
S. A. Safran, Phys. Rev. Lett. 44 (1980) 937.
W. Ri.idorff, U. Hofmann, Z. Electrochem. 30 (1934) 511.
M. L. Dzurus, G. Hennig, J. Am. Chern. Soc. 79 (1957) 5897.
F. R. McDannel, R. C. Pink, A, R. Ubbelohde, J. Chern. Soc. (London) 1951, 191.
A. R. Ubbelohde, F. A. Lewis, Graphite and its Crystal Compounds, Clarendon Press,
Oxford, 1960.
J. Conard, in: Chemical Physics of Intercalation, NATO ASI Series, A. P. Legrand, S.
landrois, Eds., Plenum Press, New York 1987, Vol. Bl72, 357.
. Yamazaki, T. Hashimoto, T. Iriyama, Y. Mori, H. Shiroki, N. Tamura, J. Mol. Struct.
41 (1998) 165
Y. Kida, A. Kinoshita, K. Yanagida, A. Funahashi, T. Nohma, I. Yonezu, Electrochim.
cta 47 (2002) 1691.
K. Tatsurni, T. Akai, T. Imamura, K. Zhaghib, N. Iwasita, S. Higuchi, Y. Sawada, J.
Electrochem. Soc. 143 (1996)1923
M. K. Song, S.D. Hong, K. T. No, J. Electrochem. Soc. 148 (2001) All59.
A. Hightower, C. C. Ahn, B. Fultz, P. Rez, Appl. Phys. Lett. 77 (2000) 238.
. Bittihn, R. Herr, D. Hoge, J. Power Sources 43-44 (1993) 409.
G. R. Hennig, Progr. lnorg. Chern. 1 (1959) 125.
. G. Hooley, Chern. Phys. Carbon 5 (1969) 321.
. Pong, U. von Sacken, J. R. Dahn, J. Electrochem. Soc. 137 (1990) 2009.
K. Yokoyama, N. Nagawa, in: New Sealed Rechargeable Batteries and Supercapacitors,
B. M. Barnett, E. Dowgiallo, G. Halpert, Y. Matsuda, Z. Takehara, Eds., The
lectrochemical Society, Pennington, NJ 1993, PV 93-23, 270.
. Kanno, Y. Kawamoto, Y. Takeda, S. Ohashi, N. Imanishi, 0. Yamamoto, J.
lectrochem. Soc. 139 (1992) 3397.
M. Winter, P. Novak, A. Monnier, J. Electrochem. Soc. 145 (1998) 428.
. Peled, D. Golodnitzky, G. Ardel, J. Electrochem. Soc. 144 (1997) L208.
. Peled, D. Golodnitzky, J. Pencier, in: Handbook of Battery Materials, J. 0. Besenhard,
d., Wiley-VCH, Weinheim, 1999, Part III, Chapter 6.
. Manev, I. Naidenov, B. Puresheva, G. Pistoia, J. Power Sources 57 (1995) 133.
. Manev, I. Naidenov, B. Puresheva, G. Pistoia, J. Power Sources 57 (1995) 211.
. Imhof, P. Novak, in: Batteries for Portable Applications and Electric Vehicles, C. F.
olmes, A. R. Landgrebe, Eds., The Electrochemical Society, Pennington NJ 1997,
V97-18, 313.
. Imhof, P. Novak, J. Electrochem. Soc. 145 (1998) 1081.
. Novak, W. Scheifele, M. Winter, 0. Haas, J. Power Sources 68 (1997) 267.
. Tatsumi, K. Zaghib, H. Abe, S. Higuchi, T. Ohsaki, Y. Sawada, J. Power Sources 54
1996) 425.
. Joho, P. Novak, M. E. Spahr, J. Electrochem. Soc. 149 (2002) Al020.
. Disma, C. Lenain, B. Beaudoin, L. Aymard, J.-M- Tarascon, Solid State lonics 98
1997) 145.
. Du Pasquier, F. Disma, T. Bowmer, A. S. Gozdz, G. Amatucci, J.-M. Tarascon, J.
lectrochem. Soc. 145 (1997) 472.
. Yamamoto, N. Imanishi, Y. Takeda, H. Kashiwagi, J. Power Sources 54 (1995) 72.
. Zaghib, G. Nadeau, K. Kinoshita, J. Electrochem. Soc. 147 (2000) 2110.
. Lampe-Onnerud, J. Shi, P. Onnerud, R. Chamberlain, B. Barnett, J. Power Sources
7-98 (2001) 133.
. Takei, N. Terada, K. Kumai, T. Iwahori, T. Uwai, T. Miura, J. Power Sources 55
1995) 191.
K. Fukuda, K. Kikuya, K. Isono, M. Yoshio, J. Power Sources 69 (1997) 165.
. Wang, T. Ikeda, K. Fukuda, M. Yoshio, J. Power Sources 83 (1999) 141.
. Yamamoto, Y. Takeda, N. Imanishi, R. Kanno, in: New Sealed Rechargeable
atteries and Supercapacitors, B. M. Barnett, E. Dowgiallo, G. Halpert, Y. Matsuda, Z.
akehara, Eds., The Electrochemical Society, Pennington NJ, 1993, PV 93-23,302.
. Imanishi, H. Kashiwagi, T. Ichikawa, Y. Takeda, 0. Yamamoto, M. Inagaki, J.
. Billaud, F. X. Henry, P. Willmann, J. Power Sources 54 (1995) 383.
.-P. Wu, C.-R. Wan, Y.-X. Li, S.-B. Fang, Y.-Y. Jiang, Electrochem. Solid-State Lett. 1
1999) 118.
Chapter 5
K. Zaghib, K. Tatsumi, H. Abe, T. Ohsaki, Y. Sawada, S. Higuchi, J. Power Sources 54

(1995) 435.
S. Passerini, J. M. Roso1en, B. Scrosati, J. Power Sources 45 (1993)333.
C. E. Newnham, S. Rinne, N. Scholey, J. Power Sources 54 (1995)516.
J.-M. Tarascon, A. S. Gozdz, C. N. Schmutz, P. C. Warren, F. Shokoohi, in Advances in
Porous Materials, S. Komarneni, D. M. Smith, J. S. Beck, Eds., Mat. Res. Soc. Symp.
Proc., 1995, 369, 595.
K. A. Hirasawa, K. Nishioka, T. Sato, S. Yamaguchi, S. Mori, J. Power Sources 69
(1997) 97.
R. Santhanam, M. Noel, J. Power Sources 63 (1996) l.
L. Kavan, Chem. Rev. 97 (1997) 3061
V.I. Decheneaux, F. Drovet, J.P. Gabano, Y. Jumet, Ger. Pat. No. 2210370, 1971.
G. Li, R. Xue, L. Chen, Solid State Ionics 90 (1996) 221.
Y. Nakagawa, S. Wang, Y. Matsumura, C. Yamaguchi, Synth. Met. 85 (1997) 1343.
P. Biensan, J. Power Sources, 81-82 (1999) 906.
R. Yazami, M. Deschamps, in: Advances in Porous Materials, S. Komarneni, D. M.
Smith, J. S. Beck, Eds., Mat. Res. Soc. Symp. Proc., 1995, 369, 165.
R. Yazami, M. Deschamps, J. Power Sources 54 (1995) 411.
R. Yazami, M. Deschamps, in: Rechargeable Lithium and Lithium-Ion Batteries, S.
Megahed, B. Barnett, L. Xie, Eds., The Electrochemical Society, Pennington NJ, 1995,
PV 94-28, 183.
M. Deschamps, R. Yazami, J. Power Sources 68 (1997) 236.
R. Yazami, M. Dechaumps, in Progress in Batteries & Battery Materials (Eds: ITE-JEC
Press Inc. and IBA Inc.), JEC Press Inc., Brunswick OH, 1996, Vol. 15, 161.
K. Sato, M. Noguchi, A. Demachi, N. Oki, M. Endo, Science 264 (1994) 556.
K. Sato, M. Noguchi, A. Demachi, N. Oki, M. Endo, Y. Sasabe, Int. Workshop on
Advanced Batteries (Lithium Batteries), Osaka, 1995, 219.
S. Yata, H. Kinoshita, M. Komori, N. Ando, T. Kashiwamura, T. Harada, K. Tanaka, T.
Yamabe, Synth. Met. 62 (1994) 153.
S, Yata, K. Sakurai, T. Osaki, Y. Inoue, K. Yamaguchi, Synth. Met. 33 (1990) 177.
S. Yata, Y. Hato, H. Kinoshita, N. Ando, A. Anekawa, T. Hashimoto, M. Yamaguchi, K.
Tanaka, T. Yamabe, Synth. Met. 73 (1995) 273.
S. Yata, Int. Workshop on Advanced Batteries (Lithium Batteries), Osaka 1995, 204.
S. Yata, H. Kinoshita, M. Komori, N. Ando, T. Kashiwamura, T. Harada, K. Tanaka, T.
Yamabe, in: New Sealed Rechargeable Batteries and Supercapacitors, B. M. Barnett, E.
Dowgiallo, G. Halpert, Y. Matsuda, Eds., The Electrochemical Society, Pennington NJ,
1993, PV 93-23, 502.
S. Yata, Denki Kakagu 9 (1997) 65.
E. Peled, C. Menachem, D. Bar-Tow, A. Melman, J. Electrochem. Soc. 143 (1996) L4.
C. Menachem, E. Peled, L. Burstein, Y. Rosenberg, J. Power Sources 68 (1997) 277. H.
Xiang, S. Fang, Y. Jiang, J. Electrochem. Soc. 144 (1997) L187.
S. Wang, Y. Matsumura, T. Maeda, Synth. Met. 71 (1995) 1759.
Y. Matsumura, S. Wang, J. Mondori, Carbon 33 (1995) 1457.
Y. Matsumura, S. Wang, T. Kasuh, T. Maeda, Synth. Met. 71 (1995) 1755.
Y. Matsumura, S. Wang, K. Shinohara, T. Maeda, Synth. Met. 71(1995) 1757.
W. S. Li, L. C. Jiang, G. Y. Xie, X. Jiang, J. Power Sources 58 (1996) 235.
Y. Mori, T. Iriyama, T. Hashimoto, S. Yamazaki, F. Kawakami, H. Shiroki, T. Yamabe,
. Power Sources 56 (1995) 205.
N. Takami, A. Satoh, T. Ohsaka, M. Kanda, Electrochim. Acta 42 (1997) 2537.
Y. Dai, Y. Wang, V. Eshkenazi, E. Peled, S. G. Greenbaum, J. Electrochem. Soc. 145
1998) 1179.
T. Fukutsuka, T. Abe, M. lnaba, Z. Ogumi, J. Electrochem. Soc. 148 (2001) Al260.
C. W. Park, S. Ho, S. I. Leee, S. M. Oh, Carbon 38 (2000) 995.
S. Lee, M. Nishizawa, I. Uchida, Electrochim. Acta 44 (1999) 2379.
. Mochida, C. Ku, Y. Korai, Carbon 39 (2001) 399.
S. Morimoto, Kureha Chemical Industry Co., Ltd., personal communication.
A. Nagai, M. Ishikawa, J. Masuko, N. Sonobe, H. Chuman, T. Iwasaki, Mat. Res. Soc.
Symp. Proc. 393 (1995) 339.
. R. Dahn, T. Zheng, Y. Liu, J. S. Xue, Science 270 (1995) 590.
L. B. Ebert, Carbon 34 (1996) 671.
. Zhou, P. Papanek, C. Bindra, R. Lee, J. E. Fischer, J. Power Sources 68 (1997) 297.
T. Zheng, J. S. Xue, J. R. Dahn, Chem. Mater. 8 (1996) 389.
T. Zheng, Y. Liu, E. W. Fuller, S. Tseng, U. von Sacken, J. R. Dahn, J. Electrochem.
oc. 142 (1995) 2581.
. S. Xue, J. R. Dahn, J. Electrochem. Soc. 142 (1995) 3668.

T. Zheng, J. R. Dahn, J. Power Sources 68 (1997) 201.
T. Zheng, W. R. McKinnon, J. R. Dahn, J. Electrochem. Soc. 143 (1996) 2137.
A. Claye, J. E. Fischer, Electrochim. Acta 45 (1999) 107.
P. Papanek, M. Radosavljevic, J. E. Fischer, Chem. Mater. 8 (1996) 1519.
E. Buiel, J. R. Dahn, Electrochim. Acta 45 (1999) 121.
. Gong, H. Wu, Q. Yang, Carbon 37 (1999) 1409.
M. Inaba, M. Fujikawa, T. Abe, Z. Ogumi, J. Electrochem. Soc. 147 (2000) 4008.
M. Noguchi, K. Miyasita, M. Endo, Tanso 155 (1992) 315.
Y. Gao, K. Myrtle, z. Meiji, J. N. Reimers, J. R. Dahn, Phys. Rev. B 54 (1996) 23.
Z. Wang, X. Huang, R. Xue, L. Chen, Carbon 37 (1999) 685.
Z. Wang, X. Huang, R. Xue, L. Chen, J. Power Sources 81-82 (1999) 328.
. Wang, Y. Zhang, L. Yang, Q. Liu, Solid State Ionics 86-88 (1996) 919.
N. Takami, A. Satoh, T. Ohsaki, M. Kanda, J. Electrochem. Soc. 145 (1998) 478.
H. Ago, M. Kato, K. Yahara, K. Yoshizawa, K. Tanaka, T. Yamabe, J. Electrochem. Soc.
146 (1999) 1262.
. Delhaes, F. Carmona, in: Chemistry and Physics of Carbon, P.L Walker and P. A.
hrower, Eds., Marcel Dekker, New York, 1981, Vol. 17, 89.
C. R. Kinney, Proc. of 1'' & 2"" Int. Conf on Carbon, Pergamon, New York, 1955, 83.
A. Marchand, J. V. Zanetta, Carbon 3 (1996) 483.
M. Cerutti, J. Ubersfeld, J. Millet, J. Parisot, J. Chim. Phys. 57 (1960) 907.
A. Satoh, N. Takami, T. Ohsaki, Solid State Ionics 80 (1995) 291.
A. Satoh, N. Takami, T. Ohsaki, M. Kanda, in: Rechargeable Lithium and Lithium-Ion
atteries, S. Megahed, B. Barnett, L. Xie, Eds.,The Electrochemical Society, Pennington
NJ, 1995, PV 94-28, 143.
K. Tatsumi, A. Mabuchi, N. !washita, H. Sakaebe, H. Shioyama, H. Fujimoto, S.
Higuchi, in: Batteries and Fuel Cells for Stationary and Electric Vehicle Applications, A.
. Landgrebe, Z. Takehara, Eds., The Electrochemical Society, Pennington NJ, 1993, PV
3-8,64.
. Tatsumi, N. !washita, H. Sakaebe, H. Shioyama, S. Higuchi, A. Mabuchi, H.
ujimoto, J. Electrochem. Soc. 142 (1995) 716.
. Zheng, J. R. Dahn, Synth. Met. 73 (1995) 1.
. Zheng, J. R. Dahn, Phys. Rev. B 53 (1996) 3061.
. Zheng, J. N. Reimers, J. R. Dahn, Phys. Rev. B 51 (1995) 734.
W. Xing, J. S. Xue, T. Zheng, A. Gibaud, J. R. Dahn, J. Electrochem. Soc. 143 (1996)
482.
. R. Dahn, W. Xing, Y. Gao, Carbon 35 (1997) 825.
. Liu, J. S. Xue, T. Zheng, J. R. Dahn, Carbon 34 (1994) 193.
. Zheng, Q. Zhong, J. R. Dahn, J. Electrochem. Soc. 142 (1994) L211.
. Ago, K. Tanaka, T. Yamabe, T. Miyoshi, K. Takegoshi, T. Terao, S. Yata, Y. Hato, S.
agaura, N. Ando, Carbon 12 (1997) 1781.
. Iomoto, A. Omaru, A. Azuma, Y. Nishi, in: Lithium Batteries, S. Surampudi, V. R.
och, Eds., The Electrochemical Society, Pennington, NJ 1993, PV 93-24,9.
. Conard, K. Tatsumi, S. Menu, P. Lauginie, ITE Letters on Batteries, New Technologies
& Medicine, 2000, 1, B27-47, and references therein.
W. Xing, R. A. Dunlap, J. R. Dahn, J. Electrochem. Soc 145 (1998) 62.
. lmoto, M. Nagamine, Y. Nishi, in: Rechargeable Lithium and Lithium-Ion Batteries,
. Megahed, B. Barnett, L. Xie, Eds., The Electrochemical Society, Pennington, NJ 1995,
V 94-28, 43.
M. Alamgir, Q. Zuo, K. M. Abraham, J. Electrochem. Soc. 141 (1994) L143.
315. E. Buiel, J. R. Dahn, J. Electrochem. Soc. 145 (1998) 1977.
. Iijima, K. Suzuki, Y. Matsuda, Synth. Met. 73 (1995) 9.
. Matsumura, S. Wang, J. Mondori, J. Electrochem. Soc. 142 (1995) 2914.
F. Chevallier, S. Gautier, J.P. Salvetat, C. Clinard, E. Frackowiack, J. N. Rouzaud, F.
eguin, J. Power Sources 97-98 (2001) 143.
. Jung, M. C. Sub, H. Lee, M. Kim, S. Lee, S. C. Shim, J. Kwak, J. Electrochem. Soc.
44 (1997) 4279.
. Guerin, A. Fevrier-Bouvier, S. Flandrois, B. Simon, P. Biensan, Electrochim. Acta 45
2000) 1607.
321. M. Kikuchi, Y. Ikezawa, T. Takamura, J. Electroanal. Chem. 396 (1995)
51.
. Wang, S. Yata, J. Nagano, Y. Okano, H. Kinoshita, H. Kikuta, T. Yamabe, J.
lectrochem. Soc. 147 (2000) 2498.
. E. Gerald II, R. J. Klingler, G. Sandi, C. S. Johnson, L. G. Scanlon, J. W. Rathke, J.
Chapter 5
Power Sources 89 (2000) 237

C. Wang, A. J. Appleby, F. E. Little, J. Electrochem. Soc. 149 (2002) A754.
A. Marchhand, in: Chemistry and Physics of Carbon, P.L. Walker, Ed., Marcel Dekker,
New York 1971, Vol. 7, 155.
V. A. Q. Tyumentsev, E. A. Belenkov, G. P. Shveikin, S. A. Podkopaev, Carbon 36
(1998) 845.
C. Kim, T. Fujino, K. Miyashita, T. Hayashi, M. Endo, M. S. Dresselhaus, J.
R. Riedel, Adv. Mater. 6 (1994) 549.
M. Kawaguchi, Adv. Mater. 9 (1997) 615.
J. R. Dahn, J. N. Reimers, A. K. Sleigh, T. Tiedje, Phys. Rev. B 45 (1992) 3773.
B. M. Way, J. R. Dahn, J. Electrochem. Soc. 1994 141(1994).
T. Tamaki, M. Tamaki, B'h IMLB, Ext. Abstr., Nagoya 1996,216.
B. M. Way, J. R. Dahn, T. Tiedje, K. Myrtle, M. Kasrai, Phys. Rev. B 1992 46 (1992).
M. Endo, C. Kim, T. Karaki, Y. Nishtmura, M. J. Matthews, S. D. M. Brown, M. S.
Dresselhaus, Carbon 37 (1999) 561.
U. Tanaka, T. Sogabe, H. Sakagoshi, M. Ito, T. Tojo, Carbon 39 (2001) 931.
C. E. Lowell, J. Am. Ceram. Soc. 50 (1967) 142.
Y. Nishimura, T. Takahashi, T. Tamaki, T. Endo, M.S. Dresselhaus, Tanso 172 (1996)
89.
C. Kim, T Fujino, T. Hayashi, M. Endo, M. S. Dresselhaus, J. Electrochem. Soc. 147
(2000) 1265.
N. Takarni, M. Sekino, T. Ohsaki, M. Kanda, M. Yamamoto, J. Power Sources 97-98
(2001) 677.
N. Takarni, T. Ohsaki, H. Hasebe, M. Yamamoto, J. Electrochem. Soc.l49 (2002) A9.
W. J. Weydanz, B. M. Way, T. van Buuren, J. R. Dahn, J. Electrochem. Soc. 141 (1994)
900
S. Ito, T. Murata, M. Hasegawa, Y. Bito, Y. Toyoguchi, J. Power Sources 68 (1997) 245.
T. Nakajima, M. Koh, M. Takashima, Electrochim. Acta 43 (1998) 883.
T. D Tran, J. H. Feikert, S. T. Mayer, X. Song, K. Kinoshita, in: Rechargeable Lithium
and Lithium-Ion Batteries, S. Megahed, B. Barnett, L. Xie, Eds., The Electrochemical
Society, Pennington NJ, 1995, PV 94-28, 110.
H. H. Schtinfelder, K. Kitoh, H. Nemoto, J. Power Sources 68 (1997) 258.
T. D. Tran, S. T. Mayer, R. W. Pekala, in Advances in: Porous Materials, S. Komarneni,
D. M. Smith, J. S. Beck, Eds., ¥at. Re. Soc. S ml?. !'roc._. 1?95, 369, 49. .
A. Omaru, H. Azuma, M. Aokt, A. Kita, Y. Ntsht, m: Ltthtum Battenes, S. Surampudt,
V. R. Koch, Eds., The Electrochemical Society, Pennington NJ, 1993, PV 93-24, 21.
348. R. L. Morrison, US Pat. No. 5558954, 1996.
T. D. Tran, J. H. Feikert, X. Song, K. Kinoshita, J. Electrochem. Soc.l42 (1995) 3297.
H. Azuma, A. Omaru, Y. Nishi, Electrochem. Soc. Japan, Fall Meeting, Nagoya, 1991,
Ext. Abstr. 132.
Y. Matsuda, M. Ishikawa, T. Nakamura, M. Morita, S. Tsujioka, T. Kawashima, in:
Rechargeable Lithium and Lithium-Ion Batteries, S. Megahed, B. Barnett, L. Xie, Eds.,
The Electrochemical Society, Pennington NJ, 1995, PV 94-28, 85.
M. Morita, T. Hanada, H. Tsotsumi, Y. Matsuda, M. Kawaguchi, J. Electrochem. Soc.
139 (1992) 1227.
Y. Matsuda, M. Morita, T. Hanada, M. Kawaguchi, J. Power Sources 43-44 (1993) 75. M.
Ishikawa, T. Nakamura, M. Morita, Y. Matsuda, S. Tsujioka, T. Kawashima, J.
Power Sources 55 (1995) 127.
M. Ishikawa, T. Nakamura, M. Morita, Y. Matsuda, M. Kawaguchi, Denki Kakagu 62
1994) 897.
M. Morita, T. Hanada, H. Tsutsumi, Y. Matsuda, in: High Power, Ambient Temperature
Lithium Batteries, W. D. Clark, G. Halpert, Eds., The Electrochemical Society,
Pennington NJ, 1992, PV 92-15, 101.
M. Ishikawa, M. Morita, T. Hanada, Y. Matsuda, M. Kawaguchi, Denki Kakagu 61
1993) 1395.
Y. P. Wu, S. Fang, Y. Jiang, R. Holze, J. Power Sources 108 (2000) 245.
A.M. Wilson, J. N. Reimers, E. W. Fuller, J. R. Dahn, Solid State lonics 74 (1995) 249.
A. M. Wilson, G. Zank, K. Eguchi, W. Xing, J. R. Dahn, J. Power Sources 68 (1997)
195.
W. Xing, A.M. Wilson, K. Eguchi, G. Zank, J. R. Dahn, J. Electrochem. Soc. 144 (1997)
410.
A. M. Wilson, G. Zank, K. Eguchi, W. Xing, B. Yates, J. R. Dahn, Chern. Mater. 9
1997) 2139.
. S. Xue, K. Myrtle, J. R. Dahn, J. Electrochem. Soc. 142 (1995) 2927.

R. A. Guidotti, B. J. Johnson, Proc. 37'h Power Sources Conf, Cherry Hill, NJ, 1996,
219.
W. Xing, A.M. Wilson, G. Zank, J. R. Dahn, Solid State /onics 93 (1997) 239.
A.M. Wilson, J. R. Dahn, J. Electrochem Soc. 142 (1995) 326.
A. M. Wilson, J. R. Dahn, in: Rechargeable Lithium and Lithium-Ion Batteries, S.
Megahed, B. Barnett, L. Xie, Eds., The Electrochemical Society, Pennington NJ, 1995,
V 94-28, 158.
G. S. Wang, G. T. Wu, X. B. Zhangh, Z. F. Qi, W. Z. Li, J. Electrochem. Soc. 145 (1998)
751.
. Niu, J. Y. Lee, Electrochem. Solid-State Lett. 5 (2002) Al07.
A. N. Dey, Electrochem. Soc. Fall Meeting, 1970, Abstr. 62.
E. Peled, J. Electrochem. Soc. 126 (1979) 40.
. Peled, in: Lithium Batteries, J.-P. Gabano, Ed., Academic Press, London, 1983,
Chapter 3.
. Joho, B. Rykart, R. Imhof, P. Novak, M. E. Spahr, A. Monnier, J. Power Sources 81-
2 (1999) 243.
M. Fujimoto, Y. Kida, T. Nahma, M. Takahashi, K. Nishio, T. Saito, J. Power Sources
3 (1996) 127.
P. Novak, F. Joho, R. Imhof, J.-P. Panitz, 0. Hass, J. Power Sources 81-82 (1999) 212.
M. Lanz, E. Lehmann, R. Imhof, I. Exnar, P. Novak, J. Power Sources 101 (2001) 177.
. Novak, J.-C. Panitz, F. Joho, M. Lanz, R. Imhof, M. Coluccia, J. Power Sources 90
2000) 52.
. Novak, F. Joho, M. Lanz, B. Rykart, J.-C. Panitz, D. Alliata, R. Kotz, 0. Haas, J.
ower Sources 97-98 (2001) 39.
. Peled, D. Bar-Tow, A. Merson, A. Gladkich, L. Burstein, D. Golodnitzky, J. Power
ources 97-98 (2001) 52.
. Endo, K. Tanaka, K. Sekai, J. Electrochem. Soc.147 (2000) 4029.
. Endo, M. Ata, K. Tanaka, K. Sekai, J. Electrochem. Soc. 145 (1998) 3757.
E. Endo, M. Ata, K. Tanaka, K. Sekai, J. Electrochem. Soc. 146 (1999) 49.
D. Rahner, J. Power Sources 81-82 (1999) 358.
M. Winter, H. Buqa, B. Evers, T. Hodal, K.-C. Moller, C. Reisinger, M. V. Santis
lvarez, I. Schneider, G. H. Wrodnigg, F. P. Netzer, R. I. R. Blyth, M. G. Ramsey,
. Golob, F. Hofer, C. Grogger, W. Kern, R. Saf, J. 0. Besenhard, ITE Batt. Lett., 1999,
(2), 129.
.-K. Jeong, M. Inaba, T. Abe, Z. Ogumi, J. Electrochem. Soc. 148 (2001) A989.
K. A. Hirasawa, T. Sato, H. Asahina, S. Yamaguchi, S. Mori, J. Electrochem. Soc. 144
1997) L81.
. Alliata, R. Kotz, P. Novak, H. Siegenthaler, Electrochem. Comm. 2 (2000) 436.
M. Dolle, S. Grugeon, B. Beaudoin, L. Dupont, J-M. Tarascon, J. Power Sources 97-98
2001) 104.
S. Yamaguchi, H. Asahina, K. A. Hirasawa, T. Sato, S. Mori, Mol. Cryst. Liq. Cryst. 1
22 (1998) 239.
J. 0. Besenhard, M. W. Wagner, M. Winter, A. D. Jannakoudakis, P. D. Jannakoudakis,
. Theodoridou, J. Power Sources 43-44 (1993) 413.
. Zhang, M. S. Ding, K. Xu, J. Allen, T. R. Jow, Electrochem. Solid-State Lett. 4 (2001)
206.
Y.-C. Chang, H.-J- Sohn, J. Electrochem. Soc. 147 (2000) 50.
. Zhang, B. S. Haran, A, Durairajan, R. E. White, Y. Podrashanzky, B. N. Popov, J.
ower Sources 91 (2000) 122.
. Zhang, P. N. Ross, Jr., R. Kostecki, F. Kong, S. Sloop, J. B. Kerr, K. Striebe1, E. J.
airns, F. McLarnon, J. Electrochem. Soc. 148 (2001) A463.
. Wang, X. Guo, S. Greenbaum, J. Liu, K. Amine, Electrochem. Solid-State Lett. (2001)
A68.
.-S. Kim, Y.-T. Park, J. Power Sources 91 (2000) 172.
. Wang, A. J. Appleby, F. E. Little, Electrochim. Acta 46 (2001) 1793.
. Wang, I. Kakwan, A. J. Appleby, F. E. Little, J. Electroanal. Chem. 489 (2000) 55.
. Bloom, B. W. Cole, J. J. Sohn, S. A. Jones, E. G. Polzin, V. S. Battaglia, G. L.
enrikson, C. Motloch, R. Richardson, T. Unkelhauser, D. Ingersoll, H. L. Case, J.
ower Sources 101 (2001) 238.
. Broussely, in: Advances In Lithium Ion Batteries, W. van Schalkwijk, B. Scrosati,
ds., Kluwer/Plenum, New York, 2002, 393.
. Kobayashi, H. Miyashiro, K. Kumei, K. Takei, T. Iwahori, I. Uchida, J. Electrochem.
oc.l49 (2001) A978.
Chapter 5
M. Koltypin, Y. S. Cohen, B. Markovsky, Y. Cohen, D. Aurbach, Electrochem. Comm. 4

(2002) 17.
B. Simon, J.P. Boeuve, M. Broussely, J. Power Sources 43-44 (1993) 65.
J. M. Tranchant, R. Messina, J. Electrochem. Soc. 143 (1996) 391.
R. Yazami, Y. F. Reynier, Electrochim. Acta 47 (2002) 1217.
E. Peled, in: Rechargeable Lithium and Lithium-Ion Batteries, S. Megahed, B. Barnett, L.
Xie, Eds., The Electrochemical Society, Pennington NJ, 1995, PV 94-28, I.
M. Jean, A. Tranchant, R. Messina, J. Electrochem. Soc. 143 (1996) 391.
M. Jean, A. Chausse, R. Messina, J. Power Sources 68 (1997) 232.
G. Amatucci, A. Du Pasquier, A. Blyr, T. Zheng, J.-M. Tarascon, Electrochim. Acta 45
(1999) 255.
M. Broussely, S. Herreyre, P. Biensan, P. Kasztejna, K. Nechev, R. J. Staniewicz, J.
Power Sources 97-98 (2001) 13.
A. Du Pasquier, F. Disma, T. Bowmer, A.S. Gozdz, G. Amatucci, J.-M. Tarascon, J.
K. Edstrom, M. Herranen, J. Electrochem. Soc.147 (2000) 3628.
D. Aurbach, M. D. Levi, E. Levi, H. Teler, B. Markovsky, G. Salitra, U. Heider, L.
Heider, J. Electrochem. Soc. 145 (1998) 3024.
D. Aurbach, B. Markovsky, M. D. Levi, E. Levi, A. Schechter, M. Moshkovich, Y.
Cohen, J. Power Sources 81-82 (1999) 95.
D. Aurbach, J. Power Sources 89 (2000) 206.
G. Nagasubramanian, J. Power Sources 87 (2000) 226.
Q. Wei, W. Lu, J. Prakash, J. Power Sources 88 (2000) 237.
C. H. Chen, J. Liu, K. Amine, Electrochem. Comm. 3 (2001) 44.
K. Amine, C. H. Chen, J. Liu, M. Hammond, A. Jansen, D. Dees, I. Bllom, D. Vissers, G.
Henriksen, J. Power Sources 97-98 (2001) 684.
N. R. Avery, K. J. Black, J. Power Sources 68 (1997) 191.
G. G. Amatucci, J. M. Tarascon, L. C. Klein, Solid State Ionics 83 (1996) 167.
H. Mao, J. N. Reimers, Q. Zhong, U. von Sacken, in: Rechargeable Lithium and Lithium
Ion Batteries, S. Megahed, B. M. Barnett, L. Xie, Eds., The Electrochemical Society,
Pennington NJ, 199S, PV 94-28, 24S.
Y. Shin, A. Manthiram, Electrochem. Solid-State-Lett. S (2002) ASS.
S. Komaba, N. Kumagi, Y. Kataoka, Electrochim. Acta 47 (2002) 1229.
D. Aurbach, B. Markovsky, I. Weissman, E. Levi, Y. Ein-Eli, Electrochim. Acta 45
(1999) 67.
D. Aurbach (Ed.), Nonaqueous Electrochemistry, Marcel Dekker, Basel, 1999.
U. Heider, R. Oesten, M. Jungnitz, J. Power Sources 81-82 (1999) 119.
Y. Ein-Eli, Electrochem. Comm. 4 (2002) 644.
A. N. Dey, J. Electrochem. Soc. 114 (1967) 823.
X. Zhang, J. K. Pugh, P. N. Ross, Jr., Electrochem. Solid-State Lett. 4 (2001) A82.
C. G. Barlow, Electrochem. Solid-State Lett. 2 (1999) 362.
F. Joho, B. Rykart, R. Imhof, P. Novak, M. E. Spahr, A. Monnier, J. Power Sources 81-
82 (1999) 243.
M. Winter, J. 0. Besenhard, P. Novak, GDCh-Monographie, 1996,3,438.
J. 0. Besenhard, I. Kain, H.-F. Klein, H. Mohwald, H. Witty, in: Intercalated Graphite,
M.S. Dresselhaus, G. Dresselhaus, J. E. Fischer, M. J. Moran, Eds., North Holland, New
York, 1983, 221.
B. Marcus, P. Touzain, J. Solid State Chem. 77 (1988) 223.
J. 0. Besenhard, H. Mohwald, J. J. Nicki, Carbon 18 (1980)
399.
P. SchoderbOck, H. P. Boehm, Synth. Met. 44 (1991) 239.
M. Arakawa, J. Yamaki, J. Electroanal. Chem. 219 (1987) 273.
G. Eichinger, J. Electroanal. Chem. 74 (1976) 183.
Z. X. Shu, R. S. McMillan, J. J. Murray, J. Electrochem. Soc. 140 (1993) L101.
Z. X. Shu, R. S. McMillan, J. J. Murray, in: New Sealed Rechargeable Batteries and
Supercapacitors, B. M. Barnett, E. Dowgiallo, G. Halpert, Y. Matsuda, Z. Takehara,
Eds., The Electrochemical Society, Pennington NJ, 1993, PV 93-23, 238.
D. Guyomard, J. M. Tarascon, J. Electrochem. Soc. 139 (1992) 937.
T. Ohzuku, Y. lwakoshi, K. Sawai, J. Electrochem. Soc. 140 (1993) 2490.
D. Guyomard, J. M. Tarascon, US Pat. No. 5192629, 1993
. 0. Besenhard, M. Winter, J. Yang, W. Biberacher, J. Power Sources 54 (1995) 228.
D. Aurbach, Y. Ein-Eli, 0. Chusid, Y. Carmeli, M. Babai, H. Yamin, J. Electrochem.
Soc. 141 (1994) 603.
G. E. Blomgren, J. Power Sources 81-82 (1999) 112.
. Barthel, H. J. Gores, in: Handbook of Battery Materials, J. 0. Besenhard, Ed., Wiley-
VCH, Weinheim, 1999, Part III, Ch. 7.

D. Guyomard, J. M. Tarascon, J. Power Sources 54 (1995) 92.
D. Guyomard, J. M. Tarascon, Solid State Ionics 69 (1994) 222.
D. Guyomard, J. M. Tarascon, Solid State lonics 69 (1994) 222.
D. Aurbach, B. Markovsky, A. Shechter, Y. Ein.Eli, H. Cohen, J. Electrochem. Soc. 143
1996) 3809
D. Aurbach, A. Zaban, A. Schechter, Y. Ein-Eli, E. Zinigrad, B. Markovsky, J.
M. Inaba, Z. Siroma, A. Funabiki, Z. Ogumi, Langmuir 12 (1996) 1535.
M. Inaba, z. Siroma, Y. Kawadate, A. Funabiki, Z. Ogumi, J. Power Sources 1 68 (1997)
21.
. Aurbach, Y. Ein-Eli, J. Electrochem. Soc. 142 (1995) 1746.
. Jeon, M. Inaba, Y. Iriyama, T. Abe, Z. Ogumi, Electrochim. Acta 47 (2002) 1975.
K. Mongaki, J. Power Sources 103 (2002) 253.
. S. Zhang, T. R. Jow, K. Amine, G. L. Henrikson, J. Power Sources 197 (2002) 18.
M. Winter, R. Imhof, F. Joho, P. Novak, J. Power Sources 81-82 (1999) 818.
X. Zhang, R. Kostecki, T. R. Richardson, J. K. Pugh, P. N. Ross, Jr., J. Electrochem. Soc.
148 (2001) A1341.
G.-C. Chung, H.-J. Kim, S.-1. Yu, S.-H. Jun, J. Choi, M.-H. Kim, J. Electrochem. Soc.
147 (2000) 4391.
.-C. Chung, J. Power Sources 104 (2002) 7.
D. Aurbach, M.D. Levi, E. Levi, A. Schechter, J. Phys. Chern. B 101 (1997) 2195.
K. Yamaguchi, J. Suzuki, M. Saito, K. Sekine, T. Takamura, J. Power Sources 97-98
159.
M. Winter, G. H. Wrodnigg, J. 0. Besenhard, W. Biberacher, P. Novak, J. Electrochem.
oc. 147 (2000) 2427.
.-C. Chung, S.-H. Jun, K.-Y. Lee, M.-H. Kim, J. Electrochem. Soc. 146 (1999) 1664.
. H. Wrodnigg, J. 0. Besenhard, M. Winter, J. Electrochem. Soc. 146 (1999) 470.
M. Winter, W. K. Appel, B. Evers, T. Hodal, K.-C. Moller, I. Schneider, M. Wachtler,
M. R. Wagner, G. H. Wrodnigg, J. 0. Besenhard, Chern. Monthly 132 (2001) 473.
. 0. Besenhard, P. Castella, M. W. Wagner, Mater. Sci. Forum 91-93 (1992) 647.
. Ein-Eli, B.Markovsky, D. Aurbach, Y. Carmeli, H. Yamin, S. Luski, Electrochim.
cta 39 (1994) 2559.
. Chusid, Y. Ein-Ely, D. Aurbach, J. Power Sources 43-44 (1993) 47.
. 0. Besenhard, M. Winter, J. Yang, International Workshop on Advanced Batteries
Lithium Batteries), Osaka 1995, 129.
M. Winter, Diploma Thesis, University of MUnster, Germany, 1993.
. Ein-Eli, S. R. Thomas, and V.R. Koch, J. Electrochem. Soc. 144 (1997) 1159.
. Ein-Eli, S. R. Thomas, and V.R. Koch, J. Electrochem. Soc. 144 (1997) L195.
. X. Shu, R. S. McMillan, J. J. Murray, J. Electrochem. Soc. 142 (1995) Ll61.
. X. Shu, R. S. McMillan, J. J. Murray, I. J. Davidson, J. Electrochem. Soc. 143 (1996)
230.
. X. Shu, R. S. McMillan, J. J. Murray, US Pat. No. 5529859, 1996.
. X. Shu, R. S. McMillan, J. J. Murray, US Pat. No. 5571635, 1996.
. Arai, H. Katayama, H. Akahoshi, J. Electrochem. Soc. 149 (2002) A217.
. Winter, P. Novak, J. Electrochem. Soc. 145 (1998) L27.
. Barker, F. Gao, US Pat. No. 5780182, 1998.
. McMillan, H. Slegr, Z. X. Sho, W. Wang, J. Power Sources 81-82 (1999) 20.
. Inaba, Y. Kawatate, A. Funabiki, S. Jeong, T. Abe, Z. Ogumi, Electrochim. Acta 45
1999) 99.
. Jehoulet, P. Biensan, J. M. Bodet, M. Broussely, C. Moteau, C. Tessier- Lescourret,
n: Batteries for Portable Applications and Electric Vehicles, C. F. Holmes, A. R.
andgrebe, Eds., The Electrochemical Society, Pennington NJ, 1997, PV97-18, 974.
. Fujimoto, M. Takahashi, K. Nishio (Sanyo) US Pat. No. 5352548, 1994; B. Simon, J.
. Boeve (SAFT) US Pat. No. 5626981, 1997; J. Barker, F. Gao (Valence) US Pat. No.
712059, 1998; Y. Naruse, S. Fujita, A. Omaru (Sony) US Pat. No. 5714281, 1998.
. Wang, S. Nakamura, K. Tasaki, P. B. Balbuenal. Am. Chern. Soc. 124 (2002) 4408.
. Matsuoka, A. Hiwara, T. Omi, M. Toriida, T. Hayashi, C. Tanaka, Y. Saito, T. Ishida,
. Tan, S. S. Ono, S. Yamamoto, J. Power Sources 108 (2002) 128.
. Aurbach, K. Gamolsky, B. Markovsky, Y. Gofer, M. Schmidt, U. Heider,
lectrochim. Acta 47 (2002) 1423.
. Oesten, U. Heider, M. Schmidt, Solid State lonics 148 (2001) 391.
. Wang, H. Nakamura, H. Komatsu, M. Yoshio, H. Yoshitake, J. Power Sources 74
1998) 142.
Chapter 5
G. H. Wrodnigg, C. Reisinger, J. 0. Besenhard, M. Winter, ITE Batt. Lett., 1999, 1(1)

111.
H. Ota, T. Sato, H. Suzuki, T. Usami, J. Power Sources 97-98 (2001) 107.
G. H. Wrodnigg, T. M. Wrodnigg, J. 0. Besenhard, M. Winter, Electrochem. Comm. 1
(1999) 148.
M. W. Wagner, C. Liebenow, J. 0. Besenhard, J. Power Sources 68 (1997) 328.
M. W. Wagner, J. 0. Besenhard, C. Liebenow, in: Batteries for Portable Applications
and Electric Vehicles, C. F. Holmes, A. R. Landgrebe, Eds., The Electrochemical
Society, Pennington NJ, 1997, PV97-18, 26.
H. Gan, E. S. Takeuchi, Eur. Pat. Appl. No. 513428, 2000.
J. F. Wolf, S.M. Savner, M. Maxfield, L. W. Lawrence, US Pat. No. 4528254, 1985.
T. Mamamoto, A. Hitaka, Y. Nakada, K. Ube, US Pat. No. 6033809, 2000.
H. Gan, E. S. Takeuchi, Eur. Pat. Appl. No. 4510457,2001.
G. H. Wrodnigg, J. 0. Besenhard, M. Winter, J. Power Sources 97-98 (2001) 592.
Y. Ein-Eli, S. F. McDevitt, J. Solid State Electrochem. 1 (1997) 227.
G. Y. Gu, S. Bouviert, C. Wu, R. Laura, M. Rzeznik, K. M. Abraham, Electrochim. Acta
45 (2000) 3127.
F. Coowar, A.M. Christie, P. G. Bruce, C. A. Vincent, J. Power Sources 75 (1998) 144.
B. V. Ratnakumer, M. C. Smart, S. Surampudi, J. Power Sources 97-98 (2001) 137.
Y. Ein-Eli, S. McDevitt, D. Aurbach, B. Markovsky, A. Schechter, J. Electrochem. Soc.
144 (1997) L180.
H. Nakamura, H. Komatsu, M. Yoshio, J. Power Sources 62 (1996) 219.
Y. Ein-Eli, S. R. Thomas, V. Koch, D. Aurbach, B. Markovsky, A. Schechter, J.
Electrochem. Soc. 143 (1996) L273.
Y. Ein-Eli, Electrochem. Solid-State Lett. 2 (1999) 212.
Y. Ein-Eli, S. F. McDevitt, R. Laura, J. Electrochem. Soc. 145 (1998) Ll.
S. Sakai, M. Yamamoto, US Patent 5891588, 1999.
M. C. Smart, B. V. Ratnakumar, V. S. Ryan, S. Surampudi, G. K. S. Prakash, J. Hu, J981h
Meeting of the Electrochemical Society, 2000, Abstract No. 138.
J.-S. Wan, C.-H. Han, U.-H. Jung, S.-1. Lee, H.-J. Kim, K. Kim, J. Power Sources 109
(2002) 47.
Y.-K. Choi, K. Chung, W.-S. Kim, Y.-E. Sung, S.-M. Park, J. Power Sources 104 (2002)
132.
A. Naji, J. Ghanbaja, P. Willmann, D. Billaud, Electrochim. Acta 45 (2000) 1893.
J. 0. Besenhard, K. v. Werner, M. Winter, US Pat. No. 5916708, 1997.
T. Yamamoto, H. Miura, US Pat. No. 5795677, 1998.
W. K. Appel, J. 0. Besenhard, S. Pasenok, M. Winter, G. H. Wrodnigg, Ger Pat No. DE
19 724 709, 1998.
W. K. Appel, J. 0. Besenhard, L. H. Lie, S. Pasenok, M. Winter, Ger. Pat. No. DE 198
58 924, 1998.
K.-C. Moller, T. Hodal, W. K. Appel, M. Winter, J. 0. Besenhard, J. Power Sources 97-
98 (2001) 595.
K.-C. Moller, S. C. Skrabl, M. Winter, J. 0. Besenhard, in: Proc. 3'd Hawaii Battery
Coriference, Hawaii, USA, 2001,228.
L. H. Lie, T. Hodal, K.-C. Moller, G. H. Wrodnigg, W. K. Appel, J. 0. Besenhard,
M. Winter, ITE Batt. Lett., 1999, 1(1), 105.
T. Nakajima, J. Fluorine Chemistry 105 (2000) 229.
K.-C. Moller, H. J. Santner, J. 0. Besenhard, M. Winter, in: Proc. 41h Hawaii Battery
Conference, Hawaii, USA, 2002, 238.
K.-C. Moller, H. J. Santner, W. Kern, S. Yamaguchi, J. 0. Besenhard, M. Winter, J.
Power Sources, in press.
H. J. Santner, K.-C. Moller, J. Ivanco, M. G. Ramsey, F. P. Netzer, S. Yamaguchi, J. 0.
Besenhard, M. Winter, J. Power Sources, in press.
R. Shimizu(NEC), Jap. Pat. JP2000149989, 2000.
M. Yoshio, H. Yoshitake, K. Abe, T. Umeno, K. Fukuda, JJ'h IMLB, Monterey CA,
2002, Abstract No. 9.
S. S. Zhang, K. Xu, J. L. Allen, T. R. Jow, J. Power Sources 110 (2002) 216.
H. Lin, D. Chua, M. Salomon, H.-C. Shiao, M. Hendricksen, E. Plichta, S. Slane,
Electrochem. Solid-State Lett. 4 (2001) A71.
E. J. Plichta, M. Hendrikson, R. Thompson, G. Au, W. K. Behl, M. C. Smart, B. V.
Ratnakumar, S. Surampudi, J. Power Sources 94 (2001) 160.
C.-K. Huang, J. S. Sakamoto, J. Wolfenstine, S. Surampudi, J. Electrochem. Soc. 147
2000) 2893.
X. Wang, E. Yasukawa, S. Kasuya, J. Electrochem. Soc. 148 (2001) Al058.
naceous and Graphitic Anodes 193
. Xu, M. S. Ding, S. Zhang, J. L. Allen, T. R. Jow, J. Electrochem. Soc. 149 (2002)

A622.
. S. Zhang, K. Xu, T. R. Jow, Electrochem. Solid-State Lett. 5 (2002) A206.
K. Xu, S. Zhang, J. L. Allen, T. R. Jow, J. Electrochem. Soc. 149 (2002) A1079.
B. Simon, S. Flandrois, K. Guerin, A. Fevrier-Bouvier, I. Teulat, P. Biensan, J. Power
ources 81-82 (1999) 312.
. Flandrois, A. Fevrier, P. Biensan, B. Simon, US Pat. No. 5554462, 1996.
. Cao, I. V. Barsukov, H. Y. Bang, P. Zaleski, J. Prakash, J. Electrochem. Soc. 147
2000) 3579.
. Guerin, A. Fevrier.Bouvier, S. Flandrois, M. Couzi, B. Simon, P. Biensan, J.
lectrochen. Soc. 146 (1999) 3660.
. Kuribayashi, M. Yokoyama, M. Yamashita, J. Power Sources 54 (1995) 1.
X. Shu, L. D. Schmidt, W. H. Smyrl, in: Rechargeable Lithium and Lithium-Ion
atteries, S. Megahed, B. Barnett, L. Xie, Eds., The Electrochemical Society, Pennington
J, 1995, PV 94-28, 196.
H. Momose, A. Funahashi, J. Aragane, K. Matsui, S. Yoshitake, I. Mitsuishi, H. Awata,
. Iwahori, in: Batteries for Portable Applications and Electric Vehicles, C. F. Holmes,
. R. Landgrebe, Eds., The Electrochemical Society, Pennington NJ, 1997, PV97-18,
76.
. Yamada, H. Tanaka, T. Mitate, M. Yoshikawa (Sharp), US Pat. No. 5595938, 1997.
M. Yamasaki, T. Nohma, K. Nishio, Y. Kusumoto, Y. Shoji (Sanyo), US Pat. No.
888671, 1999.
. Natarajan, H. Fujimoto, K. Tokumitsu, A. Mabuchi, T. Kasuh, Carbon 39 (2001)
409.
. Wang, M. Yoshio, T. Abe, Z. Ogumi, J. Electrochem. Soc. 149 (2002) A499.
. Funahashi, Y. Kida, K. Yanagida, T. Nohma, I. Yonezu, J. Power Sources 104 (2002)
48.
. Yanagida, A. Yanai, Y. Kida, A. Funahashi, T. Nohma, I. Yonezu, J. Electrochem.
oc. 149 (2002) A804.
.-S. Kim, W.-Y. Yoon, K. S. Doo, G.-S. Park, C. W. Lee, Y. Muakami, D. Shindo, J.
ower Sources 104 (2002) 175.
. Suzuki, T. Hamada, T. Sugiura, J. Electrochem. Soc. 146 (1999) 890.
. Suzuki, T. ljima, M. Wakihara, Electrochim. Acta 44 (1999) 2185.
M. E. Spahr, H. Wilhelm, F. Joho, J.-C-. Panitz, J. Wambach, P. Novak, N. Dupont
avlovsky, J. Electrochem. Soc. 149 (2002) A960.
. E. Spahr, H. Wilhelm, F. Joho, P. Novak, ITE Letters on Batteries, New Technologies
Medicine, 2002,2839, B53-370-73.
. Pan,. K. Guo, L. Wang, S. Fang, J. Electrochem. Soc. 149 (2002) A1218.
. Veeraraghavan, J. Paul, B. Haran, B. Popov, J. Power Sources 109 (2002) 377.
. Takamura, M. Kikuchi, H. Awano, T. Ura, Y. Ikezawa, Mat. Res. Soc. Symp. Proc.
93 (1995) 345.
. Yu, J. A. Ritter, R. E. White, B. N. Popov, J. Electrochem. Soc. 147 (2000) 2081.
. K. Lee, D. H. Ryu, J. B. Ju, Y. G. Shu!, B. W. Cho, D. Park, J. Power Sources 107
(2002) 90.
. Veeraraghavan, A. Durairajan, B. Haran, B. Popov, R. Guidotti, J. Electrochem. Soc.
49 (2002) A675.
. Suzuki, M. Yoshida, C. Nakahar, K. Sekine, M. Kikuchi, T. Takamura, Electrochem.
olid State Lett. 4 (2001) Al.
. Aragane, K. Matsui, H. Andoh, S. Suzuki, H. Fukada, H. Ikeya, K. Kitaba, R.
shikawa, J. Power Sources 68 (1997) 13.
. Nishimura, H. Honbo, S. Takeuchi, T. Horiba, M. Oda, M. Koseki, Y. Muranaka, Y.
ozono, H. Miyadera, J. Power Sources 68 (1997) 436.
. Momose, H. Honbo, S. Takeuchi, K. Nishimura, T. Horiba, Y. Muranaka, Y. Kozono,
. Miyadera, J. Power Sources 68 (1997) 208.
. Y. Lee, R. Zhang, Z. Liu, Electrochem. Solid State Lett. 3 (2000) 167.
. Y. Lee, R. Zhang, Z. Liu, J. Power Sources 90 (2000) 70.
. X. Wang, J.-H. Ahn, M. J. Lindsay, L. Sun, D. H. Bradhurst, S. X. Dhou, H. K. Liu, J.
ower Sources 97-98 (2001) 211.
. Tarnai, S. Matsuoka, M. Ishihara, H. Yasuda, Carbon 39 (2001) 1515.
. Egashira, H. Takatsuji, S. Okada, J. Yamaki, J. Power Sources 107 (2002) 56.
. X. Chen, J. Y. Lee, z. Liu, Electrochem. Comm. 4 (2002) 260.
. Ulus, Y. Rosenberg, L. Burstein, E. Peled, J. Electrochem. Soc. 149 (2002) A635.
. Kikuchi, Y. Ikezawa, T. Takamura, J. Electroanal. Chern. 396 (1995) 451.
. Takamura, M. Kikuchi, J. Ebana, M. Nagashima, Y. Ikezawa, in: New Sealed
Chapter 5
Rechargeable Batteries and Supercapacitors, B. M. Barnett, E. Dowgiallo, G. Halpert,

Y. Matsuda, Z. Takehara, Eds., The Electrochemical Society, Pennington NJ, 1993, PV
93-23,229.
T. Takamura, H. Awano, T. Ura, K. Sumiya, 1. Power Sources 68 (1997) 114.
H. Buqa, Ch. Grogger, M. V. Santis Alvarez, J. 0. Besenhard, M. Winter, 1. Power
Sources 97-98 (2001) 126.
H. Buqa, P. Golob, M. Winter, J. 0. Besenhard, 1. Power Sources 97-98 (2001) 122
H. Buqa, R.I.R Blyth, P. Golob, B. Evers, I. Schneider, M. V. Santis Alvarez, F. Hofer,
F. P. Netzer, M. G. Ramsey, M. Winter, J. 0. Besenhard, lonics 6 (2001) 172.
W. Lu, D. D. L. Chung, Carbon 39 (2001) 493.
T. D. Tran, L. C. Murguia, X. Soong, K. Kinoshita, in: Batteries for Portable
Applications and Electric Vehicles, C. F. Holmes, A. R. Landgrebe, Eds., The
Electrochemical Society, Pennington NJ, 1997, PV97-18, 974.
G. G. Tibbetts, G.-A. Nazri, B. J. Howie (GM), US Patent 5,587,257, 1996.
Y. Bito, S. Ito, H. Murai, M. Hasegawa, Y. Toyoguchi (Matsushita), US Pat. No.
5580538, 1996.
Y. Ein-Eli, V. R. Koch, 1. Electrochem. Soc. 144 (1997) 2968.
Y. P. Wu, C. Jiang, C. Wan, R. Holze, Electrochem. Comm. 4 (2002) 483.
D. Bar-Tow, E. Peled, L. Burstein, 1. Electrochem. Soc. 146 (1999) 824.
T. Tran, B. Yebka, YX. Song, G. Nazri, K. Kinoshita, D. Curtis, 1. Power Sources 85
2000) 269.
K. Zaghib, G. Nadeua, K. Kinoshita, 1. Power Sources 97-98 (2002) 97.
P. Novak, R. Imhof, 0. Haas, Electrochim. Acta 45 (1999) 351.
D. Aurbach, Z. Lu, A. Schechter, Y. Gofer, H. Giszbar, R. Turgeman, Y. Cohen, M.
Moshkovich, E. Levi, Nature 407 (2000) 724.
. 0. Besenhard, M. Winter, ChemPhysChem 2 (2002) 155.
ter 6
RAPHITE-ELECTROLYTE INTERFACE IN
LITHIUM-ION BATTERIES
1 2 2
M.Nazri , B.Yebka and G.-A. Nazri
iversity of Windsor, Department of Chemistry, Windsor, Ontario N9B 3P4, Canada

2. GM Research and Development Center, Warren, Michigan 48090, U.S.A.
INTRODUCTION
The recent commercialization of advanced lithium batteries is mainly
the breakthrough in the stabilization of the anode-electrolyte interface.
gh metallic lithium has an energy density (3860 mAhlg) higher than
other alternative anodes, its poor performance and safety issues
to the low melting point of lithium (180 OC), dendritic growth during
deposition (charge), and high reactivity toward the electrolytes have
ed the commercialization of rechargeable lithium-anode batteries.
There have been several approaches to solve the problem of lithium
The development of lithium alloys, particularly the binary and ternary
has received considerable attention (see Chapter 9 of this book).
er, the performance of these alloys is unsatisfactory mainly due to the
olume change (100-200%) during lithiation and delithiation processes.
xpansion and contraction processes may cause the alloy particles to
nd lose contact with the electrode substrate. In addition, the lithium
with high concentration of lithium are very reactive toward the
lytes and cause decomposition. Therefore a problem similar to that of
tallic lithium also exists for Li-alloy anodes. Recently, some success
en made using intermetallic alloys such as Cu 6Sn5 that insert lithium
tically over a wide composition range LixCu 6Sn5 (0<x<l3).1 Despite
d volumetric energy density of these alloys, their gravimetric energy
is poor and there is a significant capacity loss during multiple
.
The early transition metal oxide anodes have also been proposed for
tion in rechargeable lithium batteries -in particular, the use of lithium
lated titanium oxide.2 -7 The lithium titanium oxide anode is described
Chapter6
ero-strain insertion anode owing to negligible volume change during

hium insertion/extraction process. In addition, titanium oxide does not
pose the organic electrolytes, hence it can be considered as a safe
for rechargeable lithium batteries. The drawback of the lithiated
8
m oxide is its high redox potential (1.5 V vs. Li). Ohzuku et a/. have
very good reversibility and electrochemical performance for
012.
Lithium transition metal nitrides and phosphides also have been
9 15
ered as anodes for rechargeable lithium batteries. - The voltage
s of nitrides and phosphides during charge-discharge are between 0 to
vs. Li. There is also a large voltage hysteresis between the charge and
rge of nitrides and phosphides. The principal mechanism is believed to e
formation of lithium nitride and phosphide during reduction of ion
metal ions to the metallic state. The nano-scale transition metal rs
formed in this process are catalytically active, and during lithium
tion reaction they tend to form the original compound. The slow
cs and large hysterises between charge and discharge is a major
ack for the nitride and phosphide anodes.
In early 1990, a breakthrough was announced by researchers from
Energytech: an anode based on lithium intercalated carbon or graphite
hargeable lithium battery was used. 16 The volume change during full
m intercalation in graphite (LiC6) is less than 9% according to the in
-ray diffraction study. The specific capacity of the lithiated graphite is
iC6 (370 mAh/g), much lower than that of metallic lithium. In spite of
wer energy density, the current rechargeable lithium battery (Li-ion
) technology uses a lithiated carbonaceous anode. 17 -21 The success of
C6 anode is mainly due to the fast stabilization of the anode/electrolyte
ce and the high electrochemical reversibility of lithium
lation/deintercalation process. The LiC6 anodes have been charged and
rged over 1000 times without an appreciable capacity loss. The Li-ion
is currently replacing all other battery types in the market, particularly
field of electronics, computers, and communications. In recent years,
plication of the Li-ion batteries for more power demanding applications
as power tools, stationary, and transportation has started. The
tion of Li-ion batteries in 2002 has exceeded 500 million cells, with
as the main producer (over 70%).
Lithium intercalation in graphite is not new and has been fully studied
ysicists in search of high Tc superconductors during the last 50
2 23
• The hexagonal structure of graphite in ABAB arrangement is
in Figure 6.1. The layer B is shifted with respect to the A layer, but
nd BB layers are on exact registry with each other. When lithium
lates between the layers, the layer B glides to make an exact AAAA
re.
e-Electrolyte Interface in Lithium-Ion Batteries 197
Graphite has an ideal layered structure with weak van der Waals
g between the graphene layers. Thus, Li can be inserted between the
6 7
of graphite. According to recent Li- and Li- NMR studies, the nature
um in graphite is partly ionic. There is partial charge transfer from
to the n-orbitals of the graphene planes. The lithium in the graphite
so a high mobility, which is ideal for applications in high-power
s. In addition, the graphite has very high electronic conductivity,
electrode resistance is very low. Natural graphite is abundant, and
tic graphite also can be made from a variety of precursors. Heat
nt of various organic polymers, particularly those with conjugated
e rings, goes through a carbonization process around 1200-1500 OC
aphitization around 2300-3000 OC. During graphitization, the rate of
in a and c directions is strongly temperature dependent. Therefore,
e flakes with extended a or spherical with extended c directions can
de for different applications. The ideal graphite structure consists of
I. Schematic of graphite and lithium intercalated graphite.
of carbon atoms arranged in hexagonal rings that are stacked in an

sequence (Figure 6.1). The graphite comprises planar hexagonal
ks of carbon atoms (honeycombs) with two different crystalline forms,
nal (2H) and rhombohedral (3R). The 2H graphite structure is the
ommon and the carbon layers are arranged in the ABAB sequence,
Chapter6
the B layers are shifted to a registered position with respect to the A

. In the 3R (rhombohedral) structure, the stacking sequence is
BC, where the C layers are also shifted by the same distance with
t to the B layers, as the B layers are shifted with respect to the A
24
.
The separation of the layers in well-graphitized sample is 3.35 A, and
are held together with weak van der Waals bonds. The softness and the
ty of graphite can be attributed to the easy glide of these layers over
nother. Within each graphene layer only three carbon atoms surround
carbon atom. After forming one cr bond with each neighbor, each
atom would still have one electron and these are paired up into a
25
of 1t bonds.
EXPERIMENTAL SET UP TO EVALUATE

SURFACE REACTIONS OF GRAPHITE ANODES
Graphite Types
Several natural and synthetic graphite samples commercially available
4, BG-35, CPC, CN-39, CN-39A, SFG-15) with different morphology,
.2. SEM morphology of natural graphites BG34, BG35, KS44, and synthetic graphite
CPC, CN-39, CN39A, and SFG-15.
-Electrolyte Interface in Lithium-Ion Batteries 199
size, and surface area have been examined. Figure 6.2 shows the
ology of these graphite samples.
1. Surface area of various commercial graphite samples.
raphite 2
ace area (m )
The surface area of the graphites is listed in Table 6.1. These samples
ow surface area and, so, may be suitable for battery applications.
e of the high reactivity of lithiated graphite, lower surface area
s are preferred.
The X-ray diffractions of the graphite samples listed above are shown
re 6.3. The high and narrow 002 peak is an indication of the good
nity of the samples.
/ 0
0
N
0
......
......
MCMB.10
l 1L ± A
BG·34 A
/ SFG-15 Jl /
L KS·44 A /
L N·39A A /
/ CN·39 A /
/ CPC J\. /
/ BG·3S A /
/
10 20 30 40 so 60 70 80
2 Teta (degree)
.3. X-ray diffractions of various graphite samples.

Chapter 6
A General Procedure to Fabricate Carbon Anodes

and their Evaluation
The graphite samples need to be heated in vacuum between 700-900
remove moisture and attached functional groups, which usually exist at
ge of graphene sheets. It is very important to remove the water
ules trapped in graphite before using it in a lithium battery. Previous
rature program desorption (TPD) study has shown that almost all
e functional groups can be removed at 700-900 °C. The heat treated
te is then mixed with 8-10 wt% conductive diluent carbon, and to this
e 4-5 wt% EPDM binder (in xylene) is added to make a slurry. The
e of graphite sample, diluent carbon, and binder is ball milled using a
ercial micronizer. The slurry is spread on the surface of a copper foil
ried to remove xylene. To improve the adhesion of the sample to
, the sample is hot pressed inside of a dry box at lOOOC. The coating
s to make an electrode for lithium cells is shown in Figure 6.4.
rbon
EPDM Sam pie
ack binder Carbon
powder
.1 .
Mixture
Micronizer
..
. 4.
Coa t ing
Electrode
.4. Coating process to make electrodes for lithium cells. Anode is metallic lithium, the
is the graphite electrode, and reference electrode is lithium wire
in the cell between two layers of porous separator.
From the coated foil, electrodes are cut to make the lithium cells. Two
2
of cells can be constructed. A lOx10 cm electrode can be used to
the surface reactions and electrolyte decomposition using a GC-MS
. Figure 6.5 shows the structure of the cell used for measuring gaseous
s formed during electrolyte decomposition. The graphite electrode is
ted from the metallic lithium electrode by a porous polyethylene
opylene separator soaked with electrolyte. The cell is housed in a
hylene coated aluminum pouch.
Cathode material ( IO x iO cm 2 ) t ToMS
.5. Schematic of Li/graphite cell for in-situ GC-MS analysis of electrolyte

osition during lithium intercalation in graphite electrodes.
2
A 5 cm disc electrode can be used for evaluation of capacity and cycle
sts. Figure 6.6 shows the assembly of this cell. The electrolyte
position occurs on the surface of graphite during initial lithium
lation. In order to investigate the nature of the evolving gas, a special
sembly is constructed to allow the generated gas to escape from the
d be sampled by in-situ GC-MS analysis. Before sealing the cell
in Figure 6.5, a flexible 2 mm diameter tube is inserted in the cell. The
s connected to an on-line GC-MS system. Figure 6.7 shows the
ned GC-MS equipment used for in-situ analysis of the gaseous
ts.
Chapter6
Anodc(CE)
Elc:cuolyte Lithium (RE)
Cathode (WE)
CurTcnt collector
.6. Schematic of lithium cell for evaluation of graphite anodes. WE= working
e, CE =counter electrode andRE= reference electrode.
PLOTTER
.7. Schematic of on-line GC-MS for analysis of gases generated during lithium
tion in graphite anodes due to electrolyte decomposition.
The Lilgraphite cells, after charging and discharging cycles, are

mbled and the electrode is rinsed to remove excess salt. The IR
of the surface film is then obtained using a diffuse reflectance
ory that allows to install the sample inside the dry box and later
rred to the IR spectrometer for analysis.
In order to understand the thermal properties of the surface film, a
is collected from the surface of the graphite electrode (inside the dry
nd placed in a hermetically sealed DSC pan.
GRAPHITE CYCLING AND FILM FORMATION

The lithium intercalation/deintercalation process between the
ne planes of graphitic samples is facile. The high lithium storage in
e with a high lithium mobility, has provided a unique anode for
eable lithium batteries. The voltage of the lithiated graphite is very
o the voltage of metallic lithium (100-300 mV vs. Li). In addition, the
profile of the graphite anode during the lithium
lation/deintercalation process is fairly flat with a very small hysteresis
The graphite can accommodate one lithium per six carbons to form the
ed "stage one" (LiC6) compound. High-pressure intercalation synthesis
lded a lithium rich phase, LiC3. However, the LiC3 phase is not stable
ient pressure. It is interesting to note that the amount of lithium in 1
LiC3 is higher than that of the pure metallic lithium.
The lithium intercalated graphite is highly reactive and reacts with the
lyte solution. The by-product of the electrolyte decomposition forms
film on the surface of the electrode. The electrolyte decomposition
lithium intercalation occurs at about 0.8 V vs. Li. In parallel with the
lyte decomposition, lithium intercalation in graphite also proceeds.
m formed on the graphite surface, called solid electrolyte interface
yer), is electronically insulating and ionically conducting. The SEI
rotects the lithiated graphite from further reaction with electrolyte.
ore, the electrolyte decomposition stops in subsequent charge
ge cycles. The nature of the film formed on the surface of the
aceous anodes depends on the composition of electrolyte. Ethylene
ate based electrolytes tend to form a good compact SEI layer, and in
ommercial lithium batteries the electrolyte contain EC in addition to
carbonate solvents. The use of electrolyte with only propylene
ate solvent has not been successful, due to the co-intercalation of
and exfoliation of the graphitic anode. In practice, usually EC in
ation with other linear carbonate solvents such as dimethyl carbonate,
carbonate, or methyl ethyl carbonates are used. Figure 6.8 shows the
profile of a graphitic anode during the lithium intercalation process.
Chapter6
-0.5
·1.11
· .. :.:.:.:.... . . .
3.0 ... :k .-··
... -1.5
r\ ·.....,
.".c 1.5 r=
·0 10m0 ,._, 0.6 0.7 0.8 0,9 1.0
1.1 1.2
=
u"
=I Cell voltage (volts)
0.5 1=0.50mA 1=0.025mA

0.0 t
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
6.8. Voltage profile of graphite anode as a function of lithium concentration at 0.025

SmA, and O.lmA constant current charging.
The voltage plateau at about 0.8 volt is due to the electrolyte

position. It is important to note that when charging the electrode at a
current density, a more protective film is formed, and less electrolyte
position occurs. The voltage plateaus between the 0.3-0.01 V are an
tion of the staging phenomena usually observed for the formation of
LiC12, and other diluted phases of LiCn (n = 24, 36, etc). The insert in
6.8 shows the derivative of the voltage profile for the electrolyte
position part of the graph. The first charge-discharge of various
tic electrodes is shown in Figure 6.9. After lithiation in the first cycle,
ectrode/electrolyte interface is stable and does not react with the
lyte. The voltage of the graphite drops to very low value (300 mV vs.
intercalation of a small amount of lithium (Li0.05C6).
All samples have shown some degree of electrolyte decomposition
the first lithium intercalation. The decomposition plateau for BG-35 is
than for the other samples. In samples with a high degree of
llinity, the staging phenomenon is observed with better resolution, as
or the case of highly crystalline KS-44.
The amount of electrolyte decomposition also depends on the
sition of electrolyte. Figure 6.10 shows the voltage profile of an
de made from CPC during lithium intercalation in different electrolyte
sitions. The electrolyte containing EC-DEC-LiPF6 shows less
lyte decomposition. This suggests that the electrolyte containing cyclic
ne carbonate and linear carbonate may provide a safer operation for the
battery. The interface of graphite electrode is stabilized after
cal prelithiation or during the first charge-discharge cycle.
.9. The voltage profile of graphitic electrodes (BG-35, CN-339, CPC, BG-34, CN-
d KS44).
le charge-discharge cycles of a prelithiated graphite anode are shown

ure 6.11. The charge-discharge cycles are very reversible and there is
of an electrolyte decomposition (plateau at 0.8V). At low charging
3.5 S2/CPC
3.0
2.5
V) EC-DEC-LiPF6
2.0
PC-LiPF6
1.5
EC-PC-LiPF6
1.0
0.5
0.0
0 100 200 300 400 500 600
Q (mAh/g)
.10. Voltage profile of CPC graphite anode in different electrolytes.

Chapter6
apacities close to theoretical one (372 mAhlg) are achievable . The rate
ch the electrode can be charged depends also on the amount of active
als loaded on the substrate. For the design of a high rate (power)
0
0 10 20 30 40 50
Time (hours)
.11. Multiple charge-discharge cycles of a graphite electrode after stabilization of the

e/electrolyte interface and formation of the SEI layer.
, the coating is thinner that for the electrode designed for high
applications. The usual loading for high power battery is about 10-20
2
and for the high-energy cells the loading may exceed 100 mg/cm2.
Disordered carbons have also been used as anodes. However, their
ty is almost half that of the graphitic anodes, and their voltage profile
lithium intercalation does not show the staging processes often
ed for the graphitic anodes. Some special carbon anodes with extended
layer graphene sheets, store lithium to a higher level than the
tical capacity. In this case, the lithium is being adsorbed on both sides
graphene sheets and in micropores. However, the cycle life of this type
bon is poor and there is significant capacity loss during multiple
-discharge cycles. In addition, there is a considerable amount of
e hysterises between charge and discharge. The sloping charge
rge profile also suggests that there are various sites for lithium with
nt affinities.
Gas Generation on Graphites

The structure and degree of graphitization of the carbon electrode
critical role in the performance and safety of Li-ion batteries. The
effort is directed to understanding the relationship between the
ochemical properties of carbonaceous materials and their
2. Gas Chromatography and Mass Spectrometry analysis of the various gases

d on the surface of different graphite anodes during the first lithium intercalation
Graphite Electrolytes
BG-35
CPC
CN-39
KS-44
SFG-15
BG-34
chemical performance. The crystallinity and surface area of the

aceous materials, and the nature of the electrolyte used are important
for the electrochemical stability of Li-ion cells and for the nature and
of gases generated during the lithium insertion/extraction process
ble 6.2). Graphite anodes with high degree of crystallinity generate
during the first lithium intercalation cycle.
Chapter6
Careful analysis of various carbonaceous anodes indicates that a gel

ilm is also forming on the electrode surface as a result of electrolyte
position. The nature of this film has a significant impact on the rate
lity and the overall cell impedance.
. Infrared Spectroscopy of SEI Layer on Graphite

Anodes
The film formed on the surface of graphite anode during electrolyte
position can be probed using IR spectroscopy. After initial charge
rge, the cell is opened in the dry box, and a disc is cut from the
de. The sample is then placed in a diffuse scattering IR accessory and
inside the dry box and transferred to the IR spectrometer. The IR
a of non-cycled graphite electrodes were used as a reference. The
a of the film formed on the surface of BG-34 graphite anode after one
arge-discharge cycle in DMC-LiPF6 is shown in Figure 6.12. The
ex IR spectrum, with several absorption peaks, indicates formation of
anic moiety different from the electrolyte used. The peaks around 870-
d 1456 cm- 1 are assigned to the formation of LhC03• There is also an
tion of the formation of lithium-organic compounds such as
1
COLi with an absorption peak around 1600-1630 cm- assigned to the
tretch.
As we observed in the analysis of gaseous species generated when
is present in the electrolyte, the formation of CH2=CH2 also suggests
lowing decomposition reactions:
2Li+ + 2e- + CH30COOCH3 -LhC03 + CH3 -CH3

2Li+ + 2e- + CH30COOCH3 - CH30COLi + CH30Li
CH30COLi + CH30Li + residual H20 -LhC03 + CH30Li +
C02.
Table 6.3 reports the vibrational frequencies in DMC + 1M LiPF6. This

e following tables are self-explanatory.
e-Electrolyte Interface in Lithium-/on Batteries 209
BG-34
; &: DMC-LiPF•
1800 1600 1400 1200 1000 800 600 400
Wavenumber (cm"1)
.12. IR spectrum of the film formed on the surface of graphite anode during
intercalation in (DMC + 1M LiPF6) electrolyte.
3. Vibrational freguencies of the surface la er on BG-34 sra£hite in DMC-LiPF6•

Assignment
served (cm·1) v assigned Bands Compounds
1627br C02 stretch
1456br 1420-1470 Carbonate ion
1441 LhC03
1255SJ2 1280 Dimethz:1 Carbonate
1190m CO stretch
1160m 1150-1170 CH stretch LiOCH3
1050br 1050-1070 CH stretch LiOCH3
1020-1090 CO stretch Carbonate ion
928SJ2 969-914 Dimethz:1 Carbonate
890m
875sh 820-890 Carbonate ion
866 LhCOJ
833m 820-890 Carbonate ion
846 co3 bend LhC03
SlOsh
719sp 680-750 Carbonate ion
738 C02 asym. Li,C03
bend
539m 550-650 LiO stretch LiOCH3
eak, m: medium, s: strong, sh: shoulder, br: broad, v: very, sp: sharp)
Chapter6
TheIR spectrum of the film formed on BG-34 in EC-DMC(1:1mol) +

is more complex, indicating significant involvement of EC
position on the surface of lithiated graphite (Figure 6.13). The peaks
ponding to the formation of organa-lithium compounds as well as
1
3 are present. The peaks at 844 and 1
1483 cm- are assigned to the
m carbonate. The peak at 1083 cm- can be assigned to CH30Li, and
C02Li, various C-H bending of CH3 and CH2 groups.
1.0 .------.:---------------------:J
BG-34
EC-DMC-LiPF6
0.8
0.6
0.4
0.2
o
2000 1800 1600 1400 1200 1000 800 600 400
Wavenumber (cm-1)
6.13. IR spectrum of the film formed on lithiated graphite during the first lithium
ation process in EC-DMC (1: I mol) + 1MLiPF6 electrolyte.
In Table 6.4 the vibrational frequencies in EC-DMC-LiPF6 are

ed.
The IR spectrum of film formed on graphite during first lithium
lation in EC-DEC (1:1mol) + 1MLiPF6 is shown in Figure 6.14. The
ex spectrum with several absorption peaks shows decomposition by
cts of the EC and DEC solvents. The richness of the spectrum indicated
esides LizC03( solid), CH2=CH2(gas)• and CHrCH3(gas) there are various
organa-lithium compounds present. The peaks at around 700-800 cm-1
atched with the bending mode of Oc03 in ROC02Li. Because of the
ous nature of the film formed on the surface of the graphite, there is a
ility of ring opening decomposition of EC and formation of organa
dimers or trimers of EC (similar to the ring opening polymerization).
te-Electrolyte Interface in Lithium-Ion Batteries 211
.4. Vibrational frequencies on the surface layer on BG-35 graphite in EC-DMC-LiPF6.
Assignment
observed (cm'1) v assigned bands Compounds
1958vw
1866w 1870-1750 Organic Carbonate
1870 Ethylene Carbonate
1780sp 1870-1750 Organic Carbonate
1760 Dimethyl Carbonate
1737sh 1870-1750 Organic Carbonate
1638sh 1629 LiPF/
1566sh 1580 0-C-0 stretch RCOOLi
1483sp CH2 bend
1414sp 1441 LhCOJ
1420 LhCOJ
1369sh 1380-1360 CHbend LiOCH3
1083sp 1089 CO stretch Li,C03
1085 CO stretch RC03Li
1083 LiPF.'
977m 974 Ethylene Carbonate
909s
844br 846 C03 bending Li,C03
858 Dimethyl Carbonate
850 LiP
832 LiPF.'
780sp 745-789 LiO
730sp 738 C02 asym. Li,C03

740 bend Li,C03
C02 asym.
bend
559sp 550 LiP
557 LiPF6'
552 LiO stretching Li202
: weak, m: medium, s: strong, sh: shoulder, br: broad, v: very, sp: sharp)
The peak assignments for SEI layer formed on the surface of graphite
the first lithium intercalation are listed in Table 6.5. The peaks
pond to the formation of residual lithium carbonate with peaks around
1
cm- , formation of various organa-lithium compounds as well as peaks
C=O stretching vibrations at 1650-1800 cm-1 •
The IR spectrum of the film formed on the surface of graphite anode
lithium intercalation in PC-LiPF6 solution is shown in Figure 6.15,
e frequencies and products in Table 6.6. This spectrum is rich in by
cts of the PC decomposition. The IR signals of LizC0 3, around 850 and
1
cm- have also been observed. As already mentioned, the film formed in
se is much thicker than the film formed by EC decomposition, this
ting further electrolyte decomposition upon cycling.
Chapter 6
EC-DEC-LiPF6
3.0
25
05
Wavenumber cm-1
.14. IR spectrum of the film formed on BG-34 in EC-DEC + IMLiPF6 during first
n.
(PC-LiPFti)
-O.Il
-O.Il5
-O.'l
-0.'15
-0.8 1801) lllOCl 141)0 121)1) IC)C)C) (ij)C)
Wavawmher ao. _,
.15. IR spectrum of the film formed on graphite anode in PC-1M LiPF 6.

.5. Vibrational frequencies on the surface layer on graphite in EC-DEC-LiPF6.
Assignment
Observed vassigned bands compounds
1966s
1870 Eth 1ene Carbonate
1809s
1771sE 1750-1870 Orsanic Carbonate
1746m 1750 Diethi:1 Carbonate
1699w
1685w
1652m 1650 C02 stretch RC03Li
1555sp 1580 RCOOLi
1580 LiOH
1505sh 1500 (CHzOLi)z
1479w Eth lene Carbonate
1457w 1450-1480 CHbend LiOCH3
1445 CH, CH3 asy. Ethyl-C03Li
bend
1390w Ethylene Carbonate
1360-1380 CHbend LiOCH3
1374w 1351 COz s m. stretch Eth l-C03Li
1296sh
1267sE 1262 Dieth,rl Carbonate
1195sh CO stretch
1161sp 1162 Ethylene Carbonate
1152 CO stretch Eth,rl-C03Li
1072sp 1074 Ethylene Carbonate
1085 CO stretch Ethyl-C03Li
1100 CO stretch (CHzOLi)z
1017sp 1021 Diethyl Carbonate
1010 CO stretch Eth l-C03Li
973m 974 Eth lene Carbonate
903sh 890 (CHzOLi)z
844sp 858 Diethyl Carbonate
826 co3 bend Ethyi-C03Li
850 LiOH
774sE 774 Eth lene Carbonate
730sh 738
717m 720 C02 as m. bend Eth I-C03Li
666w 660-550 LiO stretch LiOCH3
556sE 500-600 LiO stretch Eth I-C03Li
529w 400-500 LiOH
(w: weak, m: medium, s: strong, sh: shoulder, br: broad, v: very, sp: sharp)
The IR spectrum of the film formed on the surface of lithiated graphite

DMC-LiPF6 is shown in Figure 6.16 (peak assignments in Table 6.7).
lm formed with electrolyte containing PC was always thicker and
in PC decomposition by-products. It is also observed that the film
Chapter6
d from solutions containing EC is also rich in EC decomposition by

ts.
The IR spectra of surface films formed on lithiated graphite anodes in
ns containing both the cyclic and the linear carbonates have been
d. The composition of the film is dominated by the decomposition by
ts of the cyclic carbonates. This observation may suggest that the
carbonates that selectively solvate the u+ in the electrolyte, carry their
on shell to the graphite surface, and preferentially saturate the
de-electrolyte interface with the cyclic carbonate. This observation
xplain the beneficial effect of EC in commercial electrolytes for
batteries.
6. Vibrational frequencies on the surface layer on graphite in PC-LiPF6.
Assignment
observed (cm'1) v assigned bands Compounds
1862sh
1817s
1785sh 1760 Dimethyl Carbonate
1732w
1715br
1630w 1629 LiPF6
1588sh 1580 RCOOLi
1500br
1430br 1441 LhC03
1360br COz sym.
stretch
1338br
1289br 1280 Dimethyl Carbonate
1250w 1263 LiPF6'
1233w
1134m 1122 LiPF6
1088m 1089 CO stretch LhCOJ
1083 LiPF6'
1065w 1060 LhO
960w 969-914 Dimethyl Carbonate
865sp 866 C03 bend LizC03
832 LiPF6'
870 LhO
866 LiO
791sp 793 Dimethyl Carbonate
789-745 LiO
734br 738 COz asym. bend LhCOJ
69lsh 730-689 LiO
562sp 557 LiPF6
650-550 LiO stretch LiOCH3
(w: weak, m: medium, s: strong , sh: shoulder, br: broad, v: very, sp: sharp)
e-Electrolyte Interface in Lithium-/on Batteries 215
CPC
PC-DMC-LiPF
I
0
":!':
:;::
:!:
l::i:!:
...
000 1800 1600 1400 1200 1000 800 600 400

Wawnumber (c:m-1)
6.16. IR spectrum of the film formed on the surface of graphite anode during lithium
ation in PC-DMC (l:lmol) +1M LiPF6.
7. IR frequencies of the surface layer on cycled CPC in PC-DMC- LiPF6.

Assignment
1
observed (cm" ) v assigned bands Compounds
1861sh 1750-1870 Organic Carbonate
1780br 1760 Dimethyl Carbonate
1590w 1630-1590 Carbonate ion
1555sp 1580 RCOOLi
1480sp 1480-1450 CHbend LiOCH3
1455sp 1450-1480 CHbend LiOCH3
1441 LhCOJ
1389sp 1380-1360 CHbend LiOCHJ
1254w 1250-1280 Organic Carbonate
1223w
1120s
1076w 1070-1050 CH stretch LiOCH3
1089 CO stretch LhC03
1045br
973sp 970-1020 Organic Carbonate
891sp 866 co3 bend LhCOJ
847sp 826 co3 bend Ethy-C03Li
846 co3 bend LhC03
774sp 774-793 C02 asym. Organic Carbonate
bend
709sp 720 C02asym. Ethy-C03Li
bend
625w 550-650 LiO stretch LiOCH3
537w 529 CO bend
weak, m: medium, s: strong, sh: shoulder, br: broad, v: very, sp: sharp)
Chapter 6
Thermal Analysis of SEI Layer on Graphite Anodes

The thermal characteristics of the surface film formed on the graphite
6
during lithium intercalation are very important for practical reasons?
hium battery may be subject to different thermal conditions during its
ional life. Therefore its thermal safety characteristics must be known.
ful tool for evaluating the thermal properties of films formed on a
te anode is the differential scanning calorimetry (DSC). The graphite
de was charged and discharged in carbonate-based electrolytes, and
rface film formed on the anode was collected for DSC study. Figure
shows the thermal behavior of the surface film in contact with LixC6 ,
2, 0.4, 0.6, 0.8, 1.0).
3.0 LiC6
X
X=O.S
2.5
BG-34 X=0.6
2.0
X=0.4
1.5
X=0.2
1.0
0.5
0.0
Before cycling
-0.5
-1.0 100 150 200 250 300 350 400

50
Temperature ( °C)
.17. Differential scanning calorimetry of the surface film formed on graphite anode at
states of charge (x=0.2-0.8 in LixC6).
The sample has an endothermic peak near the boiling point of DMC,
n the exothermic process starts around 150 OC. A major exotherm was
ed at 220-240 OC. The intensity of the exotherm peak correlates well
e amount of lithium remaining in the graphite. It is very important to
the safety aspects of the Li-ion cells when they are exposed to
atures above 150 oc. Further electrolyte decomposition with massive
neration may occur above 150°C, and cell temperature may rise to
the flash point of the electrolyte components. Therefore care should be

ed dealing with lithium-ion batteries with graphitic anodes,
larly in the charged state (x value close to one in LixC6).
A similar thermal behavior was observed for almost all graphitic
. The temperature-induced degradation of the SEI layers occurs first
120-140 ·c followed by an exothermic reaction and the
position of trapped LiPF6 around 210-230 ·C. We have found a direct
tion between the BET surface area of the graphite anode, amplitude of
C exothermic peak and the irreversible capacity losses measured for
ells. Figure 6.18 shows the correlation between the volume of gas
ted on the surface of lithiated graphite anodes and the surface area of
phite material used.
t'iO
50
i .w
; 3:0
20
10
•-
0 2
" Surface area (m 'fg)
6 ll 10 12 14
.18. The correlation between the volumes of gas evolved during lithium intercalation
te anode and the surface area of the graphite. Upper line: PC-DMC; middle: EC
wer: EC-DEC (all with LiPF6).
CONCLUSIONS
he following general conclusions can be made from this study.
Organic carbonates are not electrochemically stable over a lithiated
raphite electrode. However, the decomposition by-products formed on
he surface can protect the electrode, and the electrolyte decomposition
s eliminated in subsequent charge-discharge cycles.
The nature of the film formed on the surface of lithiated graphite
epends on the composition of electrolyte, and in the case of
Chapter 6
electrolyte made of organic carbonates, the by-products are organa

lithium compounds as well as some lithium carbonate.
The composition of the film is dominated by the by-products of cyclic
carbonate decomposition, and less by the linear carbonate
decomposition.
The film formed from the decomposition of cyclic ethylene carbonate
is denser and more protective as compared with films formed from
other carbonates.
The gaseous species formed during the first charge discharge are
mostly flammable and care should be taken when preparing large
lithium cells.
The thermal analysis has shown that the film formed on the surface of
lithiated graphite is not stable above 150 OC. If lithium cells are
exposed to heat, or if internal heat exceeds 150 °C, the protective film
could be destroyed and further electrolyte decomposition of the freshly
exposed surface of lithiated graphite may occur. Therefore, it can be
concluded that the current lithium-ion cells are not very safe above
150°C, and their application in such conditions must be avoided. In
addition, a safer and more robust anode is needed, particularly for large
size applications of high-energy and high-power lithium batteries.
ERENCES
M. Thackeray, J.T. Vaughey, A.J. Kahaian, K.D. Kepler, R. Benedek, Electrochem.
mm. 1( 1999) 111.
Akimoto, Y. Gotoh, Y. Oosawa, J. Akimoto, Y. Gotoh, Y. Osawa, J. Solid State
emistry 7 (1997) 129.
. Kartha, D.P. Tunstall, J.T.S. Irvine, J. Solid State Chemistry 152 (2000) 397.
E. Arroyo de Dompablo, A. Varez, F. Garcia-Alvarado, J. Solid State Chemistry 153
00) 132.
van de Krol, Thesis, 2000, Universal Press, Veeneddaal, Netherland.
Sodergren, H. Siegbahn, H. Rensmo, H. Lindstrom, A. Hagfeldt, S.E. Lindquist, J.
ys. Chern. B 101 (1997) 3087.
W. Murphy, R.J. Cava, S.M. Zahurak, A. Santoro, Solid State Ionics 9 (1983) 413.
Ohzuku, A. Ueda, N. Yamamoto, J. Electrochem. Soc. 142 (1995) 1431.
Nishijima, Y. Takeda, N. Imanishi, 0. Yamamoto, J. Electrochem. Soc. 141 (1994)
66.
Shodai, S. Okada, S. Tobishima, J. Yamaki, Solid State Ionics 86 (1996) 785.
Suzuki, T. Shodai, Solid State lonics 116 (1999) 1.
.C. Rowsell, V. Pralong, L.F. Nazar, J. Am. Chern. Soc. 123 (2001) 8598.
Pereira, L.C. Klein, G.G. Amatucci, J. Electrochem. Soc. A262 (2002) 149.
Souza, V. Pralong, A.J. Jacobson, L.F. Nazar, Science 296 (2002) 2012.
Pralong, D.C. Souza, L.F. Nazar, Electrochem. Com. 4 (2002) 516.
Nagaura, K. Tozawa, Prog. Battery and Solar Cells 9 (1990) 209.
. Holmes, A.R. Landgrebe, The Electrochemical Society Proceeding 91 (1997) 18.
Ohzuku, A. Ueda, J. Electrochem. Soc. 144 (1997) 2780.
Novack, K. Miller, K.S.V. Santhanam, 0. Heas, Chern. Rev. 97 (1997) 207.

Bicke, W.F. Chu, W. Weppner, Solid State Ionics 1997, 93.
Arakawa, J. Yamaki, J. Power Sources 54 (1995) 250.
A. Safran, Solid State Physics 40 (1987) 183.
E. Ulloa, G. Kirczenow, Comments Cond. Mat. Phys. 12 (1986) 181.
. Dahn, A.K. Sleigh, H. Shi, J.N. Reimer, Q. Zhong, B.M. Way, Electrochim. Acta
(1993) 1179.
A. Cotton, G. Wilkinson, C.A. Murillo, M. Bochmann,.Advanced Inorganic Chemistry,
hn Wiley & Sons. Inc. NY 1999.
N. Richard, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 2068.
apter 7
THE KEY ROLE OF NANOPARTICLES IN

ACTIVITY OF 3d METAL OXIDES TOWARD
LITHIUM
J-M. Tarascon, S. Grugeon, S. Laruelle, D. Larcher and P. Poizot
boratoire de Reactivite et Chimie des Solides, Universite de Picardie Jules Verne and
CNRS (UMR-6007), 33 rue Saint Leu, 80039, Amiens, France
INTRODUCTION
In response to the needs of today's mobile society and the emergence
cological concerns such as global warming, one of the major
nological challenges in this new century is undoubtedly energy
ration and storage. Ninety percent oftoday's electrical power generation
comes from fossil fuels, and we are constantly struggling to reduce the
on dioxide emissions per unit of electric power so as to help curtail
al warming. It is now mandatory that new and environmentally friendly
gy/storage sources be found. Hence, the fast developing research in that
involving, among others, fuel cells, primary and rechargeable batteries,
supercapacitors. As a result of this worldwide ecological priority,
ical concerns have come into play, and science has suffered from
itisation based on both industrial pressure and media reports, rather than
he clear and rigorous scientific identification of technological stoppers
ent in each storage system. Needless to say, this applies to battery
ms as well.
In the past two decades, intensive efforts have given birth to the
argeable Li-ion battery technology that has dominated the market place,
an be regarded as one of the great successes in modern electrochemistry
te. But these Li-based systems still suffer from the lack of suitable
rode and electrolyte materials, which they require if they are ever to
mmodate the increasing user's demands. Aware of this limitation,
ists have been acting at several levels to incrementally improve the Li
erformance. They have followed a dual approach, dealing with either
ive or negative electrode materials, with efforts centered around: 1) the
fication of existing materials through cationic/anionic substitution,
re modification and surface treatments, 2) the making of composite
Key Role ofNanoparticles in the Reactivity of 3d Metal Oxides Towards Li 221
trodes or electrolytes made of several chemical components, and 3) the

gn of new electrode materials. Such approaches were pursued at the
roscopic scale on electrode materials 1- 3 having a dual electronic-ionic
uctivity, a void structure to insert/de-insert Li ions, or the ability to alloy
Li. They led to the identification of layered LiMn 1_xCrx02 oxides4-5 or
e-dimensional iron phosphates (LiFeP04)6, that stand as a possible
native to LiCo02 or negative electrode materials such as tin-based
es (Sn02, Sn0)/-8 intermetallics (CuSb9, Cu6Sn510, •••), nitrides 11 and
phides, 12• 13 which could be used as alternatives to carbonaceous
rials, once their initial large irreversibility and poor cycle life have been
come.
Recently, our research group pursued a novel approach, deviating
the conventional route aimed at searching for compounds having voids
ture or alloying with Li. We obtained astonishing results with the
ibility for metal oxides (MO), having an interstice-free rock-salt
ture, to reversibly react with Li. 14 Not only did such findings make us
re that high-performing electrode materials may have been wrongly
garded in the past, but also they now comfort us in our approach aimed
ckling electrode materials at the microscopic rather than macroscopic
e. Through our work on metal oxides, which was extended to other
l-based anode materials, we will stress that, when working at the atomic
l, new kinds of effects are apparent, and represent completely new
rtunities in the unusual design of materials.
Thus, this chapter will be organized as follows. First, we will recall
ey observations motivating the reactivity of 3d-metal oxides towards Li,
describe the fundamentals lying beyond such an unusual reactivity.
ndly, we will rationalize these types of reactions, and evaluate their
ntial with respect to applications. Thirdly, we will stress the positive
bute of nanoparticles in enhancing kinetics, therefore enabling different
activity paths.
EXPERIMENTAL
The Co-based metal oxide powders used within this study were
mercial products (CoO), or were synthesized in-house (Co304, Mn304,
203). The other metal oxides (M= Ni, Fe, Mn, Cu) were used as
ved from commercial sources, unless otherwise specified. LiCo02 was
provided by the market.
Powder purity and crystallinity were examined by X-ray diffraction
D), operating in Bragg-Brentano geometry with a CuKa radiation. The
ders morphology and composition were investigated by scanning
ron microscopy (SEM) with a field emission gun (FEG), coupled to an
ument for energy-dispersive X-ray spectroscopy (EDS). High-
Chapter 7
lution transmission electron microscopy (HRTEM) was carried out using

croscope equipped with an EDS analyzer. Magnetic measurements were
ormed on a commercial SQUID magnetometer, at temperatures ranging
5 to 290 K. Electrochemically-driven structural changes were followed
means of an in situ electrochemical cell similar to that previously
ribed 15 , with the studied electrode material deposited behind a beryllium
dow, which acts as the positive current collector.
The materials were tested in either Swagelok™ or standard 2035-size
cells that were assembled in an argon-filled glove box at a dew point of
c. If not otherwise specified, the cell consists of a plastic positive
trode disk containing 64 wt% of MxOy materials, 8 wt% SP carbon black
28 wt.% PVDF-HFP copolymer binder; a 1 cm2 disk of Li foil as the
tive electrode member; and a Whatman GF/D borosilicate glass fibre
t separator, saturated with a 1M LiPF6 electrolyte solution in a 1:1 (by
dimethyl carbonate/ethylene carbonate.
The cells were tested using a VMP automatic cycling/data recording
em that can be operated in both galvanostatic and potentiostatic modes.
reported curve was reproduced three times. Hardly any deviation was
d among the curves.
RESULTS AND DISCUSSION
. Reactivity of 3d Metal Oxides Towards Li
The electrochemical reactivity of 3d metal oxides towards Li has long

studied. A few oxides, such as CuO, which converts to Cu through a
mposition reaction (CuO -7 Cu + 1/202), were spotted as attractive
rode materials, leading to the assembly of primary CuO/Li cells that are
commercialised nowadays.16- 18 However, with the common belief that
mposition reactions cannot be electrochemically reversible, such oxides
been disregarded as reversible Li electrode. The motivation to
estigate them, like most materials advances, is the result of several
rvations/contributions going back to 1996. In 1996, Fuji's researchers7
rted that vanadates-based electrodes displayed, when cycled down to
volt, reversible capacities greater than those expected, assuming a
ction of the constituting metal elements. Later, these results were
19 4
nfirmed by Guyomard's group • Occasionally, for some LiMV0 (M=
i,Co ..) phases, this group reported a surprising capacity enhancement
cycling, which turned out not to be fortuitous, as will be seen later. In
of such intriguing results, our group launched a complete study of the
dates, aimed at identifying the key parameters governing their unusual
ivity vs. Li. Numerous hydrated vanadates of formula RV04 .nH20,
Key Role of Nanoparticles in the Reactivity of 3d Metal Oxides Towards Li 223
r crystalline or amorphous, were surveyed with special attention given

he Fe- and Co-based ones (FeV04 -1.1H20, Co(V03h·n H20) that
layed exceedingly large (800 mAh/gr) reversible capacities.20- Among
23
umerous in situ techniques used to study the electrochemical behaviour

o(V03h·n H20 /Li cells, XANES turned out to be the most fruitful 24•
ed, it indicated that upon charge/discharge cycles, the oxidation state of
was surprisingly converting between 0 and 2, and that of vanadium
een 3 and 5. Knowing that the electrochemical behavior of V205
eeds through an insertion/deinsertion process, we hypothesized that the
pected electrochemical results could be associated to the reactivity of 3d
l oxides towards Li. Thus, our study of the vanadates was narrowed
n to that of the 3d metal oxides as MO (M =Co, Ni, Fe, Mn), which
precluded to react with lithium on the basis of Li insertion-deinsertion
i-alloying reactions.25-30
Several CoO/Li half-cells were electrochemically tested towards Li14
to our surprise, we found that they could sustain a nicely reversible
city when cycled down to zero volt. During discharge (Figure 7.1), two
rent electrochemical processes appear, a plateau at 0.8 V with an
litude of two Li, followed by a continuous uptake of 0.7 additional Li,
n to a voltage of 0.02 V. With the following charge, only about two Li
ormula unit could be removed, leading to a capacity of 700 mAh per
of CoO, a value about twice that of today's graphite negative
rodes. Since such oxide does not contain voids, or a metallic element
alloys with Li, its reactivity had to be different from the classical Li
tion/deinsertion or Li-alloying processes. In an attempt to pin down this
mechanism, Potentiodynamic Intermittent Titration Technique (PITT),
tu X-ray diffraction, and XANES measurements, together with semi in
TEM measurements, were carried out on CoO/Li cells.
PITT electrochemical measurements revealed two very clear and
nct processes. The first one (Figure 7.2) for which the chro
perometry response to the voltage step (E=0.92 V) displays a bell curve
2
e, corresponds exactly to an uptake of 2 Li (e.g., a full reduction of Co+
o).
Beyond .x=2, we observed a continuous decrease in the potential as a
tion of x, as observed in the galvanostatic mode, with a difference in the
ber of reacted Li that reached 1.8 instead of 0.7. This suggests a faradic
ess strongly influenced by the imposed current (e.g., a kinetic influence).
CoO -7 Co conversion process was further confirmed by XANES
urements24 at the Co K-edge, indicating that the spectrum of fully
ted and fully delithiated samples after 10 cycles was superimposing that
etallic cobalt foil and fresh electrode, respectively.
Chapter 7
3.5
C/2
--
3
_
2.5
-
j
V)
> 2
>
1.5
0
1
c
+- 1
0
> 0.5
0 3.5
0 0.5 1 1.5 2 2.5 3
x in "LiXCoO"
e 7.1. Voltage-composition traces for a CoO!Li cell cycled between 0.01 V and 3 a
rate of C/2 (e.g., 1 lithium per formula unit in 2 hours) shown over several s.
Capacity retention for the same cell shown in the inset.
3.5
3
0 Q
5'2.5
r:
-0.05
2 -0.1
0.5
-0.15
0 0.5 1 1.5 2 2.5 3 3.5
0 x in "Li CoO"
X
e 7.2. PITT measurements performed on a CoO!Li cell. The voltage steps are of
V with a current limit Cimin) corresponding to C/300 (e.g., 1 Li in 300 hours).
Through the electrochemical reduction process the samples were

wn, as deduced by in situ X-ray measurements,14 to become amorphous,
to remain so upon subsequent cycles. Transmission electron microscopy
then used as a complementary technique to further grasp the electrode
ution morphology upon cycling. We observed (Figure 7.3) that upon
discharge, the sintered CoO particles of dimensions ranging from 1000
000 A, decomposed into 10 to 20 A diameter Co nanograins, dispersed
a matrix.
e 7.3. TEM observations realized on CoO electrodes recovered from CoO!Li cells that were
ed at various discharged and charged states, as indicated by the full circles on the
ge/composition curves. When necessary, the SAED patterns are joined to the micrograph as
ans to indicate the electrode composition. The nanometric character of the composite
her with the growth/disappearance of a polymeric layer are nicely depicted.
At .x=2, the decomposition was completed, and from an indexation of

ollected SAED patterns, we could clearly identify the presence of Co0
ic structure) and Li 20 (antifluorine structure) matrix. Upon further
ation (beyond .x=2), the SAED patterns no longer evolved.
rastingly, we noted the growth of an organic-type layer. The layer
ppeared upon the following charge, and the Co deoxidised , leading to a
nanograin electrode, as deduced from an indexation of the collected
D patterns. Two polymorphic forms, referred to as the high temperature
Chapter 7
oO (PDF n° 48-1719) and the low temperature -CoO forms (PDF n°

300) exist. It was separately reported, as deduced by x-rays, that
allic Co nanoparticles were converting to the low temperature CoO form
31
oted a-CoO) upon oxidation. Therefore, we could not ascertain the
gnment from our HRTEM measurements, since the two CoO
morphic forms have lattice parameters differences falling within the
t of the technique itself.
2. Rationalization and Potential Applications
2.1. Thermodynamic and Kinetic Considerations of the MOM

Conversion
Whatever the a or forms, all the above results convey the message
the reduction of CoO through the first discharge plateau is reversible
rding to the following reaction.
(7.1)
The disappearance of the well-defined first discharge plateau at the

nse of a "sloping voltage plateau" upon subsequent cycling is fully
istent with the nano-sized character of the electro-produced particles,
h generate many new indefinite interfaces with various interfacial
gies. Interestingly, while never observed at room temperature, an
trochemical reaction involving the formation/decomposition of LizO was
dy observed to occur during the electrochemical oxidation/reduction of
32 33
3 or Fe304 in molten salt at 500°C. - As usual, when dealing with the
ibility of any chemical reaction, one must return to basic
modynamics. In short, one must simply calculate the associated changes
tandard Gibbs free energy. If this .:lG turns out to be negative, the
idered reaction is thermodynamically feasible. Then, from Gibbs free
alpy of reactions, one can predict the equilibrium potentials by using the
-known Nernst law .:l,G = -nEF, where F is Faraday's constant. Using
34 35
=
hermodynamic data - given in the literature for massive compounds,
an deduce for CoO a .:l,G -347 kJ/mol that should correspond to a
au located at Eeq = 1.798V vs Li+/Li • Similar calculations can be
0
36 37
nded to most of the 3d metal oxides, - , with the main result being that
tabulated Gibbs free energy values are negative. This result
mbiguously proves that such oxides are, under these conditions,
aneously reducible at Eeq values, as reported in Figure 7.4.
o Bard data
3.5
.---.. • Handbook data ·························· ······················i····················· 3
.
s
3 • Kubaschewski data
+.....
.
2.5 ------------------·--r--·-------------------r ---------------------!·----------------------r·----- 2.5
r.n
> 2 ·····················i····················+······················ -- --······ -- ----····................ 2
>'-" i aj •••f • :•oo
Q.)
1.5
.. .
.. . .... ..
0 1
> ··••••••••••••••m ••••• •• •••··••r•:••••••••••••••••r•••••••••••••••• :•••••••••••• ··· 0.5
:·'
.
- --- ---··•··-i·-·····················i······················"·······················f·····················
0.5 '
e 7.4. The equilibrium voltage values of MO < -- > M conversion reaction are
ted for various oxides. The values were obtained by first determining the reaction
nergy (dG) and then by applying the Nernst law.
Thus, by blindly considering a thermodynamic reversible process, one

ld expect the reduction of CoO to occur at 1.798 V. This contrasts with
eduction potential value of 0.8V (Figure 7.1), obtained during the first
harge of a CoO/Li cell. The deviation of about 1 Vis naturally of kinetic
re, and represents the important over potential needed to initiate and
ue the decomposition reaction. Assuming that the 1V over potential
ed to reduce CoO is of the same amplitude as for most of the other 3d
es considered, our difficulty encountered in reducing MnO (Figure 7.5),
ur inability to reduce Ti02 and at fortiori MgO, can be understood.
Since reaction (7.1) is thermodynamically allowed in the forward
ction, it should also occur backwards (that is to say, from right to left), if
applied voltage exceeds the Eeq voltage. Thus, a legitimate question is
was this reaction never observed to proceed at room temperature, when
ures of LizO and Co were mechanically made and reoxydized . The
er is of a kinetic nature, and lies in the difficult task to reconvert,
0
ugh a solid-state reaction, coarse M + Li2 0 particles into a metal oxide
y)- We believe that the reversibility experienced with various MxO/Li
is mainly nested in the high chemical reactivity of the electrochemically
e nanograins, after the electrode has been reduced once. Indeed, with a
easing particle size, an increasing fraction of atoms lies near or on the
ce, making the effect of surface on the chemical/electrochemical
ivity of increasing importance. Thus, the chemical diffusion path in
Chapter 7
ly divided powders becomes close to that of a liquid state, thereby

aining why a reaction occurring at about 500 oc can be produced at
temperature. Nevertheless, electrochemically speaking, reaction (7.1)
ins slow as seen by the substantial difference observed between the
harge and the charge voltage composition curves in cycling (Figure 7.5).
ever, caution has to be exercised, since the capacity associated with the
reduction of the 3d metal through a decomposition reaction does not
account for the measured capacity.
3.5 0.1
Eeq(Mn0/Mn°)=1,03V
+' 3 Eplateau"'0.2V/Li 0.05
2.5
CZl
> 2
0
-0.05
§g
>
1.5
.......
-0.1
..... 1 -0.15 '--'
0.5 -0.2
0 1 1.5 2 2.5 3 3.5 -0.25

0 0.5
x in 'Li01n0'
e 7.5. The voltage composition curves as deduced from PITT measurements for
Li cells cycled between 0.02 V and 3V. The voltage steps were of 10 mY with a t
limit corresponding to C/300.
.2. The Rich Chemistry Associated to the Low Voltage Process
The above observations, while leading to a full understanding of the

process, did not give any clue as to its origin. To further look into this
voltage process, we measured the dependence of I vs. the voltage scan
d (V/s) over the 0 to 1.8 V range (Figure 7.6), and observed a linearity
is reminiscent of a super capacitor-type behaviour. Interestingly, we
ed, by TEM, that the gel-like film formed on CoO was not affected
cycling was limited to the 0-1.8 V range, while vanishing when the
electrode voltage was raised to 3 V. Such a finding suggested that the
n polymer film had an electrochemical activity. However, we do not
any information about its composition. In contrast, its formation is
d in the Co-catalysed electrolyte decomposition, with some of the
lting products moving to the negative electrode, as deduced by the one

ne correlation between the phenomena occurring at the CoO and Li
38
rode, as observed by AC impedance measurements. Indeed, we noted
the reversible transition from the CoO Co conversion process to the
rolyte decomposition process was mirrored by a reversible change in the
ectrode resistivity.
0 0.5 1 1.5 2
Scan potential (Volts)
e 7.6. Cyclic-voltammetry data as a function of the scan rate for a CoO!Li cell that
ycled between 0.02 and 1.8 V. One inset shows the linear variation of the current as
ction of the scan rate. The others, a, b, and c are micrographs of CoO electrodes
ered from CoO!Li cells that were fully discharged and fully recharged 10 times
en 0.02 and 1.8 V, and fully recharged to 3 V, respectively.
Aware that the low temperature process was rooted in electrolyte

mposition reactions, we decided to study how it would depend on the
e of the electrolyte used. We went through a survey of various
rolytes encompassing the use of 1 M LiPF6 in either a single solvent,
as EC or DMC, or solvent mixtures, such as EC-DMC and EC-PC. The
of DMC (linear alkyl carbonate) as the sole solvent resulted in fast
city fading, as opposed to EC (cyclic alkyl carbonate), which enables
ined cyclability. The detrimental effect of DMC was considerably
nished when used in a 50/50 weight mixture, with either PC or EC
Chapter 7
ic molecular structures. As cells were cycled, at least during the first

es, we noted on oxidation the partial disappearance of the organic layer
was forming on reduction. Worth mentioning, however, was that the
unt of formed polymeric film was always found to be less in presence of
At this point, a legitimate question was whether such observed effects

d be the simple result of the presence of carbon within our starting CoO
ectrode composite. The observed growth of the polymeric layer in cells
e with thinner carbon-free CoO electrodes ruled out this possibility. The
difference between cells having carbon free (CoO/Li) or carbon
aining CoO positive electrode (CoOJLi) cells resides in the first few
es. The former display a lowest initial capacity that decays more rapidly
the first few cycles than the latter.
With the polymeric layer being the result of an electrolyte
mposition, one could hypothesize its growth to be strongly temperature
ndent. To check this hypothesis, we studied the cycling behaviour of
CoOJLi and CoO/Li cells at various temperatures, namely, 55°C and
(Figure 7.7).
2000
1800
1600
1400
-'
;
1200
d 1000
. 800
600
;
) 400
..
) 200
0
0 50 100 150 200
Cycle number
e 7.7. Variation of the capacity retention of CoO/Li cells, both as a function of
mperature and of the electrode content (e.g., with or without carbon). Effect of
rature sweep between 20°C and 75°C on the capacity values shown as an inset.
First, common to all the cells once the first ten cycles were achieved,
ote a continuous increase in the cell capacity that was more pronounced
increasing temperature. At 75°C, a doubling of the initial capacity was
hed after 200 cycles. Such a capacity was found to be reversibly
perature dependent, since a decrease in temperature resulted in a capacity
ease, which is fully recovered as the temperature return to its original
e39 • Worth mentioning is that more than 500 cycles were achieved
out any evidence of short-circuit, reminiscent of dendrite formations.
hermore, on cells opened after 500 cycles, we noted that the polymeric
r has expanded through all the separator, suggesting at first that the
meric layer could not be a good electronic conductor, without which the
would have a short. The feasibility of cycling CoO/Li cells at 75°C at a
ate for more than 400 cycles is somewhat amazing, bearing in mind
these are the best conditions to favour dendrite formation. This can be
ained either by: 1) the fact that the polymer film formed on the CoO
rode prevents dendrite penetration, and by the same token, cell short or
y a modification of the Li surface morphology, due to the deposit of
rolyte degradation products, to the point that it cancelled dendrite
ation. Needless to say these facts could help in engineering better
ectrolyte interfaces so that rechargeable Li batteries could finally
me a commercial reality.
Finally, by conducting cycling measurements with different discharge
ff voltages (Figure 7.7, inset), we reached the conclusion that this
erature-enhanced capacity was the result of electrochemical processes
g place at voltages close to lithium (e.g, over the 0.02-0.15 V).
iously, this result could simply be explained by assuming a temperature
nced electrolyte degradation that would lead to an electrochemically
e polymer. Therefore the amount of this polymer would increase, the
city would increase, and should remain constant whatever the cycling
erature once the film is formed. This is not as simple, since we
rved that the capacity decreases once the temperature is lowered. The
that the temperature-driven extra capacity is located near 0 V vs. Li+/Li 0,
ates that one needs to decrease the cell polarization by rising the cell
erature to tap into this low potential extra capacity. Thus, the capacity
g observed with changing the temperature is the result of polarization
ts, most likely associated to a limited ionic or electronic conductivity of
rown polymeric layer.
To further explain the cycling data at 70°C, we plotted the derivative
es (Figure 7.8) as a function of the cycle number. We noted the
arance of a 1.2V peak corresponding to the Co07Co conversion
ess, which disappeared upon cycling at the expense of a 0.8 V peak.
implies that, upon long cycling, the Co07Co conversion process
lly corresponding to the main part of the overall cell capacity vanished.
Chapter 7
0 0.5 1 1.5 2 2.5 3

U(V/LVLn
0.15V-3V
80 120 160 200 240 280

Cycle number
2.5 3
0 0.5 1 1.5 2
U(Vvs LVLi)
7.8. Derivative dx/dV plots as a function of the cycle number for CoO/Li cells.
ells were cycled at 75°C at a C rate. Two scanning ranges were used, 0.02 to 3 V
.15 to 3 V. Note the appearance of a 0.8 V to the expense of the 1.2 V peak
ponding to the MO < -- > M conversion process, and of a large extra capacity
d around 0.1 V.
The hypothesis was directly confirmed by magnetic measurements

re 7.9) that showed no changes in the sample magnetization (equivalent
0% of Co in the magnetic state), whether the composite electrode was in

harged or discharged state. In light of such results, it appeared that the
sured capacity, once the CoO conversion process became inactive, must
ue to the polymer-like forming film.
Among the remaining questions are whether, in addition to the key
played by the polymeric layer, these processes could also be rooted in:
he presence of Co surfaces electrochemically attractive to Li deposits at
ntial positive with respect to the Nernst potential (UPD-type processes)
2) in the feasibility of Co, in such highly divided electrodes, to
trochemically alloy with Li. Finding a clue to these questions depends
nding the exact nature and composition of the formed polymeric layer, a
presently being pursued by our group.
140 30 discharges
120
0
U100
4-1
0 80 30 charges
60
::E 40
20
0 ---- ----
4
----
4
---- 4----4 -- 4
0 110 2 10 3 10 4 4 10 5 10 6 10
FIELD (Gauss)
e 7.9. Variation of the room temperature remnant magnetization for fully charged
ischarged CoO electrodes recovered from CoO/Li cells that were cycled 30 and 100
, respectively, at a C rate over a 0.02 to 3 V voltage range.
Under any circumstances, and whatever the exact process, either a

bination of UPD and Li-alloying processes or others, the main result is
reactivity of the 3d metal oxides towards Li stresses the profound effect
divided particles could have within the field of energy storage. Worth
ioning is that the described effect is not specific to Co electrodes, but
occurs with Cu, Fe, Ni and Mn (which, purely coincidentally, are
uently used as catalysts in Fischer-Tropsch type reactions, for instance).
hermore, numerous classes of materials besides the 3d metal monoxides
Chapter 7
14
O, FeO, MnO, CuO), such as 3d metal sesquioxides (Co304), Li-based
27
ary oxides (LiCo02, LiNi02, ...), 3d metal sulphides (CoS) , or nitrides
4041
3N), were found to behave identically, leading to the
ation/decomposition upon cycling of Li2S and LhN, respectively,
ther with the formation of a polymeric gel-like layer. Similarly, there
recent reports describing the potential reactivity of 3d metal-based
tes or phosphates in similar terms, thus suggesting the universality of
a proposed mechanism.
3. Practical Application Considerations

A fundamental understanding was of a great help in optimising these
erials, so that 100% capacity retention was achieved with plasticized
electrodes containing 1-micron particles. Since the measured capacity,
ast within the first hundred cycles, mainly comes from the CoO Co
version process, where two electrons are liberated, rules to increase the
all capacity were established. Hence the need to search for oxides in
ch the metal oxidation state was greater than 2 as well as our interest for
04, although numerous other spinels or layer oxides (LiCo02) could be
as well. We selected Co304 as our negative electrode, and went into the
cise of assembling Li-ion cells with Co304 as carbon alternative. The
o02/Co304 Li-ion cells cycled exceedingly (Figure 7.10) well at both RT
55°C, since no capacity fade was observed.42 Furthermore, these cells,
ng a weight ratio of positive to negative of 8.5, delivered a 2.2 V output
age, resulting in an energy density of about 130 Wh/kg. This value is not
petitive with today's existing value of 180 Wh/kg for LiCo02/C cells,
use the LiCo02/Co304 cells are penalized by both the voltage of the
tive electrode, which is too high as compared to carbon, and the 20%
ersible capacity linked to the use of Co304. These are not show
pers, since, through the large oxide family, there are compounds (MnO)
0
ng a reacting voltage of 0.5V vs. Lt/Li , and others (Li 1+xMn204) that
be used as positive electrodes, while having aLi reservoir to compensate
the initial 20% capacity loss. In this respect, the Li 1+xMn2 0JMn0
em is quite attractive, although still limited in power rate.
It should be realized that the appeal of the metal oxide negative
trodes is that their nano-structure is internally created during the first
trochemical reduction, producing pristine and compact nanoparticles,
h are held together through LizO boundaries domains.43 While it may
aken for granted, let's recall that a similar type of internal grinding
rs on the Ni(OH)z electrode, widely used in today's Ni-based alkaline
ries. Indeed, during the first electrochemical oxidation, there is a drastic
re change in the Ni(OH)z particles that evolve from monolithic to
aic-type. Controlling the size of these electrochemically-made mosaic
(b)
4 ,. I
Output Cell '
Voltage , 1
---"--,, I
0
CZl
........
3 ----
,.,I
>
"-"'
2
(!.)
OJ)
C\$ 1
........
0
> 0
0 2 4 6 8 10
x in LixCo 304
0.2
\ Co3b4 /LiCo02
1 I I I I
. 0.15
:_l
<. 0.1
c::
"-"'
;>..
......
u
C\$
0.05 Rate: C/5
u Mr=8.5 T = 25°C
C\$ I I I I I I
0
0 10 20 30 40 50 60 70 80
Cycle number
re 7.10. The first cycle of a LiCo02/Co304 cell cycled between 4.1 and l.IV, as
ured by means of three-electrode configuration, is shown in (a). The cell output
ge is of about 2 V and the weight ratio (M,) of positive to negative equals 8.5. The
city retention for such Li-ion cell is shown in (b).
ains, and in the same way the domain wall boundaries that act as a
er layer, is crucial to the performance of this electrode, since, upon
sequent cycling, the so called "Ni(OHh formation cycle" remains a
able know-how specific to each battery manufacturers. This situation
Chapter 7
rs from externally-made nanostructured powder, or electrodes that can

made either by chemical/physical means, with the greatest difficulty
g to gain maximum control over the size and size dispersion of the
ting particles or cluster of atoms. Therefore there are recent reports of
nced cycling performances obtained with nanostructured LiMn 204 or
P04 electrodes made by ball milling. A remaining difficult task with
nally made nanoparticles lies in the formation of densified electrodes, h
are crucial for practical applications. Therefore, new densification ods,
for instance Magnetic Pulse Compaction (MPC), may be of great
in densifying such highly divided electrode materials.
This novel reversible Li-reactivity mechanism, mediated through the
ation of metal particles, underscores the possibility when working at
atomic level to radically change reaction kinetics and
ical/electrochemical reaction paths. Indeed, one must keep in mind that
rticle size decreases, the amount of surface atoms adheres to that of the
, and becomes greater for particle size smaller than 20 A. Thus, the
considerations of the material electronic band structure or
allographic voids are no longer valid. Our motivation to revisit well
n materials, that were previously disregarded, is investigated next.
4. Acting on Reaction Paths Through Particle Size
When the starting oxides correspond to the lowest initial oxidation

of the metal ion (CoO, NiO, Cu20), we showed that metallic clusters
ormed as soon as the reduction process starts. In contrast, when starting
higher cationic oxidation states (Co304, Fe203), one can expect
mediate reduced phases to be found in the early stage of the reduction
ess. Through all the above studies, we focused on the oxide
mposition reaction occurring at low voltage, using mainly metal oxides
ecursor, with the exception of Co304, which turned out to be one of the
promising negative electrode candidates as a carbon alternative.
ed, this phase adopts the spinel-type structure, having empty tetrahedral
ctahedral sites to host Li. It was then of special interest to really
rstand the interplay between decomposition and insertion reaction,
Co304 was fully reduced to 0 volt. There are early reports4447 regarding
easibility of electrochemically/chemically inserting Li ions in either
4 or Mn304, but the results were poorly reproducible, and the value of
ndetermined, due to the coexistence of secondary phases resulting
the lithiation process. A legitimate question was whether such issues
be related to particle size/morphology. We then initiated a study
d at sensing how the particle size could influence insertion vs.
mposition reactions in oxides having open structures. Our results will be
8 49
ented through two pedagogical examples: Co30/ and a-Fez03.
tever the studied oxide, the first step towards such an approach consists
laborating synthesis methods to enable the formation of tailor-made
cles. Hence, each following section will begin with a material synthesis
ocol.
Co304 tailored powders were prepared in two steps. First, crystallized

alkoxides were made from a derived polyol process that involved
2
ing a solution of Co+ acetate tetrahydrate in 1,2-propanediol.
ndly, the precipitated powders were recovered, washed with ethanol,
then fired at various temperatures ranging from 250°C to 800°C. These
ples will be referred to as Co304(2SO) and Co304(SOO) samples, respectively.
C is the temperature needed (as determined by TGA measurement) to
re a total combustion of the structural organic molecules. The x-ray
der patterns of the annealed samples indicated the appearance of broad
s, characteristic of the spinel phase in the 250°C sample that increased
mplitude and sharpened with increasing temperature. From the Scherrer
tion, we noted that the crystallite size evolved with temperature, and
ased from about 150 A to at least 800 A. This particle coarsening
ted in a 60 m2/g to about 3m2/g BET specific surface area decrease
n increasing the sample annealing temperature. The most astonishing
t concerns the maintenance (Figure 7.11) of the precursor spherical
cles through the annealing process. Note also the very narrow size
ibution of the particles constituting the agglomerates. Such tailor-made
4 particles were the ingredient of our in situ X-ray electrochemical
es.
In situ X-ray Co304/Li cells based on Co304(soo) powders were made
discharged at various current rates. We note (Figure 7.12) the
arance of a LixCo304 phase for discharging current rates ranging from
to C/25, and its disappearance, at the expense of a CoO phase, as the
arge current was lowered to C/75. Similarly, in situ X-ray cells having
4 samples, annealed at various temperatures, were discharged at the
(C/12) current rate. The XRD patterns collected during the discharge of
304(2solLi and Co304(soo/Li cells exhibited, respectively, the sole
nce of CoO Bragg peaks and LixCo304 Bragg peaks. A common
minator to these observed effects is the current density sensed by the
cle, with the overall result being that high density favours Li insertion,
e expense of decomposition reactions. Hence Thackeray's difficulties
untered in achieving a pure
Chapter 7
100
90
50 750
150 250 350 450 550 650
Temperature COC)
e 7.11. Weight loss, as determined by thermo gravimetric analysis, for a home

esized Co glycolate precursor that was heated at a 10°C/min. rate. SEM
graphs shown for the initial precursors as well as for materials annealed at 250°C
00°C, respectively.
um-phase, whatever the chemical used or electrochemical approaches.

er reducing conditions and using a strong reducing agent such as n
yllithium, which to a certain extent mimics a cell short circuit (e.g., high
ent density), his chemically prepared samples always contained large
unts of LixCo304.
These findings indicate that the nature of the intermediates, when
04 is electrochemically reduced vs. Lithium, is rate/surface dependent.
intercalated LixCo304 formation suggests that kinetic limitations lower
ability voltage far below the over voltage of the Co304 Co0 reaction.
a poor kinetic is nested in structural aspects that are peculiar to spinel
tures. Indeed, the insertion of Li into Co304 necessitates the migration
o ions from the Sa to the 16c sites (spinel rock-salt). This Li-driven
2
ation of Co+ from Sa to 16 c sites is energetically costly, and competes
the Co304 decomposition reaction associated to the departure of oxygen
ng to CoO, which will then decompose into Co. A similar path was
n to occur for Mn304 that either decomposes into MnO or intercalates
LixMn304.
Thus, an open question is whether such findings regarding the
rtance of particle size on Li-driven reactivity paths could be generalized
her oxides differing from the spinel structure. As a tentative answer, we
Co304 : 250°C
2000
Co304 : 800°C
BeO A
BeO + BeO coJ04
Co3 Co304
1750 BeO
1500
1250
1000
ea_n
(.)
750
500
250
x=2.0 u:co304 LixCo30

4 36 36 40 42 44 34 36 36 40 42 44
Scattering Angle (2 8) Scattering Angle (2B)
re 7.12. In situ X-ray data for CoO/Li cells discharged at very low current drains
2)/ close to equilibrium at a C rate. The cells contained either Co0250 or Co0800
rodes .
ided to investigate Fe-based oxides, and more especially a-Fe203

matite), which has the corundum structure, and is known to switch from a
agonal compact to a cubic, close-packed oxygen array at a Li
centration lower than 0.1 per unit formula.
4.2. a-Fez03
Two batches of a-Fe203 (hematite), referred to as M-Fe203 (for

rometric) and n-F 03 (for nanometric), were used in this study: M-Fe203
commercial sample (Aldrich or Prolabo), and n-Fe203 is a homemade
der prepared by forced hydrolysis in acidic medium. More specifically, ic
nitrate hydrate Fe(N03)3.9H20 was progressively dissolved in a HN03
eous solution that was heated in a glass vessel under magnetic stirring,
refluxed at boiling point for 24 hours. Then, the recovered reddish
wn powder was thoroughly washed and dried prior to being investigated.
Aside from some slight deviations in the relative peaks intensities, the
D patterns for M-Fe203 and n-F 03 (Figure 7.13) perfectly match that of
- Ill 300
-
c.
0
200
100
Chapter 7
>
:!:::
0
-
Ill
I::
Q)
1200
I:: 800
400
30 40 50 60 70 80
Scattering Angle (deg.) CoKo:
e 7.13. X-ray powder patterns for commercial and home-synthesized a-Fez0 3

her with, as inset, the SEM image for both samples.
e203• While structurally similar, these two samples strongly differ in the
h of their Bragg reflections. From the Scherrer's equation, we estimated
mean crystallite size at about 200 A for n-Fe203, and at least 1000 A for
e203• By TEM, the lozenge-shaped n-Fe203 particles were found to be
olithic and very small (200 A), in total agreement with the crystallite
deduced from XRD analysis. M-Fe203 particles are much larger (0.5
in mean diameter), as shown by SEM images. The measured BET
2 2
ific surface areas are 60 m /g and 2 m /g for n-Fe203 and M-Fe203•
The first electrochemical discharge curves collected at C/5 rate for the
hematite samples previously described are presented in Figure 7.14.
n discharged down to 0 Volt, 7.2 Li per M-Fe2 03 and 8.3Li per n-Fe203
be reacted, corresponding to capacities of 1200 and 1400 mAhlg. Both
harge curves show a nicely developed and long voltage plateau lying
w 1.0 Volt, until an x value of about 6 (x in "LixFe203"), in agreement
3 0
the theoretical capacity for the Fe + 7 Fe reaction, is reached. The
r excess in capacity measured for n-Fe203 is consistent with its much
r specific surface area, when compared to that of M-Fe203•
400Capacity (mAh/g)
-·
200 600 800 1000 ' 1200 1400
30
s. 2.5
r:l.l
2
c 1.5
s
1
Oll
0 0.5
0
0 1 2 7 8
e 7.14. First discharge for a-FezO/ Li cells, using bulk or nano a-Fez0 3 as positive
ode, together with, as insets, the open circuit voltage measurements for similar cells.
However, the most relevant part of the curve, with respect to the
ortance of nano particles, lies at low x values. Indeed, note that the onset
he 1 V plateau is very quickly reached for M-Fe203 (x = 0.1), while it
rs at x=2 for n-Fe203, suggesting a different Li reactivity mechanism
een the two samples. To better isolate such differences, PITT
riments were conducted on n-Fe203 and M-Fe203 up to x=2. Within this
position domain, the reduction of n-Fe203 clearly proceeds in two steps.
, a monophasic process characterized by a continuous decrease in the
i-equilibrium OCV is observed up to x close to 1, and then a biphasic
tion characterized by a constant equilibrium voltage close to 1.85 Volts.
The Li reactivity difference was well diagnosed by in situ X-ray
surements (Figure 7.15), with a continuous shift of the hematite Bragg
s towards low angles for the n-F 03 sample, as compared to a
ressive vanishing of the hematite Bragg peaks for M-Fe203 reflections,
he expense of a new set of reflections. These extra peaks, that were
xed within a cubic symmetry having an axis of 8.422 A, were found to
espond to a Li-Fe-0 rock-salt compound described by Thackeray et al.50
not to a FeO rock-salt phase (as confirmed by EXAFS and Mossbauer),
should have an a axis of 4.31 A (e.g., 2a = 8.62 A).
Chapter 7
12000
00oo ::o--- "-'
10000:;
/1.\"---v
o:- - :::.
In
8000
-n
0ooob
6000 .!!
37 38 39 40 41 42 43 37 38 39 40 41 42
Scattering Angle (deg.) CoKu Scattering Angle (deg.) CoKu
e 7.15. in situ X-ray data for a-F 03/ Li cells, using either bulk or nano a-F 03
ositive electrode. For reasons of clarity, only part of the pattern (33°
< 43°) depicting the evolution of the 104 and 110 Bragg peaks as a function of the
nt of reacted Li is shown. Note a disappearance of these peaks for the bulk sample,
mpared to a shift for the nano-materiaL
Thus, from these measurements, it appears that up to one Li per

ula unit can be inserted into the a-Fe203 corundum structure of n-Fe203,
lting in a 1% volume expanded a-LiF 0 3 end-member phase, having
attice parameters a=5.041(1) A and c=13.815(2) A, while no lithium can
gnificantly inserted into M-Fe 2 03. The feasibility of inducing reversible
rtion reactions in n-Fe203, and not in M-Fe203, is mainly nested in the
that small particles can more easily and uniformly expand to
mmodate strains resulting from insertion reactions than larger ones.
Finally, when either Li/n-Fe203 or Li/M-Fe203 cells were cycled at
rate between 1 and 3.5 V, we observed (Figure 7.16), as expected, that
ones containing nanometric Fe203 display a nicely reversible and
ined capacity upon cycling, while the other cells did not exhibit any
of reversibility.
. _-·r-0('1
- - "o .
0 0.
1om
8 --r--------r
45
c::::::
....- 4
, 3.5 -
-
e" il o.
0.
eo L
40.:;,
o..o. --
.a d
OJ)
r JJ -
>
0.1 201;!_
3_ u oL- -- - ---'---' €
> 2.5 -
'""-'
8 12
Cycle number
18 28
2-
1.5-
1
0
>o.5 0 0.5 1
x in LixFe203
,...-._ 5
•
+.-<
5.....:l
4.5 n-Fe203
4
rJJ
> 3.5
>
'""-'
3
OJ) 2.5
ro
,.-.+..-.>.; 2
0
> 1.5 0 0.5 1
x in LixFe203
re 7-16. Voltage-composition profile for o:-F O/ Li cells, having bulk (upper) or nano
er) o:-F 03 as the positive electrode. Their capacity retention upon cycling is shown by
nsets.
Worth mentioning is the large irreversibility displayed by the Li/n

03 cell, whose origin still remains precisely undetermined, while there is
ng suspicion for the presence of chemical groups, specific to the n-Fe2 03
hesis process, that remain adsorbed at the surface of the highly divided
ders. Further studies must be carried out to sort out this issue.
ertheless, throughout this work we stressed the importance of particle
in controlling Li reactivity mechanisms. We also resolved to unravel
ntial electrode materials that were thus far disregarded. As time elapses,
erous examples, which further emphasise the wonders of nano
pounds in igniting unexpected reactions, appear. They encompass
Chapter 7
es, nitrides, fluorides and sulphides, and another family of materials like
rmetallics. For instance, our group recently reported that the reactivity of
b3 towards Li is reversible. Although Co-Sb bonds are broken during the
harging process, leading to a full disintegration of the CoSb3 structural
ework into Co and LhSb nano particles, we show that upon recharge
e destroyed bonds, due to the high reactivity of the divided metal nano
icles, reformed and arranged to convert back to the initial CoSb3
51 2
cture Such an example, also observed with Cu6Snl , further contrasts
previous beliefs that bond-breakings were prohibited for reversible Li
tivity.
CONCLUSIONS
We have shown, through a few descriptive examples, how the field of

materials can unravel novel reactivity mechanisms so far not expected
ccur at the macroscopic scale. These mechanisms are of tremendous
ortance, since, for instance, the reported reversible reactivity of 3d metal
es towards Li was shown to be universal, and to apply to a wide family
ompounds. Furthermore, although we limited ourselves to electrode
rials, size confinement can also positively affect other battery
ponents, such as electrolyte, current collectors and others. On that issue,
ecalled a few recent reports that dealt with the positive attributes of
ng TiOz or Alz03 nano-particles to polymer electrolyte. This would
nce their conductivity several times. Consider also the make-up of nano
tured electrodes to bypass electronic limitation. The awareness of the
ortance of nano materials is rapidly growing within the battery
munity. A new path of opportunities has been opened. In the quest for
electrode materials, we no longer need materials with good electronic
uctivity or open structures. Thus, a large variety of previously
garded compounds are worth revisiting. As a new field is opened, its
rstanding suffers from the lack of theoretical support to build on.
ding this theoretical platform on nano driven Li electrochemistry stands
great challenge that requires interactive studies between chemists,
rochemists and theorists. An exciting adventure awaits us....
KNOWLEDGMENTS
The authors would like to give special thanks to L. Dupont, M. Dolle,
ayen, F. Badway, and I. Piitz for their contribution to some of the
ented data.
ey Role of Nanoparticles in the Reactivity of 3d Metal Oxides Towards Li 245
FERENCES
M.B. Armand, New Electrode Materials, in Fast Ion Transport in Solids (Van Goo!, W.,
Ed., North Holland, Amsterdam) (1973) 665.
M.S. Whittingham, Science 192 (1976) 1226.
D. W. Murphy, P.A. Christian, Science 205 (1979) 651.
A. R. Armstrong, P.G. Bruce, Nature 381 (1996) 499.
B. Ammundsen, J. Desilvestro, T. Groutso, D. Hassel, J.B. Metson, E. Regan, R. Steiner,
P.J. Pichering, ECS Fall Meeting, Hawai, Abstract N"138 (1999).
A.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, J.B. Goodenough, J. Electrochem.
oc. 144 (1997) 1609.
Y. Idota, T. Kubota, A. Matsufuji, Y. Maekawa, T. Miyasaka, Science 276 (1997) 1395.
.A. Courtney, J.R. Dahn, J. Electrochem. Soc. 144 (1997) 2045.
0. Mao, R.A. Dunlap, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 405.
K.D. Kepler, J.T. Vaughey, M.M. Thackeray, Electrochem. Solid-State Lett. 7 (1999)
07.
N. Pereira, L.C. Klein, G.G.Amatucci, J. Electrochem. Soc. 149 (2002) A262.
R. Alcantara, F. J. Fernandez-Madrigal, P. Lavela, J. L. Tirado, J-C. Jumas, J. Olivier-
ourcade, J. Mater. Chern. 9 (1999) 2517.
V. Pralong, D.C.S. Souza, K.T. Leung, L.F. Nazar, Electrochem. Comm. 4 (2002) 516.
. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J.-M. Tarascon., Nature 407 (2000) 496.
G.G. Amatucci, J.M. Tarascon, L.C. Klein, J. Electrochem. Soc. 143 (1996) 1114.
. Bates, Y. Jumel, Lithium batteries J. P. Gabano Ed., Academic Press, London (1983).
Y. Matsuda, K. Teraji, Y. Takasu, Denki Kagaku 44 (1976) 363.
. Novak, Electrochim. Acta 30 (1985) 1687.
. Sigala, D. Guyomard, Y. Piffard, M. Tournoux, C.R. Acad. Sci. Paris Il320 (1995)
23.
. Poizot, E. Baudrin, S. Laruelle, L. Dupont, M. Touboul, J-M. Tarascon, Solid state
onics 138 (2000) 31.
. Baudrin, S. Denis, F. Orsini, L. Seguin, M. Touboul, J-M. Tarascon, J. Mater. Chern. 9
1999) 101.
. Denis, E. Baudrin, F. Orsini, G. Ouvrard, M. Touboul, J-M. Tarascon, J. Power
ources 81-82 (1999) 79.
. Baudrin, S. Denis, S. Laruelle, M. Touboul, J-M. Tarascon, Solid State Ionics 123
1999) 139.
. Laruelle, P. Poizot, E. Baudrin, V. Briois, M. Touboul, J-M.Tarascon, J. Power
ources 97-98 (2001) 251.
. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J.-M.Tarascon., Nature 407 (2000) 496.
. Poizot, S. Laruelle, S. Grugeon, L. Dupont, B. Beaudoin, J-M. Tarascon, C.R. Acad.
ci. Paris II, 3 (2000) 681.
. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J-M. Tarascon, Ionics 6 (2000) 321.
. Grugeon, S. Laruelle, R. Herrera-Urbina, L. Dupont, P. Poizot, J-M. Tarascon, J.
lectrochem. Soc. 148 (2001) A285.
. Dehart, L. Dupont, P. Poizot, J-M. Tarascon, J. Electrochem. Soc. 148 (2001) A1266.
. Poizot, S. Laruelle, E. Baudrin, S. Denis, M. Touboul, J.-M. Tarascon, J. Power
ources 97-98 (2001) 235.
N.N. Obrovac, R.A. Dunlap, R.J. Sanderson, J.R. Dahn, J. Electrochem. Soc. 148 (2001)
576.
M.M. Thackeray, W.I.F. David, J.B. Goodenough, Mat. Res. Bull. 17 (1982) 785.
.M. Thackeray, W.I.F. David, J.B. Goodenough, J. Solid-State Chern. 55 (1984) 280.
Chapter 7
Standard Potentials in Aqueous Solution, A.J. Bard, R. Parsons, J. Jordan, Eds., Marcel
Dekker (1985).
0. Kubaschewski, C. B. Alcok, Metallurgical Thermochemistry, 5th Edition, Pergamon
Press (1987)
S. Laruelle, S. Grugeon, P. Poizot, M. Dolle, L. Dupont, J-M. Tarascon, J. Electrochem.
Soc. 149 (2002) 627.
P. Poizot, S. Laruelle, S. Grugeon, J-M. Tarascon, J. Electrochem. Soc. 149 (2002) 627.
M. Dolle, P. Poizot, L. Dupont, J-M. Tarascon, Electrochem. Solid-State Lett. 5 (2002)
A18.
S. Grugeon, S. Laruelle, J.M. Tarascon, J. Electrochem. Soc. submitted (2002).
N. Pereira, L.C. Klein, G.G. Amatucci, ECS and ISE Joint International Meeting, San
Francisco, CA, Sept. 2001, paper 203.
N.Pereira, L. Dupont, J.-M. Tarascon, L. Klein, G.G Amatucci, J. Electrochem. Soc. (in
press).
F. Badway, I. Piitz, S. Grugeon, S. Laruelle, M. Dolle, A.S. Gozdz, J-M. Tarascon,
Electrochem. Solid-State Lett. 5 (2002) A115.
A. Delahaye-Vidal, B. Beaudoin, M. Figlarz, Reactivity Solids 2 (1986) 223.
W.I.F. David, J.B. Goodenough, M.M. Thackeray, M.G.S.R. Thomas, Rev. Chim. Miner.
20 (1983) 636.
M.M. Thackeray, W.I.F. David, J.B. Goodenough, Mat. Res. Bull. 17 (1982) 785.
M.M. Thackeray, W.I.F. David, P.O. Bruce, J.B. Goodenough, Mat. Res. Bull. 18
(1983) 461.
M.M. Thackeray, S.D. Backer, K.T. Adendorff, Solid State Ionics 17 (1985) 175.
D. Larcher, G. Sudant, J-B. Leriche, Y. Chabre, J-M. Tarascon, J. Electrochem. Soc. 149
2002) A234.
D. Larcher, C. Masquelier, D. Bonnin, Y. Chabre, V. Masson, J-B. Leriche, J-M.
Tarascon, J. Electrochem. Soc. submitted (2002).
M.M. Thackeray, J. Coetzer, Mat. Res. Bull. 16 (1981) 591.
.-M.Tarascon, M. Morcrette, L. Dupont, Y. Chabre, C. Payen, D. Larcher, V. Pralong, J.
Electrochem. Soc. in press (2003).
D. Larcher, L.Y. Beaulieu, 0. Mao, A.E. George, J.R. Dahn, J. Electrochem. Soc. 147
2000) 1703.
apter 8
NITRIDE AND SILICIDE NEGATIVE

ELECTRODES
G. G. Amatucci and N. Pereira
Telcordia Technologies, 331 Newman Springs Road, Red Bank, NJ 07701, U.S.A.
INTRODUCTION
The search for negative electrode materials which exhibit improved
rochemical properties relative to the widely used graphite intercalation
rials has lead to a renaissance of research on inorganic negative
rodes.
The intercalation of Li+ into graphite follows the basic topotactic
tion process:
(8.1)
Beyond basic binary lithium alloys, a large focus has been on the
lopment of conversion oxides. The conversion oxides MexOy, as
ussed in a previous chapter, consist of metals (Me) which can alloy with
um. Upon the first lithiation, an electrochemically induced conversion
ess occurs creating a nanostructure consisting of nanodomains of Me in
trix of nonstructured Li2 0 (8.2):
2yLi+ + 2ye- + MexOy -+ yLhO + xMe (8.2)

zLt + ze- + yLhO + xMe !::+ yLhO + xLizMe (8.3)
The combination of the nanosized metal and buffer matrix of Li2 0

to reversible alloying reactions (8.3) that have much improved
sibility than traditional LixMe alloys. More recent studies have focused
he use of MexOy initial materials in which Me is a metal which does not
ort alloying reactions with Li. These materials were found to convert
g the cathodic scan analogous to reaction (8.2); however this
Chapter 8
ersion reaction was found to be reversible (8.4):
(8.4)
The quest for new inorganic anodes has recently proceeded into areas
ch have been largely ignored, namely, nontraditional MexXy materials,
re the anion X is N or P. The number of papers on these type of
erials is relatively small, but nonetheless the electrochemistry is found to
urprisingly rich and somewhat exotic. In this chapter, we take a close
at the work that has been done to this point on binary and ternary metal
des along with a look at the often overlooked metal silicides.
. THE NITRIDES
.1. The Ternary Nitrides
1.1. The Antifluorites (Lh,..1MNx)
Metal nitrides have been investigated for their properties related to fast
um ion conduction for the past two decades. The most famous of these
des is the highly conducting Li 3N 1- 7 which exhibits ionic conductivity on
2
order of 1'10- S/cm. A number of papers have been published on this
rial; however, because of its poor resistance to anodic decomposition at
potentials, it has found little use as a solid state ionic conductor in a
tical cell. In spite of this drawback, the initial work spurred interest in
nvestigation of alternative nitrides that could also support appreciable
conductivity as a result of the low polarizability of the nitrogen anion.
y of these structures are related to Li2x_ 1MNx which is based upon the
luorite structure. Containing pathways for facile ion movement, these
rials exhibited appreciable Li+ conductivity. Materials such as Li 3BN2, 8
iN3,9 •10 and Li 3AlN2 all share this structure type and have well
lished ionic activity. Due to their electronically insulative properties,
materials were investigated solely for use as solid state ionic
uctors.
The facile movement of the lithium ions within the antifluorite
ture gave rise to the question whether such structures could be utilized
radaic insertion intercalation electrodes for lithium batteries. This issue
addressed by the use of transition metals as a substitute for M in Lizx-
x. The resulting Lizx_ 1MNx structure gained appreciable electronic
uctivity to parallel the existing ionic conductivity. The materials, due to
ubstitution, were mixed conductors, opening the possibility for their use
e and Silicide Negative Electrodes 249
egative electrode materials. In light of this, the electrochemical activity

i2x_ 1MNx materials were investigated particularly for materials containing
Mn or Fe.
e 8.1. Crystal structure of Li 7 MnN4 from the [110] viewpoint. Dark spheres represent
nN4 can be fabricated through solid state reaction between Li 3N and Mn

Mn4N in a N2 atmosphere. The material crystallizes in the cubic
luorite structure built of isolated MnN 4 tetrahedra where Mn and Li are
ted on tetrahedral sites (Figure 8.1). Li and Mn are present in ordered
ngements with clearly observable three dimensional pathways for
sion. Despite the high formal charge associated with the Mn cation
5
+), the structure is prelithiated; therefore, the cycling of the material
mences with lithium extraction. This feature makes it particularly useful
Li-ion batteries utilizing non-lithiated positive electrodes. The
hiation of Li7MnN4 was found to proceed to Li5.45MnN4, centered on
aus at approximately 1.15 V (Figure 8.2). The initial 0.625 lithium
cted corresponded to a two-phase delithiation in a unit cell with same
metry but different lattice parameter. The remaining lithium was
ved in a single-phase reaction. The structural transformations are
rsible and the material cycles well with a stable 300 mAh/g specific
city.
In Li7MnN4 the Mn cation exhibits an unusually high nominal charge
5
n +. Even more surprising, assuming an ionic model, the Mn cation
6 7
eeds to a mixed oxidation state of Mn +/Mn + upon charge. These
ual electronic attributes combined with the material's relatively low
11
tion potential of 1.2 V spurred Suzuki and coworkers to a deeper
stigation of the electronic attributes of the material leading to its active
rochemistry.
Chapter8
2.0
i§ :::--_; --·---- --- .\. ,

0.5
200 400 600 800 1000
Capacity (mAh/g)
e 8.2. Voltage profile of Li 7MnN4 (from Ref. 12).
The investigation was accomplished through the use of core level

ron energy loss spectroscopy. It was found that in the ground state, there
a number of holes on the N 2p band. This is consistent with the strong
idization of the Mn 3d and N 2p orbitals associated with the covalency
he compound. Analysis of the lithium deintercalated Lh-xMnN 4 where
revealed a strong growth in the N-K spectra related to the amount of 2p
cupied states. Simply stated, lithium deintercalation proceeds with an
ation reaction introducing holes into the nitrogen 2p orbitals. This is in
contrast to the classical situation where holes are introduced into the
ition metal (manganese) 3d orbitals. Therefore, one can make the
ralization that the electronic redox is centered on the N rather than Mn
explains the low redox voltages associated with the Li7MnN4 electrode
rials.
As opposed to Li7MnN4, Li3FeN2 crystallizes in orthorhombic
metry. However, fundamental similarities exist as the structure is a
rted form of the antifluorite structure. The structure can be simply
acterized as Fe4N tetrahedra sharing edges to form one-dimensional
ns along the c direction.
The structural evolution during delithiation was characterized in depth
12
ishijima and coworkers. Delithiation occurred at approximately 1.2 V
he range of Li3FeN2 to LizFeN2• Four phases appeared during the
hiation reaction, all were of orthorhombic symmetry. Relithiation
rred with systematic redevelopment of the phases evolved during
hiation. The reversible lithiation processes occurred with little
/under potential and the crystallinity of the material is maintained for
It should be noted that delithiating Lh-xFeN2 beyond Li 2FeN2 resulted in
ecomposition of the crystal structure.
Question still remains to the electronic basis of the redox reaction of

eN2. Nishijima et al. concluded that the redox was classical and centered
3 4
he reversible Fe + -+ Fe + reaction. This conclusion was centered on
lts of Mossbauer spectroscopy. Half a decade later, Suzuki et al. realized y
similarities related to the structural electrochemical features of eN2
and believed the electronic reaction to be similar to that of Li7MnN4
entered electronic redox) although no supporting spectra was offered.
.2. The Hexagonal Lh(Lh.xMx)N Structures
The second major class of ternary lithium transition metal nitrides

s in a structure which is isostructural with the well known Li3N lithium
onductor. These materials take the stoichiometric form of Li 2(Li 1.xMx)N
13
its better known notation, Li3_xMxN. where M= Co, Cu, and Ni. The
structure (Figure 8.3) can be simply envisioned as u+ between layers
1
hN -. The transition metal cations substitute for the Li between the
s; however, it should be noted that this site is slightly unusual as it
ists of a linear N-M-N bond along the c axis and is not octahedral or
hedral in nature. From an electrochemical standpoint, the most
ising and most studied of the transition metal substitutions is Co. 14
..,..,...........,."'"-.:LizN Layer
(Li\M+(l·x)) Layer
01(1
e 8.3. Schematic of Li2(M 1_xMx)N structure.
Lh-xCoxN materials can be prepared by the reaction of Li 3N and Co

l under a N2-based atmosphere. The solid solution range for Lh-xCoxN is
It was quickly identified that the optimum specific capacity could
alized from a composition of Liz. 6Co0.4N.
Small amounts of Li can be inserted into Li3_xCoxN during the initial
arge reaction between 0.5 and 0.1 V (Li/Li+). The amount of lithium
ChapterS
rted is related to Li vacancy population but the specific capacity does not
ed 100 mAh/g. Practically speaking, Liz.6Co0.4N can be treated as being
he lithiated state; therefore, lithium must be deintercalated from the
cture in an initial anodic oxidation. During the first charge, Li 2.6Co0.4N
les the removal of approximately 800 mAh/g of charge at an extended
au at 1.1 V corresponding to a delithiated composition of approximately
Co0.4N. Subsequent discharge and charge results in the reversible
rtion of 800 mAh/g from 1.5 to 0.1 V (Figure 8.4). After the initial
hiation cycle, subsequent delithiations remove capacity in a
2.0,..----------------,
200 400 600 800

Capacity (mAhlg)
e 8.4. Voltage profile of Li2.6Cao.4N vs. Li metal counter electrode. Cycle numbers are
ated (from Ref. 15).
dolinear profile at voltages between 0.5 and 1.1 V. Excellent cycling

lity is observed for this material.
Ex-situ X-ray diffraction revealed that the delithiation reaction
eeds topotactically for Llx=l in Lh 6-xCo0.4N. In this region the reversible
ific capacity is 480 mAh/g and the crystal structure does tend to become
15
rphous with cycle number. More recent data suggests the initial
hiation reaction proceeds concurrent with a systematic and gradual loss
ystallinity. Regardless, extending the initial lithium removal to Llx=l.46
ts in the systematic destruction of the crystal structure and formation of
-ray amorphous compound. It is easy to understand from the inspection
e crystal structure (Figure 8.3) how excess u+ deintercalation can lead
s destabilization and loss of crystallinity. Subsequent cycles revealed an
essively stable specific capacity of approximately 800 mAh/g. To gain
insight on the faradaic activity leading to the electrochemical activity
i2.6Co0.4N, a more localized investigation of the electronic structure of
aterial had to be taken by alternative techniques.
As with the Li 7MnN4 antifluorite material discussed in the previous

ion, a number of mysteries exist with respect to the electronic structure
e optimized Liz.6Co0.4N. Utilizing a basic ionic stoichiometry model, the
2.6 0.4 1 3
ge
Li distribution of Li Co N should be +26Co +0 .4N-. Although
1
ovalent cobalt is very rare, characterization by EELS 16 revealed

rising data consistent with the existence of the monovalent cobalt cation.
zing this stoichiometric model, delithiation to Li1. 0Co0.4N would require
to adopt the even more unusual Co5 + oxidation state. EELS
surements were performed on the fully delithiated sample. Figure 8.5
s the N-K spectra of that sample. The band at 403.5 eV is ascribed to
hybridization of the nitrogen 3p band with the lithium 2s band. The
395 400 405 410 415

Energy Loss (eV)
e 8.5. N-K EELS spectra of lithiated and delithiated Li 2.6Cao.4N (from Ref. 15).
hiated Liz.6_xCo0.4N sample reveals an expected decrease of the 403.5

eak consistent with the removal of Li and subsequent decrease of Li-N
s.
The peak at 400 eV is associated to the unoccupied nitrogen 2p band.
thium is removed the intensity increases drastically, consistent with the
lopment of holes on the nitrogen 2p band during electrochemical
hiation. EELS analysis of the cobalt indicated the oxidation state to be
oximately Co3+ after oxidation. This would require the removal of the
ining electrons from the nitrogen bands to effectively charge
pensate. This interplay is clear from the schematic of the electronic
ture of Li2.6Co0.4N. Therefore, like in the case of Li7MnN4, the nitrogen
n plays a crucial role in the redox chemistry of
Li2.6Co0.4N.
Other Lh-xMxN based materials exist. Early on, Lh-xMxN materials
e M = Ni or Cu were also investigated. These materials also crystallized
e space group of P6/mmm. The Li2.6Cu0.4N sample exhibits good
cycling
ChapterS
ility, on par with that of Liz,6Co0.4N. The copper compound revealed 650
h/g in the 0-1.3 V range. The Ni system, in contrast, was found to show
versible specific capacity of only 180 mAh/g in the range of 0.0 to 1.4 V.
Most recently, a layered lithium nitride has been investigated in the
17
uit of a less toxic and more cost effective negative electrode material.
eN2 is a layered nitride with hexagonal LhN layers analogous to the Lh_ N
materials described above. A similar material Li2(Li0.7Fe0.3)N
7Fea.3N) was investigated for its electrochemical performance. The
rial exhibited 550 mAh/g at an average potential (between 0.05 and 1.2
f 0.6 V. Analogous to Li 2.6Co04N, the structure decomposes to a quasi
rphous state upon delithiation.
2.2. The Binary Nitrides
Some of the earliest work on binary nitrides as alternatives for

hite or Li has come by way of thin film work. One of the first
18
pounds to be investigated was that of Sn3N4• Sn3N4 films can be
ared by reactive sputtering_ of Sn in a Ar-N 2 gas mixture. The films
tallized in the hexagonal P 3 ml space group expected for this material.
larly, films of InN were deposited. The lithium reaction mechanism for
of these consisted of a straightforward conversion reaction to a LixSn or
alloy embedded in a matrix of LhN. Subsequent specific capacity is
loped through the alloying reaction with Sn or In, as LhN remains as a
er matrix. As expected, the alloying reaction was found to give relatively
cycling due to the electromechanical grinding induced by large volume
ges within the alloy.
Ge3N4 crystallizes in a hexagonal structure with large tunnels

retically capable of lithium intercalation parallel to the c axis. Lithium
found to be able to reversibly react with Ge3N4 at a voltage of 0.25 V
a reversible capacity ranging from 500 to 1200 mAhlg depending on the
19
allite size (Figure 8.6). Finer crystallites were found to offer much
oved specific capacity. The low voltage at which Ge3N4 is
rochemically active makes it especially suitable for use as a negative
rode in Li-ion. The cycling stability of Ge3N4 was found to be quite
with less than 15% loss after 70 cycles (Figure 8.7). In-situ and ex-situ
D revealed that Ge3N4 experienced a large loss of crystallinity during the

discharge. This, combined with an irreversible loss of 30-50% active
rial and a 50% irreversible capacity loss, suggested a portion of the
rial converted during the first cycle concomitant with an alloy
ation:
+ 3x)y Li+ + (12 + 3x)y e· + Ge3N4--+ 3y LixGe + 4y LhN + (1-y)

Ge3N4
(8.5)
().1.5V- 23mAig
-. SO% Irreversible Loss
1.5
c
y
0 0.5
;;.
0.5 1.5 2 2.5 3.5

#LiperGe
e 8.6. Voltage profile of crystalline Ge 3N 4 cycled vs. Li metal between 0 and 1.5 Vat a
ant current of 23 rnA/g.
- 1000 0-0.7V • 23mA/g

•
1 800
E
1 E 600
c.
...
u
..
0 1t>.O 400
;
-.l/l.=.. 200
Q LW LW LL LLU
0 10 20 30 40 50 60 70 80
Cycle#
e 8.7. Cycle life of crystalline Ge 3 N 4 cycled vs. Li metal between 0 and 0.7 V at a
ant current of 23 rnA/g.
Chapter 8
The reaction products were of exceptionally low crystallinity and

e characterization difficult.
The remaining cycles are believed to incorporate themselves within a
rsible alloying reaction (8.6) (supported by selective area electron
action) on crystallites composed of alloy "shells" grown on core
ctures of unreacted Ge3N4 (Figure 8.8).
ixGe + 4y LhN + (1-y) Ge3N4 !::+ 3y Ge + 4y Li3N + (1-y) Ge3N4 +

u+ +3xy e" (8.6)
The X-ray amorphous nature of the alloy combined with a substrate of

N4 is believed to bring about the uncharacteristically good cycling
ormance for en electrode based on lithium alloying.
Residual Ge3N 4
Amorphous
•Li. Ge +Li3N
e 8.8. Schematic representation of a Ge 3N 4 particle exhibiting an unreacted G N 4 core

unded by an electrochemically active converted shell.
The electrochemical behavior of zinc nitride thin films versus metallic

20
as first reported by Bates et al. It was postulated that the reaction was
lar to the irreversible conversion reaction taking advantage of the
ing properties of Zn. Zn metal alloys easily with Li and as such the Li
21 22
hase diagram exhibits five alloys, LiZn4, LizZn5, LiZn2, LizZn3 and
.
(6+3x) Lt + (6+3x) e- + Zn3N2 -+ 3 LixZn + 2 LhN (8.7)

3 LixZn + 2 LhN !::+ 3 Zn + 2 Li3N + 3x Lt + 3x e- (8.8)
Recent analysis of powder Zn3N2 revealed finer attributes of the

23
rochemistry which was not initially obvious. Zn3N2 reversibly reacts
lithium electrochemically, exhibiting a large reduction capacity of 1325
/g corresponding to the insertion of 3.7 Li per Zn. Of this initial
city, 555 mAh/g was found to be reversible. In-situ and ex-situ X-ray
action revealed the reaction mechanism with lithium was a conversion
tion of Zn 3N2 into LiZn and a matrix of LhN, the high-pressure analog
i3N:
(8.9)
Upon oxidation, LiZn transformed into metallic Zn, while Li3N

ributed to the transformation into LiZnN:
LiZn + 2/3 LhN-+ 7/3 e· + 7/3 Lt + 1/3 Zn + 2/3 LiZnN (8.10)
This reversible metal - Li 3N conversion mechanism is analogous to

mechanism (8.4) that was previously isolated to oxides and sulfides.
equent cycles proceed reversibly using the LiZnN and Zn as the
ursor materials. Indeed, the true reaction could be best described as two
llel reactions, one consisting of the reversible alloying reaction of the
Zn material, and the second consisting of a reversible conversion of the
N phase to the LiZn alloy plus Li 3N:
1/3 [LiZn !::+ e- + Li+ + Zn] (8.11)
2/3 [LiZn + f3Li 3N !::+ 3 e- + 3 u+ + LiZnN] (8.12)
The formation of LiZnN as the new end member of the

rochemical reaction with lithium was identified as the cause of the
ersible loss observed during the first cycle. Therefore, LiZnN can be
ed as a negative electrode material from the beginning as it would
bit 930 mAh/g with no initial irreversible capacity loss.
The cycle life of Zn3N2 was found to be poor. The poor cycle life was
ly attributed to the electromechanical grinding of the Li-Zn alloying
ion which contained appreciable volume change with alloy
position.
Cu3N crystallizes in the anti-Re0 3 cubic structure type. The structure

ists of NCu6 octahedra connected at the corners resulting in a 3D array
3D tunnels. Cu3N was found to involve a complex series of
rochemically induced transformations, of which the origins are not yet
24
known. The initial cathodic reduction of Cu 3N with u+ involved a
-step reaction mechanism (Figure 8.9). X-ray diffraction revealed that
attice parameter was found to expand during the first step. This was
ChapterS
2.5
2
>
';' 1.5
0.0
1 SEI
0.5
0 2 3
#LiperC N
e 8.9. First cycle voltage profile of a Cu 3N electrode cycle vs. Li metal between 0 and 3
constant current of 22.4 mA/g (equivalent to a C/6 rate).
istent with lithium intercalation into the open structure of Cu3N

urrent with copper displacement. Copper metal was detected by SAED
is region and the reaction was found to be irreversible. The second step
e electrochemical process was identified, using a combination of XRD,
D and TEM, as the conversion of the remaining copper nitride into a Cu
3N nanocomposite which promoted the formation of an organic layer at
urface of the nanocomposite. The latter two processes were found to be
rsible. The reversible nature of the Cu + Li3N nanocomposite is similar
at observed for the reversible nitride conversion mechanism of Zn3N2•
(8.13)
Cu3N electrodes showed excellent cycling efficiency at high rate

ivalent to 1.67 C) (Figure 8.10). Extended cycling seemed to promote
evelopment of electrochemical processes above 2V during delithiation
eas the cycling stability of the copper nitride conversion process was d
to be poor. These seem to be associated with the formation during
hiation of Cu20 and CuO believed to be rooted in electrolyte
adation. The reversible copper oxide conversion processes may
ribute to the increase in capacity with cycle number.
3. Summary - Nitrides
The metal nitrides exhibit an exceptionally wide range of reaction

hanisms with lithium, ranging from metal nitride conversion and
equent lithium alloying, reversible conversion, redox reactions
e and SiliridP NPuntivP PIPrtrnrlP<' 259
700
c:
.. . : 600
.§,
c
· ;:;
500
400
"-
u"'
2 "?
300
.."c::' 200
\.
-
100
i5
0 200
0 50 100 150
Cycle#
e 8.10. Cycle life of Cu 3N cycled vs. Li metal between 0 and 3 V at a rate equivalent to
c.
ciated to the anion, to simple topotactic insertion. The reaction

hanisms of these materials are closely tied to the electronic configuration
thermodynamic stability of the phases upon reduction. Some of the
l nitrides' uniqueness is a result of the low electronegativity and
lent character of the nitrogen atom. This places the Fermi level close to
of the nitrogen 2p bands as opposed to the transition metal 3d bands
ally encountered in oxides. These materials have also presented to the
archers an added complexity to material characterization to uncover
e mechanisms as many of them occur within quasi-amorphous structures.
It is still too soon to comment on the expected viability of metal
des as negative electrodes in Li-ion batteries of the future. However,
Co0.4N has shown its viability by being successfully implemented into
25
n cells. High specific capacities, exceptional cycling stability and
voltage profiles have been demonstrated within the metal nitrides
ussed in this chapter. However, the combination of all of these attributes
g with low air sensitivity and cost have yet to be demonstrated in one
rial. Regardless of the eventual outcome, the nitrides offer a new path
rds next generation negative electrodes with a rich electrochemistry
ed by few other materials.
SILICIDES
1. Introduction
Binary lithium alloys have long been considered as alternative

tive electrode materials as they exhibit greater specific capacities,
cially volumetric capacities, than carbonaceous materials. The Li-Si

m shows the highest theoretical capacity of over 4000 mAh/g with the
ChapterS
ation of Li44Si. Unfortunately, the cycle life of binary alloys is poor as

e volume changes are associated to repeated alloying/dealloying
lting in the pulverization of the electrode. One area of research to
ove the performance of Li-Si alloys focuses on silicon-based
composites. Silicon has been dispersed in different types of materials
bonaceous materials, ductile metals, TiN, etc...).
An alternative approach to the silicon-based nanocomposite considers
investigation of silicides for use both at elevated and ambient
erature. The examination of several binary silicides26' 27 showed only a
reacted with lithium significantly at ambient temperature. This section
mostly focus on magnesium silicide, Mg2 Si, which exhibits interesting
rochemical properties, such as high capacity and low potential, even at
ient temperature.
2. Ternary Li-Si-M Systems at Elevated Temperature

.1. Thermodynamics
Anani and Huggins28 proposed a methodology, based on the principles

rnary thermodynamics, to predict how the addition of a third component
binary system could alter both the capacity and the potential of the
y counterpart. They aimed at ternary alloys, using Li-Si-M alloys for
tration, with improved electrochemical performance for use as negative
rodes in thermal batteries, using FeS2 as cathode material and operating
e 400-450°C temperature range.
The reactions of an initial Si-M alloy with lithium were predicted
g the isothermal equilibrium phase diagram of the Li-Si-M ternary
m at the adequate temperature (400°C in this case). The construction of
phase diagrams was based on the free energy of formation, or Gibbs
energy, of all the binary and ternary phases known to exist in the Li-Si
ystem. An estimation of the theoretical potential, at which the
ibrium reactions (or three-phase equilibria) should occur, was obtained
the Nernst relation. Table 8.1 shows the theoretical potential and the
etical capacity of several M-Si alloys (where M = Ca, Cr, Mg, Mn, Mo,
Ni, Ta, Ti and V) used as starting materials. The maximum theoretical
ific energy (MTSE) values provided in the table were obtained as
ws:
MTSE = (26.805 · n · Eav) I W (8.14)

8.1. Estimated thermodynamic properties of several Li-Si-M ternary systems (from

28).
em • Starting Phases in equilibrium Electrode Electrode MTSE

composition potential capacity (Whlkg)
(mV vs. Li)
Li/Si Li/g
(x103)
-Si Li7 Si3 Li7 Si3-Li13Si4 158 0.92 18.1 428
Mg Mg2Si Mg2Si-Mg-Li13Si4 60 3.25 32.7 574
Mn MnSi MnSi-Mn 5Si3-LhSi3 163 0.93 10.4 316
MnsSi3 MnsSi3-Mn3Si-Li13Si4 45 1.44 11.1 350
Mn3Si Mn3Si-Mn-Li33Sis 43 4.40 19.7 474
Mo MoSi2 M oSi2-M osSi3-Li13S i4 120 2.28 24.8
502
MosSi3 MosSi3-Mo3Si-Li22Si5 3 1.96 9.7 328
Ni Ni1Si13 Ni 7SiwN iSi-Li12Si7 316 0.79 12.1 317
Nb NbSi2 NbSi,-NbsSh-LhSi3 184 1.63 19.0 432
Ta TaSi2 TaSi2-Ta5Si3-Li7 Si3 211 1.63 12.6 344
-V VSi2 VSi,-V 5Si3-Li1Si3 191 1.63 25.2 486
Ca CaSi CaSi-Ca2Si-Li22Sis 13 2.2 26.4 544
Cr CrSi2 CrSi2-CrSi-LhSi3 223 1.17 18.8 420
CrSi CrSi-Cr5Si3-Li7 Si3 205 0.93 10.8 315
Cr5Si3 CrsSi3-Cr3Si-Lit 3Si4 138 1.44 11.6 341
Ti TiSi TiSi-Ti5Si3-LhSi3 182 1.55 11.3 328
Li-Si
re n is the number of moles of lithium involved in the reaction, Eav is the

rence between the positive (FeS2) and negative electrode (M-Si alloy)
ntials and W is the weight of active material. Because the ternary Li-Si
Li-Si-Mo and Li-Si-Mg systems exhibited improved maximum
retical specific energy (544, 502 and 574 Wh/kg, respectively) with
ect to the binary Li-Si system (428 Whlkg), they appeared as potential
native negative electrode materials for thermal batteries.
2.2. The Li-Si-Mg System
The theoretical results provided in the previous section were verified

29
rimentally using the ternary Li-Si-Mg system. ' 30 Mg2 Si is the only
metallic compound existing in the Mg-Si system. Thermodynamics
icted a three-phase equilibrium at 400°C that can be depicted by the
wing reaction:
4 Mg2Si + 13 Li 8 Mg + Li 13Si4 (8.15)
According to equation (8.15), the theoretical potential and capacity of

Si upon reaction with lithium correspond to 0.06 V vs. Li/Lt and 876
ChapterS
h/g, respectively. Coulometric titration techniques with current densities

he 5-20 mNcm2 range were performed at 440°C using three-electrode
. Mg2Si was used as working electrode and a 44% Li-Si alloy served as
ter and reference electrode while the electrolyte consisted of a LiCl-KCl
ctic. The resulting Mg2Si voltage profiles showed a specific capacity of
mAh/g at an average potential of 0.07 V vs. Li!Lt; these data agree
with the theoretical values predicted by thermodynamics (Table 8.1). In
of these results, the use of Mg2Si as negative electrode would provide a
of higher specific energy than a cell utilizing a Li-Si binary alloy (based
he Li7Sh-Li 13Si4 equilibrium).
3. Silicides at Ambient Temperature
Two conflicting mechanisms have been found for the reaction of

Si with lithium at ambient temperature. One scenario involves the
calation of lithium into Mg2Si to form Li 1Mg2Si while the formation of
ique ternary LizMgSi phase intervenes in the second scheme.
.1.1. Formation of Li1Mg 2Si
Huang et al.31• 32 reported the insertion of up to 2 lithium per Mg2Si

is equivalent to a specific capacity of 700 mAh/g. The working
rode, constituted of 98 wt% Mg2Si active material and 2 wt% ethylene
ylene-diene binder, was tested versus lithium metal down to 0 V at
58 mNcm2 . X-ray diffraction showed the material maintained its
alline structure upon reaction with up to one lithium but became
rphous thereafter. As opposed to the results predicted by
modynamics at 400°C, there was no evidence of Mg and Li 13Mg4
phases mbient temperature. In the ternary Li-Si-Mg isothermal
equilibrium
e diagram at ambient temperature, a line that does not exist at 400°C
ects Mg2Si to Li. Mg2Si crystallizes in an antifluorite structure where Si
pies the octahedral sites to form a face-centered cubic arrangement
e Mg occupies the tetrahedral sites. This structure offers vacant
hedral sites favorable for lithium intercalation that supports the theory of
gle-phase intercalation process with the formation of a Li 1Mg2Si phase.
Kim et al.33 investigated the electrochemistry of Mg2Si versus lithium
more detail as it exhibited an initial capacity of 1370 mAh/g
sponding to the insertion of up to 3.9lithium (Figure 8.11). This value
exceeded the capacity reported by Huang et al.31' 32 at ambient
erature (2 lithium) and also the capacity predicted by Anani and

29 30
gins • at elevated temperature (3.25 lithium) (Table 1.1). Although the
+ third second first

\ t/
I
...J
1.5
c:.
ii 1.0
:;::;
l
r:::
.!
0
0.5
11.
0.0 0 200 400 600 800 1 000 1200 1400
Capacity (mAh/g)
e 8.11. First three cycles voltage profile of an Mg 2Si electrode cycled vs. Li metal
en 0 and 2 Vat a constant current of 10 mA/g (from Ref. 33).
rode exhibited high initial cycling efficiency (78.4% in the first

e), the electrode failed within 10 cycles.
The evolution of the structure was examined using ex-situ X-ray
action. The Mg2Si structure, which remained crystalline during cycling,
nded during lithiation and contracted during delithiation. The
nsion/contraction of the lattice with cycling was consistent with an
calation process. The emergence of Mg metal at 0.17 V and a Li-Mg
on further lithiation suggested the decomposition of the LixMg2Si
e. Although no evidence of any Li-Si phase was obtained by XRD,
er electron spectroscopy performed in the lithiated stated revealed the
xistence of a LixMg2Si phase with Li-Si and Li-Mg alloys. Similarities in
differential capacity plots of the Mg2 Si electrodes with the ones ined
for Si-based materials, such as Si metal and SiO, and Mg also
ested the formation of Li-Si and Li-Mg alloys, respectively. In brief,
um was proposed to intercalate into the vacant octahedral sites of the
Si antifluorite structure until the concentration of lithium in the LixMg2Si
hed a critical value. At this point, the material decomposes into Li-Si s
and Mg, which alloys with lithium on further lithiation.
The rapid degradation of Mg2Si electrodes was proposed to result from
large volume changes associated to the Li-Si and Li-Mg alloying
ions. Scanning electron microscopy revealed cracking and crumbling
10 cycles while XRD analyses showed an increase of the amount of
ual Mg metal and Li-Mg with respect to the Mg2Si phase with cycle
ber.
ChapterS
3.1.2. Formation of a Ternary Li2MgSi Phase

34
Moriga et al. proposed a two-step reaction mechanism for the
tion of Mg2Si with lithium at ambient temperature that differed from the
+ ..L
§ 1•• step
I
2•d step
L:r
.;
>
=----
Ill
Cl
0
> 0.1
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
x In Li.Mg2Si
e 8.12. First discharge voltage profile of an Mg 2Si electrode cycled vs. Li metal at a
2
ant current of 0.5 mA/cm (from Ref. 34).
described above. The first discharge voltage profile in Figure 8.12 shows
voltage remained quite flat during the insertion of the first two lithium
e it gradually decreased during further lithiation. The first step was
ved to involve the extraction of Mg from Mg 2Si concomitant to the
ation of a ternary LhMgSi phase (equation 8.16). The second step was
ciated to the alloying reaction of lithium with the extracted Mg metal
ation 8.17).
Mg2 Si + 2 Li+ + 2e· . LhMgSi + Mg (8.16)

Mg + y Lt + y e· . LiyMg (8.17)
XRD patterns obtained at various points of discharge revealed Mg l

and a new cubic phase emerged while Mg2Si vanished. The new cubic e
was associated to LhMgSi based on the XRD pattern simulations of
-aSi with half-occupation of the 8c Mg sites. On further lithiation, Mg
l disappeared and a body-centered cubic Li phase identified as a Li-Mg
appeared. Good correlation existed between the XRD pattern obtained
rimentally upon insertion of 3.5 Li and the JSRD pattern obtained by
lation using the LhMgSi structure with a Fm 3m space group depicted
gure 8.13. The later has a 2aFX2aFX2aF structure, where aF is the lattice
meter of the initial Mg2Si cubic structure. In the proposed structure, two
of LizMgSi were alternatively arranged. The second unit was obtained

otating the first one clockwise for 90° about the a axis. Finally, DV-Xa
ulations revealed electrons were introduced in the Mg 3s orbitals for
ge compensation during lithium insertion, resulting in the reduction of
+ ions. This was consistent with the extraction of Mg (supported by
) from Mg2 Si.
35
More recently, Roberts et al. confirmed the formation of Mg metal
LizMgSi by XRD during lithiation of Mg2Si. Because the reconstitution
ions in the Li-Si system are kinetically hindered at ambient temperature
36
only selective equilibrium is achieved, the Li-Si alloys may be difficult
etect using XRD. Therefore, although the structural characterization of
lectrode at different points of charge and discharge showed no evidence
i-Si alloy formation, the authors proposed electrochemical evidence for i
alloying with extensive cycling. The anodic differential capacity plot n
Mg2Si electrode lithiated down to 5 mV exhibited a peak in the same ge
range as obtained for pure silicon. Finally, the material was shown to form
back to Mg2Si during delithiation.
Si
Li
(A) (B)
a
Tetrahedron
occupied by Mg
/ ,L. /
;>
r
--
--....
./ ' '{3)
12.83A <"
/
--
1 / (A) ' \:J /,
/
e 8.13. Schematic illustration ofLi 2MgSi structure (from Ref. 34).
.2. CrSh
36
Lithium was found to intercalate into the CrSiz structure. Because the
ChapterS
rochemical insertion of lithium into CrSi2 is very slow at ambient

erature, ternary Li-Si-Cr samples were fabricated by reaction of pure
um with CrSi2 at 360°C using various initial Li:CrSiz stoichiometry
s. The resulting materials were characterized by XRD in order to
tify the chemical reactions of CrSiz with lithium. Table 8.2 exhibits the
med chemical reactions along with the initial Li:CrSiz ratio at which the
rent phases were observed. It also provides the theoretical capacity
ciated to these reactions and the experimental delithiation capacity.
ium intercalation occurs for initial stoichiometry ratios up to 60:40. A
er increase in the initial lithium content results first in the formation of
h and LhSh, then Cr5Si3 transforms into Cr3Si and finally Li 13Si4 and
Si5 were observed. No ternary phase was found.
The Li-Si-Cr samples prepared by solid-state reaction were tested
us lithium metal under a constant current equivalent to a C/60 rate. The
ples exhibited lithiation capacities ranging from 661 to 811 mAh/g and
rsible delithiation capacities of up to 800 mAh/g that are rooted in binary
i alloying reactions. Although the theoretical specific
cities of ternary Li-Si-Cr system are smaller than the binary Li-Si
em, the capacities obtained experimentally for the ternary system
8.2. Reconstitution reactions as found from X-ray diffraction analyses after chemical
ion of CrSi2 (from Ref. 36).
sumed chemical reaction Found for initial Calculated Experimental

stoichiometry capacity delithiation
ratio of Li:CrSi2 (mAh/g) capacity (mAh/g)
hium intercalation into CrSi2 up to 60:40
CrSh + 49 Li - 3 Cr5Si, + 7 Li,Si3 75:25
669
rSi2 + 35 Li - 3 Cr3Si + 5 Li7Si3 80:20 771
496
CrSi2 + 65 Li - 4 Cr3Si + 5 Li 13Si4 85:15 996
682
rSh + 21 Li- Cr 3Si + Lh1Si5 88:12 1197 836
eded the ones obtained with the binary system. This benefit in capacity
be related to the Cr-Si phases. The latter remain mostly
rochemically inactive after the first discharge and may act as stabilizing
uctive matrix.
.3. NiSi
Preliminary results showed nanosize NiSi synthesized by high-energy

ng reacts with lithium reversibly.37 The material consisted of
omerates of small particles with an average crystallite size of 4-6 nm as
de and Silicide Negative Electrodes 267
rmined from the width of Bragg peaks using the Scherrer equation. The
harge voltage profile (Figure 8.14), which is smooth and approximately
ows a half parabolic curve down to 0 V, shows high initial capacity, 1180
h/g. Even though 80% of the initial discharge capacity was reversible,
capacity fade during cycling is significant as NiSi electrodes retained
65% of the initial discharge capacity after 25 cycles. Based on previous
rts of Sn-Sb materials, NiSi was proposed to decompose into Li-Si
ys and Ni during the first lithiation (equation 8.18). In subsequent charge
discharge, the Li-Si alloys reversibly transform into Si metal, which
titutes the active center of the electrode, while Ni metal remains
trochemically inactive (equation 8.19).
NiSi+ Xu++ X e' -+ LixSi + Ni (8.18)

LixSi + Ni Si + Ni +Xu++ X e' (8.19)
4. Summary - Silicides
Several binary silicides have been examined as potential negative

trode materials first for thermal batteries and then for batteries operating
2.5
2.0
..J 1.5
C:.
Gl 1.0
g 0.5
0.0
0 200 400 600 800 1000 1200

Discharge Capacity (mAh/g)
e 8.14. Discharge voltage profile of a nanosize NiSi electrode cycled vs. Li metal at a
ant current of 0.08 mNcm2 (from Ref. 37).
mbient temperature. Using Mg2Si as an illustration, thermodynamics was

en to be an excellent tool for screening silicides with improved specific
city and energy with respect to the binary Li-Si system for use in
ChapterS
mal batteries operating in the 400-450°C temperature range. Mg 2Si was

found to react with lithium reversibly at ambient temperature. Even
gh this system exhibits initial high capacity, its short cycle life limits its
zation in commercial applications. The rapid failure of the electrode was
buted to the large volume changes associated to the Li-Mg and Li-Si
ated alloying/dealloying reactions occurring during cycling. These alloys
lt from the decomposition of a LixMg 2Si solid solution or a ternary
gSi phase during lithiation. There still exist a conflict as for the actual
tion mechanism of Mg2Si with lithium at ambient temperature. CrSh and
were also investigated. In general, silicides show high capacity at low
ntials that make them very attractive as negative electrodes for Li-ion
ries; however their poor cycle life needs to be improved.
FERENCES
. Knutz, S. Skaarup, Solid State lonics 9&10 (1983) 371.

M. Richards, J. Solid State Chern. 33 (1980) 127.
Bittihn, Solid State Ionics 8 (1983) 83.
. Meyer, H. Rickert, U. Schwaitzer, Solid State lonics 9&10 (1983) 689.
. F. Bell, A. Breitschwerdt, U. Alpen, Mater. Res. Bull. 16 (1981) 267.
R. Rea, D.L. Foster, Mater. Res. Bull. 14 (1979) 841.
A. Boukamp, R. A. Huggins, Mater. Res. Bull. 13 (1978) 23.
. Yamane, S. Kikkawa, M. Horiguchi, M. Koizumi, J. Solid State Chern. 65 (1986) 6.
Juza, H. H. Weber, E. Meyer-Simon, Z. Anorg. AUg. Chern. 48 (1953) 273.
H. Yamane, S. Kikkawa, M. Koizumi, Solid State lonics 25 (1987) 183.
. Suzuki, T. Sodai, Solid State lonics 116 (1999) 1.
M. Nishijima, Y. Takeda, N. lmanishi, 0 Yamamoto, N. Takano, J. Solid State Chern.
113 (1994) 205.
. Asai, K. Nishida, S. Kawai, Mat. Res. Bull. 19 (1984) 1377.
M. Nishijima, T. Kagohashi, M. Imanishi, Y. Takeda, 0. Yamamoto, S. Kondo, Solid
tate lonics 83 (1996) 107.
. Shodai, Y. Sakurai, T. Suzuki, Solid State lonics 122 (1999) 85.
S. Suzuki, T. Shodai, J. Yamaki, J. Phys. Chern. Solids 59 (1998) 331.
. Rowsell, V. Pralong, L. Nazar, J. Am. Chern. Soc. 123 (2001) 8598.
. Neudecker, R. Zuhr, Electrochemical Society Proceedings 99-24, p. 295.
N. Pereira, M. Ba1asubramanian, L. Dupont, J. McBreen, L.C. Klein, G.G. Amatucci,
ubmitted, J. Electrochem. Soc. (2002).
.B. Bates, N.J. Dudney, B. Neudecker, A. Ueda, C.D. Evans, Solid State /onics 135
2000) 33.
A. D. Pelton, J. Phase Equilibria 12 (1991) 42.
. Zint1, A. Schneider, Z. Elektrochem. 41 (1935) 764.
. Pereira, L. C. Klein, G.G. Amatucci, J. Electrochem. Soc. 149 (2002) A262.
. Pereira, L. Dupont, J.M. Tarascon, L.C. Klein, G.G. Amatucci, submitted, J.
lectrochem. Soc. (2002).
. Takeda, M. Nishijima, M. Yamahata, K. Takeda, N. lmanishi, 0. Yamamoto, Solid

tate lonics 130 (2000) 61.
.A. Huggins, Solid State lonics, in press (2002).
. Pereira, Ph-D thesis, Rutgers University, December 2001.
. Anani, R.A. Huggins, J. Power Sources 38 (1992) 351.
. Anani, R.A. Huggins, J. Power Sources 38 (1992) 363.
S Patent 4,950,566.
.K. Huang, B.V. Ratnakumar, S. Surampudi, G. Halpert, The Electrochemical Society
roceedings, Vol. 94-28, 361.
S Patent 5,294,503.
. Kim, J. Choi, H.J. Sohn, T. Kang, J. Electrochem. Soc. 146 (1999) 4401.
. Moriga, K. Watanabe, D. Tsuji, S. Massaki, I. Nakabayashi, J. Solid State Chern. 153
2000) 386.
G.A. Roberts, E.J. Cairns, J.A. Reimer, J. Power Sources in press (2002).
W. J. Weydanz, M. Wohlfahrt-Mehrens, R.A. Huggins, J. Power Sources 81-82 (1999)
37.
W. G.X. Wand, L. Sun, D.H. Bradhurst, S. Zhong, S.X. Dou, H.K. Liu, J. Alloys and
amp. 306 (2000) 249.
ter 9
ALLOYS AND INTERMETALLIC ANODES
R. A. Huggins
culty of Engineering, University of Kiel, Kaiserstrasse 2, D-24143 Kiel, Germany
INTRODUCTION
Early work on the commercial development of rechargeable lithium
es to operate at or near ambient temperatures involved the use of
ntal lithium as the negative electrode reactant. Binary phases, generally
ing a solid solution of lithium in one of the forms of carbon, are
tly employed on the negative side of lithium cells.
There is considerable interest in finding alternative materials that might
re attractive than the lithium-carbons. Improvements might involve the
to operate safely at higher current densities, less first cycle irreversible
ty loss, better cycling behavior, reduced specific volume, and lower
A number of possible alternative negative electrode materials are

sed in other chapters of this book. The intent of this chapter is not to
t a catalog of possible materials and their properties. Instead, emphasis
placed upon the principles involved in understanding the behavior of
multi-component alloys, starting with their equilibrium or near
rium behavior at both elevated and ambient temperatures. Information
g to equilibrium conditions provides the theoretical limits of the
als and capacities of electrodes in electrochemical systems. The
ties that are actually observed under operating conditions will, of
, deviate from these values due to kinetic factors.
Especially interesting have been recent results on materials that
strate significant deviations from equilibrium behavior, including the
ion of amorphous structures, at low temperatures. The use of materials
ch amorphous silicon is formed in situ during cycling appears to be
larly attractive.
d lntermetallic Anodes 271
ALLOYS AS AN ALTERNATIVE TO PURE

LITHIUM
ttention has been given to the use of lithium alloys as an alternative to

tal lithium for some time. Groups working on batteries with molten
ctrolytes that operate at temperatures of 400-450°C, well above the
point of lithium, were especially interested in this possibility. Two
irections evolved. One involved the use of lithium-aluminum alloys,
eas another was concerned with lithium-silicon alloys.3 -s
hereas this approach can avoid the problems related to lithium
, as well as the others mentioned in the literature, there are always at
o disadvantages related to the use of alloys. Because they reduce the
of the lithium, they necessarily reduce the cell voltage. In addition,
sence of additional species that are not directly involved in the
hemical reaction always brings additional weight and, often, volume.
he maximum theoretical values of the specific energy are always
compared to what might be attained with pure lithium. The energy
is also often reduced. But lithium has a large specific volume, so that
ot always the case.
n practical cases, however, the excess weight and volume due to the
alloys may not be very far from those required with pure lithium
es. It is generally necessary to have a large amount of excess lithium
argeable cells in order to make up for the capacity loss related to the
e or filament growth problem upon cycling.
ithium alloys have been used for a number of years in the high
ture "thermal batteries" that are produced commercially for military
s. These devices are designed to be stored for long times at ambient
tures before use, where their self-discharge kinetic behavior is very
hey must be heated to elevated temperatures when their energy output
ed. An example is the Li alloy/FeS2 battery system that employs a
molten salt electrolyte. In order to operate, the temperature must be
o over the melting point of the electrolyte. This type of cell typically
her Li-Si or Li-Al alloys in the negative electrode.
he first use of lithium alloys as negative electrodes in commercial
s to operate at ambient temperatures was the employment of Wood's
loys in lithium-conducting button type cells by Matsushita in Japan.
ment work on the use of these alloys started in 1983,6 and they
commercially available somewhat later.
Lithium - Carbon Alloys

was also shown in 1983 that lithium can be reversibly inserted into
Chapter 9
7
te at room temperatures when using a polymeric electrolyte. Prior
ments with liquid electrolytes were unsuccessful due to co-intercalation
cies from the organic electrolytes that were used at that time. This
m has been subsequently solved by the use of other electrolytes.
There has been a large amount of work on the development of graphites
lated carbon-containing materials for use as negative electrode materials
ium batteries in recent years. This is due in large part to the successful
pment by Sony of commercial rechargeable batteries containing
ve electrodes based upon materials of this family.
Lithium-carbon materials are, in principle, no different from other
-containing alloys. However, since this topic is treated in more detail
ere, only a few points will be briefly discussed here.
One is that the behavior of these materials is very dependent upon the
of both the nanostructure and the microstructure. Therefore, the
sition and the thermal and mechanical treatment all play important
in determining the resulting thermodynamic and kinetic properties.
als with a more graphitic structure have more negative potentials,
as those with less well organized structures typically operate over much
potential ranges, resulting in a cell voltage that is both lower and more
f-charge dependent.
Another important consideration in the use of carbonaceous materials
gative electrodes in lithium cells is the common observation of a
erable loss of capacity during the first charge-discharge cycle due to
sible lithium absorption into the structure. This has the distinct
antage that it requires that an additional amount of lithium be initially t
in the cell. If this irreversible lithium is supplied from the positive
de, an extra amount of the positive electrode reactant material must be
to the cell during its fabrication. As the positive electrode reactant
als often have relatively low specific capacities, e.g., around 140
, this irreversible capacity in the negative electrode leads to a
ment for an appreciable amount of extra reactant material weight and
e in the total cell.
There are some other matters that should be considered when
ring normal metallic lithium alloys with the lithium-carbons. The
c volume of some of the metallic alloys can be considerably lower than
the carbonaceous materials.
PRINCIPLES DETERMINING THE POTENTIALS

AND CAPACITIES OF ALLOY ELECTRODES
UNDER EQUILIBRIUM CONDITIONS
The theoretical basis for understanding and predicting the potentials
d Intermetallic Anodes 273
pacities of both binary (two element) and ternary (three element)

alloys under equilibrium and near-equilibrium conditions has now
tablished. Under these conditions, these important practical parameters
ctly related to the thermodynamic properties and compositional ranges
ertinent phases in the respective alloy systems.
ne can often understand their behavior under dynamic conditions in
f simple deviations from such equilibrium conditions. In other cases,
r, metastable phases, either crystalline or amorphous, may be present
properties are considerably different from those of the
dynamically stable phases and compositions.
Relations Between Phase Diagrams, the Composition

Dependence of the Potential, and the Capacity of
Electrodes
hase diagrams are figures that graphically represent the equilibrium

a chemical system. They can be very useful thinking tools to help
and the fundamental electrochemical properties of electrodes. There
ous types of phase diagrams, but in the most common case they are
ensional plots that indicate the temperature and compositional
ns for the stability of various phases under equilibrium conditions.
binary phase diagram is a plot of temperature versus the overall
ition for alloys composed of two different elements. It shows the
ture-composition conditions for the stability of the various phases that
m in a given system, and is commonly used in materials science. It will
that there are regions in which only a single phase is stable, and
in which two phases are stable. Although not of particular importance
ochemical experiments, there are also special conditions under which
ssible for three phases to be present in a binary system. According to
bs Phase Rule, this can only happen at a specific temperature. At both
and below that unique temperature only one-phase and two-phase
are possible. This means that at that particular temperature the three
n equilibrium must contact a horizontal (constant temperature) line in
ram. Two will touch it at the ends, and the third will only do so at a
omposition between them.
n the construction of binary phase diagrams the same rules apply
the phases present in the relevant ranges of temperature and
ition are liquids or solids. At any fixed temperature single phase
are always separated by two-phase regions, and in two-phase regions e
in the overall composition is accomplished by variation of the
amounts of the two phases at its limits, each of which maintains a
--
Chapter 9
a+ Liquid
Liquid
a l
--------'-----------------
/ U+ p \
---------------------- '-•
I
I
--
p+y \
I
II
I
p \
Pure A Pure B
9.1. Schematic binary phase diagram with an intermediate phase 13, and solid solubility
inal phases a andy.
composition.
A schematic diagram for a hypothetical binary alloy system A-B is
in Figure 9.1. In this case there are four 1-phase regions. The
hases are designated as phases a, and y. In addition, there is a liquid
at higher temperatures. It can be seen in the figure that the single phases
separated by 2-phase regions as the composition moves horizontally
rmally) across the diagram.
According to the Gibbs phase rule all intensive properties, including the
cal potential, vary continually with the composition within single phase
s in a binary system. Correspondingly, the intensive properties are
sition-independent when two phases are present in a binary system.
the equilibrium electrical potential of such an electrode, E, in an
chemical cell is determined by the lithium activity, it also varies with
sition within single phase regions, and is composition-independent
there are two phases present under the equilibrium conditions that are
ed here.
The variation of the electrical potential difference V (voltage between
ctrode and some reference potential) with the overall composition in
pothetical system at temperature T 1 is shown in Figure 9.2. It is seen
alternately goes through composition regions in which it is constant
tial plateaus) and those in which it varies. If B atoms are added to pure t
A the overall composition is initially in the solid solution phase a and
ctrical potential varies with composition. When the solubility limit is
, the addition of more B causes the nucleation and growth of the

wo phases are then present, and the potential maintains a fixed value.
he overall composition reaches composition at the ex+ /line all of the
will have been consumed and there will only be phase present.
urther compositional change within thephase, which is assumed to
a
<I t><l t>
a+ +1
Pure A PureB
2. Variation of electrical potential with composition across the binary phase diagram
Figure 9.1.
very narrow range of composition in this illustration, the electrical

l again becomes composition-dependent. When the upper
itional limit of the phase is reached, the overall composition again
2-phase (and y) range and the potential is again composition
dent. Upon reaching composition at the +y/y line the potential again
ith composition.
An Example, the Lithium-Antimony System

s a concrete example to demonstrate these principles consider the Li
em, which has been studied both experimentally and theoretically in
8 10
tail. - The phase diagram is shown in Figure 9.3.
elow 615°C there are two intermediate phases between Sb and Li,
nd LhSb. Both have rather narrow ranges of composition and are
ted simply as vertical lines in the phase diagram. Thus, if an electrode
s pure Sb and lithium is added, it successively goes through two
Chapter 9
nt reactions. The first involves the formation of the phase LhSb, and
written as:
2 Li + Sb = LhSb (9.1)
Upon the addition of more lithium a second reaction will occur that
in the formation of the second intermediate phase from the first. This
1600
1400 1320
Liquid
1200
.a
(/)
.P
....1
1000
"'"
795
BOO
630.7
600 89 99
.a .a
(I)
(/)
.!)
....1 .$:1
400 I!
....1
200 Sb-
10 20
00 30 40 50 60 70 80 90 100
Li Atomic Percent Sb Sb
.3. Lithium-antimony phase diagram.
written as:
(9.2)
This process can be studied experimentally by the use of a simple

chemical cell whose initial configuration is similar to that shown
By driving current through this cell from an external source that causes
ltage between the two electrodes to be reduced, lithium will leave the
e electrode, pass through the electrolyte, and arrive at the positive
de. If the chemical diffusion rate within the LixSb electrode is
ntly high relative to the rate at which lithium ions arrive at the
de surface, this lithium will be incorporated into the bulk of the
de crystal structure, changing its composition. That is, the value of x in

itive electrode material LixSb will increase.
f the lithium is either added very slowly, or stepwise, allowing
rium to be attained within the positive electrode material after each
Li +- conducting Li b
Li electrolyte
4. Schematic drawing of the electrochemical cell used to study the Li-Sb system.
e influence of the lithium concentration in the positive electrode upon

ential under equilibrium or near-equilibrium conditions can be
ated. This procedure was called electrochemical titration by Wagner,
st used it to study the properties of the phase "Ag2S" as a function of
11
/S ratio. The result of such an experiment on the Li-Sb system,
10
ed at 360°C, is shown schematically in Figure 9.5.
We can understand these results by consideration of the Gibbs phase
hich was mentioned earlier. It can be written simply as:
F = C-P+2 (9.3)
F is the total number of degrees of freedom, C the number of

ents, and P the number of phases. In this case there are two
ents, Li and Sb. In an experiment conducted at constant temperature
ssure two degrees of freedom are specified. Thus, if two phases are
the number of residual degrees of freedom will be zero, and all
e variables will be fixed, independent of the overall composition. In
se, this means that the electrical potential will be constant over
itional interval where two phases are present in the phase diagram. On
r hand, the potential will vary with the concentration in composition
in which only one phase is present in the phase diagram. After an
nvisibly narrow, range of solid solution, the first plateau in Figure 9.5
onds to compositions in the phase diagram in which both (almost pure)
the phase LhSb are present. Thus it is related to the reaction in
Chapter 9
on (9.1).
There is also a very narrow composition range in which only one phase,
is present and the potential varies. Upon the addition of further Li the
YS.
mv)
1oocP
L.
I I I
1000
I I ·
-Asec/o Sb
I I
2000
I 2!K)()
Li1SII
600 E
vs. AI."LiA(
800 Li 1SD 400 (mv)
600
200
400r- -
r- 0
200r- T • 360-c
1- -200
I I
0
0 2 3
y in Li.ySb
9.5. Results of coulometric titration experiment on Li-Sb system.
l composition moves into the region of the phase diagram in which two
are again present, in this case LhSb and LhSb, and the potential
s along a second plateau related to Equation (9.2).
Relation Between the Potentials and Thermodynamic

Data
The potentials of the two plateaus in Figure 9.5 can be calculated from
dynamic data on the standard Gibbs free energies of formation of the
ases, LhSb and LhSb. According to Ref. 10, these values are - 176.0
and - 260.1 kJ/mol, respectively, at that temperature.
The standard Gibbs free energy change, Gr0 , related to reaction (9.1)
ply the standard Gibbs free energy of formation of the phase LhSb,
hSb). From this we can calculate the potential of the first plateau from:
(9.4)
Eo is the potential of pure Li. This is found to be 912 mV.

he potential of the second plateau is related to virtual reaction (9.2),
(9.5)
his case:
(9.6)
ult is that the potential of this plateau is 871 mV vs. pure Li.
his is the same result that would be found if it were assumed that the
diate phase, LhSb, did not form, and that there is only one voltage
If the electrochemical titration curve were calculated from this value y,
it would have only a single plateau, at a voltage that is the weighted of
the voltages of the two reactions. This is false, due to the lack of tion
of the existence of the intermediate phase. Thus one has to be to be
aware of all of the stable phases when making predictions from
ynamic data.
his discussion has assumed that thermodynamic data are available,
one wants to predict the electrochemical properties. The opposite can
done, of course; experimental measurements can provide information
he relevant phase diagram and the thermodynamic properties of the
phases.
imilar experiments have been made on a number of other binary
systems under conditions such that equilibrium can be reached. This
ment is fulfilled within reasonable times much more easily at elevated
tures. In some cases, however, equilibrium data can also be obtained
ent temperatures, albeit with a bit more patience. Elevated temperature
a number of phases in the Li-Al, Li-Bi, Li-Cd, Li-Ga, Li-In, Li-Pb,
i-Si, and Li-Sn binary lithium alloy systems, made using a LiCl-KCl
salt electrolyte, are listed in Table 9.1.
Temperature Dependence of the Equilibrium Potential

addition to the experiments in the LiCl-KCl molten salt electrolyte,
ments have also been made on some of these systems using lower
0
ture molten salts, LiNOrKN0/ , and also with organic electrolytes at
temperature. This provided information on the temperature
nce of the potentials and capacities.
he results of experiments carried out on materials in the Li-Sb and Li
ems over a wide range of temperature demonstrate the principles
d. Each of these systems has two intermediate phases at low
tures. The data are shown in Figure 9.6.
he temperature dependence of the potentials of the plateaus due to the
e of two-phase equilibria in the Li-Sb system was straightforward,
Chapter 9
g upon two straight lines, corresponding to the reactions of equations

and (9.2) above.
9.1. Plateau potentials and composition ranges of a number of binary Li alloys at

d temperatures.
age vs. Li/Li+, V e

0.047
0.058
0.080
0.089
0.09
0.122
0.145
0.156
0.170
0.237
0.271
0.283
0.283
0.300
0.332
0.373
0.375
0.387
0.430
0.455
0.495
0.507
0.558
0.565
0.570
0.750
0.750
0.760
0.875
0.910
In the Li-Bi case, however, where the comparable reactions are:
Li + Bi = LiBi (9.7)
2 Li + LiBi = LhBi (9.8)
perature dependence of the plateau potentials is different. There is a
3
<
.•<
r
...
600
.6. Temperature dependence of the potentials of the two-phase plateaus in the Li-Sb i
systems.
in slope at the eutectic melting point (243°C), and the data for the two
s converge at about 420°C, which corresponds to the fact that the LiBi
s no longer stable above that temperature. These features can be seen
phase diagram for that system, shown in Figure 9.7. At higher
tures there is only a single reaction:
3 Li + Bi = LhBi (9.9)
n addition, the potentials of the second reaction fall along two straight
ments, depending upon the temperature range. There is a significant
in slope at about 210°C, resulting in a negligible temperature
ence of the potential at low temperatures, due to the melting of
.
s before, all the potentials are related to the standard Gibbs free
changes G,o relating to the relevant reactions.
Chapter 9
12001 ----------------------------------------------,
1000
Liquid
800
415
400 400
..,
:::;
iii
200
o;nrrrrrrrrmmmmmmnnnrrrrrrmmmmmmnnrrrrrrrrmmmmmm rrrrmmmm
0 10 20 30 40 50 60 70 80 90 100
Bi Atomic Percent Li Li
.7. Lithium-bismuth binary phase diagram.
The temperature dependence of the values of Gr0 can be seen from the
g equation:
(9.10)
Mlr 0 is the change in the standard enthalpy and Sr0 is the change
in ndard entropy resulting from the corresponding reaction. Thus it can
be at:
(9.11)
From these data, one can obtain values of the standard molar entropy
es involved in these several reactions. They are shown in Table 9.2.
llows the prediction of the potentials at any temperature within this
.2. Reaction entropies in the lithium-antimony and lithium-bismuth systems.
Reaction Molar Entropy of Reaction Temperature Range

(J/K mol) ("C)
2 Li + Sb = Li2Sb -31.9 25-500
Li + Li2Sb = Li3Sb -46.5 25-600
Li + Bi = LiBi 0 25-200
2 Li + LiBi = Li3Bi - 36.4 25-400
Experiments at Ambient Temperature

xperiments have also been performed to determine the equilibrium
of the electrochemical potentials and capacities in a number of binary
21 22
systems at ambient temperatures. ' Because of slower kinetics at
emperatures, these experiments took longer to perform. Data are
ed in Table 9.3.
LIQUID BINARY ALLOYS

Although the discussion here has involved solid lithium alloys,
considerations apply to those based on sodium or other species. In
, it is not necessary that the active material be solid. The same
es hold for liquids.
. Plateau potentials and composition ranges of lithium alloys at low temperatures

uilibrium conditions.
oltage vs. Li/Li+,v System Range ofy Temperature°C Ref.

005
055
157
219
256
292
352
374
380
42
449
485
530
601
660
680
810
828
948
956
Chapter 9
An example that was of great interest some years ago involved the so
sodium-sulfur battery that operates at about 300°C. In this case, both of
ctrodes are liquids, and the electrolyte is a solid sodium ion conductor,
only called sodium beta alumina. This configuration can be described as
S/L system. It is the inverse of conventional systems with solid
des and a liquid electrolyte, SILlS systems. The negative electrode is
sodium, and the positive electrode is the product of the reaction of
m with liquid sulfur. Thus the basic reaction can be simply written as:
(9.12)
The potential of the sodium electrode is constant, independent of the

t of sodium present. The composition of the positive electrode changes
900
BOO
u 700
•
''
'
\
\
90 ICO
SULPHUR ,
Percont
.8. Part of the sodium-sulfur phase diagram.

sodium concentration varies, however. Thus its potential is

ition-dependent.
he relevant portion of the Na-S phase diagram is shown in Figure 9.8.
en that at about 300°C a relatively small amount of sodium can be
ed in liquid sulfur. When this concentration is exceeded, a second
hase with a composition of about 78 atomic percent Na is nucleated.
s a composition that is roughly Nao.4S. As more sodium is added, the
composition traverses the two-phase region, and the amount of this
phase increases relative to the amount of the sulfur-rich liquid phase.
potential plateau is expected over this composition range. When the
concentration exceeds that corresponding to about Nao.4S the overall
ition moves into a single phase liquid range, and thus the potential
The maximum amount of sodium that can be used in this electrode
onds roughly to Nao.67S. At higher sodium concentrations, a solid
phase begins to form from the liquid solution. This tends to form at
erface between the solid electrolyte and the liquid electrode, and s
the ingress of more sodium, thus blocking the reaction. The
n of the potential with the composition of the electrode is shown in
9.9.
OPEN CIRCUIT VOLTAGE VS STATE OF DISCHARGE
2.0
I I I
Na 2S15 Na2S4 Na 2s3
1.0
DEPTH OF DISCHARGE,%
e 50 H WO%
MOLE RATIO SODIUM I SULFUR
9. Voltage versus sodium as a function of the composition of the positive electrode.

Chapter 9
MIXED-CONDUCTOR MATRIX ELECTRODES

In order to be able to achieve appreciable macroscopic current densities
maintaining low local microscopic charge and particle flux densities,
battery electrodes that are used in conjunction with liquid electrolytes
oduced with porous microstructures containing very fine particles of the
reactant materials. This high reactant surface area porous structure is
ated with the electrolyte.
This porous fine-particle approach has several characteristic
antages. Among these are difficulties in producing uniform and
ucible microstructures, and limited mechanical strength when the
re is highly porous. In addition, they often suffer Ostwald ripening,
ng, or other time-dependent changes in both microstructure and
ties during cyclic operation.
Furthermore, it is often necessary to have an additional material present
er to improve the electronic transport within an electrode. Various
dispersed carbons are often used for this purpose.
23 25
A quite different approach was introduced some years ago - in which
demonstrated that a dense solid electrode can be fabricated which has a
site microstructure in which particles of the reactant phase or phases
ely dispersed within a solid electronically-conducting matrix in which
ctroactive species is also mobile, i.e. within a mixed conductor. There
a large internal reactant/mixed-conducting matrix interfacial area. The
active species is transported through the solid matrix to this interfacial
, where it undergoes the chemical part of the electrode reaction. Since
atrix material is also an electronic conductor, it can also act as the
de's current collector. The electrochemical part of the reaction takes
n the outer surface of the composite electrode.
When such an electrode is discharged by deletion of the electroactive
s, the residual particles of the reactant phase remain as relics in the
tructure. This provides fixed permanent locations for the reaction to
lace during following cycles, when the electroactive species again
the structure. Thus this type of configuration has the additional
age that it can provide a mechanism for the achievement of true
tructural reversibility.
In order for this concept to be applicable, the matrix and the reactant
must be thermodynamically stable in contact with each other. One can te
this possibility if one has information about the relevant phase ms
as well as the titration curves of the component binary systems. The y
window of the matrix phase must span the reaction potential of the
nt material. It has been shown that one can evaluate the possibility that
onditions are met from knowledge of the binary titration curves.
ince there is generally a common component, these two binaries can

treated as a ternary system. Although ternary systems are not
ly discussed here, it can be simply stated that the two materials must
comers of the same constant-potential tie triangle in the relevant
mal ternary phase diagram in order to not interact. The potential of the
ngle determines the electrode reaction potential, of course. An
nal requirement is that the reactant material must have two phases
in the tie triangle, but the matrix phase only one.
he kinetic requirements for a successful application of this concept are
understandable. The primary issue is the rate at which the
ctive species can reach the matrix/reactant interfaces. The critical
ter is the chemical diffusion coefficient of the electroactive species in
rix phase. This can be determined by various techniques, as discussed
re.
he first example that was demonstrated was the use of the phase with
inal composition Li13Sn5 as the matrix, in conjunction with reactant
in the lithium-silicon system at temperatures near 400°C. This is an
lly favorable case, due to the very high chemical diffusion coefficient
m in the Li13Sn5 phase.
he relation between the potential-composition data for these two
under equilibrium conditions is shown in Figure 9.10. It is seen that
se Lh,6Sn (Li13Sn5) is stable over a potential range that includes the
wo-phase reconstitution reaction plateau in the lithium-silicon
0.7
0.5
c .1 ,
0.1
T
yin
10. Composition dependence of the potential in the Li-Sn and Li-Si systems.
Chapter 9
0.39
40mA.tml
D.38
-20 mA/etr/- 037
0.01
- 0.36
D.35
l !) O.JJ
-
>
- l ,.sl plateau IOCVI
::-0.01 0.33-
ci -0.02
rnAicrn2 -
!i!
l-0.03
-0.04
-!!.05 charging •
0.28
-0.06 - I I I I t t
0 10
charge stored as% o1 CCij)CIC:ily -
9.11. Charge and discharge curves of the Li-Si alloy in the matrix of the
hemically inert mixed-conducting Li-Sn alloy at different current densities.
ore, lithium can react with Si to form the phase Lit.71Si (Li12Si7) inside
solid composite electrode containing the Lh_ 6Sn phase, which acts as a
-transporting, but electrochemically inert matrix.
Figure 9.11 shows the relatively small polarization that is observed
the charge and discharge of this electrode, even at relatively high
t densities. It is seen that there is a potential overshoot due to the free
involved in the nucleation of a new second phase if the reaction goes
pletion in each direction. On the other hand, if the composition is not
quite so far, this nucleation-related potential overshoot does not
, as seen in Figure 9.12.
This concept has also been demonstrated at ambient temperature in the
of the Li-Sn-Cd system.26•27 The composition-dependence of the
ials in the two binary systems at ambient temperature is shown in
9.13, and the calculated phase stability diagram for this ternary system
wn in Figure 9.14. It was shown that the phase Li4.4Sn, which has fast
cal diffusion for lithium, is stable at the potentials of two of the Li-Cd
titution reaction plateaus, and therefore can be used as a matrix phase.
Q.l9
0.38
0.37
I• 40 mA/cm'
0.36
0.35 r
r--+-------S-i-U-171Si plateau IOCVI------------ 0.34
0.33
0.32
:g
. o.J'll
Q.30
l
0.29
:;-o-----:;10----:t20::---::J!:'o---:4 o---:s o:----t.eo::- , ;-o--•eo --. """'al'·28

charge stored as % ot capacity
12. Charge and discharge curves of the Li-Si, Li-Sn composite if the capacity is
o that the reaction does not go to completion in either direction. There is no large
n overshoot in this case.
1000 r-------------------------------------
..........
aoo-
sao- \
.. .. .
• U·Sn
• • • • •• • •
400
•• ... •
•••
•
. ........
•
200
• •
••
....
LI·Cd
0
AMA ... A
•••
0 2 3 4 5
.13. Potential versus composition for Li-Sn and Li-Cd systems at ambient
re.
havior of this composite electrode, in which Li reacts with the Cd

nside of the Li-Sn phase, is shown in Figure 9.15.
n order to achieve good reversibility, the composite electrode
Chapter9
Sn
9.14. Calculated phase stability diagram for the Li-Cd-Sn system at ambient
ture. Numbers are voltages vs. Li.
tructure must have the ability to accommodate any volume changes

ight result from the reaction that takes place internally. This can be
care of by clever microstructural design and alloy fabrication
ques.
DECREPITATION
A phenomenon called "decrepitation" can occur in materials that
o significant volume changes upon the insertion of guest species. It is
ometimes called "crumbling". These dimensional changes cause
nical strain in the microstructure, often resulting in the fracture of
es in an electrode into smaller pieces.
This can be a striking, and sometimes disastrous, phenomenon, for it is
ecifically related to fine particles, or even to electrochemical systems.
example, it has been shown that some bulk solid metals can be caused
ture, and can even be converted into powders by repeated exposure to
en gas if they form metal hydrides under the particular thermodynamic
ons present. This is, of course, different from the hydrogen
tlement problem in metals with body-centered cubic crystal structures,
100
5"
> ..
g01 60
IIi
·····························
01
80
>
.§.
Q)
•.•···•··•·· .-..=:Ch;a.rg.e.. EQUifibtium
• •
0 •
01
4
..... ...............D.is.
J: +
(...). 0 ••
2
·0.2 0.0 0.2 0.4 0.6 0.8

Fraction or Electrode Capacity
15. Charge-discharge curve of the Li-Cd system with a fast mixed-conducting phase
ium-tin system at a current density of 0.1 mA/cm2 (ambient temperature).
involves the segregation of hydrogen to dislocations within the

ructure, influencing their mobility.
ecrepitation is often particularly evident during cycling of
hemical systems. It can readily result in the loss of electronic contact
reactive constituents in the microstructure and the current collector.
nsequence, the reversible capacity decreases.
his phenomenon has long been recognized in some electrochemical
in which metal hydrides are employed as negative electrode reactants.
phenomena also occur in lithium systems employing alloy electrodes,
f which undergo very large changes in specific volume if the
ition is varied over a wide range in order to achieve a large capacity.
ecause of its potentially large capacity, a considerable amount of n
has been given recently to the Li-Sn system, which is a fine example
phenomenon. The phase diagram of the Li-Sn system shows that there
intermediate phases. The thermodynamic and kinetic properties of the
t phases in this system were investigated some time ago at elevated
17 28 21
tures and also at ambient temperatures. ' ' ' The volume
22 26 27
'
that occur in connection with phase changes in this alloy system are
he phase that forms at the highest lithium concentration, Li 4.4Sn, has a
volume that is 283% of that of pure tin. Thus Li-Sn electrodes swell
nk, or "breathe", a lot as lithium is added or deleted.
bservations on metal hydrides that undergo larger volume changes
own that this process does not continue indefinitely. Instead, it is
hat there is a terminal particle size that is characteristic of a particular
l. Particles with smaller sizes do not fracture further.

Chapter 9
Experiments on lithium alloy electrodes have also shown that the

ochemical cycling behavior is significantly improved if the initial
29
e size is already very small, and it is reasonable to conclude that this
ted to the terminal particle size phenomenon.
A theoretical paper on the mechanism and the influence of the
tant parameters related to decrepitation appeared recently ? It uses a
0
e one-dimensional model which allows the calculation of the conditions

which fracture will be caused to occur in a two-phase structure due to
ic volume mismatch. This model predicts that there will be a terminal
e size below which further fracture will not occur. The value of this
teristic dimension is material-specific, depending upon two parameters,
agnitude of a strain parameter related to the volume mismatch and the
re toughness of the lower-specific-volume phase. For the same value of
e mismatch, the tendency to fracture will be reduced and the terminal e
size will be larger the greater the toughness of the material. The of
this model calculation are shown in Figure 9.16.
50
45
40
35 -·-5
. a!
s 30
2 25
--o--10
-•-15
20 ----<>-- 20
.tj
._ 15
!'12
·c
u
10
0
0 2 3
Dilation strain parameter
9.16. Variation of the critical particle size as a function of the dilation strain and the
toughness of the phase in tension.
FORMATION OF AMORPHOUS PRODUCTS AT

AMBIENT TEMPERATURE
his chapter has been primarily concerned with understanding the

r of alloys under equilibrium or near-equilibrium conditions, from
he potential and capacity limits can be determined. Actual behavior in
lications always deviates from these limiting values, of course.
here have been a number of observations that the operation of oxide
electrodes at very high lithium activities can result in the formation of
ous, rather than crystalline, products, whose properties are different
ose of the corresponding crystalline materials.
he spectrum of materials in which amorphous phases have been
under electrochemical conditions is now quite broad, and include ls
of potential interest as positive electrode reactants, such as some m-
based materials with the general formula RV04, where R is Al, Cr,
31
r Y.
A group of interesting nitride alloys with structures related to that of
hich is known to be a fast ionic conductor for lithium, but in which
f the lithium is replaced by a transition metal, such as Co, have been
o become amorphous upon the first insertion of lithium.32 -35
xperimental evidence for the electrochemical amorphization of alloys
36
i-Si, Li-Sn or Li-Ag systems was studied by Limthongkul. In the
o cases, this was only a transient phenomenon.
specially interesting, however, have been experiments that gave
e for the formation of amorphous silicon during the initiallithiation of
er of silicon-containing precursors, including SiB3, SiO, CaSh and
39
- The electrochemical behavior of these materials after the initial
n cycle was essentially the same as that found in Si powder that was
amorphous. There was, however, an appreciable amount of
ible capacity in the first cycles of these precursors, about 1 mol of Li
case of SiB3 and the disilicides, which was evidently due to an
ible displacement reaction with Li to form one mol of amorphous
In the case of SiO the irreversible capacity amounted to about two
Li, which was surely related to the irreversible formation of LhO as
the amorphous silicon. Some of these materials with amorphous Si are
. Reversible capacity of several materials that contain amorphous silicon.
Material Reversible Capacity,

mAh/g
SiB3 440
SiO 675
Amorphous Si >1000
Chapter 9
nsiderable potential interest as negative electrode reactants in lithium

s, as their charge/discharge curves are in an attractive potential range,
ave reasonable kinetics, and their reversible capacities are very high, as
in Table 9.4.
SUMMARY
Because of safety and cycle life problems with the use of elemental
m, essentially all rechargeable lithium batteries use lithium alloys as
ve electrode reactants today. The theoretical limits of the potentials and
ies of such electrodes can be determined from a combination of
odynamic data and phase diagrams. This has been demonstrated for
l examples of binary systems.
There are two general types of reactions that can take place: homophase
ns, in which guest atoms are inserted into an existing phase, often
ctically, and reconstitution reactions in which phases nucleate and grow
terophase microstructures. The potential varies with the overall
sition of an electrode in the insertion reaction homophase case, but is
sition-independent when reconstitution reactions take place in
phase microstructures.
The electrochemical titration method can be used to investigate the
nt parameters experimentally. Experimental results are now available
number of systems, both at elevated temperatures and at ambient
ratures.
Under equilibrium, or near-equilibrium, conditions the potentials are
y related to the standard Gibbs free energies of formation of the phases
ed. Thus thermodynamic data can predict experimental results.
ise, experiments can provide thermodynamic data. As an example, the
rature dependence of potential plateaus can be used to determine the
rd entropy changes in the relevant reaction. These experimental data
orrelate with the stability of phases in the phase diagram. Furthermore,
aximum theoretical specific energy of an electrochemical system can
e determined from the equilibrium electrochemical titration curve and
ated thermodynamic data.
These principles are applicable to liquid, as well as solid, binary
als. An example is the Na-S system, which is the basis for a high
rature LIS/L battery system that uses sodium beta alumina as a solid
lyte.
Composite multiphase microstructures can be useful in some cases, if
ermodynamic and kinetic properties of one binary system are
tible with those of another binary system. The critical question is the
ial range in which the desired reaction takes place. It must be within
ility range of the other phase with fast kinetics. Experiments have
trated that this mixed-conductor matrix concept can be used at both d
temperatures and ambient temperatures.
ne of the practical problems with a number of alloy systems in which
lues of capacity are achieved by the use of large composition changes
d to the corresponding volume changes. This can lead to decrepitation
ctronic disconnection of reactant particles. The general features of this
r, and the importance of the particle size and the mechanical
es of the respective phases are now understood in principle.
n interesting recent development relating to alloys has been the
nce of experimental results showing that electrochemical lithiation
se the appearance of amorphous phases with quite different properties
ose of the equilibrium phases. Especially interesting, and potentially
nt, is the observation that displacement reactions can take place during
n of some materials containing silicon that result in the formation of
ous silicon as a product. This amorphous silicon can then react
ly with lithium at attractive potentials and with a high capacity.
RENCES
ao, L.A. Heredy and R.C. Saunders, J. Electrochem. Soc. 118 (1971) 1039 .
ay, D.R. Vissers, F.J. Martino and K.E. Anderson, J. Electrochem. Soc. 123 (1976)
ai, J. Electrochem. Soc. 123 (1976) 1196.

harma and R.N. Seefurth, J. Electrochem Soc. 123 (1976) 1763.
eefurth and R.A. Sharma, J. Electrochem. Soc. 124 (1977) 1207.
wa, 2"d!MLB, (Elsevier Sequoia) (1984), p. 259.
ami and P. Touzain, J. Power Sources 9 (1983) 365.
ppner and R.A. Huggins, Symposium on Electrode Materials and Processes for
y Conversion and Storage , J.D.E. Mcintyre, S. Srinivasan and F.G. Will, Eds.,
ochem. Soc. (1977), p. 833.
ppner and R.A. Huggins, Z. Phys. Chern. N.F. 108 (1977) 105.
eppner and R.A. Huggins, J. Electrochem. Soc. 125(1978) 7.
agner, J. Chern. Phys. 21 (1953) 1819.
en and R.A. Huggins, J. Solid State Chern. 37 (1981) 271.
en, Ph.D. Dissertation, Stanford University (1980).
Wen and R.A. Huggins, Mat. Res. Bull. 15 (1980) 1225.
Saboungi, J.J. Marr, K. Anderson and D.R. Vissers, J. Electrochem. Soc. 126 (1979)
Wen and R.A. Huggins, J. Electrochem. Soc. 128 (1981) 1636.

Wen and R.A. Huggins, J. Electrochem. Soc. 128 (1981) 1181.
Wen, B.A. Boukamp and R.A. Huggins, J. Electrochem. Soc. 126 (1979) 2258.
eppner and R.A. Huggins, J. Electrochem. Soc. 125 (1978) 7.
oench and R.A. Huggins, J. Electrochem. Soc. 129 (1982) 341C.
ng, P. King and R.A. Huggins, Solid State Ionics 20 (1986) 185.
ng, I.D. Raistrick and R.A. Huggins, J. Electrochem. Soc. 133 (1986) 457.
Boukamp, G.C. Lesh and R.A. Huggins, J. Electrochem. Soc. 128 (1981) 725.
Boukamp, G.C. Lesh and R.A. Huggins, in Proc. Lithium Batteries,
Chapter9
. Venkatasetty, Ed., Electrochem. Soc. (1981), p. 467.

. Huggins and B.A. Boukamp, US Patent 4,436,796.
Anani, S. Crouch-Baker and R.A. Huggins, in Lithium Batteries, A. N. Dey, Ed.,
trochem. Soc. (1987), p. 382.
Anani, S. Crouch-Baker and R.A. Huggins, J. Electrochem. Soc. 135 (1988) 2103.
. Wen and R. A. Huggins, J. Solid State Chern. 35 (1980) 376.
ang, M. Winter and J. 0. Besenhard, Solid State Ionics 90 (1996) 281.
A. Huggins and W. D. Nix, Ionics 6 (2000) 57.
iffard, F. Leroux, D. Guyomard, J.-L. Mansot and M. Toumoux, J. Power Sources 68
97) 698.
Nishijima, T. Kagohashi, N. Imanishi, Y. Takeda, 0. Yamamoto and S. Kondo, Solid e
Ionics 83 (1996) 107.
hodai, S. Okada, S-i. Tobishima, and J-i. Yamaki, Solid State lonics 86-88 (1996) 785.
Nishijima, T. Kagohashi, Y. Takeda, N. Imanishi and 0. Yamamoto, B'h IMLB (1996),
02.
hodai, S. Okada, S. Tobishima and J. Yamaki, B'h IMLB (1996), p. 404.
imthongkul, Ph.D. Thesis, Mass. Inst. of Tech. (2002).
lausnitzer, Ph.D. Thesis, University ofUlm (2000).
etz, Ph.D. Thesis, University of Kiel (2001).
etz, R.A. Huggins and W. Weppner, Il'hiMLB (2002), Abstract No. 47.
apter 10
CURRENT ISSUES OF METALLIC LITHIUM

ANODE
M. Ishikawa and M. Morita
Department of Applied Chemistry and Chemical Engineering,

Faculty of Engineering, Yamaguchi University, Yamaguchi, Japan
1. INTRINSIC ADVANTAGES AND DRAWBACKS

OF METALLIC LITHIUM ANODE
Metallic lithium has the highest theoretical specific capacity (3860

/g) and the most negative redox potential among all metals. These
ures have attracted the interest of battery investigators, who have
mpted to put lithium metal into practical use in rechargeable battery
ems. Its poor charge-discharge cycleability, however, and its potential
hazards have hindered thus far the development of cells with a metallic
1
um anode.
The essential reaction of metallic lithium anode is very simple:
Discharge
Li < > Lt + e-,
Charge
in spite of this simplicity, the practical application of Li metal to a

argeable anode has been very difficult due to some crucial issues. The t
important one is that Li metal usually tends to deposit as a dendritic or
sy structure during charge, and the disordered metallic deposit gives rise
poor coulombic efficiency. This happens because such a fine Li metal
n acts as an active site inducing reductive decomposition of electrolyte
ponents. Part of the deposit may become electrically isolated and
ding may also occur. Furthermore, the fine metallic Li can easily
trate into the separator and eventually cause internal short, this resulting
at generation and contingent ignition. To turn batteries with a metallic
node into actual utilization more efforts should be made to overcome
e problems.
Chapter 10
2. HISTORICAL ASPECTS IN RECHARGEABLE

METALLIC LITHIUM BATTERIES
Before the successful commercialization of Li-ion batteries, i.e.,

nd 1990, in prototype and commercialized secondary Li metal-based
ries, major strategies to compensate for poor Li cycling efficiency could
ategorized into two ways. They were: (i) compensation for the low
ge-discharge reversibility of the Li anode by loading excess amount of
etal, and: (ii) suppression of irregular (mossy or dendritic) Li deposition
sing aLi-alloy anode as listed in Table 10.1.
Although these schemes could somewhat make up for the poor
eability of Li metal, their effects were still unsatisfactory. The use of
ss amount of Li (scheme i) resulted in an increase of the total weight of
ries. Furthermore, even with excess metallic Li, cycling life was still
t: 200-300 cycles.
On the other hand, Li-alloy anodes (scheme ii) were applied mainly to
-type cells for memory back-up. Cells with Li-alloy anodes could stand
oximately 1000 cycles because memory back-up applications do not
and deep charge-discharge cycles.2' 3 However, a secondary metal
ed with Li increases the total anode weight, and induces a positive shift
he potential; for instance, Li-Al alloy anodes typically have a redox
ntial of 0.3-0.5 V vs. Li/Li+. The energy density of cells containing a
loy anode was usually much lower even than that of cells with an excess
unt of Li metal. In addition, such Li-alloy anodes cannot be applied to
drical cells, which need wound electrodes, because the Li-alloy anodes
oo brittle and hard to be wound.
Besides AI, some metals can form with Li intermetallic compounds
g as Li-alloy anodes, which can be easily handled and control
morphology of the Li deposit.4 Unfortunately, however, such anodes
crucial problems; the change in electrode volume during
ge-discharge is remarkably large, and the electrodes gradually become
erized, limiting the cycle life. "Wood's metal alloy" is a eutectic alloy
isting of Bi, Pb, Cd and Sn, and can form an intermetallic compound
Li.5 However, the pulverization of Wood's metal is still inevitable like
Li-alloy anodes.
To overcome this problem we might assemble a cell using "originally"
erized alloy powders instead of an intact alloy sheet as an anode.
rally, however, it is very difficult to optimize their binding and
rsion conditions, and there is severe trade-off between their long cycle
nd high energy density.
nt Issues of MetaUic Uthium Anodes
299
Rechargeable lithium batteries in commercialization or test production before 1990.

Baltcr)l system
Anode-Calhode
"s metal alloy·

c:alhode
loy-liS,
r grapllite-
ous V:Os
oy-Polyaniline
oy-LIOII/ MnO,
loy-Vi),
,composite oxide
e,
,
,
Lithium metal rechargeable batteries developed after 1990.
Applicalion
iah·temperBtun:
3
battety)
eralUtC operation)'
(polymer gel battety)'
(polymer gel battety6)
(polymer gel battery)'
IEPI (Central Research Institute ofEieclric Power Industry, Japan)j 'NEooiDENSO; 3USABC/Hydro-Quebec/3M; 4 USABC/ANU
T-America; 'Malsushita 8altcry Industrial; 'Lithium Technology; PolyPius
Chapter 10
-up coin-type cells, research activity for such alloy electrodes is now
g diminished. In Europe, nonetheless, Besenhard and co-workers
inue basic studies on alloy electrodes,6 and in the USA mainly USABC
veloping rechargeable batteries for high-temperature applications with
Al anode for vehicles.7
. RECENT TRENDS IN THE DEVELOPMENT OF

BATTERIES WITH ALi ANODE
After the start of commercial production of Li-ion batteries, the goal

eveloping rechargeable batteries with a metallic Li anode became more
nite than that before the '90s: the performance of Li metal batteries must
ehow exceed that of batteries with a carbon anode in practical use.
e 10.2 shows batteries under development with a Li metal anode. In n
there have been projects utilizing a Li metal anode conducted by
DO" (New Energy and Industrial Technology Development
nization) aiming at power sources both for transportation and stationary
ications.8
One of the fields where a Li metal anode is attracting much interest is
of polymer battery systems. These systems are considered promising
only for portable electronics but also for electric vehicles. As just
ribed, the practical use of Li metal anode is still considered to be worth
ffort. On the other hand, with respect to investigations on the basic
istry of Li metal electrode behavior, several important factors affecting
reversibility of Li cycling have become fairly clear based on
rochemical tests (ac impedance spectroscopy and scanning vibrating
rode technique [SVET]), chemical tests (typically ESCA [XPS] and IR),
microscopic tests (TEM, SEM and AFM, analyses of Li anode surface).
he basis of such characterizations, novel electrolyte systems as well as
rolyte additives compatible with a Li anode have been proposed.
ils will be given in the next section.
. OPTIMIZATION OF THE LITHIUM ANODE

INTERFACE
.1. Fundamentals
An essential strategy to obtain a satisfactory performance from a
llic Li anode is to optimize chemically as well as physically the
facial structure between Li anode and electrolyte. Without a design
d on this strategy, it is practically impossible to have a practical Li
Current Issues of Metallic Lithium Anodes 301
de showing high coulombic efficiency and adequate safety.
Since the reducing power of Li metal is extremely high, i.e., its
modynamic redox potential is very low, reductive decomposition of an
trolytic solvent and/or salt occurs at the pristine Li interface as soon as
e 10.3. Chemical composition of passivation films on Li metal.
Electrolyte
Without electrolyte (native film)
General PC systems
PC/LiPF6
PC/LiCl04
General EC systems
General DMC systems
General DEC systems
S021LiAlC14
General THF systems
THF/LiAsF6
General 2MeTHF systems
2MeTHF/LiAsF6
General DME systems
DME/LiAsF6
General BL systems
General DOL systems

Chapter 10
get in contact. Some components of the decomposition products form

so-called solid electrolyte interface (SEI). If the latter can protect
nst further electrolyte decomposition, as does a passivation layer in
rrosion process of some metal, the SEI can prevent perpetual side
tions of an electrolyte with Li, and hence even the thermodynamically
able Li metal can get pseudo-stability in an organic electrolyte.
With regard to a rechargeable Li anode interface, however, it should
ot only protective against side reactions, but should also not impede the
sion of Li ions across the interface during cycling. As the components
n interface layer depend obviously on the nature of electrolytes (see
e 10.3), appropriate electrolytes have to be selected to favor Li cycling.
surface layer is known to crucially affect the morphology of Li metal
the charge/discharge efficiency. It is recognized that a grained or lumpy
ace structure can lead to a high efficiency of Li cycling, while a
ritic or mossy Li structure is undesirable.
For instance, LiPF6, an electrolytic salt with perfluoro anion, leads to
ormation of a LiF layer at a metallic Li interface. 9- 11 This layer has a
protective ability, and if its thickness is appropriate, the morphology of
sited Li becomes grainy, resulting in an improved cycling efficiency.
10.4. Additives used to enhance the cycling efficiency of Li metal.
oup
HF
ganic AII3, Mgl2,

s 2-
X
2Me-furan,
derivatives
Cetyltrimethylammonium
ganic
chloride
Nonionic
Crown ethers
Benzene
C02
as
N20,CO
.2. Effect of Additives
Another approach to the improvement of Li cycling efficiency is

uding some additives in the electrolyte formulation. This method can
rol or modify the thickness, morphology and chemical composition of
Li surface layer. The main additives so far investigated for
argeable Li battery electrolytes are listed in Table 10.4.
These organic, inorganic and gaseous additives can improve the
ing efficiency of a Li anode. They can be categorized roughly into two
ps with different mechanisms: (i) formation of a protective surface
r by reaction of Li with an additive, (ii) formation of a protective
i-layer by adsorption of a non-reactive additive.
Not only major components but also minor impurities in the
trolytes can have a serious influence on the surface composition via
reactions with Li, and hence a change in cycleability can occur. In
tion, trace impurities in the controlled atmosphere used to form a Li
t would also affect the Li surface. Usually, a Li metal sheet is made in
tmosphere containing C02 to provide a LhC03 layer as a "native"
face at a Li sheet surface, while oxygen and moisture must be
oughly excluded from the atmosphere. The LhC03 interface is widely
15
wn to be favorable for Li cycling. 12 -
Addition of a small amount (5% or less) of another metal to bulk Li
reported as an alternative method for improving Li surface properties.
scheme is different from the utilization of "conventional" Li-alloys
aining relatively large amounts of another metal. Few percents of AI, Sn,
etc. can decrease the interfacial resistance between anode and
trolyte. These anodes may modify the morphology and interfacial
edance of deposited Li while keeping their redox potentials as low as
16
of pure Li.
. PHYSICAL FACTORS AFFECTING THE LITHIUM

INTERFACE
Besides chemical factors such as electrolytes and additives, some

ical factors have an influence on the Li interface structure. Temperature
stacking pressure on the anode have long been known as such factors.
as reported that cycling a metallic Li anode under a certain pressure can
17
ove Li morphology and hence can provide a good cycling efficiency.
respect to the temperature effect, a uniform surface morphology and
igher cycling efficiency can be obtained cycling Li in some
temperature electrolytes (e.g., LiPF6-based binary carbonate-type
rolytes: LiPF6-PC/DMC, LiPF6-EC/DMC) in comparison with that in
Chapter 10
corresponding electrolytes at ambient-temperature. 18" 19 Furthermore, a

efficiency can be obtained at ambient-temperature after
temperature "precycling" in such electrolytes. This result suggests
there is a "hysteresis effect" of electrolyte temperature upon Li anode
ormance. 18' 19
6. RECENT ADVANCES IN THE RESEARCH ON

METALLIC LITHIUM ANODE
Besides the above instances, many efforts have been focused on

ble electrolytes for the improvement of metallic Li anode. Fringant et
tudied multi-solvent electrolytes containing a mixture of PC and EC,
-permittivity solvents, with DMC, a low-viscosity co-solvent which
20
ributes to a good cycleability of metallic Li anode. As an electrolytic
LiAsF6 has been widely known to provide high Li cycling efficiency,
ugh it has a serious drawback, high toxicity.
Aurbach et al. have investigated the relationship among electrolyte
position, corresponding Li surface chemistry, ionic conductivity at the
ace layer and morphology of Li deposit. 19• 2123 - They suggested that
LiFlayer(f((lm LiOHand Li,CO,)
e 10.1. Chemical structure of passivation films on Li metal anode, a) under Ar

phere, b) immersion in 1.0 M LiBF4 I PC or y-BL for 3 days, c) immersion in 1.0
PF6 I PC or y-BL for 3 days.
lcarbonate-type electrolytes, such as PC, first give lithium
lcarbonate, which then react with some impurities to form LhC03 and
. It is notable that a reaction leading to a surface layer tends to be
inated by a solvent component in mixed-solvent electrolytes.
Kanamura et al. have investigated the relationship between inorganic
ponents at a Li surface and the composition of electrolytes by XPS
10 24
ysis, • and proposed chemical models with a multi-layer structure
isting of inorganic components as shown in Figure 10.1. Furthermore,
d on such information, they reported the treatment of Li metal surface
25 26
HF to suppress Li dendrite formation. •
Another topic of research on salts suitable for a Li anode is the
zation of new salts with an organic anion as alternatives to conventional
ganic salts. Among them, lithium bis(perfluoroalkylsulfonyl)imides
especially proved interesting. They have been tested in various cell
ems, where aLi imide has been used not only as a single electrolytic salt
also in mixture with another salt, e.g., LiN(S02C2F5h with LiPF6•
y useful imides have two perfluoroalkylsulfonyl groups linked to a
ral nitrogen atom asymmetrically [e.g., LiN(S02C2F5)(S02C3F7)], as
as symmetrically [e.g., LiN(S02C2F5)2], with the exception of lithium
rifluoromethylsulfonyl)imide [LiN(S02CF3)z], which is known to cause
vere corrosion of the aluminum substrate at the cathode.
Naoi et al. reported that electrolytes containing some
um
100
•••·•--·------·g-··· ----·-····--
..._
0
>-.
80 -.....-...
_ ,. . /-Q.-c:;
/ -... - =..-=:c :::;.-- ---..... --
.... - ....
c
d)
0 60
i:E
d)
. 40
E
0
:::I
0 20
u
0
0 5 10 15 20
Cycle number
e 10. 2. Addition effects of 2MeF, 2MeTp, and benzene on coulombic efficiency of Li

with LiCI04 (1 moVdm3); current density: 1.0 mA/cm2 ; charged: 0.1 C/cm2 ; a:
: PC + 2MeF (0.5 vol %), c: PC + 2MeTp (0.5 vol %), d: PC +benzene.
Chapter 10
erfluoroalkylsulfonyl)imides provide a LiF layer which is thinner than

derived from electrolytes containing LiPF6; this thinner layer can
ct
Li surface and promote efficient Li cycling.27 They emphasized that
formation mechanism of the LiF layer in lithium imide-based
rolytes should be completely different from that of LiF layer in
6-based electrolytes. In the latter, LiF is formed thanks to the small
unt of HF contained in the LiPF6 systems, while in the imide systems
comes from the direct reaction of the imide anion with Li metal.
To modify a metallic Li surface, as already stated, some additives to
rolytes have been investigated. Figure 10.2 shows the effect of
ral representative additives, 2-methylfuran (2MeF), 2-methylthiophene
eTp), and benzene, on the coulombic efficiency of Li cycling in
04-PC.
The addition of 2MeF and 2MeTp can restrain a decrease in Li
28
ng efficiency. On the other hand, a positive effect of benzene
ion appears at a higher concentration when compared to the effective
entration of 2MeF and 2MeTp. According to interface analyses such
c impedance measurement, the origin of the positive effect of these
28
ives can be explained with the help of Figure 10.3.
A)
@
@@
Reacti @ --..
High resistance film
Limetal
Good conductive film

Limetal Li metal
e 10.3. Working mechanisms of organic additives, A) without additive, B)

ve additive such as 2MeF or 2MeTp, C) adsorption-type additive such as
ne.
2MeF and 2MeTp react with Li metal to form an ion-conductive
ective layer which provides uniform Li deposition at the Li surface.
the other hand, although benzene does not react with Li metal, it is
mulated, or adsorbed, at the Li interface due to its low polarity, and the
lting quasi-layer acts as a barrier separating electrolyte components
Li metal.
Figure 10.4 shows the effect of adding a small amount of metal ion
2 29
+ and Sn +). A positive effect of these additives on the Li cycling
iency can be clearly observed. Here, it should be significant to know
difference in ionic current distribution on the Li metal anode
100 e
d ...... .,.- --...1'!1--" .. _....

......... --..... -·, "'""- -c--
90 . ,'
80
. ,'
---..--
70
60
50
0 5 10 15 20
Cycle number
e 10.4. Addition effects of Lil, Snl2, AII3 , and AII3 + 2MeF on coulombic
2
ency of Li in PC with LiCI0 4 (1 moVI); current density: 2.0 mA/cm ; charged
2
ricity: 0.2 C/cm2 ; a: PC, b: PC+ Lil (r 300 ppm), c: PC+ Snl
2 (Sn + 100 ppm),
d: AII3 (ArJ+ 100 ppm), e: PC +AII3 (Al3+ 100 ppm)+ 2MeF (0.5 vol %).
without additives; to observe this, the scanning vibrating electrode

29
nique (SVET) was applied. ' 30 The outline of the in situ SVET system
rawn in Figure 10.5; this system can offer potential gradient profiles
ced by the Li ionic currents on the surface of the Li-deposited electrode
ng discharge. The scanning vibrating (SV) electrode (a Pt wire
trode) attached to a sensor probe is vibrated (with a certain frequency
een 100-1000 Hz) along a vertical axis (Z-direction) in the vicinity of a
specimen electrode surface by a piezoelectric oscillator; typical distance
Chapter 10
een the SV electrode and the test electrode is 2-50 ).!m.

One can obtain a potential gradient value which is the difference
een the potential detected by the SV electrode at the highest position
that at the lowest position during the vertical vibration of the
lectrode.
SCSI Board
r--· '--------,
........... lll!llfl!lllll-.. - lIl
Actuator 1I
eootro er II
II
Galvanoatat
Pot.ntlosblt 1-----*1
--Signal Lile
NlSubstr11t. --- RS-2320
···-GPIB
e 10.5. Scheme of SVET instrumentation.

trode. The distance between the highest and the lowest positions was
he sub-micron order: say 0.5 f..lm. In this case, therefore, the potential
ient values correspond to the potential gaps along a distance of 0.5 f.!m
he vertical axis on the test electrode surface. This potential
ient is proportional to the ionic current in the vicinity of the
trolyte-test electrode interface because the resistance of the
trolytes can be
rded as a constant. The X-Y stage, on which the SV electrode is
nted, is moved horizontally by a stepping motor. By gathering the
ntial gradient values at various X-Y positions, a two-dimensional
ntial gradient map on the test electrode can be obtained. After all, on
basis of the above measurement criterion, a potential gradient map must
elate directly with ionic current distribution on a test electrode.
Figure 10.6 indicates the potential gradient maps of Li metal anode
ace with anodic polarization with and without additives. When the
als (Al and Sn) deposit at Li metal surface with cathodic polarization,
change into a thin Li-alloy surface layer, which can ease the
ularity of Li anode interface. Thus the distribution of Li ionic current
mes relatively regular at the surface with anodic polarization as shown
igure 10.6. Ishikawa et al. revealed the compatibility of metal ion
tives with electrolytes; for example, an additive such as All3 can fairly
nce the cycleability of the Li metal anode especially in lithium
e-based electrolytes.31'32
On the other hand, as for gaseous additives, Osaka et al. analyzed the
ct of saturating electrolytes with C02 upon Li metal cycle performance
he basis of ac impedance measurement, and found this addition effective
14 15
he enhancement of Li anode cycleability. ' Furthermore, some
33 34
actants were reported to have such an enhancement effect. •
Again, representative mechanisms of the aforementioned additives
be briefly summarized as follows: (a) formation of organic polymer
ive layer with a good ionic conductivity as a protective interface derived
28
organic compounds such as 2-methylfuran (2MeF) or its analogues,
adsorption of inactive additives at a Li interface to inhibit a reaction
ing a low-conductive passivation films resulting in uneven Li
osition, i.e., nonpolar additives such as benzene and dekalin28 as well as
ic or nonionic) polymer or oligomer additives such as surfactants,33• 34 (c)
lloy layer formation by co-deposition of Li with another metal such as
rom an electrolyte containing a small amount of metal salt additive, e.g.,
, to control the morphology of Li deposit (in other words, to suppress
dritic deposition of Li),2932- (d) formation of an uniform inorganic
ective layer from a dissolved gas such as C0212- 15 or inorganic acid, e.g.,
s.z6
These working mechanisms appear to have independent origins, so
Chapter 10
we are likely to observe "synergistic effects" when we use some

ives with different mechanisms simultaneously. In fact,
imultaneous use of All 3 and 2MeF results in a better Li cycleability
n compared to the effect of each additive (Figure 10.4).29 In addition,
e of
-1
"D
JS
t3 0.5
J
01)
All3: 200 ppm
....
c 2$
e
t o.s
c:l ! l ..
00
Sni2: 100 ppm
.....
c
m
J Q.S
J 35
10.6. Potential gradient maps of Li-deposited Ni substrate (X-Y plate) in

2Me-THF!LiC104 during discharge (25 mV vs. LilLi+); 2 mA/cm 2 and 0.2 C/cm2
-deposition.
above additives have been known to provide "sustained" effect, even

transfer of aLi anode from an electrolyte with an additive to another
trolyte without additives, the Li anode "treated" with the additive, in
e cases, still retains the effect of the additive.35
This means that "pretreatment" of a Li anode with some additives
enhance the cycling efficiency in the subsequent cycles in a "clean"
trolyte without additives?5 This treatment sequence may be promising
use we could avoid some undesirable effects of such additives upon
r battery components: for instance, internal self-discharge originated
the shuttle effect of redox active additives between anode and cathode,
n unforeseen negative effect of the additives on cathode performance.
FERENCES
T.B. Reddy, S. Hossain, "Rechargeable Lithium Batteries (Ambient Temperature)" in D.
Linden, T.B. Reddy Eds., Handbook of Batteries, 3'd Edition, McGraw-Hill, New York,
2002, Chapter 34, pp. 34.1-34.62.
J.O. Besenhard, J. Electrochem. Soc. 94 (1978) 77.
R.A. Huggins, "Lithium Alloy Anodes" in J.O. Besenhard Ed., Handbook of Battery
Materials, Wiley-VCH, Weinheim, 1999, pp. 359-381.
Y. Toyoguchi, J. Yamaura, T. Matsui, T. Iijima, Proc. Electrochem. Soc. 88-6 (1986)
659.
J.O. Besenhard, J. Yang, M. Winter, J. Power Sources 68 (1997) 87.
GL. Henriksen, A.N. Jansen, "Lithium/Iron Sulfide Batteries" in D. Linden, T.B. Reddy
Eds., Handbook of Batteries, 3'd Edition, McGraw-Hill, New York, 2002, Chapter 41, pp.
41.1-41.23.
P. C. Symons, P. C. Butler, "Advanced Batteries for Electric Vehicles and Emerging
Applications-Introduction" in D. Linden, T.B. Reddy Eds., Handbook of Batteries, 3rd
Edition, McGraw-Hill, New York, 2002, Chapter 37, pp. 37.1-37.25.
K. Kanamura, H. Tamura, Z. Takehara, J. Electroanal. Chern. 333 (1992) 127.
K. Kanamura, H. Tamura, S. Shiraishi, Z. Takehara, Electrochim. Acta 40 (1995) 913.
K. Kanamura, S. Shiraishi, Z. Takehara, J. Electrochem. Soc. 141 (1994) 2379.
Y. Malik, D. Aurbach, P. Dan, A. Meitav, J. Electroanal. Chern. 282 (1990) 73.
D. Aurbach, 0. Chusid, J. Electrochem. Soc. 140 (1993) L155.
T. Osaka, T. Momma, T. Tajima, Y. Matsumoto, Denki Kagaku 62 (1994) 451.
T. Osaka, T. Momma, T. Tajima, Y. Matsumoto, J. Electrochem. Soc. 142 (1995) 1057.
M. Ishikawa, K. Otani, M. Morita, Y. Matsuda, Electrochim. Acta 41 (1996) 1253.
T. Hirai, I. Yoshimitsu, J. Yamaki, J. Electrochem. Soc. 141 (1994) 611.
M. Ishikawa, Y. Takaki, M. Morita, Y. Matsuda, J. Electrochem. Soc. 144 (1997) L90.
M. Ishikawa, M. Kanemoto, M. Morita, J. Power Sources 81-82 (1999) 217.
Chapter 10
C. Fringant, A. Tranchant, R. Messina, Electrochim. Acta 40 (1995) 513.

D. Aurbach, Y. Ein-Eli, A. Zaban, J. Electrochem. Soc. 141 (1994) Ll.
D. Aurbach, A. Zaban, Y. Gofer, Y. Ein-Eli, I. Weissman, 0. Chusid, 0. Abramson, J.
D. Aurbach, A. Zaban, Y. Gofer, 0. Abramson, M. Ben-Zion, J. Electrochem. Soc. 142
1995) 687.
K. Kanamura, H. Tamura, S. Shiraishi, Z. Takehara, J. Electrochem. Soc. 142 (1995)
340.
K. Kanamura, S. Shiraishi, Z. Takehara, J. Electrochem. Soc. 143 (1996) 2187.
S. Shiraishi, K. Kanamura, Z. Takihara, Langmuir 13 (1997) 3542.
K. Naoi, M. Mori, Y. Naruoka, W.M. Lamanna, R. Atanasoski, J. Electrochem. Soc. 146
1999) 462.
M. Morita, S. Aoki, Y. Matsuda, Electrochim. Acta 37 (1992) 119.
M. Ishikawa, S. Yoshitake, M. Morita, Y. Matsuda, J. Electrochem. Soc. 141 (1994)
L159.
Y. Matsuda, M. Ishikawa, S. Yoshitake, M. Morita, J. Power Sources 54 (1995) 301.
M. Ishikawa, S. Machino, M. Morita, J. Electroanal. Chern. 473 (1999) 279.
M. Ishikawa, S. Machine, M. Morita, Electrochem. 67 (1999) 1200.
T. Hirai, I. Yoshimitsu, J. Yamaki, J. Electrochem. Soc. 141 (1994) 2300.
K. Naoi, M. Mori, M. Inoue, T. Wakabayashi, K. Yamauchi, J. Electrochem. Soc. 147
2000) 813.
M. Morita, M. Ishikawa, Final Report on Electrochemistry of Ordered Interfaces: by
Grant-in-Aid for Scientific Research on Priority Area (No. 282) from The Ministry of
ducation, Science, Sports, Culture, Japan, 2000, 57.
apter 11
TRENDS IN CATHODE MATERIALS FOR

RECHARGEABLE BATTERIES
1 2 3
M. Pasquali , S. Passerini and G. Pistoia
1
Dipartimento ICMMPM, Universita di Roma La Sapienza, Rome, Italy
2
Enea Casaccia, Rome, Italy
3
Consultant, Rome, Italy (E-mail: pistoia@libero.it)
. INTRODUCTION
This chapter will not provide an exhaustive review of the work on

des for rechargeable Li batteries, but will try to focus on the most
ficant recent advances in both fundamentals and applications.
Commercial Li-ion batteries were successfully introduced into the
1
et by Sony in 1991. They used LiCo02 as a cathode and more than a
de later this is still the cathode of choice. Nonetheless, the research on
de materials has been more intense than ever and remarkable success
een met both at the practical and at the theoretical level. Improving the
o02 synthesis has contributed to double the specific energy of
mercial batteries in 10 years: from 80 to 165 Wh/kg. A new cathode,
n 204, has been introduced by NEC24 a in 1996 and Sanyo4 b in 2001,
ugh for a limited market. Some other materials, derived from LiCo02,
LiNi 1_x-yCoxMy0 2 (x+y<0.25, M=Mg or Al, preferably), have become
us candidates as future cathodes.
A wealth of knowledge has been acquired in terms of: synthesis
iques, reaction mechanisms, effect of particle size, interfaces, structural
fications, effect of doping, electrochemical and thermal stability, etc.
knowledge has allowed to revisit the criteria of choice of an efficient
de. The classical criteria based on thermodynamic and kinetic
iderations are:
1. The cathode must intercalate (if layered) or insert (for mono- and
tri-dimensional structures) u+.
2. To show a high OCV, it has to have low electron energy (hence low
Fermi level) and low site energy for Lt.
3. The electrode potential should have limited variations as a function
Chapter II
of Li+ content.
4. The number of sites for Lt has to be high and the hosting molecule
has to feature low weight and high density (high gravimetric and
volumetric capacity).
5. The coupled diffusion of electrons and Lt in the host, as a function
of concentration gradient, must be fast enough to grant a good rate
capability (high power).
6. The Lt intercalation/insertion has to be reversible, i.e., with no or
reversible structural changes, to allow cyclability.
7. The cathode must be stable towards the electrolyte over the entire
operating voltage range.
To these criteria, one has to add the prerequisites of:
8. Low cost.
9. Lack of toxicity.
10. Ease of synthesis.
However, it is becoming evident that the above are necessary but non
cient criteria. One has also to add:
11. A favourable interface formed in contact with the electrolyte (a
cathode may be stable in a given electrolyte, but affected by a
highly resistive interface).
12. The synthesis has to be not only simple, but also reproducible and
must allow the production of a material with the desired particle
size (the latter feature being of the utmost importance).
13. It should be possible to prepare a material encapsulated by another
one to protect it from unwanted reactions with the electrolyte.
14. The material has to be easily processed into a practical electrode.
Indeed, some materials may be difficult to fabricate as sheets (e.g.,
a LiNi02 slurry tends to be thixotropic).
Furthermore, the choice of a cathode is oriented by the applications.
instance, in portable devices, a material can be tolerated which is
ively expensive and moderately performing at high rates; conversely for
power applications, such as in EV, cheapness and rate capability are
datory.
In Table 11.1, materials which have particularly attracted the attention
esearchers in the last 10-15 years are reported together with their
ating voltage ranges. More details on them will be given in the
opriate sections. Before starting the description of selected cathode
rials, we think it useful reviewing the principal methods of synthesis.
section will be followed by two other sections dealing with the effect of
cle size on cathode performance and with the properties of the cathode
ion interfaces. These aspects are of a general character and can apply to
athode materials.
s in Cathode Materials for Rechargeable Batteries 317
ll.J. Cathode materials of interest and their V range.
V Range Typical Materials
5 LiMnz_xMx04 (x .5)
4 LiNiOz, LiCoOz, LiCol-x-yNixMy0z,
LiMnz04, Lil+yMnz-xMx04 (low x andy)
4-3 LiMnOz, LixMnl-yMyOz, Li[LixMyMnl-x-y]Oz
3.5 LiFeP04
3 Mn Spinels, LixMnOz, LixVyOz
2 Sand Polysulfides
1.5 FeSz
. METHODS OF SYNTHESIS
For many years, the only way of synthesizing cathode materials was
hermal treatment at high temperatures of mixtures of reactancts.
sion limitations in the solid state impose both long reaction times and
nding between subsequent firings. Even so, the possibility of having less
pure final products remains high. This can be partially alleviated if a
ant has a low melting point, so that the melt can impregnate the solid
impregnation technique). However, especially if the solid reactant has
particles and limited porosity, the core of the solid may not be reached
e viscous melt, again leading to incomplete reactions.
Since the 90's, alternative methods of synthesis have been proposed.
neral, they tend to form a better mixture of the precursors and to reduce
the time and temperature of the reaction. The most popular of these
sses is sol-gel synthesis. Originally, this term was used for the
lysis of an alkoxide and subsequent heating of the hydrolysis product
ve an oxide. At present, we call solution-sol-gel (or sol-gel, for short)
process in which a solution gives, through a chemical reaction, a sol
h converts into a gel. The latter is typically converted by annealing or
nt-exchange into the desired ceramic (normally an oxide) .
The sol-gel process allows a high degree of control over the physico
ical characteristics of the product obtained with high purity and
geneity. This is made possible by the initial solution step, as the
mixing of the reactants may reach the molecular level. Then, lower
eratures and shorter firing times are necessary to achieve materials with
Chapter 11
electrochemical characteristics. Quite often, this approach allows one

tain small particle size materials and, so, faster reactions through faster
sion of intercalating ions. However, the particle size issue is a rather
ate one, as lowering this parameter may bring about such shortcomings
w packing density, poor thermal stability, higher reactivity with the
rolytes and enhanced solubility. We shall return to this topic in the
wing sections.
In a typical sol-gel process, a mixture of reactants is dissolved in
r, and a solution with a chelating agent is then added and mixed. Slow
oration of the solvent generates a sol and moderate heating of the latter
s a gel, which is then is calcinated at the appropriate temperature. An
ple of sol-gel process is reported in Figure 11.1 for a synthesis of
n 204, using adipic acid as a chelating agent.5
Li(CH3C02) • 2H20
Mn(CH3C02)2 • 6H20 Adipic acid
aqueous solution aqueous solution
I I
I Mixing
Adipic acid/metal acetate

aqueous solution
l Drying at 80-
90°C
Adipic acid/metal sol
I Drying at 90-1oooc
Gel precursors
1 Calcination at 300-800°C in air
Polycrystalline LiMn204
powders
11.1. Typical sol-gel scheme (from Ref. 5).
Control of the pH and concentration of the chelating agent is required

ptimized characteristics. Several chelants have been proposed, e.g.
acrylic acid) (PAA), poly(vinylalcohol) (PVA), and monomeric acids
ccinic, adipic, glycolic, stearic, citric, propionic, oxalic. A technique
tly described is the so-called combustion synthesis. The reactants (e.g.
nd Mn nitrate in the case of Mn spinel) are dissolved in a solution

ther with a fuel (e.g., urea or PAA). The powder formed on reaction is
and then calcined at the appropriate temperature while igniting.
bustion is thought to accelerate the reaction and to bring it to
pletion.6 -9
A peculiar technique combining the features of sol-gel and
10
bustion is the so-called Pechini process applied to Mn spinels. To a
ion containing Li and Mn compounds, a solution of citric acid and
lene glycol is added. The polymeric gel obtained is pre-ignited at
ively low temperature before being calcined. As in the case of sol-gel
hesis, particles in the submicron range, with surface areas higher than
e typical of thermal treatments, are obtained.
Solution techniques with reduction or oxidation processes have also
11
exploited. For instance, Mn spinel can be obtained by reduction of the
12 2 13
anganate with such reducing agents as fumaric acid or Mn +.
2
versely, Mn + aqueous solutions can be oxidized with lithium peroxide,
om temperature and in the presence of LiOH, and the precursors can be
at t<500°C to give Li4Mn 5012 with unusually good performance in the
14
range (Figure 11.2). A material which can be prepared by reduction in
15
ion and shows excellent electrochemical behaviour is V02(B).
ed, the nanocrystalline product so obtained can insert into its
Capacity (mAh/g)
0 50 100 150 200
2
O.lmA/cm
3 --------------
/t'\
,
lOth 5th lst
3
[\. .............-···-·-·-·-·-·-. ..
0.5mA/cm
2
/t""
40th 20th IOth 1st
2-
1 --.---.--.---.--,---.--.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
x in Li 4+Mn sO 12
Figure 11.2. Discharge curves of Li 4 Mn5012, synthesized by a solution technique, in the

3 V range (adapted from Ref. 14).
Chapter 11
odimensional tunnels 0.9 Li per vanadium 16 (290 mAhlg) vs. 0.5 Li per
dium for the thermally prepared analog.
A peculiar synthesis of LiNi02 (and LiNi1_xCox02) is carried out by
izing, in a LiOH solution, Ni(OH)2 to NiOOH with Na2S208, followed
17
he H+;u+ exchange in the same solution. This form of LiNi02 shows a
4 -7 2
electron redox reaction (Ni + Ni +) to which a specific capacity of 300
2
/g corresponds in the potential range 4.2-1.5 V. 30 cycles at 0.9 mA/cm
eported, with a final capacity of 260 mAh/g.
Ion-exchange has been frequently exploited in the last years,
cially to obtain LiMn02 or similar materials. Orthorhombic LiMn02 was
prepared by Dahn et al. by ion exchange in an acetonitrile solution from
18
OH and LiOH. Bruce et al. have substituted Li for Na in NaMn02 and
4Mn02 19by21 ion exchange in alcoholic solutions at moderate
eratures. - The products so obtained show high capacities and long
e lives. They will be more extensively treated in another section of this
ter.
An interesting solution procedure, sometimes utilised to tailor
rials 22with
25 structures otherwise difficult to achieve, is hydrothermal
hesis. - By controlling such parameters as the pH of the medium,
erature, solvent, time and, optionally, size of the templating cation (if a
ial architecture is sought), modified oxide cathodes may be obtained. An
ple is the layered MxMn1_yCoy02 (M=Na or K) which has an electronic
uctivity two orders of magnitude higher than pure KxMn02 .Z Also, a
4
ed Mn oxide can be stabilized 26 by placing pillars (e.g., of vanadium

e) between the Mn02 sheets. The opposite is also possible, that is
26
lizing the layered structure of a vanadium oxide with Mn ions.
27
Other peculiar synthesis techniques include spray drying , emulsion
28 9
ng and freeze drying? In all of them the precursors are initially
lved (suspended) in aqueous solutions enabling the formation of sub
on particles. Whether or not these rather complex techniques can really
seful is difficult to say as limited data on battery cycling have been
rted.
30 31 32
Recently, the synthesis of Li cobaltate and Mn spinels · by
owaves has been reported. The former has been cycled for over 300
s in a lithium-ion cell. Li-rich and fluorinated spinels have been
2
hesised in a domestic microwave oven in less than 40 minutes?
rtunately, no electrochemical data were provided.
A technique which has been emerging 33 in the last34 few
35 years is
hanochemistry. LiM02 (M=Ti, Mn, Fe) and spinels · have been
hesized by ball milling mixtures of Li compounds and Mn oxides. It is
ising that a definite compound may be formed in a limited milling time.
nstance, in 10 minutes LhC03 and Mn02 give rise to spinel formation.
cceleration of the solid state reaction is due to the close contact of the
ents imposed by grinding and plastic deformation of particles.
equent thermal treatment to the desired stoichiometry and degree of

allinity may be carried out. Tarascon's group has shown a satisfactory
cycle even for a non-annealed spinel after 8 hrs of grinding. Sub
2 35
on particles of high surface area (12-18 m /g) are obtained. "
. EFFECT OF PARTICLE SIZE AND

MORPHOLOGY ON CATHODE BEHAVIOR
The choice of a synthesis technique to produce a cathode material of

ntial use in Li batteries should take into account the kind of particles
ned by that technique. Indeed, the size (and its distribution),
hology and density of the particles play a fundamental role on their
rmance. It is also possible to modify the particle size by simple
ing after the synthesis and, in some instances, the behaviour of the
fied material may be surprisingly different.
When evaluating if a material with given particle characteristics is
ble as a cathode, one has to assess its behaviour not only in typical room
erature cycling but also in high temperature storage and cycling. Indeed,
ese latter conditions, unwelcome reactions with the electrolytes, e.g.,
mposition and dissolution processes, may more easily be identified. A
k of its thermal stability is also necessary, as this important parameter
deteriorate with smaller particle sizes.
Compounds based on nanoparticles constitute a relatively new class of
des with, at least, one excellent characteristic, i.e., elevated cyclability.
comes from much shorter diffusion paths for Lt diffusion and the
rrence of smaller dimensional changes upon cycling. Some examples
rovided below.
Nanosized V205, synthesized by the combustion flame-chemical vapor
36
ensation process giving particles of -40 nm, has a capacity of 170
lg (14 cycles), while coarse V 205 has only 130 mAhlg. Furthermore, the
ersible capacity loss after the first charge is negligible for the former,
6% for the latter.
LiMn 204 cannot be cycled in the 3 V range, as the Jahn-Teller (J-T)
t changes the cubic phase (da=1) into a tetragonal one (da=l.16). The
ting large volume increase causes particle fracture and loss of contact.
4 3
ntly, this problem was alleviated by using spinels having a Mn +/Mn +
11
higher than 1, such as Li4Mn5012• However, a surprising result has
reported by Goodenough et al. who could cycle a nearly stoichiometric
l (Li1.03Mnl.9104, Mn oxidation state=3.64), after ball milling, in spite
apparent J-T distortion.37• 38 It has been observed that ballmilling (1-2
produces three effects: generation of lattice strains, partial oxidation of
to 3.76) and formation of nanosized grains within the particles. The
induced by the J-T effect is supposed to be better accommodated in
Chapter 11
mall particles, i.e., it is less anisotropic also by virtue of the pre-existing

n.
A similar effect has been reported for a spinel created by the
rochemical cycling of orthorhombic LiMn02 (o-LiMn02) obtained at
temperature.39 This spinel has a nanodomain structure with particles of
0 nm within the pristine particles. At low rates (3.3 rnA/g), a capacity
of
mAh/g can be obtained, which is 95% of the theoretical value.
ording to the authors, the J-T effect is suppressed in this case due to
n disorder (partial occupation of the lithiated spinel tetrahedral sites)
the occurrence of nanodomains.
The above mentioned excellent cyclability of o-LiMn02 could only be
rved after several cycles. Indeed, the initial capacity was below 40
lg and reached a steady value after 30-40 cycles. Lee and Yoshio have
prepared o-LiMn02 by high temperature synthesis (1000°C followed by
40
ching) and observed the same low initial capacities. However, after a
milling to bring the surface area from 0.55 to 9.25 m 2/g, these authors
d obtain a material providing 193 mAh/g in the first cycle. The
sional limitations due to the relatively large particles (5-15 f-Lm) were
come with the smaller ones obtained by milling.
The above positive effects stemming from the use of small particles
balanced by the drawbacks mentioned in the previous section. In
ical batteries, relatively large particles are now preferred. For instance,
i-ion cells by Sony have LiCo02 produced by a synthesis process using
4, Li 2C03 and PVA resin as precursors, so to obtain a material with
cles of 20 f-Lm, instead
41 of the 1-3 f-Lm which are obtainable in the
entional synthesis. In this way, a better packing density and an
nced safety are obtained.
The importance of particles with a predetermined shape and size is
sed in some articles related to LiNixCo1 _x02• Cho et al. were able to
hesize this material starting from coprecipitated NixCo1_x(OH)2 with
42
rical particles. Reaction of the latter with LiOH at 650-750°C
uces LiNixCo1_x03 2 with the same 3 spherical particles of -15 f-Lm and a tap
ity of 2.9 g/cm (vs. 2.6 g/cm for the analogous product obtained by
mal synthesis). This material cycles much better than the one obtained by
ary thermal synthesis and, furthermore, has a higher volumetric energy
ity as the uniform spherical particles improve the packing density.
The same cathode material (with the exact composition
0.74Co0.2602) was investigated, in its electrochemical and43thermal
vior, with 3 different particles sizes, namely 5, 13 and 25 f-Lm. When
d at the 1C rate, the 5-f-Lm material shows a fairly constant capacity of
mAhlg for 70 cycles. Under the same conditions, the cathodes with 13
25 f-Lm show a final capacity of 160 mAh/g, which is still remarkable at
igh rate used. On the other hand, DSC experiments have shown a large
hermic peak for the 5-!.lm material, while the heat evolved by the two
r materials is rather limited. The conclusion is that the large surface area
2
m /g) of the 5-f..lm cathode and the subsequent reaction with the
rolyte prevent its use in practical batteries.
A similar solution technique ("controlled crystallization") was used by
et al. to obtain spherical LiNi 0.8Co0.202.44 These authors could obtain
cles with a very high tap density (3.24 g/cm\ while a commercial
le not based on spherical particles has a density of 2.27 g/cm3• The
er material could maintain a stable capacity of 170 mAh/g when cycled
2
5 mA/cm for 50 cycles.
A study of commercial Mn spinels has again stressed the importance
e morphology factor in determining the electrochemical performance.
ng the different samples, those having uniform spherical particles
ting from the agglomeration of spherical crystallites showed better
45
bility.
The issue of capacity fading for the Mn spinel upon storing and
ng, especially at high temperatures, has received a wealth of attention.
we wish to point out the reported implication of the spinel particle size
is phenomenon. Tarascon et al. have investigated in detail this aspect
ound a correlation between surface area and capacity loss on storing or
ng at 55°C.46 They prepared samples with different surface areas and
d that as this parameter increases the irreversible self discharge also
ases, due to the enhancement of both electrolyte oxidation and Mn
lution. A low surface area of 0.5-1 m2/g is considered optimal by these
ors.
The issue of spinel surface area vs. capacity fading has also been
essed by Cho through measurements of the surface area by both a
entional BET method and methylene blue (MB) method.47 With the
the area is measured through N2 adsorption, while the second is based
e adsorption of MB molecules. The difference between the two lies on
act that N2 can access the nanopores of the particles, while the larger
molecules, having the same size of the electrolyte molecules, -15 A,
ot. When evaluating the electrochemical characteristics of 5 samples
particles in the 5 to 50 f..lm range, a precise correlation was found
een irreversible first cycle capacity (and capacity fading) vs. MB
ce area. As the latter increased, higher values for the irreversible
city and capacity fading on cycling were measured. Interestingly, BET
MB surface areas did not correlate and even had opposite trends for the
es in the 13-25 f..lm range.
There are some discrepancies in the literature about the influence of
cle surface area on storing and cycling the spinel at high temperatures,
me authors have noticed a better performance with higher surface areas.
It seems that these disagreements may be partly reconciled if one takes
ccount the fact that the BET measurements do not provide a value of
Chapter II
real area accessible by the electrolyte. On this basis, one may perhaps
nalise the results of Yoshio et al. on cycling at 55°C Li-rich spinels of
2 50
ace areas 1.2 and 6.2 m /g. In 100 cycles, the former had a 5% capacity
while the latter had a loss of 8%. With a 511 ratio in surface areas, one
ht have expected a larger difference. It may be conceivable that the two
ples, prepared by different techniques, have a different pore structure in
s of number and size.
All of the above remarks emphasise that any synthesis or post
hesis process should produce particles with an appropriate size and
phology if they are to be used successfully in rechargeable batteries.
niques leading to compounds with exceedingly high surface area or
d particle size distribution are unlikely to be of practical interest.
4. NATURAL VS. ARTIFICIAL CATHODIC FILMS
Any cathode in an electrolyte solution is covered by a film, which

ly determines its electrochemical behavior. There are now several
niques enabling the characterization of such films, e.g., SEM, TEM (and
M, HRTEM), electron diffraction, AFM, FfiR, XPS, EDAX, EIS and r
electrochemical techniques.
The most reactive materials, e.g., LiNi0 2, have a pristine film mostly
51
d on Li 2C03 from the reaction with atmospheric C02• However, the
re of the film changes in contact with the electrolyte. Aurbach's and
scon's groups have explored this topic, taking into consideration
51 53
cially LiNi02 and LiMn204• - According to its own reactivity and the
of electrolyte, each material tends to form a film with specific
acteristics. For instance, the highly nucleophilic LiNi02 is very reactive
rds the electrophilic alkyl carbonates and tends to form thick resistive
s comprising LiF and Li-alkyl carbonates of general formula ROC02Li.
mparison, LiMn 204 is less reactive, as shown by FfiR and impedance
, but it is covered by a complex layer containing, in a LiPF6-based
1 51 54
rolyte, Lip5 , moieties with P (e.g., LizP0 3 F) ' and a protonated, Mn
55
ctive form of A.-MnOz (A-HzMnz_z04).
Li ions migrate through the surface layer before reaching the active
rial for the charge transfer process. So, it is obvious that the
acteristics of the film, especially its conductivity, play a fundamental
in the kinetics of the cathode. LiF is reported to be highly resistive and
explains why, in LiPF6-based electrolytes, the impedance is higher than
1
her electrolytes.5 The surface layer, besides allowing Lt migration, has
rmit electron tunneling. It has also been proposed that this layer is an
ronic conductor, so that the charge transfer may occur at the interface
een the electrolyte and the surface layer. 5 6
As the cathode-electrolyte reaction is a surface process, it is magnified
gh surface area materials. This has especially been observed for Mn

els.55 A sample with a surface area of 5 m2/g, prepared from LiN03 as a
ting agent, was stored at 100°C in solutions of EC-DMC containing
rent Li salts (LiPF6, LiBF4, LiAsF6 or LiC104). With LiPF6, remarkable
dissolution was observed, which was larger for the oxidized state (A.
2) in comparison with the reduced state (LiMn204). The
nic/inorganic layer observed by IR was connected with the dissolution
ess and the underlying electrolyte reaction at the cathode. HF, present in
ion as an impurity or formed at the cathode by a secondary reaction,55 is
onsible for the spinel dissolution and the increase of Mn concentration in
lectrolyte. This process, which is believed to be the main cause of the
l capacity fading upon cycling at high temperature, will be dealt with
in more detail.
Under proper conditions, the formation of a surface film on the spinel
be beneficial with regard to the capacity loss on storage. Pistoia et al.
shown that both undoped and Ga-doped spinels, if stored in the charged
for several days at 55°C in a LiBF4-EC/PC solution, have a much lower
city loss on extended cycling than samples stored only few days. 57 It
s that, with time, the gradual construction of an interface occurs, as
n by impedance measurements, which to a large extent preserves the
rochemical characteristics.
Once it is ascertained that the performance of a cathode largely
nds on the type of layer formed in contact with the electrolyte, and,
generally, on the interactions with the electrolyte, it is conceivable to
cially create a film which can modify such interactions. Tarascon et al.
applied this concept to LiMn204 in an attempt to reduce the loss of
city on storing and cycling at high temperature. They have attempted to
psulate the spinel particles with B-containing glasses through such
iques as: ball-milling the spinel with Liz0.2B203 and firing at 800°C or
ing the spinel with H3B03 followed by annealing at the same
erature. Alternatively, the spinel was treated with acetylacetone, which
ed a monolayer around each particle.53 The treatment at 800°C
ously decomposed acetylacetone, but it was proposed that its
tingldissolving effect on Mn would leave a surface layer mainly
osed of LizMn03. The tetravalent Mn of the latter is resistant to
lution. Storage of these surface-modified spinels at 55°C (4 weeks)
ght the capacity loss from 45-50% to 18-23%. Cycling was not
oved over the uncoated sample.53 Spinel surface modification could also
tained by the use of composite LiMn204-polypirrole cathodes, possibly
gh partial Mn dissolution preventing further capacity fading. Such
des have a higher capacity and a better cyclability at 55°C.58
Cho's group has extensively investigated the coating of several
de materials. In particular, for the Mn spinel they proposed a coating
3 4
on such Co-containing molecules as LiCo0259 a and Co 0 . b59 A
Chapter 11
mally prepared spinel was coated by sol-gel solutions from which, by

g at the appropriate temperature, the above materials were obtained.
yses of the particles by X-Ray diffraction and electron probe have n
that the pure spinel transforms into LiMn2_xCox04, with a higher Co
entration at the particle surface. In other words, a Co-rich coating was
ed, which preserved the particles from the dissolution reaction with HF
iPF6-containing solutions at high temperature. As a consequence, a
tly improved cyclability was observed. For instance, a coated spinel
d sustain 50 cycles at 60°C without showing an appreciable loss, while
ncoated one lost 41% of its initial capacity. It has been shown that Mn
lution and capacity loss is accompanied by major structural changes.59b
spinels tend to approach compositions of the type Li 2Mn409 and
n5012, characterized by low values of the cubic parameter a (e.g., 8.12-
A). The uncoated spinel shows, after cycling, a=8.12 A vs. an initial
e of 8.23 A, while no change was observed for the coated one. 59 b A
el coated with LiNi 1_xCox0 2 (.x=0.2 and 1) has also shown improved
60
ge and cycling characteristics at 65°C.
The technique of reducing (or eliminating) electrode/electrolyte
ndary reactions has successfully been applied to other cathodes. Both
o02 and LiNi02 were covered by CVD with a carbon layer (diamond
for the former, amorphous for the latter).61• 62 In such a way, it was
ible to charge both cathodes up to 4.6 V without electrolyte
mposition. As a result, initial capacities above 200 mAh/g were
ned for both compounds. However, in limited cycling experiments
cles), only LiNi02 showed a very low capacity loss.
Coating of LiCo02 was also addressed by applying such oxides as
63 66
2, Ah03, and Zr02 on the cathode surface. - LixCo02, when
hiated, undergoes several phase transitions. In particular, at x=0.5, a
gonal-monoclinic transition (H -7M) is observed which gives a
tantial contribution (about 50%) to the total anisotropic expansion along
axis (about 2.6% with respect to the initial c value). It is accepted that
hase transitions in delithiating LiCo02 are reversible up to the Co02
position.67" Nonetheless, the large anisotropic volume changes
ot be tolerated by the particles, so that fractures and contact losses
e. Furthermore, a remarkable Co dissolution at high potentials has been
enced.67<a-c) This is why commercial cells are only charged up to about
V. Coating with the oxides mentioned above changes the surface of
o02 and this has a dramatic effect on the cycle life performance even if
harge cut-off voltage is brought well above 4.2 V. For instance, a stable
city of 168 mAh/g was obtained for 70 cycles with Zr02 coating in the
ng range 4.4-2.75 V.63" In Refs. 63a-65, this behavior is attributed to a
ng-induced structural stabilization. In particular, the best oxides, e.g.
and Ah03, would greatly limit or even suppress expansion along the c
thanks to a strain-preventing shell formed around the LiCo02 particles.
This view was rejected by Chen and Dahn, who found, for
crystalline ZrOz-coated LiCo02, the same variation of the c axis
66
ured for the uncoated sample. According to these authors, the
llent electrochemical performance of this material has to be possibly
ected with a reduction of side reactions involving oxygen loss from
o02 to the electrolyte.
A reconsideration of the effect of coating LiCo02 with oxides has led
s group to attribute the capacity stabilization to a reduced Co
lution at high potentials.63b Even after charging to 4.8 V, Zr02-coated
o02 showed a much lower Co dissolution with respect to the uncoated
A final example of performance improvement by cathode coating is

ded by o-LiMn02• On cycling, this material transforms into a spinel
oxide of general formula Li 1+xMn2_x04 (O<.x::;0.33; the upper limit
sponding to Li4Mn 5012). As cycling goes on in the extended range 4.5-
, formation of the rock salt phase LizMn 2 04 occurs, with all Mn in the
ent state and, so, quite prone to dissolution through a disproportionation
ion producing Li 2Mn03 and MnO. An uncoated o-LiMn02 does indeed
after 50 cycles the presence of the LizMn03 peak, while a significant
ase of this peak was observed in a sample coated with CoO or
68 69
3. • The stabilization of the spinel formed upon cycling results in a
city of about 170 mAh/g after 50 cycles, while a bare sample gives 120
lg.
. SPECIFIC CATHODE MATERIALS
.1. LiNi02 and LiCo02
The early report by Goodenough et al. on a new layered compound,

Oz, cyclable at 4 V, received limited attention in 1980 as there were not,
e time, electrolytes suitable for that potential range.70• However, a
de later, LiCo02 became the cathode used in the first commercial Li-ion
ies and is still in use.
The structural features of LiCo02, and of its homologous LiNi02,
received a great deal of attention. They are built of M02 slabs (with M
tahedral coordination) separated by Li layers (with Li also octahedrally
inated), as shown in Figure 11.3a. Lt can entirely be removed from the
s of the 03 hexagonal structures, through different phase changes. The
generally accepted phase-composition diagrams are shown in Figure
,b. Lt removal from LiCo02 or LiNi02 causes an enlargement of the
ayer distance (equal to c/3), due to slabs repulsion, for up to -50%
iation, followed by a decrease. This is depicted in Figure 11.4c,d.
The large variations along the c axis cause volume reductions of 6-
Chapter 11
at full delithiation. This is too much to grant a practical reversibility and,

fore, the utilisation of the full capacity (273 mAh/g). It has been pointed
that both Co02 and Ni02, obtained by electrochemical delithiation, can
67
ert 90-95% of Li in a low-rate discharge. a This means that the phase
itions, and the related volume changes, are reversible. However, as
ted out in the previous section, volume expansion causes particles
king and contact losses impairing true reversibility in practical
itions, this adding to other unpleasant effects as electrolyte
mposition and partial cathode dissolution at high potentials. This last
omenon has especially been evidenced for LiCo0 67a. Lee et at. have
2
ed a remarkable reduction of cobalt dissolution by preparing solid
2
tions of LiCo0
67 with Aurbach et al. have obtained similar
LhSn03. b
ts by using a boron-based additive which forms a protecting film on
Oz. 67c
In the following, we shall review interesting findings regarding the
tural characteristics and electrochemical behavior of LiCo02, LiNi02
their derivatives. Stoichiometric LiCo02 is easily prepared by soft
istry or high temperature processes. The preparation of LiNi02,
ever, is not straightforward, as it tends to become Li-deficient while
2
lent Ni is present in both the Lt layers and Ni02 slabs. Ni + in the Lt
s has detrimental effects. Indeed, in the early oxidation stages it gives
to the smaller Ni 3+ (r=0.56 A vs. 0.65 A for Ni2+), so that the structure
pses around this ion and the following Lt intercalation is hindered, as
shown by the irreversible capacity loss in the first cycle.73·74 Other
easant features of LiNi02 are: limited thermal stability in the oxidized
and poor capacity retention during cycling. This is the reason why, in of
its lower price and toxicity with respect to LiCo02, the latter is the ode
of choice for practical batteries. However, in recent years the
hesis of nearly stoichiometric LiNi02 has been reported.74 This material
cycle fairly well for as many as 1200 cycles, without showing structural
ifications.75 However, the safety issue remains, as a large exothermic
is manifested by the charged cathode at about 200°C (the exact
erature depending on the state of charge).76 Therefore, only compounds
partial Ni substitution are of practical interest. Before going into the
ls of LiNi02 or LiCo02 derivatives, we wish to stress that the structural
ges of the pure materials upon u+ removal is still a matter of debate. Me
n et al., by in situ XRD experiments using a synchrotron X-ray source,
questioned the formation of monoclinic phases in LiNi02 (Figure
a). According to these authors, the first-order transitions give rise to 3
es with hexagonal structure.71 Bianchi et az.14 have pointed out that
e coexistence and the extent of the 2 unit cell volume change upon
hiation strongly depend on the Ni + content in the Lt layers. By
ving 60% of Lt from Li0.975Ni1.02502, the volume 74 decreases by 1.4%,
e with Lio.992Ni1.00802 the variation is only 0.4%.
ds in Cathode Materials for Rechargeable Batteries 329
LiM02 A.-Mn02
Symmetry:Hexagonal Symmetry: Cubic
Space Group: R 3 m Space Group: Fd3m
c d
o-LiMn02 LiFeP04
Symmetry: Orthorhombic Symmetry: Orthorhombic
Space Group: Pmnm Space group: Pnma
e 11.3. a) shaded octahedra: M0 6 , white octahedra: Li0 6; b) note two of the pathways
Li+ transport; c) hatched octahedra: Mn06, circles: Li+; d) tetrahedra: P04 ,
edra:Fe06, circles: Li+ (in octahedral sites). a-c) from Ref. 70b; d) from Ref. 70c.
For LiCo02, the second monoclinic phase at low Li+ content (Figure
1
b) is not observed by Me Breen et al./ who found instead
her hexagonal phase. The phase composition above -4.3 V has recently
reconsidered by other authors. Chen and Dahn 77 " have found that the
gonal 03 structure·becomes a stage-2 compound of formula Li 0.12Co02,
hexagonal as previously proposed by Ceder et az.77b At full delithiation,
2 with an 01 structure is formed. This latter finding is in agreement with
esults of Yang et al.71c and Tarascon et alY" The latter authors, in a
equent paper, assign a mixture of two 01 phases to Co02.78 However,
01 and 03 structures, Li is in octahedral (0) coordination and the unit cell

ins 1 or 3 Co02 sheets, respectively. In P3, Li has prismatic (P) coordination
he unit cell has 3 sheets.
Chapter 11
15.5 Hl H3
HI +
•
6.0
Cl
II 4.1S
!!
0
> 4.0 (a)
3.5
3.0
o.o o.1 0.2 o.3 o.4 o.5 o.e o.7 o.8 o.g 1.0
x In L1 1_"NI02
---- - ;:!/
(b)
5.0 lit HI+H2 H2 Ml H2 M2
----
Cl
! 4.5
0 4.0
)
3.5
3 0
" o.o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 O.SI 1.0
X In Ll 1_wCo02
.. \ ••• ••••••
•••••
- 13.5
u
14
;;:
o -Monoclinic
"""""
(c)
13
•• •+CoO2
12.5 ....._ ......._......_......._.......... _._. _._............
-0.2 0 0.2 0.4 0.6 0.8
X in LixCo02
14.5
14.5 oo
co 0
14.4 CtJ 0 OofJOO
0 0
14.4 cP
-< 14.3 OJD
(.)
14.3 cx:fX)
13.6 (d)
()
14.2 13.5
14.2
0.0 0.2 0.4 0.6 0.8 1.0

x in Li1_.Ni02
11.4. (a) Phase diagram of Li 1.xNi02 (from ref.71); (b) Phase diagram of Li 1_xCo02
Ref. 71); (c) Variation of the c axis during delithiation of LiCo02 (negative x values:
olyte decomposition) (from Ref. 67a); (d) Variation of the c axis during delithiation of
2 (from Ref. 72).
79
rding to Venkatraman and Manthiram, Co02 has the P3 structure. e
authors observe formation of a P3 phase for x=0.45 in LixCo02, h
coexists with the 03 phase in the range 0.2<x<0.45 and is
acterized by a faster decrease of the c axis as x decreases. The P3 phase
oxygen more rapidly than the 03 phase (P3-like Co02 should be
en as CoOu) and this can be connected with the practically usable
city of LiCo02.
Examples of the cyclability of pure LiNi02 and LiCo02 are given in
e of the references of the next sub-sections as well as in the references
ng with coated LiCo02 in Section 11.4.
We have already noted that LiCo02 is the cathode used in most of the
ently commercialised Li-ion batteries. LiMn204-based cells have a
inal market. Batteries with LiCo02 have now reached energy densities
cess of 150 Wh/kg and 350 Wh/1, cycle lives in excess of 1000 cycles
low self-discharges (<3%/month). The industrial processes are well
lished and high levels of mass production of portable batteries have
achieved through huge investments. On this basis, it is very difficult to
ider the substitution of LiCo02 with another cathode, unless the
native is highly attractive in terms of performance and price. One has
to consider that the price of cobalt has been drastically decreasing in
t years and that in small portable batteries the relative cost of active
rials is lower.
.1.1. LiNit.xCox02 Solid Solutions
Taking LiNi02 as a reference, a number of cations have been used to

titute Ni, e.g., Co, Mg, AI, Fe, Ti, Ga. In some cases, 0 has partly been
ced by For S. Co is the most obvious substitute for Ni as they form the
1.xCox02 solid solution at all x values. In view of the need of exploiting
potentialities of LiNi02, Ni-rich stoichiometries (70-80% Ni) have
cially been investigated. For .x>0.3, LiNi 1_xCox02 are strictly layered
75
rials, with no Ne+ in the Lt layers. Delmas' group has extensively
ed this system, with special emphasis to the structural evolution upon
ging. These authors have reported that, in the mixed Ni/Co oxide, Ni
75
are first oxidized. By further NMR experiments of the same authors, a
3
pound of formula LixNi03Co0.702, was said to have only Ni + oxidized in
ange 0.85<x<l. However, in the following 0.70<x<0.85 range, Ni/Co
4
ing was evident (Ni4+ + Co3+ Ne+ + Co +) so that the question as to
80
h ion is preferentially oxidized on delithiation becomes less relevant.
prevailing charge carriers are electron holes and even at x=0.40, where
LiCo02 is a metallic conductor, the presence of Ni hinders long range
80 81
ronic delocalization. By using XANES spectra, Me Breen et al. have
luded that Ni is first oxidized, while, with the same technique, Lee and
Chapter 11
have found that both ions are simultaneously oxidized. These latter
have also examined the relationship between irreversible capacity
d depth of first charge. They found that most of the capacity is lost at
ginning of the first charge and is not influenced by the phase
ons occurring during further charging/discharging. The presence of
uces the irreversible capacity, which is due to the oxidation of Ni2+ in
layers and preferential removal of u+ around the Ni ions, this
a local structural collapse.82
Co not only improves the layered characteristics of LiNi02, but also
es its thermal stability in the charged state. This is because the Co-O
s stronger than the Ni-0 bond (binding energies: 1067 KJ/mol for
nd 1029 KJ/mol for Ni02).83 As a consequence, a thermogravimetric
s has shown progressively reduced weight losses to 1000°C as Co
s Ni in LiNi 1_xCox02• 83 The stronger Co-O bonds also contribute to
e the structure during Lt intercalation/deintercalation. In addition to
excess helps to reduce the amount of Ni2+ in the Lt layers, as these
rated by Li+, and thus a limited irreversible capacity in the first cycle
ood cyclability are observed. 83
An investigation on the thermal stability of delithiated Li 1_yNi 1_xCox02
.3), has produced surprising results. All samples have shown
al damages on storage, demonstrated by a decrease of the cia ratio
temperatures as low as 50°C. The rate of change of the cia ratio
es with time (several days), temperature and oxidation degree, and
es with the Co content: a Ni-free sample does not show any
n.84 The thermal stabilization of oxidized Li 1_yNi 1_xCox02 at higher Co
s is confirmed by DSC experiments. In samples charged to 4.2-4.3 V,
e exothermic peak at about 195°C of pure Li 1_yNi02 is transformed
85
small peak centred at 220°C for Li 1_yNi Cyclic
0.7Co0.302.
metry tests of this latter compound have shown the disappearance of
agonal-monoclinic phase transitions and delithiation seems to occur
85
gle phase up to 4.4 V. Several authors have attributed the im roved
6 87
hemical stability of the Co-containing cathodes to this feature. •
ccording to Ref. 79, the original 03 phase coexists, for x<0.25 in
5Co0.1502, with another hexagonal phase, 03'. At full delithiation,
he new 03' phase exists. Oxygen loss would occur only after
on of this phase. This behaviour is different from that of LiCo02 n
loss at x<0.45, see Section 11.5.1) and would explain why
5Coo.1502 can reach a practical capacity of 180 mAh/g vs. 140 mAh/g
o02•
he references of this sub-section give details on the best syntheses to
e efficient materials. Here, we only wish to point out an unusual low
ature synthesis based on the reaction at 400°C (instead of the usual
86
0°C) between f3-Ni1_xCoxOOH and LiN03• Since Ni is already in the
t state in the starting material, a lower temperature may be used. A
o.ssCoo.tsOz composition gives in a one-phase discharge (i.e., with no

2
M transitions) 190 mAhlg at 0.5 rnNcm and a very limited capacity
on cycling.
.1.2. Partial Substitutions in LiNh.xCox02
To further improve the electrochemical and thermal stability of

Oz, several cations have been proposed for partially substituting Co/Ni,
e S and F have been proposed for 0. A cation of particular interest is
used by several authors at low concentrations (around 0.05 atoms/mol)
87 90
mpounds of the type Li(Nio.7sCoo.zs)t-xM&Oz. - The role of this ion is
ted in the literature, even after carrying out Rietveld analyses to
mine the site occupancies. Delmas et al. found that in
.s6Coo.ogMg0 _0502, part of Mg (0.02 atoms/mol)
88 resides in the Li layers,
the remaining Mg is in the Co/Ni slabs. Mg +, whose radius (0.72 A)
2
milar to that of u+ (0.73 A), prevents occupancy of the Lt layers by

and, as it cannot be oxidized, does not cause any local structural
150 ------r--------- ---------- ------------

140
13Dt--- -1----------i--- ---- --
120
l 110
!! 100
. 90+------- ---------+---------- ------------
f 80+------- ---------+---------- ------------
15 70
j
80
= +------- ---------+---------- --------------
30
20 +------- ---------+---------- ------------
10
0
0 100 200 300 400 500 800 700 BOO 900 1000 1100 1200
Cycle number
11.5. Cycling of a 40 Ah Li-ion SAFf battery with a LiNi 1.x-yCoxAly02 cathode

5%). C/3, 80% DOD (from Ref. 95b).
pse and, so, is beneficial for the electrochemical behaviour.

ermore, during an extended cycling, all of the Mg ions migrate
some Ni) from the Co/Ni slabs to the u+ layers. In aLi-ion cell cycled
imes at 1 Crate, the capacity loss was strongly reduced, while the Mg
homologous showed a rapid decline. At the end of cycling, an XRD
sis showed limited variations of the cell parameters for the Mg
ining compound, in agreement with the low capacity fading. Chowdari
l., studying the cathode LiNi 0.8Co0.2.2xTixMgx02, also inferred the
Chapter 11
ence of Mg in the interslab sites from the relative heights of the (003)
(104) peaks in the diffractograms: a lower (003)/(104) ratio for x=0.05
0.075 was taken as a proof of the distribution of Mg between Li and
91
o layers.
Conversely, Cho et a/.89 and Chang et a/.90 have refuted the presence
g in the 3b sites. According to the former authors, as the Mg content
ases from 0 to 0.06 atoms/mol, the Mg and Ni occupancy of the Li sites
mes negligible (Rietveld analysis). Chang et al. also found a very low
concentration in the Li sites, through Rietveld, chemical and
mogravimetric analyses. Despite the disagreement regarding Mg
pancy, all of the authors agree that these materials have excellent
abilities with high and stable capacities (e.g. 160-170 mAhlg 89).
1
Mn, Ti and AI substitutions also give excellent cyclability.92 -95 LiNi _x
Aiy02 is now receiving special attention at the technological level as it is
ved to be a candidate for a new generation of Li-ion batteries. 95 Indeed,
n be produced on a large scale and is commercially available. A 10 Ah
n cell was cycled at AEA for 270 cycles (C/5) with a 90% capacity
tion. The specific energy of this cell was 150 Wh/kg in a "soft
aged" version. 95 "A similar performance (up to 145 Wh/kg) was reported
AFI' for a 40 Ah cell.95b,c This cell can be long cycled with a rather
tant energy output- see Figure 11.5.
Dahn et al. have reported the structure and electrochemical behaviour
2
i[NixCo1_2xMnx]02 (x=l/4 and 3/8), prepared by substituting Ni + and
92
+ for Co3 in LiCo02. " This new compound, which maintains the a
+
e02 structure of LiCo02, can deliver a stable capacity of 160 mAhlg
een 2.5 and 4.4 V, and proves more thermally stable than LiCo02•
The double cationic and anionic substitution in LiNi02 has also been
96
osed, e.g., in: Li .075Ni 0.755Co0.1701.9F0.1 • This material has an initial
city of 182 mAh/g and a capacity fading of 2.6% in 100 cycles at 1
cm2 .
Delmas' group has partly substituted Ni in LiNi02 with such ions as

73
AI or Mg. 75' They confirm the presence of the last ion in the Lt layers
gh Rietveld analysis of XRD data and magnetic measurements: only
iamagnetic Mg2+, and not the paramagnetic Ni2+, is found in the layers.
bstitution of 5% Mg for Ni is enough to suppress lithium/vacancy
ring in the interslab space and, so, all phase transitions observed upon
hiation.73 A limited amount of Mg substitution (<0.02 atoms/mol)
ted in capacities of 150 mAh/g.
AI has been used by several authors for substituting both Ni and
7 98
' Ohzuku et al. have found that Al in Li1-xAlu4Nh,402 produces a
e-phase behaviour and allows charging up to 4.5 V with excellent
97
sibility of the 160 rnAh/g initial capacity. The presence of non
3
izable Ae+ limits the degree of Ni + oxidation, thus avoiding the
iderable shrinking of the c axis at high x values and consequent loss of
city. In contrast, Al substitution in LiCo02 does not have a positive
t and fast capacity fading is observed. Al substitution increases the
ge in both LiNi02 (by -0.1 V) and LiCo02 (by -0.2 V).
2
Instead of counterbalancing the tendency of Ne+ to form Ni + in
4
i02 with di- or tri-valent ions, in one report Ti + was used to compensate
99
he electric charge deficiency and structural instability due to N?+. The
bility of LiNi1_xTix02 (0.025:SX:::;0.2) was rather good with initial
cities ranging from 170 to 240 rnAh/g as a function of x.
A double Mg-Ti substitution has been reported to have a dramatic
t on the thermal stability of LiNi02. Indeed, compounds of formula
1_xTix12M&1202 show a progressive disappearance of the large
hermic peak of pure, fully oxidized LiNi02 as x increases. For .x=0.3, no
76
herm was observed. " These materials, however, have large irreversible
cities. Including Co among the dopants, the materials become highly
hiometric (99% Li+ in the Lt layers for LiNio.7Co0.2Ti0.05Mg0.0502) and
an excellent cycle performance in addition to thermal stability.76 b This
eature has been explained by arguing that the highly ionized Ti4+ and
(plus unremovable Lt) increase the ionic character of these
pounds and prevent depletion of electrons from the oxygen sites during
ge.76c
6 100 101
There are reports of 0 substitution with p9 • or S. The latter seems
e particularly effective in stabilizing the capacity of a compound2 with
ula Li1.QJNi01.98S0.02 at -140 rnAh/g in 100 cycles at 0.4 rnAJcm .
.2. Manganese Spinels
The advantages of a Mn spinel cathode (structure in Figure 11.3b)

LiCo02 are well known: lower price, lack of toxicity and better thermal
lity. On the other hand, the spinel has a lower specific capacity (around
120 rnAh/g vs. 130-140 for the cobaltate) and serious problems in
ge and cycle life at high temperatures. Some years after the first
mercial release (1996), NEC has introduced a second generation of
24
l batteries with improved capacity and high temperature cyclability. "
e 11.6 depicts the cycling behaviour of a battery for HEV
4
cations. " Note that, at 60°C, 87% of the initial capacity was retained
500 cycles.
Sanyo has also developed in 2001 aLi-ion polymer battery with a
Chapter 11
95
90
.g 85
80
z:. 75
-o-
· 8 70 ... -&- 45'C ............·-·-·--- --·-··-··-··"·-·-··
65 -o- 60'C ..----··-·-······-·-·····---·--··········.......
600 1.000 2.000 3.000
Cycle (Charge 4.1V, Discharge 2.5V)
number
e 11.6. Cycling of a NEC battery with a Mn spinel cathode. Rate: 1 C (from Ref. 4a).
ode based on Mn spinel. This cathode also contains LiCo02 which

rols the gas evolution during storage at high temperature. A good
4
ng performance at room temperature for 2000 cycles is reported. b
Thus far, such batteries have found only a niche in the high-tech
et (computers, cellular phones, cameras, etc.), but will probably find
acceptance in industrial and EV/HEV applications.4•·95b
Researchers in this area are steadily working to better understand the
plexity of this material and to overcome the problems cited above. The
literature on this material presents several discrepancies with regards to
aspects as: phase transition of the as-prepared material just below room
erature; phase transition upon delithiation in the 4 V range; origins of
apacity fading during storage/cycling; Mn dissolution as a function of
tate of charge and products formed during dissolution.
LiMn204 undergoes a structural phase transition at -280 K. Even the
hanism of this process is debated. Some authors have supposed, as a
equence of the Jahn-Teller effect, the formation of a tetragonal
71 102
e, • while others have reached the conclusion that a cubic
103 104 3
rhombic transition occurs. ' Substitution of a small amount of Mn +
3 4
foreign ions, e.g., Li, Ni, Co, Cr, lowers the Mn +/Mn + ratio below
103 104
, which suppresses the transition. ' However, Basu and Seshadri
rejected the view of a transition linked to a J-T effect and suggest that it
105
stead due to charge (holes and electrons) ordering. Yoshio et al. have
106
ed out that oxygen deficiency is the primary cause of the transition.
The simple delithiation, i.e. u+ removal from the tetrahedral 8a sites
g charge, is also much debated and has been variously interpreted as a
ess involving one, two or three phases. Me Breen et al. have shed light
106
is question. In situ X-Ray data have shown that stoichiometric spinel
n204 gives rise to a single phase for 0.5<x<l and to a two-phase system
x<0.5. Li-rich spinels (and spinels doped with another ion) are single
e throughout the entire x range, and, finally, oxygen-defective spinels
Mn204-S) show two biphasic plateaux separated by a voltage knee at
5. The two-phase regions show the fastest capacity decay on
cycling.
Much debate is also ongoing regarding the failure mechanisms upon
ge or cycling at high temperature. The hypotheses are: 1) Mn2+
olution; 2) J-T effect with formation of the tetragonal spinel LizMn204;
ructural instability and 4) decomposition reaction of the electrolyte at
voltage. It is quite possible that such effects are interrelated. For
nce, points 1-3 can occur if the following sequence is considered: in a
harge down to -3 V under non-equilibrium conditions, LizMn 204 is
ed; this can then disproportionate according to the reaction:
(11.1)
the latter dissolves in the solution; as a consequence, a structural

anization has to occur which can greatly impair the cyclability.
tion (11.1) is thought to be more important at low voltages, i.e., when
atio Mn 3+/Mn4+ is maximized. Indeed, several authors report higher Mn
bility values in the fully discharged state, 107• 108 or at low states of charge
109 110
). ' However, others report higher solubility at full charge. 111' 112 In
case, dissolution implies a complex reaction mechanism based on
rolyte decomposition and reaction with A.-Mn02• The cathode would
be covered by such by-products as MnF2, Li 0.5Mn02, LizPOF354 and
1
In all cases, the particle surface is modified and a polarization ensues
contribute to degrade the electrochemical characteristics. Impedance
have been particularly useful in showing the resistance increase due to
52 113
ation of surface layers. •
2
It has been demonstrated that the amount of dissolved Mn + is
ally limited. Inoue and Sano report a weight loss of 1.1% for a totally
arged spinel exposed to a LiPF6-based solution at 80°C for 2 days.108
ever, this spinel has a capacity loss of 59% after storage. According to
authors, the modest solubility has a dramatic effect on the structural
rity, creating lattice defects that adversely affect the Lt diffusion paths.
dition, structural degradation during the first charge, especially at high
, has been suggested as a cause of performance loss for stoichiometric
114
els.
Another proposed mechanism for rationalizing the capacity fade on
ge at high temperature is the formation of a partially protonated,
ive A.-Mn02 in solutions containing LiPF6, where HF is easily formed
ecomposition.55 In this study, by Tarascon et al., the protonated phase
bably Mn-defective) was found to form both in the charged and in the
arged state. Gradual accumulation of A.-Mn02 was also noticed
Chapter 11
hite et al. in prolonged cycling (up to 700 cycles) at room temperature

1 Ah Li-ion battery with a spinel cathode. 115 These authors have also
d evidence, by EDAX experiments, that Mn metal plated onto the C
e, as a consequence of the dissolution promoted by HF generated during
rolyte oxidation.
The Mn spinel structure undergoes considerable contraction/expansion
ng delithiationllithiation, regardless of the number of phase
formations involved. The cubic parameter a typically contracts from
to 8.05 A (2.5%). As a general rule, this level of contraction/expansion
create problems with the integrity of the particles and their
connection. By substituting Mn with other metals (e.g., Ni, Co, AI), it is
ible to limit the variation of a, and, so, of the lattice volume. Doped
els showing lower volume changes also show reduced capacity
s. 116" This has been confirmed by Shin and Manthiram. 116b They have
sured the !!.a difference between the two cubic phases formed for x<-0.5
xMn204. Lower capacity losses correspond to lower !!.a, the latter being
red by such substituents as Li, Co or Ni.
Other effects brought about by metal dopants will be later discussed.
We have stressed in Section 11.4 the role of the surfaces in
rmining the cathode characteristics. The surface changes stemming from
ge and/or cycling of the Mn spinels have been the subject of extensive
stigations aiming at understanding the reaction mechanisms and their
ucts. Direct evidence of electrolyte decomposition during room
erature storage at high voltages in LiBF4-EC/PC was obtained by SEM
57
riments showing spinel grains heavily covered by fibers. In LiPF6-
DMC, the original 100-nm grains of fully charged cathodes were
red, after storage at 70°C, by fine structures of 10-30 nm. 117 XPS
yses have shown that MnF2 was formed on the surface and that the
ation state of Mn changed from Mn02 to MnO.
The surface composition after cycling was examined by several
ors. There is a general consensus on the formation of tetragonal spinel
118
n204, as also indicated by electron diffraction. " According to Cho and
keray, usb reaction (11.1) then occurs, this accounting for the formation
dissolution of Mn2+. Other authors suppose that the Mn+4 compounds
ed in the disproportionation are cation-defective spinels of the type
107
n409• According to Yoshio et al., an oxygen-defective spinel,
119
204-z• would be formed.
The approaches taken to counteract the capacity fading include:
ng additives to the solution or to the cathode, encapsulating the particles
a protective layer or doping the spinel with foreign ions.
A soluble additive suggested by Sano et al. is (CH)3SiNHSi(CH)3, a
cule able to react with water traces to produce NH 3 which can neutralize
Mn dissolution was drastically reduced with this additive and so was the
120
city fading on storage at high temperatures. The cathode of the above
ioned NEC battery is formed by blending the Mn spinel with 10-15%
121
0 .sCoo.zOz?' The latter material acts as a scavenger for HF, thus
ressing Mn dissolution.
We have already mentioned the effect of coating the spinel particles
such materials as Co304 or LiCo02 prepared by sol-gel synthesis. 5 9 The
layer formed on the spinel, possibly with the LiMn 2.xCox04 composition,
sistant to HF attack and prevents Mn dissolution. As a result, a steady
city of 105 mAh/g was obtained on cycling at 60°C.
3
Substituting part of Mn + with other ions was suggested by the
ence of a J-T effect in the 4 V range. The spinel was initially doped with
ent ions as Mg2+ or Zn 2+. 122' 123 The capacity was indeed stabilized
3
ugh a penalty in its absolute value was paid, as the amount of Mn +
h determines the capacity was decreased, according to the general
3 4
ula: Li[Mn\Mn + 1+(n-4)xMn + 1_(n-J)x]04. The numerous reports on this
have shown that the doping ions also have other beneficial effects: they
124
ce the Mn solubility in the electrolyte; often give higher energies for
125
M-0 bond; limit the extraction of u+, and give rise to a single-phase
126
ge-discharge process. All the above effects contribute to the structure
lization, which is ultimately the basis for a good cyclability.
Several elements have been used as dopants. Substitution with Li is
ost obvious choice: it is generally agreed that Li+ replaces Mn 3 + in the
octahedral sites, from which it cannot be extracted. However, Howard et
y observing that Li+-doped spinels have an experimental capacity larger
the theoretical one, have suggested, also on the basis of neutron
127
action data, that Li+ may also reside in tetrahedral 8b sites. In another
128
rt, occupancy of octahedral 16c sites is suggested. Lii+xMn2_x04
ls, even though showing a better cyclability than the undoped one,
problems on storage at high temperature, with an enhanced formation
107
i2Mn03/LizMn 4 09, and do not compare favourably with other doped
129
ls in terms of capacity retention.
Divalent metals, as Zn and Ni, or trivalent as Co, Cr, Ga and Al show
est performance. Ni and Cr are particularly effective in reducing the Mn
124
lution and, so, the spinel instability. Capacity losses as low as 0.02-
mAhfg·cycle at room temperature and 50°C have been measured during
129 130 132 133
cycling. ' ' ' Reports indicate that AI, Zn and Ga may reside in
131 132
8a tetrahedral sites. ' In principle, they could create diffusion
ems for u+ along the 8a-16c-8a paths, but, at least for the limited
itutions considered, this does not occur and, instead, both show a well
129 132 133 131
lized capacity. ' ' According to Thackeray et al. substitution in
hedral sites may retard the Mn diffusion to these sites, a phenomenon
mpanying its dissolution.
Excellent data have often been reported for double doping. Not only
Chapter 11
57
different cations be used to replace Mn, e.g., Ga-Cr, butS or F may be
to partially replace 0. Tarascon et al. have proposed LiA10.2Mn1.804_sFz
0.06
0.05 13% Capacity Loss
0.04
0.03
PLiON™Li-ion
LAMOFvs. Li 4Ti5 012
O.ot 55°C C/7
approx. 6 months
0.00 +"........4 ................+-'-......."+..............+-'-.......'+'-'-l..l...j..............................-'+

0 50 100 150 200 250 300 350 400
Cycle#
e 11.7. Cycling of aLi-ion cell with LiAio. 2 Mnl.804 _sFz as a cathode (from Ref. 134).
134
h z;::;(}.5). The improvement of cyclability of this material at 55°C over
-doped spinel is dramatic as shown in Figure 11.7, where cycling of a
mer Li-ion cell at 55°C is reported.
Sun et al. have substituted 0 with a limited amount of S:
135
0.24Mn1.7603.98S0.0z. AI stabilizes cycling in the 4 V range, whereas S
is tive in stabilizing cycling in the 3 V range. Overall,
capacities eding 200 mAh/g could be obtained. Leaving aside for the
moment the
ical interest of a cathode showing a sharp voltage drop of 1 V, the
lity in the 3 V range is highly surprising. Furthermore, the authors have
n that, even upon long cycling down to 2.3 V, the cubic structure of the
l spinel is preserved, i.e., there is no J-T effect causing the tetragonal
rtion, perhaps due to a more flexible structure of the oxysulphide. A
136
lar result has been found by using Cr instead of Al. Kang and
denough have commented on these results. According to them, sulphur
(r=l.70 A vs. 1.26 A for oxygen) would only reside on the spinel
ce and the J-T distortion should be present as the 3 V plateau does
137
r.
Yoshio et al. have shown that heavily doping the spinel with Co or Ni,
in LiM0.5Mn 1504, also provides a good stabilization of the 3 V
138
city.
ds in Cathode Materials for Rechargeable Batteries 341
C>
c: 250
<
E
-
200
>-
(3 150
C tl
c.
Ctl
.) 100
Q) 50
C>
...
Ctl
c:
.)
0
0 0 20 40 60 80 100
Cycle number
e 11.8. Cycling of a cell with a Li,Mn 0.975Li0.02502 cathode (from Ref. 21c).
.3. Layered LixMn02 and Similar Cathodes

A LiMn02 compound with the same layered structure as LiNi02 and
o02 would give a theoretical capacity of 285 mAh/g based on the
4
+/Mn + couple. However, LiMn02 is not thermodynamically stable as the
ed structure, but as the orthorhombic phase, o-LiMn02, of space group
m. Layered LiMn02 can be prepared by soft chemistry routes or by
ng stabilizing ions during the thermal synthesis. This material does not
the a.-NaFe02 structure of LiCo02 (hexagonal symmetry), but rather a
oclinic symmetry with space group C2/m. This occurs because the J-T t
3
brought about by the Mn + ion (tf with a single electron on the eg
) causes a cooperative distortion of the Mn06 octahedra.
A number of syntheses have been reported to prepare LiMn02 and
lar materials. The work done until 2001 has been reviewed by
139
undsen and Paulsen.
According to Ceder and Mishra, the orthorhombic structure is more
e than the layered one due to the strong antiferromagnetic interactions
3 140
een Mn + ions and the ensuing localization of the electronic states.
3
if one could partly substitute Mn + with non-magnetic ions, the layered
ture could be stabilized. Following this suggestions, layered LiMn02
een prepared by ceramic synthesis by slightly doping the material with
139 141
such as AI or Cr. ' Doping levels of 3-5% allow one to obtain
n02 with enhanced cyclability in the voltage range 4.4-2.0 V. Both o
n02 and layered LiMn02 transform when cycled into a more stable
Chapter 11
el-like structure, through a minor cation rearrangement, as all of these

pounds have the same cubic close packed sublattice of oxygen ions. Al d
layered LiMn02 is no exception to this rule, as shown by the
acteristic two-plateau discharge curve. Cr-doped LiMn02, however,
not convert into a spinel, although showing some phase transitions.
Capacities as high as 160-190 mAh/g could be obtained at 55°C at the
of 30 mNg, with Cr giving higher capacities than Al. The reason why
oes not favour transformation into spinel lies in its strong octahedral site
139
lization energy. Whether or not a spinel is formed is important, as the
drop between the two plateaux would prevent applications in practical
ries.
39 68 69
We have already mentioned the relevant work of Jang, Cho, · and
40
hio on a-LiMn02, which also has high and stabilized capacities. a
n02 is not a true layer structure as it is formed by alternate corrugated
rs of Li and Mn (see Figure 11.3c).
A truly layered LiMn02 can be obtained 21at low temperature by ion
ange from NaMn02 with LiBr in hexanol. " This material however
not cycle well and subsequent work by Bruce et al. was devoted to
pounds partially doped with Li, Ni or Co. These have the a-NaFe02
ture of LiCo02 (space group R 3m) with Li in the 3b sites, Mn and the
ng ion in the 3a sites and 0 in the 6c sites. The performance of these
pounds of general formula LixMn1 _yMy02 depends upon the starting Na
pound used, degree of doping and ion exchange procedure.21(b,c),I 42·143 It
been found that a Na compound of formula Nao. 518Mn0.908Coo.o2502, ion
anged with Li by refluxing in ethanol at 80°C, cycles fairly well. This is
buted to the presence of Na+ and transition metal vacancies in the final
pound, which facilitates the relief20 of strains induced by the J-T effect in
cles composed of nanodomains. When the ion exchange is carried out
xanol at 150°C, the amount of defects is quite limited and the structure
ss prone to accommodate the J-T effect, so that the electrochemical
vior is poorer. 143 In Figure 11.8, the cyclability of layered
n0.975Li0.02502, prepared
21 by Bruce et al. using low-temperature ion
ange, is reported. b
By using different Na precursors, 144 both layered and 19 non-layered Li
atives may be obtained. Doeff et al., and Bruce et al. starting from
44Mn02 with a tunnel structure, have obtained by ion exchange the
sponding Li compounds which could insert u+ with capacities of 100-
mAh/g. Jeong and Manthiram have shown that this Na compound, h
more correctly should be written as Nao.5Mn02+lh is obtained with a ed
structure if fired below 500°C and with a tunnel structure above C.
The layered ion-exchanged samples have a better cycling life than the
el ones (225 145mAh/g of initial capacity vs. 170 mAh/g) and better
city retention.
Ohzuku's group has thoroughly

146 148 investigated layered Li[Ni 112Mnl!2]02
ned by ceramic synthesis. - Its XRD is closely related to that of
02 but both the structure and the electrochemical behavior are different.
material gives a stable capacity of -200 mAh/g when cycled at low rate
een 2.5 and 4.5 V. The synthesis of this material is critical and requires,
rticular, the use of a double Mn-Ni hydroxide with an exact 1:1 ratio for
148
wo ions. b An improved behavior, This with enhanced
material rate
has capability, was
the peculiar
n by Li[Co 113Ni 113Mn c
148
113]02.
acteristic of showing a zero-volume change up to a deintercalation of

Li+/mol.
Dahn et al. have devoted particular attention to doped layered LiMn02
ng the 02 structure instead of the 03 typical of LiCoOiLiNi02 and
149 152
d LiMn02. - A layered 02 structure consists of 2 Mn02 sheets
sed in the unit cell, instead of 3 for 03. We have already pointed out
he 03 layered structure of monoclinic LiMn02 is easily converted into
of the lithiated spinel. The 02 structure, however, is quite different.
P2-type sodium manganese oxides, 02 layered lithium oxides may be
ned which do not convert into spinel, as this would require a
angement of the oxygen lattice - a very unlikely event at room
erature. Several 02 phases have been prepared as Li213(Liv6Mn5,6]02,
149 150 They show specific
Coo.l5Mn0.85]02, and Li2,3
[Ni113Mn213]02.
city of 150-180 mAh/g at low rates (typically C/40).
Stabilization of the 03 layered structure of LiMn02 has been
pted with the use of solid solutions between LhMn03 and LiM02
Cr, Ni, Co). Dahn et al. have prepared materials of general formula
xLi<113_2xi3)Mn<213_x/3)]02, that may be thought of as derived from
4
113Mn213]02, i.e. LhMn03, by partial substitution
4 of Lt and Mn + with
while maintaining the remaining Mn as Mn +. These compounds have
03 structure of LiCo02 when synthesised at temperatures above
C.153<•-dl Charging them up to 4.8 V, Ne+ is first oxidized to Ni 4+ and,
oxygen is lost while more u+ is extracted in a plateau between 4.5 and
. These oxygen-deficient cathodes, with .x=1/4, 1/3, 5/12, can reversibly
calate Lt with capacities as high as 225 mAh/g to 2V. Excellent rate
bilities are also reported. According to the authors, considerations of
rmance, safety and cost candidate these materials as possible substitutes
LiCo02. However, from the data on cyclability, a loss of 0.25
fg·cycle may be calculated. This value is too high for practical use and s
to be improved.
The same solid solution between LhMn03 and LiNi02 has been used
un et al. to prepare a material with formula Li[Li0. 13Ni0.30Mn057]02,
h can provide a stable specific capacity of 190 mAh/g when cycled
153
een 2.0 and 4.6 V. e A further improvement can be obtained if AI is
153
to partially replace Ni and Mn, as in Li[Li0.15NixAlo.ss-2xMn0.3+xl02. f
Chapter 11
The solid solution between LizMn03 and LiCr02 has also received
h attention and, in particular, Li[Li0.2Cr0.4Mn0.4]02 has proved
139
ctive. Just like for the other solid solutions, the doping transition metal
3 6
e electrochemically active species (Cr + 7Cr + in this case), while Li
Mn form an inert matrix.
5.4. 3 V Cathodes
.4.1. Manganese Oxides
During the early 90's, a cylindrical rechargeable battery with a Li

e, a Li0.3Mn02 cathode and a solution based on dioxolane was
154
mercialized. This battery, whose safety resulted from the
merisation of the solvent at 125°C (thermal shutdown), could deliver
cycles but only at low charging rates (<C/9). At higher rates, a failure
hanism based on the reaction of small Li grains with the electrolyte was
155
rved.
In spite of this failure (and of the previous ones), the development of a
node rechargeable battery remains a realistic goal, e.g., through the
156
zation of a glassy film to protect the anode. Also in view of this
pective, the research on 3 V cathodes has been actively ongoing,
cially for Mn-based oxides. Both layered (e.g., buserite and birnessite)
tunnel structures of Mn02 (e.g., ramsdellite/rutile, a-Mn02, hollandite,
157 161
rokite), have been reconsidered. - These compounds are
acterized by a high initial capacity which rapidly decreases in the next
cycles, but recently some reports on the stabilization of the capacity
appeared. For instance, Pranger et al. have found that a sol-gel Co
159
d birnessite can deliver a stable capacity of 177 mAh/g in 40 cycles.
son and Thackeray have reported that a-Mn02 stabilised by doping with
and treated with NH3 can deliver 150-220 mAh/g (depending on the
160
.
More interest has been devoted to lithiated Mn oxides. Yoshio et al.
tried to resolve the uncertainty linked to the actual composition of what
onventionally called CDMO (composite dimensional manganese
162
e). A single orthorhombic phase, Li0.33Mn02, can be obtained by
ng a LiNOiMn02 mixture (Li/Mn ratio =113) at a temperature lower
350°C for less than 30 hours. Higher Li/Mn ratios, temperatures and
s favor the formation of Mn spinel. The latter was indeed observed by
163
keray et al. Li0.33Mn02 obtained under the conditions described above
ers a stable capacity of 180 mAh/g. However, an even better
rmance is reported by Yoshio et al. with orthorhombic Li 0.5Mn02
ned by lithiating Mn02 with Lil (Li/Mn ratio =1) at 150-200°C. The
ors report capacities of -250 mAh/g while cycling at 0.4 rnNcm2 in the
.0 V range. 164
Recently, Li-rich spinels (e.g., Li 4Mn50u) and 0-rich spinels have
reconsidered as 3 V cathodes. The former has been examined by
thiram (see Section 11.2),u while the latter, with a typical composition
n 204.54, has been shown by Kilroy et al. to cycle with capacities above
165
mAh/g.
Surprisingly stable capacities have been reported for the Mn spinel in
V range by Goodenough et al. These authors have prepared a nearly
hiometric spinel by a sol-gel synthesis with glycolic acid, producing a
137 2
rial with an average particle size of 0.8 f-Lm. Cycling at 1 rnNcm ,
cities of -110 mAh/g have been obtained in spite of the evident two e
behaviour testifying to a J-T effect. The authors, although unable to
ide a sound explanation of this unusual behaviour, tended to attribute it
e small particle size. Indeed, the same authors made experiments on
milled spinel samples obtained by both sol-gel and thermal syntheses
found even higher and stable 3 V capacities. In this work, also
ioned in Section 11.3, the conclusion was reached that the J-T distortion
d be accommodated by strained, nanosized particles.37 Excellent cycling
vior was also reported at 60°C and 80°C.
4.2. Cr and V Oxides
Vanadium oxides, e.g., V205, V6013 and LiV308, have long been
tigated as cathodes for 3 V batteries with aLi anode.
V205 is very amenable to sol-gel processing by different routes. Using
ell known ion-exchange process that yields a precursor HV03 solution
self-assembles into the V205 hydrogel, a variety of materials of different
sity have been synthesized in form of films, powders and free-standing
branes. In the gel stage, the V205 solid network is bicontinuous with
connected pores filled with water. Depending on the way the liquid
nt is removed (drying process) any of the following three types of solid,
phous vanadium pentoxide can be produced:
1. Xerogel (XRG): nominal surface area= 1-10m2/g.
2. Aerogel (ARG): nominal surface area= 350-450 m2/g.
3. Aerogel-like (ARG-like): nominal surface area= 100-300 m2/g. 166
All materials are reported to have high specific capacities (see Table
) in near-equilibrium or very slow insertion/release tests.167 However, as
ate increased the capacity of xerogel decreased to 380 mAh/g (at C/20),
mAh/g (at C/5), and 200 mAh/g (at 1C). The dependence of the capacity
e rate indicates that diffusion limitations are key factors determining
the
rmance of the xerogel material.
Chapter II
In an effort to further improve the intercalation kinetics and rate

ability of these materials, there has been some investigation of the doping
he precursor hydrogel. Reaction of the hydrogel V205 with metallic Cu,
Al, Ni, Mg, and Zn has been found to enhance the electronic
168
ductivity by up to three orders of magnitude. X-Ray absorption
troscopy studies have shown the formation of metallic copper
169
oclusters upon lithium insertion in Cu0.1V205 xerogel. The process is
pletely reversible at the atomic level and the starting compound is
nerated upon releasing the inserted lithium. Cuo.Nz05 aerogel-like based
trodes in coin cell batteries were found to have capacities of 170 mAh/g
ive material) at 1 C rate for more than 400 cycles.
17
°
Clio. I V205 xerogel
wed similar capacity-cycle life behavior. In addition the latter material
e 11.2. Specific capacities of amorphous vanadium pentoxides prepared by sol-gel

sses. For comparison, the specific capacity of crystalline V2 05 is also indicated. Data
ned in low rate discharges (C/100).
Material Typical Specific Capacity
VzOsXRG 560mAh/g
Cu-doped V z05 XRG 300 mAh/g
VzOsARG 560 - 780" mAh/g
Vz05 ARG-Like 560 mAh/g
c-V205 148 mAh/g
a. Obtained by chemicallithiation
wed very good performance in pulsed insertion test.171

V205 is the cathode most used when investigating organic-inorganic
id materials. 26 In its layered structure it can even intercalate large
ecules as polyaniline. 172 However, both the electrochemical and thermal
ility of these composites need to be improved.
LiV308 is a pseudo-layered material formed by V308- puckered sheets
ed by non-mobile u+ ions residing in octahedral sites.173 It can be
ared by both solution and ceramic techniques, and, in both cases,
mized syntheses lead to high (well above 200 mAh/g) and reversible
cities.17 4-m Particularly effective is the introduction of H2 0 and C02
ecules between the layers, which allows their expansion and, so, an
eased mobility and enhanced distribution of Li+.174 A sample prepared in
manner has an initial capacity approaching 300 mAh/g (at C/3) and can
deliver 260 mAh/g after 200 cycles.
Interesting results have also been obtained with the isostructural

178 179 80
308 ' and Na1.xKxV308 (x<0.4)/ while KV308 gives poor results.
181
latter, however, is not the only composition in the phase diagram
een V205 and KV03. One of the four possible compositions, i.e.
181
013, also gives rechargeable capacities in excess of 200 mAh!g.
K-containing vanadates have also been prepared by hydrothermal
182
ions between KMn04 and hydrated VOS04. Depending upon the
ant ratios and the mixture pH, two different layered materials can be
ined, i.e. Ko.16Mlloo4V204.94·14H20 and Ko.44Vz04.96· It seems that at a
value of 2.3 both phases coexist and provide the optimum reversible
city of 190-200 mAh!g. Another interesting vanadate is the Al-doped
183
V6017, providing more than 250 mAh!g at low rates (C/10).
All the above vanadium oxides do not contain removable Li+ and can
be used in cells with aLi anode. However, chemicallithiation has been
rted for 177both
178 LiV308 and NaV308 to obtain Li4V308 and LhNaV308,
ctively. • Li4V308 was obtained by using LizS as a lithiating agent,
eas LhNaV308 was obtained with buthyllithium. Li can be reversibly
cted from both compounds, enabling them to be used in lithium-ion
. For instance, with LhNaV308, 178 capacities of 210-230 mAh/g could be
ned on cycling at 25-50 mA/g.
In comparison to V oxides, the search on Cr oxides has been more
ed. Starting from Cr03, Arora et al. could prepare, by decomposition in
utoclave under 02 pressure, CrOx and, by lithiating the latter with Lil,
184
Ox. a Both compounds can give high capacities but only at low rates.
ever, a moderate doping with Co (typically 0.2 atoms/mol) is effective
abilizing both structures, so enhancing the rate capability and allowing
184
tain capacities of 230-250 mAhlg. b
.5. A Special Case: LiFeP04

Lithium iron phosphate, LiFeP04, with the phospho-olivine structure
ure 11.3d) is emerging as a promising cathode for lithium and lithium
atteries. Its high theoretical capacity (170 mAhlg), flat voltage at -3.4
w cost and safety (low toxicity and environmental impact) make this
rial very appealing.
185
The interest on this material was raised by Padhi et al. that showed
ossibility of chemically remove lithium from the structure of LiFeP04
leaving a new phase, FeP04, isostructural with heterosite,
sMn0.35P04. The lithium electrochemical extraction from LiFeP04
eeds via a hi-phasic process in which the final FeP04 structure is
ned through minimum displacements of the ordered phospho-olivine
ework. Nevertheless, the lithium electrochemical extraction/insertion in
Chapter II
material is limited to 0.6 Li+/mol even if driven at a very low specific

nt (2.1 mA/g, i.e., C/81 rate).
186
Ravet et a/. renewed the interest on LiFeP04 by improving its
rochemical performance via a new synthetic route. They added sucrose e
precursors to act as a source of carbon to be formed during the thermal
hesis (700°C). In such a way they were able to produce LiFeP04
cles covered with a carbon layer. This material, cycled at 80°C in
mer electrolyte cells, delivered about the theoretical capacity even at
current rates.
Andersson et al. have investigated the electrochemical lithium release
nsertion processes in thermally synthesised, additive-free LiFeP04 by in
187
X-ray diffractometry and Mossbauer spectroscopy. They found that
5% of LiFeP04 remains unaltered depending on the material particle
and surface area. In a following work, they showed that the amount of
188
um involved in the first extraction is larger than in the second.
hermore, they found a direct correlation between the temperature and the
cted amount of lithium. On these bases they proposed that lithium
sion within each particle is the limiting step.
Following this finding, research groups at NTT and Sony have
stigated the effect of synthesis temperature on the electrochemical
erties of LiFeP04.189• 190 They found that the reduction of the synthesis
erature decreased the material particle size (ranging from 0.2 to 30 j.lm),
2
ased the specific surface area (about 10 m /g) and enhanced the
rochemical performance (160 mAh/g for a material synthesised at
C).
Similar results have been obtained by Prosini et al. that have
192
191
hesised LiFeP04 in the presence of high surface area carbon. The
•
ion of carbon was found to reduce the particle size to less than 10 j.lm.
authors report an enhancement of the electrochemical performance in s
of high rate delivered capacity while the theoretical capacity (170 lg)
was obtained in tests performed at 80°C and C/10 rate. According to
authors, the rate limitations in LiFeP04 could be due to the poor
m diffusion or the poor electronic conductivity of the material. They
osed that the addition of carbon has a beneficial effect because it
ces the particle size and increases the overall electronic conductivity of
aterial.
The effect of the particle size reduction (i.e., the shortening of the
phase path for ions and electrons) has been also evidenced by
193 194
denough on hydrothermally synthesised LiFeP04• Croce et al. have
a different approach that consisted of the dispersion of a very fine
llic powder (copper and silver) in the LiFeP04 precursors following a
168 169
ess earlier reported. ' The authors stated that the metallic particles
s nucleation sites (thus reducing the active material particle size) and
nce the electronic conductivity, thus allowing the improvement of the
a)
.-.
:::1 3.8
..;
;o.
3.4
:..!..
c 3.0
!2.6
0 0.2 0.4 0.6 0.8 1.0
xLi/UFeP04
b) 160
0.8
-
140 'eG
0...
0.6
r - 5C
- 120
.... 100 ..e....
lo<
....J 80 'Cl
0.4 •
60
•c.
t,)
0.2 40
20
0 0
0 200 400 600 800
Cycle number
re 11.9. a) First cycle of a cell with a LiFeP0 4 cathode (C/10); b) Prolonged

ng at 5 C (from Ref. 196).
rmance. Prosini et al. reported the synthesis of nanocrystalline LiFeP04

eating its amorphous form, the latter being prepared by chemical
195
tion of FeP04• The nanocrystalline material had a globular structure
a grain size of about 100-150 nm and delivered a capacity of 162
/g at 17 mNg (room temperature).
Following this reasoning, Nazar et al. have synthesised a LiFePOJC
posite with active material particles in the scale of nanometers, via a sol
196
ynthesis. Such a composite was able to deliver about 90% of the
retical capacity at C/2 rate and ambient temperature. The reported cycle f
this material is also extremely good with about 90% of the theoretical
city delivered at the 100th cycle at C/5 rate. Even more impressive is the
t of the combined rate performance-cycle life test performed at 5C rate
hich the material was able to deliver 70% of the theoretical capacity
ut 110 mAh/g) at the 800th cycle (see Figure 11.9). These results
Chapter 11
rly confirm that the reduction of the material particle size and the
ovement of the electronic conductivity lead to a better electrochemical
ormance of LiFeP04.
5.6. Sulphur and Organic or Inorganic Sulphides
Integrated circuits for microelectronics which can be powered by 2 V

ces, instead of 3 V, have been developed. This makes it possible to use
rochemical couples thus far neglected, provided that they can deliver
capacities and are cheap and safe.
Sulphur can provide a specific capacity of 1675 mAhlg and an energy
600 Wh/kg in a Li/S cell. However, realization of an efficient cathode
been rather elusive thus far due to the low conductivity and high
bility of sulphur. In recent years, efforts to exploit its potentialities have
renewed and interesting results reported.
Sulphur is reduced in two steps. During the first, at -2.4 V, several Li
sulphides are formed, e.g., LhS4, LhS8, etc., while in the second step, at
2.0 V, Li2S is formed.197 The sulphides are soluble and can diffuse to the
ode where they are reduced. The reduced forms can diffuse back to the
197
ode and reoxidize. A proper cathode formulation, including an
ronic conductor and a binder, affords capacities as high as 1364 mAhlg
198
om temperature (82% utilization). A further improvement is reported
the use of filamentary nickel. By ball-milling the Ni with S, NiS is
ed and the so-modified electrode can retain 99% of its initial capacity
199
110 cycles.
The sulphur electrode has stimulated considerable interest at the
strial level. Prototypes with a relatively high cathode loading, >4
2 2
lcm vs. 2.5-3 mAh/cm for commercial lithium-ion batteries, have been
zed and provide about 370 mAh/g on extended cycling. It has been
ested that a stable passivation layer on Li forms, resulting in gravimetric
volumetric energy densities, in liquid and polymeric electrolytes,
200
parable with those of commercial batteries.
In a recent development, the S cathode is coupled with a Li electrode
ined by deposition and separated by a glass electrolyte film from the
201
rolyte. The interfacial resistance of Li/glass/electrolyte is much lower
more stable than that of unprotected Li and seems to offer the potential
ractical secondary batteries with a Li anode. The combination of this
cted Li and S (LhSx) is rather promising. In a flat-type cell, this
stry has afforded specific energies of 420 Whlkg and 520 Wh/1,
201
wing the developers to foresee viable marketing possibilities. a
Organo-sulphur cathode materials have also been investigated, e.g.,
(carbon sulphides) obtainable in various forms according to the
202
hesis procedures. These compounds have been tested in prototype
ries under conditions simulating those typical of cellular phones. A
(phenylenedithiazol) has also been reported, whose electrochemical203
ity is based on an S-S bond and an unsaturated dithiazolium ring. Its
llent first-discharge capacity (420 mAh/g at low rate) was only partly
rsible, perhaps due to solubility problems.
Peled's group has investigated in detail the Li/FeS2 couple in a
204 205
posite polymeric electrolyte based on PE0. ' Thin-cathode batteries,
harging at the mean potential of 1.7 V (at 135°C), have a reversible
city of 625 mAh/g (2.8 Li/FeS2) and a projected specific energy of 170
kg in a bipolar configuration. 500 cycles with 100% DOD have been
rted. The Li/FeS2 couple has been proposed as a good candidate for EV
cations. The same couple has also been cycled at room temperature in a
206
meric electrolyte based on PVdF.
.7. High Voltage Materials
Another way to increase the energy output of lithium batteries is by

sing on materials having redox reactions above -4.3 V (a commercial
n battery based on LiCo02 has an average potential of 3.65 V on
arge). Heavily doped Mn spinels, LiMxMn2_x04 (M=Fe,Cr,Ni,Co,Cu
heir mixtures), have been identified as the compounds which may offer
possibility. All of these materials retain the spinel space group Fd3m and
advantage of the redox reaction of the doping ion. West et al. have
ined several substituted spinels and found that some of the Co
tituted ones, e.g., Li2Co0.sMn 4 3.208 and LiCoMn04, give the best
207 208 3
ts. · In them, the Co +/Co + couple creates a discharge plateau at
3 4
t 5 V, in addition to the 4 V plateau of Mn +/Mn +. Initial capacities of
compounds are 135 and 105 mAh/g, 2 respectively, with moderate
ements after 35 cycles at 0.5 mA/cm .
Ohzuku's group has examined the low-rate first cycle of the spinels
112Mn213]04 (M=Ni,Co,Cr,Fe,Cu) and found capacities of 120-125
/g for the Fe and Ni 2sginel,
9
with the latter providing most of its capacity
single 4.5 V plateau. The promising behaviour of Ni-spinels has been
rmed in subsequent reports. Hernan 210 et al. have examined spinels of
position LiM03Mnu04 (M=Fe,Co,Ni). XPS data have shown that Ni
a divalent state. Its oxidation to the tetravalent state results in high
cities (around 120 mAh/g at 4.5 V). Thackeray et al. have studied
211
ls of composition LiNixCUo.s-xMn5.04 (O:Sx::;0.5). With increasing Ni
entration, the capacity increases (to a maximum of 117 mAh/g) while
pper-plateau potential decreases (from 4.9 V to 4.6 V). In the last two
ences the cathodes were cycled for about 50 cycles at 0.25 mA/cm2 . We
calculated for these tests capacity losses of 0.4 mAhfg·cycle
Chapter II
3%/cycle), a very high value in comparison with that shown by the

entional 4 V cathodes (0.01-0.05%/cycle). A better capacity retention
212
recently been reported for sol-gel spinel LiNi0.5Mn1.504• • 213 In
cular, in Ref. 214 the synthesis was optimized by first firing the gel at
C and annealing at 600°C to reduce oxygen loss and improve Ni
ent. The material so obtained has an initial capacity of 127 mAh/g at 0.5
2
cm , which is only reduced to 124 mAh/g after 60 cycles. This is
ained by the low surface area of the annealed spinel (2 m2/g), so that
rolyte (LiPF6-PC) decomposition is minimized.
Recently, a remarkable improvement has been reeorted for the
ability of LiNi0.5Mn1.504• In a first report, Yoshio et al.2 4 have reported
this material, prepared by a sol-gel technique and coated with nanosized
, could be cycled at C/3 and 55°C for 50 cycles with a stable capacity of
mAh/g. ZnO is proposed to protect the spinel from the electrolyte and,
articular, to act as a HF scavenger. In a subsequent report, Yoshio et
5
have described an improved synthesis technique for LiNi0.5Mn1.504, the
alled composite carbonate process, allowing to obtain a more
ogeneous material formed by agglomerated nanoparticles of 50-100 nm.
result, it shows an excellent performance at high temperature without
nO-coating: 96% capacity retention in 50 cycles.
. CONCLUDING REMARKS
The research on rechargeable cathode materials has been very fruitful
e the introduction of the first commercial Li-ion batteries. Here, we shall
o summarize the most important advances.
LiCo02, the cathode still used in mass production, has been optimized
41
s preparation, especially as far as particle size is concerned (-20 -tm).
esting reports on the coating of this cathode to improve the thermal
lity and to allow the use at higher voltages (with higher capacities) have
ared.59• 63 As mentioned in Section 11.5, the price of this material has
decreasing, consolidating its predominant position in batteries for
ble electronics. In 2002, the price of LiCo02 was about twice that of
1_xCox02 and about four times that of LiMn 204, the two major
idates for future developments. However, if the price is related to the
gy (Wh) of a given battery, a LiCoOz-based battery is only 30% more
nsive than batteries with cathodes containing Ni or Mn.95b This can be
ated in small cells, but not in the large ones intended for EV and
strial applications. Indeed, in view of the fact that for such large
ries mass production processes are not as well established as for the
l ones, it can be foreseen that only Ni- and/or Mn-based cathodes will
sed.
As already mentioned, small Li-ion batteries based on Mn spinel are

commercialized, although to a limited market share. The lower spinel
city with respect to LiCo02 does not allow to foresee a higher energy.
limited cyclability at hi.fh temperature has recently been improved,
ks to the use of additives. The safety has also been improved but needs
er efforts. A remarkable advancement is represented by the cyclability
37 137
e previously prohibitive 3 V range. •
Perhaps, the most rapidly emerging material is now LiNit-x
Aly02.95 It contains more than 75% of the relatively cheap Ni, can be
ared on a large scale and affords high and stable capacities. LiFeP0 4
appears of great interest, as shown in Section 11.5.5. Close attention are
receiving layered Mn-containing materials, such as Li[Li 0.2Cr0.4Mn0.4]02
3 6 139
h the Cr +/Cr + couple) and Li[NixLi( 113.2xf3)Mn<213_x,3)]02 (with the
4 153
/Ni + couple) •
Broussely has calculated that an optimized Li-ion cell based on a C
e of 350 Ahlkg and a LiCo02 cathode of 140 Ahlkg would give 166
95
kg and 412 Wh/1. b These values are not far from those attainable in the
nt generation of Li-ion batteries. They may be taken as a benchmark for
ible new batteries based on different cathodes, especially those having r
or higher voltages than C/LiCo02 (3.6 V). So, a new cathode with an
age 2.5 V (usable in the newly developed integrated circuits) should
a capacity of at least 250 Ah/Kg and a density higher than 6.5 g/cm3 to
erform C/LiCo02• Some new cathodes mentioned in Section 11.5 have
high capacities but none are so dense.
On the other hand, a cathode discharging at 4.25 V should be better
3
115 Ahlkg and 5 g/cm • Capacities well in excess of the above value
been reported for heavily doped spinels (up to 140 mAhlg at >4.5
, while their density should be similar to that of pure spinel (-4.2
m\ Electrolyte decomposition at the cathode could negatively affects the
mance, but this problem can be tackled with improved syntheses and
214
oating techniques. '215 Therefore, the new generation of high voltage
des looks quite promising.
In the meantime, the presently available cathode materials perform
enough if one notes that, in only 10 years, Li-ion batteries have gained a
share of the market for portable devices.
FERENCES
. Nagaura, K. Tozawa, Prog. Batt. Solar Cells 9 (1990) 209.
. Numata, C. Amerniya, J. Iriyama, T. Miura, M. Shirakata, NEC R§D 41 (2000) 8.
M. Tsunoda, Y. Oshima, M. Yoshinaga, T. Shirasu, NEC R§D 41 (2000) 13.
a) C. Amerniya, J. Kurihara, M. Yonezawa, NEC R§D 42 (2001) 241; (b) S. Narukawa,
. Nakane, N. lrnachi, S. Fukuoka, M. Yamasaki, II'" IMLB, Monterey, CA, USA, June
Chapter 11
2002, Abs. 2.
Y.-S. Lee, Y-K. Sun, K.-S. Nahm, Solid State lonics 109 (1998) 285.
H.-B. Park, J. Kim, C.-W. Lee, J. Power Sources 92 (2001) 124.
C. Julien, M.A. Camacho-Lopez, T. Mohan, S. Chitra, P. Kalyani, S. Gopukumar, Solid
State 1onics 135 (2000) 241.
S. Chitra, P. Kalyani, T. Mohan, R. Gangadharan, B. Yebka, S. Castro-Garcia, M.
Massot, C. Julien, M. Eddrief, J. Electroceram. 3:4 (1999) 433.
W. Yang, G. Zhang, J. Xie. L. Yang, Q. Liu, J. Power Sources 81-82 (1999) 412.
Y.-S. Han, H.G. Kim, J. Power Sources 88 (2000) 161.
A. Manthiram, J. Kim, Chern. Mater. 10 (1998) 2895.
S. Bach, J.P. Pereira-Ramos, N. Baffier, Solid State lonics 80 (1995) 151.
Q. Feng, H. Kanoh, K. Ooi, M. Tani, Y. Nakacho, J. Electrochem. Soc. 141 (1994)
L135.
J. Kim, A. Manthiram, J. Electrochem. Soc. 145 (1998) L53.
A. Manthiram, A. Dananjay, Y. Zhu, Chern. Mater. 6 (1994) 1601.
C. Tsang, A. Manthiram, J. Electrochem. Soc. 144 (1997) 520.
. Maruta, H. Yasuda, M. Yamachi, J. Power Sources 90 (2000) 89.
.N. Reimers, E.W. Fuller, E. Rossen, J.R. Dahn, J. Electrochem. Soc. 140 (1993) 3396.
A.R. Armstrong, H. Huang, R.A. Jennings, P. G. Bruce, J. Mater. Chern. 8 (1998) 255.
A.D. Robertson, A.R. Armstrong, P.G. Bruce, Chern. Mater. 13 (2001) 2380.
a) A.R. Armstrong , P. G. Bruce, Nature 381 (1996) 499; (b) A.D. Robertson, A.R.
Armstrong, A.J. Peterson, M.J. Duncan, P.G. Bruce, 2001 ECS and ISE Joint
nternational Meeting, San Francisco, CA, USA, September 2001; (c) A.D. Robertson,
A.R. Armstrong, A.J. Peterson, M.J. Duncan, P.G. Bruce, /11h IMLB, Monterey, CA,
USA, June 2002, Abs. 74.
M.S. Whittingham, J.-D. Guo, R. Chen, T. Chirayil, G. Janauer, P. Zavalij, Solid State
onics 75 (1995) 257.
M.S. Whittingham, S. Yang, K. Ngala, P. Zavalij, 2001 ECS and ISE Joint International
Meeting, San Francisco, CA, USA, September 2001.
.K. Sharma, G.J. Moore, F. Zhang, P. Zavalij, M.S. Whittingham, Electrochem. Solid-
tate Lett. 2 (1999) 494.
R. Chen, M.S. Whittingham, J. Electrochem. Soc. 144 (1997) L64.
M.S. Whittingham, Solid State Ionics 134 (2000) 169.
a)Y. Li, C. Wan, Y. Wu, C. Jiang, Y. Zhu, J. Power Sources 85 (2000) 294; (b) K.
1
Konstantinov, G.X. Wang, J. Yao, H.K. Liu, S.X. Dou, ll h IMLB, Monterey, CA, USA,
une 2002, Abs. 100.
a) S.-T. Myung, H.-T. Chung, J. Power Sources 84 (1999) 32; (b) S.-T. Myung, S.
Komaba, N. Hirosaki, N. Kumagai, K. Arai, T. Kodama, Y. Terada, I. Nakai, ll 1 h 1MLB,
Monterey, CA, USA, June 2002, Abs. 90.
Y.-M. Chiang, Y.-1. Jang, H. Wang, B. Huang, D. Sadoway, P. Ye, J. Electrochem. Soc.
145 (1998) 887.
H. Yan, X. Huang, H. Li, L. Chen, Solid State lonics 113 (1998) 11.
H. Yan, X. Huang, L. Chen, J. Power Sources 81-82 (1999) 647.
.S. Whitfield and I.J. Davidson, J. Electrochem. Soc. 147 (2000) 4476.
M.N. Obrovac, 0. Mao, J.R. Dahn, Solid State 1onics 112 (1998) 9.
N.V. Kosova, N.F. Uvarov, E.T. Devyatkina, E.G. Avvakumov, Solid State /onics 135
2000) 107.
a) S. Soiron, A. Rougier, L. Aymard, J.-M. Tarascon, J. Power Sources 97-98 (2001)
02; (b) W.T. Jeong, J.H. Joo, K.S. Lee, 1 J'h IMLB, Monterey, CA, USA, June 2002,
bs. 68.
. Singhal, G. Skandan, G. Amatucci and N. Pereira, Electrochemical Society
roceedings Volume 2000-36, A.R. Landgrebe and R.J. Klinger, Eds., p. 244, 2001.
.-H. Kang, J.B. Goodenough, L.K. Rabenberg, Chern. Mater. 13 (2001) 1758.
S.-H. Kang, J.B. Goodenough, L.K. Rabenberg, Electrochern. Solid-State Lett. 4 (2001)
A49.
Y.-I. Jang, B. Huang, H. Wang, D. R. Sadoway, Y.-M. Chiang, J. Electrochern. Soc. 146
1999) 3217.
Y. S. Lee, M. Yoshio, Electrochern. Solid-State Lett. 4 (2001) A166.
Y. Nishi, J. Power Sources 100 (2001) 101.
. Cho, G. Kim, H.S. Lim, J. Electrochern. Soc. 146 (1999) 3571.
. Cho, B. Park, J. Power Sources 92 (2001) 35.
. Ying, C. Wan, C. Jiang, Y. Li, J. Power Sources 99 (2001) 78.
H. Huang, C.H. Chen, R.C. Perego, E.M. Kelder, L. Chen, J. Schoonman, W.J.
Weydanz, D.W. Nielsen, Solid State Ionics 127 (2000) 31.
G. Amatucci, C.N. Schmutz, A. Blyr, C. Sigala, A.S. Gozdz, D. Larcher, J.-M. Tarascon,
. Power Sources 69 (1997) 11.
. Cho, Solid State Ionics 138 (2001) 267.
V. Manev, T. Faulkner, J. Engel, in: Proceedings of the 1' 1 Hawaii Battery Conference,
anuary 1998.
H. Huang, C.A. Vincent, P.O. Bruce, J. Electrochern. Soc. 146 (1999) 481.
Y. Xia, N. Kumada, M. Yoshio, J. Power Sources 90 (2000) 135.
D. Aurbach, J. Power Sources 89 (2000) 206.
D. Aurbach, K. Gamolsky, B. Markovsky, G. Salitra, Y. Gofer, U. Heider, R. Oesten, M.
Schmidt, J. Electrochern. Soc. 147 (2000) 1322.
G. Amatucci, A. Du Pasquier, A. Blyr, T. Zheng, J.-M. Tarascon, Electrochirn. Acta 45
1999) 255.
T. Aoshima, K. Okahara, C. Kiyohara, K. Shizuka, J. Power Sources 97-98 (2001) 377. A.
Du Pasquier, A. Blyr, A. Cressent, C. Lenain, G. Amatucci, J.-M. Tarascon, J. Power
ources 81-82 (1999) 54.
C.H. Chen, J. Liu, K. Amine, J. Power Sources 96 (2001) 321.
A. Antonini, C. Bellitto, M. Pasquali, G. Pistoia, J. Electrochern. Soc. 145 (1998) 2726. A.
Du Pasquier, F. Orsini, A.S. Gozdz, J.-M. Tarascon, J. Power Sources 81-82 (1999)
07.
a) J. Cho, G.B. Kim, H.S. Lim, C.-S. Kim, S.-1. Yoo, Electrochern. Solid-State Lett. 2
1999) 607; (b) J. Cho, T.-J. Kim, Y.J. Kim, B. Park, Chern. Cornrn. 2001, 1074.
.-C. Park, Y.-M. Kim, S.-C. Han, S. Ahn, C.-H. Ku, J.-Y. Lee, J. Power Sources 107
2002) 42.
. Endo, T. Yasuda, A. Kita, K. Yamaura, K. Sekai, J. Electrochern. Soc. 147 (2000)
291.
. Endo, T. Yasuda, K. Yamaura, A. Kita, K. Sekai, J. Power Sources 93 (2001) 87.
a) J. Cho, Y. J. Kim, T.-J. Kim, B. Park, Angew. Chern. Int. Ed. 40 (2001) 3367; (b) Y.
. Park, personal communication.
a) J. Cho, Y. J. Kim, B. Park, Chern. Mater. 12 (2000) 3788; b) J. Cho, Y. J. Kim, B.
ark, J. Electrochern. Soc. 148 (2001) A1110.
. Cho, C.-S. Kim, S.-I. Yoo, Electrochern. Solid-State Lett. 3 (2000) 362.
. Chen, J. R. Dahn, Electrochern. Solid-State Lett., 5 (2002) A213.
a) G. Amatucci, J.-M. Tarascon, L.C. Klein, J. Electrochern. Soc. 143 (1996) 1114; (b)
.H. Lee, H.S. Kim, K.T. Lee, S.M. Oh, 1 I' 11 IMLB, Monterey, CA, USA, June 2002,
bs. 102; (c) B. Markovs1y, A. Rodkin, Y. S. Cohen, E. Levi, D. Aurbach, H.-J. Kim, M.
chmidt, U. Heider, 1 J'h IMLB, Monterey, CA, USA, June 2002, Abs. 327.
. Cho, Chern. Mater. 13 (2001) 4537.
. Cho, Y. J. Kim, T.-J. Kim, B. Park, Chern. Mater. 13 (2001) 18.
a) K. Mizushima, P.C. Jones, P.J. Wiseman, J.B. Goodenough, Mater. Res. Bull. 15
1980) 783; (b) M.M. Thackeray, "The Structural Stability of Transition Metal Oxide
nsertion Electrodes for Lithium Batteries", in Handbook of Battery Materials, J.O.
esenhard Ed., Wiley-VCH, 1999; (c) A.S. Andersson, J.O. Thomas, J. Power Sources
Chapter 11
97-98 (2001) 498.

J. McBreen, X.-Q. Yang, M. Balasubramanian, X. Sun, Electrochemical Society
Proceedings Volume 2000-36, A.R. Landgrebe and R.J. Klinger, Eds., p. 252, 2001.
W. Li, J.N. Reimers, J.R. Dahn, Solid State lonics 67 (1993) 123.
C. Pouillerie, L. Croguennec, Ph. Biensan, P. Willmann, C. Delmas, J. Electrochem. Soc.
147 (2000) 2061.
V. Bianchi, S. Bach, C. Belhomme, J. Farcy, J.P. Pereira-Ramos, D. Caurant, N. Baffier.
P. Willmann, Electrochim. Acta 46 (2001) 999.
C. Delmas, M. Menetrier, L. Croguennec, I. Saadoune, A. Rougier, C. Pouillerie, G.
Prado, M. Grune, L. Fournes, Electrochim. Acta 45 (1999) 243.
a)Y. Gao, M. V. Yakovleva, W.B. Ebner, Electrochem. Solid-State Lett. 1 (1998) 117;
b) M.V. Yakovleva, Y. Gao, ll 1 h IMLB, Monterey, CA, USA, June 2002, Abs. 209; (c)
1
. Gao, M.V. Yakovleva, 11 h IMLB, Monterey, CA, USA, June 2002, Abs. 315.
) Z. Chen, J.R. Dahn, 202"d ECS Meeting, Salt Lake City, Utah, USA, October 2002;
(b) A. Vander Van, M.K. Aydinol, G. Ceder, J. Electrochem. Soc. 145 (1998) 2149; (c)
X.Q. Yang, X. Sun, J. McBreen, Electrochem. Comm. 2 (2000) 100.
.-M. Tarascon, G. Vaughan, Y. Chabre, L. Seguin, M. Anne, P. Strobel, G. Amatucci, J.
olid State Chern. 147 (1999) 410.
. Venkatraman, A. Manthiram, 202"d ECS Meeting, Salt Lake City, Utah, USA,
October 2002
D. Carlier, M. Menetrier, C. Delmas, J. Mater. Chern. 11 (2001) 594.
M. Balasubramanian, X. Sun, X.Q. Yang, J. McBreen, J. Electrochem. Soc. 147 (2000)
2903.
K.-K. Lee, K.-B. Kim, J. Electrochem. Soc. 147 (2000) 1709.
G.X. Wang, J. Horvat, D.H. Bradhurst, H.K. Liu, S.X. Dou, J. Power Sources 85 (2000)
79.
R.V. Chebiam, F. Prado, A. Manthiram, J. Electrochem. Soc. 148 (2001) A49.
. Cho, H. Jung, Y. Park, G. Kim, H. Lim, J. Electrochem. Soc. 147 (2000) 15.
Y. Fujita, K. Amine, J. Maruta, H. Yasuda, J. Power Sources 68 (1997) 126.
M. Broussely, P. Biensan, B. Simon, Electrochim. Acta 45 (1999) 3.
C. Pouillerie, F. Perton, Ph. Biensan, J.P. Peres, M. Broussely, C. Delmas, J. Power
Sources 96 (2001) 293.
. Cho, Chern. Mater. 12 (2000) 3089.
C.-C. Chang, J. Y. Kim, P.N. Kumta, J. Power Sources 89 (2000) 56.
B.V.R. Chowdari, G.V. Subba Rao, S.Y. Chow, Solid State lonics 140 (2001) 55.
a) Z. Lu, D.D. MacNeil, J.R. Dahn, Electrochem. Solid-State Lett. 4 (2001) A200; (b)
M. Yoshio, H. Noguchi, J. ltoh, M. Okada, T. Mouri, J. Power Sources 90 (2000) 176.
H. Arai, M. Tsuda, Y. Sakurai, J. Power Sources 90 (2000) 76.
S. Madhavi, G.V. Subba Rao, B.V.R. Chowdari, S.F.Y. Li, J. Power Sources 93 (2001)
156.
a) J.S. Weaving, F. Coowar, D.A. Teagle, J. Cullen, V. Dass, P. Bindin, R. Green, W.J.
Macklin, J. Power Sources 97-98 (2001) 733; (b) M. Broussely, Lithium Battery
Discussion, Arcachon (France), May 2001; (c) M. Broussely, ll'h IMLB, Monterey, CA,
USA, June 2002, Abs. 415.
K. Kubo, S. Arai, S. Yamada, M. Kanda, J. Power Sources 81-82 (1999) 599.
T. Ohzuku, K. Nakura, T. Aoki, Electrochim. Acta, 45 (1999) 151.
S. H. Park, K.S. Park, Y. K. Sun, K. S. Nahm, Y.S. Lee, M. Yoshio, Electrochim. Acta
6 (2001) 1215.
. Kim, K. Amine, Electrochem. Comm. 3 (2001) 52.
A.R. Naghash, J. Li, Electrochim. Acta 46 (2001) 2293.
S. H. Park, K.S. Park, Y. K. Sun, K. S. Nahm, Y.S. Lee, M. Yoshio, Y.K. Lee, K.W.
Han, Electrochemical Society Proceedings Volume 2000-21, M. Doyle, E. Takeuchi and
K.M. Abraham, Eds., p. 45, 2001.
A. Yamada, M. Tanaka, K. Tanaka, K. Sekai, J. Power Sources, 81-82 (1999) 73.

C.-H. Shen, R. Gundakaram, R.-S. Liu, H.-S. Sheu, J. Chern. Soc. Dalton Trans. 2001,
7
a) D. Capsoni, M. Bini, G. Chiode!li, V. Massarotti, C. Azzoni, M.C. Mozzati, A.
Comin, Phys. Chern. Chern. Phys. 3 (2001) 2162; (b) G. Rousse, C. Masquelier, J.
Rodriguez-Carvajal, E. Elkaim, J.P. Lauriat, J.L. Martinez, Chern. Mater. 11 (1999)
629.
R. Basu, R. Seshadri, J. Mater. Chern. 10 (2000) 507.
Y. Xia, T. Sakai, T. Fujieda, X.Q. Yang, X. Sun, Z.F. Ma, J. McBreen, M. Yoshio, J.
A. Blyr, C. Sigala, G. Amatucci, D. Guyomard, Y. Chabre, J.-M. Tarascon, J.
T. Inoue, M. Sano, J. Electrochem. Soc. 145 (1998) 3704.
Y. Kogetsu, M. Kohno, T. Hatanaka, T. Saito, J. Yamaura, 2001 ECS and 1SE Joint
nternational Meeting, San Francisco, CA, USA, September 2001.
T. Saito, M. Machida, Y. Yamamoto, M. Nagamine, 2001 ECS and ISE Joint
N. Nishimura, T. Douzono, M. Kasai, H. Andou, Y. Muranaka, Y. Kozono, J. Power
ources, 81-82 (1999) 420.
D.H. Jang, Y.J. Shin, S.M. Oh, J. Electrochem.Soc. 143 (1996) 2204.
G. Pistoia, A. Antonini, R. Rosati, D. Zane, Electrochim. Acta, 41 (1996) 2683.
H. Huang, C.A. Vincent, P.G. Bruce, J. Electrochem. Soc. 146 (1999) 3649.
R. Premanand, A. Durairajan, B. Haran, R. White, B. Popov, J. Electrochem. Soc. 149
2002) A54.
a) J. Lee, J. K. Hong, D.H. Jang, Y.-K. Sun, S.M. Oh, J. Power Sources, 89 (2000) 7;
b) Y. Shin, A. Manthiram, 202'ui ECS Meeting, Salt Lake City, Utah, USA, October
002.
.T. Quinlan, K. Sano, T. Willey, R. Vidu, K. Tasaki, P. Stroeve, Chern. Mater. 13
2001) 4207.
a) M.M. Thackeray, C.S. Johnson, A.J. Kahaian, K.D. Kepler, J.T. Vaughey, Y. Shao
Horn, S.A. Hackney, J. Power Sources 81-82 (1999) 60; (b) J. Cho, M.M. Thackeray, J.
lectrochem. Soc. 146 (1999) 3577.
S. Ma, H. Noguchi, M. Yoshio,J. PowerSources97-98 (2001) 385.
H. Yamane, T. Inoue, M. Fujita, M. Sano, J. Power Sources 99 (2001) 60.
T. Numata, C. Amemiya, T. Kumeuchi, M. Shirakata, M. Yonezawa, J. Power Sources
7-98 (2001) 358.
R.J. Gurnmow, A. de Kock, M.M. Thackeray, Solid State Ionics 69 (1994) 59.
A. de Kock, E. Ferg, R.J. Gummow, J. Power Sources 70 (1998) 247.
E. Iwata, K. Takahashi, K. Maeda, T. Mouri, J. Power Sources 81-82 (1999) 430.
D. Song, H. Ikuta, T. Uchida, M. Wakihara, Solid State Ionics 117 (1999) 151.
G. Pistoia, G. Wang, Solid State Ionics 66 (1993) 135.
H.F. Howard, S.W. Sheargold, P.M. Story, D. Zhang, Electrochemical Society
roceedings Volume 2000-36, A.R. Landgrebe and R.J. Klinger, Eds., p. 36, 2001.
C. Vogler, A. Butz, H. Dittrich, G. Arnold, M. Wohlfahrt-Mehrens, J. Power Sources 84
1999) 243.
G. Pistoia, A. Antonini, R. Rosati, C. Bellitto, G.M. Ingo, Chern. Mater. 9 (1997) 1443.
Y.-K. Sun, C.S. Yoon, C.K. Kim, S.G. Youn, Y.-S. Lee, M. Yoshio, I.-H. Oh, J. Mater.
hern. 11 (2001) 2519.
.-S. Kim, J.T. Vaughey, C.S. Johnson, M.M. Thackeray, 202"d ECS Meeting, Salt Lake
ity, Utah, USA, October 2002.
a) M. Yoshio, H. Noguchi, Y. Todorov, Y. Hideshima, Denki Kagaku, 66 (1998) 1198;
b) H. Noguchi, H. Nakamura, M. Yoshio, Electrochemical Society Proceedings Volume
Chapter II
2000-21, M. Doyle, E. Takeuchi and K.M. Abraham, Eds., p. 18, 2001.

C. Sellitto, M. Pasquali, to be published.
G. Amatucci, N. Pereira, T. Zheng, I. Piitz, J.-M. Tarascon, J. Power Sources, 81-82
1999) 39.
Y.-K. Sun, Y.-S. Jeon, J. Mater. Chern. 9 (1999) 3147.
Y.-K. Sun, I.-H. Oh, J. Power Sources 94 (2001) 132.
S.-H. Kang, J.B. Goodenough, J. Electrochem. Soc. 147 (2000) 3621.
M. Okada, Y.-S. Lee, M. Yoshio, J. Power Sources 90 (2000) 196.
a) B. Ammundsen, J. Paulsen, Adv. Mater. 13 (2001) 943; (b) B. Ammundsen, J.
ausen, I. Davidson, C.-H. Shen, J.-M. Chen, L.-Y. Jang, Y.-F. Lee, J. Electrochem. Soc.
149 (2002) A436.
G. Ceder and S.K. Mishra, Electrochem. Solid-State Lett. 2 (1999) 550.
B. Ammundsen, J. Desi1vestro, T. Groutso, D. Hassel, J.B. Metson, E. Reagan, R.
Steiner, P.J. Pickering, J. Electrochem. Soc. 147 (2000) 4078.
T.E. Quine, M.J. Duncan, A.R. Armstrong, A.D. Robertson, P.G. Bruce, J. Mater. Chern.
10 (2000) 2838.
A. Robertson, A.R. Armstrong, A. J. Fowkes, P.G. Bruce, J. Mater. Chern. 11 (2001)
113.
M.M. Doeff, A. Anapolsky, L. Edman, T.J. Richardson, L.C. De Jonghe, J. Electrochem.
oc. 148 (2001) A230.
Y. U. Jeong, A. Manthiram, Electrochem. Solid-State Lett. 2 (1999) 421.
T. Ohzuku, Y. Makimura, Chern. Lett. 2001, 744.
T. Ohzuku, Y. Makimura, 200I ECS and ISE Joint International Meeting, San
rancisco, CA, USA, September 2001.
a) T. Ohzuku, D. Endo, Y. Makimura, 42"d Battery Symr:osium in Japan, Yokohama,
ovember 2001, p.l18; (b) Y. Makimura, T. Ohzuku, 11' IMLB, Monterey, CA, USA,
une 2002, Abs. 117; (c) N. Yabuuchi, T. Ohzuku, JJ'h IMLB, Monterey, CA, USA, June
002, Abs. 122.
.M. Paulsen, C.L. Thomas, J.R. Dahn, J. Electrochem. Soc. 146 (1999) 3560.
.M. Paulsen, C.L. Thomas, J.R. Dahn, J. Electrochem. Soc. 147 (2000) 861.
.M. Paulsen, J.R. Dahn, J. Electrochem. Soc. 147 (2000) 2478.
.M. Paulsen, D. Larcher, J.R. Dahn, J. Electrochem. Soc. 147 (2000) 2862.
a) Z. Lu, D.D. MacNeil, J.R. Dahn, 42"d Battery Symposium in Japan, Yokohama,
ovember 2001, p. 42; (b) Z. Lu, D.D. MacNeil, J.R. Dahn, Electrochem. Solid-State
ett. 4 (2001) A191; (c) z. Lu, L.Y. Beaulieu, R.A. Donaberger, C.L. Thomas, J.R.
ahn, J. Electrochem. Soc. 149 (2002) A778; (d) Z. Lu, J.R. Dahn, J. Electrochem. Soc.
49 (2002) A815; (e) Y.-K. Sun, S.H. Park, S.S. Shin, C.W. Park, ll'h IMLB, Monterey,
A, USA, June 2002, Abs. 287; (f) S.H. Park, S.S. Shin, Y.-K. Sun, II'h IMLB,
Monterey, CA, USA, June 2002, Abs. 76.
. Dan, E. Mengeritsky, Y. Geronov, D. Aurbach, I Weissman, J. Power Sources 54
1995) 143.
D. Aurbach, E. Zinigrad, H. Teller, P. Dan, J. Electrochem. Soc. 147 (2000) 1274.
.-M. Tarascon, M. Armand, Nature 414 (2001) 359.
M.S. Whittingham, P.Y. Zavalij, Solid State Ionics, 131 (2000) 109.
M.J. Duncan, F. Leroux, J.M. Corbett, L.F. Nazar, J. Electrochem. Soc. 145 (1998) 3746.
. Pranger, S. Bach, J. Farcy, J.-P. Pereira-Ramos, N. Baffier, 200I ECS and ISE Joint
.S. Johnson, M.M. Thackeray, J. Power Sources 97-98 (2001) 437.
. Jouanneau, S. Sarciaux, A. Le Gal La Salle, D. Guyomard, Solid State lonics 140
2001) 223.
M. Yoshio, H. Nakamura, Y. Xia, Electrochim. Acta 45 (1999) 273.
M.M. Thackeray, M.H. Rossouw, R.J. Gummow, D.C. Liles, K. Pearch, A. de Kock,
W.I.F. David, S. Hull, Electrochim. Acta 38 (1993) 1259.
H. Nakamura, K. Motooka, H. Noguchi, M. Yoshio, J. Power Sources 81-82 (1999) 632.

W.P. Kilroy, W.A. Ferrando, S. Dallek, J. Power Sources 97-98 (2001) 336.
B.B. Owens, S. Passerini, W.H. Smyrl, Electrochim. Acta 45 (1999) 215.
B.B. Owens, W.H. Smyrl in Electrochemical Microsystems Technology, J. W. Schultze,
T. Osaka and M. Datta, Eds., Taylor &Francis, London/New York (2002).
F. Coustier, J. Hill, B.B. Owens, S. Passerini, W.H. Smyrl, J. Electrochem. Soc. 146
1999) 1355.
M. Giorgetti, S. Mukerjee, S. Passerini, J. McBreen, W.H. Smyrl, J. Electrochem. Soc.
148 (2001) A768.
F. Coustier, J.-M. Lee, S. Passerini, W.H. Smyrl, Solid State Tonics 116 (1999) 279.
F. Coustier, G. Jarero, S. Passerini, W.H. Smyrl, J. Power Sources 83 (1999) 9.
M. Lira-Cantu, P. Gomez-Romero, J. Electrochem. Soc. 146 (1999) 2029.
A.D. Wadsley, Acta Crystallogr. 10 (1957) 261.
G. Pistoia, M. Pasquali, G. Wang, L. Li, J. Electrochem. Soc. 137 (1990) 2365.
V. Manev, A. Momchilov, A. Nassalevska, G. Pistoia, M. Pasquali, J. Power Sources 54
1995) 501.
A.Yu, N. Kumagai, z. Liu, J.Y. Lee, J. Power Sources 74 (1998) 117.
B. Chaloner-Gill, D.R. Shackle, T. N. Andersen, J. Electrochem. Soc. 147 (2000) 3575.
M.E. Spahr, P. Novak, W. Scheifele, 0. Haas, R. Nesper, J. Electrochem. Soc. 145
1998) 421.
M. Pasquali, G. Pistoia, Electrochim. Acta 36 (1991) 1549.
G. Pistoia, G. Wang, D. Zane, Solid State Tonics 76 (1995) 285.
V. Manev, A. Momchilov, A. Nassalevska, G. Pistoia, M. Pasquali, J. Power Sources
3-44 (1993) 561.
P. Liu, J.-G. Zhang, J.A. Turner, J. Power Sources 92 (2001) 204.
E. Andrukaitis, I.R. Hill, G.L. Tor1one, J. Power Sources 97-98 (2001) 494.
(a) P. Arora, D. Zhang, B.N. Popov, R.E. White, Electrochem. Solid-State Lett. 1 (1998)
49; (b) D. Zhang, B.N. Popov, Y.M. Podrazhansky, P. Arora, R.E. White, J. Power
ources 83 (1999) 121.
A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144, (1997)
1188.
N. Ravet, J.B. Goodenough, S. Besner, M. Simoneau, P. Hovington, M. Armand, 19rJh
CS Meeting, Hawaii, October 1999.
A.S. Andersson, B. Kalska, L. Haggstrom and J.O. Thomas, Solid State Tonics 130
2000) 41.
A.S. Andersson, J.O. Thomas, B. Kalska, L. Haggstrom, Electrochem. Solid-State Lett. 3
2000) 66.
M. Takahashi, S. Tobishima, K. Takei, Y. Sakurai, J. Power Sources 508 (2001) 97.
Yamada, S.C. Chung, K. Hinokuma, J. Electrochem. Soc. 148 (2001) A224.
P.P. Prosini, D. Zane, M. Pasquali, Electrochim. Acta 46 (2001) 3517.
P.P. Prosini, M. Lisi, D. Zane, M. Pasquali, Solid State Tonics 148 (2002) 45.
.B. Goodenough, V. Manivannam, Denki Kagaku 66 (1998) 1173.
F. Croce, A. D'Epifanio, J. Hassoun, A. Deptula, T. Olczac, B. Scrosati, Electrochem.
olid-State Lett. 5 (2002) A47.
P.P. Prosini, M. Carewska, S. Scaccia, P. Wisniewski, S. Passerini, M. Pasquali, J.
H. Huang, S.-C. Yin, F. Nazar, Electrochem. Solid-State Lett. 4 (2001) A170.
. Shim, K.A. Striebel, E.J. Cairns, 2001 ECS and ISE Joint International Meeting, San
rancisco, CA, USA, September 2001.
H.S. Ryu, K.Y. Kang, K.W. Kim, B.Y. Hur, J.H. Ahn, J.Y. Lee, H.J. Ahn, 2001 ECS and
SE Joint International Meeting, San Francisco, CA, USA, September 2001.
.-C. Han, H.-S. Kim, Y.-M. Kang, H.-J. Ahn, J.-Y. Lee, 2001 ECS and ISE Joint
Chapter II

H. Kim, I8'" International Seminar&Exhibit on Primary&Secondary Batteries, March
2001, Fort Lauderdale, Florida, USA.
a) S.J. Visco, IS'" International Seminar&Exhibit on Primary&Secondary Batteries,
March 2001, Fort Lauderdale, Florida, USA; (b) E.S. Nimon, M. Sukeshini, S.J. Visco,
II'" IMLB, Monterey, CA, USA, June 2002, Abs. 221.
J. Broadhead, T. Skotheim, J. Power Sources 65 (1997) 213.
H. Uemachi, Y. Iwasa, T. Mitani, Electrochim. Acta 46 (2001) 2305.
E. Peled, D. Golodnitsky, E. Strauss, J. Lang, Y. Lavi, Electrochim. Acta 43 (1998)
1593.
E. Strauss, D. Golodnitsky, E. Peled, Electrochim. Acta 45 (2000) 1519.
E. Shembel, 0. Chervacov, D. Reisner, L. Neduzhko, Y. Polischuk, I. Maksyuta, N.
Globa, T. Pastushkin, K.K. Kylyvnyk, P. Novak, I I'" IMLB, Monterey, CA, USA, June
2002, Abs. 419.
H. Kawai, M. Nagata, H. Kageyama, H. Tukamoto, A.R. West, Electrochim. Acta 45
1999) 315.
H. Kawai, M. Nagata, H. Tukamoto, A.R. West, J. Power Sources 81-82 (1999) 67.
T. Ohzuku, S. Takeda, M. lwanaga, J. Power Sources 81-82 (1999) 90.
L. Heman, J. Morales, L. Sanchez, E. Rodriguez Castellon, M.A.G. Aranda, J. Mater.
Chern. 12 (2002) 734.
Y. Ein-Eli, J.T. Vaughey, M.M. Thackeray, S. Mukerjee, X.Q. Yang, J. McBreen, J.
H. Narai, Y. Idemoto, N. Kaura, 42"d Battery Symposium in Japan, Yokohama,
November 2001, p. 42.
X. Wu, S.B. Kim, J. Power Sources 109 (2002) 53.
Y.-K. Sun, Y.-S. Lee, M. Yoshio, K. Amine, Electrochem. Solid-State Lett. 205 (2002)
A99.
Y.S. Lee, Y.K. Sun, S. Ota, T. Miyashita, M. Yoshio, Electrochem. Comm. 4 (2002) 989.
pter 12
SPINEL CATHODE MATERIALS FOR

Y. Xia" and M. Yoshiob
"Department of Chemistry, Fudan University, Shanghai, China bDepartment

of Applied Chemistry, Saga University, Saga 840-8502, Japan
1. INTRODUCTION
Secondary lithium batteries have long been studied because they exhibit
ighest specific energy among all rechargeable batteries. Nevertheless, the
ed safety and rechargeability associated with the use of metallic lithium
prevented their widespread acceptance in the market. Unsatisfactory
s with lithium metal have directed the research towards the development e
so called "rocking-chair" or "lithium-ion" batteries (LIB), based on a
13
ition metal oxide cathode and a carbon anode. - In 1990, Sony Energytec
3
commercialized this type of batteries. Today, lithium ion batteries are
in cellular phones, notebook-size personal computers, video cameras,
to a limited extent, in electric vehicle (EV) applications. Most commonly,
egative electrode acts as a "lithium sink" and the positive LiAzBy
rode acts as a "lithium source". Layered lithium transition metal oxides
pinel lithium manganese oxide have been selected as preferred cathode
rials for lithium-ion batteries. Table 12.1 summarizes the cathode
rials that have been, or possibly will be, used in LIB. LiCo02 is the
de of choice for small-size batteries. In view of the economic and
onmental advantages, spinel-structured manganese oxides have
nstrated to be the most promising positive electrode materials for
-size batteries for EV and industrial applications, as they are cheaper, less
, and show higher safety.
In this chapter, we will describe the fundamental aspects of spinel
Chapter12
12.1. Cathode Materials for Li-ion Batteries
erial
o02
i02
n204
o.2Nio.s02
no.sNio.s02
P04
ode materials with emphasis on their electrochemical performance.

ly, the structure of LiMn 204 and metal-ion substituted spinels
xMn2_x04, and their Li insertion/extraction characteristics, will be
duced, followed by addressing the issue of how to optimize cycleability
capacity. Finally, the capacity fading at elevated temperature will be
ussed, together with the new finding of oxygen-deficient spinels.
2. LiMn204 AND METAL-ION DOPED SPINELS

LiMxMnz_x04
The structure of stoichiometric spinel LiMn 204 has been refined

space group Fd3m, that can be described as layers of close-packed en
atoms, in which lithium and manganese ions occupy tetrahedral (8a)
octahedral (16d) sites, respectively, as shown in Figure 12.1. Li ions in
the etrahedral sites can be removed during charge in the 4 V range,
and n 204 transforms into A.-Mn02, also having a cubic unit cell.4
During arge to 3 V, Li ions are inserted into the 16c octahedral
sites, and n 204 with a tetragonal lattice is formed. 5 A fundamental
problem ibiting a wider use of the Mn spinel as a 4 V cathode for
lithium-ion ries is a remarkable capacity fading at elevated
temperatures. Doping
foreign metal ions, including Lt, Mg2+, Ga3+, Zn2+, Al +,3 Ni 2+, Cu2+, Co3+,
and Fe3+ has been suggested to stabilize its structure, thus improving the
11
ted temperature performance.6- The doping ions substitute for Mn+3 and,
hey reduce the capacity delivered at 4 V. A heavy doping results in a
ganese oxidation state approaching 4+. These heavily doped compounds
Spinel Cathode Materials for Lithium-Ion Batteries 363
--- [Lil-xMxls.[Mn2]16d[04h2e
(Ga3+, Zn 2+)
(3V, 4V)
[Li]s.[ Mn2-xll6d[04]32e
(Li+, Mg 2+, Ga 3+, AP+,etc)
(3V, 4V)
[Li]s.[MxMn2-xll6d[04]32e
(Ni2+, Cu 2+, Co 3+, cr-3+, Fe 3+, etc)
(4V, 5V)
[Li]s.[Mn2]16d[Ny04-y]32e
[Li]s.[Mn2]16d[04h2e (N = F, S. 3V, 4V)
(32e) • Mn (16d) e Li
(8a)
e 12.1. Structure of spinel LiMn 204 •
4.8 ..........................:::::::::::::::···········--"
4.4
l€ 3.6 I
I
;> 3.2
' ,--------------
2.8 -;. - - - - - - - - - - - - - -
, u,l\fn.012
'
''
2.4 t
20 40 60 80 100 120 140 160
Otpacity (mAh/g)
12.2. Typical charge/discharge curves of three kinds of spinel structures.
er no capacity at 4 V, but show a capacity at 3 V or 5 V. Figure 12.2 shows

pical charge/discharge for three kinds of cubic-structured spinel
rodes. In this case, M is an ion which can be further oxidized.
Li4Mn5012 (Li[Li033Mn .d04) is the upper limit of Li 1+xMn2_x04•
Chapterl2
ium insertion into its structure occurs at 2.8 V with a rechargeable

12 14
city of 140 mAh/g. - Recently, it has been demonstrated that this
toichiometric spinel has the potential for use in true solid-state
15
um/polymer batteries.
LiMnz04 shows a poorer cyclability in the 3 V region than in the 4V,
h is considered to be due to the cubic/tetragonal phase transition. The
city can be stabilized by cation substitution in 16d sites and optimization
article size, morphology and microstructure. Figure 12.3 shows the cycle
of LiA10. 1Mn .904 synthesized from two different Mn304 precursors:
one Tosoh, prepared by hydrothermal method, has a much finer particle
size
0.1 J..tm, and another from Chuo, prepared by heating Mn02 or MnS04,
a particle size of <1 J..tm. Although XRD analysis indicates virtually no
tural differences among the two spinels, TEM and SEM analysis reveal
le differences in the microstructure, which result in different reversible
cities of the electrochemical cells. XRD and TEM analysis confirm that
0.1 Mn .904 prepared with Mn304 from Chuo, showing a better capacity
16
tion, has a lower tetragonal/cubic phase ratio after cycling, in spite of
act that lithium insertion causes the average Mn oxidation state to fall
w 3.5 (e.g., 3.3).17a-c
150
125
100 Cl
75
O"t"ftTl"l·n·
· ;50
-= (;a}
I: 25
§. 100
.....
1"-' rc::Ll:n:t: .:a:l:
so
2.5
00
(bl
10 20
l 0
t:l
4Vte:nl
3 Vtest
Cycle number
Figure 12.3. Cycle tests of LiAI 0_ 1MnJ.904 in the 4 V region (3.0-4.3V) and 3 V region
(2.2-3.6V) at 1/3 Crate: (a) using Tosoh Mn 30 4; (b) using Chuo Mn 3 04 •
The end-member of a doped spinel Li[MxMn2_x]04 is LiM0. 5Mn1.504 for

5
alent cation, e.g., LiNio.sMn04 ' ,LiCu05Mn1. 0420, and LiMMn04 for a
18 19
pinel Cathode Materials for Lithium-Ion Batteries 365
21
lent cation, e.g., LiCoMn04 and LiCrMn04• If the substituent is an
izable ion, like Ni2+, Co3+, Cr3+, Fe3+, Cu 2+,22- 26(a-e) lithium deintercalation
3 4
companied by oxidation ofMn + to Mn + at4 V and oxidation ofM ions at
4
, e.g., Ne+ to Ni + in LiNi05Mn04 and Co3+ to Co4+ in LiCoMn04• The
t potential depends on the type of dopant used. Among them,
05Mn1.504 shows the best cycleability even at 50°C, particularly if
26
ared by the composite carbonate process. b Moreover, it seems that
ng the other 5 V materials with Ni increases the cycleability, e.g., in
19 2
Mn04 transformed into LiMn1.25Cr0.5Ni 0.2504. ' 6c According to in
situ FS studies, poor cycleability of LiCrMn04 would be due to the

existence
2
(VI) at a high charge state. 6d· 26e
However, some samples prepared by classical techniques show faster
city fading when cycled at high voltage and elevated temperature. This is
ase of sol-gel LiNi0.5Mn1.504 • When cycled at 55°C, this spinel retains
27
27% of its capacity after 50 cycles. A most useful approach consists in
oating of LiNio5Mn1.504 with nano-sized Zn0. 27 This allows a capacity
175,----.---- ---- ---, ---.
150
'@I "'(a)
:=
g
125
_e.
·c too
§.
u
75
50L--- _L L _j
0 10 20 30 40 50
Cycle number
12.4. Specific capacity vs. cycle number for the Li/lM LiPF -EC/DMC/Zn0-coated
6
Mnl.p4 cell at 113 Crate [cathode obtained by sol-gel method] ((a) at 55°C) and for
LiPF -EC/ DMC/LiNi Mnl. 0 [cathode obtained by the composite carbonate process]
6 05 5 4
°C and (c) 25°C).
Chapter12
tion of 99% in the same conditions. ZnO is proposed to act as a

cavenger to protect from Mn dissolution and electrolyte decomposition.
ever, as anticipated above, a significant improvement can be achieved
a proper synthesis technique. Figure 12.4 shows the cycle life at high
erature of uncoated LiNi 05Mn1.504, prepared by the composite carbonate
26
ess, b and ZnO-coated one, prepared by sol-gel. The capacity retention
e former, characterized by high homogeneity and nanosized particles,
oaches that of the latter.
These 5 V cathode materials have an advantage when used in
bination with such electrode materials as SnO, SnMex. Li4Ti5012, as they
pensate for the loss of energy density due to the high intercalation
28 29
ntial of the anode. '
2 2
Cations with an octahedral-site preference (e.g., Ni +, Cu +, Co3+, c?+)
expected to occupy the 16d sites, whereas cations with a high
3 2
rence for tetrahedral site occupation (e.g., Ga + and Zn +) are expected to
0 31
py the 8a sites, as confirmed by Rietveld analysis? • However, it has to
ressed that Zn occupies tetrahedral sites only, whereas Ga distributes over
30
tetrahedral and octahedral sites. ' 31 Both Ga- and Zn-doped spinels show
proved cycling performance at elevated temperature.
. OPTIMIZING THE CAPACITY AND

RECHARGEABILTY BY USING A Li-Mn-0
PHASE DIAGRAM
The theoretical capacity of LiMn 2 04 is 148 mAh/g, corresponding to 1

3
xtracted from LiMn204, which is accompanied by the oxidation of Mn +
4
n +. However, a remarkable capacity decline is observed during cycling.
mentioned above, the capacity fading is alleviated by using lithium,
ent and trivalent metal ion-doped spinels. Normally, the oxidation state of
ubstituents is less than 4+, thus increasing the average oxidation state of
ining manganese. This leads to a reduction of the delivered capacity and,
equently, raises the question: how to optimize both the capacity and
argeability of a spinel to be used as a practical 4 V positive electrode for
n batteries? The capacity at 4 V for the metal-modified LiM/+Mn2_y04
be estimated by the equation:
C = 26.8 [1-(4-z) y]IMW (Ahlg) (12.1)

pinel Cathode Materials for Lithium-Ion Batteries
367
e z is the valence of the doping ion and MW is the molecular weight of

+Mn2_y04.32 The relationship between the capacity and the doping ion
ation state with two levels of doping is presented in Figure 12.5. For a
doping level, a lower oxidation state of the doping ion gives rise to a
r capacity.
Herein we further discuss in detail the relationship between the capacity
ycleability of lithium-ion doped spinels. A nonstoichiometric spinel may
ithium-rich" or "vacancy-rich". Li4Mn 5012 is the limit composition of the
um-rich" series, and LizMn409 of the "vacancy-rich" series, both with a
alence of 4.
The Li-Mn-0 phase diagram is displayed in Figure 12.6, where the
h of the straight line .xy is assumed to represent the Li/Mn molar ratio and
verage Mn valence, xz the average Mn valence, and yz the Li/Mn molar
160
0
140
e iMn,o,
f-
LiMill Mn19 0" -------
120
100
> --• 0
80
•
Li+ 0
/ AI"' eos•
'
er>+
'
!it O Ni>+,Mg2'
u 60 LiM02 nuO"
40
20
l 2 3 4
Doping Ion Valence
12.5. Relationship between capacity and doping level and oxidation state of doping
ions.
The lines parallel to the line yz show changes in the average Mn valence
3.5 to 4.0, and the lines parallel to the line xz variations of Li/Mn molar
Chapter12
from 0.5 to 0.8. It is evident that the stoichiometric spinel LiMn204 and
nonstoichiometric spinels, Li 2Mn409 and Li4Mn5012, are on the three
28
ces of the equilateral triangle. If the Li/Mn molar ratio is represented as
d the average Mn valence as m:
(i) the composition of each compound on the straight lines xy, xz and yz
be expressed as:
Li 2_ 1 1\.1D. 2 _10 (xy) (12. 2)

4
1+- 1--
l+n 1-tn
Li 8n J\IIn fi 04 (yz) (12. 3)

n+4 n+4
Li m-35]\fu J(m-35)0 4 (xz) (12.4)

1--- J---
05- m. 05+m
and (ii) any compound lying within the xyz triangle can be written as:
Li sn Mn 8 04 (0.5n0.8, 3.5m4.0) (12.5)

n+m n+m
Each compound in the LiMn 204-Li4Mn5012-Li2Mn409 triangle plane

be synthesized directly by a solid-solid reaction in which Li/Mn molar
, reaction temperature and time are all controlled. It should be noted that
heating temperatures may result in an oxygen deficient spinel LiMn 204_y·
for "lithium-rich" spinels it is very difficult to obtain a formula
37 38
XMn2_x04, as reported in the literatures, ' in which the total number of
and Li atoms is just equal to 3 per formula unit. In fact, the number of Li
Mn atoms per formula unit depends not only on the Li/Mn molar ratio in
starting materials but also on the oxygen content, as the compounds
ed on the line xz have the same Li/Mn molar ratio (Li/Mn=0.50) but have
ferent composition as a result of different heating temperatures. Therefore
chemical formula of "lithium-rich" spinels should be referred to as
Mnz_y04 (ya).
The initial capacity can be evaluated through the knowledge of the
n molar ratio, n, and the average Mn valence, m [equation (12.6)]:
C = 1184(4-m)/(m+n) (mAhlg) (12.6)
The capacity decreases with increasing vacancy rate or lithium content

e spinel matrix, while the rechargeability is improved. This indicates that
capacity and good rechargeability cannot coexist in these materials, e.g.,

spinel with a Li/Mn molar ratio of 0.55, the capacity may be limited to
mAh/g. Despite this, it should be possible to optimize both of them in
s of the Li-Mn-0 phase diagram given above. For example, if the
mum spinel electrode requires a capacity in the range 110-125 mAh/g and
d rechargeability, its formula should be located in the shaded area.
Mn: 3.5-4.0
Mn: 3.5-4.0
Lithium-rich
12.6. Semiquantitative Li-Mn-0 phase diagram for optimization of the capacity and
bility.
. CAPACITY FADING ON CYCLING

A fundamental problem prohibiting a wider use of spinels as cathodes
thium-ion batteries is that their capacity fades when they are cycled in an
voltage range. Several factors that contribute to the fading have been
ssed in the literature: (i) electrochemical oxidation of the electrolyte at
39 42 2
voltage; - (ii) dissolution of Mn + from LiMn204 into the electrolyte
31 42 39
ion; - (iii) structural failure in the two-phase (4.2 V) region, with
39 43
formation to a more stable defect spinel; ' (iv) fracture of the particle
44
ce due to local Jahn-Teller distortion at high discharge rates, and (v)
45
n site exchange, etc. It can be envisaged that all above suggested
Chapter12
hanisms will also to some extent occur in metal-ion doped spinels, with a
rence in their insertion/extraction reactions (two-phase for the former vs.
phase for the latter). Capacity fading has to be referred primarily to the
el structure. Lithium-ion intercalation in the stoichiometric LiMn204
ework leads to a one-phase reaction (cubic a= 8.163 to 8.247 A) in the
e O<.x<0.55 in Li 1.xMn204 (4.0 V plateau), and a two-phase reaction
ic a= 8.154 and 8.072 A) in the range 0.55<.x<0.9 (4.2 V plateau), while
the lithium-rich compound Li 1+xMn2 04+y• with a high capacity
lization, a one-phase reaction (cubic a= 8.105-8.223 A) was observed
the entire intercalation range 0.25<Li<l.04.33.4 6 Moreover, the
eability of LiMn204 critically depends on the operating temperature.
39
n 204 loses less capacity during cycling at 0°C. In situ XRD experiments
irmed that split peaks were still observed after 50 cycles at 0°C
esponding to two phases in the high voltage region), indicating that the
phase structure is more stable for lithium ion insertion/extraction at this
erature. On the other hand, during cycling at 50°C, split peaks were
ly detected and each diffraction line shifted almost continuously to
er diffraction angles, indicating a continuous contraction of the unit
Figure 12.7 shows the cell lattice parameter changes as a function of
8.28 Before cycling

•
$
8.24 0
•
Before cycling
• • •
•
• •
='" 8.20 After cycling
i
s
.
•• •
•
8.16
!a0
1.1
Q,j
:=
8.12
8.08
• •
••
0
1a 8.04 0
.1
8.00
0.0 0.2 0.4 0.6 0.8 1.0
X inLi,Mn2 04
e 12.7. Variation of the cubic lattice parameter as a function of oxidation depth of spinel in
st charge and after 50 cycles at 50°C (113 C rate).
ation depth (in the first charge and after cycling at 50°C for 50 cycles).
hermore, the lattice parameter, a, decrease from 8.246 A of original
n204 to 8.235 A after cycling at 50°C. Both results are consistent with a
ge to a defect structure, in which more Mn vacancies are present in 16d
, e.g., "lithium-rich" or "vacancy-rich" spinels are formed. Mn
lution and structural changes are facilitated at high temperature.47• 48
arently, all above proposed mechanisms are mainly associated with the
al structure and morphology. Accordingly, many researches have been
ed to these aspects. The most common way to stabilize the structure is by
ng with metal ions, as previously discussed. A second way is to reduce
the issolution by minimizing the LiMn20Jelectrolyte interface through
51
the
49
hesis of spinels with small surface area; (a,b),so coating with Lh0.B203,
27
or LiCo02 52; "'b using a salt containing less HF, e.g. LiBF4 " or Li
53
53
alatoborate (LiBOB) b instead of LiPF6; or using (CH)3SiNHSi(CH)354
h is able to react with water traces to act as a scavenger for HF, thus
ressing Mn dissolution.
140
I
!;>
·
8
100
0 50
Cycle number
12.8. Effect of the surface area on the cycle profile of spinel electrode materials at 50°C
2 2
rate). (a) LiMnz04 (3.0 m /g), (b) LiuMnz04.19 (6.3 m /g) and (c) LiuMn2 04.17 (1.2
Figure 12.8 shows cycling of a stoichiometric spinel and two

m-rich spinels, prepared by different methods, with the same chemical
Chapter12
position of LiuMn204.2 but different surface areas. Stoichiometric spinel

s, as expected, much of its initial capacity. The spinel with the small
2
ace area of 1.22 m /g loses 5% of its initial capacity over the first 100
2 49
es compared to 8% of the one with an area of 6.2 m /g. " The different
hesis techniques may have reduced the difference between the two
els, but low surface area is beneficial especially in terms of Mn
53
olution. The solubility of the common spinel is reported around 70 ppm, b
2
ighly optimized metal ion doped spinels with surface area of 0.2-0.3 m /g
w solubilities of 2-3 ppm after storage in the electrolyte at 60°C for 4
ks.
We have recently prepared spinels with a high stabilization at elevated
eratures by taking into account three factors: low surface area, doping
metal ions, oxygen stoichiometry. A two-step process has been
loped based on the: 1) synthesis at relatively high temperature to keep the
ace area low, thus reducing Mn dissolution, 2) annealing at lower
erature under 02 atmosphere, so to have spinels with a stoichiometric
49
unt, or a slight excess, of oxygen. b Figure 12.9 shows charge-discharge
of lithium-ion cells carried out at 60°C. The cathodes consisted of either
asi-stoichiometric spinel (Li,Mn,Mgh04+8 (Og):::;0.02), prepared as above,
spinel with an oxygen defect (Li,Mn,Mgh04_0•
Even at 60°C, the spinel prepared according to the two-step process
s a remarkably constant capacity. In the following, the problems created
n oxygen deficiency will be stressed.
. .,
'O l
0
0 oxygen defect spinel
£ • oxygen stoichiometric spinel

2 150
8.. 100 •
e o·o·o·o·o·oooo
· o·
·o
o•o•o••••
§""' 50
u
QL---._-- -- -- -- ----L---
0 10 15 20
Cyclfl number
e 12.9. Cycle performance of Li-ion batteries at 60°C: (•) oxygen quasi-stoichiometric

, (Li,Mn,Mgh04+8, and (o) oxygen-deficient spinel (Li,Mn,Mg)304_0,as a cathode at 1/3 C
node: graphitized mesophase carbon (MCMB6-28). CR2032 type coin cells.
. OXYGEN-DEFICIENT SPINEL LiMn204.y
LiMn204 has commonly been synthesized using solid-state reactions,

solvent-process techniques, e.g., sol-gel, precipitation, and Pechini
ess. In fact, it is quite difficult to obtain the truly stoichiometric LiMn
204 out anion or cation vacancies. The spinel synthesis mechanism
involves ation of a defect spinel at <600°C (the temperature is
dependent on the ion process, with particular reference to the lithium
salt), followed by a formation from a defect to a well-ordered spinel
proceeding via a loss of en. When the annealing temperature is as high as
800°C, LiMn 204 loses en to give an oxygen-deficient spinel LiMn 2 04 _y
and, on further heating
46
00°C, LiMn02 is formed. During cooling, oxygen can reversibly be
up, but the lost oxygen cannot completely be recovered. The net oxygen
ht loss increases with increasing annealing temperature.36• 46 The
erature corresponding to transformation into an oxygen-deficient spinel
a well-ordered spinel is lower as the lithium content in the starting
rial is higher.38 This is why no general consensus on the mechanism of
city fading on cycling has yet been reached. The structural and
rochemical results vary from sample to sample, even for samples
claimed oichiometric LiMn 204 by means of chemical analysis. The
key factor mining these different results is oxygen stoichiometry.
The oxygen hiometry plays a very important role in the structures and
electrochemical erties of LiMn 204 spinel materials. Herein, the
charge/discharge profile, e transition during lithium-ion
insertion/extraction, and low-temperature e transition will be related to
the oxygen stoichiometry.
.1. Charge/Discharge Profile of Spinels with Different

Li/Mn Ratios and Oxygen Contents
Figure 12.10 shows charge/discharge curves of four kinds of

YMn204±S spinels with different Li/Mn ratios (obtained by varying
Li/Mn ratio in the starting materials) and various oxygen contents (by
olling the synthesis conditions, including temperature, heat treatment
and atmosphere, during both the solid state reaction and annealing), i.e.,
le A is quasi-stoichiometric spinel Li099Mn 204.02, sample B
en-deficient Li0.98Mn 203.97, sample C lithium- and oxygen-rich
Chapter12
5Mn 204.10, and sampleD lithium-rich but oxygen-deficient

Li1.05Mn203.92• capacity fading of the quasi-stoichiometric spinel mainly
occurs on the V plateau. In contrast, fading occurs on both 4.0 V and 4.2
V plateaus for ompounds with an oxygen-deficiency (samples B and D).
The lithium oxygen-rich compound shows the highest capacity
stabilization (sample Capacity fading in oxygen-deficient samples is
much faster than in those
out an oxygen-deficiency.55 For the samples with an oxygen deficiency, a
8
small discharge plateau at about 3.2 V was also observed (B and D)? A
nt research reported that there is a linear relationship between the capacity
56
2 V (or 4.5 V) and oxygen deficiency. The two arrows in FigurelOB
w that the oxygen-deficient spinel loses capacity in both the low voltage
the high voltage region.
4.4
A
4.0
l.1
4.4
B
4.0
.u
l.1
u
10 40 0 80 100 110 10
Capacity (mAlt/g)
e 12.10. Charge/discharge curves of four kinds of Li 1±yMn204 ± 5 spinels at 113 Crate.

le A: stoichiometric spinel Li0.99Mnz04.02· Sample B: oxygen-deficient Li 0.98Mn203.97,
le C: lithium and oxygen-rich Liz.05Mn204. 10, and Sample D: lithium-rich but
n-deficient Liz.osMnz03.92·
.2. Phase Transitions During Charge/Discharge

It has been demonstrated that the capacity loss for stoichiometric spinel
n 204 is mainly due to the unstable two-phase region of the 4.2 V plateau.
hown above, this two-phase characteristic, and the capacity fading, can be
3
ressed by making the sample lithium-rich (one-phase model)? Moreover,
57
ree-phase model was also proposed in the literature. Recent research d
on systematic in situ XRD studies has shown that the phase transition
cally depends on the oxygen stoichiometry, this affecting the cycling
55
ile. When a lithium-rich compound is heat-treated to yield an
en-deficient compound, it shows clear evidence of two biphasic regions in
the 4.0 V and 4.2 V plateaus, and significant capacity fading was observed
oth plateaus. In the compound with no oxygen deficiency, the two-phase ns
were suppressed, and the capacity fading was also reduced. These results ide
strong evidence relating the capacity fading during cycling to the itions
among three cubic phases.
.3. Phase Transition at Low Temperature
In 1995, Yamada and Tanaka found that stoichiometric LiMn204

rgoes a first-order phase transition from a cubic (Fd3m) to a tetragonal
md) structure during cooling (at about 280 K) attributed to a Jahn-Teller
58
rtion. They concluded that the mass fraction of the tetragonal phase
ases with decreasing temperature, but saturates below 260 Kat a value of
. On the other hand, other research groups proposed a low temperature
e with -100% orthorhombic symmetry (Fddd) at temperatures below 250
60
More recently, it has been demonstrated that this phase transition is
ly related to the population of oxygen vacancies. 5 5• 61 The existence of
en vacancies is the sole and necessary condition for this phase transition.
ther words, if this phase transition is observed, the sample is not an
en-stoichiometric one, but contains oxygen vacancies. Equivalently, in a
tly stoichiometric sample, this phase transition should not occur.
Figure 12.11 shows the DSC traces for four kinds of compounds
een -30°C and 50°C. For the spinel sample with composition close to the
hiometric LiMn204 (sample A), no exothermic and endothermic peaks

observed in the DSC curves. On the other hand, oxygen-deficient
n203.97 (sample B) shows a phase transition. For lithium-rich
Mnz04.JO (sample C), no exothermic or endothermic peaks were detected.
Chapter12
58
is in agreement with Yamada's data. However, after an
en-deficiency was created by heat treatment, Li1.05Mn 2 03.97 (sample D)
s clear exothermic and endothermic peaks. The peak amplitude increases
increasing degree of oxygen deficiency. Therefore, the disappearance of
hase transition observed in the lithium-rich samples was actually due to
limination of the oxygen-deficiency. As soon as the oxygen-deficiency
created, the phase transition reappeared. Therefore,
the
en-deficiency is the primary condition and the content of lithium is only a
ndary factor.
l U
---
100
so c oo lin g
0 A
-SO
-100
-ISO
H - -eati ng B
--
200
D cooli ne
'-"
100
.. .. .. ., .
0:1
OJ
-100
-200
_, . . . ,. ._.._ ..L... -'
-
U ..ati
- :ng
-300 ._ ....
-30 -20
_.__
-10 0 10 20 30 40 so
Temperature (C)
e 12.11. DSC traces between -30°C and 50°C for four kinds of Li 1±yMn204 ±s spinels:
le A (Li0.99Mnz04.oz). Sample B (Lio.9sMnz03.97), Sample C (Lii.osMnz04.IO), and Sample
.osMnz03.92)-
. SUMMARY
The capacity fading on cycling of spinel cathode materials, especially at

temperatures, is mainly due to structural instability and Mn dissolution, h
can be improved by metal-ion substitution and morphology
ification. It is possible to obtain spinel cathode materials with high
city stabilization by controlling the Li/Mn (or Li/M+Mn) ratio, the heat
ment temperature and time, and the atmosphere during both the solid state
ion and annealing. In particular, the oxygen deficiency should be
ded, as it is the primary factor affecting the structural stability. The
city fading on cycling of spinel cathode materials is not only dependent
e spinel material itself, but also on the electrolyte and anode materials.
arkable progresses have recently been made, this allowing the realization
actical batteries.
For instance, a spinel material with a capacity of 110 mAh/g has been
in aLi-ion battery developed by Shinkobe Inc. for Nissan Hypermini EV
62
Tino HEV. "·bRecently, it has also been applied in 2.5-3 Ah medium-size
63
ries with soft cover for electrically assisted bicycles by Panasonic Inc.
of the advantages of using spinel for EV or HEV applications is the high
capability, up to 50 C, which is twice that of MH batteries or Ni-based
n batteries. Spinels used in HEV and bicycles are oxygen-stoichiometric
les with low surface area, to improve the high temperature performance.
A cooperative research among electrochemists, inorganic and organic
ists, ceramists, and among research groups, material makers and battery
rs may provide new advancements in this field. We foresee that high y,
low-cost lithium-ion batteries with Mn-based materials will grow fast e
market in the near future.
FERENCES
M. Armand, in Materials for Advanced Batteries, D.W. Murphy, J. Broadhead and C.H.
Steele, Eds., Plenum Press, New York, 1980, p.145.
T. Nagura, K. Tozawa, Prog. Batteries Solar Cells 9 (1990) 209.
K. Brandt, 1hiMLB, Boston, MA, USA, May 1994, Abstract FRI-30.
J. Hunter, J. Solid State Chern. 39 (1981) 142.
M. M. Thackeray, A. De Kock, M. H. Rossouw, D. C. Liles, D. Hoge, R. Bittihn, J.
Chapter12
Electrochern. Soc. 139 (1992) 363.

G Li, H. Ikuta, T. Uccida, M. Wakihara, J. Electrochern. Soc. 143 (1996) 178.
M. Yoshio, Y. Xia, N. Kumada, S. Ma, J. Power Sources 101 (2001) 79.
J. M. Tarascon, E. Wang, F. K. Shokoohi, W. R. McKinnon, S. Colson, J. Electrochern.
Soc. 138 (1991) 2859.
A. D. Robertson, S. H. Lu, W. F. Averill, W. F. Howard, Jr., J. Electrochern. Soc. 144
(1997) 3500.
G Pistoia, A. Antonini, R. Rosati, C. Belletto, GM. Ingo, Chern. Mater. 9 (1997) 1443.
G Pistoia, A. Antonini, R. Rosati, C. Bellitto, J. Electroanal. Chern. 410 (1997) 115.
M. M. Thackeray, A. De Kock, Mat. Res. Bull. 28 (1993) 1041.
M. H. Rossouw, A. De Kock, L.A. de Picciotto, M. M. Thackeray, Mat. Res. Bull.
25 (1990) 173.
A. De Kock, M. H. Rossouw, L. A. de Picciotto, M. M. Thackeray, Mat. Res. Bull. 25
(1990) 657.
Y. Xia, T. Sakai, C. X. Wang, T. Fujieda, K. Tatsumi, K. Takahashi, A. Mori, M. Yoshio, J.
Electrochern. Soc. 148 (2001) All2.
Y.-K. Sun, C.S. Yoon, C.K. Kim, S.G Youn, Y.-S. Lee, M. Yoshio, I.-H. Oh, J. Power
Sources 108 (2002) 97.
Y.- S. Lee, Y.- K. Sun, M. Yoshio, Chern. Lett. 498 (2001).
Y.- S. Lee, C.S.Yoon, Y.- K. Sun, M. Yoshio, Electrochern. Solid-State Lett. 5 (2002)
Al.
Y.-S. Lee, M. Yoshio, Electrochern. Solid-State Lett. 4 (2001) A155
Q. Zhong, A. Bonakdarpour, M. Zhang, Y. Gao, J. R. Dahn, J. Electrochern. Soc. 144
(1997) 205.
Y. Todorov, C. Wang, B. I. Banov, M. Yoshio, Batteries for Portable Applications and
Electric Vehicles, C. F. Holmes and A. R. Landgrebe, Eds., The Electrochem. Soc. Inc.,
Pennington, NJ, 1997, PV97-18, p. 176.
Y. Ein-Eii, W. F. Howard, Jr., J. Electrochern. Soc. 144 (1997) L205.
Y. Todorov, Ph.D. Thesis, Saga University, Japan, 1999.
T. Ohzuku, S. Takeda, M. Iwanaga, J. Power Sources 81-82 (1999) 90 .
J. H. Lee, J. K. Hong, D. H. Jang, Y. K. Sun, S.M. Oh, J. Power Sources 89 (2000) 7.
K. Amine, H. Tukamoto, H. Yasuda, Y. Fujita, J. Electrochern. Soc. 143 (1996) 1607.
Y. Gao, K. Myrtle, M. Zhang, J. N. Reimers, J. R. Dahn, Phys. Rev. B 54 (1996) 3878.
H. Kawai, M. Nagata, H. Tukamoto, A. R. West, J. Power Sources 81-82 (1999) 67.
Y.S.Lee, Y.K.Sun, S.Ota, T.Miyashita, M.Yoshio, Electrochern. Cornrn. 4 (2002) 989.
Y.-S. Lee, Y.M.Todorov, K.Konishi, M. Yoshio, ITE Lett. 1 (2000) 883.
Y. Terada, K. Yasaka, F. Nishikawa, T. Konishi, M. Yoshio, I. Nakai, J. Solid State
Chern. 156 (2001) 286.
K.Yasaka, Master Thesis, Science University of Tokyo (2001).
Y.-K. Sun, Y.-S. Lee, M. Yoshio, K. Amine, Electrochem. Solid-State Lett. 5 (2002) A99.
Y. Xia, T. Sakai, T. Fujieda, M. Wada, H. Yoshinaga, J. Electrochem. Soc. 4 (200l)A9.
T. Ohzuku, K.Ariyoshi, S.Yamamoto, Y. Makimura, Chern. Lett. 1270 (2001).
M. Yoshio, H. Noguchi, Y. Todorov, Y. Hideshima, Denki Kagaku 66 (1998) 1198.
H. Noguchi, H. Nakamura, M. Yoshio, Electrochemical Society Proceedings Volume
2000-21, M. Doyle, E. Takeuchi, K.M. Abraham, Eds., p. 18,2001.
Y. Xia, M. Okada, M. Nagano, M. Yoshio, Batteries for Portable Applications and
Electric Vehicles, C. F. Holmes and A. R. Landgrebe, Eds., The Electrochem. Soc. Inc.,
Pennington, NJ, 1997, PV97-18, p. 494.
Y. Xia, M. Yoshio, J. Electrochem. Soc. 143 (1996) 825.
Y. Xia, M. Yoshio, J. Power Sources 57 (1995) 125.
R.I. Gummow, A. de Kock, M.M. Thackeray, Solid State lonics 69 (1994) 59.
Y. Xia, M. Yoshio, J. Electrochem. Soc. 144 (1997) 4186.
M.M. Thackeray, L.A. de Picciotto, A.de Kock, P. Johnson, V. Nicholas, K. Adendorff,
Mater. Res. Bull. 21 (1987) I.
Y. Gao, J.R. Dahn, J. Electrochem. Soc. 143 (1996) 100.
Y. Xia, Y. Zhou, M. Yoshio, J. Electrochem. Soc. 144 (1997) 2593.
D. H. Jang, Y. J. Shin, S.M. Oh, J. Electrochem. Soc. 143 (1996) 2204.
D. Guyomard, J. M. Tarascon, Solid State lonics 69 (1994) 222.
G Pistoia, A. Antonini, R. Rosati, D. Zane, Electrochim. Acta 41 (1996) 2683.
S. J. Wen, T. J. Richardson, L. Ma, K. A. Striebel, P. N. Ross, Jr., E. J. Cairns, J.
Electrochem. Soc. 143 (1996) L136.
M. M. Thackeray, Y. Shao-Horn, A. J. Kahaian, K. D. Kepler, E. Skinner, J. T. Vaughney,
and S. A. Hackney, Electrochem. Solid-State Lett. I (1998) 7.
J. M. Tarascon, W. R. McKinnon, F. Coowar, T. N. Bowmer, G Amatucci, and D.
Guyomard, J. Electrochem. Soc. 141 (1994) 1421.
Y. Xia, H. Noguchi, M. Yoshio, J. Solid State Chern. 119 (1995) 216.
X. Wang, Y. Yagi, Y-S Lee, M. Yoshio, Y. Xia, T. Sakai, J. Power Sources 97-98 (2001)
27.
T. Inoue, M. Sana, J. Electrochem. Soc. 145 (1998) 3704.
Y. Xia, N. Kumada, N. Yoshio, J. Power Sources 90 (2000) 135.
M.Yoshio, X. Wang, H. Nakamura, 203rd ECS Meeting, to be presented (2003).
G G Amatucci, C.N. Schmutz, A. Blyr, C. Sigala, A.S. Gozdz, D. Larcher, J. M.
Tarascon, J. Power Sources 69 (1997) II.
G G Amatucci, A. Blyr, C. Sigala, P. Alfone, J. M. Tarascon, Solid State lonics 104
(1997) 13.
.Cho, GB.Kim, H.S.Lim, C.-S.Kim, S.-In.Yoo, Electrochem. Solid-State Lett. 2
(1999) 607.
S.-C.Park, Y.-S.Han, Y.-Skang, P.S.Lee, S.Ahn, H.-M.Lee, J.-Y.Lee, J. Electrochem.
Chapter12
Soc. 148 (2001) A680.

Y. Xia, Y. Hideshima, N. Kumada, M. Nagano, M. Yoshio, J. Power Sources 74 (1998) 24
J. Liu, K. Xu,T. R. Jow, K. Amine, 20J!ld ECS Meeting, Abs. No. 135 (2002).
H. Yamane, T. Inoue, M. Fujita, M. Sano, J. Power Sources 99 (2001) 60.
Y. Xia, T. Sakai, T. Fujieda, X. Q. Yang, X. Sun, Z. F. Ma, J. McBreen, M. Yoshio, J.
Electrochem. Soc. 148 (2001)A723.
Y. Yagi, Y. Hideshima, M. Sugita, H. Noguch, M. Yoshio, Jpn. Electrochem. 68 (2000)
252.
X. Q. Yang, X. Sun, S. J. Lee, J. McBreen, S. Mukerjee, M. L. Daroux, X. K. Xing,
Electrochem. Solid-State Lett. 2 (1999) 157.
A. Yamada, J. Solid State Chem. 122 (1996) 160.
K. Oikawa, T. Kamiyama, F. Izumi, B. Chakoumakos, H. Ituka, M. Wakihara, J. Li, Y.
Matsui, Solid State Ionics 109 (1998) 35.
J. Rodriguez-Carvaja1, G Rousse, C. Masquelier, M. Hervieu, Phys. Rev. Lett. 81 (1998)
4660.
X. Q. Yang, X. Sun, M. Balasubramanian, J. McBreen, Y. Xia, T. Sakai, M. Yoshio,
Electrochem. Solid-State Lett. 4 (2001) All7.
K. Hironaka, T. Aiba, T. Kai, T. Matumura, K. Koseki, T. Horiba, Y. Muranaka,
Shin-Kobe Tech. Report 10 (2000) 3.
T. Kai, H. Ando, Y. Muranaka, T. Horiba, K. Hironaka, Shin-Kobe Tech. Report 11 (2001)
9.
http://www.panabyc.eo.jp/
apter 13
LAYERED MANGANESE OXIDES AS

CATHODES
B. Ammundsen
Pacific Lithium New Zealand Limited, PO Box 90725, Auckland, New Zealand
. INTRODUCTION
The first commercial lithium-ion batteries used LiCo02 as the cathode
e material, and this material continues to be used in most lithium-ion
ries manufactured despite the high cost and safety hazards associated
cobalt. Apart from LiCo02, only the isostructural nickelate LiNi02, and
particularly the Co-substituted nickelate LiCoxNi 1_x02, have been
idered to have sufficient energy density and cycling stability to be of
mercial interest. However the nickelates present safety and toxicity
erns which are still greater than LiCo02• Manganese oxides offer lower
and toxicity than cobalt or nickel, and have been demonstrated to be
on overcharge. A lithium manganese oxide based cathode should
fore, at least in principle, provide significant technological advantages
ithium-ion system over LiCo02, LiNi02, or LiCoxNi 1_x02•
One of the reasons that LiCo02 functions so well as a lithium-ion
de material is that it has a well-ordered and stable layered crystal
ture which is easily prepared and handled in air, and enables a fast and
sible lithium intercalation. Nickelates also have a layered crystal
ture, although the synthesis conditions are more difficult, requiring
nation and sintering under a controlled oxygen-rich atmosphere. The
readily prepared lithium manganese oxide is LiMn 204, which does not
a layered crystal structure but a spinel structure (refer previous chapters
is volume). Although the spinel crystal structure of LiMn 204 permits
intercalation of lithium ions, its stoichiometry means that it has a lower
city than LiCo02• Furthermore, the stoichiometric spinel LiMn 204
s a large capacity fade with cycling. Lithium-rich spinel compositions
xMn2_x]04 or materials substituted with cations such as Co, Cr, Al,
Chapter 13
or Zn provide better cycling stability and are therefore generally applied

reference to the stoichiometric material. However the Li-rich and
tituted materials give capacities only around 115 mAh/g, and even
lized formulations may still exhibit stability problems in aLi-ion battery
a carbon anode. These disadvantages have not yet been fully overcome,
have limited the application of spinel in commercial lithium-ion
ries.
A layered lithium manganese oxide cathode material with the
hiometry LiMn02 would be expected to overcome at least some of the
ations of the spinel manganese oxides. Layered crystal structures based
ithium manganese oxide have therefore been widely investigated over
last few years as an alternative to spinel. The underlying idea is to
bine the electrochemical properties of the layered structures, particularly
higher theoretical capacity, with the advantages of a manganese-based
istry. The problem with layered LiMn02, however, is that its synthesis
t trivial.
The structural chemistry of manganese oxides is very complex - more
plex than that of cobalt or nickel oxides. Even the binary oxide MnOz
s in a variety of structures, as well as intergrowths of these structures.
efore it is not surprising that lithiated manganese oxides can also be
ared in a variety of structures. Thermodynamically stable spinel
pounds have been reported with a range of different cation distributions
in the compositional range LiMn204-LhMn40g-Li4Mns012. Other
ly prepared lithium manganese oxides are an orthorhombic form of
n02 and a monoclinic phase Li2Mn03• However LiMn02 with a layered
al structure analogous to LiCo02 is not thermodynamically stable, and
fore cannot be simply prepared using the high temperature processing
es used to manufacture other lithium manganese oxides, LiCo02, or
um nickelates.
This chapter examines the various approaches that have been taken to
are and stabilize the layer structure for lithium manganese oxide in order
velop high energy storage capacity and cycling stability in lithium-ion
. Strategies have included preparation by soft chemistry synthesis or by
tive substitution, but neither approach has fully successfully dealt with
nstability of the layered manganese oxide structure when cycled. As an
rstanding of the relationships between crystal structural chemistry and
rochemical performance has developed, recent work has focused on g
the properties of tetravalent rather than trivalent Mn as a basis for
loping novel complex or solid-solution cathode materials with improved
city, cycling stability and safety. This latter approach provides a
nological solution to many of the stability issues that have previously
ented layered lithium manganese oxides from entering into commercial
ications.
ed Manganese Oxides as Cathodes 383
. OVERVIEW OF THE STRUCTURE AND

STABILITY OF LAYERED MANGANESE OXIDES
.1. Layered vs. Orthorhombic Structure of LiMn0 2
When prepared by typical high temperature solid state methods,

o02 has the well-ordered a-NaFe02 rocksalt structure with R3 m
metry shown in Figure 13.1(left). The stacking sequence of the oxygen
attice in the rocksalt structure is ABCABCABC. The cations are located
tahedral sites, ordered to make alternating layers of transition metal and
m, thus yielding AyB a CA c BaC (where greek letters denote
ition metal layers and small latin letters denote Li layers). The transition
l cations (M) and oxygen anions form rigid MOt) sheets. The Lt ions
een the sheets are more mobile, and can be reversibly extracted and
erted electrochemically.
LiMn02 is not thermodynamically stable as the layered structure.
er equilibrium high temperature synthesis conditions, preparations of
hiometric LiMn02 crystallize in the orthorhombic structure (space
p Pmmn) illustrated in Figure 13.l(right). In this structure, which is
ue among lithiated transition metal oxides, Li and Mn ions occupy
edral sites arranged in an alternating zig-zag configuration of edge
ng [Li06] and [Mn06] octahedra. The phase with this structure is
monly referred to as "orthorhombic LiMn02", or a-LiMn02•
e 13.1. Crystal structures of layered LiCo02 (left) and o-LiMn02 (right). Small white
es represent Li ions, small black spheres represent transition metal ions, and large gray
es represent oxygen.
Chapter 13
However, the difference in stability between the orthorhombic and

red structures is rather small, and soft chemical routes such as ion
13
ange or hydrothermal syntheses can yield layered LiMn02• - These
hesis methods are described in more detail below in Section 13.3.2. The
lting materials do not have a perfect a.-NaFe02 structure. The trivalent cf
ions cause a cooperative distortion of the Mn06 octahedra due to Jahn
er stabilization, leading to a monoclinic unit cell (space group C2/m)
angles and lattice constants distorted from the higher rhombohedral
metry.
The physical principles underlying the thermodynamic stability of
orhombic LiMn02 over the layered structure have been examined using
4
principles structure calculations. Mn3+ ions introduce two features
h distinguish LiMn02 from other lithiated transition metal oxides: a
magnetic moment (four unpaired electrons in the high-spin
iguration), and Jahn-Teller distortion of the Mn06 octahedron due to the
le occupancy of an eg level. Structure calculations show that although the
red monoclinic structure should be more stable based on ionic size
3
iderations, strong antiferromagnetic interactions between Mn + ions, and
consequent localization of electronic states, stabilize the orthorhombic
ture of LiMn02 over the layered structure. Chemical substitution of
n02 by either nonmagnetic ions or ions with valence other than 3+ could
fore stabilize the layered structure by perturbing the magnetic Mn-Mn
action and/or electronic charge on the Mn ion. This approach has been
essfully implemented in practice using several different chemical
ies, as described in Section 13.3.3 below.
.2. Instability with Respect to the Spinel Structure
Although strategies have been developed to synthesize layered lithium

ganese oxides, a further stability problem arises when the materials are
as cathodes. When LiMn02 materials are cycled, layered (and also
rhombic) structures tend to transform toward the spinel structure.
el LiMn204 differs from the layered LiMn02 by having a three
nsional crystal structure rather than the two-dimensional layered crystal
ture. The Li ions reside in tetrahedral sites in channels of the spinel
2 crystal structure, rather than the interlayer octahedral sites of the
ed LiMn02• However the transformation between a layered or
rhombic structure and a spinel structure requires only relatively minor
angement of the cations. Figure 13.2 shows the spinel crystal structure
iMn204, drawn with the oxygen planes in the same relative orientation
e structures in Figure 13.1 to illustrate the similarity between the basic
tures. In relation to the R 3m layered structure, the spinel structure
can
e 13.2. Crystal structure of spinel LiMn 204 (refer Figure 13.1 for key).
iewed as having cation layers with octahedral sites alternately filled

-quarters and one-quarter by Mn, resulting in three-dimensional
nels with Li ions in tetrahedral sites.
Structure calculations have also helped to understand this instability of
ed and orthorhombic manganese oxide structures with respect to the
formation to spinel. Calculations have shown that the spinel structure is
getically preferred over the layered or orthorhombic structure for most
ted transition metal oxides when the stoichiometry Li0.5M02 is reached
g extraction of Li.5·6 This is true for cobalt oxide as well as manganese
e, and yet cobalt oxide cycles well without showing spinel
formation. This suggests that the transformation to spinel occurs more
ly in manganese oxide due to different kinetic factors. Ab initio
lations suggest that delithiated LixMn02 layered materials transform to
l in a two stage process.7 In the first stage a portion of the Mn and Li
rapidly migrate into tetrahedral sites surrounded by Li vacancies. The
ation barriers for this are low, partly because the migration of Mn into a
hedral site is assisted by a stabilization gained by charge
oportionation of Mn3+ into Mn2+ (tetrahedral) and Mn4+ (octahedral)
es. By comparison, in LiCo02 the activation energy for the same
ess was calculated to be prohibitively high. The second stage of the
ed-to-spinel transformation involves a more difficult coordinated
angement of Mn and Li ions to form spinel, which takes place more
ly due its complexity and higher activation barriers. This calculated
l agrees with experimental observations of LiMn02 cathodes examined
ffraction and X-ray absorption experiments. Experimental data typically
that significant structural rearrangements to a disordered, less two
nsional structure take place during the first charge (delithiation) of the
de material, followed by a gradual re-ordering into an increasingly
Chapter 13
8 11
el-like structure over subsequent cycles. -
Figure 13.3(A) in Section 13.3.3 shows the consequence of the phase
formation toward the spinel structure on the discharge voltage profile of
pical LiMn02 material (here shown for o-LiMn02 prepared by high
erature processing). The transformation toward the spinel structure is
mpanied by evolution of a voltage profile characteristic of spinel
ganese oxide, which drops sharply from approximately 4 V vs. Li/Li+ to
below 3 V near the Li 0.5Mn02 stoichiometry, i.e. at the halfway point of
harge. The voltage step in the spinel structure is associated with a change
ntercalation mechanism. For O<x<l in spinel LixMn 204, Li ions are
ted directly into tetrahedral sites with high binding energy for lithium, a
ess which takes place around 4 V vs. LilLi+. For l<x<2 in spinel
n204, Li ions are inserted into octahedral sites to form a tetragonally
rted rock salt structure of stoichiometry Li2Mn204 (=LiMn02) in a two
e process which occurs at a potential of just below 3 V vs. LilLi+. Such a
drop in voltage midway through discharge poses problems for power
agement in actual systems. Furthermore, while a crystalline spinel
ture provides fast diffusion for lithium into tetrahedral sites at around
conversion of LiMn02 materials can lead to a less ordered spinel
ture with lower 4 V capacity and poorer rate capability. Additionally,
wo-phase process that occurs at around 3 V typically has much poorer
tics than the insertion process at 4 V, manifesting as polarization of the
ge and incomplete discharge as the current rate is increased. This also
s problems for practical constant-power applications that require higher
ent rates as the voltage lowers toward the end of discharge.
On the other hand, spinel-type cathodes formed in situ by cycling
n02 materials often show better cycling stability over a large lithiation
e (up to nearly LiMn02 or LizMn204) than spinel materials prepared ex
by conventional methods. The retention of high capacity- in many cases
e than 200 cycles have been reported - is surprising considering that
entionally prepared LiMn2 04 spinel materials almost always show large
ersible capacity loss over only a few cycles when cycled over both the 4
3 V regions. Data from high resolution electron microscopy have shown
the cycling stability may largely result from the formation of antiphase
domains (at a length scale of 20-50 nm) in cathode materials starting
12 13
LiMn02 phases. • The antiphase nanodomains arise from nucleation
he spinel structure at different points in the original crystal structure,
wed by growth of the spinel domains from the point of nucleation to
at phase boundaries where there is incoherence of the spinel crystal
ture between neighboring domains. The formation of this nanodomain
ture, in which there is a high density of phase boundaries, provides an
icity which can better accommodate transformation strains and
otropic changes in volume arising from cycling the material over such a
voltage range. Some cation inversion (Mn ions in tetrahedral Li sites)
also been found in the spinel structure obtained from LiMn0 2 phases,
13
h is also considered to contribute to the cycling stability, but would
an adverse effect on rate capability.
It has been pointed out that substitution of LiMn02 with species
ng strong octahedral site stabilization energies could hinder the second
e of the layered-to-spinel transformation, i.e. the stage that requires
7
erative re-arrangement of transition metal ions in the structure. This
oach has been shown to be partially successful for materials made by
soft chemistry routes and high temperature processing (see below).
.3. Alternative Layered Structures
In the materials described above, the transformation to spinel is

ble because the layered, orthorhombic, and spinel structures have the e
close-packed oxygen sublattice, so that only relatively minor cation re
gements are needed for the transformation. An alternative approach has
to prepare a layered lithium manganese oxide with a different oxygen
attice, so that transformation to spinel would require a coordinated
e rearrangement impossible to achieve at ambient temperature. Such a
14
rial can be prepared by soft chemistry routes, as described in Section
.4 below. The material with the different oxygen sublattice in this case is
nated an 02 structure following the notation for layered structures
15
n by Delmas et al., in which the letter refers to the type of site
pied by the alkali metal cation (in this case Li in octahedral sites), and
umber refers to the number of M02<-l layers included in the unit cell. By
ast the R 3 m layered structure is an 03 structure in which the oxygen
ing is essentially the same as in the stable spinel structure.
It has been shown experimentally that 02 layered lithium manganese
14
es do not convert to the spinel structure during cycling. However the
rials were at first found to have relatively poor current rate capability,
h was attributed to poor crystallinity caused by high densities of
ing faults, resulting from the ion-exchange process used to make the
rials. More crystalline 02 materials with improved electrochemical
14 16 17
erties have since been prepared using Ni and Co substitution. • •
The collapse of a layered to a spinel structure could also be hindered
ubstituting larger cations in the Li layers to "pillar" the oxide. Such an
oach has been described, for example using K ions as the pillaring
18
es. However a problem remains that species such as K ions tend to be
cted during the charge cycles.
Chapter 13
2.4. Layered Lithium Manganese Oxides with

Tetravalent Manganese
A lithium manganese oxide phase which is layered,

modynamically stable and easy to prepare, but until recently frequently
looked, is LhMn03. Unlike LiMn02, the manganese in LhMn03 is
valent. The structure of LhMn03 can be viewed as a rocksalt phase with
nating Li and [Li113Mn213] layers, allowing the formula to be written as
i 113Mn213]02. In this phase the [Li 113Mn213] layers show a
honeycomb ordering of the cations. The material can be prepared very
simply by temperature solid state processing in air. The layer
structure of n03 is stabilized compared with LiMn02 partly by the
smaller ionic
s of tetravalent Mn compared with trivalent Mn, partly by energy
ring effect of the ordering of Li ions in the [Li 113Mn213] layers, and also
ss strong magnetic interactions between Mn ions.
Li[Li 113Mn213]02 is, however, electrochemically inactive - the
ction of any lithium would normally require highly oxidized manganese h
is not observed in these types of structures. Li[Li 113Mn213]02 has been as
a precursor to prepare layered manganese oxides by acid leaching of
ithium out of the structure, and the resulting materials have been shown
19 20
able to reversibly insert lithium in lithium cells. ' In some cases the
rials seem to convert more slowly to a spinel-type phase than LiMn02
rials, which may be due to a unique protonated and hydrogen-bonded
1
structure formed during acid-leaching.Z However, the materials still
3 4
e on a Mn +/Mn + redox couple, which ultimately leads to spinel
ation. Furthermore, the acid-leached materials require lithiation before
can be used with a carbon anode.
Although tetravalent Mn cannot itself be oxidized in layered oxides, it
be combined with other oxidizable transition metals in solid solutions to
layered oxides having interesting properties as cathode materials. It is
just now beginning to be appreciated that this type of approach can be
ed to provide at least part of the benefits of manganese in terms of cost,
y and non-toxicity, while avoiding the instability issues associated with
n02 materials. As described above, the first stage of the transformation
yered LiMn022 to spinel4 is associated with the disproportionation of
lent Mn to Mn + and Mn + within the crystal structure, allowinMn to
2
ate through tetrahedral sites as Mn +. On the other hand, Mn + has a
high activation barrier for diffusion through the tetrahedral site.
red oxides containing only tetravalent Mn are therefore expected to be e
in the layered structure. Practical applications of this approach are
ribed in Section 13.4.
. LAYERED MANGANESE OXIDES BASED ON

LiMn02
.1. Orthorhombic LiMn02
The structure of the orthorhombic phase o-LiMn02 is sometimes

idered to be "layered", and the material will be discussed briefly here
use the LiMn02 stoichiometry means that it shares certain common
acteristics with other layered LiMn02 systems.
The synthesis and electrochemical properties of o-LiMn02 materials
been investigated in detail by various groups over the past decade. The
orhombic phase being the thermodynamically stable structural form for
LiMn02 stoichiometry, its preparation is relatively straightforward as
as excess oxygen is excluded. Any oxidation gives rise to spinel and
n03 phases. The electrochemical performance of o-LiMn02 materials is
ever very dependent on material microstructure, and therefore dependent
he synthesis approach. o-LiMn02 materials can be broadly categorized
those made by low temperature routes, and those made by high
erature processing.
Many low temperature synthesis methods have been reported, with the
ctive of increasing the electrochemical activity of the o-LiMn02
rial. Ohzuku et al. were the first to report a relatively high capacity of
mAh/g over a voltage range of 4.25- 2.0 V for an o-LiMn02 product
ared by heating an equimolar mixture of y-MnOOH and LiOH at 300-
°C under dry nitrogen.22 Reimers et al. also reported high capacities for
Mn02 products synthesized by an ion-exchange reaction of y-MnOOH
iling LiOH solution with subsequent drying at 105 °C in air, or heating
0 °C under argon.Z3 However, in all cases the crystal structure of the
rials was found to transform on charging the cell (delithiation of the
n02) to give a disordered crystal structure of the spinel type and a
arge voltage profile of the kind shown in Figure 13.3(A).
rochemical and X-ray diffraction data have shown that several charge
arge cycles are generally required to complete the transformation to the
l structure - depending on current rate and voltage limits the
formation is more or less fast. In particular, charging cells to higher
ge (more complete delithiation of the cathode) has been found to
24
ase the speed of the transformation. However the capacity retention on
ded cycling of the cathode materials formed from o-LiMn02 materials
ared by low temperature routes has generally been poor.
o-LiMn02 materials made using high temperature processing have
rally been prepared by solid state reaction of a manganese oxide and
m salt under an inert atmosphere. These o-LiMn02 materials also
form to spinel-type structures when delithiated during charging, but
Chapter 13
rally show better cycling stability over a wide voltage window than the
rials prepared at low temperature. Gummow et al. prepared o-LiMn02
ucts by solid state reaction of y-Mn02 with LiOH at 600- 620°C in an
n atmosphere in the presence of carbon black as a reducing agent, and
25
ned capacities of up to 160 mAhlg over 4.45- 2.5 V. Davidson et al.
ned capacities up to 140 mAhlg over 4.4 - 2.0 V for HT o-LiMnOz
26 prepared by reaction of Mn02 and Li2C03 at 800- 1000°C under
ucts
n. Croguennec et al. examined in detail the effect of crystallinity on
lectrochemical performance of HT o-LiMn02 phases, using solid state
ions between Mn203 and LiOH.H20 at 700°C designed to produce
7 28
rials with very small crystallites (<0.5 )..tm).Z • Capacities up to 200
29
/g were obtained for materials with crystallite sizes of around 0.3 )..tm.
small-crystallite-size o-LiMn02 phases were characterized by cationic
der identified as stacking faults in the crystal structure, which were
30
idered to be important for the electrochemical activity. However Jang
. later reported high capacities for o-LiMn02 materials prepared by a
olled high temperature synthesis process giving small crystallite size
low content of stacking faults, indicating that crystallite size is the most
al parameter in determining the electrochemical performance of o
31
n02 materials.
The problem for application of o-LiMn02 as a cathode material
ins its instability in transforming to a disordered spinel structure during
ng. To the best of our knowledge the stabilization of the o-LiMn02
ture against transformation to a spinel-type structure has never been
ted. The speed of transformation to spinel, and the rate capability
lem on the 3 V plateau, can both be controlled to some extent by
fying the microstructure of the o-LiMn02 material. However the sharp
ge drop in the middle of discharge remains a problem for applications.
.2. Lithium Manganese Oxide with the 03 Layered

Structure Prepared by "Soft Chemistry" Routes
Although LiMn02 with a layered-type crystal structure is metastable
respect to the orthorhombic phase, it has been prepared successfully by "
chemistry methods involving low temperature processing such as ion
ange. As explained in Section 13.2.1, the layered LiMn02 materials do
rystallize in the R 3 m space group, but show a monoclinic distortion of
attice (space group C2/m) due to cooperative ordering of Jahn-Teller
3
rted [Mn +06] octahedra. Monoclinic LiMn02 shows similar behavior to
Mn02 when used as a lithium-ion cathode. Spinel phase nucleates in the
rial on first charge, and transformation to a disordered spinel phase is
ly complete within a few charge-discharge cycles. The consequences
he discharge voltage profile and rate capability are analogous to those

d for the orthorhombic material.
Despite its large commercial and scientific interest, it took more than
ears of development before layered LiMn02 was successfully prepared.
e early 1980's Delmas et al. investigated ion exchange reactions and
ction/insertion in layered transition metal oxide bronzes.15 Two groups
kali transition metal bronzes were shown to exist. One group consists of
tures denoted 03, 01, 06 and P3, and the other of structures denoted P2
02. In the first group all M02 layers have the same orientation, while in
atter the orientation alternates. Phase transitions are possible between
bers of the same group by gliding of the layers, but the orientation of
ayers cannot change. The concept was successfully demonstrated using
32
rent LiCo02 and NaxC02 phases. ' 33 Due to the smaller ionic radius of
i-Na exchange is energetically preferred and in principle it should be
ghtforward to prepare layered LiMn02 (which has the 03 structure) by
xchanging Li for Na in 03 NaMn02 •
However, Li-Na exchange in molten salt was not successful. A
pse of the layered toward a three dimensional structure was reported
attributed to the use of relatively high temperatures for the exchange
34
ions. Success was eventually achieved by Capitaine et al. through ion
ange of NaMn02 in LiCl dissolved in methanol at 90°C, however the
exchange kinetics were poor, and the process required around one
1
h. More rapid and complete ion exchange was reported by Armstrong
Bruce by refluxing NaMn02 in LiBr, dissolved in n-hexanol, at 150°C.2
ever the layered LiMn02 material did not cycle well. A large amount of
m could be extracted on first charge, but subsequent discharges showed
and declining capacity. Later investigations of cycled material by on
diffraction and electron microscopy confirmed that the material rgoes
10
structural transformation to a disordered spinel over only a few s.
During the first charge cycle Li is initially removed in a two-phase ss
where both the original monoclinic layered and a hexagonal phase ist.
After approximately 30% of the lithium is removed, a single gonal
phase is observed up until almost complete delithiation of the n02• At
the very end of the first charge cycle a collapse of the interlayer ng was
observed. Subsequent cycling showed increasing amounts of
l phase appearing in the material.
Armstrong et al. have subsequently reported that partial substitution of
r Ni for Mn in the NaMn02 precursor can significantly improve the
ng properties of the LiMn02 material obtained by ion exchange.35 The
itution of around 10% Co or Ni results in a slightly Li-deficient
Mn 1_x0 2 material with a R 3m lattice symmetry rather than monoclinic
rtion. Although the materials still transform to a spinel structure with
ng, the transformation is slowed and the resulting materials show
Chapter 13
oved capacity, rate capability and cycling stability relative to non

tituted LiMn02 • Recent advances in the conditions required for synthesis
also been reported, including lower temperatures and the use of
37
native solvents such as ethanoe 6 or NMP. The results are interesting
the ion exchange process of manufacture remains a complex and
lenging process for large scale manufacture, and synthesis of the doped
um bronze precursor materials is not trivial in itself.
The structural variety of manganese oxides, and the small stability
rences in free enthalpy between the stable and metastable phases allow
a ty of different structures and intergrowth structures to be prepared by
temperature processing routes. Some reported products can be described
ayered structures. Tabuchi et al. reported hydrothermal synthesis of
3
red monoclinic LiMn02•
Whereas LiMn02 materials prepared at high temperatures often have
capability limitations, the low temperature materials may show much r
capacities and rate capabilities. This often originates from shorter
sion lengths for Li in such materials, which typically have lower
allinity and higher surface areas than materials sintered at high
erature. If the diffusion lengths are short, then lithium may be inserted
extracted at crystallographic sizes and into defects that would not be
ssible in crystalline materials, contributing to large capacities. On the r
hand such materials often have low volumetric densities, diminishing
nergy density of the cathode and the battery. Furthermore large surface
s can speed up unwanted reactions on the interface between electrolyte
cathode, leading to poor safety and poor cycling stability. These
lems can counter the often impressive capacities reported for cathodes
ared by low temperature routes. At present such cathode materials have
o be applied in commercial lithium-ion batteries.
.3. Substituted LiMn02 with the 03 Layered Structure

Prepared by High Temperature Processing
Chemical substitution in LiMn02 not only slows the transformation

the layered structure to spinel during cycling of the cathodes.
titution can also stabilize the layered structure over the orthorhombic
ture in the initial preparation of the LiMn02 material, allowing the
ed phase to be directly obtained using high temperature processing. As
ioned in Section 13.2.1, this occurs by perturbing the magnetic Mn-Mn
actions and/or by changing the electronic charge on the Mn ions. In
ice, as little as 3-5% substitution has been found to be sufficient to
bilize the orthorhombic structure in favor of the layered C2/m
oclinic structure. Based on theoretical considerations, a number of
ns should be able to perform this function, but only AI, Cr and Ga have
shown to work in practice. This is at least partly because the high
erature conditions using inert atmosphere are not suitable for the use of
tituting elements like Co and Ni, which tend to reduce to divalent oxides
g the attempted synthesis. AI, Cr and Ga have in common that the
lent oxidation state is stable under the synthesis conditions.
The stabilization of the layered structure over the orthorhombic
ture in a high temperature process was first reported by Davidson et al.,
38 3
Cr substitution. Substitution of as little as 5% of the Mn by Cr + in
n02 resulted in crystallization of a layered monoclinic phase by a simple
state reaction at 1000°C under argon. The original report was unclear in
ifying the crystal structure of the material, but subsequent work
rmed that it was indeed layered with the same monoclinic C2/m
ture reported for LiMn02 obtained by ion-exchange.39 It was reported
the Cr-substituted material seemed to convert to a spinel-type structure
cycling, but extended cycling data was not shown.
Later, Jang et al. reported that monoclinic layer-structure LiMn02
also be crystallized in a stable form at high temperature (ca. 950°C) by
ituting 5-10% of the Mn with AI.40 The materials were synthesized
a precursor powder prepared by atomizing and freeze-drying a
ension of Al and Mn hydroxide, coprecipitated from nitrate precursors,
OH solution. During cycling, the Al-substituted materials were again d
to convert to a spinel-type structure. Despite the transformation of the
ture and consequent evolution of the discharge voltage profile, however,
Al-substituted cathodes could be cycled over the voltage range 4.4 -
with good retention of capacity, and showed unusually good cycling
lity at 55°C.41
The formation of monoclinic layered oxides by high temperature
42 3
ssing was investigated in further detail by Ammundsen et al. .4
red monoclinic LiMn02 was successfully prepared by substituting
een 3 and 10% Al 3+, Ga3+, or Cr3+ in simple low-cost solid state44
ods of preparation which could be implemented on an industrial scale.
monoclinic unit cell parameters for the substituted materials showed a
matic decrease in unit cell volume and in the extent of the monoclinic
rtion by comparison with non-substituted LiMn02 prepared by ion
ange.
AI- and Cr-substituted materials were evaluated as cathodes to
mine whether stabilization of the layered crystal structure in a well
ed highly crystalline material through chemical substitution could
late to increased stability of the layered structure against transformation
42
inel during cycling. '43 Figure 13.3 shows the discharge voltage curves
monoclinic layered materials substituted with 5% AI and 3% Cr,
ared with the o-LiMn02 material prepared under the same high
erature processing conditions in the absence of a substituting element.
Chapter 13
o-LiMn02 cathode undergoes rapid conversion to spinel-type structure,

the 4 V and 3 V voltage steps clearly defined in the discharge by cycle
The results for Al-substituted LiMn02 prepared by the high temperature
41
state methods agree with those of Chiang et al., showing that
titution by AI serves to slow the transformation of the crystal structure
layered to spinel-type, but does not prevent it from occurring. The ts
for the Al-substituted material prepared at high temperature are
tatively similar to those observed for 03 layered materials prepared by
i ion-exchange in lattices substituted with Co or Ni.
The quite different evolution of the discharge voltage curves for Cr
tituted material compared with those for o-LiMn02 or the Al-substituted
0 50 100 150 (A) 200
o-LiMn0
2
(B)
4.0
AJ-doped man. LiMn02
>
-;; 3.5
g
C)
.l!l -----1·····
:
3.0
'iii
0 2.5 (C)
2.0 Cr-doped man. LiMn0
2
4.0
Q) 3.5
C)
3.0
'iii
0 2.5
2.0 50 100 150 200

0 Capacity (mAh!g)
13.3. Discharge voltage profiles for LiMn0 2 phases prepared by high temperature
sing: (A) o-LiMnOz, (B) monoclinic layered LiMn02 with 5% AI substitution,
onoclinic layered LiMn02 with 3% Cr substitution. Data recorded in lithium cells
between 4.4 and 2.0 V at 55°C, 30 mA/g current rate. Cycle numbers are indicated.
rial indicates that something different occurs during cycling. Little

tional capacity is evolved at 4 V for the Cr-substituted LiMn02, even
200 cycles, indicating little formation of spinel-type tetrahedral
rtion sites for Li ions. X-ray diffraction on cycled cathodes confirmed
43
absence of spinel phase in the cycled Cr-substituted materials.
sequent work by other authors has confirmed that Cr substitution is
liarly effective in blocking the transformation to spinel structure with
45
ing. .46 Instead a quasi-layered phase with some disordered
lacement of a portion of the Mn ions out of their original octahedral r
sites seems to form. This is similar to the intermediate structure
osed by structure calculations for a layered material which has
rgone the first stage of the transformation toward a spinel-type
ture, but which has been unable to complete the second stage of
dinated rearrangement of the cations to form a true spinel structure, as
ribed in Section 13.2.1.
To summarize, substituted 03 layered LiMn02 materials prepared at
temperature, like layered LiMn02 materials prepared by soft chemistry
es, also undergo structural phase changes with cycling to form defective
el-type structures, with consequent changes in the voltage profile.
titution with Cr has been demonstrated to be particularly effective in
biting transformation of the layered structure to spinel, but the Cr
tituted LiMn02 materials still undergo structural rearrangements away
an ideal ordered R 3 m structure. More work will be needed to
essfully stabilize a true layered LiMn02 structure by this approach.
3.4. 02 Layered Structures

As an alternative to preparing layered lithium manganese oxide with a
o02-type (03) crystal structure, it is possible to synthesize a layered
e with a different oxygen sublattice, so that transformation to spinel
ld require a coordinated lattice rearrangement impossible to achieve at
ient temperature. While the direct preparation of a lithium manganese
e with the latter kind of structure has never been reported, the structural
ty of sodium manganese bronzes allows such compounds to be made
g ion-exchange methods.
Preparation of lithium manganese oxides with 02 layered structures
first reported by Paulsen et al., who synthesized materials by ion
anging P2-structure layered sodium manganese oxides ("P" in this case
14
nating prismatic Na sites). However, although it was demonstrated
the 02 materials do not convert to spinel during cycling, the materials
theless cycled poorly with insufficient rate capability. This was
uted to poor crystallinity caused by high densities of stacking faults
Chapter 13
ting from the ion-exchange process. During Na-Li exchange, the Mol>
ts glide with respect to one another, creating the stacking faults.
However substitution of a fraction of the Mn with Co or Ni improves
the ion exchange kinetics and the electrochemical properties. A layered
rial of composition Li2,3 [Mn0.85Co0.15]0z showed a smoothly sloping
ge profile and good cycling performance without transforming to
14
el. Substitution with Ni allowed a successful ion exchange at room
erature resulting in a crystalline layered phase with a novel T2 type
16
ture. T2 Lizi3[Ni 113Mnz,3]0z shows Ni 113Mnz,3 honeycomb type
ring in the transition metal planes and unusual occupation of tetrahedral
47
layer sites by Li ions. On first charge the material converts to the 02
ture. The material can then be cycled in the 02 structure over hundreds
cles without collapse of the layer structure, with capacities of up to 180
lg if cycled between 2.5 and 4.6 V. However, unlike the Co-substituted
rial, only one-third Li per formula unit Li213 [Ni113Mn213 ]02 (i.e. half the
able Li) can be removed from the lattice on charge. In order to obtain a
city of 180 mAhlg, an additional 1/3 Li needs to be inserted into the
e on discharge (requiring an additional source of Li not normally
able in lithium-ion cells). Furthermore, although the 1/3 Li removed on
charge is reversibly re-intercalated at around 3.8 V, the additional 113 Li
ercalated at around 2.8 V. Although the material does not convert to a
l, the voltage profile is almost the same as that of 03 phases that do
ert to spinel. In the case of Li 213 [Ni113Mn 213]02 the two steps with a 1 V
in the middle are due to different redox couples being involved. The 3.8
ercalation is related to reduction of nickel, while the 2.8 V intercalation
sponds to reduction of Mn4+ to Mn3+.
Substitution of Co for Ni in Li213 [Nil/3Mnz,3]0z has been found to
t in formation of a stacking faulted 02 structure with incomplete Li-Na
ange.48.49 However substitution of Co for both Ni and Mn allows ion
ange to go to completion to give materials of composition Li213 [CoxNi113.
213.x12]02, and allows increased amounts of Li to be extracted on first
e, enlarging the reversible capacity available in a sloping discharge
17
een 4.5 and 3.8 V.
The 02 layered cathode materials are still far from commercialization
o the complicated methods of preparation, and lower capacity at higher
nt rates in their current state of development. Although the substitution i
and/or Co improves the rate performance, the resulting T2 and 02
rials need to be cycled over a wide voltage range and require an
ional source of Li to capture their full capacity, which is delivered in
teps with a large voltage drop mid-discharge.
The structural studies carried out on the 02 and T2 layered oxides do,
ver, provide interesting data on the behavior of Mn, Ni and Co ions
sharing the same oxide lattice. In particular, the data show that best
tural and electrochemical results are achieved when Mn is fully
valent in the initial material. Ni in the T2 layered manganese

pounds is divalent, while Co is trivalent. These observations have
ng on the subject matter of Section 13.4.
. LAYERED SOLID SOLUTION CATHODE

MATERIALS BASED ON TETRAVALENT
MANGANESE
.1. Overview of Solid Solution Approaches
In order to facilitate the preparation of layered LiMn02 and to attempt

prove the stabilization of the layered structure of the delithiated phases st
transformation to spinel, there have been many investigations of er
levels of substitution by elements such as Cr, Co and Ni. This type of
oach overlaps to some extent with that of investigators who have tried to
r the cost and improve the safety properties of LiCo02 and LiNi02 by
ily substituting these materials with manganese. The underlying idea
d all these different approaches is the same: to combine the cost and
y advantages of manganese oxide with the thermodynamic preference of
LiM02 phases to stabilize in a layered structure and, in the case of Co
Ni, to include something of the electrochemical performance of the
02 phases.
The last few years have seen an increasing emphasis on layered
plex oxides of manganese with Cr, Co and Ni which are "lithium-rich",
here the molar ratio of lithium to transition metal is higher than one. A
ral trend is that "lithium-rich" solid solution compositions are easier to
are than stoichiometric LiMn 1_xMx02 compounds, and the "lithium-rich"
rials have improved electrochemical properties. It is now clear that
properties, which include improved rate capability and better structural
lity on cycling, are associated with a stable tetravalent state for
anese.
The ability of the layered rock salt structure of complex oxides to
rb an excess of Li ions is associated with Mn, because Mn is able to
t a tetravalent oxidation state in the as-prepared compounds to
ensate the charge. The excess Li ions appear in the transition metal
, as observed for LizMn03. The chemistry of these lithium-rich layered
s therefore relates to the crystal chemistry of Li2Mn03• As already
ined in Section 13.2.4, LizMn03 is an easily prepared layered oxide
113Mn 213]02 in which cation planes alternate as lithium layers and layers
h are one-third lithium and two-thirds manganese. Ammundsen and
50
en proposed a useful classification of solid solution layered oxides
on manganese according to their position within the triangle
Chapter 13
i 113Mn213]02 - LiMn02 - LiM02, shown in Figure 13.4. Solid

solutions he line between LiMn02 and LiM02 (i.e. Li[MnxM1_x]02), where
the ansition metal ratio is equal to one, are denoted
"lithium hiometric". Solid solutions on the line between Li 2Mn03 and
LiM02
Li[MnxLix12M 1_3x12]02), where the maximum amount of Li is
rporated into the transition metal layers, are denoted "lithium saturated".
d solutions within the triangle (i.e. Li[LiyMnxMI-x-y]02 with y<x/2),
e different amounts of Li are incorporated into the transition metal
s, are called "lithium-rich". The metal M in LiM02 can be a single
ies, or a mixture of transition metals such as Ti, Cr, Co, Ni or Fe, and/or
transition metals such as Mg, Al, or Ga. A number of these systems will
escribed in this section.
lithium lithium
stoichiometric saturated
phases phases
LiMn02
e 13.4. The Li2Mn03 - LiMn02 - LiM02 solid solution system showing the
fication of phases according to their lithium stoichiometry.
All compounds within the phase triangle shown in Figure 13.4 obey
ules described by Dahn et al. for these layered rock salt solid solution
51
es. Firstly, the sum of the cation occupations on the sites in the
ition metal layers (the 3b sites in the space group R 3m) is equal to one.
ndly, the sum of the cation oxidation state times the cation occupations
e transition metal layer sites equals three.
It should be noted that the results of different investigations into solid
ions over the past decade have shown considerable variation, depending
gly on the conditions used to prepare the materials. These variations can
to some confusion about the intrinsic performance of certain

istries. A general trend in complex layered oxides based on solid
ions is that solid state reactions using simple binary oxides or salts as
ng materials often do not lead to the desired phases or to phases with
esired electrochemical activity. Increasingly, more complex synthesis
ods are applied, such as sol-gel, coprecipitation, or ball milling
ods, which allow the preparation of well-mixed solid solution phases
superior electrochemical performance.
Another important factor influencing the outcome of any solid
ion preparation is the conditions of temperature and atmosphere used.
solutions based on trivalent Mn tend to only be achievable using
en-free atmospheres, but oxygen-free atmospheres are not normally
conditions for forming materials with trivalent Co or Ni. On the other
, solid solutions based on tetravalent Mn are easily achieved in air
spheres, up to 1000°C.
The examples that follow have in common that the as-prepared
rials contain tetravalent Mn in combination with a high proportion of
or more other, redox-active, elements. In these materials the Mn remains
e in the tetravalent state during charge and discharge, while the other
ent(s) change(s) oxidation state. The situation is therefore very different
the compounds described in preceding sections, which depend on the
sible oxidation of Mn3+ to Mn4+ for their activity as lithium-ion cathode
rials. The Mn in the materials described in this section therefore takes a
ve role as a framework element, providing an inert and stable matrix for
verall material. The resulting materials benefit from the manganese
e component in terms of cost, thermal stability and safety, without
ring from the structural instabilities associated with trivalent Mn. This
oach of developing solid solution materials based on tetravalent
ganese therefore shows highest promise for practical application of the
red manganese oxide" concept.
.2. Solid Solutions with LiCr02

One of the first demonstrations of the improvement in electrochemical
erties that can be gained through "lithium-rich" preparations was in
solutions with LiCr02•
As described in Section 13.3.3, Davidson et al. showed that single
e layered materials could be made in the solid solution (1-x)LiMn02 -
r02 for any value of x down to x=0.05, providing the preparations are
under oxygen-free atmospheres?8•39 Practically the whole solid
ion range is achievable because the mechanism is a simple substitution
n3+ for Cr3 +. Compounds with x<0.7 crystallize in the monoclinic C2/m
Chapter 13
red structure, showing clearly that the Mn in these compounds is

lent, producing a Jahn-Teller cooperative distortion. Only compounds
x?.O.7 adopt the layered R 3m structure of LiCr02, due to sufficient
ion of Mn3+ ions to prevent cooperative ordering of the distorted Mn06
hedra. (Dahn et al. have shown that materials with 0.5<x<0.75 can be
ared with hexagonal layered structures
52 by using sol-gel precursors and
r synthesis temperatures of 700°C, but it is likely that these lower
erature preparations result in cation vacancies compensated by some
+).
Cathode materials in the solid solution LiMn02 - LiCr02 undergo
ly complete delithiation on first charge, similar to that observed for the
oclinic LiMn02 materials described in Section 13.3. This indicates that
Mn and Cr are oxidized from 3+ to 4+ states in these materials. During
charge any monoclinic phase is first converted to hexagonal phase, and
the hexagonal phase is delithiated in a single-phase process. The
ppearance of monoclinic phase in the first part of charge suggests that
3
+ is oxidized before Cr +. Materials of composition around the mid
ber of the LiMn02 - LiCr02 solid solution, LiCr 05Mn0.502, can
be ed with reversible capacities around 150 mAh/g between 4.4 and 2.5
V, ered in a sloping discharge profile with an average voltage of about
3.3 owever the capacity drops off rapidly to below 100 mNg when
current are increased to provide discharges over two hours or
less. hermore, the voltage profile evolves with cycling in a manner
similar to described in Section 13.3.3 for lightly Cr-substituted
LiMn02. xMn 1_x02 materials with compositions x?.0.75 were found to
have more e voltage profiles with cycling, but cells containing cathodes
based on compositions show large impedance and lower capacities, <100
mAh/g
at low current rates.
The substitution of additional lithium ions into solid solutions with
r02 to give lithium-rich compositions changes and improves the
tural and electrochemical properties of the materials.53·54 Results are
cularly interesting for lithium-saturated compositions in the LizMn03 -
r02 system. In particular the stability of the crystal structure with cycling
arkedly improved over "stoichiometric" LiMn 1_xCrx02 compositions,
iding a monotonic discharge voltage profile that does not evolve with
ng. Detailed structural studies of the mid-member of the lithium
ated solid solution LizMn03- LiCr02 (i.e. Li[Lio.2Cro.4Mno.4]02), using
S and NMR, have shown that the material contains both LizMnOrlike
LiCrOz-like structure at the local level, although the material is single
e to X-ray diffraction.55·56
What was initially surprising is the high reversible capacities which
be achieved for lithium-saturated Li2Mn03 - LiCr02 materials, up
20 mAh/g in the voltage range 4.4- 2.5 V. Figure 13.5 shows charge
discharge curves for the first two cycles of a Li[Li0.2Cr0.4Mn0.4]02

rial. It is clear from these curves that an irreversible process occurs on
irst charge, giving rise to around 25% capacity loss that is not recovered
he first discharge. Subsequent cycling however is nearly 100%
ient, with very little capacity loss or change in voltage profile over more
200 cycles, even at
55°C.55
X-ray absorption spectroscopy data have shown that the high capacity
ned from these materials is due to Cr being oxidized to higher than a 4+
during charge.55 57 ' Cycling oxidizes and reduces Cr ions between 6+
3+ states, while the LizMn03 component acts as an inert matrix. The Mn

ins tetravalent throughout cycling, which means that Mn ions remain
ed into the LizMnOrtype structure throughout charge and discharge.
situation is different from LiMn02 - LiCr02 solid solutions, where the
nce of Mn 3+ allows Mn to migrate out of the transition metal layer sites
g cycling. In fact, in Li2Mn03 - LiCr02 materials it is likely that the
Cr are the mobile species, owing to the large changes in oxidation
state.
4.5
Q)
Cl 3.5
ctl
15
> 3
2.5
0 50 100 150 200 250
Capacity (mAh/g)
e 13.5. First and second cycle charge-discharge voltage curves for a material of
osition Li[Li0.2Cr0.4Mno.4]02 cycled between 4.4 and 2.5 V (3 mA/g current rate, 25°C).
ability of the LizMn03-type layered matrix to support reversible

ation of Cr between 3+ and 6+ states without significant loss of
tural integrity is quite remarkable, and illustrates a significant advantage
sing LizMn03 as a base for solid solution compounds rather than
n02•
The capacity retention of LizMn03 - LiCr02 materials when cycled
at illustrates another interesting feature of materials based on
tetravalent
r than trivalent manganese. The poor cyclability of manganese oxides at
Chapter 13
temperatures is frequently attributed to manganese dissolution through

chanism that requires Mn3+ to be present at the surface of the material.
izMn03 is used as the solid solution, trivalent Mn ions are not present
the material is less susceptible to dissolution processes.
4.3. Lithium-Rich Solid Solutions with LiCo02
Since LiCo02 is the dominant commercial lithium-ion cathode

uct, the solid solution LiCo02 - LiMn02 is of obvious interest. LiCo02
nown to stabilize the layer structure and significantly improve the
rochemical properties of LiNi02• Beneficial effects of Co substitution
he stability and rate capability of monoclinic LiMn02 prepared by low
erature soft chemistry methods have also been reported? 5 However a
modynamically stable layered material prepared at higher temperatures
d on a LiCo02 - LiMn02 solid solution has never been reported. This
is probably due to phase separation of spinel lithium manganese oxide
es when preparations are made in air, or of CoO when preparations are
e in inert atmospheres.
On the other hand, using Li2Mn03 as the base of the solid solution
ws stable single-phase materials to be prepared by simple high
erature processing in air.58• 59 This is because although compounds in
h both Co and Mn are trivalent cannot be prepared, LiCo02 can be
ared with tetravalent Mn in combination with charge-compensating Li in
ransition metal layer.
However the solid solution LizMn03 - LiCo02 differs in one
rtant regard from the LizMn03 - LiCr02 system. Unlike Cr, the
ation state achievable by Co on charge is limited to Co4+. This represents
sic limitation in realizable capacity for the LizMn03 - LiCo02 system.
he mid-member Li[Li 0.2Co0.4M11o.4]02 the maximum capacity that could
xpected would be around 125 mAh/g, corresponding to 0.4 moles of Li
cted, at which point both Co and Mn would be in their maximum
evable oxidation states of 4+. The achievable capacities reported in
ice have been lower, and the system is therefore of less interest for
ical applications.
.4. Solid Solutions with LiNi02 and LiCoxNh-x02
Unlike the LiCo02 - LiMn02 system, single-phase compositions can

repared in the (1-x)LiNi02 - xLiMn02 system up to x=0.5. Solid
ion materials of this type were first investigated by Rossen et al.,f!J who
rted reversible capacities of around 150 mAh/g in the voltage range
2.5 V, delivered in a smooth sloping discharge profile, although with

61
ficant rate limitations. Spahr et al. described LiMn0.5Ni0.502 prepared
oxidative coprecipitation process which showed similar characteristics,
ycling data reported for the material showed poor capacity retention.
uku et al.62 and Lu et al.63 have recently reported data for layered
n0.5Ni 0502 materials prepared from coprecipitated precursors of Ni and
salt, calcined and then sintered at temperatures of 1000°C and 900°C,
ctively. The resulting materials again show smooth discharge voltage
les, and reversible capacities between 150 and 180 mAh/g, depending
harge voltage cutoff and discharge current rate.
It is interesting to understand why compositions in the solid solution
i02 - LiMn02 can be prepared up to .x=0.5, while LiCo02 -
LiMn02 rials cannot be synthesized. It now seems that the answer
lies in rent oxidation states that Co and Ni ions prefer to adopt when
sharing iM02 lattice with Mn. Recent theoretical calculations [64], as
well as
y absorption and NMR data,65 have shown that the Ni in LiMn 05Ni 0.502
ainly divalent, while the Mn is tetravalent. Substitution of Mn into
i02 therefore takes place with Mn going into the lattice as Mn 4+,
ring reduction of an equivalent number of Ni3+ ions to Ni2+. This
ess explains why substitution is limited to a maximum Mn/Ni ratio of 1.
umably Mn cannot be easily substituted into LiCo02 because the
/Mn4+ oxidation states are not adopted. The chemistry of the LiMn02 -
02 system is also different from that of LiMn02 - LiCr02, where Mn
3
+
titutes for Cr3+ across the full solid solution range. And because Mn
4 3
titutes into LiNi02 as Mn + rather than Mn +, the resulting as-prepared
rials retain hexagonal symmetry, whereas stoichiometric materials in
LiMn02 - (1-x)LiCr02 solid solution adopt a monoclinic
symmetry
e .x=0.3 due to the trivalent Mn ions.
4
The important consequence of the Mn +/Ne+ oxidation state of
n0_ 5Ni 0502 is that Mn remains stable in the tetravalent state during
ng of the material. The reversible capacity is due to oxidation and
ction of Ni between the 2+ and 4+ states. LiMn 05Ni 0.502 materials can
fore be cycled without conversion from a layered to a spinel-type
al structure. As with LhMn03 - LiCr02 solid solution materials,
n 05Ni 0.502 remains kinetically stable against transformation to spinel
use Mn is not present in oxidation states lower than 4+. An advantage of
LiMn02 - LiNi02 system is that oxidation of Ni 2+ to Ni4+ can
occur
out changes in the Ni site, whereas the oxidation of Cr 3+ to Cr6+ results
splacements of Cr due to the preference of Cr6+ for a tetrahedral rather
an octahedral coordination.
Another interesting feature of the LiMn05Ni 0.502 materials is that the
ge curves are almost identical to those of LiNi02, although different
x couples are active. In LiMn 0.5Ni 0502, the charge and discharge
4
cities can only be explained by a reversible oxidation of Ne+ to Ni +,
Chapter 13
reas in LiNi02 Ni cycles between 3+ and 4+ oxidation states. Hence the

ntial for LiMn 0.5Ni 0.502 might be expected to be lower than for LiNi02.
n if Mn participated in the redox process the higher 3 4
potential would be
cult to explain, since the redox potential of Mn +/Mn + is below that of
4
/Ni +. Theoretical2 calculations have suggested that the surprisingly high
4
ntial for the Ni +/Ni + redox in LiMn0.5Ni 0.502 may be due to changes in
64
Mn-Ni interactions as the oxidation states change from 4+12+ to 4+14+.
4 2
e lattice containing Mn + and Ni + the Mn-Ni interactions are essentially
4 4
ctive, while they are repulsive for Mn +-Ni +, effectively introducing an
tional energy term for the intercalation of Li that would increase the
ntial.
Materials of composition LiNi0.5Mn0.502 have however proven
cult to prepare reliably, and often show evidence of spinel lithium
2
ganese oxide secondary phases. It is also difficult to prevent Ni + ions
appearing in the Li layers, which degrades the performance of the
rials as cathodes. Incorporating excess Li has been shown to address
63 66 67
these problems. ' ' The main benefit of lithium rich compositions is
mprovement in the ordering of the layered structure compared with the
hiometric materials, i.e. less transition metal (mainly Ne+) on lithium
rs. Best results have been achieved for phases with a limited amount of
ss lithium. In the LizMn03 - LiMn02 - LiNi02 system there is basically
lance between increasing the quantity of LizMn03, to obtain improved
r character, and not adding so much that the capacity of the material is
red below practical values. On the other hand, an interesting feature of
ithium-rich compositions is that if the cathodes are first charged to a
ively high voltage, an irreversible plateau appears at around 4.5 V at
cities
4 63
beyond that which can be accounted for by oxidation of all the Ni
i +. The length of the plateau seems to be correlated to the amount of
um in the transition metal layer. The plateau disappears during
equent cycles, but allows higher capacities to be achieved in subsequent
ng by reversible insertion of some additional Li around 3.3 V,
lating with reduction of a portion of the Mn + to Mn3+.
4
Solid solutions in the LizMn03 - LiMn02 - LiNi02 system have

also
investigated when Co is introduced. ' 66' 67 In fact, even though LiCo02
51
MnOz materials cannot be prepared, it is possible to substitute Co into

m-stoichiometric LiMn02 - LiNi02 materials. The result is particularly
tive when the Mn/Ni ratio is maintained equal to 1, giving materials of
bility, and improved cyclability. In this case the Co serves a similar role
does in LiCoxNi 1_x02 materials. Ohzuku et al. have shown particularly
electrochemical results for the composition Li[Co113Mn113Ni 113]02,
e it is assumed 62that 69 the transition metal ions are ordered in a
erlattice" structure. ·
Paulsen et al. have shown that lithium-rich formulations in the
Mn03- LiMn02 - LiNi02 system also benefit from addition of Co.
66 67
·
4.5
Q)
Ol 3.5
.!9
0
> 3
2.5
0 50 100 150 200

Capacity (mAh/g)
e 13.6. First and second cycle charge-discharge voltage curves for a material of
osition Li[Lio.05Co0.15Mno.4Ni0.4 ]02 cycled between 4.4 and 2.5 V (30 mA/g current rate,
.
e complex oxides contain Co +, Mn +, Ni + and Ne+. As with lithium

3 4 2
hiometric compositions, good electrochemical results have been

ned for materials in which the Mn!Ni ratio is near to 1. Again, the effect
e Co is to further improve the layered structure, and consequently the
sible capacity, rate capability, and cyclability. Representative first cycle
ge and discharge profiles are shown in Figure 13.6 for a lithium-rich
position comprising approximately 15% Co.
Materials based on the LhMn03 - LiMn02 - LiNiOz solid
solution m show attractive characteristics for practical applications as
cathodes
thium-ion batteries. The manganese component inherently lowers the
of production relative to cobaltates and nickelates. The capacities which
be achieved in a lithium-ion cell are close to those of LiCoOz in a similar
ating voltage range. Unlike LiCo0 2 and LiCoxNi 1_x02 materials which
ire strict voltage control to prevent charge above 4.2 V for safety and
lity reasons, it has been shown that cells containing cathode materials in
LhMn03 - LiMn02 - LiNi02 solid solution system, both with and
out addition of Co, can be charged to higher voltages without
promising safety, thereby allowing higher capacities to be achieved. The
mal stability of the charged cathodes in contact with electrolyte has been
51 63 66 68
onstrated by both DSC and ARC measurements. • • - Remarkably, as
as 5% [Mn0.5Ni 0.5] in LiCo02 has been found to significantly improve
thermal stability relative to LiCo02. 68 The impressive safety
acteristics, derived from the Mn component, make these types of
rials particularly attractive for use in larger battery formats.
The impressive electrochemical properties of these types of materials,
edly superior to that currently provided by LiMn02 materials, means
Chapter 13
they are likely to be the first "layered manganese oxide" materials to be

in commercial applications.
.5. Further Work
The advantages of preparing solid solution materials based on

valent rather than trivalent manganese have been demonstrated for a
ber of compositions. In these materials Mn has given up its role as an
rochemically active center and is present as a stabilizing framework
ent. Based on this concept, other innovative cathode materials could be
neered. It has been pointed out that other elements that can take on the
xidation state in layered oxides might also be selected on the basis of
weight, processability and non-toxicity, as a substitute for Mn, taking
64
ole of stabilizing component. An example of such an approach has
reported using a small amount of Li2Ti03, which is isostructural with
0
n03, to stabilize layered LiMn0.5Ni 0.50z?
. SUMMARY AND OUTLOOK
Technological interest in manganese-based cathode materials will

nue as battery manufacturers seek to distinguish their products in an
asingly competitive market by lower pricing, improved safety and
r toxicity, as well as increased energy density and performance.
hermore, over the next few years manufacturers will seek to deliver
m-ion batteries into new markets beyond portable electronics, and in
cular into markets requiring larger cells and batteries. The cost and
y requirements of large cells are likely to require a manganese-based
ode solution. Although intensive efforts are still under way to
mercialize batteries using spinel lithium manganese oxide, alternative
ganese-based materials that can offer improvements in capacity and
ng stability will be required to be competitive in commercial
cations.
Although layered LiMn02 materials offer the higher capacity, their
r drawback remains the structural instability against transformation
rds disordered spinel structures, low density, and/or poor rate capability. t
substitution helps to improve the electrochemical properties, but the
amental instability associated with the trivalent Mn is still a problem.
Solid solution materials using a stable tetravalent manganese oxide
ix to enhance the structural stability during cycling represent the most y
near-term solution for commercialization of the "layered lithium
ganese oxide" concept. In particular, layered complex oxides in the
Mn03 - LiMn02 - LiNi02 system, with or without added

cobalt, strong near-term promise with competitive capacities, good
cycling lity, and smooth voltage profiles with high average voltage.
These rials offer significant cost and safety advantages over materials
such as
o02 and LiCoxNiJ_xOz.
FERENCES
. Capitaine, P. Gravereau, C. Delmas, Solid State lonics 89 (1996) 197.
. R. Armstrong, P. G. Bruce, Nature 381 (1996) 499.
M. Tabuchi, K. Ado, H. Kobayashi, H. Kageyama, C. Masquelier, A. Kondo, R. Kanno,
. Electrochern. Soc. 145 (1998) lA9.
. Ceder, S.K. Mishra, Electrochern. Solid-State Lett. 2 (1999) 550.
. Ceder, A. Vander Ven, Electrochirn. Acta. 45 (1999) 131.
. Wolverton, A. Zunger, J. Electrochern. Soc. 145 (1998) 2424.
. Reed, G. Ceder, A. VanDer Ven, Electrochern. Solid-State Lett. 4 (2001) A78.
. Vitins, K. West, J. Electrochern. Soc. 144 (1997) 2587.
. M. Koetschau, J. R. Dahn, J. Electrochern. Soc. 145 (1998) 2672.
Y. Shao-Horn, S. A. Hackney, A. R. Armstrong, P. G. Bruce, R. Gitzendanner, C. S.
ohnson, M. M. Thackeray, J. Electrochern. Soc. 146 (1999) 2404.
.-J. Hwang, H.-S. Park, J.-H. Choy, G. Campet, Chern. Mater. 12 (2000) 1818.
Y.-I. Jang, B. Huang, H. Wang, D. R. Sadoway, Y.-M. Chiang, J. Electrochern. Soc. 146
1999) 3217.
H. Wang, Y.-I. Jang, Y.-M. Chiang, Electrochern. Solid State Lett. 2 (1999) 490.
. M. Paulsen, C. L. Thomas, J. R. Dahn. J. Electrochern. Soc. 146 (1999) 3560.
. Delmas, J. Braconnier, A. Maazaz, P. Hagenmueller, Rev. Chirn. Miner. 19 (1982)
43.
. M. Paulsen, C. L. Thomas, J. R. Dahn, J. Electrochern. Soc. 147 (2000) 861.
. Lu, R. A. Donaberger, C. L. Thomas, J. R. Dahn, J. Electrochern. Soc. 149 (2002)
l083.
. Chen, M.S. Whittingham, J. Electrochern. Soc. 144 (1997) L64.
M. H. Rossouw, M. M. Thackeray, Mater. Res. Bull. 26 (1991) 463.
. S. Johnson, S. D. Korte, J. T. Vaughey, M. M. Thackeray, T. E. Bofinger, Y. Shao
orn, S. A. Hackney, J. Power Sources 81-82 (1999) 491.
. Paik, C. P. Grey, C. S. Johnson, J.-S. Kim, M. M. Thackeray, Chern. Mater. (2002) in
ress.
. Ohzuku, A. Ueda, T. Hirai, Chern. Express 7 (1992) 193.
. N. Reimers, E. W. Fuller, E. Rossen, J. R. Dahn, J. Electrochern. Soc. 140 (1993) 3396.
Koetschau, M. N. Richard, J. R. Dahn, J. B. Soupart, J.-C. Rousche, J. Electrochern.
oc. 142 (1995) 2906.
. J. Gummow, D. C. Liles, M. M. Thackeray, Mater. Res. Bull. 28 (1993) 1249.
J. Davidson, R. S. McMillan, J. J. Murray, J. E. Greedan, J. Power Sources 54 (1995)
32.
. Croguennec, P. Deniard, R. Brec, A. Lecerf, J. Mater. Chern. 5 (1995) 1919.
. Croguennec, P. Deniard, R. Brec, P. Biensan, M. Broussely, Solid State lonics 89
1996) 127.
. Croguennec, P. Deniard, R. Brec, J. Electrochern. Soc. 144 (1997) 3323.
Chapter 13
. Croguennec, P. Deniard, R. Brec, A. Lecerf, 1. Mater. Chern. 7 (1997) 511.

Y.-1. Jang, B. Huang, H. Wang, D. R. Sadoway, Y.-M. Chiang, ECS Meeting Abstracts
Vol. MA 98-2 (1998) p.132.
C. Delmas, J. Braconnier, P. Hagenmueller, Mater. Res. Bull. 17 (1982) 117.
A. Mendiboure, C. Delmas, P. Hagenmueller, Mater. Res. Bull. 19 (1984) 1383.
. Fuchs, S. Kemmler-Sack, Solid State lonics 68 (1994) 279.
A. R. Armstrong, R. Gitzendanner, A. D. Robertson, P. G. Bruce, Chern. Cornrnun.
1998) 1833.
. E. Quine, M. J. Duncan, A. R. Armstrong, A. D. Robertson, P. G. Bruce, 1. Mater.
Chern. 10 (2000) 2838.
. Coowar, PCT International Publication No. WO 00/15557 (2000).
. J. Davidson, R. S. McMillan, J. J. Murray, 1. Power Sources 54 (1995) 205.
. J. Davidson, R. S. McMillan, H. Slegr, B. Luan, I. Kargina, J. J. Murray, I. P.
wainson, 1. Power Sources 81-82 (1999) 406.
.-1. Jang, B. Huang, Y.-M. Chiang, D. R. Sadoway, Electrochern. Solid-State Lett. I
1998) 13.
.-M. Chiang, D. R. Sadoway, Y.-1. Jang, B. Huang, H. Wang, Electrochern. Solid-State
ett. 2 (1999) 107.
. Ammundsen, J. Desilvestro, T. Groutso, D. Hassell, J. B. Metson, E. Regan, R.
teiner, P. J. Pickering, in New Materials for Batteries and Fuel Cells, D. H. Doughty, L.
. Nazar, M. Arakawa, H.-P. Brack, K. Naoi, Eds., Mat. Res. Soc. Syrnp. Proc. Vol. 575,
Materials Research Society (2000), pp 49-58.
. Ammundsen, J. Desilvestro, T. Groutso, D. Hassell, J. B. Metson, E. Regan, R.
teiner, P. J. Pickering, 1. Electrochern. Soc. 147 (2000) 4078.
. Ammundsen, W. Gao, D. Hassell, Z.-Y. Liu, J. B. Metson, R. Steiner, P. J. Pickering,
CT International Publication No. WO 00123380 (2000).
.-J. Hwang, H.-S. Park, J.-H. Choy, G. Campet, 1. Phys. Chern. B 104 (2000) 7612.
. Choi, A. Manthiram, 1. Electrochern. Soc. 149 (2002) A1157.
. M. Paulsen, R. A. Donaberger, J. R. Dahn, Chern. Mater. 12 (2000) 2257.
. Lu, R. A. Donaberger, J. R. Dahn, Chern. Mater. 12 (2000) 3583.
. Lu, J. R. Dahn, 1. Electrochern. Soc. 148 (2001) A237.
. Ammundsen, J. M. Paulsen, Adv. Mater. 13 (2001) 943.
. Lu, D. D. MacNeil, J. R. Dahn, Electrochern. Solid-State Lett. 4 (2001) A200.
. R. Dahn, T. Zheng, C. L. Thomas, 1. Electrochern. Soc. 145 (1998) 851.
. Storey, I. Kargina, Y. Grincourt, I. J. Davidson, Y. Yoo, D. Y. Seung, 1. Power
ources 97-98 (2001) 541.
. Grincourt, C. Storey, I. J. Davidson, 1. Power Sources 97-98 (2001) 711.
. Ammundsen, J. Paulsen, I. Davidson, R.-S. Liu, C.-H. Shen, J.-M. Chen, L.-Y Jang,
.-P. Lee, 1. Electrochern. Soc. 149 (2002) A431.
. Pan, Y. J. Lee, B. Ammundsen, C. P. Grey, Chern. Mater. 14 (2002) 2289.
. Balasubramanian, J. McBreen, I. J. Davidson, P. S. Whitfield, I. Kargina, 1.
lectrochern. Soc. 149 (2002) A176.
. Numata, C. Sakaki, S. Yamanaka, Chern. Lett. (1997) 725.
. Numata, C. Sakaki, S. Yamanaka, Solid State lonics 117 (1999) 257.
. Rossen, C. D. W. Jones, J. R. Dahn, Solid State lonics 57 (1992) 311.
. E. Spahr, P. Novak, B. Schnyder, 0. Haas, R. Nesper, 1. Electrochern. Soc. 145
1998) 1113.
. Ohzuku, Y. Makimura, Chern. Lett. (2001) 642.
. Lu, D. D. MacNeil, J. R. Dahn, Electrochern. Solid-State Lett. 4 (2001) A191.
. Reed, G. Ceder, Electrochern. Solid-State Lett. 5 (2002) A145.
W.-S. Yoon, Y. Paik, X.-Q. Yang, M. Balasubramanian, J. McBreen, C. P. Grey,

lectrochem. Solid-State Lett. 5 (2002) A263.
. M. Paulsen, B. Ammundsen, L. Kieu, European Patent Application No. EP 1 189 296
2 (2002).
. M. Paulsen, B. Ammundsen, S. Argue, II'hiMLB, Abstract 208, Monterey, CA (2002).
. D. MacNeil, Z. Lu, S. D. Beattie, G. D. Murphy, J. R. Dahn, JJ'hiMLB, Abstract 109,
Monterey, CA (2002).
. Koyama, I. Tanaka, H. Adachi, Y. Makimura, N. Yabuuchi, T. Ohzuku, ll 1h/MLB,
bstract 121, Monterey, CA (2002).
. S. Johnson, J.-S. Kim, A. J. Croft, A. J. Kahaian, J. T. Vaughey, M. M. Thackeray,
lectrochem. Comm. 4 (2002) 492.
hapter 14
CATHODES BASED ON LiCo02 AND LiNi02
1 2
Byungwoo Park\ Yang Jeong Kim , and Jaephil Cho
1
School of Materials Science and Engineering, and Research Center for Energy
Conversion and Storage, Seoul National University, Seoul, Korea
epartment of Applied Chemistry, Kumoh National Institute of Technology, Gumi, Korea
.1. LiCo02 CATHODE MATERIALS
.1.1. Introduction
Li-ion cells consisting of LiMe02 (Me: a 3d-transition metal

ment) and carbon materials have been of interest because of their
ability to be safely operated for thousands of cycles whilst retaining a
h energy density. Materials with the above formula more extensively
mined for positive electrodes include LiCo02, LiNi02, LiCo 1_xNix0 2,
LiMn02• Among these, more research has been done on LiCo02
ause of its high energy density and good cycling performance. The
ctrochemical properties of cathodes based on LiCo02 are shown in
ble 14.1. 1-5 LiCo02 has served as an archetypal cathode material for
ondary Li batteries ever since the discovery by Mizushima et al.6 that
can be reversibly removed (deintercalated) from and reinserted
ercalated) into LixCo02• The layered form of LiCo02, which has a
mbohedral symmetry belonging to the space group RJm, is ideally
ted to accommodate large changes of the Li content, x. This crystal
cture consists of close-packed oxygen layers stacked in an ABCABC
uence with Co and Li ions residing in octahedral sites in alternating
ers between the oxygen planes.7 Figure 14.1 illustrates the crystal
cture of LiCo02 • As the Li concentration is changed in LixCo02,
ancies are either created or filled within the Li planes.
hodes Based on LiCo02 and LiNi02 411
le 14.1. Cycling stability of cathodes based on LiCo02 •
Cathode Material
LiCo02
LiAlo.25Coo.1s02
LiMgo.osCoo.9s02
LixCo02 exhibits many of the essential properties required for a

able cathode material. The voltage of LixCo02 is sufficiently high to
rantee a high-energy density, although it is not too high to cause
ctrolyte decomposition. Furthermore, LixCo02 undergoes small
ctural changes as Li is deintercalated to a concentration of
roximately 0.5. In addition to its favorable electrochemical properties,
Co02 also exhibits phase transformations that are typical of many
rcalation compounds.
Figure 14.2 shows the evolution of the lattice constants a and cas a
28
ction of x in LixCo02, and the global phase diagram for LixCo02. In
X-ray diffraction indicates a sequence of three distinct phase
sitions as x is varied from 1 to 0.4. Two of the transitions are situated
htly above and below x=0.5, and are caused by an order-disorder
89
sition of the lithium ions. • Reimers and Dahn,8 for example,
erved that Li ions are ordered in rows at x=0.5. They carefully
asured the stability region of this ordered phase and determined its
er-disorder transition temperature to be approximately 60°C (Figure
2). The order-disorder (monoclinic-hexagonal phase) transition has
n studied as a function of temperature, allowing the determination of
order-disorder diagram. The other phase transition is shown to be first
er (hexagonal I + II) involving a significant expansion of the lattice
89
stant c of the hexagonal unit cell. •
2 Chapter 14
e Li+
c Co +
3
0 oz-
B
A
c
--A
ure 14.1. Schematic diagram of the layered LiCo0 2 structure.
Another transformation of an electronic nature occurs at a high Li

10
centration: LiCo02 is a semiconductor while LixCo02 at
centrations below x=0.75 is metallic, u suggesting that a metal
ulator transition occurs at the intermediate Li concentration.
iguingly, superimposed on this metal-insulator transition is the two se
coexistence region between two crystallographically identical host
agonal structures. At room temperature, this two-phase coexistence
ion is found to exist between the Li concentrations of x=0.75 and
89 12
3, ' as shown in Figure 14.2. Amatucci et al. reported that LixCo02
be completely and reversibly deintercalated to form Co02• They found

t in the Co02 phase the different 0-Co-0 slabs, held together by van
Waals forces, slide with respect to each other, so that the close-packed
gen layers have an ABAB stacking sequence, as opposed to the
CABC sequence of LiCo0 2•
(a)
2.82 I
-$ 2.81
1::1 II
2.80 (b) Mono.

14.4
14.3
-<
';:;' 14.2
14.1
I
14.0
(c)
60
040 ci
II I+II I
II s:::
izo
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

xinLi.Co01
re 14.2. Lattice constants a (a) and c (b) as a function of the lithium concentration, x,
,.Co02• (c) Phase diagram for Li..Co02 (Ref. 8).
4 Chapter 14
.1.2. Capacity Fading Mechanisms in LiCo02 Cathode

Materials
The theoretical capacity of a LiCo02 cathode is 274 mAh/g.

wever, in practical cells, the capacity is limited to about 140 mAh/g.
is value corresponds to the range 0.5<x<l in LixCo02, with the cutoff
tage of approximately 4.2 V (with respect to aLi metal). To achieve
re than 50% of its theoretical capacity, one must charge the LixCo02 to
ve 4.2 V. However, this generally causes a rapid loss of capacity with
reasing cycle number. Cycling above 4.2 V leads to a dramatic
rease in the capacity fade, which has been closely related to structural
nges involving a decrease in the unit-cell volume,1• 13·19 and an increase
20
cobalt dissolution into the electrolyte.
As shown in Figure 14.2 [(a) and (b)], the lattice constant a shows
y minor changes during Li deintercalation. This is quite the opposite to
lattice constant c, which exhibits an expansion of up to 2-3% at x =
, followed by a decrease for larger x.9' 12 Therefore, the lattice strain is
h anisotropic and directly correlated with the lithium concentration. A
Co02 cathode, like most oxides, can tolerate some elastic strain prior to
cture.1 This property has been mainly attributed to the mechanical
ure associated with the large lattice expansion along the c axis
ection. Figure 14.3 shows individual highly-strained particles where
ocracks are clearly visible (indicated by arrows), after 50 cycles
1
ween 2.5 and 4.35 V. It is also observed that the fractured particles
metimes contain an unstrained region, apparently relieved from the
ss by the fracture event (Figure 14.3(b)).
LiCo02/liquid electrolyte/Li cells were cycled to upper voltages in
range 4.1 - 4.5 V. The cells were then floated at each voltage for two
eks to allow the system to reach equilibrium. The cells were
assembled and quantitatively analyzed for the presence of cobalt on the
de through the use of atomic absorption spectroscopy. Figure 14.4
ws that for voltages of 4.1 and 4.2 V versus Li metal, cobalt
solution in the liquid electrolyte (1 M LiPF6 in ethylene carbonate
): dimethyl carbonate (DMC) (67:33 vol.%)) is very low, as is the
acity fade.20 However, a marked increase in cobalt loss is found when
cells are cycled to 4.3 V. Cycling to 4.4 V further increases the cobalt
s, and 4.5 V shows another sharp increase in the amount of cobalt
solution. Cobalt dissolution is indeed related to the capacity loss with
ling above the traditional reversible voltage limit of 4.2 V.
thodes Based on LiCo02 and LiNi02 415
(a)
ure 14.3. Severely damaged LiCo02 particles from a cycled cathode showing
rofractures (arrows) and a strong diffraction contrast from the internal strain and
ended defects (Ref. 1).
0.5 •
0.4
. "s'
- ";
,Q
u
Q
0.3
0.2 • •-----· -----·

4.1 4.2 4.3 4.4 4.5
Voltage (V)
ure 14.4. Cobalt loss (percentage) as a function of cycling voltage (Ref. 20).
6 Chapter 14
.1.3. Electrochemical Properties of LiCo02 Cathode
LiCo02 powders are typically prepared by reacting a mixture of

OH·H20 and Co304 in the mole ratio of 1.05:1 at 900°C for 24 h. An
ess amount of LiOH · H20 is used to compensate for the loss of Li
ing firing. Figure 14.5 shows the XRD patterns of LiCo02 powders
ore cycling. The starting materials have a well-ordered a-NaFe02
ucture.
Figure 14.6 shows the first charge and discharge curves for LiCo02,
ich indicate an initial capacity of 174 mAh/g between 4.4 and 2.75 Vat
ate of 0.1 C (14 rnA/g) in the half cell Li/LiCo02• The flat potential
teau at -3.93 V is characteristic of the two-hexagonal-phase region.8 9' o
additional plateaus at -4.1 and -4.2 V are attributed to the order
order transition (changes between monoclinic and hexagonal phases)
urring at x=:0.5.8 9• Figure 14.7 shows the cycle-life performance of a
Co02 cathode with charge cutoff voltages of 4.4 V, 4.2 V, and 4.1 V at a
e of 0.5 C. The capacity losses at the charge cutoff voltage of 4.4 V, 4.2
and 4.1 V are -52%, -20%, and -6%, respectively, after 50 cycles.
20 30 40 50 60 70
Scattering Angle 29 (degree)
ure 14.5. XRD patterns of the LiCo0 2 powders before cycling.

4.0
c
,-...
=n
s0 3.5 LiCoo2
> Crate= 0.1 C (= 14 rnA/g)
u
3.0
2.5 L-,J...._J.._..L.,_..J...._...J........I........L......J........L.---L.---L.---L---L----I.......JL........L-..L.-.L..-.J
o w rn
Capacity (mAh/g)
ure 14.6. Voltage profiles of the LiCo0 2 cathode at a rate of 0.1 C (4.4- 2.75 V).
180
r
170 Charge cut-off voltage
160
+ 4.4V
! A 4.2V
e 4.1V
150
140
-
=ii' 130
LiCo02
120
u
cu 110
Oil
-"='
100
s"' 90
80
Crate= 0.5 C
70
0 10 20 30 40 50
Number of Cycle
ure 14.7. Cycle-life performance of the LiCo02 cathode with a charge cutoff voltage
4.4 V, 4.2 V, and 4.1 Vat the rate of 0.5 C. The discharge cutoff voltage is 2.75 V.
An evaluation of the cycle-life performance of LiCo02 was also

ried out between 4.4 and 2.75 V in a coin-type Li-ion cell (C/LiCo02)
gure 14.8). The lower initial capacities of the C/LiCo02 cells (154
h/g) compared to the half-cells are due to the irreversible capacity loss
he mesophase pitch-based carbon fiber (MCF) anode.
8 Chapter 14
For measuring the rate capability of a LiCo02 cathode, the charge

e is fixed at 0.5 C, and the discharge rates are increased from 0.2 to 0.5,
2, and 3 C (Figure 14.9). At higher C rates, the discharge voltage
reases more appreciably than the initial capacity.
Figure 14.10 shows the cyclic voltammograms (CVs) of LiCoOz
ducted up to the second cycle. In this figure, H and M represent the
agonal and monoclinic phases, respectively. Li diffusion in the oxide
key factor that determines the rate at which a battery can be charged
discharged. With increasing interest in higher power density batteries,
derstanding the mechanisms of diffusion in insertion electrodes is
portant. The dependence of the Lt diffusion coefficient in the LiCo02
ticles on the electrode potential is shown in Figure 14.11- note the
iation from -3x10·10 to -7x10·8 cm%ec.Z 1 The Li diffusivities in the
dic (Li deintercalation) and cathodic (Li intercalation) scans are quite
ilar. Three minimum values corresponding to the CV peaks in Figure
10 can be observed. It should be noted that the Li diffusivities exhibit
nimum values at these peak potentials.
C/LiCo02
LiCo02
Charge Rate: 0.5 C
Discharge Rate: 1 C
Cycle Number : 200 150 80 40 20 1

2.5 -L-L --L- L-L- -L
0 20 40 60 80 100 120 140 160
Capacity (mAh/g)
ure 14.8. Plots of the capacity retention for C/LiCo02 cells as a function of cycle
ber. The charge and discharge rates are 0.5 and I C, respectively.
thodes Based on LiCo02 and LiNi02
419
C/LiCo02
0.2C
- -3 : : : :: - - - _
_- - -- - _- _- : :
0.5C
'
..
IC ' ''
... ' \
.... \ 1\
... ... 2C ', \ \I
... \
1
,u
II
JC ', 1\1
,u
'
'
2.5 J--L J--L J--L
0 20 40 60 80 100 120 140 160
Capacity (mAh/g)
ure 14.9. Rate capabilities of LiCo02 at various discharge rates of 0.2, 0.5, 1, 2, and 3
The charge rate was fixed at 0.5 C.
1500
1000
_, 500
d
0
="' ""'-500
u
1-1000
3.2 3.4 3.6 3.8 4.0 4.2 4.4
Voltage (V)
ure 14.10. Cyclic voltammograms of a LiCo02 cathode. Scan rate: 0.02 mY/sec. The
bols Hand M denote the hexagonal and monoclinic phase, respectively.
0 Chapter 14
ure 14.11. Dependence of the apparent diffusion coefficient of Li in the LiCo0 2

icles on the electrode potential (with aLi anode) (Ref. 21).
.1.4. Doping the LiCo02 Cathode
Although LiCo02 has excellent electrochemical properties, other

terials such as LiNi02 and LiMn204 are being extensively studied as
er-cost and more environmentally-friendly substitutes. Efforts have
o been made to synthesize solid solutions of LiMyCo1.y02 (M = AI, Mg,
2 5 13 22 23 22
.) isostructural with LiCo02. - ' ' ' Ceder et a/. reported that first
nciples calculations could orient the search for possible cathode
terials. Through such calculations, they have identified a large class of
candidate materials in which non-transition metals are substituted for
alt. For one such material, LiAlyCo 1.y02, they have predicted and
erimentally verified that AI substitution raises the cell voltage while
reasing both the density of the material and its cost.
Figure 14.12 shows the open-circuit voltage (equilibrium potential)
a function of the Li content in LiCo02, LiA10.25Co0.7502, and
3
l0.5Co0.502. The open-circuit voltage of the cells increases
tematically with increasing AI content in the oxide. As shown in
ure 14.13, during the first charge of LiAlo.25Coo.7502 between 2.0 and
hodes Based on LiCo02 and LiNiOz 421
2
V at 0.4 mA/cm , 182 mAh/g is extracted at room temperature, and
9 mAh/g at 55°C, which corresponds to 61 and 73% of the theoretical,
pectively. The first discharge capacities are 127 mAh/g and 144 mAh/g
room temperature and 55°C, respectively. Rapid capacity fade is
erved upon cycling at room temperature. After nine cycles, the
charge capacity decreases to 51% of the initial value. The Al-doped
Co02 cathode shows both a reduced initial capacity relative to LiCo02,
a deteriorated capacity retention during cycling.
4.4 &.............. A
-·········· y=O.S
o----Q o---o- ·-.......A..... ,.......
E =--o. ··A···AA
4.0 y = o.ff·-oo-o.--o-o. <e
A
y=O •0
3.6
u
! 3.2
0.4 0.6 0.8 1.0
ure 14.12. Open-circuit voltage (OCV) as a function of the Li content in pure LiCo0 2
in Al-doped materials (Ref. 3).
--Charge at 55°C
200 -o- Discharge at 55°C
.......... Charge at RT
-o-Discharge at RT
Cycle Number
ure 14.13. Specific capacity vs. cycle number for LiA10.25Co0.7502, tested against aLi
2
al anode at 0.4 mA/cm between 2.0 and 4.5 V at room temperature and 55°C (Ref. 3).
2 Chapter 14
.2. LiNi02 CATHODE MATERIALS
.2.1. Introduction
Compared to LiCoOz, LiNi02 is cheaper and less toxic. It delivers

rger reversible capacity with a comparable cycle life. It is considered
be one of the best cathode materials for Li-ion cells, and is used in
phite-based cells. The electrochemical properties of cathodes based on
Ni02 are shown in Table 14.2.2427 - However, its synthesis requires
cial care to obtain good and reproducible performances.
• Its structural
24 25 2830
electrochemical behavior has been reviewed. • -
le 14.2. Cycling stability of cathodes based on LiNi02•
Cathode Material
LiNi02
LiNi 0,,0 Coo.3o02
iNio.zsCoo.soMno.2s02
.2.2. Structure of LiNi02
The ideal structure of LiNi02 is layered, isostructural with LiCo02.

e XRD patternsf LiNi0 2 are indexed to a hexagonal symmetry with
space group R3m, as shown in Figure 14.14. However, the Li-Ni-0
tem is characterized by the existence of a Li 1_,Ni 1+,02 (0<z<1) solid
ution. At low z values (0<z<0.2), the structure can still be described as
2
yered structure with the presence of Ni + cations within the Li layers.
e amount, z, of interlayer Ni + cations depends on the
2
hodes Based on LiCo02 and LiNi02

423
"' Z'
'§ LiNi02
.cl
1;
'-'
B
j ;:::;
=0
15 20 25 30 35 40 45 50 55 60 65 70
Scattering Angle 28 (degree)
ure 14.14. XRD patterns of the LiNi0 2 powders, before cycling.
thesis conditions, and is lowered when using oxidizing atmospheres

3
the synthesis because of the stabilization of Ni +. The extra interlayer
+ cations are strongly detrimental to the electrochemical performance,
hey induce a lower capacity and larger polarization (voltage difference
1 35
ween the charge and discharge curves)? -
.2.3. Electrochemical Properties and Capacity Fading

Mechanisms of LiNi02
As shown in Figure 14.15, LiNi02 exhibits quite a large irreversible

acity loss during electrochemical cycling, much larger than that of
Co02• Most of the irreversibility is observed at the end of the first
rge, and a model based on the formation of inactive domains was
36 33
posed. Another model, '37 based on a local collapse of the interlayer
3
ce due to the oxidation of interlayer Ne+ to smaller size Ni + cations
i2• = 0.68 A; rNi'• = 0.56 A), was proposed. Such a local contraction
uld impede Li reintercalation in the six close-neighboring sites of
3
rlayer Ni + cations, due to the larger size of the Li+ ion ( rLi. = 0.74
which would lead to a fast increase of the irreversible capacity with
reasing quantities of initial interlayer cations.
4 Chapter 14
4.5
4.0
"'
i >
3.5
=
u"'
3.0 Crate= 0.1 C (= 18 rnA/g)
2.5 0 50 100 150 200 250

Capacity (mAh/g)
200
Io:l
150
100
o:l
u
r"';
-5 50
Crate= 0.2 C (= 36 rnA/g)
i5"'
0
0 10 20 30 40 50 60 70
Cycle Number
ure 14.15. (a) Initial voltage profiles of LiNi0 2 between 4.3 and 2.75 V at a rate of 0.1
8 rnA/g), and (b) cycle life ofLiNi02 at a rate of 0.2 C (36 rnA/g).
Several reversible structural transitions occur upon Li

ntercalation,31' 3438
' which are different from those occurring in LiJ_
o0 2 (Figure 14.16). CVs up to 4.4 V show H1 H M H H2 H
H3 sformations and high initial charge and discharge
capacities. wever, a large capacity fade on long-term cycling was also
observed gure 14.15(b)). This behavior can be attributed to the
migration of Ni ons from the layers to the interlayer space,
12 39
thereby altering the sequent • electrochemical behavior. In
addition, since metal oxides h a low ductility undergo mechanically-
induced microcracks resulting m lattice strains, the reversibility
of Li intercalation sharply
eriorates. Microcrack formation in each LiNi02 particle is induced by

sotropic lattice changes along the a and c axis in the initial hexagonal
8 40
se (Figure 14.17)? • A similar behavior is observed in LiCo02.
2000
1000
=' .
' - 0
.
8 -1000
-2000
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4
Voltage (V)
ure 14.16. Cyclic voltammograms ofLiNi0 2• The scan rate was 0.02 mY/sec.
14.50
0
"' 14.45
j
:r:l 14.40
u
..
"'
: 0
14.35
.'l.U.. 14.30
14.20
2.90
2.89
2.88
0
2.87
<:!
=s0 2.86
" '
u=
..
"'
: 0
2.85
2.84
.'.lU..
2.83
2.82
M M M U U
X in Lil·XNi02
ure 14.17. Evolution of the lattice constants a and c as a function of x in Lil-xNi0 2•

6 Chapter 14
.3. METAL-OXIDE COATING ON LiCo0 2

POWDERS
.3.1. Introduction
The electrochemical cycling of LiCo02 at a high cutoff voltage

ults in a significant deterioration of the stability of the cathode. As
mmented above, this capacity loss is related to non-uniform structural
20
nges 1• 13- 19 and an increase in cobalt dissolution into the electrolyte.
improve the electrochemical performance of LiCo02 above 4.2 V, an
ovative approach was adopted. This involved coating the particles
h some metal oxides to avoid the unwanted surface reactions and
tect the bulk.5 •16-19.41-47 This method changes the surface properties of
cathode materials. Aurbach et al.48 reported that the electrochemical
avior of LixMOy (M = Ni, Mn) cathode materials is strongly dependent
their surface chemistry. Recent reports on surface modifications to
hode materials such LiCo02 and LiNi02 with Alz03, Zr02, MgO, Zn02,
2 03, and Sn02 have shown that surface coating is an effective way to
bilize the structure of the materials and improve their electrochemical
formance.
.3.2. Preparation of Metal-Oxide Coated LiCo02

Powders
A sol-gel coating of various metal oxides having a wide range of

49
cture toughness (Zr02, Alz03, Ti02, B2 03, and Si02) -51 is applied on
LiCo02 particle surfaces. Metal ethylhexanate-diisopropoxide
(OOC8H 15)(0C3H7)z) is dissolved in 2-propanol, followed by
tinuous stirring at RT. The LiCo02 powder (average particle size of
0 flm) is then mixed with the coating solution such that the total
centration of metal in the coating solution corresponds to 5 wt% of the
Co02 powders used. The resulting solution is further aged at 50°C to
w the M-OR groups to react with the LiCo02 surface-OR groups,
ding to strong bonds between the metal-oxide gel and the LiCo02
ticle surface. After drying, each coated LiCo02 powder is fired at
°C for some hours. Field-emission SEM confirms that the metal-oxide
ting is continuous. X-ray diffractiO,!l of the as-prepared samples show
ngle-phase a-NaFe02 structure (R3m), where the cobalt and lithium
s are located at the 3a and 3b octahedral sites, respectively, in a close

ked oxygen array. However, the XRD patterns of the coated samples
not show any evidence of a crystalline metal oxide. Enhanced
ctrochemical results have been observed by a metal-oxide coating from
42
hn's group. However, they have shown that the coating layer contains
ocrystalline Zr02 (15-25 nm) on the surface of the LiCo02 powders
ough their XRD pattern and TEM image. This discrepancy may be due
he difference in the coating precursors. The coating solutions used by
hn's group are a zirconium oxide polymeric precursor or an aqueous
ution of ZrO(N03)z ·xH2 0.
.3.3. Cycle-Life Performance of Metal-Oxide Coated

LiCo02
To test the cycle-life performance of each cathode material, coin e

half cells (Li/LiCo0 2) are cycled at the 0.1 C rate (14 rnNg) for the t
2 cycles between 4.4 and 2.75 V. Cycling is then continued at the C
rate for up to 70 cycles. As shown in Figure 14.18, the initial
charge capacities of the ZrOz- and Al20rcoated samples are similar to
t of the uncoated one, and the cycle-life performance trend is
consistent
h the fracture toughness of the coated metal oxide (Zr02>A}z03
i02>B203>Si02). In addition, the Zr02-coated sample does not show
noticeable capacity fading over 70 cycles. In contrast, bare LiCo02 es
-40% of its original capacity after only 30 cycles.
The lattice constant c as a function of x in the hexagonal Li 1_xCo0 2
m x=O to 0.7 during the first charge is shown in Figure 14.19. Even
ugh the lattice constant a shows a slight variation during charging in
h the bare and coated Li 1_xCo0 2, the lattice-constant c shifts depend on
choice of the metal-oxide coating. The stability of the c-axis
ension correlates well with the fracture toughness of the coated oxides,
occurs in the order of Zr02>A}z03>Ti02>B203>Si02 . This result
firms that suppressing the non-uniform strain during the delithiation
duces a new-class of cathode materials, which are extremely tolerant to
trochemical cycling. Note that the lattice constant c of the ZrOz
ted LiCo02 exhibits a negligible shift in the range O<x<O.7. In
ition, a sol-gel coating of Zr02 can effectively inhibit cobalt
olution from the LiCo02 cathodes. As shown in Figure 14.20, cobalt
solution in ZrOz-coated LiCo0 2 is greatly reduced (-140 ppm),
pared to that of an uncoated LiCo02 (-730 ppm), after a charge
8 Chapter 14
4.S
4.0
E 3.S Crate= 0.1 C (= 14 mA/g)
3.0
u
2.S
2.0
0 so 100 1SO 200 2SO 300 3SO 400
Capacity (mAh/g)
180
!
t>
-
=u - 140
"'., 130
0 11
.. .
120
s"' 110
10 20 30 40 so 60 70
Cycle Number
ure 14.18. (a) Initial charge and discharge curves for various metal-oxide coated
o02, and comparison with uncoated LiCo02 (first cycle). (b) Cycle-life performances
ZrOz-, Al203-, TiOz-, B 203-, Si02-coated, and uncoated LiCo02. The cells were
ed at a rate of 0.1 C for the first two cycles, followed by a 0.5 Crate between 4.4 and
Vat 21oc.
.8 V and one-week storage at 25°C. Amatucci et al. reported that an

rease in cobalt dissolution above 4.4 V is related to the lattice changes
12
uced by removal of the binding lithium. Figure 14.21 depicts how
losing the LiCo02 particles in a fracture-toughened metal-oxide
ting layer can effectively suppress the non-uniform lattice-constant
nges during Li deintercalation. It should be noted that, in thin film
o02 (both bare and metal-oxide coated), the lattice constants as a
ction of the cell voltage from OCV to 4.4 V show negligible c axis
hodes Based on LiCoOz and LiNiOz 429
ansion (details in Section 14.4.2.5). Dahn's group suggests that a Zr02

ting on the powder has no effect on LixCo02 expansion although the
Orcoated LiCo02 powders exhibit an improved cycling behavior to 4.5
2
This disagreement is probably due to the difference in the coating
ution. Dahn's group assumes that a Zr02 coating may inhibit side
ctions involving oxygen loss from the LixCo0 2 to the electrolyte and
ce improve the cycling stability.
, -..
o<
,="'
'-'
Q
14.25
u
Col
14.20
';I
14.15
....:l
14.10
14.05
14.00
2.818
2.816
, -..
o<
,= <:::1
' '
... 2.814
2.812
Q
u
';I
C!l
...
Col
2.810
....:l
2.808
2.806 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

x in Li1.xcoo2
ure 14.19. Lattice constants a and c in Zr0 2-, Al203-, TiOz-, B203-, SiOz-coated, and
LiCo02 as a function of x in Li1_xCo02 during the first charge. The precursor is
al ethylhexanate-diisopropoxide (M(OOC8H 15)(0C3H7)z) dissolved in 2-propanol.
Chapter 14
800 ------ ---- ------ ---- --

• Bare
After one week at 25°C
600
.
400 _.....-+ Alp3

...........
.
200 .••.••••.•.••...•..••.••.••••.••-<] Zr0
2
....................--
OL- L----- -------L------ ------L---

4.4 4.6 4.8
Voltage (V)
igure 14.20. The amount of cobalt dissolution in the electrolyte (I M LiPF6-EC/DMC),
om various metal-oxide coated and uncoated LiCo02, after an initial charge and being
mmersed for one week at 25°C.
ZrOrcoated LiCo02
u
0%
deintercalation
hexaQ9(1al
UCo02
igure 14.21. Schematic illustration of the suppression of lattice expansion from

exagonal to monoclinic phase by a nanoscale metal-oxide coating.
.4. THIN-FILM LiCo02 CATHODES
.4.1. Introduction
Thin film microbatteries52-54 have potential applications in

croelectronics. In particular, the fabrication of lithiated intercalation
des in a thin-film form is of great interest because of their possible use
a positive electrode in all-solid-state lithium rechargeable
crobatteries. This is due to the need to miniaturize various electronic
ices, such as monolithic hybridization with CMOS-RAM, back-up
wer systems for computer chips, small sensors, and hazard cards. 55- 58
n-film cathodes have also received a great deal of attention as they
w to study the intrinsic electrochemical properties of lithiated
sition-metal oxides. Indeed, composite powders with polymer binders
carbon blacks may not represent the characteristic electrochemical
perties of these oxides. 57- 67
.4.2. Electrochemical Properties of Alz03-Coated

LiCo02 Thin Films
4.2.1. Cycle-Life Performance of Alz03-Coated LiCo02 Thin Films
To test the cycle-life performance of bare and 30 nm-thick AhOr

ted LiCo02 thin films on a Pt/Si substrate, half cells are cycled in the
2
tage range 2.75 - 4.4 Vat 0.2 mNcm , as shown in Figure 14.22. A
30
-thick Ah03 coating significantly improves the electrochemical
bility of the LiCo02 thin-film cathodes. The charge and discharge
acities of bare LiCo02 get deteriorated to only -3% and -25% of the
inal capacity after 100 cycles (Figure 14.22(a)), while Ah03-coated
o02 retains -64% and -75% of the initial capacity at a current rate of
mNcm2 (Figure 14.22(b)). The charge capacities of the bare LiCo02
s (Li deintercalation) exhibit faster deterioration than the discharge
acities (Li intercalation). The effect of an Al203 coating on the cycle
performance is more pronounced at a higher-current rate (Figure
23). The insets in Figure 14.22 show the same cycle-life performance,
uding the extra capacity from the constant voltage mode (at 4.4 V and
5 V).
In the bare samples, most of the charge capacity after -40 cycles
es from the constant voltage charge of 4.4 V, indicating that the Li
2 Chapter 14
60r- ---.-- --.--- . o

(a) Bare LiCo02 50
50
2
0.2mA/cm
10
40 0 -- --
0 20 40 60 80 100
30
c ---o--- Discharge Capacity
a
·20 -o- Charge Capacity
10
60
(b) 30 nm Coating 50
50 40
30
20
to
0
0 20 40 60 80 100
Cycle Number
ure 14.22. Cycle-life performance (charge and discharge capacities vs. cycle number)
(a) bare, and (b) 30 nm-thick Al203 -coated LiCo02 thin films between 4.4 and 2.75 V
2
0.2 mA/cm (= 6 C). The inset shows the same cycle-life performance, including the
a capacity from the constant-voltage mode.
ntercalation kinetics are slower than the Li intercalation kinetics. On

other hand, the Ah03-coated samples exhibit similar charge and
charge capacities, and the curve in the inset of Figure 22(b) has an
ost constant slope. This means that Li intercalation and
ntercalation in the coated films become more symmetrical. The Ah03
ting on the surface modifies the properties of the cathode surface
osed to the electrolyte and changes the Li-diffusion kinetics. The
tage profiles of the bare and 30 nm-thick Ah0 3-coated LiCo02 thin
s are shown in Figure 14.24. The profiles of the bare LiCo0 2 thin
s become steeper with increasing the cycling number up to 80 cycles
thodes Based on LiCo02 and LiNi02
433
100 (a)
o Alp -coated LiCo0
0... 3 2
80
,-.._ ··o···················O····················o···················O
=
'-'
0
;I 60
.=
fj o Bare LiCo02
0 -D
40
_e. 0 0
·c'II
§' 20
u
0
50
(b)
-< 40 0-=::-:B......... o Alp -coated LiCo0

3 2
_e. Q-···················9··········· 0
·c
a
30 o Bare LiCo0
'II 2
1,.:.1.1 20
' II
.. c :
(,J
Q"' 10
:...!.
·..a.. 0
200 400 600
2
800
Charge/Discharge Rate (jLA/cm )
ure 14.23. (a) Capacity retention vs. current rate after 100 cycles, and (b) initial
charge capacity vs. current rate of uncoated and Ah03-coated LiCo02 thin films.
2
0.2 mA/cm • In contrast, the change in the voltage profiles of the
zOrcoated LiCo02 thin films during 80 cycles is more gradual.
wever, compared with the cycling profiles of bare LiCo02 thin films, a
ht increase in cell potential is observed in the initial cycle of A}z03-
ted cathodes. This is attributed to the insulating effect of the
minum-oxide layer. From the 2nd cycle on, the polarization in the
rge and discharge processes of coated sample is not significant, as the
Al-0 coating layer on LiCo0 2 acts as a solid electrolyte (details in
tion 14.4.2.2).
4 Chapter 14
(a) Bare LiCo0 2

4.4 .·-··- :-..:.·------ -----·-·-::-:';"'" ........ ·.: ,_
E 4.0
-a;c:J '.
=
3.6 \
'·
I '.
U I '
3.2 \ ·,
·,·,
' ·,
•. ·,, •• I
2.8 '··-··--- ·-·-·-·-·-· ....... 0 '
4.4
E 4.0
-a;c:J
= 2
£ 3.6 0.2mA/an
'
=
u --1st Cycle
\
\
- - - 2nd Cycle
3.2 ·····20th Cycle
I
i
-·-··40th Cycle
I
-··-··80th Cycle
2.8
0 10 20 30 40
Capacity (jLAh/cm1
ure 14.24. Voltage profiles of (a) uncoated, and (b) 30 run-thick Al 20rcoated LiCo02
2
s between 4.4 and 2.75 V at a rate of 0.2 mA/cm • At each charge/discharge cutoff
, the cell voltage was potentiostated until the current decreased to 0.02 mA/cm2 .
4.2.2.Alz03 Coating Layer Acting as a Solid Electrolyte
The AhOrcoating thicknesses were varied from 10 nm to 300 nm

xamine the possible questions of electronic insulation and Li migration
ough the A}z03 layer. Figure 14.25(a) shows that the initial discharge
acities are almost independent of the coating thickness (0-300 nm
ge). In addition, the variation in the A}z03 thickness ranging from
nm to 300 nm does not affect the capacity retention: -70% and -60%
2
ntion, respectively, at 0.2 and 0.4 mNcm , while the capacity retention
the uncoated LiCo02 thin film is only -25% after 100 cycles, as shown
Figure 14.25(b). This is possibly because the A}z03 coating layer acts
a solid electrolyte with a low electronic conductivity and a reasonably
h Li-ion conductivity. For example, the Li-ion conductivity of a
7 68
Liz0-0.3A}z03 solid electrolyte is -10- S/cm at room temperature.
50
"s: (a) Initial Discharge Capacity
l
a
· o
40
45
< II
c . 35
till
30
- .=...
s 25
1 ...... 20
-o---
100
(b) Capacity Retention
...,.........o-0
---o---!:c-
-6i
---a-- --------Q
-a- 0.2 mA/cm2

- -o- 0.4 mA/cm2
0 40 80 120 300
Alp3 Thickness (nm)
re 14.25. The effect of the Al20rcoating thickness on (a) initial discharge capacity
(b), capacity retention after 100 cycles, all excluding the extra capacity from the
tant-voltage mode. The cells were cycled between 4.4 and 2.75 V at 0.2 and 0.4
2
cm •
6 Chapter 14
4.2.3.Suppression of Cobalt Dissolution by Ah03 Coating
To clarify the reason why the Alz03-coated LiCo02 thin films

wed better electrochemical performance than the bare ones,
uctively coupled plasma-mass spectroscopy (ICP-MS) analysis was
ried out after floating at 4.2 V, 4.4 V, 4.5 V, 4.6 V, and 4.7 V,
pectively, for 12 days. As shown in Figure 14.26, the amount of cobalt
solution from the uncoated LiCo02 films significantly increases with
reasing the charge-cutoff voltages, while an increase in cobalt
solution from the AlzOrcoated ones is not significant. An increase in
4
fraction of Co + in Li 1_xCo02 augments the reactivity with the
ctrolytes and HF, if present. An Alz03 coating can effectively reduce
alt dissolution, even at 4.7 V, to a much smaller level (-100 ppm) than
t of the uncoated LiCo02 films ( -1000 ppm).
1200
After floating for 12 days
D
I
1000
_ I
I
. I
Bare LiCo02 ,'
-
800 I
I
I
I
. 600 I
I
I
I
I
I
,0
,,
l
400 , ,
. , ' o·
,
200 Al 0 -coated LiCo0
-----o·-
--- ---
0
o--o
o-- ---
0
0 2 3 2
4.2 4.3 4.4 4.5 4.6 4.7

Voltage (V)
gure 14.26. The amount of cobalt dissolution in the electrolyte, from bare and Ah03-
ated LiCo02, after an initial charge and being immersed for 12 days at 25°C.
4.2.4. CV Measurement of Ah03-Coated LiCo02 Thin Films
Figure 14.27 shows the cyclic voltammograms of the bare and 30

-thick AhOrcoated LiCo02 thin films at a scan rate of 0.1 mV/sec,
ried out during the 1st cycle, and after the 40th and 80th
rging/discharging cycles. The CV s of the Ah03-coated LiCo02 thin
ms during the 1st cycle clearly show three sets of well-defined current
ks, i.e. -3.96 and -3.87 V (a, a'), -4.07 and -4.05 V (b, b'), -4.19 and
16 V (c, c'). The first set is due to the first-order phase transition
ween the two hexagonal phases, and the second and third sets are
sed by the phase transition between the hexagonal and monoclinic
ses.8' 9 These values are in good agreement with those obtained from
uncoated LiCo02 films and powders. However, these thin-film results
16 18
different from the CVs of AhOrcoated LiCo02 powders, ' and the
sons for this disparity need to be identified. The cathodic peak (-3.87
t a') of the Ah03-coated film during the 1st cycle is narrower than that
the bare one, although the anodic peaks (-3.96 Vat a) are similar. In
ition, in the bare LiCo02 thin films, the widths of both the cathodic
anodic peaks are remarkably broadened, and the peak positions
ome more separated as the cycle number increases to 80, while the
ths of the cathodic and anodic peaks of the Al 20rcoated samples are
tively sharp. This indicates that the Ah03-coated LiCoOz thin films
more reversible than the bare ones, which is probably related to the
pression of cobalt dissolution from the cathode (Figure 14.26).
4.2.5. Changes in the Lattice Constants cas a Function of the Cell

Potential
Figure 14.28 shows the changes in the lattice constants c as a

ction of the cell potential during the first charge in the Li 1_xCo0 2 thin
18
s and powders.16- Both the bare and Ah0 3 -coated LiCo02 thin films
ibit negligible c axis expansion even with a strong (003) texture.67
s is in contrast to the LiCo02 powders, which exhibit -2.6% c axis
ansion at -4.17 V. Further studies are necessary to identify the
chanisms allowing LiCo02 thin films to have a limited c axis
ansion.
8 Chapter 14
150 (a) Bare LiCo0 2 a
100 0.1 mV/sec
50
=t: 0 ··-··-··-..........
u=
-50
--During 1st Cycle
······After 40 Cycles
-100 a' -··-·· After 80 Cycles
150 (b) 30 urn-Coated LiCo02 a
100
50
=5I . . .
< 1 , )
0
u
-50
-100 a'
3.4 3.6 3.8 4.0 4.2 4.4

Cell Potential (V)
ure 14.27. Cyclic voltammograms of (a) bare, and (b) 30 nm-thick Al203 -coated
o02 thin films, during the 1st cycle, and after the 40th and 80th cycle. The sweep rate
0.1 mV/sec between 3.0 and 4.4 V. The symbols Hand M denote the hexagonal and
oclinic phases, respectively.
4.2.6. Li Diffusivities of Ah03-Coated LiCo02 Thin Films
Figures 14.29 and 14.30 show the apparent Li diffusivities as a

ction of the cell potential of the Li1.xCo02 cathodes during Li
ntercalation (charging) and intercalation (discharging), respectively.
Li diffusivities of AlzOrcoated LiCo02 are slightly lower than those
bare one during the 1st Li deintercalation, as shown in Figure 14.29.
14.5
14.4
,-...
·s - -········ ·········
.... 14.3
B ··•·· LiCo02 Powder
-o- Bare LiCo02 Film
"
u='
14.2 t' - -o-· Al 0 -coated LiCo0 Film
0
2 3 2
c.J
E 14.1
C'iS !
= --
14.0
atOCV
4.1 4.2 4.3 4.4
2.7 3.9 4.0
Cell Potential (V)
ure 14.28. Changes in the lattice constants c as a function of the cell potential for bare
Al203-coated LiCo02 thin films during the first charge, with a comparison with
der samples. For thin films, each cell was charged at a 0.1 mA/cm 2 rate to the
determined voltages, then potentiostated until the current density decreased to
2
tA/cm •
The Li diffusivities in the bare LiCo02 thin films during Li

ntercalation (Figure 14.29) are in good agreement with the previous
57
ults obtained using either thin-film Li 1.xCo02 cathodes •58 •62 •69 or
70 72
wder electrodes. - Note that the plots of Li diffusivity vs. cell
ential show maxima at -4.13 V, corresponding to the monoclinic phase.
addition, two minima are observed at the cell potential corresponding to
phase transition between the hexagonal and monoclinic phase, in
60
eement with previous reports by Jang et al..
As the cycle number increases, the Li diffusivity decreases owing to
degradation of the thin-film LiCo0 2 cathodes. As shown in Figure
29, during Li deintercalation the Li diffusivities of the bare LiCo02
n films (in the range of approximately 4.15 and 4.3 V) decrease to
12 2 11 2
10- cm /sec after 80 cycles, compared to -2x10· cm /sec in the
ted films. During Li intercalation, the Li diffusivities in bare samples
0 Chapter 14
(a) Bare LiCo02
(b) 30 run-Coated LiCo0 2
-o- Before Cycling

-• - After 20
Cycles
··6·· After 40 Cycles
-·.,-· After 60 Cycles
-·0.- After 80 Cycles
3.9 4.0 4.1 4.2 4.3

Cell Potential (V)
ure 14.29. Lithium diffusivities as a function of cell potential for (a) bare, and (b) 30
thick AI20rcoated LiCo02 thin films, determined by GITT during Li deintercalation
rging).
11 2
-lxl0- cm /sec after 80 cycles, while the deterioration of the Li
usivities in the coated ones remains at -3xl0-11 cm2/sec (Figure 14.30).
se results suggest that an A}z03 coating on the LiCo02 thin films can
rate the electrochemical cycling with a high cutoff voltage of 4.4 V,
nks to the suppression of cobalt dissolution by A}z03 coating.
During Li Intercalation
-a- Before Cycling

-• - After 20
Cycles
··-b.·· After 40 Cycles
-·'f'-· After 60 Cycles
-·<>·- After 80 Cycles
3.9 4.0 4.1 4.2 4.3

Cell Potential (V)
gure 14.30. Lithium diffusivities as a function of cell potential for (a) bare, and (b) 30
-thick Al20rcoated LiCo02 thin films, as determined by GITT during Li intercalation
scharging).
.4.3. Future Work
Further studies to detail the effect of a metal-oxide coating on the

rface reactions, such as the microstructural changes in Li 1_xCo02, the
mposition profile of the Li-Al-0 coating layer, and the electric-potential
tribution near the interface between the LiCo02 cathode and the Li-Al-
coating layer, are currently underway. Also, the coating effects of
rious oxides need to be identified.
Chapter 14
KNOWLEDGEMENT
We are grateful to Joon-Gon Lee, Hyemin Kim, Tae-Joon Kim, and

oungsoo Kim for their help to improve the clarity of this manuscript.
FERENCES
H. Wang, Y.-1. Jang, B. Huang, D. R. Sadoway, Y.-M. Chiang, J. Electrochern. Soc.

146 (1999) 473.
H. Tukamoto, A. R. West, J. Electrochern. Soc. 144 (1997) 3164.
Y.-1. Jang, B. Huang, H. Wang, D. R. Sadoway, G. Ceder, Y.-M. Chiang, H. Liu, H.
Tamura, J. Electrochern. Soc. 146 (1999) 862.
W.-S. Yoon, K.-K. Lee, K.-B. Kim, J. Electrochern. Soc. 147 (2000) 2023.
M. Mladenov, R. Stoyanova, E. Zhecheva, S. Vassilev, Electrochern. Cornrnun. 3
(2001) 410.
K. Mizushima, P. C. Jones, P. J. Wiseman, J. B. Goodenough, Mater. Res. Bull. 15
1980) 783.
H. J. Orman, P. J. Wiseman, Acta Crystallogr., Sect. C: Cryst. Struc. Cornrnun. 40
(1980) 12.
J. N. Reimers, J. R. Dahn, J. Electrochern. Soc. 139 (1992) 2091.
T. Ohzuku, A. Ueda, J. Electrochern. Soc. 141 (1994) 2972.
J. van Elp, J. L. Wieland, H. Eskes, P. Kuiper, G. A. Sawatzky, F. M. F. de Groot, T. S.
Turner, Phys. Rev. B 44 (1991) 6090.
J. Molenda, A. Stoklosa, T. Bak, Solid State lonics 36 (1989) 53.
G. G. Amatucci, J. M. Tarascon, L. C. Klein, J. Electrochern. Soc. 143 (1996) 1114.
H. Wang, Y.-1. Jang, B. Huang, D. R. Sadoway, Y.-M. Chiang, J. Power Sources 81-
82 (1999) 594.
Y.-K. Sun, J. Power Sources 83 (1999) 223.
Y.-M. Choi, S.-1. Pyun, Solid State lonics 99 (1997) 173.
J. Cho, Y. J. Kim, B. Park, Chern. Mater. 12 (2000) 3788.
J. Cho, Y. J. Kim, T.-J. Kim, B. Park, Angew. Chern. Int. Ed. 40 (2001) 3367.
J. Cho, Y. J. Kim, B. Park, J. Electrochern. Soc. 148 (2001) A1110.
J. Cho, C.-S. Kim, S.-1. Yoo, Electrochern. Solid-State Lett. 3 (2000) 362.
G. G. Amatucci, J. M. Tarascon, L. C. Klein, Solid State Ionics 83 (1996) 167.
K. Dokko, M. Mohamede, Y. Fujita, T. Itoh, M. Nishizawa, M. Umeda, I. Uchida, J.
Electrochern. Soc. 148 (2001) A442.
G. Ceder, Y.-M. Chiang, D. R. Sadoway, M. K. Aydinol, Y.-1. Jang, B. Huang, Nature
392 (1998) 694.
S.-T. Myung, N. Kumagai, S. Komaba, H.-T. Chung, Solid State Ionics 139 (2001) 47.
K. Kubo, M. Fujiwara, S. Yamada, S. Arai, M. Kanda, J. Power Sources 68 (1997)
553.
C.-C. Chang, J. Y. Kim, P. N. Kumta, J. Electrochem. Soc. 149 (2002) A331.
J. Cho, G. Kim, H. S. Lim, J. Electrochem. Soc. 146 (1999) 3571.
Z. Lu, D. D. MacNeil, J. R. Dahn, Electrochem. Solid-State Lett. 4 (2001) A200.
M. M. Thackeray, J. Electrochem. Soc. 142 (1995) 2558.
P. G. Bruce, Chem. Commun., 1817 (1997).
P. Koksbang, J. Barker, H. Shi, M. Y. Sakdi, Solid State lonics 84 (1996) 1.
H. Arai, S. Okada, H. Ohtsuka, M. lchimura, J. Yamaki, Solid State lonics 80 (1995)
261.
T. Ohzuku, A. Ueda, M. Nagayama, J. Electrochem. Soc. 140 (1993) 1862. A.
Rougier, P. Gravereau, C. Delmas, J. Electrochem. Soc. 143 (1996) 1168. J.
Barker, R. Koksbang, M. Y. Saidi, Solid State lonics 89 (1996) 25.
R. V. Moshtev, P. Zlatilova, V. Manev, A. Sato, J. Power Sources 54 (1995) 329.
H. Arai, S. Okada, Y. Sakurai, J.l. Yamaki, Solid State lonics 95 (1997) 275.
J.P. Peres, PhD thesis, Univ. of Bordeaux I, France, (1996).
W. Li, J. N. Reimers, J. R. Dahn, Solid State lonics 67 (1993) 123.
T. Ohzuku, A. Ueda, M. Kouguchi, J. Electrochem. Soc. 142 (1995) 4033.
J. Cho, H. Jung, Y. Park, G. Kim, H. Lim, J. Electrochem. Soc. 147 (2000) 10.
Z. Wang, C. Wu, L. Liu, F. Wu, L. Chen, X. Huang, J. Electrochem. Soc. 149 (2000)
A466.
Z. Chen, J. R. Dahn, Electrochem. Solid-State Lett. 5 (2002) A213.
Z. Wang, L. Liu, L. Chen, X. Huang, Solid State lonics 148 (2002) 335.
J. Cho, T.-J. Kim, Y. J. Kim, B. Park, Electrochem. Solid-State Lett. 4 (2001) A159.
D. Rahner, W. Plieth, M. Kloss, "Modified electrode material and its use," U.S. Patent
No. 6348259, March 30, 1999.
T. Miyasaka, "Non-aqueous lithium ion secondary battery," U.S. Patent No. 6037095,
March 30, 1998.
H.-J. Kweon, G.-B. Kim, D.-G. Park, "Positive active material for rechargeable lithium
battery and method of preparing same," U.S. Patent No. 6183911, October 29, 1999.
D. Aurbach, K. Gamosky, B. Markovsky, G. Salitra, Y. Gofer, U. Heider, R. Oesten,
M. Schmidt, J. Electrochem. Soc. 147 (2000) 1322.
W. D. Jr. Callister, Materials Science and Engineering: An Introduction, John Wiley &
Sons: New York, 4th ed., Appendix C, p. 787 (1997).
M. C. Frede!, A. R. Boccaccini, J. Mater. Sci. 31 (1996) 4375.
J. Sehgal, S. Ito, J. Non-Cryst. Solids 253 (1999) 126.
P. Birke, W. F. Chu, W. Weppner, Solid State lonics 93 (1997) 1.
J. B. Bates, G. R. Gruzalski, N.J. Dudney, C. F. Luck, X. Yu, Solid State lonics 70-71
(1994) 619.
4 Chapter 14
C. Julien, G.-A. Nazri, Solid State Batteries: Materials Design and Optimization,
Kluwer Academic, Boston, 1994.
C. Wolverton, A. Zunger, J. Electrochem. Soc. 145 (1998) 2424.
D. G. Fauteux, A. Massucco, J. Shi, C. Lampe-Onnerud, J. Appl. Electrochem. 27
(1997) 543.
B. Wang, J. B. Bates, F. X. Hart, B. C. Sales, R. A. Zuhr, J. D. Robertson, J.
J. B. Bates, N. J. Dudney, B. J. Neudecker, F. X. Hart, H. P. Jun, S. A. Hackney, J.
M. Antaya, J. R. Dahn, J. S. Preston, E. Rossen, J. N. Reimers, J. Electrochem. Soc.
140 (1993) 575.
Y.-1. Jang, B. J. Neudecker, N. J. Dudney, Electrochem. Solid-State Lett. 4 (2001)
A74.
J.-K. Lee, S.-J. Lee, H.-K. Baik, H.-Y. Lee, S.-W. Jang, S.-M. Lee, Electrochem. Solid
State Lett. 2 (1999) 512.
K. A. Striebel, C. Z. Deng, S. J. Wen, E. J. Cairns, J. Electrochem. Soc. 143 (1996)
1821.
Y. Iriyama, M. Inaba, T. Abe, Z. Ogumi, J. Power Sources 94 (2001) 175.
J. D. Perkins, C. S. Bahn, J. M. McGraw, P. A. Parilla, D. S. Ginley, J. Electrochem.
Soc. 148 (2001) Al302.
P. J. Bouwman, B. A. Boukamp, H. J. M. Bouwmeester, P. H. L. Notten, J.
Electrochem. Soc. 149 (2002) A699.
. D. Perkins, C. S. Bahn, P. A. Parilla, J. M. McGraw, M. L. Fu, M. Duncan, H. Yu, D.
S. Ginley, J. Power Sources 81-82 (1999) 675.
Y. J. Kim, T.-J. Kim, J. W. Shin, B. Park, J. Cho, J. Electrochem. Soc. 149 (2002)
1337.
A.M. Glass, K. Nassau, J. Appl. Phys. 51 (1980) 3756.
J. M. McGraw, C. S. Bahn, P. A. Parilla, J. D. Perkins, D. W. Readey, D. S. Ginley,
Electrochim. Acta 45 (1999) 187.
J.-S. Hong, J. R. Selman, J. Electrochem. Soc. 147 (2000) 3190.
M.D. Levi, K. Gamolsky, D. Aurbach, U. Heider, R. Oesten, J. Electroanal. Chem.
477 (1999) 32.
D. Aurbach, D. L. Levi, E. Levi, H. Teller, B. Markovsky, G. Salitra, J. Electrochem.
Soc. 145 (1998) 3024.
ter 15
POLYANION-BASED POSITIVE
ELECTRODE MATERIALS
C. Masquelier, S. Patoux, C. Wurm and M. Morcrette
oire de Reactivite et Chimie des Solides, UMR CNRS 6007, Universite Picardie Jules
Verne, 33, rue Saint-Leu, 80000 Amiens, France
INTRODUCTION: FROM OXIDES TO

IONICALLY CONDUCTING POLYANIONIC
FRAMEWORKS
ver the last 15 years the vast maJonty of fundamental and
ogical contributions on the search for better positive electrode
s has been devoted to transition metal oxides such as LixM02 (M =
1
, Mn), LixMn204, LixV205 or LixV 308. The first two classes of
s, built on close-packed oxygen stacking adopt bi-dimensional and tri
onal crystal structures, respectively. Lithium ions may be easily
ated or extracted from these structures in a reversible manner. These
are reasonably good ionic and electronic conductors and lithium
3
n/extraction proceeds while operating on the M4+/M + redox, located at
4V vs. Li+/Li. Cost considerations brought some special attention to
ible use of LiFe02 that may be prepared in the same crystallographic
ment as of LiCo02 through ion exchange from a-NaFe02. 2
4 3
hemical extraction of lithium would give access to the Fe +/Fe + redox
but no stability of Li 1_xFe02 was successfully demonstrated yet.
he peculiar structure of the two-dimensional oxides Li 1_xM02 (M =
Fe, Mn) lead to structural instabilities when the number x of extracted
is high (end of charge). Irreversible motion of transition metals within
um layers may occur and lead to important capacity loss on cycling. In
l use, for instance, only 150 mAhlg out of the theoretical 273 mAhlg
or LiCo02 in commercial Li-ion batteries.
3 6
or this reason, Goodenough and Thackeral· had envisioned in the
ghties the possible use of three-dimensional oxides such as the spinel
04• In such spinel structure, Mn06 octahedra are strongly connected
other through edge-sharing and delimitate a three-dimensional
Chapter 15
rk of conduction paths for lithium motion. One of the major problems

he spinel LiMn 204, though, is that there is a -1V difference for the same
Mn3+ couple between the extraction of lithium from the tetrahedral 8a
-4V vs. Li+/Li) and the insertion of lithium in the octahedral 16c sites
vs. Lt/Li). This limits then the theoretical capacity at 4V on lithium
tion from LiMn204 (equivalent to Li05Mn02) to values similar to the
al capacities delivered by LiCo02 or LiNi02•
The concept of three-dimensional insertion/extraction positive
des was then extended in the late 80's to new families of 3-D
78 9 10
work structures, by Goodenough's group ' and by Delmas. ' These
work structures are not built from close-packing of oxygen atoms but
n polyanions such as tetrahedral XO/- groups (X = P, Mo, W, S).
The NASICON framework (Natrium Super Ionic CONductor), which
e described in much details through this contribution was considered as
larly attractive from structural stability and fast ion transport
erations. The originality of the idea of using this class of materials as
on electrodes comes from the fact that previous studies had been
ng on developing NASICON compositions as solid electrolytes in Na/S
atteries.
[001]
_f_
j-*_
j_
J-.\11
>1: l\.'12 M3
15.1. Schematic representation of two NASICON compositions Na3Fez(P04h (left) and

P04h (right).
The NASICON structure of general formula AxMM' (X04)3 is built on a

imensional framework of M06 and M'06 octahedra sharing all their
s with X04 tetrahedra and vice versa (Figure 15.1). The basic
X04h repeating unit, named as "lantern", is made of three tetrahedra
cted to two octahedra. Each lantern is connected to six other lanterns
generates overall a large interstitial space that may accomodate
n 0 to 4 alkali cations per structural formula. The exceptional
bility of this structural arrangement to numerous chemical substitutions
n-Based Positive Electrode Materials 447
e An+, Mn'+ or xn"+ sites has been widely demonstrated. Importantly,

ber of alkali cations per structural formula may be adjusted depending
stable oxidation states of transition metals and on the X element
Below, the elements which may give NASICON compositions are
A= Li, Na, Mg, Ca

M or M' = Sc, Ti, V, Cr, Mn, Fe, Zr, Nb, In
X = Si, P, S, As
or M' elements form octahedra, while X elements form tetrahedra.
or instance, niobium, titanium and iron, mostly stable in air at their
n states +5, +4 and +3 respectively lead, when combined with the
nion, to compositions such as NbTi(P04)3, LiTiz(P04h. Na2FeTi(P04)3
3Fe2(P04)3 that all adopt the NASICON structure and are easy to
as pure powders. Other examples include Fe2(S04)3 and
Si04)3.
preparation of these numerous compositions is well documented in
ature and was aimed at trying to establish close relationships between
structures, alkali ion contents within the interstitial space, chemical
11
f substituting elements, and ionic conduction. Within the Na1+xZr2P3_
(Og 3) solid solution, a pronounced maximum of ionic conductivity
n first discovered by Hong 12 for x=2 while the two end members
13 14
P04)3 (x=O, ]) and Na4Zr2(Si04h (x=3, ) are poor ionic
conductors.
emical and structural characteristics responsible for the fast ion
rt in the NASICON 15 structure were clearly addressed in a fundamental
ution of Goodenough and are summarized as follows :
ighly covalent three-dimensional framework generating a large

terstitial space,
lasticity of the framework for better accomodation of local
ompositional changes,
eak framework - alkali cation interactions,
lectrostatic interactions between neighbouring alkali cations,
hree-dimensional network of interconnected conduction pathways,
on reducible transition element when in contact with metallic
lkali cation.
the transition metal, located in octahedral environment in the present

ay be reduced or oxidized, the above considerations indicate that the
ON structure is very interesting as well as a host for reversible alkali
rtion/extraction.
s will be developed in this contribution one of the major drawbacks of
cture as an insertion electrode is its very poor intrinsic electronic
ivity (major advantage for an electrolyte !!): no direct ..-M-0-M-..
Chapter 15
nic delocalization is possible as the M06 octahedra are isolated from

ther and separated by the X04 groups (Figure 15.1). As an example, the
distance in NaZr2(P04)3 is of 4.75 A. An interesting study aiming at
ying the electronic conduction within mixed-valence NASICON
16
res was reported by Tillement. In this pioneer work, Tillement
bes in much details the possibility of preparing NASICON structures
nusual oxidation states for transition elements, such as NbiV, Tim, Fen.
as achieved either through synthesis in sealed container under vacuum
ugh electrochemical insertion of Na+ ions. We will develop in further
5 14 4 13
these important early results on the localization of Nb + +, Ti + + and
redox couples vs. Na+/Na. 17- 19
In this contribution, we will attempt to establish an overall review as
ete as possible on the ongoing research on alkali insertion/extraction
three dimensional framework structures of transition metal elements.
ASICON structure will be used first as a didactic example to illustrate
y concepts brought to our attention from the recent works of Padhi :
The influence of the chemical nature of the XO/. anion (X= P, As, S,
Mo, W) on the position of a given Mn+IM(n-I)+ redox couple vs. Lt/Li.
8
This was first addressed in 1989 by Manthiram
For a given XO/- polyanion, the relative positions of several transition
metal redox couples.
The concept of using polyanionic structure materials as positive
des in rechargeable lithium batteries will then be discussed through s
structures that were examined recently such as LiMP04 (M = Fe, Co,
ixMM'(X04h (M,M' = Nb, Ti, Fe, V; X= S, P, As), LixMXz07 (M =
Ti; X= P, As), LixMOX04 (M = V, Nb; X= P, As, S), FeP04.nHzO,
07)3. Depending on the MIX ratio, these materials offer theoretical
ties ranging from 170 mAh/g (Lh.xFeP04) to 128 mAh/g
Fez(P04)3) to 116 mAhlg (Lii+xFePz07).
TUNING THE Mn+IM(n-t)+ REDOX COUPLE IN THE

NASICON STRUCTURE BY CHANGING THE
CHEMICAL NATURE OF THE XO/- GROUPS :
INDUCTIVE EFFECT
The first reports dealing with lithium or sodium electrochemical 20

on in a NASICON-type polyanionic framework are those of Nadiri,
1 2 7
Torarde and Manthiram on molybdates and tungstates LixFe2(M04)3
3 2
Mo, W). Electrochemical insertion of lithium operates on the Fe +/Fe +
couple, located at 3V vs. Li+/Li for the whole compositional range
2). A two-phase reaction takes place between the monoclinic (P2/a)
Fe2(M04)3 and the orthorhombic (Pnma) form of LhFe2(M04)3. This

hase may be prepared by chemicallithiation using Lil.
23
ruce undertook a similar study on the insertion of Na+ in defect
compositions and confirmed the similar behavior of Fe2(Mo04)3 and
04)3. Interestingly, the cell voltage is of 2.7 V, as measured vs. Na+/Na,
V lower than when lithium is inserted vs. Lt/Li. This indicates that the
2+ redox couple is located at the same value, the 0.3 V difference being
related to the difference between the Li+/Li and Na+/Na redox couples
15.2a). This observation is a clear signature on the weak interactions
n the alkali cations and the oxygen atoms of the NASICON
ork, which is of importance for good ionic conduction.
E
a)
G
Li•/Li
Na•/Na c;;;;;,;;;iiiiiiiiiiiijii;l
.2. Illustration of the cell voltages measured for a) lithium or sodium insertion into
4 h and b) lithium insertion into Fe2 (Mo04)3 and Fe2 (S04)3.
significant early contribution on the evaluation of the insertion

8
es of polyanionic structures was given by Manthiram who
trated the good reversibility of the Fe2(S04)3/LhFe2(S04)3 system,
g once again on the Fe3+/Fe2+ redox couple, located this time at 3.6 V
Li. A 0.6 V difference was hence pointed out for the same redox
into same structural arrangements that differ only in the nature of the
olyanion (Figure 15.2b). This was explained in a convincing manner
nature of the inductive effect of xo4 groups that, depending on the
egativity of X, weakens or strengthens the covalency of the Fe-0
n the NASICON structure, each transition metal is involved in six M-
quences. As a result, substitution of S for Mo
3 or2 W weakens the Fe-0 8
nd increases the difference between the Fe +/Fe + and Li+/Li couples.
further illustration of the inductive effect on the position of the
2+ couple in the NASICON structure is given from the rhombohedral
LhFe2(P04)3 that may be prepared through ion exchange from the
Chapter 15
m counterpart Na3Fe2(P04)3•24 26- In this case, lithium insertion proceeds

complicated series of subtle phase transitions located at an average
ial of 2.8 V vs. Li+/Li, 27-29 for the complete reduction of Fe3 + into Fe 2+.
3 2
ng shift in the position of the Fe +/Fe + couple (0.8 V) is hence observed
2
ing from S04 z- to POl groups. The polarization of the 0 - anion
ds Pv is weaker than towards svr and this leads to Fe-0 bond strengths
r in LhFez(P04)3 than in Fez(S04h One should note that the above
rison is drawn for NASICON structures that host different amounts of
m ions (0.$X2 for LixFez(S04) 3 ; 3,$X5 for LixFez(P04) 3). This could
small impact on the relative positions of the Fe 3+/Fe2+ couples. Padhi
30
d, additionally, that lithium insertion in the mixed composition
S04)z(P04) occured at -3.3 V vs. Lt/Li, i.e. at intermediate values
en the pure sulfate (3.6 V) and the pure phosphate (2.8 V).
These early considerations have prompted the identification of new
2
e electrode materials with attractive operating voltage for the Fe3+/Fe +
in lithium batteries and theoretical capacities in the range of 120
. They provided as well an elegant measure of Fe-0 bond strengths
ed by XO/- polyanions through the inductive effect, summarized
(Figure 15.3):
2.8 v 3.0V 3.6 v

3 2
5.3. Positions of the Fe +/Fe + redox couple vs. Li+/Li in NASICON compositions
X04h·
1. Relative Positions of Various M"+/M<n-l)+ Redox

Couples (M =Fe, Ti, V, Nb) in Nasicon Phosphates
2
We indicated above how the value of a given redox couple (Fe 3+/Fe +)
be influenced by the chemical nature of the tetrahedral XO/- group. In
ction, we fix the X04 n- polyanion as a phosphate POl group. A very
riety of AxMM'(P04) 3 compositions having the NASICON structure is
ible and the number x (0.$X 5) of alkali cations is governed by the
cal nature and the oxidation states of M and M'. Besides the
erable amount of contributions in the literature on how the ionic
tion of NASICON compositions would be nfluenced by the amount of
cations and by the nature of (M,M'), only few studies have been
d so far on their behavior as hosts for reversible alkali
-Based Positive Electrode Materials 451
31
/extraction. The major advances were reported by the groups of
Quartoii and Goodenough and are summarized below.
4
V.,7 V.,
Is, 1.5 t.v 2o;o 4>3

2 2.5 3 3.5 4 4.5
X in B-L M 2 (PO 4 )3 (M = V, Ti, Fe)
.4. Voltage-composition profiles recorded vs. li+/Li during lithium

xtraction into/from various NASICON compositions li..M2(P04)) (M =Fe, V, Ti)
Position of the Ti4+rre+ Couple vs. Li+/Li and Na+/Na

9 10 32
elmas • · showed the possible insertion of lithium or sodium into
04)3 and NaTh(P04)3, respectively, to reach the end-members
04)3 (A= Li, Na) into which titanium is at the oxidation state +ill.
32 33
determined frrst, recently confrrmed by Aatiq and Patoux that
insertion into LiTiz(P04)3 occurs at 2.48 V vs. Li+/for the Ti4+/Te+
uple according to a two-phase mechanism between LiTiz(P04)3 and
33
04)3 (Figure 15.4). The in situ X-ray diffraction data of Figure 15.5
discharge the progressive disappearance of LiTiz(P04)3 peaks at the
f those of Li3Tiz(P04)J.
or LhTiz(P04)3, the appearance of an XRD peak at around
0° (d = 2.30 A, (303)) indicates that the space group R-3c (into which
04)3 crystallizes) is no longer valid. On going from LiTiz(P04)3 to
04)3, Delmas pointed out a remarkable increase in the da ratio of the
al unit-cell, due to stronger electrostatic repulsions between
4)3] lanterns along [001] when the M1 site, located halfway between
erns, is emptied (Figure 15.1).
Chapter 15
1 §fj (J 0
lg
31
1 8-l\.00
1 gg
M'
* )I J\.
ql )... - [A]
'
[U]
::;. :,.; .
·
:,..,.
..,.;_
"A
36 38 40
IAI
24 26 28 30 32 34
0
29 ( )- CuKa
15.5. In situ X-ray diffraction patterns recorded during a full discharge and charge cycle
h(P04)3 I Li cell.
In a first approach, Li3Tiz(P04)3 adopts the crystal structure of

34
(P04)3 determined by Masquelier where u+ ions occupy a set of
dral sites labelled as M3, shifted by - 0.8 A along [001] from the M2
n. From a careful investigation coupling solid state Li NMR and
33
r neutron diffraction, Aatiq et al. indicate that the situation is even
omplex as lithium appears to be distributed on two similar tetrahedral
3' and M3", occupied in the ratio 2/3 and 113, respectively.
As shown in Figure 15.6, the isotypic composition NaTiz(P04)3 is also
towards the electrochemical insertion of sodium according once again
o-phase mechanism. The reversible redox process occurs at 2.15 V vs.
a for the Ti4+/Te+ couple. This value is once again a clear
stration of very weak interactions between the alkali cations and the
work in the NASICON structure as the -0.3V difference between the
ng voltages of Li/Lh+xTiz(P04)3 and Na/Na1+xTiz(P04)3 reflects the
nce between the Lii and Na+/Na couples. Patoux showed recentl/ 5
dium insertion into NaTiz(P04)3 results into Na3Tiz(P04)J isostructural
a3Fe2(P04)3 where the M(1) site is still fully occupied and M(2) at 2/3.
4
3.6
zGIS I
3.2
-
2.8
fli
GIS 2.4
..=.
1
2
1.6
..e 1.2
1.5 2 2.5
2
x in
Na
Ti ..
l
(PO
) J
l
4.5 s
.6. Voltage- composition profiles recorded vs. Na+/Na during sodium

extraction into/from various NASICON compositions Na,.M2(P04h (M = Fe, V, Ti).
. Relative Positions of the Ti4+m3+, Fe3+/Fe2+, Nb 5+1Nb4+,

Nb4+1Nb3+, V4+N3+ and V3+N2+ Couples vs. Na+/Na and
Li+/Li in the NASICON Structure
first series of detailed investigations had been conducted in the early
Quarton's group who demonstrated Na+ reversible insertion into
mixed NASICON compositions NaxMM'(P04)3 (M = Fe, Ti, Nb, 17 18
Peculiar attention was given to the compositions NaxTiNb(P04 h, •
16 19
Nb(P04)3,18' 19 Na1+xScNb(P04)3 and Na2+xFeTi(P04)3. '
36
An
1
nt result was that the position of a given M +JM<n· >+ redox couple is
11
independent of the chemical nature of the second transition metal

M', due to the absence of direct conectivity between two {M,M')06
ra. For instance, the Ti4+rre+ couple was found to lie at the same f
-2.2 V vs. Na+/Na in NaxTiNb(P04)3, Na2+xFeTi(P04)3 and
33
z(P04)3. Additional experiments performed recently by Patoux on
M(P04)3 (M =Fe, Cr) are depicted in Figure
15.7. 2 2
3he case of Na +xFeTi(P04)3 is instructive as Fe + 7 Fe + reduction
4
trst at -2.45 V vs. Na+/Na before reduction of Ti + 7 Te+ at 2.2 V.
9
a obtained for Na 1+xFeNb(P04)/ and Na3+xFe2(P04)3 (Figure
3 2
15.6) the position of Fe +/Fe + vs. Na at -2.5 V. Besides this, it is
of
nce to note that only one sodium ion may be intercalated into
Chapter 15
1) E • :us V .._ Na•/Na

1) Fe+-+ Fel+
1) E •:UO V ...Na•/Na
2) Ti4+ -+ Til+
1) E •l.15 V ...Na•/Na
';
-;::
:u
c c::u
. ..c. l
1.8
0
15.7. Voltage- composition profiles recorded vs. Na+/Na during sodium
n/extraction into/from NASICON Na2+xTiM(P04h (M =Fe, Cr).
35
(P04)3 as a result of the complete filling of both M(l) and M(2) sites.
5 4 3
ent was the ftrst to locate the Nb +/Nb + and Nb4+/Nb + redox couples in
SICON structure, at 1.90 V and 1.20 V vs. Na, respectively.
As shown in Figure 15.6, the NASICON composition Na3V2(P04)336 may
either oxidation of V3+ -7 V4+ through sodium extraction or to partial
2
ion of V3+ -7 V + through sodium insertion. The difference between
two couples is of 1.8 V (3.4 V and 1.6 V vs. Na+/Na, respectively).
in situ X-ray diffraction during cell operation, Patoux showed that the
tion of sodium from Na3V2(P04)3 to NaV2(P04)3 is a two
phase n, the latter phase being isotypical with NaTh(P04)3.
A similar approach was followed by Goodenough's group to evaluate
ative positions of these same transition elements couples when lithium
rted or extracted into/from the NASICON framework. To this end,
M'(P04)3 compositions were prepared through Na+- u+ ion exchange
the stable37 39sodium analogs 37immersed
39 in molten
27 29 40LiN03 at 300°C.
b(P04)3, " Lh,FeTi(P04)3, ' Li3Fe2(P04h, ' " and LhV2CP04)
3
4143
ong the most significant examRles. in situ X-ray diffraction
Recent neutron diffraction4 <a.b),44 and
45
ments demonstrated that the the composition LhV2(P04)3 is
ctural with LhFe2(P04)3 and LhTh(P04)3. Lithium extraction leads to
P04)3 isostructural with LiTh(P04)3 (Figure 15.8). This occurs at an
ting potential of 3.75 V vs. Li+/Li, among the highest for polyanionic
res. The overall data relative to lithium insertion or extraction is
tly consistent with the one obtained for sodium
LiV,(I :(4 V)
-
Li 3V1(P04)3 (3.4 V)
24 28 32 36 40
29 (")- CuKa
5.8. X-ray diffraction patterns collected in situ for NASICON LiM 2(P04h and
4)J (M = Ti, V)
Vs Li • I Li
1.5 1.7 2.2 2.5 2.8 3.7
V4•t3•
Vs. Na • INa
Nb4•/3• Nb5•t4• Ti4•t3• fe3+12•
5.9. Positions ofMn+tM<n·l)+ redox couples in NASICON AxMM'(P04)J.
n/extraction and is summarized in Figure 15.4 and Figure 15.9.

ll these studies, summarized in Figure 15.10, have identified a wide
of redox potentials for the NASICON frameworks containing le
/oxidizable transition metals. The values of these couples are
y dependent on the nature of the X04 tetrahedral polyanion and on the
n metal itself. The NASICON framework, additionally, is found to be
ably stable over a large range of alkali ion content. Electrochemical
n of lithium has allowed for instance to isolate for the first time
ON structures (Li5Fe2(P04)3 and Li5Vz(P04)3) containing 5 alkali
27 40
per structural formula. •
Chapter 15
0 2 3 4 5
x(Li)
V 2 (504)
3 .... .-... ¥. Li2 V2 (504 )3
Li Fe (50 )
3.6 v
Fe (50 )
..............
2 4 3 2 2 4 3
2 5 et 2 . 2 et 1 . 5 V
TiNb(P0 4h .,.. ... .... ... .. . ... ... ..3. .... ... 4Li3 TiNb(P0 )
2
LiTi4
(P0 ) ..... :. ... Li Ti (P0 )
3 3 2 4 3
2.8 et 2.5 v
Li2 FeTi(P04)3 .,......::::......... Li4 FeTi(P04 h
.....2....8....v
.....
......3........v... .....1.........v......
15.10. Selected examples ofNASICON compositions obtained electrochemically (in

rom various starting electrodes (in blue).
EXTENSION OF THE CONCEPT TO OTHER

POLYANIONIC STRUCTURES
The third part of this contribution deals with a review, as complete as

le, on latest discoveries and achievements in shifting the concept of
polyanionic insertion hosts, from an academic curiosity to attractive
ates for real systems. As stated above, these materials take full benefit
inductive effect that, for a given transition element redox couple, lead
her operating voltages than for "simple" oxides. The presence of the
y" X04 groups is of course penalizing, compared to the oxides, in terms
oretical specific capacity that can be achieved. However, as we showed
"didactic" NASICON family, this is partly compensated by the high
ty of these frameworks that allow 46 to fully extract or insert lithium
t irreversible structural changes. The theoretical capacity of 138
for the reaction LiTh(P04) 3 Li3Th(P04)3 at 2.5 V vs. Li+/Li is
vely attained experimentally, once smart electrode preparation is
med to favor electron transport into these mostly insulating materials.
ost exciting example, developed in details further below is that of
04 out of which several groups have recently succeeded in obtaining
mental capacities very close to the theoretical value of 170 mAh/g.
.1. Selected examples of iron-containing phosphate structures that reversibly insert t

potentials ranging from 2.5 to 3.5 v vs. u+/Li.
M/P =1
M/P =1
M/P = 2/3
M/P = 2/3
M/P = 1/2
he selected examples given in Table 15.1 are relative to a same given

2
+ redox couple for two kinds of polyanions, Pol- and P207 4-. This
es the wide variety of potentials accessible
38 depending on the crystal
e, as previously addressed by Padhi. Table 15.1 illustrates as well the
penalty in terms of theoretical capacity when the Fe/P ratio decreases
FeP04 to LiFeP207.
he examples developed below include several structural families,
phosphates, of iron, vanadium, titanium and niobium.
. The Monoclinic Forms of Fe2(S04h and LhM2(X04h

(M =Fe, V, X= P, As)
epending on the preparation procedure, Fe2(S04)3 may adopt two
crystal structures where the Fe2(S04h lantern units are stacked
tly in three-dimension. The rhombohedral form is of NASICON
47
e, fully described in the space group R-3. The lantern units are
parallel to the [001] direction 48of the hexagonal unit-cell. In the
inic form, space group P2/n, the lantern units are stacked
llel, alternately along the directions -(2b ± c) of the monoclinic unit
This arrangment of M06 octahedra and S04 tetrahedra results in a
n structure than in the NASICON form, less favourable for alkali ion
rt. This is illustrated by the schematic representations of Figure 15.11
Chapter 15
fferentiate the bottlenecks in both monoclinic and NASICON forms of

04)3 : 6-membered "rings" for the monoclinic form, 8-membered "rings"
NASICON form.
24 51 52 53 and
The monoclinic forms of Li3Fe2(P04)3, ' ' ' LhFe2(As04)3 •
54 27
55 56
P04)3 ' are easily prepared49by solid state reaction at 700-900°C. They
structural with Lhln2(P04)3.
Rhomboedrique (R-3)
Monoclinique (PZ/n) Echange d'ions
Vo1e ceramique
A B
,,,
;;;
,,,
,,,
15.11. Schematic representations of the monoclinic (left) and hexagonal-NASICON

orms of Fe2(S04)J.
8
Manthiram was the first to report on the insertion of lithium into 56
linic Fe2(S04) 3 at an attractive potential of 3.6 V vs. Li+/Li. Okada
d that NASICON and monoclinic Fe2(S04)3 had similar electrochemical
ures. Insertion of lithium into monoclinic Fe2(S04)3 occurs through a
hase reaction, the end-member LhFe2(S04)3 being of orthorhombic
51
etry (Pcan).
In the same study, Okada investigated3 for 2the first time the behavior of
linic LhFe2(P04)3 and located the Fe +/Fe + couple at 2.8 V vs. Li+/Li.
point, Li3Fe2(P04)3 was envisaged as an interesting buffer material to
ed with Fe2(S04) 3 for protection against overdischarge. More detailed
27 29
gations, ' revealed though that two hi-phasic reactions take place at
and 2.73 V vs. Lt/Li with the existence of a definite composition
56
P0 4)3 in between. In situ X-ray diffraction during a discharge process
ed that the framework is very well-maintained and that the intermediate
involved are extremely similar to the pristine one. From
iodynamic Cycling 29 with Galvanostatic Acceleration technique,
ette determined that these phase transitions are kinetically limited
by electron transport within each phase rather than from phase front
ion (Li diffusion) (Figure 15.12).
0 3.5 . . . . . . . . ,. . . . . . . . . f.:.. . . . . :. . . . . . . .
- . Ml-+-+.-+-f-4-..JJ
-513 ....
-10 2.5
; ,----
!····..···············....1·························t····················.. ···t··"
3.5 4 4.5 5
10 20 x in A-Li Fe (PO )
Timelh • 2 4 3
.12. Discharge in PCGA mode (left) and galvanostatic cycling at C/10 (right) of a
c Li3F (P04)3 I Li cel1.29
dditionally, Morcrette showed that reversible capacities as high as

Ah/g at C/10 regime could be attained with an excellent capacity
n once particle sizes and surface were optimized. Tailored particles
ained chemically by a low temperature synthetic route and their
was then modified by ball milling of carbon/Li e2(P04)3 mixtures.
uch attention has been given recently to the monoclinic LhV2(P04)3
be oxidized throuh lithium extraction at potentials ranging from 3.4
V vs. Li+!Li. 55,56,s -60 LhV2(P04h shows a complicated series of four
ve two-phase transitions on Li extraction towards Lio. 1V2(P04)3
15.13). From in situ X-ray diffraction, Morcrette45 identified that all
rmediate compositions were fully indexed in the monoclinic space
P2yn, with a unit-cell contraction of 11VN = -6.8% between
04)3 and LiV2(P04)3 and a slight increase on further oxidation
to
P04)3• Despite these numerous phase transitions and non-negligible
changes, the monoclinic Li 3V2(P04)3 system is structurally
highly
le and, once vanadium dissolution in the electrolyte will have been
d, will stand as a promising positive electrode for Li-rechargeable
.
ithium may be inserted as well into the monoclinic arsenate
2
s04)3, towards LisFe2(As04)3 through reduction of Fe3+ into Fe +. The
27
of the potential-composition curve is very similar to that of
P04h, with the existence of an intermediate composition
As04)3 between two two-phase reactions. These reactions take place at
nd 2.62V vs. Li+/Li, i.e. are better separated than for LhFe 2(P04)3
but age voltage is very similar. Substitution of AsOl for PO/ has
very nfluence on the global inductive effect on iron due to the
61
similar egativities of As and P. This was confirmed by Gaubicher •
62
who
Chapter 15
same redox potential values (4.0 V vs. Li+/Li) for the V 5+N4+ couple in
P04 and LixVOAs04.
5
;::3
4.5
4
ta
'+=I
5 3.5
p
0..
3
0 0.5 1 1.5 2 2.5 3 -20 0 20
x in LiX V2 (PO4 )3 -dxldV(V 1)
15.13. Cycling of a monoclinic U 3V2(P04h I Li cell in GIIT mode (left)

and onding derivative curves (right).45
2. The Rhombohedral Form of Li3Fe2(As04)3
A second form of Li (As04)3 maJ be prepared through ion exchange

he sodium analog Na e2(As04)3•63' This structure is unusual for being
n Fe4018 clusters into which one Fe(1)06 octahedron is surrounded by
Fe(2)06 octahedra through edge sharing. Alkali cations are distributed
o independent crystallographic sites, A(1) fully occupied, and A(2)
ed at 5/6 according to the structural formula :
64 65 3
ire and Masquelier showed that partial chemical reduction of Fe +
2
e + could occur through adding of extra sodium ions in the interstitial
according to the structural formula :
The electrochemical insertion of lithium into this unusual form of

(As04)3 allowed to demonstrate the preferential reduction of Fe(1).
g the first discharge, two clearly distinct redox processes occur as
lation plateaus located at 3.1 and 2.4 V vs. Li+/Li. As seen in the inset,
t process at 3.1 V is highly reversible. It corresponds to the insertion of
per LhFe2(As04)3 unit, hence to the total occupation of the A(2) site
ce, by analogy with the sodium analog, to the reduction ofFem(1) into
The second process corresponds to reduction of Fe on the Fe(2) sites
versible structural changes occur.
. The Diphosphates and Diarsenates LixMX207 (M =

Fe, V, Ti; X= P, As)
s Table 15.1 indicates, LixMX 207 compositions 66• 67 are clearly not
itive electrode materials in terms of theoretical capacities to be
d. They provide however an interesting panel of crystal structure
ments and electrochemical insertion mechanisms that deserve to be
68
mentioned here. In a comparative study, Wurm described the
tion of LiFeP207, LiFeAs207 and LiVP207 as pure microcrystalline
s via an aqueous solution route that allowed the formation of finely
ed small particles.
he redox couples that are accessible are Fe3+/Fe2+ for Li 1+xFeX207
and,
4 2 67 33
ingly, V +N 3+ and V3+N + for Lia xYP207 • Patoux showed the
4 3
te reduction of Ti + to Ti + from TiP207 to LiTiP207• LiFeP207
and
49 69 70
7 are isostructural ' ' and crystallize in the space group P2J. As
in Figure 15.14a, each Fe06 (or V06) octahedron is linked to five
t P207 groups, one of them acting as a "chelating" sequence around Fe
LiFeAs207 adopts its own crystal structure,71 described in the space
C2. There is in this case no bidentate corner sharing, and thus each
tahedron is linked to six different As207 groups (Figure 15.14b).
.14. Partial representation of the crystal structures of (a) LiMP 207 (M =Fe, V) and
s207.
7
adhi showed firse that the electrochemical insertion of lithium into
07 occurred at 2.9 V vs. Li+/Li i.e. at a slightly higher operating voltage
r LhFe2(P04h. However, at this early stage of investigations on
onic structures, only part of the theoretical capacity (-60%) was
Chapter 15
68
ed due to the poor electronic conductivity of such material. Wurm
d that the electrochemical activity of these materials is drastically
ced by an intimate mixing of finely divided particles of pristine
als with conductive carbon, through ball milling.
3.6
Fe, P
il
;.::3
.;
>
03
· . p 2.2
.@ 2.0
1.8
1.6
1.0 1.1
1.2 1.3 1.4
xdans LiX MX2 07
15.15. Potential-composition curves recorded in GITT mode during the first discharge
MX207 (M =Fe, V ; X= P, As) I Li cells.
This mandatory step resulted in producing amorphous Li-M-X-0

ns at the surface of crystalline LiMX207 crystallites and a continuous
e decrease is first observed as lithium insertion proceeds into the
hous part of the composite electrodes ("A" region of Figure 15.15).
Crystalline LiMX207 particles react with lithium through two-phase
nisms, characterized by insertion plateaus located at 2.90 V for
3 2 3 2
207 (Fe +/Fe + couple), 1.99 V for LiVP207 (V +N + couple) and 2.55 V
FeAs207 on first discharge. The Li-driven second phase is crystalline,
larger unit-cell in the case of isostructural Li 1+xMP207 (M = Fe, V). It
orphous in the case of Li 1+xFeAs207 that shows, interestingly, a
allisation process on charge (when Li is extracted from Li 2FeAs207)
bsequent cycling at -2,90 V.
As for Li3V2(P04)3, LiVP207 offers the possibility of operating on the
+ redox couple if lithium is extracted. The electrochemical extraction of
hium from LiVP207 occurs as a two-phase process at a relatively high
rium potential of 4.26V vs. Li+/Li (Figure 15.16). The compound
ed at the end of the oxidation (either electrochemically or chemically) is
exempt of lithium and the transformation between LiVP207 and VP207
placive, i.e. without breaking of V-0 and P-0 bonds. The crystal
re of VP207 was determined from Synchrotron X-Ray powder
72
tion. The framework of LiVP207 is maintained upon lithium
on, and VPz07 crystallizes in space group P2J. Compared to LiVP 207,
a very lirrge cell distortion due to a modification of the shape of the
that contained lithium ions along with an oxidation of the vanadium
iV that induces shorter V-0 bond lengths.
10
5
-"? - 0
X
------:::::::::--;::--..:::=!
-5
3.75 4 4.25 4.5

Potential (V) vs Li
.16. Electrochemical data recorded in PITI mode during lithium extraction/insertion

LiVP2 68.
atoux33 showed the reversible insertion of lithium into Li.xTiP207 73/ 4

ceeds through two reversible re4ox processes, at 2.63 V and 2.57 V vs.
Figure 15.17). From potentiodynamic intermittent titration technique
itu XRD, a complex series of phase transitions was reported in details.
mation of Lio.96TiP207 at the end of discharge corresponds to a global
volume expansion of ll.VN = + 4.4% from the pristine material
The in situ XRD data revealed the good reversibility of the system as
D pattern of the pristine material TiP207 was recovered at the end of
oxidation. This is a common feature to many polyanionic framework
es built on phosphate groups that easily accommodate relatively large
expansions/contractions over extensive electrochemical cycling.
Amorphous or Crystalline Iron Phosphates:

FeP04.nH20 and Fe4(P207)J.4H20
75 76
ecent contributions from Whittingham's and Masquelier's groups
77 8
as those of Hong and Prosini' indicate reversible insertion of
into amorphous hydrated Fern phosphates such as FeP04.2H20 and
7h.4Hz0. As seen in Table 15.1, these compositions present attractive
al capacities, among the highests for iron phosphates. Such
nds are easily prepared from precipitation in aqueous media which
powders of submicronic particle size with high surface area.
Chapter 15
3.4 Oxl Ox2
3.2
c
3
.,;
..
;.. 2.8
>. .
2.6
-;:
c
....
0
. 2.4
2.2
2
1.8 0.2 0.4 0.6 0.8 1 1.2
0
15.17. Potential-composition curve ofTiP 20, recorded in a GIIT mode vs. u+/Li.
The ot draws the incremental capacity voltammogram.42
29 76
As stated by Morcrette and Masquelier, the poor electronic
ction of Fern phosphates may be partly compensated by selecting highly
d powders intimately mixed, or "coated" (through ball-milling) to
ctive carbon. The beneficial role of constitutional water, that may act as
moter for faster ionic diffusion has been demonstrated. Within the
ing voltage window used (2V-4V vs. Li) these water molecules look
cted as very good electrochemical stability was demonstrated for
ive cycling. This stability suggests that, as for the crystalline form of
·2H20, where the 6-fold coordination of Fe is ensured by 4 oxygen
ing to P04 groups + 2 oxygen belonging to H20 groups, a similar local
ement occurs for the amorphous FeP04 • nH20 powders.
Despite a whole panel of kinetic responses that depend strongly on
ic parameters such as powder crystallinity, morphology, texture, and
de preparation, all the materials investigated in Ref. 76 lead to a very
r operating voltage vs. Li, i.e. comprised mainly between 3.4 V and 2.6
for crystalline LiFeP0 4 and Li3Fe2(P04)3, the position of the
Fe3+/Fe2+ couple is thus strongly modified by the inductive effect of
the P04 that diminishes the strength of the Fe-0 bonds, even for
amorphous sitions.
Capacity (mAh/g)
40 20 40 60 80 100 120 140
Amorphous
::3 FeP04.2H20 _......
>>"'
'-' 3.5 Galvano, C/S
c;; 3
·.::::
11)
0
tl.. 2.5
2 _L
0 0.2 0.4 0.6 0.8

x in FePO .2H 0
4 2
5.18. Li insertion in amorphous FeP0 4.2H20 at a C/5 regime. Electrolyte: LiPF6

(1:1).
Capacity (mAh/g)
4 0 20 40 60 80 100 120 140
Amorphous
;::1 Fe4(P207),.4H20
">' 3.5
_......
GITT
e:,
3 3
'i::
B
0
tl..
2.5 Eeq. = 3.2 V
2
0 0.5 1.5 2 2.5 3 3.5 4
x in Li Fe (P 0 ) .4H 0
X 4 2 73 2
5.19. Potential-composition curve of amorphous Fe 4 (P207)3.4H2 0 recorded in a

GITT Li+/Li.77
oth profiles of Figures 15.18 and 15.19 show S-shape voltage profiles
any well-defined intercalation plateau over the whole composition
3 2
The position of the Fe +/Fe + redox couples for both materials were
termined, after a first full cycle in GITT mode, by measuring the
ium potential after relaxing the battery for 2 days at half state of
ge. The obtained values are 3.0 V and 3.2 V for amorphous FeP04·
and Fe4(P207)3-4H20, respectively. This difference may arise from
t local environments around Fe atoms. In this regard, it is interesting
that the higher position of the Fe 3+/Fe2+ redox couple in FeiP207 h-
Fe/P = 2/3) may be related to the fact that Fe is statistically surrounded
Chapter 15
re P04 groups than in FeP04 · 2H20 (Fe/P = 1): it is therefore more

d by the inductive effect.
61 62 79 80
A very complete series of studies was initiated by Gaubicher • • •
llowed by Dupre 81'82and Kerr83 on various LixVOX04 structures84' 85
or the reversible insertion of lithium. As for the NASICON structure,
uctive effect of xo4n· polyanion generates attractive operating voltages
e V5+N4+ and V4+N3+ redox couples in lithium batteries. However, as
discuss extensively, each M06 octahedron is connected to 4 X04
dra only, instead of 6 for the NASICON structure. This generates
r global inductive effect in VOX04-type structures than in AxM2(X04)3.
We will focus here on the relative positions of these couples as a
on of the crystal structure and of the chemical nature of the polyanion.
(010)
5.20. Schematic representation of the crystal structure of a- VOP0 4. V06

octahedra er, P04 tetrahedra smaller.
From a structural point of view, two main classes of VOX04 (X = P,

compositions, noted as a. andshould be distinguished. The a.-VOP0 4
re (Figure 15.20) is tetragonal, space group P4/n and is built of [V05]oo
87
of V06 octahedra sharing comers along the quaternary axis [001].86 •
ur remaining oxygen comers of the V06 octahedra each belong to a
nt X04 tetrahedron. V0As04 adopts only the a. form. The -VOP04
re (Figure 15.21) is orthorhombic, space group Pnma. The V06
dra are strongly distorted and form chains along [100].
(010) (100)
.21. Schematic representation of the crystal structure of-VOP0 4. V06 octahedra

, P04 tetrahedra smaller.
s mentioned earlier, only VOP04 may be prepared in the 13 form.

80
her indicated however that this form may be stabilized for X = As by
g 13-LiVOAs04 from a mixture of VOAs04.2H20 and LhC03 heated
r up to 600°C. As a consequence, the position of the V 5+N 4+ redox
n this structure could be determined by i) electrochemical insertion of
into VOP04 or ii) electrochemical extraction of lithium from
s04. For the two above reactions, the experimental values are
ly close, equal to 3.98 V and 4.02 V, respectively. This is an
al signature of the same electronegativities of P and As. The location
V5+N4+ couple at -4V vs. Li+/Li is at an intermediate value between
as found for LiV2(P04)3 - V2(P04)3 (>4.5 V) and what is known for
61
3.4 V). This was interpreted in details by Gaubicher as a result of the
structural arrangment in I3-VOP04, itself intermediate between a
ON structure (only ..-V-0-X-.. sequences) and an infinite oxide (only
V-...) sequences.
4
Nb5+/Nb + Couple in LixNbP05
he chemical and electrochemical insertions of lithium into two

al forms of NbP05, a-NbOP04 (P4/n) and I3-NbP05 (P2/c) was
35
communicated by Patoux.
mong the complex polyanionic framework groups built of M06
ra (M = Nb, W, Mo, V ...) and X04 tetrahedra (X= P, S, Mo...), the
86
structure of a-NbOP04 is built on one of the simplest arrangements.
cribed in the space group P4/n (Z = 2) with a = 6.388(1) A and c =
) A and consists of P04 tetrahedra sharing their corners with Nb06
ra, which themselves share four vertices with tetrahedra and two with
milar octahedra along [001] (Figure 15.22). Thus, the a-NbOP04
Chapter 15
re can be described as ..0-Nb-0-Nb-0-... type chains, which are cross

by P04 forming a three-dimensional network. 87 The formulation
P04) accounts for this peculiar connectivity. The Nb06 octahedra are
ed with 4 medium Nb-0-(P) bond lengths (1.97 A) in the (a,b) plane,
nger (2.32 A) and one shorter (1.78 A) alternately along the c axis(-
0- sequences). Similar features are encountered for VOP04 and
P04, which are isostructural with a-NbOP04.
a b
5.22. Schematic representation of the crystal structures of a-NbOP04 (left) and 13-
(right).
On the other hand, two kinds of more regular Nb06 octahedra are
ntered in r3-NbP05 that crystallises in the monoclinic space group P2/c
8) with a = 13.028(1) A, b = 5.278(1) A, c = 13.225(1) A and r3 =
88 89
7(1)0 • • This second high temperature form, is characterised as the
er m=2 of the "MonoPhosphate Tungsten Bronze with pentagonal
s" (MPTBp) family of general formula Ax(Nb03)2m(P02)4; (A= Li, Na,
The structure is described as Re03-type chains of Nb06 octahedra,
ted by slabs of P04 tetrahedra delimiting pentagonal tunnels (Figure
. In this structure, double strings of Nb06 octahedra parallel to the b
e isolated, and linked to others by tetrahedra only.
Both structures present distinct behavious with regard to lithium
on (Figure 15.23). r3-LixNbP05 rapidly and reversibly uptakes lithium
g to a sustained capacity of 90-120 mAhlg (i.e. operating to the
4
Nb + couple) for more than 100 cycles at - 2 V vs. Li+/Li. This
bility was confirmed by in situ X-ray diffraction in the range 0x0.7
dditionally revealed a series of single- and two-phase insertion
nisms. On the other hand, a-LixNbOP04 displays an insertion plateau at
V vs. Li+/Li indicative of a two-phase process that was found to be
reversible. Both materials, when discharged at voltages lower than 1
sent irreversible reactions with complete amorphization and reduction
allic Niobium.
0 U OA G.6 0.1 1
x in cx-Li,.NbOP04
0 o.l 0.4 u o.a

X in fl-LixNbP05
.23. Potential-composition plots ofGITT data recorded during U insertion into a.

(left) and (3-NbP05 (right).
The Olivine LiMP04 (M =Fe, Mn, Co)

39 91
ince the demonstration by Padhi et al. ' that lithium can be
d reversibly from triphylite at -3.5V vs. Li+/Li, the material of choice
nsertion/extraction among the so-called polyanionic structures is with t
the triphylite LiFeP04• Lithium is extracted from the octahedral sites o
phase process which, although very attractive for providing a very
age plateau at 3.44V vs. Li+/Li, is penalizing through a kinetically
41
ront phase migration between LiFeP04 and FeP04.
he triphylite LiFeP04 adopts the olivine structure-type, built on an
2
hexagonal packing into which Li+ and Fe + occupy half of the
ral sites and P 1181h of the tetrahedral sites. The peculiar distribution of
2
Fe + within the octahedral sites generates [Te0c2] layers as shown in
5.24. This connectivity has a special impact on both electronic and
nductivity in LiFeP04• Fe06 octahedra share corners between each
Chapter 15
'r-----'' [Te0c2]layer #1
l .iO,, share edges

as chains along
[010)
FeOr. isolated }
from each other
within each
dense layer
15.24. Structure representation of LiFeP04•
not edges, and electronic delocalization is made difficult. Lithium

ctivity is not three-dimensional and proceeds mostly along [001].
son demonstrated, by means of in situ Mossbauer spectroscopy and in
-ray diffraction, the two-phase character of the electrochemical
92
tion of lithium. '93
The theoretical capacity of 170 mAh/g at a flat potential plateau, the
xicity and the cost of iron makes it a very good candidate to be used in
4
rial lithium batteries. Several patents have been published recentl/ ,
HQ. Since Ravet's public anouncement of reaching this theoretical
95
ty at high charge/discharge rates, many efforts have focused on
sing the electronic conductivity at the surface of the mostly insulating
04. Recent significant advances have focused on
increasing the electronic conductivity of composite electrodes through

carbon coating.96 "97'98 '99
100
lowering the synthesis temperature to -500°C.
using hydrothermal reaction, - or solution routes104 •105
101 103
106 107
adding foreign metallic elements such as Cu or Ag or Nb. Note that
107
n this last case, the term "doping" may be unappropriate to explain
he dramatic increase in electronic conductivity of LiFePOJNb
compared to that of LiFeP04•
Careful synthesis procedures are needed to stabilise Fe2 + in the
108
.25. SEM image of optimized :UFeP04 particles (-l -tm).
e: i) prevent the formation of stable Fern phosphates such as LiFeP 207,

P04h and/or amorphous Fern phosphates if low temperature synthesis
en for producing small particles, ii) avoid the formation of iron
res such as FezP if high temperature is chosen for improving the
coated electronic conductivity.
ure LiFeP04 may be easily obtained at 500°C by annealing under
g atmosphere the solid resulting from the evaporation (-150°C) of an
solution containing lithium and iron (III) precursors. The particles
duced are around lJ.Lm in their largest dimension (against 50J.Lm for d
route presented by Padhi et al.) (Figure 15.25). A classical hand
with carbon black to prepare the electrode gives very poor
hemical performances. These are much enhanced by mechanical ball
09
which had to be optimised (duration, absence of oxygen in the
er, etc...).
160
140
'i ii
:c 120 ...
<E
. _,
100 ...
,e. 80
!:! •
'D
c . 60 ;::.
f l ...
u
fl
40
20
OL---
.... 1.1 ... ...
.
x In Ll FePO
. .. --
0 1 u 2 3
Rate (number of Li atoms exchanged per h)
.26. Galvanostatic cycling of a LiFePOJLi cell at various rates at room at 2o•c.

Chapter 15
A similar "solution" route was used to coat chemically carbon at the

e of LiFePO,. in an in situ way. Acrylonitrile was introduced in the
aqueous solution with all the precursors needed for its polymerisation.
ration and polymerisation occur at the same time on first heating in air,
which the polymer is decomposed into carbon particles coated at the
e of LiFePO,.. The electrochemical perfonnances (Figure 15.26) are
ced in this case as well, given appropriate thennal treatment of the
site electrode is done: at sufficient temperature to roduce conductive
, not too high though to avoid reduction of Fen in Fe .
For a carefully optimized LiFePOJC composite electrode, Morcrette et
wed the complete transfonnation of LiFePO,. into FePO,. by means of
109
X-ray diffraction. The diffractogram obtained at the end of charge of
lectrode is compared to that obtained from a non-optimized electrode in
15.27.
Optimised Standard
,.....
4.S
...t
;: 4 -·:
I
I
3.5 3.S
-
3 i!:.3
2.S
:
: ! 2.S
2
1.5
0
>
2
l.S
0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
x lnLI FePO
• • •
x Ia Li FePO
•
1>8 llruker, CDK.
15.27. X-ray diffraction patterns ofLiFePOIC pristine electrode and at the end of
or two different electrodes. 110
Padhi showed as well38 that lithium extraction from isostructural

sitions LiFel-A.MnxP04 was also possible, locating the Mn,.../Mn 3+ redox
at 4.1 V vs. Li+/Li. Since then, Yamada 111 "113 reinvestigated the Li 1_
nxPO.. phase diagram in details. Li did as well 11.._116 and reported good
ty (-140 mAhlg) for pure LiMnP04 117 coated with carbon black. These
d results appear however difficult to reproduce. Investigations by

18
Lloris119 and Okada120 on Li 1.xCoP04 located the Co4+Co3+ redox
t 4.8 V vs. Li, too high to be sustained by the electrolytes presently
e.
CONCLUSIONS
olyanionic structures built on transition metal M06 octahedra and X04

ra offer a wide variety of crystalline structures hosts for the reversible
/extraction of alkali cations. We tried through this contribution to
rate that these structures are progressively shifting from a conceptual
to possible candidates for new generations of positive electrode
s for lithim batteries.
2
s illustrated in Figure 15.28, V4+N 3+ and Fe3+/Fe + redox couples
en identified in various structural types and are the most interesting
high potential vs. Li+/Li is desired. Depending on the type of
ion chosen though, titanium-containing phosphates are also attractive
of stable potential delivered at 2.5 V vs. Li. This offers a very wide
f operating voltage for individual cells, tuned by the inductive effect
04 groups and by the crystal structure as well.
4 -·-·-·-·-·-·-·-·-·-·-·-·-· ............................................. ·-·-·-·-·-·-·-·-·-·-·-·-·-· ........................

Vl+fV'-+ Vl+fV'-+
1 A-
LI1Fe2(P04)1 B-LI F8z(P04)1 LIFeP04
UFeP10r
A-
LIVP10r B-L11V1(P04h FeP04.2HzO
LI1V2(P04h
28. Derivative ofpotentiodynamic curves obtained for several polyanionic structures.
he extensive work undertaken since 2000 on coating the olivine

with conductive carbon has led to the successfull use of this material
% of its theoretical capacity. As evocated by Chen though 99 efforts
Chapter 15
till have to be focused on increasing the tap density of these composite

des for their use in practical systems.
ERENCES
Guyomard, in "New Trends in Electrochemical Technology: Energy Storage Systems
Electronics", T. Osaka and M. Datta (Editors), Gordon & Breach Science Publishers,
apter 9 p. 253 (2000).
B. Nalbandyan, I. L. Shukaev, Russ. J. lnorg. Chern. 32 (1987) 3.
B. Goodenough, M. M. Thackeray, W. I. F. David, P. G. Bruce, Rev. Chim. Miner. 21
984) 435.
. M. Thackeray, W. I. F. David, J. B. Goodenough, Mater. Res. Bull. 17 (1983) 785.
M. Thackeray, W. I. F. David, P. G. Bruce, J. B. Goodenough, Mater. Res. Bull. 18
983) 461.
M. Thackeray, P. J. Johnson, L.A. de Picciotto, P. G. Bruce, J. B. Goodenough,
ter. Res. Bull. 19 (1984) 179.
Manthiram, J. B. Goodenough, J. Solid State Chern. 71 (1987) 349.
Manthiram, J. B. Goodenough, J. Power Sources 26 (1989) 403.
Delmas, F. Cherkaoui, A. Nadiri, P. Hagenmuller, Mat. Res. Bull. 22 (1987) 631.
Delmas, A. Nadiri, Mat. Res. Bull. 23 (1988) 65.
Collin, J.P. Boilot, in "Superionic Solids and Solid Elecrolytes ", Academic Press,
89) p 227 .
Y-P. Hong, Mat. Res. Bull. 11 (1976) 173; L. Hagman, P. Kierkegaard, Acta Chern.
and. 22 (1968) 1822.
Hagman, P. Kierkegaard, Acta Chern. Scand. 22 (1968) 1822.
G. Sizova, A. A. Vronkov, N. G. Shumyatskaga, V. V. Pynkhin, N. V. Belov, Dokl,
ad. Nauk SSSR, Ser. 205, Issue 1 (1972) 90.
Tillement, These de doctorat de l'Universite Paris VI, Mars 1992.
B. Goodenough, H. Y-P. Hong, J. A. Kafalas, Mat. Res. Bull. 11 (1976) 203.
Tillement, J. Angenault, J.C. Couturier, M. Quartan, Solid State Ionics 44 (1991) 299.
Tillement, J.C. Couturier, J. Angenault, M. Quartan, Solid State Ionics 48 (1991) 249.
Tillement, J. Angenault, J.C. Couturier, M. Quartan, Solid State lonics 53-56 (1992)
1.
Nadiri, C. Delmas, R. Salmon, P. Hagenmuller, Rev. Chim. Miner. 21 (1984) 537.
M. Reiff, J.H. Zhang, C.C. Torardi, J. Solid State Chern. 62 (1986) 231.
C. Torardi, E. Prince, Mat. Res. Bull. 21 (1986) 719.
G. Bruce, G. Miln, J. Solid State Chern. 89 (1990) 162.
d'Yvoire, M. Pintard Screpel, E. Bretey, M. de Ia Rochere, Solid State Ionics 9&10
83) 851.
Pintard-Screpel, F. d'Yvoire et Francis Remy, C. R. Acad. Sc. Paris 286 (1978) 381.
Delmas, R. Olazcuaga, F. Cherkaoui, R. Brochu, G. Le Flem, C. R. Acad. Sc. Paris
7 (1978) 169.
Masquelier, A. K. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Solid State
ern. 135 (1998) 228.
. Andersson, B. Kalska, P. Eyob, D. Aemout, L. Haggstrom, J.O. Thomas, Solid State
ics 140 (2001) 63.
Morcrette, C. Wurm, C. Masquelier, Solid State Sciences 4 (2002) 239.
K. Padhi, V. Manivannan, J. B. Goodenough, J. Electrochem. Soc. 145 (1998) 1518.
Rangan, J. Gopalakrishnan, Inorg. Chern. 34 (1995) 1969.
elmas, A. Nadiri, J. L. Soubeyroux, Solid State Ionics 28-30 (1988) 419.

Aatiq, M. Mem!trier, L. Croguennec, E. Suard, C. Delmas, GFECI, Orleans, March
2 ; J. Mater. Chern. 2002, in press
Masquelier, C. Wurm, J. Rodriguez-Carvajal, J. Gaubicher, L.F. Nazar, Chern. Mater.
2000) 525.
atoux, C. Masquelier, Chern. Mater. 14(5 (2002) 2334.
naidi, S. Launay, M. Quarton, Solid State Ionics 93 (1997) 273.
opalakrishnan, K. Rangan, Chern. Mater. 4 (1992)745.
K. Padhi, K. S. Nanjundaswamy, C. Masquelier, J.B. Goodenough, J. Electrochem.
144 (1997) 2581.
. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Electrochem. Soc. 144 (1997)
8.
yob, A.S. Andersson, J. 0. Thomas, J. Mater. Chern. 12, (20001)1.
. Gaubicher, C. Wurm, G. Goward, C. Masquelier, L. Nazar, Chern. Mater. 12 (2000)
0; (b) J. Gaubicher, F. Orsini, T. Le Mercier, S. Llorente, A. Villesuzanne, J.
enault, M. Quarton, J. Solid State Chern. 150 (2000) 250.
atoux, C. Masquelier, Chern. Mater. in press (2002)
. Cushing, J. B. Goodenough, J. Solid State Chern. 162 (2001) 176.
Masquelier, G. Rousse, S. Patoux, C. Wurm, M. Morcrette, submitted to Chern. Mater.
3).
Morcrette, J. B. Leriche, S. Patoux, C. Wurm, C. Masquelier, submitted to Elec. Solid
e Letters, 2002
. Nanjundaswamy, A. K. Padhi, J.B. Goodenough, S. Okada, H. Ohtsuka, H. Arai, J.
aki, Solid State Ionics 92 (1996) 1.
Christidis, P.J. Rentzeperis, Z.fiir Kristallographie 144 (1976) 341.
Christidis, P.J. Rentzeperis, Z. fiir Kristallographie 141 (1975) 233.
ran Qui, S. Hamdoune, Y. LePage, Acta Cryst. C43 (1987) 201.
ran Qui, S. Hamdoune, Acta Cryst. C43 (1987) 397.
. Maksimov, L.A. Muradyan, E. A. Genkina, V.I. Simonov, Sov. Phys. Dokl. 31(5)
5) 370.
. Bykov, A.P. Chirkin, L. N. Demyanets, S. N. Doronin, Solid State Ionics 38 (1990)
Rousse, J. Rodrfguez-Carvajal, C. Wurm, C. Masquelier, Chern. Mater. 13 (2001)

7.
. Winand, A. Rulmont, P. Tarte, J. Solid State Chern. 87 (1990) 83.
ato, H. Ohkawa, K. Yoshida, M. Saito, K. Uematsu, K. Toda, Solid State Ionics 135
0) 137.
toux, C. Wurm, M. Morcrette, G. Rousse, C. Masquelier, 1lth International Meeting
ithium Batteries, Monterey, CA (2002), accepted for publication into J. Power
ces.
1
kada, K. S. Nanjundaswamy, A. Manthiram, J. B. Goodenough, 36 h Power Sources
f, June 6-9 (1994).
arker, M.Y. Saidi, US Patent 6,203,946 (March 20, 2001).
. Saldi, J. Barker, H. Huang, J.L. Swoyer, G. Adamson, Electrochem. Solid State
5(7) (2002) A149.
azar, H. Huang, S.-C. Yin, T. Kerr, 11th International Meeting on Lithium Batteries,
terey, CA (2002).
aubicher, T. Le Mercier, Y. Chabre, J. Angenault, M. Quarton, J. Electrochem. Soc.
(12) (1999) 4375.
aubicher, F. Orsini, T. Le Mercier, S. Llorente, A. Villesuzanne, J. Angenault, M.
rton, J. Solid State Chern. 150 (2000) 250.
Yvoire, M. Pintard-Screpel, E. Bretey, Solid State lonics 18&19 (1986) 502.
Yvoire, E. Bretey, G. Collin, Solid State lonics 28-30 (1988) 1259.
Chapter 15
Masquelier, F. d'Yvoire, G. Collin, Journal of Solid State Chemistry 118 (1995) 33.
Gabelica-Robert, P. Tarte, Solid State Chern., Proceedings of the 2nd European Conf
9 June 1982, p.475.
Belkouch, L. Monceaux, E. Bordes, P. Courtine, Mat. Res. Bull. 30 (1995) 149.
. Wurrn, M. Morcrette, L. Dupont, G. Rousse, C. Masquelier, Chern. Mater. 14(7)
002) 2701.
Riou, N. Nguyen, R. Benloucif, B. Raveau, Mat. Res. Bull. 25 (1990) pp 1363.
. H. Lii, Y. P. Wang, Y. B. Chen, S. L. Wang, J. Solid State Chern. 86 (1990) 143.
Wang, C. Wu, S. Liu, J. Solid State Chern. 113 (I) (1994)37.
. Rousse, C. Wurm, J. Rodriguez Carvajal, M. Morcrette, C. Masquelier, Int. J. lnorg.
ater. 3 (2001) 881.
V. Pechkovskii, E. D. Dzyuba, G. I. Salonets, V. N. Yaglov, A. I. Volkov, Rus. J.
rg. Chern. 20(3) (1975) 329.
Sanz, J.E. Iglesias, J. Soria, E. R. Losilla, M.A. G. Aranda, S. Bruque, Chern. Mater. 9
997) 996.
Song, S. Yang, P.Y. Zavalij, M.S. Whittingham, Mater. Res. Bull. 37 (2002) 1249.
Masquelier, P. Reale, M. Morcrette, C. Wurm, L. Dupont, D. Larcher, J. Electrochem.
c. 149(8) (2002) Al037.
.S. Hong, K.S. Ryu, Y.J. Park, M.G. Kim, J.M. Lee, S.H. Chang, J. Mater. Chern. 12
002) 1870.
. Prosini, M. Lisi, S. Scaccia, M. Carewska, F. Cardellini, M. Pasquali, J.
ctrochem. Soc. 149(3) (2002) A297.
Gaubicher, Y. Chabre, J. Angenault, A. Lautie, M. Quarton, J. of Alloys and
mpounds 262-263 (1997) 34-38.
Gaubicher, These de Doctorat, Universite Pierre & Marie Curie, Paris (1998).
Dupre, J. Gaubicher, J. Angenault, G. Wallez, M. Quarton, J. Power Sources 97-98
01) 532.
Dupre, J. Gaubicher, T. Le Mercier, G. Wallez, J. Angenault, M. Quarton, Solid State
ics 140 (2001) 209.
A. Kerr, J. Gaubicher, L.F. Nazar, Electrochem. Solid State Lett. 3(10) (2000) 460.
V. Lavrov, V. P. Nikolaev, G. G. Sadikov, M. A. Porai-Koshits, Sov. Phys. Dokl.
(9) (1982) 680 0
H. Lii, C.H. Lii, C.Y. Cheng, S. L. Wang, J. Solid State Che., 95 (1991) 352.
M. Longo, P. Kierkegaard, Acta Chern. Scand. 20 (1966) 72.
G. Amos, A. Yokochi, A. W. Sleight, J. Solid State Chern. 14 (1998) 303.
G. Amos, A. W. Sleight, J. Solid State Chern. 160 (2001) 230.
Leclaire, H. Chahboun, D. Groult, B. Raveau, Z. for Kristallographie 177 (1986).
Chahboun, D. Groult, M. Hervieu, B. Raveau, J. Solid State Chern. 65 (1986) 331.
K. Padhi, K. S. Nanjundaswamy, C. Masquelier, S. Okada, J.B. Goodenough, J.
ctrochem. Soc. 144(5) (1997) 1609.
S. Andersson, B. Kalska, L. Haggstrom, J.O. Thomas, Solid State Jonics 130 (2000)
S. Andersson, J.O. Thomas, B. Kalska, L. Haggstrom, Electrochem. Solid State Lett.

), (2000) 66.
. Goodenough, A.K. Padhi, K.S. Nanjundaswamy, C. Masquelier, US Patent
910,382 (1999).
Ravet, J. B. Goodenough, S. Besner, M. Simoneau, P. Hovington, M. Armand,
stract# 127, 196'h Meeting of the Electrochemical Society, Hawai:, Oct. (1999).
Ravet, Y. Chouinard, J. F. Magnan, S. Besner, M. Gauthier, M. Armand, J. Power
urces 97-98 (2001) 503.
Huang, S.C. Yin, L. F. Nazar, Electrochem. Solid State Lett. 4(10) (2001), A170.
Dominko, M. Gaberscek, J. Drofenik, M. Bele, S. Pejovnik, Elec. Solid State Lett.
1) (2001) A187.
hen, J.R. Dahn, J. Electrochem. Soc. 149(9), (2002) All84.

Yamada, S.C. Chung, K. Hinokuna, J. Electrochem. Soc. 148 (3) (2001) A224.
Yang, P.Y. Zavalij, M.S. Whittingham, Electrochemistry Communications 3 (2001)
.
Yang, Y. Song, P.Y. Zavalij, M.S. Whittingham, Electrochemistry Communications
002) 239.
ranger, F. Le Cras, C. Bourbon, H. Rouault, Electrochem. Solid State Lett. 5(10)
2) A231.
. Prosini, D. Zane, M. Pasquali, ElectrochimicaActa 46 (2001) 3517.
Prosini, M. Carewska, S. Scaccia, P. Wisniewski, S. Passerini, M. Pasquali, J.
trochem. Soc. 149(7) (2002) A886.
roce, A.D. Epifanio, J. Hassoun, A. Deptula, T. Olczac, B. Scrosati, Electrochem.
d State Lett. 5(3) (2002) A47.
. Chung, J. T. Bloking, Y.M. Chiang, Nature materials 1 (2002) 123.
Wurm, These de doctorat, Universite Paris-XI Orsay, July 2002.
51
Morcrette, C. Wurm, J. Gaubicher, C. Masquelier, Proceedings of the 1 Lithium
ery Discussions, Arcachon (France), 27-30 May 2001 .
Morcrette, Y. Chabre, G. Vaughan, G. Amatucci, J-B. Leriche, S. Patoux, C.
quelier, J-M. Tarascon, Electrochimica Acta 47 (2002) 3137.
Yamada, Y. Kudo, K.Y. Liu, J. Electrochem. Soc. 148 (7) (2001) A747.
Yamada, S.C. Chung, J. Electrochem. Soc. 148 (8) (2001) A960.
Yamada, Y. Kudo, K. Y. Liu, J. Electrochem. Soc. 148 (10) (2001) All53.
i, H. Azuma, M. Tohda, Electrochemical and Solid State Letters 5(6) (2002) A135.
i, H. Azuma, M. Tohda, J. Electrochem. Soc. 149(6) (2002) A743.
i, Y. Kudo, K. Y. Liu, H. Azuma, M. Tohda, J. Electrochem. Soc. 149(11) (2002)
14.
eller, J. L. Durand, Acta Cryst. 13 (1960) 325.
mine, H. Yasuda, M. Yamachi, Electrochem. Solid State Lett. 3(4),(2000) 178.
. Lloris, C. Perez Vicente, J. L. Tirado, Electrochemical and Solid State Letters 5(10)
2) A234.
kada, S. Sawa, M. Egashira, J. Yamaki, M. Tabuchi, H. Kageyama, T. Konishii, A.
hino, J. Power Sources, 97-98 (2001) 430.
ter 16
ERSTANDING PHASE TRANSFORMATIONS

IN LITHIUM BATTERY MATERIALS BY
RANSMISSION ELECTRON MICROSCOPY
Y. Shao-Hom
Massachusetts Institute of Technology

Cambridge, Massachusetts 02/39
INTRODUCTION
Downsizing of electronic components demands energy storage

s with high energy and power densities, and thus continuously drives
earch and development efforts. The most commonly used batteries in
le computers and cellular phones are lithium rechargeable batteries'. In
atteries, chemical energy stored in the positive electrode is released and
ted into electrical energy during discharge through an intercalation
ion) process by which lithium ions are incorporated within the host
re of the positive electrode materials. During charge, the process is d
(de-intercalation) and electric energy is applied to remove lithium om
the positive electrode. Lithium transition metal oxides that host
lithium ions on the interstitial sites have been studied as positive de
materials and the intercalation and de-intercalation processes of ions
are accompanied by redox of transition metal ions. The energy
of lithium rechargeable batteries is dependent on the voltage upon
lithium ions are inserted and the number of interstitial sites that can
modate lithium ions in the host structure. Ideally, intercalation and de
lation of lithium ions and redox of transition metal ions, should leave
t structure intact. In practice, however, variation in lithium contents in
st structure leads to lattice expansion or contraction, migration of
on metal ions and local or global lattice distortion as a result of
g of lithium and vacancies or Jahn-Teller distortion of transition metal
hese structural changes can 1) induce stresses and strains within lithium
on metal oxide crystals, 2) affect the electronic conductivities of
battery materials, and 3) lead to irreversible phase changes and thus
dy of Battery Materials 479
e in energy outputs of lithium batteries. Therefore, much research has

cused on characterizing structural changes of lithium transition metal
associated with lithium intercalation and de-intercalation, and
ng material chemistry to suppress phase transformations and to
e lithium battery performance.
4 58 11
Layered 03 LixCoo/- , LixNi02 - and LixMnOt , spinel
12 14 15
04 - and olivine LixFeP04 16
are the most studied positive electrode
ls in the last two decades ' 17• Layered LixCo02 and LixNi02 are stable
thium intercalation or de-intercalation despite the fact that several
ansformations are observed from lithium composition x = 1 to x = 0.
ificance, lithium and vacancy ordering3, Jahn-Teller induced18, and
sulator 19 transformations have been reported. On the other hand,
LixMn02 host structure is not stable upon lithium intercalation or de
ation and it transforms a spinel-related structure20-22. Spinel LixMn204
e is stable upon electrochemical cycling but it can undergo 13 many
ng phase transformations induced by Jahn-Teller distortion or charge
23 25
g - . Extensive research efforts have focused on understanding the
al changes underwent in these lithium transition metal oxides and their
on lithium battery performance.
Conventional bulk analytical techniques, such as X-ray, neutron
diffraction and electrochemical characterization, are used routinely to
overall, averaged chemical and structural information of lithium
electrodes. Although the bulk diffraction techniques enable accurate
nation of distances and angles of atoms in the volume-averaged crystal
es, they are insensitive to spatially localized phase transformations and
transitions of lithium and vacancy or charges in lithium electrode
s. In particular, Garcia-Adeva et al. have recently pointed out the
ties in revealing nanoscale heterogeneity or aperiodic phase separation
26
entional powder diffraction techniques . Electrochemical methods,
lithium cell voltage monitoring, are very sensitive to changes of local
ments available for lithium ions at the electrode-electrolyte interface
y provide little information on the details of structural features
ble for the electrochemical responses. Therefore, bulk analytical
es alone are inadequate to identify critical material features and phase
mations in lithium electrode materials that limit battery performance.
ation of transmission electron microscopy (TEM) analyses, bulk
on and electrochemical techniques has proven invaluable in this
TEM and associated techniques enable microstructural
rization of lithium battery materials from both local structural and
ry points of view.
Although TEM techniques have been used27 extensively in materials
solid state chemistry and mineral science -30, application of these
es in lithium battery research has been comparatively recent 21·31-52.
Chapter 16
techniques provide high-spatial resolution single-crystal diffraction

ation from individual lithium transition metal oxide crystals
ementary to X-ray and neutron powder diffraction as electrons can be
d easily. In addition, one can use TEM imaging to detect the phase
ution in the electrode that provides valuable insights to heterogeneous
transformation mechanisms, and imperfections in the lithium transition
oxide crystals, which is introduced either during material synthesis or
chemical cycling. Moreover, high-resolution TEM imaging and image
tion can permit direct visualization of atomic arrangements and
ications associated with structural defects and modifications, as shown
erstanding stacking faults in orthorhombic LiMn02 48 and imaging of
9
m ions in LiCoOl . Furthermore, recent electron energy loss
oscopy (EELS) studies have provided evidence for charge compensation
and LixNi0.8Co0.z0z49, and transition metal
51
nisms of LixCo02
53
ing in LiNi 1_xAlx02 • Therefore, combination of bulk analytical studies
EM studies has led to significant advances in identifying phase
rmation mechanisms of lithium electrode materials, understanding of
re-property relationships, and providing directions to design of new
materials.
In this chapter, we focus on understanding phase transformations of
d 02layered LixCoOz, 03 LixNi02, 03 LixMnOz and spinel Lix[Mn]z04
went during electrochemical cycling or upon materials synthesis. Much
information is derived from electron diffraction data obtained in the
ission electron microscope (TEM). We begin with an overview of
only used TEM techniques. We then illustrate how TEM techniques can
ed to reveal and understand electrochemically induced phase
rmations in lithium electrode materials and understand their effects on
ctrochemical performance of lithium batteries. Electron diffraction has
ed first experimental evidence for order-disorder phenomena of lithium
acancy in 03 and 02 derived LixCo02 and 03 derived LixNi02• In
n, high spatial resolution of TEM techniques is powerful in identifying
ed phase transitions such as revealing the presence of tetragonal, Jahn
distorted Liz[Mn 2]04 in Lix[Mn2]04 electrodes discharged to 3.0 V and
which supports one of previously proposed hypotheses responsible for
ty degradation of 4 V Lix[Mn 2]04 electrodes during electrochemical
g. Moreover, electron diffraction has revealed and quantified
ediate cation configurations between the spinel and the layered
urations in as-prepared low-temperature LiCo02 (LT-LiCo02) and
layered 03 LixMn02 electrodes, which leads to electrochemical
ties of LT-LixCoOz and LixMn02 electrodes significantly different from d
03 LixCo02. A rapid, irreversible phase transformation from the d
to a spinel-related structure in cycled LixMn02 electrodes leads to
cant modification in the electrochemical voltage profile and reduction
sible capacity, which limits the commercialization of layered LixMn02

m batteries. In all these cases, TEM techniques identify critical phase
mations in lithium transition metal oxides that control the
hemical performance of lithium batteries.
TRANSMISSION ELECTRON MICROSCOPY

Electron diffraction and TEM imaging are effective in probing
d structural modifications in lithium electrode materials on the
ter-scale.
. Electron Diffraction of Crystals

Kinematical theory that assumes that electron beam is scattered
by the material (the amplitude of the scattered electron wave is much n
the incident electron wave) is used here. We begin with scattering of s
from atoms. The wave function If/ of the electron at non-relativistic es
satisfies the time independent Schrodinger equation, for stationary
h2 2
lf/ + (E + V)lf/ = (1)
V
8.1Z' moe
2 0
Where eE is the total energy of the electron and -eV is the potential
Within the kinematical approximation, the amplitude of the wave
d from an atom at a point P (at r from the origin) can be expressed by:
If/ s = 2mn0e exp(27dkr1) Jvc r,) )dr (2)
(
exp 2m"K'r
2 1 1
h r atom
(r 1 ) is the atomic potential at a point r 1, k is the wave vector and K' =

sing
f .,.-(8)the
= Born approximation,
'(.i:e we obtain the atomic scattering
r Z- fx) (3)
de for electrons:
(
x is the atomic scattering amplitude for X-rays and Z is the atomic
The interaction between the atom and electrons consists of scattering
oth nucleus and the electron cloud. This is different from X-ray
on, which interacts with atomic electrons and the atomic scattering r
X-rays can be expressed by:
Chapter 16
e2 (4)
fx-rays ((}) = --2fx
m0 c
s shown in Equations 3 and 4, the scattering power of atoms for

ns is therefore much greater than for X-rays. In addition, the atomic
ing factors for electrons increase smoothly with atomic number except
scattering angles but not as quickly as X-rays. Therefore, electron
tion is more sensitive in detecting light atomic number elements than
diffraction. It should be noted that the electron scattering factor of
m relative to transition metals is significantly larger than that of X-rays
esult of strong interactions between electrons and atoms, as shown in
16.1.. Therefore, electron diffraction can be used to probe the presence
erstructures induced by ordering of light elements or charges, which is
lly proven difficult for X-ray or neutron powder diffraction.
o.s -----;:::============::-!
--e- x-ray diffraction
- Electron Diffraction
" 0.4
::;;
:::;
..
0 0.3
N0 o.2
.".,
l!!
!
..
rl 0.1
...
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Sine/A
16.1. The ratios between the X-ray and electron scattering factors of lithium to
ese atoms lfLI!Mn) as a function of SinS//.,. Note that the ratio, fd!Mn, for electron
on is considerably larger than that of X-ray diffraction.
he scattered electron wave from one crystal can be described by:

exp(- 2m.K. ·
(5)
lf/ - exp(21likrf) LN f
(.e) r.)
S f I I 1
(1 ((})is the atomic electron scattering factor of site i in the crystal. The
te intensity of the scattered wave can be obtained by taking the product
wave function with its complex conjugate.
Electron diffraction patterns of a powder sample can be obtained by
d area electron diffraction from regions as small as 0.5 J..lm or by
fraction or convergent beam electron diffraction, in which focused

beam is used to obtain diffraction patterns from regions down to 1
gle-crystal electron diffraction data obtained from lithium transition
xide crystals can be viewed as planes in its reciprocal space, which
of the transmitted electron beam and individual diffraction spots.
d procedures on electron diffraction pattern indexing can be found in
29
M textbook by Fultz and Howe . Because of strong dynamic
on effects in electron diffraction of most crystals, electron diffraction
es are often used qualitatively, and quantitative analysis of electron
on intensities is less reliable than X-ray diffraction and neutron
on.
. TEM Imaging
Diffraction contrast in the TEM images results from the variation in
racted intensity of the material. Bright-field images (the transmitted
tensity) and dark-field images (the diffracted beam intensity) provide
tion about microstructure features of the materials. Imperfections such
cations, stacking faults, domain and phase boundaries within the
displace the unit cell and thus introduce a phase change in the scattered
which leads to intensity variation in the dark-field and bright field
Dark-field imaging has been particularly useful to show and study
structures within lithium transition metal oxide crystals induced by
intercalation or de-intercalation.
In addition to diffraction contrast, imaging of lattice fringes has been
nly used in studying phase transformations in lithium battery materials
nanometer scale. The intensity distribution of the incident beam and
racted beam gives a modulation with the spacing of the plane being
ed for a given crystal orientation and thickness, which reveals the
f the lithium electrode material in form of sine fringes as shown as
2 2 (6)
I= 1+ R - :-
2Rsin( S0 )
d is the spacing of the plane being diffracted, x is the direction of

ion in the image, and R and S 0 are constants for a given thickness and
orientation). It should be noted that the positions of the lattice fringes
apparent relation to the position of atomic planes. Further detailed
tion on lattice image formation can be found in the TEM
27 30
ks ' •
Chapter 16
3. TEM Sample Preparation
The requirements of a good TEM sample include 1) electron beam

arency, which limits maximum thickness to 5 J..tm, and 2) clean surface
of contaminants). Lithium transition metal oxides used in lithium
es are in form of powder and the crystal sizes are dependent on
sis routes, typically in the range of 20 nm to 10 J.m. TEM samples of
oxides can be obtained from the following preparation procedure: 1)
the powder in agate mortar and pestle, 2) suspend the powder in an c
solvent in an ultrasonic bath, 3) deposit suspended particles onto a
copper grid coated with lacey carbon. If crystal sizes of lithium
ode materials are significantly greater than 1 IJ.m, mechanically thinning
se crystals is needed. Microtoming is an effective method to produce
EM sections of as-prepared lithium transition metal oxides. The material
embedded in epoxy resin. Ultrathin sections on the order of 60 nm cut
iamond knives using an ultramicrotome are floated on water and then
rred to TEM copper grids with lacey carbon support.
4. Caveats to TEM Techniques
Although electron diffraction and TEM imaging techniques are

ful in probing localized structural defects and phase transitions, they
few limitations in comparison to bulk analytic techniques. Three of the
mportant are 1) the sampling statistics is very poor, 2) only crystals with
ess less than 1J.,tm can be analyzed and 3) kinematical theory is only
tive. Therefore, in order to ensure the fact that crystals analyzed in the
are representative of the bulk materials, it is necessary to combine TEM
with bulk analytical information including X-ray powder diffraction
ectrochemical data. In addition, there are concerns with regard to the
ty of materials during TEM sample preparation and under electron
This is a serious issue when studying highly de-intercalated lithium
materials in the TEM. Performing TEM data collection at liquid
en temperature can remove some materials instability problems related
mal breakdown under electron beam.
LITHIUM AND VACANCY ORDERING
dering of lithium and vacancy has been first proposed in intercalation

unds of layered chalcogenides such as LixTiS254 and LixTaSt). At room
ature, lithium ions that reside in octahedral sites are mobile. Lithium
re constrained to occupy certain sites preferentially as a result of long
repulsive, Coulomb interactions among lithium ions, which leads to
on of superlattices at various lithium contents parallel to the transition

ayers. Although direction observation of lithium ordering by X-ray
on is difficult, characteristic peaks in the incremental capacity dx/dV
ots are observed at specific lithium compositions where ordering is
d in LixTiS254 • As a similar electrochemical feature is noted in
o02 cells where x is near 0.5, Reimers and Dahn3 have proposed
and vacancy ordering in Li0.5Co02 obtained from lithium de
ation of layered 03 LiCo02. First direct experimental evidence of
ordering is the presence of superstructures in the electron diffraction
of layered LixNi02 42' 56• Subsequently electron diffraction and TEM
have successfully revealed order-disorder phenomena of lithium and
37 8
in layered 03-Li05Co02 and 02-LixCo0/ electrode materials.
er, first-principles studies support the concept of lithium and vacancy
57
in these layered compounds - 64• Van der Ven and Ceder have shown
ium ordering reduces lithium ion mobility as evidenced by a two-order
de of reduction in the chemical diffusion coefficient in LixCo02 at
57
compositions • Lithium and vacancy ordering can be suppressed by
65 66
ubstitution in the transition metal layers ' , which eliminates phase
mations and greatly affects electrochemical cycling of lithium
. Therefore, understanding order-disorder phenomena of lithium and
in lithium battery materials is not only essential to further improve
ing performance of lithium rechargeable batteries but also important to
new insights to lithium-lithium interactions in these transition metal
its impact on their kinetic properties and design of better lithium
e materials. In this section, we focus on understanding order-disorder
mations of lithium and vacancy in these layered materials by electron
on.
Lithium and Vacancy Ordering in 03 LixNi02
03 layered LiNi02 consists of lithium and nickel sheets stacked

vely in-between AB CA BC AB close-packed oxygen arrays. The
e is typically defined in a hexagonal cell that has a trigonal symmetry
ce group R 3 m. Lithium de-intercalation from LiNi02 leads to many
ansformations that occur as a function of lithium concentration67- 69•
diffraction of LixNi02 has lent to first experimental evidence of
and vacancy ordering in lithium transition metal oxides at various
concentrations, x = 0.25, 0.33 and 0.65, as shown in Figure 16.2.42•56 •
be noted that at lithium compositions of x = 0.25 and x = 0.33, order
of lithium and vacancy is associated with noticeable inflexion points
voltage profile of LifLixNi02 cells, similar to lithium ordering in
. However, lithium ordering in LixNi02 near x = 0.65 is different and
4
ounced modification in the voltage profile is associated with this

Chapter 16
ng transformation. X-ray powder diffraction studies42'67.69 of LixNi02

that this composition falls in the wide monoclinic, single-phase region
he lithium composition the range of 0.5 :$ x :$0.75. The stability of the
linic phase, important to electrochemical performance, has been a
t of debate.
;] •
r-:-·--·::-::
.
c::.a·:--i
I ··•• ••••• I
1
INtO.:.
...
:t .t · -········..-·········...,--...··-,-·-·...- -,.-· ·-·-,X
0.0 O.l 0.. O.J O.J t .t
16.2. The voltage profile of Li,Ni0 2 vs. lithium as a function of lithium content x from
et al.56 . Note that the electron diffraction patterns obtained from various points in the
profile show different superstructures, which is not visible in the X-ray powder
on patterns of Li,Ni02 samples.
It has been proposed that this symmetry reduction to monoclinic is

and stabilized by either a cooperative Jahn-Teller induced distortion of
68 42
octahedra or ordering of lithium and vacancy • Using selected area
n diffraction, Peres et al. have provided experimental evidence of a
superstructure of ordered lithium and vacancy in the monoclinic phase
etectable in the X-ray powder diffraction patterns) in a Li0.63Ni1.0202
e. This superstructural model can be used to accommodate a range of
concentrations, in which lithium ordering stabilizes the monoclinic
2 64
. Recent first-principle studies of layered LixNi02 have shown that
and vacancy ordering is coupled with monoclinicity and weak Jahn
distortion of Ne+ ions in this phase region. In contrast to the 2x2x2
42
tructure of the monoclinic cell found by electron diffraction studies ,
63
nt superstructures have been predicted from first-principles studies :
perstructure with 2xlxl of the monoclinic cell at Li0.75Ni02 and the
with .J3 x .J7 x1 of the parent hexagonal cell at Li0.4Ni02. In addition to
tructures in the monoclinic phase region, one 2..fJ x2.J3xl
63
tructure of the hexagonal cell is found at Li0.5Ni02 via computation
experimental evidence is found for this composition at this time.
y of Battery Materials 487
re, discrepancy exists in the detailed nature of the superstructures of

lithium and vacancy in the monoclinic LixNi02 between first es
and electron diffraction results and further investigations are
. Lithium and Vacancy Ordering in 03 LixCo02

03 layered LiCo02, isostructural to 03 LiNi02, has become the most
nly used positive electrode material in lithium rechargeable batteries
e concept of reversible lithium removal from and insertion into
2
was first introduced by Goodenough et al. . Intensive research and
ment efforts have been devoted to understanding the physical
es and phase transformations of LixCo02 (0x1) in order to
3 4 19 44 59 70 76
the energy output and cycle life of lithium batteries • • • • • · •
de-intercalation from LiCo02 lead to several phase transformations as
as follows: 1) 0.75x0.94, a first-order transformation from a
hedral
2 insulating Li0.94Co02 phase to a rhombohedral metallic
0 phase'3 '4 19' 73; 2) x ::::: 0.5, a phase transformation from the
4
hedral to monoclinic symmetrl· , 3) 0.17 < x < 0.33, a phase
mation from 03 rhombohedral to H1-377 '78 and 4) 0 < x < 0.12, a
ansformation from H1-3 to 0175' 77'78 • Of particular interest has been to
nd the phase transformation from the rhombohedral to monoclinic
ry in LixCo02 at x::::: 0.5, which has been attributed to the ordering of
ions and vacancies within the layers (in-plane ordering) by X-ray
3 58 78
diffraction experiments and by first-principle studies • , as shown in
Ordering
6.3.. • • • •
•
or: rb ·o o
4.4 E (V) •
o
•ami'. • • • •
0 0 0 0
:::i 4.2 ./. • • • • • o Vacancy

ellthlum
+-
:::i
. 4.0
. &s
3.8
3.6
0.4 0.5 0.6 o.7 0.8 0.9 1.0

x in u.coo2
.3. Typical voltage profiles of Li/LixCo0 2 cells and the inflexion point is correlated
-disorder of lithium and vacancy in Li,Co02 (x ""0.5)3 •37 .
Chapter 16
37
Experimental electron diffraction analysis of LixCo0 2 with x z 0.5
ectron diffraction simulation of disordered Li0.5Co02 with space group
and ordered Li0.5Co02 with space group P2/m have confirmed the
ed in-plane lithium ordering. The in-plane ordering long the [110]hex.
uivalent directions allows the formation of mazed crystals and electron
tion patterns with appearance that resembles those of the spinel
re can be fully explained by lithium and vacancy ordering and
37
tence of two or three variants, as shown in Figure 16.4. • Therefore, it
37
eved that electron diffraction data do not suggest the layered to spinel
rmation in the LixCo02, and the stability of the layered LixCo02 upon
removal is in agreement with the excellent cycling performance of
Co02 batteries. In addition to in-plane lithium ordering, evidence of
ng of lithium and vacancy arrangements perpendicular to the c axis of
xagonal parent cell is also found, which is consistent with first
les predication 78 •
We now compare and discuss the difference in lithium ordering
ements between LixNi02 and LixCo02 systems. The number of ordered
arrangements found in LixNi02 obtained from electron diffraction and
rinciples is significantly more than LixCo02. Lithium ordering occurs at
02 with x = 0.5 and x = 113 where lithium ordering in LixNi02 has been
ed at x = 0.25, 0.4, 113, 0.5, 0.6 and 0.75 ' 63• In theory, in-plane
42
ng of lithium and vacancy should lead to deviation in the alb ratio of the
ohedral layered structure as the electrostatic interactions of lithium ions
the a and b directions are different. The deviation in the alb ratio from
al value (1.732) decreases the crystal symmetry from rhombohedral to
linic, where the crystal structure can be modified by varying the alb
nd/or ·It is interesting to note that the structural changes associated
ithium ordering at the Li0.5Co02 composition and in the LixNi02
; x:::; 0.75) system are different. Changes in the alb ratio would
erably modify the cobalt-oxygen bonds and distort the Co06 octahedra,
is energetically unfavorable as both Co3 + and Co4+ ions prefer to
ideal or nearly ideal octahedra. Therefore, the Li0.5Co02 layered
re adopts a negligible change in the alb ratio (1.733) but a significantly
nt (107.89°) value from the rhombohedral symmetry upon lithium
cancy ordering. In contrast, the LixNi02 (0.50:::; x:::; 0.75) structure has
l change in (109.75°) but a large increase in the alb ratio (1.764) as
3
eller distorted Ni + ions can accommodate the distortion in the a and b
on upon lithium and vacancy ordering.
6.4. Left, three possible orientations for lithium and vacancy ordering in the lithium
d the monoclinic unit cell of Li05Co02 and right, an experimental diffraction pattern
[103lmon. zone axis of the monoclinic unit cell or the [OOllhex. zone axis of the
l unit cell, which revealed the reflections associated with lithium and vacancy
37
rom all three variants •
Lithium Ordering in 02 derived LixCo02
Layered 02 LiCo0 19 2
with a unique AB CB AB oxygen packing, first
by Delmas et al. , is metastable to layered 03 LiCo02 with 03-type
BC oxygen packing. Recently 02 LiM02 (M = Ni, Mn, Co, etc.)
nds have been reinvestigated for lithium rechargeable battery
ions80 -89 as their structural derivatives upon lithium de-intercalation
ercalation are stable with respect to the spinel-related framework.
removal from 02-LiCo02 leads to the formation of various novel
as shown by X-ray powder diffraction, T#2, T#2' and 06, which are
82 86 90
related to the 02 crystal structure ' ' , as shown in Figure 16.5.. Of
ar interest is the T#2-LixCo02 phase that is stable over a large
itional range (0.52 < x 0.72) and the stability of this structure is not
understood. The r2 structure has an orthorhombic unit cell and
roup Cmca, for instance, 'Lio.61Co0z' with aorth. = 2.8097 A,
corth. = 9.9082 82 . . 87
0
4.8500 A, and 0 . Previ Ous stud1 es have reported

A
ium ions are located on the 8e type tetrahedral sites in the T#2
e, where two 8e sites are available per cobalt ions and fewer than half
e sites are occupied by lithium ions. As simultaneous occupancy of
8e sites is unlikely due to small site-to-site distances (1.4 A and 2.4
ium ordering on the 8e sites is expected in the T#2 structure.
eless, it should be noted that the voltage-capacity profile for the entire
positional range is relatively smooth (Figure 16.5.). The absence of
Chapter 16
on points in the voltage profile is significantly different from lithium

cancy ordering at specific lithium compositions in 03-LixNi02 and 03-
02 systems (such as x = 0.25 and 0.33 in Figure 16.2. and x = 0.5 in
16.3.), where pronounced modification in the lithium intercalation
e is noted.
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

x In LI Co02
16.5. Typical galvanostatic charge and discharge voltage profiles of Li/02-Li.Co0 2

d schematics of the 02-LiCo02 structure with both lithium and cobalt ions on the
ral sites, the T'2-Li.Co02 structure with cobalt ions on the octahedral sites and lithium
8e tetrahedral sites and the 06-Li.Co02 structure with both lithium and cobalt ions on
38
hedral sites .
Although no superlattice reflection is found in X-ray and neutron

r diffraction patterns of the T#2 phase, electron diffraction has revealed
ce of several commensurate and incommensurate superstructures in the
38
hase • A 2aorth.x 2borth.x 2corth.' a 2aorth.x borth.x earth. and an
mensurate with q = y a;rth. (y = 0.23 and 0.36) superstructures have
dentified, which can be explained by different lithium and vacancy
ng configurations. For instance, superreflections present half way
n the diffraction center and the fundamental reflections in the [512 lorth.
e [02 1 lorth. diffraction patterns collected from a "Li0.7Co02' sample
the presence of the 2 aorth. x 2 borth. x 2 corth. superstructure and one
ordering scheme is proposed to explain this superstructure, as shown
re 16.638 •
Be Variant II
Ntl
16.6. The [51 2 lorth. and the [02 T lorth. diffraction patterns collected from the
02' sample reveal the presence of the 2 aorth. x 2 borth. x 2 corth. superstructure, in
ly fundamental reflections are indexed. Lithium and vacancy ordering configurations
o describe the doubling of the aorth. and borth. cell parameters are presented. Note that
ants are allowed as two possible orientations exist for the 8e Li04 tetrahedra. The
of the corth. cell parameter results from the alternation of these variants in the
38
space along the corth. axis .
In addition to commensurate superstructures, incommensurate

cities, q = 0.36" a;rth. and 0.23" a;rth., are found in the [0 T lJorth. and
h. electron diffraction patterns collected from the 'Li0.63Co02' samples,
38
vely, as shown in Figure 16.7. • The presence of these
ensurate superstructures suggests that 1) incommensurate lithium and
y ordering is preferred to commensurate ordering of lithium and
y in large superstructures such as 3 aorth. x borth. x c orth. and
borth. x c orth. and 2) the lithium-lithium interaction can be fairly long
reater than 10 A, in the T#2 structure. In addition, it is of significance
t out that combination of transmission electron microscopy and
diffraction showed that two or three T#2 variants could coexist in
t regions within one T#2 crystal, which led to the development of a
(mazed) microstructure similar to what has been observed in
02 (Figure 16.4.).
Chapter 16
16.7. The [0 ll]onh. and [OOl]onh. electron diffraction patterns collected from the
o02 ' samples, where the first-order, incommensurate reflections are marked by
8
. Two variants of the incommensurately ordered structure are highlighted in the
. electron diffraction pattern.
The lack of substantial peaks of ordering transitions within the single

Tlt2-LixCo02 domain in the differential capacity versus voltage curve
be attributed to the presence of a large number of commensurate and
mensurate superstructures with similar Gibbs free energies of different
38
cal compositions • The presence of many ordered phases in the Tlt2
2
is supported by a recent first-principles studl • It is believed that the
ce of various lithium and vacancy ordering configurations improves the
ty of the Tlt2 structure over a range of lithium compositions.
heless, it should be mentioned that the nature of superstructures present
ossible lithium ordering configurations in Tlt2-LixCo02 were complex
rther investigations are needed.
Many fundamental questions with regard to lithium and vacancy
ng in layered intercalation compounds remain to be answered: 1) what
strain and the stress concentrations in the crystals induced by lithium
cancy order-disorder transitions and formation of domains, and 2) how
ering transformations directly affect the cycling of lithium batteries.
SPATIALLY LOCALIZED PHASE TRANSITIONS
As electrons can be focused much easier than X-rays, combination of

maging and electron diffraction can provide structural information from s
as small as the probe size of electron beam, which can be focused to
5 nm. As electrochemical reactions occur at the interface between the
Battery Materials 493
aterials and the electrolyte, the performance of lithium batteries is

ve to any phase transformations on the surfaces of electrode
which the secondary phase often cannot be detected by X-ray or
der diffraction due to a low volume fraction in the electrode. TEM
ave shown particularly powerful in detecting new phases formed in
calation compounds during electrochemical cycling as reported by
91
t al.34 ' 40 and Dupont et al. .
n-Teller Induced Phase Transition in Lix[Mn2]04
ix[Mn2]04 batteries cycle poorly between 3 and 4 V plateaus and

ling performance results from destruction of the spinel framework
tetragonal phase transition induced by cooperative Jahn-Teller
13
3 V vs. lithium • Gradual loss in capacity is still observed if
04 batteries are cycled within the 4 V plateau (the cubic phase
It has been proposed that the capacity loss of 4 V Li1Lix[Mn2]04
m temperature is at least partially attributed to the formation of
h[Mn2]04 transformed from cubic Li[Mn2]04 at the end of
However, the tetragonal Li2[Mn2]04 phase is not detected by X-ray
action of the cycled Lix[Mn 2]04 electrodes in the 4 V range. If the
nal transition is localized, especially on the surface of crystals, it
rably affect the reversibility of the lithium intercalation process
the volume fraction of the tetragonal phase is insignificant in the
electrode. Using high-spatial resolution TEM techniques, the
localized tetragonal domains at the surface of cubic Lix[Mn2]04
34
ls in cycled electrodes .4° is confirmed.
ransmission electron microscope image of a Lix[Mn 2]04 crystal, extracted from a

cell discharged to 3.3 V vs. lithium, showing a phase boundary near the surface of
ed by arrows. The electron diffraction patterns collected from the surface and the
ystal using a 20 nm electron beam probe are shown on the right40 .
Chapter 16
TEM data of a spinel Lix[Mn2]04 crystal, extracted from a

[Mn2]04 cell discharged to 3.3 V vs. lithium, are presented in Figure
The dark fringes in the image marked by arrows are associated with the
and tetragonal phase boundary. The phase change across the boundary is
med by electron diffraction analysis of two regions. The diffraction
obtained from the crystal center (right bottom) has a four-fold
etry, corresponding to the cubic phase. In contrast, the pattern from the
l surface (right top) exhibits a two-fold symmetry, corresponding to the
nal phase. In addition, the absence of { 220} reflections and the non
lence of the (040) and (004) reflections are consistent with tetragonal
alt Liz[Mn2]04• Therefore, TEM analysis has provided direct
mental evidence for localized, cubic to tetragonal transformation
3
d by the Jahn-Teller distortion of Mn + during cycling of 4V
[Mn2]04 batteries. Substitution of monovalent, divalent and trivalent
n Li[MxMn2.x]04 can increase the average valence state of Mn ions,
3
ss the cooperative Jahn-Teller distortion of Mn + ions and thus the
tetragonal transition, and eventually improve the cycling performance
Li[MxMn2.x]04 spinel electrodes92. Although TEM studies have
ed the formation of tetragonal Liz[Mn2]04, further research is still
to correlate this phase transformation to capacity fading of
[Mn2]04 cells during electrochemical cycling.
Phase Transformations Between 2D Layered And 3D

Spinel-Related Configurations
The layered structure and the lithiated-spinel structures have close

oxygen arrays (which, in practice, usually deviate slightly from ideal
close-packing) but differ in the arrangement of the transition metal and
ions. The layered LiM02 structure has one set of unique close-packed
, i.e., the (001) planes, which is parallel to one of four equivalent close
planes of the lithiated-spinel Liz[M2]04 structure, i.e., the (111),
- --
), (111) and ( 1 11) planes. In the lithiated-spinel Liz[M2]04
re, the transition meal ions are distributed between the alternate close
oxygen layers in a 3:1 ratio. Therefore, if one-quarter of the transition
ions in the lithiated-spinel structure is transposed with one-quarter of
hium ions, the lithiated-spinel structure is then transformed to the
d structure. An intermediate cation distribution occurs if these
on metal ions are partially transposed with the lithium ions. X-ray and
n powder diffraction patterns of layered LiM02 and lithiated-spinel
]04 structures (M = transition metal cation) often share strong
lance, particularly when the cia ratio approaches 4.899, which leads to
ity in identifying whether the cation configuration in lithium electrode

ls is 2D layered or 3D spinel-related. In the case of layered 03-
with a cia ratio of 4.99, it is reasonably easy to distinguish the
structure. It is especially difficult to distinguish structures with cation
tions intermediate between those of the ideal layered and ideal spinel
rations by X-ray powder diffraction. Single-crystal electron diffraction
s of individual crystals provide phase relationships among different
planes, which can determine whether the structure is layered or spinel
In addition, the electron scattering intensity can be used to quantify
diate cation distributions in lithium transition metal oxides.
The lithiated-spinel structure can be converted to the layered structure
our transition metal ions on the 16d sites with coordinates ( 1/.i, Y2,
1
/.i),
), (%, Y2, %) and(%, 0, 1/.i), are transposed with the four lithium
atoms
16c sites with coordinates (1/.i, 0, 1/.i), (1/.i, Y2, %), (%, 0, %) and (%, Y2,
n intermediate structure is produced when a fraction of these transition
nd lithium ions are transposed and such an intermediate structure can
rmined quantitatively by examining and comparing the intensities of
ts of the { 111} planes in the lithiated-spinel structure and the (003)
n the layered structure.
he structure factor Fr 0031 of the (003) atomic planes in the layered
e with space group symmetry R 3m and with a cubic close-packed
array and the oxygen ions located at (0, 0, 0.24) have the following
ships:
F(003) = 3{JM - fLi )+ 1.124 fo (7)
fM is the scattering factor of a transition metal ion, fo is the scattering

f an oxygen and fu is the scattering factor of a lithium ion. When
ring an equivalent number of atoms in the layered structure to the
of atoms in one unit cell of the lithiated-spinel structure and ignoring
tering of oxygen (1.124!0 is significantly smaller than 3fM), then
F/ooJ) = 16 (JM - fLi) (8)
total scattering intensity of the (003) planes in the layered structure is
2
I(oo3; oc 2(F(003! ) = 512 (JM- fu )2 (9)
In the case of an ideal lithiated-spinel Li 16[M16]032 structure with
roup Fd 3 m, the structure factor F{llll and the total scattering
y /{lu) for the {111} atomic planes have the following relationships
ttering of oxygen is neglected):
Frw; = s(!M - fu) (10)
(11)
Chapter 16
equations indicate that the total scattering intensity from the (003)
ions in the layered structure equals that from the { 111} reflections of
hiated-spinel structure.
Now we consider intermediate cation distributions that deviate from
eal lithiated-spinel and layered configurations, which can be described
following formula: { (Lit6-4x)J6c[Li4x] t6cthayerl{ (Mt6-4x)J6ct[M4xh6c }Jayer2032
nel notation. When x = 0, the structure has the ideal lithiated-spinel
uration, (Lit6)t6c[Mt6h6ct032, or (Liz)t6c[M2h6ct04. When x = 1, it has the
d configuration, { (Lit2)t6c[Li4h6d}tayert { (Mdt6d[M4h6c}tayer2032
atively, in the layered notation, (Li3)33[M3hb06). The magnitude of the
re factor and the scattering intensity for the set of the { 111} planes
ncreasing transition metal density, e.g., the (1 T 1) reflection can be
bed as a function of x:
F(lll) = (8fM +8xfM -8fu -8xfu) (12)
agnitudes of the structure factors and the scattering intensities for each
other three sets of the { 111} planes with decreasing transition metal
ies, can also be described as a function of x:
F ( lll) = (8 f M - 8 xf M - 8 f Li + 8 xf Li ) (13)
tio between the scattering intensities of the (111) and (1 T 1) planes,

d as m, can be related to x by the following equation:
2
1(111) (1- x)
m = = -'------'-""7"" (14)
(1 + x)2
wn in Figure 16.10., m decreases as x increases, i.e., where the specific

ancies of lithium and transition metal ions in the structure vary from the
d-spinel to the lavered confilmration.
thiated-spinel o.•
o.a
0.7
o.e
0.5
o.•
0.3
02
0.1
o.o 0. 1 0.2 o.3 o.• o.s o.e o.7 o e o.e 1 .0

Lhhl•tod-Splnel L•yered
6.9. The scattering intensity ratio of the (111) and (1 T 1) planes, m, decreases as the
istribution varies from the ideal lithiated-spinel to the ideal layered configuration31 •
One can determine the values of x, the degree of cation distribution

n the layered and the lithiated-spinel configurations, by measuring the
intensity ratios of the { 111} reflections. The caveat to this approach is
dynamical diffraction conditions often encountered in transmission
microscopy studies would tend to reduce the intensity differences
the { 111} reflections. Taken together with the experimental fact that
ations in the { 111} scattering intensity are observed, it is apparent that
fraction conditions of thin crystals with slight deviation from
ct zone axis are consistent with the kinematical approximation. This
h has proven particularly powerful
31 35 in analyzing the structural features
LiCo02 upon heat-treatments ' and the structural stability of layered
21
Mn02 during electrochemical cycling .
Intermediate Layered-Spinel Cation Configurations in

LT-LiCo02
LT-LiCo02 synthesized at 400 oc has been studied extensively as a

electrode material for lithium batteries 95. X-ray powder diffraction
d vibration spectroscopy studies have determined that LT-LiCo02 has
6 97
ated-spinel Liz[Co2]04 structure (space group Fd3m/ • • LT-LiCo02
es do not perform well in lithium cells, and there is considerable
sis between the lithium de-intercalation and intercalation processes,
ontrasts with the superior electrochemical performance of electrodes
yered 03 LiCo02 structure (space group R 3m) • In addition, the
95
profile of LiJLT-LixCo02 cells is considerably different from that of

Mn2]04 spinel cells, which calls into question the ideallithiated-spinel
configuration of LT-LiCo02. TEM analyses of LT-LiCo02 crystals
own that the cobalt distribution in LT-LiCo02 samples is not ideal,
t it can be considered to be intermediate between an ideal layered
o hb02 configuration and an ideal lithiated-spinel (Liz)16c[Co2h6ct04
31
ration .
Crystal morphology of LT-LiCo02 crystals and two single-crystal
diffraction patterns collected from two different LT-LiCo02 crystal
wn in Figure 16.10.. TEM imaging reveals that the average crystal
LT-LiCo02 is on the order of 20 nm, which supports that the
ical approximation can be used in electron diffraction intensity
. Both diffraction patterns are indexed according to the lithiated-
tructure with cubic symmetry (Fd3m) along the [10 1] direction; they
be indexed consistently with trigonal symmetry (R 3m). The (111)
1) reciprocal points in one pattern (top right of Figure 16.10.) have
scattering intensities with an intensity ratio of 0.95, as expected for an
Chapter 16
thiated-spinel structure. In contrast, the (111) and (Ill) reflections in

her pattern (bottom left of Figure 16.10.) have significantly different
ing intensities with an intensity ratio of 0.25. The scattering intensity
on observed in the { 111} planes, which can be quantified digitally from
ctron diffraction patterns, is then used to determine x, the parameter to
be non-ideal cation distribution intermediate between the layered and
ed-spinel structures based on Equations 12-14 and Figure 16.9.. There
nificantly more crystals that have a configuration closer to the lithiated
structure (x < 0.5) than to the layered structure (x > 0.5) in LT-LiCo02,
wn in bottom right of Figure 16.10.. These electron diffraction results
ly confirm that the lithiated-spinel structure is the major phase in LT
2 products but also provide an explanation for the previously reported,
electrochemical properties of LT-LiCo02 in lithium batteries. The
of cobalt and lithium ions in layered or lithiated-spinel LiCo02 res
will restrict the two-dimensional or three-dimensional pathways for
diffusion that are normally accessible in the ideal layered and ideal
structures, respectively. Therefore, the poor electrochemical behavior
LT-LiCo02 cells can be attributed to the intermediate cation distribution
n the layered and the lithiated-spinel structures in LT-LiCo02•
' U 03 Ol U
Uhlllled-SpFui'
00 0.7 Oj DO I
la,...-.:
16.10. A high-resolution TEM image (top left) of LT-LiCo02 crystals showing the
crystal size on the order of 20 nm, and two single-crystal electron diffraction patterns
similar scattering intensity ratios of the (111) and (Ill) planes 31• The number of
analyzed in the LT-LiCo02 sample is plotted as a function of x (bottom right).
Layered to Spinel Transformation in Cycled LixMn02
Extensive research efforts have been focused on developing layered

Mn02, isostructural to layered 03 LiCo02 in order to reduce the cost
prove the safety aspects of positive electrode materials in lithium
11
s9• • The voltage profile of LiiLixMn02 cells is modified considerably
rst intercalation process and subsequent voltage profiles of cycled
Mn02 cells resemble those of the layered and the spinel compounds21 • It
ved that some Mn ions in the Mn02 slabs migrate into the lithium
uring lithium de-intercalation, resulting in localized, spinel-type cation
ments. However, similarities between the X-ray powder diffraction
of the lithium-spinel Li2 [Mn2]04 and layered LixMnOz phases lead to
ities in understanding phase transformations underwent in LixMn02
es, which prevents confirmation of the spinel-type configuration with
y. Combination of high-resolution lattice imaging and electron
21
ion has been able to reveal the formation of localized, spinel domains
ycled LixMnOz.
A typical high resolution TEM image and the corresponding electron
ion pattern of one as-prepared LiMn02 crystal and a randomly
crystal in the Li0.5Mn02 electrode obtained after one charge and one
discharge are compared in Figure 16.12.. It should be pointed out that
s of lattice fringes and the corresponding reflections (right of Figure
indicate the presence of two sets of close-packed planes in the
02 structure in contrast to one allowed in the layered LiMn02 structure
Figure 16.11.). The occurrence of more than one set of close-packed
nes corresponds to the presence of the spinel-type configuration. As
anganese ions have migrated into lithium planes during lithium de
ation and they are trapped in the lithium layers upon lithium re
ation, therefore, localized cation distributions resemble that of the
In addition, it can be argued that the non-equivalent scattering
y of the { 111} reflections of the spinel-type structure in the LixMn02
can be attributed to a non-uniform cation distribution within
ual LixMn02 crystal with discrete domains, where each domain has a
istribution of either the layered or the spinel-type configuration. The
d cation distribution over each crystal can be regarded as being
diate between that of the layered and the lithiated-spinel structures,
can be quantified according to Equation 12-14 and Figure 16.9..
re, combination of high-resolution TEM imaging and electron
ion has clearly revealed the structural instability of layered LixMn02
e and provided direct evidence of the spinel-type atomic arrangements
ycled LixMn02 electrode materials. Furthermore, Chiang et al.46 have
d the formation of antiphase boundaries upon transformation from
Chapter 16
d LiMn02 to the spinel configuration during electrochemical cycling.

r high-resolution TEM imaging and theoretical calculations are needed
ify the presence and the energetic of nano-meter-scale antiphase
ns in lithium transition metal oxides.
16.11. Left: A typical high-resolution TEM micrograph and the corresponding electron
on pattern {[1 T 0]1ayered zone axis}of layered LiMn02 crystals and Right: A high
on TEM image and the corresponding electron diffraction pattern { [0 1 !]spinel zone
a randomly selected crystal in the Li05Mn02 electrode after one charge and one partial
e21 •
PERSPECTIVES
Conventional TEM imaging and electron diffraction studies of lithium

materials have been invaluable in revealing phase transformations that in
lithium battery materials during lithium de-intercalation or
lation. The information has led to improved understanding of structure
ty relationships of lithium intercalation compounds and optimization of
performance and design of new electrode materials. Here we discuss
l recently developed TEM techniques that can be used to further
ve our fundamental understanding of lithium intercalation compounds in
batteries.
1. High-Resolution TEM Imaging and Simulation

Although X-ray and neutron powder diffraction techniques are
itive to lithium ion arrangements, columns of lithium in host structures
tentially be imaged (along with the columns of other atoms present) by
g use of their interactions with an electron beam in the high-resolution
HRTEM). The results of the electrons' interaction with lithium

n metal oxide lattice are encoded on the emergent electron wave as
in the phase (and, to a lesser extent, the amplitude) of the original
t electron wave. The phase of this exit-surface wave (ESW) is a
n of the oxide potential projected in the direction of the incident
98
beam • In the HRTEM, the ESW phase is imaged with an objective
t imposes additional phase changes due to aberrations such as defect
s (defocus) and spherical aberration. Focal-series reconstruction
f the ESW from a series of images obtained over a range of defocus
can mitigate the effects of spherical aberration, and move beyond the
98
resolution of a TEM to its information limit • This method has
d resolution of oxygen atoms in staurolite 100 and BaTi0 3101 , and
it has been used to successfully image and resolve individual lithium
ayered 03 LiCo02•
As shown in Figure 16.12., image simulation via the "weak-phase
approximation indicates that resolutions higher than 1 A are needed to
39
individual lithium ions in layered Co02 matrix • The ESW phase
econstructed from experimental images of a thin edge of a LiCo0 2
along the [110] zone axis shows all three types of atoms in LiCo02
16.12., right a)39• Comparison with a simulated ESW phase image
n a crystal model of 17 unit cells thick (48 A) at 0.9 A resolution
16.12., right b), reveals that all the atom peaks have the characteristics
d by the simulations. The 0-Co-0 units can clearly be distinguished
xperimental image, with each fuzzy cobalt peak flanked by a pair of
oxygen atoms, and the weak spots lying between them can be
d as the electron scattering intensities of the lithium ions (arrowed).
. .
,.•,. .• .•
• •• • •
, ,•
•
,,•
•
.• .•
t • • ••
..
6.12. Left: Simulated Images of LiCo0 2 in the [110] zone axis under thin-crystal
hase-object") conditions for increasing resolution (marked) show that atoms of cobalt e
visible at 2 A resolution, oxygen at 1.4 A, and lithium at 1 A. Right: Experimental
of Li columns confirmed by simulation a) LiCo02 ESW-phase image, reconstructed
experimental images, shows lithium ions in rows (arrowed) between 0-Co-0 groups.
phase simulation for 48 A thickness and 0.9 A resolution •
39
Chapter 16
Studies with such atomic details can provide knowledge of the

ution of vacancies and light atoms such as lithium and oxygen in
m transition metal oxides, which is invaluable to engineering of lithium
e materials.
2. EELS Analysis and Imaging
Much attention has been focused on characterizing localized atomic

ements and phase transformations by electron diffraction and imaging
ques. It is also important to analyze the chemical and electronic
ties of lithium battery materials upon lithium intercalation or de
lation. Although high-resolution images are sensitive to differences in
ral factors, the combined effect of differences in structural factors and
ion from the exact Bragg condition leads to ambiguity in revealing
nces in local chemical compositions from TEM images. EELS could
e quantitative information on light elements such as lithium and
n, and determine the valence states of transition metal and oxygen ions
49 51
ium transition metal oxides. Recently Graetz et a1. • have used EELS
es to study and confirm charge compensation mechanisms in layered
02 and LixNi0.8Co0.202 upon lithium de-intercalation, which has been
60 78
usly predicted from first-principle studies " • In addition, imaging with
can provide information about how different ions are distributed in
transition metal oxides on the nanometer-scale. Combination
ation on the chemistry, structure and electronic properties of localized
ns in lithium battery materials will enable us to optimize and advance
battery performance.
3. In-Situ TEM imaging
One way that offers the possibility of overcoming material stability

andling limitations of TEM sample preparation is to develop in-situ
characterization and observe the phase transformations in electrode
als in an electrochemical cell in-situ. The in-situ cell could be built
ning solid or liquid electrolytes. The difficulties in design of in-situ
chemical cells reside in the facts that the whole cell has to be thin h
to be electron transparent and hermetically sealed. Ross recently d a
preliminary design of a liquid in-situ cell to study the electroplating s,
102
which is a critical step in integrated circuit fabrication • Although it
und too thick to be electron transparent, it held promise for real-time
of electrochemical systems. More importantly, in-situ characterization
offer an effective route to study many interfacial phenomena between
trode and electrolyte, and provides directions to improve the safety

age aspects of battery systems. Interfacial reactions may include the
n of the solid electrolyte interface on the negative and/or positive e
before electrochemical reactions, the chemical stability of electrode s
at the electrolyte-electrode interface during electrochemical
s and storage, etc. At present, these interfacial reactions are not well
nor clearly understood. Observations in-situ before, during and after
hemical reactions would provide information essential to better
nding of these phenomena.
wledgments
The author is grateful to A. Mayes, G. Ceder, A. Vander Yen and C.

for comments insightful discussions. The International Research
Award INT 0000-429 from National Science Foundation and partial
from Department of Mechanical Engineering at the Massachusetts
of Technology and Office of Naval Research Young Investigator
00014-03-10448 are greatly acknowledged.
NCES
den & T. Reddy (eds.) Handbook of Batteries (McGraw-Hill, New York, 2002).
ushima, P. C. Jones, P. J. Wiseman & J. B. Goodenough, Materials Research
5,783-789 (1980).
eimers & J. R. Dahn, J. Electrochem. Soc. 139,2091-2097 (1992).
uku & A. Veda, J. Electrochem. Soc. 141,2972-2977 (1994).
ales, C. Perez-Vicente & J. L. Tirado, Materials Research Bulletin 25, 623-630
ahn, U. Von Sacken, M. W. Juzkow & H. Al-Janaby, J. Electrochem. Soc. 138,

(1991).
ussely, F. Perton, P. Biensan, J. M. Bodet, J. Labat, A. Lecerf, C. Delmas, A.
J. P. Peres, Journal of Power Sources 54, 109-114 (1995).
mas, M. Menetrier, L. Croguennec, I. Saadoune, A. Rougier, C. Pouillerie, G. Prado,
& L. Foumes, Electrochimica Acta 45, 243-253 (1999).
rmstrong & P. G. Bruce, Nature 381, 499-500 (1996).
Bruce, A. R. Armstrong & H. Huang, Journal of Power Sources 68, 19-23 (1997).
pitaine, P. Gravereau & C. Delmas, Solid State lonics 89, 197 (1996).
. Thackeray, P. J. Johnson, L.A. De Picciotto, P. G. Bruce & J. B. Goodenough,
Research Bulletin 19, 179 (1984).
F. David, M. M. Thackeray, L. A. De Picciotto & J. B. Goodenough, Journal of Solid
mistry 67, 316-323 (1987).
. Thackeray, W. I. F. David, P. G. Bruce & J. B. Goodenough, Materials Research
,461 (1983).
Padhi, K. S. Nanjundaswamy & J. B. Goodenough, Journal of the Electrochemical
4, 1188 (1997).
Tarascon & M. Armand, Nature 414, 359 (2001).
. Thackeray, J. 0. Thomas & M.S. Whittingham, MRS Bulletin, 39 (2000).
Chapter 16
Delmas, J.P. Peres, A. Rougier, A. Demourgues, F. Weill, A. Chadwick, M. Broussely,

n, P. Biensan & P. Willmann, Journal of Power Sources 68, 120-125 (1997).
Menetrier, I. Saadoune, S. Levasseur & C. Delmas, Journal of Materials Chemistry 9,
140 (1999).
Vitins & K. West, Journal of the Electrochemical Society 144, 2587-2592 (1997).
Shao-Hom, S. A. Hackney, A. R. Armstrong, P. G. Bruce, R. Gitzendanner, C. S.
& M. M. Thackeray, J. Electrochem. Soc. 146, 2404-2412 (1999).
F. Wang, Y. I. Jang & Y. M. Chiang, Electrochemical and Solid State Letters 2, 490-
99).
odriguez-Carjaval, G. Rousse, C. Masquelier & M. Hervieux, Physical Review Letters
0 (1998).
Rousse, C. Masquelier, J. Rodriguez-Carjajal & M. Hervieu, Electrochemical and
ate Letters 2, 6-8 (1999).
Rousse, C. Masquelier, J. Rodriguezcarvajal, E. Elkaim, J.P. Lauriat & J. L. Martinez,
try of Materials 11, 3629-3635 (1999).
J. Garcia-Adeva, D. R. Conradson, P. Villella & S. D. Conradson, J. Phys. Chern. B,
), ' 2003 107, 6704 -6716 (2003).
Hirsch, A. Howie, R. Nicholson, D. W. Pashley & M. J. Whelan. Electron Microscopy
Crystals (Krieger Publishing .Company, 1977).
B. Williams & C. B. Carter. Transmission Electron Microscopy- A Textbook for
ls Science (Plenum Press, New York, 1996).
Fultz & J. M. Howe. Transmission Electron Microscopy and Diffractometry of
ls (Springer, Berlin, 2002).
R. Buseck, J. M. Cowley & L. Eyring (eds.) High-Resolution Transmission Electron
opy and Associated Techniques (Oxford Science Publications, 1992).
Shao-Hom, S. A. Hackney, C. S. Johnson, A. J. Kahaian & M. M. Tackeray, Journal of
ate Chemistry 140, 116-127 (1998).
Shao-Hom, S. A. Hackney, C. S. Johnson & M. M. Thackeray, J. Electrochem. Soc.
2-589 (1998).
Shao-Hom, Ph.D dissertation, Michigan Technological University, Houghton,
n, USA (1998).
Shao-Hom, S. A. Hackney, A. J. Kahaian, K. D. Kepler, E. Skinner, J. T. Vaughey &
hackeray, Journal of Power Sources 81-82, 496 (1999).
Shao-Hom, S. A. Hackney, A. J. Kahaian & M. M. Thackeray, Journal of Solid State
try 168, 60 (2002).
Shao-Hom, M.A. O'Keefe, E. C. Nelson, L. Croguennec & C. Delmas. in Fall Meeting
aterials Research Society, Symposium G, Boston, Massachusetts (2002).
Shao-Hom, S. Levasseur, F. Weill & C. Delmas, J. Electrochem. Soc. 150, A366-A373
Shao-Hom, L. Croguennec, D. Carlier, F. Weill, M. Menetrier & C. Delmas, Chemistry

rials 15, 2977-2983 (2003).
Shao-Hom, L. Croguennec, C. Delmas, E. C. Nelson & M.A. O'Keefe, Nature
ls 2, 464-467 (2003).
M. Thackeray, Y. Shao-Hom, A. J. Kahaian, K. D. Kepler, E. Skinner, J. T. Vaughey
Hackney, Electrochemical and Solid-State Letters 1, 7-9 (1998).
M. Thackeray, C. S. Johnson, Kahaian, A.J., K. D. Kepler, J. T. Vaughey, Y. Shao S.
A. Hackney, J. Power Sources 81-82, 60 (1999).
. Peres, F. Weill & C. Delmas, Solid State Ionics 116, 19 (1999).
Wang, Y. I. Jang, B. Huang, D. R. Sadoway & Y. M. Chiang. in 9th International
um on Intercalation Compounds abstract N°lll (Molecular Crystals and Liquid
Science and Technology, Arcachon (FRANCE), 1998).
. Wang, Y. I. Jang, B. Y. Huang, D. R. Sadoway & Y. T. Chiang, J. Electrochem. Soc.
3-480 (1999).
. Wang, Y. I. Jang, B. Y. Huang, D. R. Sadoway & Y. M. Chiang, Journal of Power

82, 594-598 (1999).
hiang, H. F. Wang & Y. I. Jang, Chemistry of Materials 13, 53-63 (2001).
risch H., R. Yazami & B. Fultz, Electrochemical and Solid State Letters 5, Alii
roguennec, P. Deniard, R. Brec, M. T. Caldes Ricos & L. Brohan, Mol. Cryst. Liq.
1, 101-108 (1998).
aetz, C. C. Ahn, R. Yazami & B. Fultz, Journal of Physical Chemistry B 107, 2887-
03).
abrisch, R. Yazami & B. Fultz, Electrochemical and Solid State Letters 5, A111-A114
raetz, A. Hightower, C. C. Ahn, R. Yazami, P. Rez & B. Fultz, Journal of Physical

y B 106, 1286-1289 (2002).
ightower, C. C. Ahn, B. Fultz & P. Rez, Applied Physics Letters 77, 238-240 (2000).
hao-Horn, L. Croguennec, A. Gloter, M. Guilmard, C. Colliex & C. Delmas,
y of Materials, submitted (2003).
. Thompson, Physical Review Letters 40, 1511-1514 (1978).
. McKinnon & J. R. Dahn, Solid State Communications 48, 43-45 (1983).
elmas, M. Menetrier, L. Croguennec, S. Levasseur, J. P. Peres, C. Pouillerie, G.
Fournes & F. Weill, Inter. J. of lnorg. Mater. 1, 11 (1999).
ander Yen & G. Ceder, Electrochemical and Solid State Letters 3, 301-304 (2000).
atti, Physical Review B 61, 1795-1803 (2000).
olverton & A. Zunger, Physical Review B- Condensed Matter 57, 2242-2252 (1998).
olverton & A. Zunger, Physical Review Letters 81, 606-609 (1998).
olverton & A. Zunger, Journal of Power Sources 82, 680-684 (1999).
arlier, A. Vander Yen, C. Delmas & G. Ceder, Chemistry of Materials 15, 2651-2660 e
Dompablo, A. Vander Yen & G. Ceder, Physical Review B 66, art. no.-064112
. Arroyo y de Dompablo & G. Ceder, Chemistry of Materials 15, 63-67 (2003).

adoune & C. Delmas, Journal of Solid State Chemistry 136, 8-15 (1998).
evasseur, M. Menetrier, E. Suard & C. Delmas, Solid State Ionics 128, 11-24 (2000).
i, J. N. Reimers & J. R. Dahn, Solid State lonics 67, 123-130 (1993).
hzuku, A. Ueda & M. Nagayama, J. Electrochem. Soc. 140, 1862-1870 (1993).
irano, R. Kanno, Y. Kawamoto, Y. Takeda, K. Yamaura, M. Takano, K. Ohyama, M.
Y. Yamaguchi, Solid State Ionics 78, 123-131 (1995).
. S. R. Thomas, P. G. Bruce & J. B. Goodenough, Solid State lonics 17, 13-19 (1985).
ikkawa, S. Miyazaki & M. Koizumi, Journal of Solid State Chemistry 62,35 (1986).
ichta, M. Salomon, S. Slane, M. Uchiyama, D. Chua, W. B. Ebner & H. W. Lin,
f Power Sources 21, 25-31 (1987).
olenda, A. Stoklosa & T. Bak, Solid State Ionics 36, 53-58 (1989).
Kemp & P. A. Cox, J. Phys.: Condens. Matter 2, 9653-9667 (1990).
. Amatucci, J. M. Tarascon & L. C. Klein, J. Electrochem. Soc. 143, 1114-1123
. Aydinol, A. F. Kohan & G. Ceder, Journal of Power Sources 68, 664-668 (1997).
ander Yen, M. K. Aydinol & G. Ceder, J. Electrochem. Soc. 145, 2149-2155 (1998).
ander Yen, M. K. Aydinol , G. Ceder, G. Kresse & J. Hafner, Physical Review B 58,
7 (1998).
elmas, J. J. Braconnier & P. Hagenmuller, Materials Research Bulletin 17, 117
egel, J. Hirschinger, D. Carlier, S. Matar, M. Menetrier & C. Delmas, J Phys. Chern.

66 (2001).
Chapter 16
Carlier, I. Saadoune, E. Suard, L. Croguennec, M. Menetrier & C. Delmas, Solid State

44, 263 (2001).
Carlier,l. Saadoune, M. Menetrier & C. Delmas, J. Electrochem. Soc. 149, Al310
M. Paulsen, C. L. Thomas & J. R. Dahn, J. Electrochem. Soc. 147, 861 (2000).

M. Paulsen & J. R. Dahn, J. Electrochem. Soc. 147, 2478 (2000).
M. Paulsen, C. L. Thomas & J. R. Dahn, J. Electrochem. Soc. 147, 2862 (2000).
M. Paulsen, J. R. Mueller-Neuhaus & J. R. Dahn, J. Electrochem. Soc. 147, 508 (2000).
M. Paulsen, R. A. Donaberger & J. R. Dahn, Chern. Mater. 12, 2257 (2000).
Lu & J. R. Dahn, J. Electrochem. Soc. 148, A710 (2001).
Lu & J. R. Dahn, J. Electrochem. Soc. 148, A237 (2001).
Mendiboure, C. Delmas & P. Hagenmuller, Materials Research Bulletin 19, 1383
Dupont, M. Hervieu, G. Rousse, C. Masquelier, M. R. Palacin, Y. Chabre & J. M.

n, Journal of Solid State Chemistry 155, 394 (2000).
J. Gummow, A. De Kock & M. M. Thackeray, Solid State lonics 69, 59-67 (1994).
M. Tarascon, E. Wang, F. K. Shokoohi, W. R. McKinnon & S. Colson, Journal of the
chemical Society, 2859 (1991).
J. Gummow, Ph.D dissertation, University of Cape Town (1993).
J. Gummow, M. M. Thackeray, D. Wif & S. Hull, Materials Research Bulletin 27,
7 (1992).
Li, J. N. Reimers & J. R. Dahn, Physical Review B 49, 826 (1994).
Huang & R. Frech, Solid State lonics 86-88, 395-400 (1996).
A. O'Keefe, Ultramicroscopy 47, 282-297 (1992).
Schiske. in Image Processing And Computer-Aided Design (ed. Hawkes, P. W.) 82-90
Press, London, 1973).
. H. Downing, M. Hu, H. Wenk & M. A. O'Keefe, Nature 348, 525-528 (1990).
. L. Jia & A. Thust, Phys. Rev. Letts 82, 5052-5055 (1999).
M. Ross, IBM J. Res. Develop. 44, 489 (2000).
ter 17
LIQUID ELECTROLYTES: SOME

HEORETICAL AND PRACTICAL ASPECTS
M. Nazri
versity of Windsor, Department of Chemistry, Windsor, Ontario, N9B 3P4, Canada
INTRODUCTION TO SOLVENTS
The basic requirements of a suitable electrolyte for electrochemical
s are high ionic conductivity, low melting and high boiling points,
cal and electrochemical stability, and safety. Electrolyte conductivity
ectrochemical stability are key parameters in selecting an electrolyte
odem electrochemical devices such as advanced batteries, fuel cells,
capacitors, sensors, and electrochromic displays. These parameters,
ctivity and electrochemical stability, will receive particular attention in
apter. Although progress has been made in enhancing the conductivity d
electrolytes, particularly the polymeric ones, liquid electrolytes are sed
in most electrochemical systems. The solvent properties, and ics of
ion solvent interactions, must be understood in designing new lytes. In
this chapter, a short but general introduction to properties of ts and ion-
solvent dynamics is discussed.
The history of electrolyte development goes as far back as the work of
philosophers in search for a universal solvent, the so-called
hest". In search of Alkahest, many solvents and chemical rules were
ered such as "like dissolves like" (similia similibus solvuntur) as
in Table 17.1. Later, the theory of osmotic pressure by van't Hoff
1911), and the theory of electrolyte dissociation by Arrhenius (1859-
were discovered. Many speculations about the nature of solute-solvent
tions and the influence of solvent media on the rate of chemical
n were proposed in the early eighteen-century. The role of solvents on
cal equilibrium, on tautomerism (i.e. keto-enol tautomerism), and the
menon of solvatochromism (shift of UVNis absorption bands due to
nges of the index of refraction) were discovered. 1' 2 Scheibe et al. have
Chapter 17
ted the solvating ability of solvents to their degree of influence on

3
n rate, chemical equilibrium, and shift in absorption spectra.
In the early 19th century, it was much argued that solvents and
lytes should not be considered as a continuum, usually characterized
ir physical properties such as dielectric constant, density, index of
ion, etc., but as a discontinuum with mutually interacting molecules. It
scovered in many instances that the density of the solvent in solvation
around small cations is greater than that of the bulk solvent, due to the
ion-solvent interactions. The interaction forces between solvent
ules are too big to treat them like gases but not significant enough to
er them as solids. Although the nearest neighbor distances in liquid
lids are very close, the volume of liquid is about 10% greater than the
forms, due to the significant defects that exist in the coordination
r of ions in the liquid state.
7.1. Various solvents with different degree of hydrophobicity.
und Well soluble in
arbons hydrophobic Hydrocarbons, Ethers,

drocarbons Halohydrocarbons
Carboxylic Esters
drocarbons
Alcohols, 1,4-Dioxane
des Acetic Acid
s Alcohols, Water
ylic Acids
c Acids
hydrophilic Water
n addition, the solvent molecules in the solvation shell of an ion have

residency time, and the dynamics of solvent exchange processes vary
ide range, depending on the size and charge of the ions and the
cal nature of the solvent. In some cases the solvation shell is stable
h 3)6]
that the complex can be considered as a molecular entity, e.g.
H +. On the other hand, in some cases the solvent exchange is
2
than the time scale of NMR, and solvent molecules in the solvation
annot be distinguished from those in bulk solution. The complexity of
tions in liquid state makes the theoretical treatment of electrolyte
uid Electrolytes:Some Theoretical and Practical Aspects 511
re a difficult task, and still most reliable data are those obtained
mentally, using NMR, vibrational spectroscopy, and fast spectroscopy
ues.
When dealing with mixed solvents, the solubility parameter relates to
4
lar enthalpy of vaporization for a gas of zero pressure by
Hildebrand parameter= o = [( Hv -RT)Nm1 112 (17.1)
V m is the molar volume and Hv is the molar enthalpy of

zation. The o is the measure of work necessary to separate solvent
les to create a cavity to accommodate the solute.
COLUMBIC FORCES BETWEEN IONS AND

DIPOLAR SOLVENTS
n addition to the intermolecular forces, such as van der Waals forces
e well-known hydrogen bonding forces responsible for the non-ideal
or of solvents and electrolytes, there are other important forces
n ions and neutral molecules with permanent dipole moments (dipolar
s) that can alter the physical and chemical properties of solvents. The
mportant forces which exist between solvent molecules and solute s
are discussed in the following section.
. Ion-Dipole Forces
olvents with unsymmetrical charge distribution (dipolar solvents)
a permanent dipole moment, - The magnitude of the dipole moment
s on the value of charges on the opposite sides of the molecule and
eparation distance as defined by = q·r, where q is the charge
ing equal value of charge on each side of the molecule), and r is the
5
ion distance between charges. Most organic solvents have permanent
3
°
moment in a wide range from 0 to 18xl0. Coulomb-meters (Cm). In
lar, cyclic and linear organic carbonates, used in lithium batteries and
larly dealt with here, have a large 3difference in permanent dipole
t, (i.e.
10"
30 °
llpropylene carbonate = 16.7 X 10" Cm, and J.l.dietbyl carbonate= 3 X
he dipole moment, relative dielectric constant, melting point, and
point of several common organic solvents are listed in Table 17.2.
ter molecule is also listed as a reference for comparison.
Chapter 17
6 8
7.2. Common dipolar organic solvents and their physical properties. -
nt tmJ°C tru1-°C e u (10-3°Cm)
ide
ol
acid
exanol
ne Carbonate
ethane
trile
yl sulfoxide
ne
e
enzene
nzene
ethylphophoric
amide
henone
e
acetate
xanone
ne
drofuran
carbonate
yl carbonate
propyl carbonate
isopropyl carbonate
opyl carbonate
opropyl carbonate
e carbonate
xane
oethene
ether
e
disulfide
oromethane
ne
lting point, tbp = boiling point, E = dielectric constant, 1.1 = dipole moment.
solvent molecule with a permanent dipole when placed in the field

on, orients itself to minimize its mutual potential energy so that the
ve end of the solvent becomes directed toward the ion and the
ve end orients itself away from the ion. The ion-dipole interaction
9
is given by
2
Uion-solvent =- (ze!-l cos 9)/(4 1tco) r (17.2)
ze is the charge of the ion, e is the dipole angle relative the line joining
to the center of dipole, and Eo is the vacuum permittivity. For a
endate ion-solvent configuration, the solvent molecule and the ion are
on the same line, and e = 0. The ion-dipole forces are very important
relative stability of the solvation shells. It also plays a major role on
ed solvation when multi-component solvents are used.
. Dipole-Dipole Interactions
olvent molecules with permanent dipole moments interact with each
ia directional electrostatic interactions. The dipole-dipole orientation
gly temperature dependent, and thermal energy prevents optimum
tion of solvents toward each other. The dipole orientation is
ally favored, and at higher temperatures all other dipole orientations
more populated. The dipolar interaction is mainly responsible for the
10 11
tion of dipolar solvents. ' The dipole-dipole interaction is given by 12
(17.3)
KB is the Boltzmann constant, and T is the absolute temperature.
. Dipole-Induced Dipole Interactions
solvent with permanent dipole moment can induce a dipole moment

eighboring solvents. The magnitude of the induced dipole moment s
13
on the polarizability of the solvents, a, and is defined as
1-l induced = 4 1tEo aE (17.4)

Chapter 17
E is the electric field strength. The induced moment is directional and

long the induced dipoles. The energy of dipole-induced dipole
13
tion between solvent molecules is given by the following equation
2 2 6
Udipole-induceddipole =- (UtJl2 + a2Jlt )/(41teo)r (17.5)
In addition to the solvent-solvent molecules induced dipole moment

tion, there is also strong induced polarization of solvent by ions in the
lytes. The magnitude of the ion-solvent induced polarization depends
polarizability of the solvent molecules and the degree of distortion of
olecule's electron cloud. The energy of ion-induced dipole interaction
13
ned as
2 2 4
Uion-induceddipole =- (z e a)/(41tEo)2 r (17.6)
4. Instantaneous Dipole-Induced Dipole Interaction

Forces
n general, the electron movement in one species (atom or molecule)
s a small dipole moment that polarizes the electron cloud in its
oring system. This induced interaction is called the dispersion
tion and strongly depends on polarizability and ionization potential of
14
ting species according to the following equation
(17.7)
Iis the ionization potential. The dispersion energy between two

r molecules then will be reduced to
Udispersion = - (3a2I)/4 6
r (17.8)
6
The dispersion energies are very short range as expressed in 1/r . This
important for molecules with polarizable 1t electrons. The molecular
ation of polyaromatic molecules are mainly related to the dispersion
among polyaromatic molecules with conjugated 1t electrons. From the
ons above, it is clear that the dispersion energy is also important for the
on of molecules with high polarizability values. For example, the
sion interaction energy of two molecules with a = 3x1o- m , I =
30 3
J, and r = 3x10- m is calculated to be -11.3 kJ/mol or -2.7

19 10
ol. For a similar molecular system the dipole-dipole energy is only

J/mol or -1.1 kcal.mo1. 15
refractive
Chapter 17
Dipolar organic solvents, usually with a large dielectric constant and a

high dipole moment, do not strongly participate in hydrogen bonding,
their hydrogen is not connected to a strong electronegative atom (0, N,
). Their C-H bond is not strongly polarized. However, they may
pate in electron pair donors (EPD), electron pair acceptor (EPA)
s.
ELECTRON PAIR DONOR-ELECTRON PAIR

ACCEPTOR INTERACTIONS (EPD/EPA
INTERACTIONS)
Molecules with high energy, occupied orbitals, and molecules with a
ently low energy unoccupied orbitals, form electron pair donor/
n acceptor complexes. The long-wavelength absorption nature of
PA complexes is associated with the electron transfer from the donor
ule to the acceptor. This type of optical absorption is called the
e-transfer" or CT absorption. Three classes of EPD molecules have
ccepted based on the involvement of n, a, and 1t orbitals. For the n
the highest energy orbital is the lone-pair of a heteroatom (R20, R 3N,
etc.). The a types are the involvement of the a-type bond (R-Hal), and
n-type association is the involvement of a pair of n-electrons of
rated molecules (alkyl benzenes, polycyclic aromatics, etc.).
A similar classification also exists for electron acceptor molecules,
n, a, or 1t orbitals are involved. Therefore nine different combination of
PA association have been distinguished. The reaction enthalpy of the
PA associations is used as their bond strength, and varies between -42
18
kJ/mol (-10 to 45 kcal/mole). The enthalpy off ormation of a n-EPD/
, such as Et20-BF3, is -50kJ/mole, -11.9 kcal/mole, and for a n
-EPA such as benzene/1,3,5,-trinitobenzene is -8 kJ/mole or -1.9
ole.
The n-donor solvents strongly solvate cations in electrolytes. They are
19
olvents for salts and they are known as coordinating solvents. The n
solvents are good media for inorganic reactions. Examples of n-donor
ts are pyridine, dimethyl sulfoxide, triamide, N,N-dimethylformamide,
e, methanol, and water.
20
Gutmann and coworkers '21 have developed a semi-quantitative scale
degree of nucleophilic properties of EPD-solvents, called the donor
r (DN) or donicity. The donicity is defined as the negative H value
adduct formation between the antimony pentachloride (SbCl5) and the
n pair donor solvent (EPD), in a dilute solution of non-coordinating
hloroethane solvent, according to the following equation
id Electrolytes:Some Theoretical and Practical Aspects 517
<:1-<:IIz-<:liz-<:1
EPD + Sb<:ls ====================EPD+- -sb<:ls
At room temperature
solvent Donor Number= DN=Donicity =- IIEPD-sbcis (kcal/mole)
The donor number is a very useful concept and correlates well with
ctrochemical half-wave redox potentials, kinetic rates, and the NMR
20
al shift signals for various EDP solvents. - 26 The donor numbers of
ommon organic solvents are listed in Table 18.3 of <:hapter 18.
A quantitative electron accepting ability (accepting number, AN) for
n pair acceptor solvents has also been developed to determine their
philic properties. This concept is developed by Gutmann and
31
ers based on P-NMR chemical shift of triethyphosphine oxide
27
0) in presence of an EPA-solvents at room temperature. - 29 The
n used to determine the acceptor number AN is
8corr (A)-8corr (n-C H14 ) 100 =s: 2 3481

6
· Ucorr· · ppm
8corr(Et 3 PO- SbCl 5 )- 8(n- C6 H 14
)
(17.11)
31
N is obtained from the relative P-NMR chemical shift values
to the 1:1-adduct Et3P=0 - Sb<:l 5 dissolved in 1,2-dichloroethane.
emical shift for n-hexane has been used as reference, ANn-hexane= 0.
N values of a number of common organic solvents are also listed in
18.3 of chapter 18. The acceptor numbers are dimensionless number
to those of Sb<:Is.
SOLVATION
he interaction between ions and solvent molecules is important in
ytic media. The shell formed around the ions by solvent molecule is
solvation (hydration in aqueous electrolytes). The energy of solvation
hange in Gibbs energy when an ion is transferred from vacuum into a
: Gsolvatiom is the solvation ability of the solvent. In order to dissolve a
a solvent, the interaction energies between solvent molecules, g
the all-electrostatic, polarization, and dispersion energies, as well
attice energy of the salt, must be overcome.
he electrostatic field that solvent molecule experience in close
ty of the ion in a solvation shell is in the order of 106-107 V/cm. This
Chapter 17
ous field increases the density of the solvent in a solvation shell as

red to that of the bulk solution. The effect of solvation in different
on chemical reaction rates is well known, and in most cases the
t media need to be mentioned in defining the reaction kinetics. Criss
alomon have provided a comprehensive compilation of Gibbs free
30
es of solvation.
1 Solvation Number
The solvation number is the number of solvent molecules in the
y shell attached to the ion in a way that their translational and
nal motions exhibit changes from the bulk solution. In some cases, the
t molecules in the second sphere around the ion are also considered as
ion number. The residency time of a solvent molecule in the solvation
is short, so that in most cases they can be monitored in the time scale
17
R spectroscopy. The 0-NMR spectroscopy of Ae+ and alkali ions in
4
us solution has indicated a slower than 10 /sec solvent exchange for
4
nd faster than 10 /sec exchange for the alkali ion hydration. 31• 32 NMR
oscopy has proved to be a very useful tool for studying the dynamic of t
molecule in solvation shells of various ions.
The solvation number is usually measured by electrolytic transport
ds. The solvation number is higher for the harder ions (small size), as
3 34
red to the softer ions (large size), with similar charge? • The data
electrolytic transport measurement and from conductance
rements indicate that solvation numbers increase according to: Li+>
+>Rb+>Cs+; the alkaline earth ions are more solvated than the alkali
nd their order is: Mg++>Ca++>Sr++>Ba++. Anion solvation numbers for
35
n ions follow the order: F>CI">Br->L
2. Selective or Preferred Solvation

In mixed solvents containing salt, e.g. binary or ternary solvent
lytes, the interaction of solvent-solvent molecules and ion-solvent
tions are complicated. It is often observed that the ratio of solvent
ules in the solvation shell is different from that of the bulk solution.
ns in the electrolyte tend to solvate themselves more with the solvent
ules that produce more negative Gibbs energy of solvation, Gsolvation·
referred solvation results in lack of homogeneity in the solvation shell
the bulk of mixed solvent electrolytes. When the cation and the anion
ferentially solvated with the same solvent, it is called homoselective
on, and when the cation is solvated preferentially by one solvent and
ion by other solvent molecules, it is called heteroselective
36 37
on. '
everal methods have been developed to study selective solvation in
mponent electrolytes. The IR and NMR spectroscopy have been used
y spectroscopists and the conductance method and Hittorf transference
6
ements by electrochemists?
THE BASIC REQUIREMENTS OF

ELECTROLYTES FOR LITHIUM BATTERIES
suitable electrolyte for lithium batteries must have the following
eristics:
High ionic conductivity to minimize the cell resistance and resistive

heating of the device.
High chemical stability to prevent decomposition of electrolyte on

the surface of a highly reducing anode materials, e.g. metallic lithium
or lithiated graphite, LiC6, and a highly oxidizing cathode material
such a delithiated Li1_xM02 electrodes, where M = Co, Ni, Mn, or
mixed transition metals.
Electrochemical stability to tolerate the high voltage difference

between anode and cathode (>4 V). Not many organic solvent can
tolerate such wide voltage range without being reduced or oxidized.
Low melting point to provide sufficient conductivity at sub-ambient

temperatures and prevent solidification and phase separation.
High boiling point to provide safety and prevent explosions resulting

from high pressure build-up in the cell.
Non toxicity to be accepted environmentally for ease of handling,

mass production, and waste treatment.
Low cost to be affordable and compete with existing power sources

using aqueous electrolytes.
Chapter 17
1. Electrolyte Conductivity
We wish to stress in the final part of this chapter points 1 and 3 of the
list. To this end, some data on the anodic stability of several
ates and their conductivity in a broad temperature range will be
ted here.
mong various solvents presented in Table 17.2, the organic
ates, particularly the cyclic carbonates, are the most suitable solvents
plication in a battery electrolyte. However, the viscosity of the cyclic
ates is high and needs to be modified to improve their low temperature
mances. Addition of a linear carbonate with high fluidity enhances the
conductivity of multi-solvent blend electrolytes. The conductivity of
l organic carbonates at various salt concentrations and at different
ratures has been measured. The salt concentration was chosen from
to 1M LiPF6• Normally, below 0.4M salt concentration a marked
se in conductivity of the electrolyte is observed and considered not to
actical. Above the 1M concentration, there is significant salt
itation at low temperatures, t<-2o·c, that is detrimental to most
chemical devices. The temperature range was selected from -40 to
in our experiments. This is a practical temperature range for most
chemical devices. It has been shown that there is no electrolyte with a
solvent that can provide all the requirements of a practical electrolyte.
ore multi component electrolytes containing multi-blend solvents are
Several binary and ternary organic carbonate solvents are used to
ce the performance of the electrolytes, particularly their low
rature conductivity. In fluid dynamics, the mixing rules for the
38
ity and dielectric constant have been defined
(17.12)
xi is the mole fraction of solvent I and lJi is the viscosity of the

nent i.
(17.13)
Yi = xNiN m = volume fraction of component I, and Vi is the molar

e fraction of component i, and V m is the molar volume of mixture
ole)
(17.14)
There are several reviews concerning the conductivity and properties

naqueous electrolyte solutions.39 -43 Electrolytes are composed of a
salt (LiC104, LiPF6, LiCF3S03, etc) dissolved in high purity cyclic

ear organic carbonates. The salt used in lithium batteries also must be
in a wide voltage range (0-4.5 volts). LiPF6 is the most studied salt
used in almost all advanced high voltage lithium cells. LiPF6 has a
tory electrochemical stability, but its thermal stability is of concern,
table up to ca. 15o·c as shown by thermogravimetric analysis. LiPF6
extremely moisture sensitive and reacts with residual water in the
lyte and forms HF that is detrimental to the performance of the cell,
larly on the cathode side. The released HF also reacts with the anode
LiF, which is a poorly ionic conductor and increases the impedance
lectrode/electrolyte interface. There is an extensive search to develop
44 52
thium salts for lithium batteries. -
inary and ternary solvents containing LiPF6 have been studied in an
t to improve the low-temperature conductivities of electrolytes to be
Li batteries. Table 17.3 shows the various solvents and electrolytes
ed by the author from the conductivity point of view.
.3. Solvents and electrolytes.
UPF•
: LiPF•
iPF•
1. Conductivity of Single Solvents vs. Concentrations and

Temperatures
he first class of electrolytes contained only single carbonate solvents

fferent concentration of salt. PC- DMC- and EMC-based electrolytes
en particularly investigated.
mong them, LiPF6-EMC has proven quite interesting. The
tivity of this asymmetric linear carbonate solvent containing various s
of LiPF6 is shown in Figure 17.1. The conductivity of ethyl methyl
ate (EMC) based electrolytes containing LiPF6 remain high even at
mperatures (-4o·q. This solvent has a very low melting point. It is
ing that substituting one of the methyl groups in DMC with another
group significantly affects the melting and boiling temperatures of
ased electrolytes (Table 17.2). This substitution introduces significant
r in the electrolyte system that markedly changes the glass transition
fluidity of this solvent. Asymmetric solvents such as EMC are good
Chapter 17
ates for low temperature electrolyte and investigation of the

ochemical properties of materials at low temperatures.
-50 0 50
Temperature ( C)
17.1. Conductivity of ethyl methyl carbonate (EMC) containing, ( +) 1M, (.&) 0.8M,
M, and (o) 0.4M LiPF6.
.2. Conductivity of Binary Solvents-LiPF6 Electrolytes vs.

Concentrations and Temperatures
It is almost impossible to find a single solvent electrolyte meeting

requirements for application in practical devices. This is why mixed
c carbonates have been more often investigated. According to the
n rule, the conductivity of an electrolyte is inversely proportional to its
ity. Therefore, multi-blend electrolytes with solvents of different
ities are preferred. Data on the conductivities of mixtures of cyclic and
carbonate solvents are here reported.
Figure 17.2 shows the conductivity of EC-DMC-based electrolyte for
s salt concentrations and in a wide temperature range (-40- 50°C). It is
able that the conductivity of this mixture is very different than the
solvent electrolyte. DMC-LiPF6 has very poor conductivity below
and that of EC-LiPF6 is very negligible below 30°C, as EC melts at
However, the mixed solvent is fluid down to -30°C, and has
cant conductivity from -30 to 50°C.
id Electrolytes:Some Theoretical and Practical Aspects
523
"' "
§
""
.
....
en....
20,00
s 15,00
. e 10,oo
' - "
>
•; :J
.§ 5,00
s:::::
0 0,00 -50
u 0 50 100
Temperature ( C)
7.2. Conductivity of ethylene carbonate- dimethyl carbonate (EC-DMC) containing

.._) 0.8M, (•) 0.6M, and (o) 0.4M LiPF6.
The conductivity of EC-EMC containing various amounts of LiPF6

0 to 50°C is shown in Figure 17.3. Again, high ionic conductivity
erved at temperatures well below the melting point of the EC solvent.
utual interaction of solvents in mixed electrolytes has opened a new
search of electrolytes with unique properties for low temperature
hemistry.
"'s" 15,00
"" 10,00
en
'-"
s
·- ..>..
..
>
......
u
.g
s:::::
0 0,00
u
-50 -30 -10 10 30 50
Temperature ( C)
.3. Conductivity of ethylene carbonate- ethyl methyl carbonate (EC

ntaining, ( +) 1M,(.._) 0.8M, (•) 0.6M, and (o) 0.4M LiPF6.
Chapter 17
The most striking feature of the EC-EMC mixed electrolyte is that its
ctivity does not change significantly for 1M, 0.8M, and 0.6M LiPF6
oncentrations, even at high temperatures. Using this remarkable
lyte, the cost of salt can be reduced by 40%. The other remarkable
ty of this electrolyte is that the EC with melting point of 39°C does not
separate at temperatures significantly below its melting point.
.3. Conductivity of Ternary Solvents- LiPF6 Electrolytes vs.

Concentrations and Temperatures
In order to optimize the performance of the electrolyte in lithium

es, several electrolytes with a ternary solvent mixture have been
d. The conductivity of PC-EC-DMC containing various amounts of
as a function of temperature is shown in Figure 17.4. Several
ting observations are worth mentioning. First, the conductivity of
ns with 1M, 0.8M and 0.6M LiPF6 are almost the same over the entire
rature range of --40 to 50°C. Second, the overall conductivity of the
solvent is higher than that of the single and binary solvent
lytes. Third, there is no effect of salt precipitation and phase
tion of EC. The conductivity for 0.4M solution is markedly lower than
6-1M solutions. This ternary solvent is suitable for an advanced
battery, because it provides sufficient conductivity at low
ratures. It also contains ethylene carbonate, which selectively solvates
hium ion, and its decomposition on the lithiated graphite anode
es a good protective SEI layer. The nature of the SEI film formed in
-..
] 20,00
Cl)
5 15,00
>..
·-.§ 10,00
......
5,00
§ 0,00
u 100
-50 0 50
Temperature ( °C)
7.4. Conductivity of propylene carbonate- ethylene carbonate- dimethyl

e (PC-EC-DMC) containing, (+)1M, (.6.) 0.8M, (•) 0.6M, and (o) 0.4M LiPF6•
ctrolyte is discussed in Chapter 5. Further, the electrolyte contains

nt EC concentration that can be used with a graphite anode, because
vent does not co-intercalate between the graphite layers. The ternary
s can be tailored to reduce the amount of lithium salt used in lithium
s.
he conductivity of multicomponent electrolytes with PC-EC-EMC is
in Figure 17.5. The conductivity values for salt concentrations in the
f 1M to 0.6M are very similar between -40 and 50°C, making these
ytes quite interesting for applications in Li batteries.
s 15,00
Cfl
5 10,00
.......
:.>...
.........
5,00
§
u 0,00
-50 -30 -10 10 30 50
Temperature ( °C)
7.5. Conductivity of propylene carbonate- ethylene carbonate- ethyl methyl

e (PC-EC-EMC) containing, (+)1M, (A) 0.8M, (•) 0.6M, and (o) 0.4M LiPF6 •
he multi-blend solvent provides the opportunity to use less salt and

the overall cost of the lithium battery. In most commercial cells,
ytes with 1M LiPF6 are still in use. This high salt concentration is
etrimental to the performance of the cells at low temperatures (-30-
due to the possibility of salt precipitation.
. The Electrochemical Stability of Organic Electrolytes
hen cycling aLi battery, several electrochemical reactions may take t

the electrodes as a function of the voltage range. In general, by
nto account all compounds that may be present in a cell, we have the
ng scheme:
Chapter 17
Oxygen reduction
OH reduction
Oxide formation ..,
Solvent oxidation .
5
--r.ithium deposition
Li-M alloy formation (UPD)
Solvent Reduction and formation of SEI layer
-----------H 20 reduction
17.1. Schematic of various electrochemical reactions and their voltage ranges.
The data collected on scheme 17.1 refer to a platinum microelectrode

a voltage scan from the cathodic to the anodic region.
The decomposition of organic electrolytes at the negative electrode
en a major concern, as there is no organic electrolyte that can tolerate
ucing power of metallic lithium or lithium doped graphite. In fact, the
that current anodes work so well is mainly because of the formation of
ective layer on the surface of anode. This protective layer is formed by
position of organic solvents (SEI layer). The nature of SEI depends on
ganic solvents used in the battery. The SEI is electronically insulating,
s significant lithium ion conductivity and allows the lithium ions to
the anode. The SEI layers that are formed in carbonate-based
lytes are remarkable in their protection of the anode.
There have been extensive investigations to find a stable organic
lyte above 4 V. The carbonates have shown stability even exceeding 5
ble 17.3) and this remarkable stability provides the opportunity to use
oltage cathodes materials (layered LiCo02 and 3-D spinel LiMn204
es). The average voltage of aLi-ion battery (LiCc/!LiCo02) is 3.75V
operating voltage range is between 3 to 4.5 V.
Anodic stability tests have been performed in electrolytes containing
d 0.8M LiPF6 (Pt microelectrode, Li reference and counter, 70°C). The
of electrolyte oxidation for each solvent is reported in Table 17.4.
id Electrolytes:Sorne Theoretical and Pr
.4. Oxidation potentials of organic solve n c
Electrolyte solvent
Propylene carbonate
Ethylene carbonate
Dimethyl carbonate
Diethyl carbonate
Methyl ethyl carbonate
Gamma-butyrolactone
Diethoxyethane
Tetrahydrofuran
2-methyltetrahydrofuran
I ,3 dioxolane
Acetonitrile
CONCLUSIONS
ome fundamental properties of single solvents, mixtures and salt

(s) combinations have been discussed. There is a complex interplay
the electrolyte components linked to the ion-solvent and solvent
interactions. Such factors as ion-dipole and dipole-dipole interaction,
solvation determine the suitability of a given solution from the
int of its transport properties. However, the parameters making a
lectrolyte usable in a given battery are numerous and this chapter has
hose for electrolytes to be used in Li batteries. Among them,
tivity and anodic stabilility have been stressed here.
Conductivity data for different combinations of cyclic and linear
tes have shown the interesting behavior of electrolytes containing the
etric linear ethyl methyl carbonate. Especially the ternary mixture EC
C (with LiPF6 as a salt) shows good conductivies over a wide
ture range. Furthermore, lowering the LiPF6 concentration from 1 M
.4 M does not cause a loss of conductivity, this obviously having a
implication on the battery cost. The carbonates have excellent anodic
es (above 5 V vs. Li+/Li, with the exception of PC), this allowing their
only in the present generation of Li-ion batteries but also in the 5 V
s now being widely investigated.
RENCES
Magat, J. Phys. Chern. Al62 (1932) 432.

. Sheppard, Chern. Abstr. 37 (1943)1654.
cheibe, E. Felger, G. RoBler, Ber. Dtsch. Chern. Ges. 60 (1927) 1406.
Chapter 17
. Hildebrand, J. M. Prausnitz, R. L. Scott, Regular and Related Solutions,

nNorstranf-Reinhold, Princeton, (1970).
J. F. Bottcher, Theory of Electric Polarization, Vol. I, Second Edition, Elsevier
ientific Publishing Co., New York (1973).
C. Weast, M. J. Astle, CRC Handbook of Data on Organic Compounds, Vol I and II,
C Press, Florida, (1985).
C. Weast, (ed.), Handbook of Chemistry and Physics, 66th Edition, CRC Press, Florida
986).
L. McClellan, Table of Experimental Dipole Moments, Freeman Co., San Francisco,
963).
H. March, M.P. Tosi, Coulomb Liquids, Academic Press, New York 1984.
L. Amev, J. Phys. Chern. 72 (1968) 3358.
Rabinowiz, A. Pines, J. Am. Chern. Soc. 91 (1969) 1585.
H. Keesom, Z. Physik23 (1922) 225.
Debye, Z. Physik 22 (1921) 302.
. Mahanty, B.W. Ninham, Dispersion Forces, Academic Press, New York (1977).
H. Yoder, J. Chern. Educ. 54 (1977) 402.
E. Hill, W.E. Vaughan, A.H. Price, M. Davice, Dielectric Properties and Molecular
haviour, Van Norstrand Reinhold Co., London (1969).
E. Thomas, R.M. Darling, J. Newman, Mathematical Modeling of Lithium Batteries,
Advaces in Lithium-ion Batteries, (eds. Schalkwijk, W.A., Scrosati,B.), Kluwer
ademic I Plenum Publishers, Boston (2002).
Paetzold, Z. Chern. 15 (1975) 377.
S. Drago, L.B. Parr, C.S. Chamberlain, J. Am. Chern. Soc. 99 (1977) 3203.
Gutmann, Coordination Chemistry in Non-Aqeous Solvents, Springer, Wien, NY
68).
Gutmann, Coord. Chern. Rev. 2 (1967) 239.
Mayer, Pure Appl. Chern. 41 (1975) 291.
Gutmann, Pure Appl. Chern. 15 (1973) 141.
Schmid, V. A. Sapunov, Chemie Verlag Nethreland(1982).
. Bender, Chern. Soc. Rev. 1986, 201.
Mayer, V. Gutmann, W. Gerger, Pure Appl. Chern. 51 (1979) 1697.
Schmid, J. Sol. Chern. 12 (1983) 135.
M. Criss, Salomon, M., Thermodynamic Measurements- Interpretation of
ermodynamic Data, in A. K. Covington, T. Dikinson, (eds): Physical Chemistry of
anic Solvent Systems, Plenum Press, London, NY (1973).
. Gordon, The Organic Chemistry of Electrolyte Solutions, Wiley, New York (1975).
A. Jackson, J. F. Lemons, H. Taube, M. Alei, J. A. Jackson, J. Chern. Phys. 41 (1964)
02.
S. Amis, J. F. Hinton, Solvent Effects on Chemical Phenomena, Vol.l, Academic
ss, New York (1973).
S. Amis, Solvation of Ions, in Solutions and Solubilities, Vol. III, Part 1, of the series
chniques of Chemistry, M. R. J. Dack, (ed.), Wiley-Interscience, NewYork (1975).
. Hinton, E. S. Amis, Chern. Rev. 71 (1971) 627.
Strehlow, H. Schneider, W. Knoche, Ber Bunsenges. Phys. Chern. 77 (1973) 760, and
re Appl. Chern. 25 (1971) 327.
Strehlow, H. Koepp, H. Schneider, Z. Phys. Chern. 44 (1966) 49.
E. Blomgren, J. Power Sources 14 (1985) 39.
M. Abraham, M. J. Alamgir, J. Electrochem. Soc. 137 (1990) 1657.
Klessen, R. Aroca, G. A. Nazri, J. Phys. Chern. 100 (1996) 9334.
id Electrolytes:Some Theoretical and Practical Aspects 529
. Bloomgren, in Lithium Batteries, (ed): J. Gabano, Academic Press, New York

3, p13.
. Gores, J. Bartha!, J. Solution Chern. 9 (1980) 939.
Matsuda, J. Power Sources 19 (1987) 20.
. Dudley, D.P. Wilkinson, G. Thomas, R. LeVae, Woo, H. BJorn, C. Horvath, M. W.
ow, B. Denis, P. Juric, P. Aghakinan, J. R. Dahn, J. Power Sources 35 (1991) 59.
Watanabae, T. Nohma, I. Nakane, S. Yoshimura, K. Nishio, T. Saito, J. Power
rces 217 (1993) 43.
. S. S. Prabhu, T. P. Kumar, P. N. N. Namboodiri, R. J. Gangadharan, Appl.
trochem. 23 (1993) 151.
urbach, M. Daroux; P. Faguy, E. B. Yeager, J. Electroanal. Chern. 225 (1991) 297.
. Lee, Y. Zu, A. Hermann, Y. Geerts, K. Mullen, A. J. Bard, J. Am Chem.Soc. 121
99) 3513.
esten, U. Heider, M. Schmidt, Solid State lonics 148 (2002) 391.
Yamaki, Liquid Electrolytes, in Adavances in Lithium-ion Batteries, (W.A. Van
alkwijk, Scrosati, B., Eds.), Kluwer Academic Plenum Publishers, Boston (2002).
apter 18
ADVANCED LIQUID ELECTROLYTE

SOLUTIONS
D. Aurbach and A. Schechter
Department of Chemistry, Bar-llan University,Ramat-Gan 52900, Israel
1. INTRODUCTION
1.1. Types of Advanced Liquid Electrolyte Solutions

.1.1. Organic Polar Aprotic Solvents
The major families of this class are ethers, esters, and alkyl carbonates,
h are highly important for the field of high energy density batteries.
e solvents and their formulas are summarized in Figure 18.1. Other
ortant solvents are acetonitrile (AN), dimethylsulfoxide (DMSO), N-N
thyl formamide (DMF), methylene chloride and nitromethane. The latter
solvents are mainly important for electroanalytical use and organic
hesis. As discussed later in this chapter, the type of salt plays a major
in determining the electrochemical window of a polar aprotic system.
monly used anions, in conjunction with Li+, are PF6-, Cl04- and BF4-. All
r aprotic solutions may contain atmospheric contaminants such as 02,
and C02, which are reactive and may play an important role in
rmining the electrochemical behavior.
.1.2. Inorganic Polar Aprotic Solvents
Oxyhalide solvents such as SOCh, S02Ch, and S02 are used in

rochemical systems together with a salt such as LiA1Cl4• Their major use
r batteries.1 These species may serve as both the electrolyte solvent and
athode active material in Li batteries. There are reports on the study of
rochemical reactions in S02 as a solvent.2 There are many solvents of
ced Liquid Electrolyte Solutions 531
MOyXz type (M=S, P, N, and even metals such as V and Cr; X=Cr, Bf,
hat were mentioned in the literature in connection with possible battery
ications. However, SOC}z and S02 seem to be the most important
1
ents in this group for electrochemical systems.
.1.3. Nonpolar Electrolyte Solutions
As shown by Gileady et al} it is possible to form non-polar electrolyte

ions based on aromatic hydrocarbons such as ethyl benzene, toluene,
4
ene and mesitylene. The electrolyte comprises A}zBr 6 and KBr. By
g a nearly molar concentration of these species, it is possible to obtain
ific conductivities of the same order of magnitude as that measured in
ETIIERS
Diethyl ether (DEE) CH,CH,OCH,CH,
Tetrahydrofuran (TifF)
6
2-Methyl-THF (2Me ·
T HF) q
Diethoxyethane CH,CH,OCH,CH,OCH,CH,
1-3 Dioxolane
0
ALKYL CARBONATES
Ethylene carbonate (EC)
)a
\._/
Propylene carbonate (PC)

)a
Dimethyl carbonate (DMC) 9
CH30COCH3
Diethyl carbonate (DEC) y

CH3CH20COCJi2CH3
Ethyl methyl carbonate (EMC) II
CH3CH70COCH3
ESTERS
Methyl fonnate (MF) II
HCOCH3
y-Butyrolactone (BL)
0
Methylacetate 9
CH3COCH3
(a)
18.1. The most important organic polar aprotic solvents for Li batteries.
Chapter 18
r aprotic solutions (around 5 mS/cm) at ambient temperature. It was

d that these electrolytes do not dissolve in the above solvents to form
or ion pairs, but rather they form clusters which are charged
egates of species such as AlzBr7-K+.
5
.1.4. Molten Salts
Molten salt electrolyte systems comprise salts in the liquid state

ten), which form ionic phases and are highly ionically conductive, have
de potential range of stability, and thus reach specific conductivities
parable to those of room temperature, concentrated aqueous solutions
<x<IO S/cm).These systems can be divided into two classes:
I. High temperature molten salts.
. Ambient temperature molten salts.
.2. Scope of Subjects of Interest

Dealing with nonaqueous electrolyte solutions, the most important
s for discussion are the following:
1. Physical properties.
2. Factors that determine solution conductivity and diffusion of the
ions.
3. The electrochemical windows of the solutions.
4. The major electrode reactions of the electrolyte solutions and the
factors that influence them: the nature of the solvent, the cation and
anion used, the level of contamination, and the importance of the
electrode's material.
5. The most important applications and what the directions are for
improvement of existing systems in order to meet current and future
needs.
This chapter is written in light of the above points of interest.
. GENERAL PROPERTIES OF LIQUID

ELECTROLYTE SOLUTIONS
.1. On the Physical Properties of Polar Aprotic

Electrolyte Solutions
6
Table 18.1, prepared by Blomgren, presents physical properties of
est for a number of important polar and nonpolar aprotic solvents.
Solvent
;,:
:;;
1:>..
N,N-diethylacetarnide
N,N-dimethylformamide
Dimethyl sulfoxide
il iS; w
Hexamethylphosphoric triarnide
N-Methyl-2- ii:
pyrrolidinone ;,:
5·
I'yridine "'
Sulfolane
Acetone
Acetonitrile
Benzonitrile
y-Butyrolactone
Diethyl carbonate
Diethyl sulfite
Dimethyl carbonate
Ethyl acetate
Ethylene carbonate
Ethylene glycol sulfite
Methyl acetate
Meth_yl formate
Ul VJ VJ
Nitromethane
ylene carbonate
Trimethyl phosphate
Diethyl ether
Diglyme
1,2-Dimethoxyethane
1,4-Dioxane
1,3-Dioxolane
2-Methyl-tetrahydrofuran
Tetrahydrofuran
Tetrahydropyran
Triglyme
Benzene
Carbon tetrachloride
Chloroform
1,1-Dichloroethane
Methylene chloride
Toluene
Trichloroethylene
Cyclohexane
n-Heptane
n-Hexane
Sulfur dioxide
Sulfuryl chloride
-
-
'1:1
Thionyl chloride
.Oo
dvanced Liquid Electrolyte Solutions 535
itical property of solvents for electrolyte solutions is their

ty to dissolve salts and completely separate the ions from each
r. This ability depends on the solvent polarity. The parameters
ed to this polarity, which are easily available and which appear in
le 18.1, are the dielectric constant and the dipole moment. The
7
ectric constant usually decreases as the temperature increases.
ever, these parameters are not sufficient for determining and ranking the
ty of solvents to form ion-conducting electrolyte solutions. Other
meters had to be invented for that purpose, as summarized in the next
on.
.2. How to Evaluate Solvents for Electrolyte Solutions
The solvating power of a solvent is a fairly complex quantity that

nds on various factors, such as solvent basicity, acidity, polarity,
ture, etc., whose relative contributions may vary considerably with the
trate under consideration. In other words, different reactions may
nd differently to the same changes in a solvent environment. The initial
undamental requirement for the development of a more generally
cable solvent concept is the need to try to separate the various factors
onsible for the solvating power of a solvent. It is important to find
ria for the solvent classification that can be correlated not only to salt
bility and apparent conductivity, but also to the impact of the solvents on
hermodynamics and kinetics of the electrochemical reactions. There are
ral approaches to defining a typical solvent property that can represent
olarity and can be correlated to the thermodynamics and kinetics of
ions conducted in its solutions (i.e., a linear free-energy relationship).
8
comprehensive review of such approaches by C. Reichardt, he divides
into three categories:
1. Comparative equilibrium measurements of reactions in solvents.
2. Comparative spectroscopic measurements based on solvatochromic
effects of solvents.
3. Comparative measurements of reaction rates in various solvents.
Table 18.2, taken from Reichardt's review, describes representative
sponding data from the above three categories. Especially interesting are
9 9
parameters: donor number (DN), acceptor number (AN), and Er30,
h appear in this table.
Refs. 8,10,11 present both the compound, whose solvatochromic effect
es the polarity scale ET30, and the typical absorption in the UV-VIS of
ompound in three different solvents of different polarities (in decreasing
r): ethanol, acetonitrile, and 1-4 dioxane. It was clearly demonstrated
in that as the solvent is less polar, the absorption of the 2,6-diphenyl-4-
6,-triphenyl-1-pyridino) phenolate in the visible, which defines the
Chapter 18
e 18.2. Typical approaches for defining solvent polarity. Taken from Reichardt<8l.
bol Physical quantity measured Solvent-dependent standard process

me) m
ilibrium
surements
esmotropic Equilibrium constant Keto-enol tautomerism equilibrium
tant) of ethyl acetoacetate at ca. 20 °C
o Free energy of the reaction NH/OH tautomerism equilibrium of Schiff
reaction in standard state Bases of pyridoxal-5'-phosphate at 25 oc.
Reaction enthalpy 1:I Adduct formation between antimonyi(V)
nor number) -!!..HEPD-St£1 chloride as standard EOA and EPD solvents
5
in I ,2-dichloroethane at 25 °C.
m surements
action rates
Relative rate constant k 1 SN I Solvolysis of tert-buty! chloride at 25 °C.
Relative rate constant k2 SE2 reaction of tetramethyltin with bromine at 20 oc
Endo-exo product ratio Diels-Alder [.4+,2] cycloaddition of cyclo-
pentadiene to methyl acrylate at 30 °C.
m ctroscol!
ic
surements
Molar transition energy Charge-transfer absorption of 1-ethyl-4-
methoxycarbonylpyridinium at 25 °C.
r30 Molar transition energy n-n• Absorption of pyridinio-phenolate at 25 oc.
Molar transition energy n-n• Absorption of a positively solvatochromic
undecamethine-merocyanine dye at 25 oc
Molar transition energy n-n* Absorption of a negatively solvatochromic
nonamethine-merocyanine dye at 25 oc.
Absorption wave number n-n• Absorption of several compounds particularly
nitro-substituted arenes (e.g. 4-nitro-
anisole (1-methoxy-4-nitro-benzene)).
Equilibrium constant, Mixed parameter, calculated
rate constant, molar from various solvent dependent processes.
transition energy
19 19
Relative F-NMR F-NMR absorption of 1-fluoro-4-nitrosobenzene.
chemical shift
31 31
acceptor Relative P-NMR P-NMR absorption of
ber) chemical shift triethylphosphane oxide.
scale, appears at a lower wave number. The scale itself is defined by the
wing equation:
30 (kcal/mole)= heVL (18.1)

re his Plank's constant, cis the speed of light, fJ is the absorption's

e number, and Lis Avogadro's number. The importance of this solvent
meter lies in the ease of its measurement and the widely successful
rience obtained to date in correlating this parameter to reaction rate
ficients and equilibrium constants in many solvent systems.
9 12 13
The donor number (DN), introduced by Gutmann and Mayer, • •
sents the measure of the donor properties of solvents and is defined as
numerical value of the heat of adduct formation between the donor
cule and the reference acceptor SbC15 in diluted 1,2-dichloroethane
ion:
SbC15 +solvent[SbC1 5·solvent] in 1,2-dichloroethane (18.2)
The values range between zero and about 60 kcal/mole. Polar solvents
as nitromethane are very weak donors. Solvents such as acetonitrile and
ylenecarbonate are moderately strong donors. Ketone, ester, and ether
ents show medium donor properties. Amides and sulfones are strongly
solvents with donor numbers of 25 to 30, and aliphatic amines are very
g donors.
The characterization of the acceptor properties of solvents is a more
cult problem. The definition of an analogous thermochemical quantity is
possible because many solvents contain atoms with lone pair electrons,
hus may undergo adduct formation with strong reference acceptors.
9 12
This problem was solved by Mayer • by the use of triethylphosphine
5 5
oxide (Et3P + --- 0 - ) as a basic reference donor, in a solvent acting as an
ptor.
Triethylphosphine oxide contains a highly basic oxygen atom, which is
y accessible to solvent electrophilic attack. This causes a polarization of
= 0 bond and a downfield shift of the 31P NMR signal. The observed
ical shifts (8) referred to the reference solvent n-hexane, and
polated to infinite dilution, may be taken as a measure of the acceptor
erties of the solvents. The acceptor number (AN) is defined as follows:
AN = _8_(E_t..::...•P_O_in_so_l_ve_n_t_S_)_-_8_(E_t..:;_.P_O_in_h_ex_a_n_e) (18.3)
8(SbC15 ·Et 3 POinC 2 H 4 Cl 2
)
The importance of solvent parameters such as DN and AN and the

ntage of their use over physical-electrostatic parameters was further
14
onstrated by Mayer et al., who studied correlations between the DN
AN of solvents and thermodynamic properties of reactions carried out in
. These include redox potentials, equilibrium constants, and parameters
n pair equilibria. According to the Born theory, redox potentials should
13
nd linearly on the reciprocal of the solvent's dielectric constant.
Chapter 18
2
However, when plotting Ev. values of a redox couple such as Cd/Cd +
E of the solvents, the results provide a very scattered picture. In contrast,
s been clearly shown by Mayer et al. that redox potentials of metals (e.g.,
n2+, Cd/Cd 2+, Eu/Eu 2+) can be nicely correlated to the donor numbers of
solvents, and that the Ev. values measured behave linearly vs. the
ents' DN.9
Table 18.3, taken from Refs. 8 and 9, lists some solvent evaluation
meters, in light of the above discussions and the definitions in Table
, for a number of solvents of interest.
In conclusion, extensive work on solvent properties has revealed that
le physical properties, such as dielectric constant or dipole moment, are
dequate measures for solvent polarity (which can correlate well with the
ence of solvents on thermodynamic and kinetic reaction parameters in
). Better solvent parameters, which correlate well with the impact of the
en solvent on electrochemical and chemical reactions, are donor and
ptor numbers or parameters based on solvatochromic effects, because
reflect not only pure electrostatic effects, but rather the entire electronic
erties of a solvent. We should also mention the efforts to develop tools
18.3. Er30 values, donor numbers (DN), and acceptor numbers (AN) of various
nts. Taken from Refs. 8 and 9.
Solvent Er30 DN AN
c acid 51.2 52.9

ne (AC) 42.2 17.0 12.5
ne 34.5 0.1 8.2
ichloroethane (DCE) 39.4 0.0 16.7
ylamine 35.4 9.4
yl ether 34.6 19.2 3.9
imethylformamide (DMF) 43.8 26.6 16.0
hyl sulfoxide (DMSO) 45 29.8 19.3
ol (ET) 37.9
amide 51.9 24 39.8
xane 56.6 0.0 0.0
l acetate 16.5 10.7
thyl-2-pyrrolidone (NMP) 27.3 13.3
lene carbonate 46.6 15.1 18.3
onitrile 43.6 16.1
yl chloride 0.4
63.1 16.4 54.8
nced Liquid Electrolyte Solutions 539
he prediction of solution properties in polar aprotic solvents based on

e characteristic measurements of thermodynamic properties in reference
ents (e.g., the concept of transfer parameters, G. MI, S of solution, as
15 16
loped by Coetzee, and also dem.onstrated by others ).
2.3. On the Conductivity of Polar Aprotic Solutions and

Diffusion of Ions in Them
The elementary condition for obtaining conductivity in solutions is

ously the dissolution of the electrolyte and considerable charge
ration. The most common and important conduction mechanism relates
he motion of separated ions in the liquid medium, retarded by friction
ent-solute and ion-ion interactions). The ions thus reach a constant drift
d (S) under the electric field (Stock's law 17) .
The mobility of ions in solutions (i.e., the proportion between their
d and the electric field applied to the solution) is proportional (at
1
tant temperature) to their diffusion coefficient (Einstein equation \
c properties of solutions are their specific (X) and molar (A=xJC,
oncentration of the electrolyte) conductivity, resulting from the sum of
ontributions of the various ions. The molar conductivity of solutions
strong electrolytes depends slightly on the concentration ca.JC ). 17 18
'
n there is no full dissociation of the electrolytes in solutions, either in

s of weak electrolytes or solvents of low polarity, we observe a variety of
plicated phenomena, including formation of solvated and unsolvated ion
, triple ions, etc. The behavior of polar aprotic solutions with
plicated inter-ion and ion solvent interactions is well covered in theory
9 0 21
, Fuoss/ Shedlovsky/ and Venkatasetty ).
An important approach to forming conductive nonaqueous solutions is
se of solvent mixtures. There are many important highly polar aprotic
ents whose polarity leads to strong, intermolecular interactions, and thus
gh viscosity. Hence, the advantage of the high electrolyte dissociation
to the high polarity is balanced by the high viscosity, which leads to the
vely low conductivity, since the latter property depends inversely on the
17 18
nt viscosity. ' Combining solvents of low dielectric constant and low
sity with solvents of high dielectric constant but high viscosity may
mixtures of higher conductivity, compared with the single solvent
ms. Typical examples are mixtures of alkyl carbonates and ethers that,
her with Li salts, form highly conductive electrolyte solutions for
m batteries. In most cases, the conductivity of these systems vs. the r
ratio of the solvents has a maximum around equimolar concentrations
e solvents.
Chapter 18
The dependence of the conductivity on the salt concentration may also

omplicated. As the salt concentration increases, the total amount of
ge carriers in the solution increases. However, high salt concentration
s to high viscosity and strong ion interactions, which increase the friction
n mobility under the electric field. Thus, in many cases the dependence
e conductivity of polar aprotic solutions on the salt concentration is a
tion with a maximum.
It should be noted that there are well-established theories that calculate
ffect of salt concentration and temperature on the specific conductivity
6
olar aprotic electrolyte solutions (e.g., a summary by Blom ren, the
22 23 4
eel-Amis equation, Angell et a/., and Barthel and coworkers ) .
A special challenge is the development of electrolyte solutions for low
eratures.Z5 For instance, nonaqueous batteries are required to function at
eratures as low as -40°C. It was found that some ternary and quaternary
ures of solvents, e.g., cyclic or linear alkyl carbonates (EC, PC, DMC,
, EMC) and esters (MF, methyl propionate) with salts such as LiPF6,
F4, etc., can function as electrolyte solutions for Li-ion batteries at very
temperatures (<-30°C).25 However, when preparing electrolyte solutions
batteries, one should take into account that the most important factor
ed to the battery performance is the stability of the electrode-
% 2Me-THF (vol)
0 20 40 60 80 100
10
e 8 •
'
Q
-=e LiAsF6(DN)
e 6 IMLiA5F6
in 2Me THF/PC
'\.
•
2
u LiAsF6(PC)
L iC 104(DN)
0
0 0.5 I 1.5 2
M/Li SALT.
e 18.2. Examples of the variation in the conductivity of polar aprotic solutions as a

on of solvent ratio and salt concentration. LiAsF6 and LiC104 solutions in PC and DN,
iAsF6 solutions in mixtures of PC and 2MeTHF, as indicated. Refs. 26 and 27.
ion interfaces. Hence, some esters that can have a positive impact on the
uctivity may interfere badly with the passivation of Li or Li-C anodes.
, the choice of solvent mixtures for low temperature applications
ires a rigorous study of the electrodes' surface chemistry and stability in
mixtures at ambient and high temperatures. Figure 18.2 shows typical
ples of the behavior of polar aprotic solutions showing dependence of
onductivity on the salt concentration, and on the solvent ratio, in binary
ions containing low dielectric constant-low viscosity and high
26 27
ctric constant-high viscosity solvents (e.g., 2Me-THF and PC). '
28
Table 18.4 shows typical parameters related to the conductivity of
ted polar aprotic solutions and demonstrates the effects of salts,
ents, solvent mixtures, and temperature. There are many publications
ng with the conductivity of polar aprotic solutions. Typical examples are
. 29-32.
. ELECTROCHEMICAL WINDOWS OF
NONAQUEOUS ELECTROLYTE SOLUTIONS
In order to review briefly the electrochemical stability of nonaqueous

rolyte solutions, classification of the electrode types used is very
rtant. The relevant electrodes may be divided into the following
ories.
Noble metal electrodes and special carbon electrodes

Noble metal electrodes include metals (M) whose redox couple M/M'+ t
involved in direct electrochemical reactions in all nonaqueous systems
nterest. Typical examples that are the most important, practically
king, are gold, platinum, and glassy carbon. Conductive-doped diamond
33
rodes also belong to this category, since the carbon material in these
is not involved in redox reactions.
Reactive metal electrodes
Reactive electrodes refer mostly to metals from the alkaline (e.g., Li,
and the alkaline earth (e.g., Ca, Mg) groups. These metals may react
taneously with most of the nonaqueous polar solvents, salt anions
ining elements in a high oxidation state (e.g., Cl04-, AsF6·, PF6-,
F3") and atmospheric components (02, C02, H20, N2). Note that all the
Salt Solvents
liPF6 ECJPC
2Me-TIIF/ECJPC
liAsF, ECIDME
PCIDME
2Me-TIIF/ECJP
liCF,SO, ECJPC
DMEIPC DMEIP
2Me-TIIF/ECJP
liN(CF1SOJ2 ECJPC
ECIDME PCID
2Me-1HF/ECJP
liBF. ECJPC
2Me-1HF/ECIP
ECIDMC ECID
'1:1
"....
01)
uao.
a.
LiPF,
s:
! :!.
a
LiAsF,
c;·
I i!
LiCF,SO,
LiN(CF,S
LiBF4
VI
" ""
U l
Chapter /8
solvents have groups that may contain C-0, C-S, C-N, C-CI, C-F, S-0,
, etc. These bonds can be attacked by active metals to form ionic species,
hus the electrode-solution reactions may produce reduction products that
more thermodynamically stable than the mother solution components and
h less soluble. Consequently, active metals in nonaqueous systems are
34
ys covered by surface films.
When introduced to the solutions, active metals are usually already
red by 'native' films (formed by reactions with atmospheric species),
then these initial layers are replaced by surface species formed by the
5
ction of solution components? In most of these cases, the open circuit
ntials of these metals reflect the potential of the MIMX/Mz+ half-cell,
e MX refers to the metal salts/oxide/hydroxide/carbonates which
prise the surface films. The potential of these half-cells may be close to
of the M!Mz+ couple.
At potentials positive to the OCV, the electrochemical window may be
limited because dissolution of the active metal takes place either via
ation of Mz+ ions through the surface films or by the breakdown of the
ce films, exposing the active metal to the solution and enabling a 'direct'
6
lution process? At potentials lower than the OCV, the situation is more
plicated. Since there is no possibility of a cathodic breakdown
anism of the surface films, the possibility of obtaining active metal
sition depends either on the feasibility of migration of the M'+ ion
gh the surface films, or on electron tunneling through them to form
l deposits outside the surface layer. Consequently, in the case of alkaline
metals (Ca, Mg, etc.) whose ions are bivalent, and whose migration
gh thin layers of their salts is difficult (if not impossible), there may be
37
ossibility of metal deposition. '38 In such cases, the cathodic limit of the
rochemical window may be determined by reduction reactions of
ions components. In the case of alkaline metals such as lithium, the
rochemical window is very narrow because Li dissolution and
sition take place above and below the OCV at very low overpotentials,
39
ions migrate through the surface films that cover Li electrodes.
Non-active/slightly reactive metal electrodes
Nonactive/slightly reactive electrode materials include metals whose

ivity towards the solution components is much lower as compared with
e metals, and thus, there are no spontaneous reactions between them and
olution species. On the other hand, they are not noble, and therefore their
ic dissolution may be the positive limit of the electrochemical windows
any nonaqueous solutions. Typical examples are mercury, silver, nickel,
er, etc. It is possible to add to this list both aluminum and iron, which
react by themselves spontaneously with nonaqueous solvent
40
cules or salt anions containing central atoms of high oxidation states.
ever, they are not reactive due to passivation phenomena. Hence, the
rochemical window of nonaqueous solutions with electrodes such as
inum or stainless steel may be determined by the electrochemical
41
ion of the solution components.
A typical example is aluminum, which is used as a current collector for
odes in lithium batteries. The stability of aluminum in many Li salt
ions at potentials as high as 4.5 V vs. Li/Lt is due to the formation of
ly insoluble AI halides on the aluminum surface, which remain stable
42
hus protect this active metal from corrosion.
Intercalation electrodes
Intercalation electrodes may include either carbon electrodes or

ition metal oxide and sulfide electrodes, which intercalate with solution
ponents such as metal cations. The most important example is
calation of lithium ions into many types of carbonaceous materials at
potentials (0-1.5 V vs. Li/Li+) 43 and transition metal oxides and sulfides
44
gh potentials (2.5-4.5 V vs. Li/Li+) • These electrochemical intercalation
ions are the basis for rechargeable Li ion battery systems and are usually
imiting anodic reactions in the Li salt solutions/LixMOy or LixMSy
ms.
Red-ox electrodes
Redox electrode materials mostly include electronically conducting

mers such as polypyrrole, polythiophene, or polyaniline that may
rgo reversible oxidation-reduction reactions accompanied by doping of
45
olymer with the counter-ion that balances the charge. These processes
he anodic limit for these electrodes in many nonaqueous systems. In
cases, such as conductive polythiophene or polyparaphenylene
rodes, it is also possible to dope the polymer at low potentials and thus
ge it negatively, when this negative doping is charge balanced by the
nce of cations into the polymer's matrix.45 The electrochemical window
ese systems may be determined solely by the redox activity of the
rode materials, which may leave a wide range of potentials between the
processes in which the electrode is electrochemically inert.
.1. On Cathodic Reactions of Polar Aprotic Electrolyte

Solutions
The cation used in the electrolyte plays an important role in

Chapter 18
mining the cathodic behavior of polar aprotic solutions. Figure 18.3

s typical cyclic voltammograms of ether (THF), alkyl carbonate (PC),
ester (BL) solutions with tetrabutylammonium perchlorate (TBAC104)
gold electrodes, and a CV of a Li salt in BL with a gold working
46
rode (WE). The comparison among the CVs in Figure 18.3 is highly
rtant because it demonstrates very clearly the effect of the salt cation on
lectrochemical behavior of these solutions. In TBA salt solutions, as
as in any nonaqueous tetraalkylammonium salt solutions, there is no
vation of non-active electrodes. The high cathodic currents seen for the
ant CVs in Figure 18.3 reflect massive reduction of solution species. In
case of ethers, TBA is reduced. When the solvents are alkyl
I 30JLAICiT?
Li + SaII Solution
o.v
o+ ,.....
---, ,.----.
I
\ I
I
/ 1/
I
,...-'\ I
0. / vi DME or THF
O'(F;t;;z;;;::::===P=C::=::==:;::: :::
I ,f o+\1AVr:..::::==BL=. =::::=::::::::::::=:: _
I '
t
11 /' TBA Solt
1 I
I I I
t, l
I
ov
,i 2V 3V
4l/
18.3. Steady state cyclic voltammograms obtained with four electrolyte solutions. The
lower ones were obtained from 0.2 M TBAP solutions with THF, BL and PC, as
ted. The upper one was obtained from 0.2 M LiAsF6 or LiC104 solutions in BL. The
ial scale refers to the LilLi+ 0.1 M reference electrode. The potential sweep rate was 20
The current scale is common to all CVs. Solid lines: dry solutions (-15-30 ppm of
dashed line: water contaminated solutions (0.001 M). The potential scan in all cases
from open circuit potential (2.7-3 V in these systems). Ref. 46.
onates or esters, which are more reactive than ethers, the major cathodic
46
ions in TBA salt solutions relate to solvent reduction.
When the cations belong to the alkaline metals (Li, Na, K), reduction
ace oxygen, water, C02 solvents, and salt anions form insoluble oxides,
oxides, carbonates, halides, and organic salts, which passivate the
rodes and form surface films that block massive reduction of solution
47
ies. -53 Hence, a first polarization of a non-active electrode in a polar
tic Li (or Na) salt solution from open circuit potentials (usually around 3
. Li/Li+) to low potentials, is characterized by irreversible cathodic peaks
nd 2 V and 1.5 V (corresponding to traces of 0254 and H2055 reduction)
rimposed on a broad cathodic wave related to reduction of solvent
46 56
cules and salt anions, which form surface films. -
The steady state voltammetric response of Li (and Na) salt solutions
non-active electrodes includes mostly underpotential deposition and
ping of the active metal on/from the noble metal electrode, possible alloy
ation and decomposition (active metal + non-active metal), and bulk
sition-dissolution of the active metal. As discussed in detail and well
lished in the literature, UPD processes and alloy formation depend on
46 51 52
nature of both the host (substrate) and the deposited metals. ' •
mes 18.1-3 show reaction paths of alkyl carbonates, commonly used salt
ns and contaminants, respectively, in the presence of Li-ions with Li
l, lithiated carbon and/or non-active metals and carbons at low
ntials, as well as the product distribution, based on Refs. 26, 27, 46-53,
7.
Finally, we mention the fundamental basic cathodic reactions of the
most important polar aprotic inorganic solvents, thionyl chloride and
(due to their application for Li batteries\ as listed below:
2 SOC}z +4e- S0
SOz + 2e- Szol-
.2. Oxidation Reactions in Polar Aprotic Electrolyte

Solutions
2.1. Introduction
The study of the oxidation reactions of polar aprotic solutions and

Chapter 18
me 18.1: Possible reduction patterns of alkyl carbonate on Li (Refs. 27, 49,

9, 61, 67)
PC+ e·+ u.-+ ------.. CH3CHCH20Ca 2·u.-+(f>C-I./) supported
. ?
CH3CHCH 20C02Li + H - CH3CH2CH20C02Li. not supported
? . d
. 0C0 Li. (CHrCH-CH 0C0 Li )z not
2CH3
CHCH2 2 2 2 supporte
2CH3CHCH2 0C 02Li CH3CH(OC02 Li)CH 20C02Li+ CH3CH=CH 2 t supported
2EC ze". ZLt (CH20C02L!)2 + CH2 = CH2 t supported
not supported
?
l..i.zO + EC ---r...iOCH 2CH2 0C02Li.
ucleophilic Mechanism (relevant to PC as well):
_ _, CH 2 -0, 1
_.+ .
RO --.../" I .--C=O ------.. ROCH2-CH2-0C02Li pooriiDcleophile
CH2 -0
secood
RO-yH2
nucleC1'hilic
.,ock &
II nucleophile
CH2 -0COz Li ROCH 2CH20R + 1
C,
Q,r\Q
Ro-.J 8" 8"
Li'"
LizC03.. ionpairing
competition
IIDCleophilic
yHz-OCOzlit attack
CH2 -0C02li
(Relevant to PC as well)
---
(In a similar manner other acvclic alkvl carbonates: DRC.. RMC.. PMC. reacts).
. 2ROC0
2 Ii+ H20
-
HF
R- Li.
R·+ Li.0 ---
vanced Liquid Electrolyte Solutions 549
heme 18.2: Surface reactions of commonly used Li salts (Refs. 26, 49, 50, 53,
, 58, 61-67)
LiAsF6 + 2Li+ + 2e" 3LiF J, + AsF3 (sol)
AsF3 + 2xLi+ + 2xe-LixAsF3-x J, + xLiF i

PF6- + 3Li+ + 2e"3LiF J, + PF3
LiPF6 ..====""' LiF + PF5

PFs + HzO PF30 + 2HF
PFs + 2xLi+ + 2xe-LixPFs-x J, + xLiF J,
PF30 + 2xLi+ + 2xe- LixPF3_xO J, + xLiF J,
L'+
1 -
' e LiF J-, LixBFy J, (in general)
BF4-
LiCI04 + 8Li+ + 8e-4Liz0 + LiCI

LiCI04 + xLi+ + xe-LiCIOc4-v,x) + YlxLizO. (x = 2, 4, 6)
LiN(SOzCFJ)z + ne- + nLi+ Li3N + LizSz04 + LiF + CzFxLiy
LiN(SOzCF3)z + 2e- + 2Li+ LizNSOzCF3 + CF3SOzLi
LizSz04 + lOe- + lOLi+ 2LizS + 4Liz0
LiC(SOzCF3)3 + 2e- + 2Li+ LizC(SOzCF3)z + LiSOzCF3, etc.
LizSz04 + 4e- + 4Li+ LizS03 + LizS + LizO
2LiS03CF3 + 2Li+ + 2e- 2LizS03 + CzF6
R-CF3 + 2Li+ +2e"RCFzLi + LiF
LizS03 + 6Li+ +6e- LizS + 3Liz0
heme 18.3: Reaction patterns of common contaminants (Refs. 26, 47, 48, 50,
53-55, 61, 62, 64-66)
COz + e- + u+ COzLi
COzLi + COz0 = C-0-COzLi
O=C-0-COzLi + e- + Li+ CO i + LizC03 J,
2LiOH J-+ COzLizC03 J, +

HzO LizO J, + COzLizC03 J,
ROLi J, + COzROCOzLi J,
HzO + e- + u+ LiOH + Y2 Hz
LiOH + Li+ +e- LizO + Y2 Hz
H + e- + Li+ LiH
Chapter 18
N2 + 6e·+ 6Li+2Li 3N
02 + e·+ u+Li02
Li02 + e- + Li+Li202
Li 202 + 2e·+ 2Li+2Li 20
HF + e·+Li+LiF + Y2 H2
D of solutions with high anodic stability have become very important in

nt years due to the R&D of Li-ion batteries with cathode redox potentials
igh as 5 V. In recent years we have seen intensive efforts to study
rously the oxidation reactions of polar aprotic electrolyte solutions,
41
ortant in the field of Li batteries. '68 -71 These have included mostly
ures of alkyl carbonates and commonly used Li salts. Of especial
ortance for these studies has been the application of spectroscopic tools
70 71
as in situ FfiR spectroscopy and mass spectrometry. ' In general, the
ification of the specific oxidation reactions of nonaqueous solutions is h
more difficult than the identification of the cathodic reactions of the e
solutions. This is due to the fact that oxidation of most of the polar tic
solutions usually forms species that are soluble in the mother tions
and cannot be easily separated from them for identification. In rast,
cathodic reactions of polar aprotic solutions usually form gases and ace
films which can be identified by spectroscopic means, as described in
previous section, as well as described in the relevant references listed
ein. In the case of ethers and alkyl carbonate solutions, oxidation forms
71
ble polymers. '72 The solutions whose oxidation was studied most
68 75
rously were alkyl carbonates and ethereal systems. -
With noble metal electrodes, small scale oxidation processes of alkyl
71
onate solutions may be detected at potentials above 3.5 V (Li/Li+),
41
e massive, intensive oxidation processes occur at potentials >4.5 V •
During oxidation of alkyl carbonates, evolution of C02 was clearly
cted by DEMS. Ring opening and a change in the nature of the carbonyl
p as an intermediate step were suggested, following in situ FfiR
76 5
surements. As shown by Eggert and Heitbaum in DEMS studies/
n an LiC104 electrolyte is used, the anion is also involved in the
ation process of the system at the 4-4.5 V potential range, and
mediates such as Cl02 and radical or molecular oxygen are then formed.
es of water were also identified (by DEMS) among the oxidation
ucts of a PC-LiC104 solution. This product may be formed through
ogen abstraction by an oxygen radical from the solvent molecules.
In the case of BF4- or PF6- salts , the oxidation products also include
inated derivatives of propylene and/or propylene oxide (in the case of

solutions).76 Another solvent whose electrochemical oxidation was
ously investigated was 1-3 dioxolane (DN).77 It was found that when
ketal is diluted with other ethers, it exhibits high anodic stability, similar
HF (4.5 V >Eox> 4 V vs. Li/Lt). However, since this solvent is a ketal, it
ghly sensitive to Lewis acids and thus can polymerize at potentials as
as 3 V in the single solvent solutions of salts such as LiC104• At these
ntials, adsorbed hydrogen on the electrode (formed by trace water
ction) is oxidized to H+, which readily initiates polymerization of
olane.
.2.2. On the Role of the Electrode's Material
The electrode's material has a high impact on the anodic stability of

rochemical systems based on polar aprotic solutions, as summarized
w:
. It is only with noble metals such as Au and Pt and doped
(conducting) diamond that the electrode's material is not involved in
the anodic reactions of the electrochemical systems.
. With electrodes such as Ag, Ca, Ni, etc., metal dissolution may be
the major anodic process that limits the electrochemical window of
the system.
. Graphite electrodes can intercalate with commonly used anions at
potentials >4 V (Li/Ln. Such an intercalation process, when
possible, becomes the limiting anodic reaction.78
. In the case of transition metal oxides or sulfide electrodes and Li salt
solutions, the limiting anodic process is, of course, Li insertion into
the electrode's active mass. It is important to note that we have
evidence that massive oxidation of solution species is largely
inhibited with commonly used transition metal oxide cathodes, even
at high potentials (>4.5V), due to passivation phenomena.79
. With active metal electrodes such as Li, Na, Mg, and Ca, the limiting
anodic reactions are, of course, dissolution of the active metal.
However, since these electrodes are always covered by surface films,
their processes are complicated. The surface films on Li or Na may
conduct the monovalent cations and, thus, Li and Na
dissolution occurs at low overpotentials. 52
. Aluminum, which is also an active metal, demonstrates high anodic
stability in most of the commonly used polar aprotic solutions, due to
passivation phenomena (e.g., by AI halides). Therefore, the metal is
widely used as a current collector for cathodes in Li and Li-ion
batteries.42
Chapter 18
.3. A Short Summary of the Anodic Limit of

Nonaqueous Electrolyte Solutions
Table 18.5 provides oxidation potentials of a number of Li salt

ions with three types of electrodes: Pt, glassy carbon and composite
80
on black. An expected correlation between the donor number of the
ent and its oxidation potential was81 found: the higher the donor number,
ower is the oxidation potential. Further comprehensive data on the
rochemical windows of polar aprotic solutions can be found in Refs.
3.
Some conclusions related to the data in Refs. 80-83 are listed below.
1. In general, the determination of oxidation potentials is not trivial,
since all of the relevant reactions are irreversible. Therefore,
oxidation potentials may relate to arbitrary decisions; normally, the
potential is determined at a point where the anodic current of a
system is significantly high.
2. Some of the early data in the literature refer to aqueous SCE or
85
Ag!Ag+ reference electrodes, whereas more recent data, such as that
presented in Table 18.5, refer to the non-aqueous LilLi+84reference
electrode. However, as shown by Auborn and Ciemieki, the SCE
electrode potential is 3.24 V vs. Li/Li+ and thus data from different
sources can be compared. It appears that the early data provided by
85
Mann relating to aqueous reference electrodes correspond well
to later data relating to non-aqueous reference electrodes (for the
same solvents and salts).
3. Alkyl carbonates have a much higher anodic stability than do ethers
(>1 V difference).
4. The anion and the electrode material have a pronounced effect on
the oxidation potentials of the non-aqueous systems. The metals to
which the highest potentials can be applied in non-aqueous systems
are obviously the noble metals (Pt, Au). The limiting reaction when
the anions are halides (Cr,Bf, r) was found to be their oxidation to
the elemental form. When the anion is Cl04-, its oxidation onset at
potentials above 1.5 vs. Ag!Ag+ may promote further intensive
solvent degradation, as in the case of AN. It is important to note that
using BF4- instead of Cl04- in AN (which is an important and useful
non-aqueous solvent in electrochemistry) extends its anodic stability
by 2 V. Similar results were obtained from Pt and glassy carbon
electrodes, indicating that no catalytic effects take 82 place in the
electrochemical oxidation of polar aprotic solutions.
5. It appears that acetonitrile is one of the polar aprotic solvents of the
highest anodic stability.
6. The order of anodic stability found for commonly used Li salts is

LiAsF6, LiBF4, LiPF6 > LiC104, LiS03CF3, LiN(S02CF3h.
4. LIQUID ELECTROLYTE SOLUTIONS FOR

RECHARGEABLE Li (METAL) BATTERIES
4.1. Problems with the Use of Li Metal Anodes
Li surfaces are naturally covered by bilayer surface films. The inner

is comprised mostly of LizO, while the outer part is comprised of a
ure of LiOH and Li 2C03. When Li electrodes are introduced into
rolyte solutions, the following changes take place.
1. Solvent molecules can percolate through the surface films and react
with the active metal.
2. Species such as LiOH, LhN, Li20, and even Li carbonate, which are
part of the native passive layer, may be highly nucleophilic in aprotic
media and react with electrophilic solvent molecules such as esters
48
and alkyl carbonates. '86 For instance:
LizO + y-BLLiO(CHz)3COOLi (18.6)

LiOH +PCCH3CH(OC0 2Li)CH20H (18.7)
Li3N + ECLi 2NCH2CH20C02Li (18.8)
3. Part of the pristine surface species and the products of the

nucleophilic reactions above dissolve in the solution, thus allowing
other solution species to react with the active metal.
4. Traces of water may hydrate most of the surface species formed on
Li, and hence, diffuse through the surface films and react with the
87
active metal to form LiOH, Li20, LiH and H2.
5. Consequently, the surface films become very non-uniform at the
micrometric and nanometric scales (both laterally and vertically).
6. When fresh Li is exposed to a solution, there is an immediate reaction
with solution species under highly non-selective conditions. The
solution reduction products precipitate and form an initial surface
layer through which electron transfer to solution species continues at
an increasing level of selectivity.
As the film reaches a certain thickness, it blocks electron transfer and
lectrode reaches passivation. Further reduction of surface species may
nue within the surface films. In addition, at the outer solution side of the
, there is a continuous dissolution-precipitation of surface species,
ndary reactions of surface species with solution components, and highly
Chapter 18
18.5. Oxidation potentials (V vs. Li/Lt) for several electrolyte solutions. Taken from
128 and Refs. therein.
vent Salt Pt GC" TABb
LiCl04 5.3 5.3 4.7

DME (90/10) LiCl04 4.5 4.5
DME (75/25) LiC104 4.5 4.5
DME (50150) LiCl04 4.6 4.7 4.6
E LiCl04 4.5 4.5 4.6
e-THF/THF/2-MeF"·a (49/49/2) LiCl04 4.1 4.0
e-THF/THF (50150) LiCl04 4.1 4.0
e-THF/THF (90110) LiC104 3.9 4.0
e-THF LiC104 4.0 4.2
F LiC104 4.2
LiCl04 LiAsF6 5.6 5.2
DME (90110) LiAsF6 4.9 4.9
DME (75/25) LiAsF6 4.7 4.8
DME (50/50) LiAsF6 4.7 4.7
E LiAsF6 4.6 4.6
e-THF/THF/2-MeF (49/49/2) LiAsF6 4.0 3.9 4.0
-THF/THF (90/10) LiAsF6 4.0 4.1
-THF/THF (75/25) LiAsF6 4.0 4.0
-THF/THF (50150) LiAsF6 4.0 4.1
-THF LiAsF6 4.1 4.1
F LiAsF6 4.2 4.2
-F(50/50) - LiAsF6 •5.0c 4.9 do
... LiAsF6 4.9
- glassy carbon, TAB - Teflomzed acetylene black., MF = methyl formate, 2MeF = 2-
l furan.
tive reduction of solution species by electron tunneling through the

ce films at certain locations of high electrical conductivity. The result of
description is, again, highly non-uniform, multilayer films that comprise
mpact, inner part, which contains species of very low oxidation states
O, Li-C, Li 3N, LiH, Li halides, etc.), a middle part that contains surface
ies of higher oxidation states (ROC02Li, LizC03, ROLi, LiOH,
OLi, etc.), and a porous outer part of the surface films' solution
88 92
face. • The surface species formed on Li, which are all Li salts,
93
pitate as thin films (submicronic thickness ) that can conduct Li ions.
efore, when an electrical field is applied, the surface films formed on Li
39
y polar aprotic solution behave like a solid electrolyte interphase (see
famous, well-established SEI model in Ref. 94)
ter 18 555
Hence, a general property of Li electrodes in solutions based on

ents such as esters, ethers, alkyl carbonates, DMSO, S02, and SOCiz, is
on a thermodynamic basis, the Li electrodes are unstable because all
ction processes of polar aprotic solvents by Li are thermodynamically
rable. However, kinetically speaking, Li electrodes are apparently stable
to passivation by the surface films. In most cases, the surface films are
n conductive and, therefore, Li deposition-dissolution in polar aprotic
ions can be driven easily at relatively low overpotentials. However, as
ribed above, the surface films are highly non-uniform and, thus, the
nt distribution of Li deposition-dissolution processes is very non
rm. Thereby, in most polar aprotic solutions, Li dissolution leads to a
down and repair of the surface films and forms holes and cracks that
(a)
------,;;--- .
----. . ....
• ,., .
..
" ' .,, .
.. :
.t
,f ' !
' · I'
-'·:"' ,
· ..
18.4. Non-uniform Li deposition and dissolution in polar aprotic solutions. (a) A

l AFM image of non-uniform Li dissolution (5x5 1-1-m, 122 nm height difference) in a
PF6 solution. Ref. 96. (b) A typical AFM image of non-uniform Li deposition (25x25
30 nm height difference) in an EC-DMC LiC(S02CF3h solution. Ref. 97.
Chapter 18
se fresh Li to solution, and Li deposition is dendritic. Figure 18.4

ents AFM images that demonstrate typical, non-uniform Li deposition
dissolution processes.95 -97
Dendritic Li deposition and non-uniform Li dissolution is a major
lem in the use of Li metal anodes in rechargeable batteries when using
of the commonly available polar aprotic solutions. There are only a few
rolyte solutions in which Li deposition/dissolution is sufficiently
orm for battery application.
4.2. On the Choice of Electrolyte Solutions for

Rechargeable Li (Metal) Batteries
In the early stages of R&D of Li batteries it already became clear that

98
eposition is dendritic in most of the relevant polar aprotic solutions. It
discovered that LiAsF6 is a superior salt for the reversibility and high
ng efficiency of Li electrodes.99 It was suggested that reduction of the
- anion or species such as AsF3 or AsF5 in ethereal solutions form,
her with ether reduction products, polymeric films that improve the
vation of lithium anodes (the "brown film" concept). 100 Later on, it was
0
ested that reduction of AsF6- forms a variety of species: LiF, Li3As, As ,
57 77 101
sFy, all of which dope the Li surface and reduce its reactivity. • •
Over the years, a large variety of electrolyte solutions containing
rent solvents, salts and additives were tested in connection with the
ng efficiency of Li electrodes.102-104 The solvents included cyclic
s,105 esters/ 06 alkyl carbonates,107 S02 108 and sulfolane.109 The salts
ded LiC104, LiAsF6, LiBF4, LiN(S02CF3h and LiC(S02CF3)3, and the
tives included active gases (e.g., C02),50 metal cations (e.g., Ga3 +,
110 111
+), and organic species (alkanes, aromatic compounds). It was found
modifying lithium surface films with very small quantities of Li 112 or a
3N
nd metal such as alurninum 113 may improve its performance as an anode
rial. Also, application of pressure to Li electrodes (e.g., using compact
rochemical cells in which the separator presses on the Li electrodes),
resses dendritic Li deposition and increases Li cycling efficiency.114
It became clear that the major problem of Li electrodes relates to the
that the surface films formed on Li in most of the commonly used
rolyte solutions cannot accommodate the volume and morphological
ges of the Li metal during dissolution or deposition, as illustrated in
115
re 18.5. In cases where the surface films are comprised mostly of Li
they are not elastic enough to protect the active metal in conditions of
uniform Li dissolution or deposition. Only in cases where the surface
s contain elastomers, as is the case of 1-3 dioxolane solutions, the surface

s are elastic and provide good passivation to the Li electrodes in the
mic situation of Li electrochemical deposition/dissolution cycling (see
re 18.5).77'n 5
Highly important and critical in the use of Li metal anodes in
argeable batteries are safety considerations. Abuse cases, such as short
it of Li batteries, which means an immediate, very high current flow,
to a pronounced heat dissipation. A rapid release of heat inside a lithium
ry can drive highly dangerous, exothermic redox reactions between Li
the solution components. The lithium in the battery can very easily reach
elting point, which should accelerate the thermal reactions in the cell.
a scenario is called "thermal runaway" and results in the explosion of
116
ed Li batteries. As a Li battery is cycled, the Li surface may be
her, and hence, internal thermal reaction inside the battery in abuse cases
ld be more intensive and dangerous. Dangerous "thermal runaway"
tions of Li batteries can also occur upon overcharging and
116
discharging of secondary Li batteries. Hence, even in cases where the
rmance of Li batteries is excellent in terms of cycling efficiency and
and temperature range of operation, the above safety considerations
be a critical limitation in their practical use.
Over the years, there were three electrolyte solutions that were
idered as suitable for the practical development of rechargeable Li
al) batteries, and each deserves special discussion.
117
2.1. EC-PC/LiAsF6, MOLl Energy Inc. Li-MoS2 Cells
With this electrolyte solution, the Li passivation is dominated by the

50
eduction products (e.g., CH20C02Li)z) plus doping by the reduction
ucts of AsF6-, as explained above. We attribute the possibility to obtain
cient Li cycling efficiency for practical battery application in this
on to the stack pressure applied to the Li anodes in the AA-"jelly
4
d" configuration of the batteries produced by MOLl Inc.n However,
Ll's Li-MoSz batteries could not be fully commercialized due to safety
ems.n 7
2.2. THF, 2Me-THF, 2-Methyl Furane/LiAsF6ns.ng
This solution was also found to be very good for reversible, high
ng efficiency of Li anodes. While relatively high cycling efficiency
Chapter 18
Lithium in alkyl carbonates solutions

sahalioJl
sahalioJl
ROCOz
surface films
Li in 1-3 Dioxolane solutions
e 18.5. A schematic illustration of surface film formation on lithium electrodes in alkyl

nates and in 1-3 dioxolane solutions. Ref. 115.
d be measured with Li anodes in 2Me-THF/LiAsF6,120 the combination

e three-solvent mixture (2Me-furan in a few percent by volume) was d
to be superior. The identified surface species formed on Li in these
tions include Li-alkoxides, which are the reduction products of THF and
-THF, and salt anion reduction products (LiF, LixAsFy, possibly
121
ding Li 3As and As\ 2Me-furan may be reduced on the Li surface to
polymeric species. It also acts as a scavenger for trace Lewis acids in
ions. Another component in the surface films on Li in this system may
100
e brown film (As-O-R) cross-linked polymer.
JPL (Pasadena, USA) developed prototypes of AA rechargeable Li
122
batteries with these electrolyte solutions. In spite of their high energy
ity, these batteries could not be commercialized, due to safety
ations, as explained above, and the possible danger in the use of such
ries in abuse cases.
77
.2.3. 1-3 Dioxolane/LiAsFJTributylamine
From early studies of secondary Li battery systems, 1-3 dioxolane

) was considered to be a "magic" solvent in which Li anodes behave
ly reversibly. The morphology of Li deposition in ON solutions was
123
d to be very smooth and dendrite free. The surface films formed on Li
is solvent include Li formate, alkoxides such as CH3CH20CH20Li and
omers of polyDN with -OLi edge groups. We believe that the presence
hese oligomers makes the surface films elastic, and thereby, enhances
115
passivating properties.
When the salt is LiAsF6, doping of Li surfaces by the various possible
sFy products further improves the passivation of the Li electrodes.
ever, DN, which is a ketal, readily polymerizes in the presence of trace
is acids such as AsF3 (which may be unavoidably present in LiAsF6
tions). Therefore, it is necessary to stabilize the DN/LiAsF6 solution with
77
se such as three-butyl amine (TBA). In addition to the excellent cycling
iency of Li electrodes in these solutions, this electrolyte system also
ides a unique, intrinsic safety mechanism for Li batteries, that works as
124
ws. Up to l20°C, Li with DN/LiAsFJTBA is stable, and Li is fully
ivated. At temperatures above l20°C, DN reacts with LiAsF6 (a thermal
x reaction). One of the products of the reaction is HF. As it is formed, it s
to a very fast polymerization of the solvent, which shuts down the
rochemical system because of the relatively low electrical conductivity e
solid-gel matrix in which polyDN is a major component. Commercial
i0.3Mn02 AA batteries containing this solution were produced and
125
ibuted. In any case of short circuit, high currents flow and heat up the
ry so that its internal temperature exceeds l20°C, solvent (DN)
merization occurs, as explained above, and the battery shuts down before
mal runaway takes place (see Figure 18.6a). Upon overcharging, the
77
stabilizer is electrolyzed (>3.6 V vs. Li/Lt), trace Lewis acids are
ed, the solvent is polymerized by the acidic species thus formed, and the
ry again shuts down before any thermal runaway can occur (see Figure
b). These batteries possess high energy density (>140 Wh/Kg, practical),
power density upon discharge, a wide temperature range of operation
°C60°C), and extended cycle life (300-400 cycles at 100% DOD).
A major disadvantage of these batteries has been their requirement for
charging (C/9-12 hrs). At too high charging rates, Li is deposited in
l grains, which facilitate side reactions between Li deposits and solution
ies (no dendrite formation and internal shorts by Li deposits). The cycle
ecreases as the charging rate increases, because the electrolyte solution
126
pears by continuous reaction with Li deposits. Tadiran Inc. gave up
mass production of these batteries because they could not compete with
n battery technology that allows for very fast charging rates with no low
e life penalty.
While there are reports from time to time on research on rechargeable
etal battery systems, we have come to the conclusion that there is no
e in secondary batteries with Li metal anodes and liquid electrolyte
Chapter 18
24 ..... ·············-·············......······-. .. 110

Sri
<
J"
Ill
90
s
..
i
..
c 12 60
ra: E
30 1-<
6
100 ISO 100

a. Time min
12 --------------·---·--··········· ................. 180
10 150 u
..
> 8 120 :!'
c
>
g
!! c.
4
90 t E
60
1 30
0 llO 0
0 60 Time, min 180
b.
e 18.6.(a) Short-circuit behavior of practical AA LiiDN, LiAsF 6, TBAILiMn02 batteries

0 cycles. (b) Overcharged behavior of the same cells at a current of 1 A after 60 cycles.
25.
ions based on the current state-of-the-art technology. There is no

to avoid side reactions between any liquid electrolyte solution and I1
sits (even in the absence of dendrite formation), especially at high
ging rates.127 I1 metal anodes, however, can be used in rechargeable
128
ries with polymeric/solid or gel electrolytes (a full discussion of which
yond the scope of this chapter).
. LIQUID ELECTROLYTE SOLUTIONS FOR

RECHARGEABLE LI-ION BATTERIES
.1. The Requirements of Suitable Solutions for Li-Ion

Batteries
In Il-ion batteries, I1 metal anodes are replaced by U-C intercalation
ounds. Il-ion batteries use high voltage cathodes: I1Ni0 2, I1Co02,
n 204, etc., whose redox potentials are above 4 V.129 I1Mn204 doped with
130 131
ls such as Cr and Cu have redox activity around 5 V. ' Hence, the
rolyte solutions for these batteries should have high anodic stability. The
ic side, i.e., the Li-C/solution interface, also requires special properties
e electrolyte solutions in terms of their low potential surface chemistry.
So far, the most commonly used anode materials for Li-ion batteries are
hitic carbons. All graphitic carbons have a weak lattice comprising
hene planes bound together in layered structures by relatively weak
s. Graphitic lattices can be easily destroyed during intercalation of
132
m by phenomena such as co-intercalation of solvent molecules, and a
-up of internal pressure due to surface reactions that form gaseous
133
ucts. Hence, the stability and reversibility of graphite electrodes upon
tercalation depends mostly on passivation phenomena.
As explained in Section 18.3 above, polarization of any carbon
rode to low potential in any polar aprotic, Li salt solution leads to
sive reduction processes of solution species on the electrode's surface,
the precipitation of surface films. These surface films consist mostly of
lts, and they conduct Li-ions (and thus, do not interfere badly with Li
tion-deinsertion into/from the carbon). A mandatory condition for the f
an electrolyte solutions for Li-C anodes is that the reduction of the ion
species on the carbon surface occurs at potentials higher than that of
134
sertion potentials. The products of solution reduction should have a
cohesion and a strong adhesion to the carbon surface. Hence, they
ld precipitate as quickly as possible to form passivating surface films,
h efficiently block electron transfer from the negatively charged
rode to solution species.
Finally, it should be noted that the 3D structure of the carbon and the
hology of the particles plays a major role in determining its performance
135
Li insertion anode material. Disordered carbons are much more robust
graphitic carbons, and hence, the reversibility of Li insertion into them is
136
less dependent on the solution composition. The smoother the
ce through which Li is inserted into the carbon (e.g., the edge plane
s of graphite particles), the faster are the surface film formation
sses, and the better the passivation. As a graphitic carbon includes more
stratic disorder in its 3D structure, it is more robust, and hence, it is
enced less by solution-related detrimental processes during intercalation
lithium. 135
.2. Standard Electrolyte Solutions for Li-ion Batteries

Over the years, a large matrix of solvents, salts and additives was tested
136 138
graphite electrodes. - As explained in the previous section, graphite
e most sensitive carbon material. Composite electrodes consisting of
Chapter 18
hetic graphite flakes seem to be the best probe to test suitable solutions
Li-ion batteries, because what works well with these electrodes should
even better with other types of graphite and with disordered carbons. 135
e intensive studies converged into the development of standard
rolyte solutions which are commonly used throughout the world in the
production of commercial Li-ion batteries. These solutions include
6 as the electrolyte and alkyl carbonate solvents. EC is an obvious
ponent, 139 together with cosolvents that are acyclic alkyl carbonates from
40 141
DEC/ DMC, EMC,59 and PMC60 list. The alkyl carbonate solvents
he most suitable because their anodic stability is higher than that of other
lies of solvents such as ethers and esters. Also, at low potentials both Li
Li-C compounds reach a very good passivation in a variety of alkyl
onate solutions. It should be noted that neither ethers nor esters promote
passivation of Li-C electrodes. 132- 138 In the case of ethers, these solvents
not sufficiently reactive to form passivation layers at high enough
ntials. Thus, ether molecules can easily cointercalate into graphite with
ns, which leads to the destruction of the graphite's structure
oliation). 133 The reduction products of esters are not cohesive and
sive enough to form sufficiently passivating surface films on carbon
rodes.
Highly interesting, and deserving of discussion herein, is the failure of
hite electrodes in PC solutions. No graphite carbons can insert Li
sibly in pure PC solutions with Li salts. 134 It appears that some
uctive processes take place at relatively high potentials (in which PC is
ced, 1.5-0.8 vs. Li/Li+), and the electrodes cannot reach all of the Li
132
calation stages. Reversible behavior of Li-graphite electrodes is
ned in mixtures of ethers and PC, PC solutions containing crown ethers
142
, 12 Cr 4), and in EC-PC mixtures 143• In the latter solutions, the
ersible capacity is very strongly dependent on the morphology and
ee of disorder of the graphite particles. 135 The fact that in solutions based
C, which is similar in structure to PC, the behavior of Li-graphite
rodes is so reversible, is striking. PC is reduced on graphite below 1.5 V
species such as CH3CH(OC02Li)CH20C02Li and CH3CH=CH2
pylene gas), while EC is reduced to species such as (CH2 0C02 Lih and
46
ene gas. •50 The EC reduction products, when formed, precipitate
kly on the graphite surface to form cohesive and adhesive surface films
passivate the electrode and block electron transfer, thus preventing any r
surface reactions except for Li insertion (which include Li-ion
ation through the surface films). The fast precipitation of the surface
in this case prevents detrimental side processes such as cointercalation
lvent molecules.
In the case of PC, its reduction product contains the methyl group,
h interferes badly with the good cohesion and adhesion of the surface
es and formation of compact Li2C03 film, as seen in the case of EC

ed surface film. As a result, an intensive PC reduction takes place at low
tials before passivation is completed. This allows a build up of internal
ure (due to propylene gas formation), which cracks the graphite particles
135
revents their passivation by stable surface films. The surface films in
case rather lead to electrical isolation of a great part of the active
. 133• 135 Figure 18.7 illustrates the behavior of graphite electrodes in PC
EC solutions.135 In PC-ether solutions, the surface chemistry of graphite
rodes is different: LizC03, which is an excellent passivating agent,
mes a dominant surface species, and thus, efficient electrode
86
vation can be developed in these solutions.
The change in the surface chemistry in these cases relates to the
ble formation of carbonate ion (C0 3= or LiC03-), which may be a strong
ophile in aprotic media. At high PC concentration, it attacks another PC
cule nucleophilically to form CH 3CH(OC02Li)CH20C02Li, while at a
enough PC concentration it interacts with Li-ions and precipitates as
03• 86 The difference in the behavior of graphite electrodes in PC and EC
ions, as well as the conclusion on the impact of the surface
istry/surface film composition of the electrodes' behavior, serves as a
tool for understanding the failure and stabilization of graphite
odes in solutions, and for designing optimized electrolyte solutions. A
ssue is the formation of passivating, compact and stable surface films of
Li-ion conductivity at high enough potentials, before detrimental
sses such as solvent cointercalation, build up of internal pressure, and
iation/cracking of graphite particles take place.
Regarding the choice of electrolyte, LiPF6 was selected as the standard
which is commonly used in Li-ion batteries. Below we list some reasons
is, some of which relate to problems with alternative salts.
1. This salt is much less poisonous than LiAsF6, which is
commercially better for both Li and Li-C anodes.
2. A very good passivation of the aluminum current collectors used for
the cathodes is reached in LiPF6 solutions, probably by AlF3 (e.g.,
by reactions of trace HF, unavoidably present in all LiPF6
solutions, with aluminum).42
3. The conductivity of alkyl carbonates/LiPF6 solutions is relatively
high. (See Table 18.4.)
4. The thermal stability of battery systems based on LiPF6 solutions is
acceptable, compared with systems containing LiC104 (a very good
salt in terms of electrode performance at ambient temperatures),
which are explosive. 144
5. The mass production of LiPF6 is relatively easy and its price is
acceptable.
However, LiPF6 solutions have a pronounced disadvantage, which is
Chapter 18
unavoidable contamination with HF. 145 LiPF6 decomposes into LiF and
in an equilibrium reaction. 146 The latter species readily hydrolyzes
protic substances to form POxFy, P(OR)xFy species and HF, which reacts
ROLi or ROC02Li surface species formed by alkyl carbonate reduction,
replaces them with highly resistive LiF surface films. HF also reacts
the LixMOy cathode materials. The impact of the HF reactions on the
trodes' passivation is negative and the electrodes' impedance increases
F surface reactions are more intensive.
In conclusion, the standard electrolyte solutions for Li-ion batteries
LiPF6 solutions in mixtures of alkyl carbonates that always contain EC as
ghly polar component and a precursor for very good passivating agents
61 62
, (CH20C02Li)2) ' and solvents from the DMC, EMC, DEC
39 141
- It should be emphasized that ternary mixtures such as EC-DMC
enable a very good low temperature conductivity of LiPF6 solutions to
eached.25 The choice of these systems is mostly due to an acceptable
ivation and stabilization of both anodes and cathodes for Li-ion batteries
a wide electrochemical window (i.e., high anodic stability > 4.5 V vs.
t with lithiated transition metal oxide cathodes).41 Major problems with
use of these solutions in practical batteries relate mostly to HF
amination, which interferes badly with the electrodes' passivation. In
tion, at elevated temperatures there are accelerated surface reactions on
anodes and cathodes, which increases their impedance and lead to
rent capacity-fading of Li-ion batteries.147' 148 There is, therefore, a
g driving force for R&D of improved electrolyte solutions for Li-ion
ries, as summarized in the next section.
5.3. Advanced Electrolyte Solutions for Li-ion Batteries
Special challenges in R&D of new electrolyte solutions for Li-ion

ries include the following:
1. Increasing the anodic stability.
2. Increasing low temperature conductivity.
3. R&D of non-flammable solvents.
4. The use of acid scavengers.
5. R&D of more stable, non-contaminating salts.
6. Enhancing the passivation of the electrodes, especially at elevated
temperatures.
7. Overcharge protection.
anced Liquid Electrolyte Solutions 565
I( ., ,
s;r- !- (CH _ O+C02 Li)2
2ne- +2nEC+2nLit----+"'?i=.r _..,.,;; ::i: -'2
Current collector CCH
2
CH3
I
CH - OC02Li
I
C -OC02Li
The methyl group in
the RC02Li species interferes
--
wi th a +C H-
good
cohesion
CH=C
c==:)
Worse
passivation
Gas bubbles are
Formed inside the
particles
aphite particles
e
litted ¢::=::J
e More
reactive
s
essure su rface
D
exposed,
further surface reactions
.-f?C¥t 1 Sf 7?' , Ele trochemically

Current coll ector ISOlated
graphi te flakes
re 18.7. A schematic presentation of stabilization and failure mechanisms of graphite

trodes in alkyl carbonate solutions as indicated. Ref. 135.
It should be emphasized that the possibility of finding replacements for

currently used alkyl carbonate solvents and LiPF6 is difficult. Much
ier is the use of functional additives that improve the electrodes'
rfaces and enhance conductivity. The R&D of special additives for Li-ion
Chapter 18
ry solutions is indeed highly innovative, and opens the door for

ounced improvements in the battery's performance and safety features.
w we list typical solvents, salts, and additives that were suggested in
nt years as desirable components in electrolyte solutions for Li-ion
ries.
We find in the literature a number of reports on substituted alkyl
onate solvents that improve performance. Trans butylene carbonate was
d to be a suitable single solvent in which graphite electrodes behave
149
ly reversibly. Chlorinated and fluorinated alkyl carbonates were also
150 151
d to be very suitable ' for graphite anodes. The major importance of
e solvents is their use as a replacement for EC, whose freezing point is
high (RT). There are studies of pyrocarbonates (ROC02C02R) and
152
rbonates (ROC02ROC02R) as cosolvents. Their major advantage
es to modification of the surface chemistry of both anodes and cathodes
an improvement of the high temperature performance. Alkyl, aryl and
sulfites were tested as cosolvents that react predominantly on the
es' surface, enhance passivation, and decrease the irreversible capacity
153 154 155
lithiated carbon anodes. Alkyl nitrates, sulfates and
156
phonates were also suggested as passivation-improving additives
se presence in standard solutions reduces the irreversible capacity of Li
nodes. Another 157 additive that was the focus of attention is vinylene
onate (VC). VC is more reactive than EC on both anodes and
odes. It is reduced on the anodes to ROC02Li and poly Li alkyl
onate species, and may polymerize on the LiMOx cathodes to form
carbonates. The presence of VC in solutions reduces irreversible
157
city and increases stability, especially at elevated temperatures. Alkyl
phates and phosphazenes 158 were also suggested as non-flammable
ents for Li-ion batteries. Alkyl phosphates were suggested recently as
159
lizers for LiPF6 solutions at high temperatures.
Aromatic species
160 such as biphenyl were suggested as overcharging
ction agents. At excessive potentials, they may be reduced into
al anions at the negative or oxidize to radical cations at the positive
rode, after which the radical ions migrate/diffuse to the counter
rode to reversibly form the pristine compounds. We should also mention
161
tives such as alkyl silanes and LhC03, which can be considered as an
162
scavenger. Active gases were also tested as additives for electrolyte
ions. C02 in solutions137reacts on Li-C surfaces to form LhC03, an
llent passivating agent. S02 solutions react on Li-C to form LhS,
163
03, LhS204, and LhO, which are also good passivating agents.
ntly, Li organo borate complexes were suggested164 as promising additives
mproving high temperature performance. What is described above in
includes only a few typical examples of R&D of new additives for Li
battery solutions. In recent years we have seen intensive efforts to
lop additives that improve the passivation of the Li-ion battery

trodes.
Finally, we should review attempts to introduce new families of salts.
165
ck developed the LiPF3(CF2CF3h (LiFAP) salt. It was proven that this
is much more stable than LiPF6. Its solution is less acidic and more
166
mally stable than LiPF6 solutions. Both graphite anodes, LiCo02 and
nz04 cathodes, behave better in LiFAP solutions than in LiPF6 solutions,
cially at elevated temperatures. The LiN(S02CF2CF3)2 (BETI) salt was
introduced. 167 However, we found that for Li-ion battery systems its
tions are inferior in their electrochemical performance to LiPF6 or LiFAP
166
tions.
Recently, the 'LiBOB', Li bis(oxalo)borate salt was suggested as a
168
le, non-acidic electrolyte. Both Li-C and LixMOy cathodes seem to
ve highly reversibly in its solution. However, more work is needed to
irm that this salt is really an alternative to LiPF6•
6. MOLTEN SALTS
6.1. Introduction
Molten salts may be useful non-aqueous electrolyte systems for a
ty of applications, including batteries. 169' 170 Ionic liquids of anhydrous
, namely molten salts, usually have high conductivity and wide stability
large potential ranges. 171
Especially important are the room temperature molten salts. The most
ly studied are the chloroaluminate melts with organic cations such as
yl-ethylimidazolium, N-butyl pyridinium, pentamethylimidazolium, and
dimethyl-1,2,4 triazolium.172
Regarding the conductivity of molten salts, unlike conventional
queous solutions, where the medium is almost unaffected by the
rical field, in molten salts the properties of a medium which consists
of ions are affected by the applied electric field. It is often found that
quivalent conductivity increases as the electrical field applied is higher,
to a certain point where a plateau is reached. 173 Regarding the
uctivity of molten salts, several conduction mechanisms have been
osed to explain their high electrical conductivity. 174' 175 With organic RT
en salts, a maximum conductivity is usually reached with neutral
ms.176
Chapter 18
6.2. Electrochemical Properties of Molten Salt Solutions

and Battery Applications
The potential windows of molten salts are very important for their
rochemical applications. The limiting potentials depend on the
position of the melts and the temperature, which affect the standard
ntial of reduction and oxidation of their components.
In the case of room temperature molten salts, the anion is typically a
ure of aluminum halide complexes in equilibrium. These anions can
rgo several oxidation reactions, depending on the composition of the
. The cathodic reactions may involve reductions of the organic
177
ns.
It was possible to obtain reversible Li and AI dissolution and
172
sition in some RT organic molten salts. Hence, these molten salts
be considered as possible electrolyte systems for rechargeable
queous lithium and aluminum batteries.
Recent work on the behavior of cathodes for Li batteries (e.g.,LixMOy,
178 179
Mn, Co) • clearly demonstrates that RT molten salts may be
ising electrolytes for secondary Li-ion batteries as well, comprising
ns such as ethyl methyl imidazolium and anions such as cr, BF4-, and
4-.
Finally, there are reports on the use of these systems in super (EDL)
180
citors.
FERENCES
C. R. Schlaikjer, in Lithium Batteries, J. P. Gabano, Ed., Academic Press, London

1983), Chapter 13, p. 304.
. Garcia, A. J. Bard, Chern. Phys. Lett. 120 (1985) 437.
M. Elam, E. Peled, E. Gileadi, J. Electrochem. Soc. 131 (1984) 2058.
. Peled, M. Brand, E. Gileadi, J. Electrochem. Soc. 128 (1981) 1697.
. Galasio, R. Galasio, J. Thonstad, in Nonaqueous Electrochemistry, Wiley-VCH,
Weinheim (2002), Chapter 12, p. 342.
G. Blomgren, in Nonaqueous Electrochemistry, D. Aurbach, Ed., Marcel Dekker, N.J.
1999), Chapter 2, p. 53.
M. Swarc, in Ions and Ion Pairs in Organic Reactions, Vol. 1, M. Swarc, Ed., Wiley
nterscience, New York, (1972), Chapter 1.
. Reichardt, Angew. Chern. Int. Ed. Engl. 18 (1979) 87.
. Mayer, in Proceedings of the Workshop on Lithium Nonaqueous Battery
lectrochemistry, E. B. Yeager, B. Schumm, Jr., G. Blomgren, D. R. Blankenship, V.
eger, J. Akridge, Eds., The Electrochemical Society, Inc., PV 80-7, 1980, p. 13.
. Dirnrot, C. Reichardt, T. Siepmann, F. Bohlmann, Justus Liebigs. Ann. Chern. 661
1963) 2; K. Dirnrot, C. Reichardt, ibid. 727 (1969) 93.
Z. B. Maksimovic, C. Reichardt, A. Spiric, ibid. 270 (1974) 100.
U. Mayer, Pure Appl. Chem. 51 (1979) 1697.
V. Gutmann, The Donor-Acceptor Approach to Molecular Interactions, Plenum Press,
New York, 1978.
U. Mayer, Monatsch. Chem. 108 (1977) 1479.
J. F. Coetzee, in Proceedings of the Workshop on Lithium Nonaqueous Battery
Electrochemistry, E. B. Yeager, B. Schumm, Jr., G. Blomgren, D. R. Blankenship, V.
Leger, J. Akridge, Eds., The Electrochemical Society, Inc., softbound series PV 80-7,
1980,p. 70
G. R. Hedwig, D. A. Owensby, A. J. Parker, 1. Am. Chem. Soc. 97 (1975) 3888.
P. W. Atkins and J. D. Paula, Physical Chemistry, 7th ed. Oxford University Press,
Oxford, Melbourne, Tokyo, 2002, Chapter 24.
S. H. Maron, J. B. Lando, Fundamentals of Physical Chemistry, Macmillan, New York;
Collier Macmillan, London, 1974, Chapter 13, Eq. 40.
R. M. Fuoss, 1. Am. Chem. Soc. 80 (1958) 5059.
R. M. Fuoss, T. Shedlovsky, 1. Am. Chem. Soc. 71 (1949) 1496.
H. V. Venkatasetty, in Proceedings of the Workshop on Lithium Nonaqueous Battery
Electrochemistry, E. B. Yeager, B. Schumm, Jr., G. Blomgren, D. R. Blankenship, V.
Leger, J. Akridge, Eds., The Electrochemical Society, Inc., softbound series PV 80-7,
1980, p. 46.
J. F. Casteel, E. A. Amis, 1. Chem. Eng. Data 17 (1972) 55.
C. A. Angell, J. C. Tucker, 1. Phys. Chem. 78 (1974) 278.
J. Barthel, H. G. Gores, P. Carlier, F. Feuerlein, M. Utz, Ber. Bunsenges. Phys. Chem. 87
(1983) 436.
M. C. Smart, B. V. Ratnakumar, S. Surampudi, 1. Electrochem. Soc. 146 (1999) 486; M.
C. Smart, B. V. Ratnakumar, S. Surampudi, ibid. 149 (2002) A361.
0. Youngman (Chusid), Y. Gofer, A. Meitav, P. Dan, D. Aurbach, Electrochim. Acta 35
(1990) 625.
D. Aurbach, Y. Gofer. 1. Electrochem. Soc. 138 (1991) 3529.
T. B. Reddy, S. Hossain, in Handbook of Batteries, 3'd Edition, D. Linden and T. B.
Reddy, Eds., McGraw Hill, Inc., N.Y., 2002, Chapter 34.
J. Barthel, H.-J. Gores, in Chemistry of Nonaqueous Solutions, G. Mamantov and A. I.
Popov, Eds., VCH, New York, 1994, p. 1.
J. T. Dudley, D.P. Wilkinson, G. Thomas, R. LeVae, S. Woo, H. BJorn, C. Horvath, M.
W. Juzkow, B. Denis, P. Aghakian, J. R. Dahn, 1. Power Sources 35 (1991) 59.
M. Salomon, E. J. Plichta, Electrochim. Acta 30 (1985) 113.
K. Hayashi, Y. Nemoto, S. Tobishima, J. Yamaki, Electrochim. Acta 44 (1999) 2337.
G. M. Swain, A. B. Anderson, J. C. Angus, MRS Bulletin 23 (1998) 56.
D. Aurbach, in Nonaqueous Electrochemistry, D. Aurbach, Ed., Marcel Dekker, N.Y.
1999, Chapter 6, p. 289.
. 0. Besenhard, G. Eichinger, 1. Electroanal. Chem. 60 (1976) 1.
M. W. Verbrugge, B. J. Koch, 1. Electrochem. Soc. 141 (1994) 3053.
D. Aurbach, R. Skaletsy, Y. Gofer, 1. Electrochem. Soc. 138 (1991) 3536.
Z. Lu, A. Schechter, M. Moshkovich, D. Aurbach, 1. Electroanal. Chem. 466 (1999) 203.
E. Peled, in Li Batteries, J. P. Gabano, Ed., Academic Press, N.Y. and London 1983,
Chapter 3, p. 43.
D. A. Shifter, P. J. Moran, J. Kruger, Electrochim. Acta 42 (1997) 567.
K. Kanamura, 1. Power Sources 81-82 (1999) 123.
M. Morita, T. Shibata, N, Yoshimoto, M. Ishikawa, Electrochim. Acta 47 (2002) 2787.
. R. Dahn, A. K. Sleigh, H. Shi, B. M. Way, W. J. Weydanz, J. N. Reimers, Q. Zhong,
U. von Sacken, Lithium Batteries- New Materials, Developments and Perspectives, G.
stoia, Ed., Elsevier-North Holland, Amsterdam 1992, p. 1.
Chapter 18
T. Ohzuku, A. Veda, M. Nagayama, Y. Iwakoshi, H. Komori, Electrochim. Acta 38

(1993) 1159.
P. Novak, K. Muller, K. S. V. Santhanam, 0. Hass, Chern. Rev., 97, 207 (1997); M.
Gazard in Handbook of Conducting Polymers, Vol. 1, T. A. Skotheim, Ed., Marcel
Dekker, New York, 1988, Chapter 19.
D. Aurbach, H. E. Gottlieb, Electrochim. Acta 34 (1989) 141.
D. Aurbach, J. Electrochem. Soc. 136 (1989) 906.
D. Aurbach, J. Electrochem. Soc. 136 (1989)1606.
D. Aurbach, M. L. Daroux, P. Faguy, E. Yeager, J. Electrochem. Soc. 134 (1987) 1611.
D. Aurbach, Y. Gofer, M. Ben-Zion, P. Aped. J. Electroanal. Chern. 339 (1992) 451.
D. Aurbach, M.L. Daroux, P. Faguy, E. Yeager, J. Electroanal. Chern. 297 (1991) 225.
D. Aurbach, M. Moshkovich, Y. Gofer, J. Electrochem. Soc. 148 (2001) E155.
D. Aurbach, Y. Ein-Eii, Langmuir 8 (1992) 1845.
D. T. Sawyer, G. Chlericato, C. T. Angelsi, E. J. Nanni, T. Tsuchiya, Anal. Chern. 54
1982) 1720.
B. Burrows, S. Kirkland, J. Electrochem. Soc. 115 (1968) 1164.
Y. Ein-Eii, B. Markovsky, D. Aurbach, J. Power Sources 54 (1995) 281.
D. Aurbach, A. Zaban, 0. Chusid, I. Weismann, ElectrochimicaActa 39 (1994) 51.
D. Aurbach, E. Granat, Electrochimica Acta 42 (1997) 697.
D. Aurbach, A. Schechter, B. Markovsky, Y. Ein-Eli, V. Koch, J. Electrochem. Soc. 143
1996) L273.
Y. Ein-Eii, S. F. McDevitt, B. Markovsky, A. Schechter, D. Aurbach, J. Electrochem.
Soc. 144 (1997) L 180.
D. Aurbach, A. Zaban, A. Schechter, Y. Ein-Eii, E. Zinigrad, B. Markovsky. J.
D. Aurbach, B. Markovsky, A. Schechter, Y. Ein-Eii, H. Cohen, J. Electrochem. Soc.
143 (1996) 3809.
D. Aurbach, 0. Chusid, I. Weissman, Electrochimica Acta 41 (1996) 747.
E. Goren, 0. Chusid, D. Aurbach, J. Electrochem. Soc. 138 (1991) L6.
0. Chusid (Youngman), D. Aurbach, J. Electrochem. Soc. 140 (1993) Ll.
D. Aurbach, 0. Chusid, J. Electrochem Soc. 140 (1993) L155.
D. Aurbach, Y. Ein-Eii, and A. Zaban, J. Electrochem. Soc. 141 (1994) Ll.
M. Arakawa, J. I. Yamaki, J. Power Sources 54 (1995) 250.
G. Eggert, J. Heitbaum, Electrochim. Acta 31 (1986) 1443.
F. Joho, P. Novak, Electrochim. Acta, 45 (2000) 3589.
M. Moshkovich, M. Cojocaru, H. E. Gottlieb, D. Aurbach, J. Electroanal. Chern. 497
2001) 84.
. E. Dubois, G. Tourillon, P. C. Lacaze, J. Electrochem. Soc. 125 (1978) 1257.
D. H. Jang, S.M. Oh, J. Electrochem. Soc. 144 (1997) 3344.
K. Kanamura, S. Toriyama, S. Shiraishi, z. Takehara, J. Electrochem. Soc. 143 (1996)
548.
G. Eggert, J. Heitbaum, Electrochim. Acta 31 (1986) 1443.
B. Rasch, E. Cattaneo, P. Novak, W. Vielstich, Electrochim. Acta 36 (1991) 1397.
D. Aurbach, Y. Gofer, M. Ben-Zion, J. Power Sources 39 (1992) 163.
. A. See!, J. R. Dahn, J. Electrochem. Soc. 147 (2000) 892.
D. Aurbach, B. Markovsky, M. D. Levi, E. Levi, A. Schechter, M. Moshkovich, Y.
Cohen, J. Power Sources 81-82 (1999) 95.
. Ossola, G. Pistoia, R. Seeber, P. Ugo, Electrochim. Acta 33 (1988) 47.
L.A. Zvat'kova, V. N.Atanasev, G. A. Krestov, T.V. lvanova, Elektrokhimiya 29 (1993)
46.
R. Herr, Electrochim. Acta 35 (1990) 1257.
M. Ue, K. Ida, S. Mori, J. Electrochem. Soc. 141 (1994) 2989.
K. T. Ciemieki and J. A. Auborn, in Proceedings of the Symposium on Lithium Batteries,

A. N. Dey, Ed., ECS Meeting, Washington, The Electrochemical Society, Inc., softbound
series PV 84-1, The Electrochemical Society, Inc., Pennington, N.J., 1984, p. 363.
C. K. Mann, in Electroanalytical Chemistry, A. J. Bard, Ed., Marcel Dekker, New York,
1970, Vol. 3.
D. Aurbach, M.D. Levi, E. Levi, A. Schechter, J. Phys. Chern. B. 101 (1997) 2195.
K. Wang, P. N. Ross, Jr., F. Kong, F. McLarnon, J. Electrochem. Soc. 143 (1996) 422.
D. Aurbach, E. Zinigrad, A. Zaban, J. Phys. Chern. 100 (1996) 3089.
A. Schechter, D. Aurbach, Langmuir 15 (1999) 3334.
K. Kanamura, H. Tamura, Z.-1. Takehara, J. Electroanal. Chern. 333 (1992) 127.
K. Kanamura, S. Shiraishi, Z.-1. Takehara, J. Electrochem. Soc. 143 (1996) 2187.
K. Kanamura, S. Shiraishi, H. Tamura, and Z.-1. Takehara, J. Electrochem. Soc. 141
(1994) 2379.
E. Peled, D. Golodnitsky, and J. Penciner, in Handbook of Battery Materials, J. 0.
Besenhard, Ed., Wiley-VCH, New York, 1999, Chapter 6, p. 419.
D. Aurbach, Y. Gofer, Y. Langzam, J. Electrochem. Soc. 136 (1989) 3198.
D. Aurbach. Y. Cohen, Electrochem. Solid-State Lett. 2 (1999) 16. D.
Aurbach, Yaron Cohen, J. Phys. Chern. B 104 (51) (2000) 12282.
V. S. Bagotzky, A. M. Skundin, Russ. J. Electrochem. 31 (1995) 308; translated from
Electrokhimiya 31 (1995) 342.
K. M. Abraham, S. B. Brummer, in Li Batteries, J.P. Gabano, Ed., Academic Press, N.Y.
and London 1983, Chapter 14, p. 371.
V. R. Koch, J. L. Goldman, D. L. Natwing, J. Electrochem. Soc. 129 (1982) 1.
D. Aurbach, I. Weissman, A. Schechter, H. Cohen, Langmuir 12 (1996) 3991.
S. Tobishima, M. Arakawa, T. Hirai, J. Yamaki, J. Power Sources 28 (1989) 449.
K. M. Abraham, J. L. Goldman, D. L. Natwig, J. Electrochem. Soc. 129 (1982) 2404.
D.P. Wilkinson, H. BJorn, K. Brandt, D. Wainwright, J. Electroanal. Chern. 355 (1993)
193.
V. R. Koch, R. H. Young, Science 204 (1979) 499.
E. Plichta, M. Salomon, S. Shane, M. Uchiyama, J. Power Sources 21 (1987) 25.
Y. Matsuda, M. Morita, J. Power Sources 26 (1989) 579.
I. R. Hill, B. G. Anderson, M. Goledzinowski, R. J. Dare, J. Electrochem. Soc. 142
(1995) 3267.
B. J. Carter, B. Jeffries, S. P .S. Yen, in Proceedings of the Symposium on Lithium
Batteries, A. N. Dey, Ed., The Electrochemical Society, Inc., softbound series PV 87-1,
1987, p. 145.
Y. Matsuda, M. Morita, H. Nigo, in Proceedings of the Symposium on Primary and
Secondary Lithium Batteries, K. M. Abraham and M. Salomon, Eds., The
Electrochemical Society, Inc., softbound series PV 91-3, 1991, p. 272.
J. Yamaki, S. Tobishima, in Proceedings of the Symposium on Primary and Secondary
Lithium Batteries, K. M. Abraham and M. Salomon, Eds., The Electrochemical Society,
nc., softbound series PV 91-3, 1991, p. 235.
C. D. Desjardins, T. G. Green, G. K. Maclean, J. Power Sources 12 (1988) 489.
N. Kumagai, Y. Kikuchi, K. Tanno, J. Appl. Electrochem. 22 (1992) 620.
D.P. Wilkinson, D. Wainwright, J. Electroanal. Chern. 335 (1993) 193. D.
Aurbach, J. Power Sources 89 (2000) 206.
P. Biensan, B. Simon, J.P. Peres, A. de Guibert, M. Broussely, J. M. Bodet, J. Power
Sources 81-82 (1999) 906.
L. Dominey, in Nonaqueous Electrochemistry, D. Aurbach, Ed., Marcel Dekker, N.Y.
1999, Chapter 8, p. 437.
K. M. Abraham, D. M. Pasquariello, F. J. Martin, J. Electrochem. Soc. 133 (1986) 662.
K. M. Abraham, D. M. Pasquariello, D. A. Schwartz, J. Power Sources 26 (1989) 247.
Chapter 18
C. D. Desjardins, T. G Cadger, R. S. Salter, G. Donaldson, E. J. Casey, J. Electrochem.

Soc. 132 (1985) 529.
D. Aurbach, A. Zaban, Y. Gofer, 0. Abramson, M. Ben-Zion, J. Electrochem. Soc. 142
1995) 687.
K. M. Abraham, J. Power Sources 14 (1985) 179.
J. L. Goodman, L.A. Dominey, V. Koch, J. Power Sources 26 (1989) 519.
D. Aurbach, I. Weissman, A. Zaban, Y. Ein-Eii, E. Mengeritsky, P. Dan, J. Electrochem.
Soc. 143 (1996) 2110.
P. Dan, E. Mengeritsky, Y. Geronov, D. Aurbach, I. Weissman. J. Power Sources 54
1995) 143.
D. Aurbach, E. Zinigrad, H. Teller, P. Dan, J. Electrochem. Soc. 147 (2000) 2486.
D. Aurbach, E. Zinigrad, H. Teller, Y. Cohen, G. Salitra, J. Electrochem. Soc. 149 (2002)
A1267.
T. Osaka, in Nonaqueous Electrochemistry, D. Aurbach. Ed., Marcel Dekker, N.Y. 1993,
Chapter 7, p. 441.
J. B. Goodenough, in Advances in Li-ion Batteries, W. A. Van Schalkwijk and B.
Scrosati, Eds., Kluwer Academic/Plenum Publishers, N.Y., London and Moscow 2002,
Chapter 4, p. 135.
D. D. MacNeil, Z. Lu, J. R Dahn, J. Electrochem. Soc. 149 (2002) A1332.
Y. Ein-Eii, M.A. Rzeznik, S. Mukerjee, X. Q. Yang, J. McBreen, J. Electrochem. Soc.
145 (1998) 3383.
M. Winter, J. 0. Besenhard, in Handbook of Battery Materials, J. 0. Besenhard, Ed.,
Wiley-VCH, Weinheim, New York, Chapter 5, 1990, p. 383.
D. Aurbach, B. Markovsky, K. Gamolsky, E. Levi, Y. Ein-Eii, Electrochimica Act, 45
1999) 67.
E. Buiel, and J. R. Dahn, Electrochim. Acta 45 (1999) 121.
D. Aurbach, H. Teller, E. Levi. J. Electrochem. Soc. 149 (2002) A1255.
D. Aurbach, 0. Youngman Chusid, Y. Carmeli, M. Babai, Y. Ein-Eii, J. Power
ources 43 (1993) 47.
D. Aurbach, B. Markovsky, Y. Ein-Eii, Y. Carmeli, H. Yamin, S. Luski, Electrochim.
cta 39 (1994) 2559.
R. Yazami, Electrochim. Acta 45 (1999) 87.
D. Guyomard, J. M. Tarascon, J. Power Sources 54 (1995) 92.
. M. Tarascon, and D. Guyomard, Solid State Ionics 69 (1994) 293.
D. Guyomard, J. M. Tarascon, J. Electrochem. Soc. 140 (1993) 3071.
Z. X. Shu, R. S. McMillan, J. J. Murray, J. Electrochem. Soc. 140 (1993) 922.
R. Pong, U. Von Sacken, J. R. Dahn, J. Electrochem. Soc. 137 (1990) 2009.
M. N. Richard, J. R. Dahn, J. Electrochem. Soc. 146 (1999) 2068.
U. Heider, R. Oesten, M. Jungnitz, J. Power Sources 81-82 (1999) 119.
. E. Sloop, J. K. Pugh, S. Wang, J. B. Kerr, K. Kinoshita, Electrochem. Solid-State Lett.
(2001) A42.
D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi, H. J. Kim, Electrochim.
cta 47 (2002) 1899.
M. Broussely, in Advances in Li-ion Batteries, W. A. Van Schalkwijk and B. Scrosati,
ds., Kluwer Academic/Plenum Publishers, N.Y., London and Moscow 2002, Chapter
3, p. 393.
G. C. Chung, H. J. Kim, S. H. Jun, J. W. Choi, M. H. Kim, J. Electrochem. Soc. 147
2000) 4398.
X. Z. Shu, R. S. McMillan, J. J. Murray, J. J. Davidson, J. Electrochem. Soc. 142 (1995)
161.
M. Inaba, Y. Kawatate, A. Funabiki, S. K. Jeong, T. Abe, Z. Ogumi, Electrochim. Acta
5 (1999) 99.
H. Gan, E. S. Takeuchi, "Dicarbonate Additives for Nonaqueous Electrolyte
Rechargeable Cells," (Wilson Greatbatch Ltd., USA), Application for US Patent,

N. 2001:45110.
G. H. Wrodnigg, J. 0. Besenhard, M. Winter, J. Power Sources 97-98 (2002) 592.
H. Gan and E. S. Takeuchi, "Nitrate Additives for Nonaqueous Electrolyte Rechargeable
Cells," (Wilson Greatbatch Ltd., USA), Application for US Patent, N. 2000:752090.
H. Gan and E. S. Takeuchi, "Sulfate Additives for Nonaqueous Electrolyte Rechargeable
Cells," (Wilson Greatbatch Ltd., USA), Application for Eur. Patent Appl. Ep./2001, N.
2001:451045 7.
H. Gan and E. S. Takeuchi, "Phosphonate Additives for Nonaqueous Electrolyte
Rechargeable Cells," (Wilson Greatbatch Ltd., USA), US Patent Application, US 2002,
N. 2001:453468; US Patent 6,200,701.
D. Aurbach, K. Gamolsky, B. Markovsky, Y. Gofer, Electrochim. Acta 47 (2002) 1423. K.
Xu, M. S. Ding, S. Zhang, J. L. Allen, T. R. Jow, J. Electrochem. Soc. 149 (2002)
A622.
S. S. Zhang, J. Ku, T. R. Jow, Electrochem. Solid-State Lett. 5 (2002) A206.
H. Mao, "Polymerizable Aromatic Additives for Overcharge Protection in Secondary
Nonaqueous Lithium Batteries," (MOLl Energy (1990) Ltd. Can.): Eur. Pat. Appl.;
Ep./1997, An. 1997:257375.
H. Buqa, M. V. S. Alvarez, J. 0. Besenhard, M. Winter, J. Power Sources 97-98 (2001)
126.
A. M. Stux and J. Barker, "Carbonate Additive for Inhibiting Decomposition of Lithium
Salts in Nonaqueous Battery Electrolyte and Electrolytes Containing this Additive,"
(Valence Technology, Inc., USA). US Patent Application, US, 1998, An. 1998:44815.
Y. Ein-Eii, S. R. Thomas, V. R. Koch, J. Electrochem. Soc. 143 (1996) L195.
D. Aurbach, B. Markovsky, K. Gamolsky, M. Koltypin, U. Heider, M. Schmidt, ITE
Battery Letters, Vol. 2, No.3, B46 (2001).
M. Schmidt, U. Heider, A. Kuehner, R. Oesten, M. Jungnitz, N. lgnat'ev, P. Sartori, J.
Power Sources 97-98 (2001) 557.
J. S. Gnanaraj, M.D. Levi, Y. Gofer, and D. Aurbach, J. Electrochem. Soc. (2002), in
press.
K. Naoi, M. Nori, Y. Naruoka, W. M. Lamanna, R. Atanasoski, J. Electrochem. Soc. 146
(1999) 462.
K. Xu, S. S. Zhang, R. R. Jow, W. Xu, C. A. Angell, Electrochem. Solid-State Lett. 5
(2002) A26.
E. J. Cairns, R. K. Steunenberg, in "Progress in High Temperature Physics and
Chemistry," Vol. 5, C. A. Rouse, Ed., Pergamon Press, New York, 1973, p. 63.
E. J. Cairns, "Batteries in Molten Salt Technology,' D. G. Lovering, Ed. Plenum Press,
N.Y. (1972), p. 287.
C. L. Hussey, in Advances in Molten Salt Chemistry, Vol. 5, G. Mamantov and C. B.
Mamantov, Eds., Elsevier, Amsterdam, 1983, pp. 55-78.
A. Webber, G. E. Blomgren, in Advances in Li-ion Batteries, W. A. Van Schalkwijk and
B. Scrosati, Eds., Kluwer Academic/Plenum Publishers, N.Y., London and Moscow,
Chapter 6, p. 185.
T. A. Zawodzinski, R. A. Osteryoung, lnorg. Chem. 29 (1990) 2842.
B. R. Sundheim, in Fused Salts, B. R. Sundheim, Ed., McGraw-Hill, New York, 1964,
Chapter 3.
G. J. Janz, Molten Salts Handbook, Academic Press, New York, 1967, p. 344.
J. S. Wilkes, J. A. Levisky, R. A. Wilson, C. Li Hussy, Inorg. Chem. 21 (1982) 1263.
P.R. Gifford, J. B. Palmisano, J. Electrochem. Soc. 134 (1987) 610.
Y. S. Fung, R. Q. Zhou, J. Power Sources 81-82 (1999) 891.
K. Vi, N. Koura, Y. Idemoto, K. Iyzuka, Denki Kagaku 65 (1997) 161.
A. B. McEwen, H. K. Ngo, K. Le Compte, J. L. Goldman, J. Electrochem. Soc. 146
1996) 1687.
pter 19
POLYMERIC ELECTROLYTES:
AN OVERVIEW
J. B. Kerr
Lawrence Berkeley National Laboratory,Berkeley, California, 94720, USA
INTRODUCTION
Ionically conducting polymers have been the focus of much
mental and applied research for many years. Polyelectrolyte
ranes have found significant technological use in the production of
1
ne and caustic soda, as separators in fuel cells2'3 and in
45
dialysis · for example. The discovery of ionic conductivity in
67
hylene oxide solutions of alkali metal salts ' led the way for the
uction of polymer electrolyte in devices such as lithium batteries and
89 0
chromic windows. ' Since those early days many books/ -13 book
8 9 14 19
rs and reviews ' - have been published on these materials and the
is referred to these for more detailed information.
15
In a recent article Scrosati and Vincent have listed the desired
ties of a polymer electrolyte for use in lithium batteries. These are
ate conductivity, high cation mobility, good mechanical properties,
interfacial contact with electrodes, wide electrochemical stability,
cal and thermal stability, safety and ease of processing. This set of
ties is understandably difficult to obtain in a single material and a
r of different classes of polymer electrolyte materials have been
4
d which possess most but not all of the desired properties. Wrighe has
elaborated on this classification and defined them as:
Solvent-containing gel and hybrid systems.
Solvent-free, ion-coupled systems.
"Single-ion" systems and systems with reduced anion mobility.
Decoupled systems.
These classifications are useful for distinguishing different
nisms of ion transport in the bulk separator. However, the complete
ic Electrolytes: An Overview 575
ty requirements are most easily appreciated in the context of the

battery application and it is a purpose of this chapter to attempt to he
theoretical considerations of ion transport with the practical
ments of the battery in order to provide a rationale for material
on. For this we shall explore a mix of quantum chemistry, molecular
ics, synthesis, mechanical and transport property measurements with
modeling and chemical reactivity considerations. Although this is by
ans an exhaustive survey of useful techniques and approaches, it is
that it will provide an overview of the methods available to study the
ms involved in preparation and use of polymer electrolytes.
LITHIUM ION TRANSPORT IN LITHIUM

BATTERIES
Rigorous performance demands are made by battery engineers on
c electrolyte solutions in rechargeable lithium batteries. These
als, whether liquids, gels or polymers, are in intimate contact with
y reducing lithium metal or lithiated carbon and with reactive, highly
ng metal oxides related to materials widely used for catalysis in other
of technology. Figure 19.1 shows a schematic diagram of a typical
battery system which shows the basic components and the role of the
Load
""
l nter raccstSEI)
- "' / x ummmmu uu H i
:-.Li -t---- Li+ ----+!.! ! !m;<; mlllllllllllllllllllll
Electrolyte HH vxoy
Li+ ----E m:;-LL=='=() l ,,== =
X \ jjjUjjjjHjjijjjjillllllllll
\
"Dry" Polymers, Solvent swollen gels.
Binary salt systems, single-ion conductors
9.1. Schematic diagram of a rechargeable lithium battery (relative thickness of the

te exaggerated relative to the electrodes).
Chapter 19
olyte in separating the electrodes while facilitating the passage of

m ions between them.
In any rechargeable lithium battery capable of achieving desirable
mance three main processes must occur. Lithium ions must pass from
ode into the solution, ions must be transported across the separator
ntly (good lithium ion transport properties) and react or be intercalated
ntly and reversibly at the cathode. Since the battery should be
ged, the process must be reversed with very high efficiency (>99.99%)
er to achieve a high cycle life. In addition the electrolyte must not react
he electrodes while the battery is at open circuit thereby providing a
calendar life (15 years for electric vehicle use). Since no organic
und is thermodynamically stable to 0 volts vs. lithium metal, these
ements are stiff indeed! Fortunately, the formation of surface films at
ectrodes limits the extent of these reactions and reduces the rates to
that allow the useful cycle and calendar life of the battery to be
20
ed. An extensive review on this topic has recently appeared to which
ader is referred. For the present discussion however, it is important to
hat these surface films provide layers with very different transport
ties from the bulk electrolyte and through which the lithium ions must
hese layers are frequently referred to as the solid electrolyte interphase
and they play a crucial role in both the cycle life and also in the rate
lity of the battery.
Although chemical and thermal stability of the bulk electrolyte would
to be a prerequisite for long life, the most commonly used liquid
lytes in Li-ion batteries (ethylene carbonate/dimethyl carbonate/LiPF6)
21
t stable on standing. They have been shown to react on standing to
ransesterification products, a reaction that is catalyzed by the presence
ds or bases in the electrolytes. Such acids and bases can be formed
h oxidation or reduction reactions at the electrodes or they may be
t in the electrolyte itself either as an impurity (HF) or an intrinsic
ty of one of the components. LiPF6, LiBF4 and LiAsF6 have all been
to generate the Lewis acids PF5, BF3 and AsF5 which are known to
ze the ring-opening polymerization of ethylene carbonate22 to
hylene ether carbonates and other polymer materials with the
panying production of C02 gas. Even though these electrolytes appear
unacceptably reactive, the reactions play an important role in the
ion of the SEI and hence the cycle and calendar life of the battery.
Prior to the development of intercalation anodes for lithium ion
es, much attention was given to the behavior of the rechargeable
metal electrode with liquid electrolytes such as 1,3-dioxolane, 23- 25
nd 2-methyltetrahydrofuran (2-MeTHF)25 .27 containing LiC104 , LiBR4
AsF6 salts. The Lewis acidity of the AsF5 formed from the AsF6- was
to play a crucial role in forming surface layers that enhance the cycling
ncy. Impressive cycle lifetimes were obtained for the lithium electrode
certain conditions (LiFoM > 100). Parameters such as the purity of the
lyte and the lithium metal, prior history (exposure to heat or light),
ce of surface-active additives or impurities and cell stack pressure
ll found to affect the results. With lithium metal electrodes, loss of y
was found to be due to chemical reactions and morphological
s that convert the lithium into a form that is not easily recoverable.
an be insoluble lithium salts or isolation of lithium metal through
e growth that produces "mossy" lithium. The suppression of lithium e
growth was and still remains a critical safety and reliability issue uld
not be easily solved with liquid electrolytes. The reaction of and
flammable organic solvents with finely divided lithium powder und to
be an insurmountable safety problem. Polymer electrolytes, er, are
not volatile and although they do react with lithium metal and ng
28
metal oxides the slow delivery of fuel to the reaction site limits of
reaction and prevents runaway reactions. This safety feature plus a ion
that polymer electrolytes inhibit dendrite growth have lead to the
interest in solvent-free materials over the last twenty years.
. Solvent-Free, Ion-Coupled Systems

The commonest polymer electrolyte investigated has been based on
ylene oxide which is commercially available in a relatively pure state
onable cost. It was found by Wright that the polymer could dissolve
metal salts to provide ionic conductivity. Figure 19.2 illustrates how
ymer achieves the dissolution of the salts by complexation of the
ons via binding interactions between the ether oxygens and the metal s
shown in Figure 19.2 it is possible to bind the ions via one or more
depending on the metal ion and the structure of the polymer chain
The figure shows that the number of donor atoms bound to the metal
y also vary and in a dynamic fashion that gives rise to the movement
ions along the polymer chains. The interaction of the anion is also
where the anion binds or is ion-paired to the cation to form a neutral
a negatively charged cluster species. The figure shows contact ion
though solvent separated ion pairs are also possible. Intuitively, it is d
that the conductivity of the material will be favored by the on
of a higher concentration of charged species. Reduction of the ion
nding strength may be achieved by the use of anions such as
uoromethylsulfonyl) imide ((CF3S02)2N-), often referred to as TFSI
de, which delocalize the negative charge over the large anion
29
e. The anion in this case is dissociated from the cation and does not
significantly with the polymer chains. Its motion requires free
Chapter /9
19.2. Solvation of lithium and other metal ions by polyethylene oxide.
e between the polymer chains.

In parallel with the decreased binding of the lithium ion with the anion,
rength of binding (or solvation) to the polymer increases. Thus,
gh there is a higher concentration of free lithium ions, they are more
bound to the polymer and mobility is restricted by the need to break
onger bond in order to move along the polymer chain or from one
to another. It was found early in the study of polymer electrolyte salt
ns that the most significant contribution to conductivity occurred h
the amorphous phase where there is segmental motion of the er
chains. Although the crystalline phases are of significant interest in to
understand the thermodynamics and conformations of the ion
30 31
er complexes, ' the ionic conduction through these immobile phases
ch slower. Thus the semi rystalline PEO-lithium salt electrolytes
t reduced conductivities at temperatures below the melting point of the
ex. Similarly, the conductivity falls precipitously at temperatures
the glass transition temperature (tg) where the segmental motion of the
er ceases and the chains are immobilized. This evidence demonstrates
ong coupling between the polymer chain motion and the mobility of
ium ions.
A rule of thumb for the design of a polymer electrolyte is therefore to
t crystallization and to reduce the tg as much as possible. The larger
ference in temperature between the operating temperature and the tg,
re vigorous is the segmental motion. Unfortunately, both crystallinity

gh tg are linked to good mechanical properties (shear modulus, tensile
h, creep compliance), which are necessary to prepare practical
ls that will act as separators as well as electrolytes. A major goal of
r electrolyte design is to de-couple the ion transport from the
nical properties.
. Effect of Polymer Molecular Structure (Architecture)
EO :- $4-8/lb
MO :- M.Wt. -150k
ymethylene-linked PEG400
9. 3. Linear chain polymers.
A first approach is then to alter the structure of the polymer to remove

stallinity, at least in the temperature range of interest. Figure 19.3
the chemical structures of commercial PEO, PEMO (also known as
hylene-linked polyethylene glycol 400 or amorphous PE032 ) and a
(PPO =
33
PO/AGE co-polymer patented by HydroQuebec
pylene oxide; AGE = allyl glycidyl ether). These are all linear chain
rs and the latter two polymers exhibit significantly less crystallinity
wer melting points than PEO. In fact, the polymer complexes with
exhibit no sign of a melting transition in DSC measurements down to
. The introduction of the oxymethylene link, the propylene oxide
r the allyl ether group prevent crystallization due to the irregularity of
ins that disrupts the packing of the chains into regular forms necessary
Chapter 19
stallization. The glass transition temperatures of these linear polymers l

less than -65°C and hence they are viscous liquids with moderate
nical properties for high M.Wt. (> 500k) samples.
All of the polymers shown in Figure 19.3 may be cross-linked after
ormation by reaction with radical species such as those provided by
or UV radiation. However, this is quite uncontrolled chemistry that
sult in a non-uniform material as the reactive radical intermediates are
unselective. Over reaction results in a high density of cross-links, g
to an increase in tg, reduced segmental motion and lower
ctivity. Non-uniform cross-linking (or curing) yields non-uniform t
densities through the membrane which will lead to rapid failure (e.g. te
growth). The allyl ether group, however, may be selectively reacted
34 35
ydrosilyl cross-linkers ' to provide uniform films from the more
lled and selective chemistry. It is worth noting that commercially
ble PEO contains butylated hydroxytoluene (BHT) to inhibit attack on
lyether chains by radical species formed from oxygen or light. Higher
grades of PEO are actually prepared by radiation cleavage of very
6
.Wt. PEO (5-8·10 ) which illustrates the sensitivity of these materials
ation.
Polymer backbone flexibility decreases n=2-6
9. 4. Comb Polymers. Optimum side chain length= 5-6 EO units.
Figure 19.4 shows a series of comb polymers with different backbones

e chains containing ethylene oxide (EO) units similar to those in PEO.
aterials are completely amorphous at all temperatures. The optimum

of the side chains is found to be between 5-6 EO units due to the
ance of crystallinity as the length of the side chains grows and
les PEO. For shorter side chains the influence of the backbone is
er and leads to lower conductivity. The polymer backbone influences
nductivity due to the stiffness (polystyrene) and polarizability
crylate). In these cases short side chains (2-4 EO units) lead to
36
d conductivity. '37 The PEPEx and PVEx polymers show less
38 44
vity to the side chain length as their backbones are more flexible. -
45 47 48
osphazene - and polysiloxane ' 49 backbones exhibit similar
50 53
ties as do highly branched polymer architectures. - The tg values for
mb polymers are lower than for the linear polymers and their
nical properties are much more liquid-like due to the molecular shape.
The conductivities of the polymers in Figures 19.3 and 19.4 are shown
re 19.5 as a function of temperature for polymer-salt complexes with
TFSI at a salt concentration of one lithium ion to 20 or 30 oxygens in
ymer. One can immediately note the lower conductivity of the PEO
lyte at low temperatures due to the crystallinity. The linear amorphous
r solutions show higher conductivity at low temperatures consistent
e lack of crystallinity while the comb materials show even higher
tivity at low temperature, which is consistent with greater segmental
due to the more freely moving side chains that have free end
1
.5 All of the materials show the familiar convex, bent shape that has
F form related to the polymer chain motion. For rechargeable lithium
s, acceptable performance for electric vehicle and typical consumer
ao•c so•c 20"C -2o•c
-PEM0-20
-3 .......................,...
, -<>- PVES-20
; ··•··PEPE5-20
j ---PEO-LITFSI [30:1]
················ • -HQ data-30 TFSI
PEO = Polyethylene Oxide
PEA= Polyacrylate ether
-5 .........................,... PVE = Polyvinylether
PEPE= Polyepoxide ether
-6 ..................... .
PEMO = oxymethylene
-7 '
[
'
i
linked PEG400
-a crystalline PEO ' HQ =EO/PO copolymer
0.004 0.0045
0.0025 0.003 0.0035
1fT
9.5. Comparison of conductivities of comb-branch polymer electrolytes (20: I LiTFSI)

ar and networked co-polymers (HQ data, USP #5,755,985).
Chapter 19
4
ations requires a conductivity of between 10- and 10-3S/cm. Thus PEO
ms adequately at 80°C but no system provides acceptable transport
ties at ambient temperatures. In order to provide adequate performance
temperatures it is important to elucidate the factors that control the
dence of transport properties on temperature.
3. Effect of Polymer Solvation Structure
The foregoing results provide an example of how change of the

ular structure may lead to improved ion transport through an increase
polymer chain mobility. Much research effort has been expended in
nthesis of novel polymer structures to increase the lithium ion mobility
o true, dry polymer electrolyte has been shown to provide better
mance than the comb materials shown here. The effects of the
cture change upon the conductivities are generally greatest at
ratures below 40°C. Above this temperature one may note that all the
als exhibit nearly identical conductivity dependence with temperature.
s Arrhenius dependence and is consistent with a process with an
ion energy involving the making and breaking of chemical bonds.
all of the polymers use ethylene oxide units (EO) to solvate the lithium
seems reasonable that the rate determining process involves breaking
rming oxygen-lithium bonds that occur in the same way for all
ers containing EO solvation units. Thus, at low temperatures close to
0
o ·-, 4-:o
l'·-\
,#'\,.
·'-A h .
.r' ' rr emus
Controlled by Control
Segmental Motion -related to
- related to Tg solvation
9.6. Schematic contributions of segmental motion and solvation to ionic mobility.

ss transition the segmental motion of the polymer chains is the slow

s. At higher temperatures this motion increases to a point where the
gement of the solvation sphere becomes rate determining. This is
ted schematically in Figure 19.6 which shows an exaggerated motion
mb polymer side chain combined with the solvent rearrangement as
n moves from one chain to the next. Recent neutron scattering
54 56
ments have provided support for this scheme - through the
cation of two distinct types of relaxation.
The mechanism of the solvent reorganization around the ion is of
importance as it controls the height of the activation barrier to
ent of the bound ion and hence the slope of the temperature
ence of conductivity. In order for the ion to move the solvated ion
through an activated complex. This may be formed by dissociation of
gen-lithium bond to form a less coordinated complex followed by
ion of a new u+-0 bond to a different point in the chain or on a new
An alternative mechanism would involve an increase in the
nation number in the activated complex with either the same chain or a
chain (or segment). The rearrangement involving multiple chains will
er as a result of the decreased entropy of the intermediate complex in
e of either mechanism and may imply that chain to chain transfer is
ermining.
f the nature of the solvation is rate determining for ion mobility there
be a change in the temperature dependence of the conductivity upon a
of the groups involved. Figure 19.7 shows a comparison of a number
ymers complexed with LiTFSI. In particular, those that contain
ene oxide units (CH2CH(CH3)0), carbonate (OCOO) and trimethylene
(TMO, CH2CH2CH20) appear to show different temperature
ences at temperatures above 40°C where segmental motion is not rate
ining. In comparison, a liquid or gel electrolyte shows less pronounced
ature dependence in this range than any of these electrolytes due to the
ar mechanism of ion transport. For liquids and gels the solvent
les travel with the ions so the bond breaking and forming is less
ant to the ion transport.
hese results indicate that there is scope to improve the ambient
ature performance of polymer electrolytes by variation of the
ng groups. However, this involves considerable synthetic effort to
ce new groups such as TMO, carbonate or new donor atoms such as
n, sulfur or phosphorus. The use of molecular modeling is an
ble aid to guide the synthesis direction since it may eliminate
ising structures before an extensive effort is made to prepare
riate polymers. There is an extensive literature on different levels of
ng for polymer electrolytes. 10 •17 •57 -62• Among this work are some
tions relevant to the polymer structures whose conductivities are
Chapter 19
UTFSI salt (20:1) PPO units, carbonates

-2 .-=-+- ;,p;,,,J,; and acrylates trap U ions
-e- PESA(20:1]
and lead to lower rmbility.
-3 EO-polymers PEM0[20:1]
--><-· PEPE5{20:1]
1MO units appear to lead to
-4 • • '• • PCE3E(20o1)
higher rmbility
• ,..._ PP3A(20:1]
·5 _,_PEPP3(20:1)
1P1 PEPP3
·10
0.0025 0.003 O.OOJS
1fT
19.7. Temperature dependence of conductivity for polymer electrolytes containing

t solvation groups.
in Figure 19.7. Quantum chemical calculations have been carried out

mate the binding strength of lithium ions to polyalkyl oxides as a
on of the coordination number (1-6) and the nature of the solvating
63
(EO, PO and TM0). These calculations show that the binding energy
TMO groups are higher for TMO than for PO and EO at coordination
rs up to 6 when the steric crowding of the TMO forces a longer Li -0
ength and the binding energies drop below those of the EO and PO.
s consistent with early literature reports that polyTMO polymers did
64
solve lithium salts. Methods of estimating the height of the activation
for movement of the lithium ion along the polymer chain have been
ped based on a mechanism that involves an intermediate with a lower
65 66
nation number ' and a comparison of polymer structures containing
d TMO groups indicates that the barriers are lower for the TMO
67
ers. Figure 19.8 illustrates the movement of the lithium ions along the
er chain where the arrow points to the lithium ion and illustrates the
d of calculation.
The theoretical calculations indicate that the migration of the lithium
ong the polymer chain will be faster for a TMO polymer compared to
polymer. Comb polymers have been prepared with varying contents of
..
d TMO units and the temperature dependence of conductivity, the
ransition temperature as a function of salt concentration are shown in
'
F U ··t Jg ·g·r
•
28.2
(!Ui)
1 -·A , 1 .
6
I U "-t (wtcr)
7
'
G U ·-tgt "
E U ·., JJl • .t:.t (ccnlcr)
9.8. Calculated reaction pathway for migration of the lithium cation along a
lene oxide (PEO) chain. The PEO is modeled by diglyme and the cation moves by
and breaking Li-0 bonds. The energies are in kcal/mol from HF/6-310*
67
ons.63 •
19.9. The polymers in this case are the polyepoxide ether polymers
ifferent proportions of EO and TMO units in the side chains which
67
om all EO to all TMO units. ' 68 It can be clearly seen that as the
ion of TMO in the polymer increases the low temperature
tivity increases and the dependence of tg upon the salt concentration
ses. However, it is very striking that the temperature dependence of the
tivity above 40°C is unaffected indicating that the motion along the
s calculated is not rate determining.
There are a number of significant differences between these comb
rs (presence of PO units in the backbone, chain ends with OH groups)
linear polymer chain used for the calculations but it appears that the
nism used in the calculations does not dominate the rate.
ements of the linear PTMO-LiTFSI complexes show a slightly
Chapter 19
wer slope but no overall increase in conductivity over the linear PEO
s which indicates that the differences in the comb structure are not
cant. The decreased dependence of the tg on salt concentration
ed for both comb and linear TMO polymers indicates that the ionic
linking between chains is reduced with TMO. Although this does not
the problem of how to obtain high conductivity at room temperature, it
important result as the TMO structures may be used to mitigate the
of salt concentration gradients that occur in lithium-polymer batteries,
larly in the presence of surfaces such as electrodes or ceramic filler
es.
Conductivity of Polyepoxide Ethers
different contents of TMO and EO units.
Salt is LiTFSI at 0: Li 10:1 ratio

t
TMO,
-20
T
g
of Polyepoxide Ethers
as function of salt concentration (LiTFSI)

1
q/ <.\ ............;. "'·""f qcrM
. ... j , '-
\
3
-3o 1 · :.;;.:.ri>ei'MQEioi'iVio
........ --fPE(TM6Eo>2
Me
-40
I• u• ncrea·.· se .,· · "•e·PETMd,!E03 ..
.:............-.--· ......L..........L.. I!! '. fMO ratio - PEEC>3.

< • f'>Yi\'it::J i"!
-50
.L PET OEOTMb 8.
'j: :l:I£:> -:.:

E
'PE(TMOE0)2i -60
-
-70
-- L- ....
026 0.003 0.0032 0.0034 0.0036 0.0036 0.004 -80 20 40 60 80 100 120
0
1FT O:Ll
9.9. Conductivity as a function of temperature and glass transition temperature as a

of Li TFSI salt concentration for comb polymers with structures like the PETM03
ut with some or all of the TMO units exchanged for EO.
The results of these combined synthetic and theoretical efforts have yet
blish the physical phenomena that limit the mobility of the ions.
er, the possibilities have barely been explored as there are many
res such as carbonates, polyethylene imide (PEl) and polyalkylsulfide
1
that may be synthesized to test the limitations of these electrolyte
s. To date, the barrier to ion transport remains undefined in the simple
s. Efforts have been made to investigate the influence of different
such as the use of block copolymers 69-74 that provide different
5
nments for the solvated ions, rigid polymer systems such as glassi or
76
line electrolytes that contain defects as well as "polymer-in-salt"
77
s. These approaches show promise to provide information that will
te the nature of the limiting process in the ionic motion although they
oduce further complexity into the system. The presence and effect of
nt phases on the behavior of the polymer electrolyte is of great interest
y for the effect on ionic mobility but also for the understanding of the
cial behavior at electrode surfaces.
. Effect of Surfaces on Polymer Electrolyte Behavior
There has been much interest in the behavior of polymer electrolytes in

ation with fillers such as nano-particulate fumed silica, alumina and
78 84
eramic materials. - Most of these studies have been carried out with
M.Wt. PEO and have reported improvements in conductivity,
rence number and interfacial behavior with lithium metal electrodes
ddition of the ceramic fillers. The effect of the fillers has been shown
artly due to a suppression of the crystallinity of the PEO electrolyte
reases in conductivity have been observed at temperatures above the
point of the polymer system which indicate some different
nism is in play. It was noted many years ago that the method of
lyte preparation and the presence of fillers could alter the mechanical
85 86
nsport properties ' and the presence of solvent impurities and the
l history of the semi-crystalline materials can lead to wide variations
measured properties. Unfortunately, most of the studies are
cated by the fact that commercial PEO already contains about 3%
silica plus some residual calcium salts from the manufacturing
. The fumed silica is added as a desiccant and binds impurities such as
nd other solvents to its surfaces.
The observation of increased ion transport with added nanoparticles at
temperatures is not universal. Johansson and co-workers observed no
e upon the addition of fumed silica to LiTFSI solutions in PEO and
84
. Their observation is more consistent with reports in the rheology
re that would lead one to expect a decrease in ion transport properties.
mple, Tsagaropoulos and Eisenberg have reported on the observation
cond glass transition at higher temperatures upon addition of fumed
87 88
nano-particles to various polymers. ' They have ascribed this
r to the inhibition of polymer segmental motion by interactions
n the polymer and the particle surface which leads to the development
obile layers. As the proportion of filler particles increases the
ile layers overlap leading to a distinct phase from the polymers in the
the material and hence to the observation of a second tg at values as
Chapter 19
as 100°C above the first one (see Figure 19.10). The second tg was
ed using dynamic mechanical analysis but the behavior has been
med by recent neutron scattering measurements which showed the
89
nce of two distinct relaxation processes in the presence of the fillers.
ehavior is dependent upon the nature of the polymer-surface
tion, the polymer molecular weight and any cross-linking, be it
nt or ionic as in the case of a polymer electrolyte. It has been reported
or high M.Wt. polymers the second tg is not observed but the tg
ses due to the restriction of the polymer chain motion by multiple
90
es. As has been noted by Eisenberg these effects are not confined to
olymer systems but also play a role in polymer gel systems and hence
ely to influence transport properties in composite electrolytes and
des in lithium ion and lithium-polymer gel batteries as well as solid
polymer electrolytes.
>50%, no Tg
© © 0 ooo
o0oo
0 = filler particle
= immobilized
© © <10% polymer
=restricted
1 oneTg r polymer
©©© oooo
oo 0 >20%, smaller
©© -10%,
two Tg's upperTg
9.10. Eisenberg model of the effect of nanoparticles on the mobility of polymer
The effect of surfaces and thin film geometries on the properties of

91 97
ers . is of general interest to materials scientists due to the relevance
properties of composites, laminates and the field of adhesion.98•99 In
l these effects derive from the inhibition of the polymer motion by the
e either through formation of some form of binding or by geometric
ion of some modes of motion. This accounts for the inhibition of
lization and may in some cases result in an increase in mobility of the
er chains close to the surface compared to the bulk. The nature of the
tion between the surface and the polymer may lead to increases of the
ic Electrolytes: An Overview
+
589
,......., 10-3 +
+ Ultra-pure
<> Lab-pur ied
b
• Hydrophobic Si02 filled
Hydrophilic Si02 filled
E
.!::!
<> +
+ (a)
• •
<>
'-"
• ••
(I) <>
+
<>
<> +
>- 10-4 ... +
... ...
=ti:::; ...
"0
c
... ...
u ...
+ _.
0
0)
0 10"5 b
...J ...
...
--- •
... <>
10-6 I I
2.7 2.8 2.9 3 3.1 3.2 3.3 3.4

1
1000!T (K- )
E'
(b)
0
10 Hz
19.11: (a) Conductivity; (b) compression elastic modulus of low molecular weight
P(EO)z0LiTFSI at 80°C. Amount of fumed silica added is 10%.
Chapter 19
motion and lower tg values if the surface is appropriately treated to

92
ize the polymer-surface interaction. On the other hand very strong
er-surface interactions can lead to the formation of an immobile layer
surface. If this surface happens to be an electrode it implies that the
nism of ion transport through the surface layers at the electrode is
nt from the bulk polymer where segmental motion predominates. The
of nano-particle fillers may therefore be useful for understanding some
complex processes that occur at the electrode surface in the SEI layers.
A recent study of the effects of added nanoparticles upon the ion
ort and mechanical properties of PEO electrolytes has provided results
llow reconcilement of the apparently contradictory mechanical and
ctivity measurements. ° 10
Firstly, PEO was purified to remove the
silica added by the manufacturer to provide a true measure of the PEO
ties when complexed with lithium salts. This material was compared
ommercial PEO containing the 3% fumed silica (lab-purified) and with
ercial PEO containing added nanoparticles fillers (hydrophilic Aerosil
and hydrophobic Aerosil R805), 10% by weight. The conductivities are
against temperature in Figure 19.11a for the LiTFSI solutions in these
amples while Figure 19.1lb shows the compression elastic modulus as
tion of frequency at 80°C for the same LiTFSI-PEO samples.
The results clearly show that the conductivity is higher in the absence
oparticles while the modulus is lower. Crystallization occurs below
as seen by the break in the conductivity curve in Figure 19.11(a) and
esence of the filler materials does not suppress this. The compression
us increases and the conductivity decreases as fillers are added in these
ments where care was taken to exclude moisture. A striking result,
er, is that the addition of the hydrophobic filler material provides
as much increase in mechanical strength as the hydrophilic material
conductivity decrease is much smaller, indicating the different effects
surface groups. In this case the hydrophilic filler has SiOH groups on
rface while with the hydrophobic group the surface OH groups are
with octyl groups to provide the hydrophobicity. The results illustrate
ects of the polymer-surface interaction upon the segmental motion and
case the ionic motion appears to largely avoid the hydrophobic surface
It might be noted that similar interfacial effects might be expected
iblock copolymers that promote microphase separation and generate a
on-polar interface.70 '72-74 '101
Addition of impurities such as water through exposure of the samples
osphere has been shown to have an effect on conductivity for certain
ystems. In the case of the LiTFSI-PEO system exposure of the
ranes to atmosphere did not result in appreciable changes in the
ctivity in the presence or absence of ceramic fillers. However, for
triflate and lithium perchlorate systems such exposure led to an
e in the conductivity particularly in the presence of the ceramic fillers.

fect appears to be greater at low temperatures where crystallization is
ed although increase of conductivity was observed above the melting
Clearly, adventitious water is absorbed by the filler material to create a
n the surface of the particles that can provide a pathway for the ions to
more easily than through the bulk of the polymer. Addition of filler s
such as activated alumina have long been used to scavenge residual
102
n organic solvents and electrolytes ' 103 and this desiccant effect is
the source of the reported improvements in interfacial behavior with
electrodes. The concentration of the water and other polar impurities
filler surfaces provides a physical basis for the proposal of Kumar that
83
a change in dipoles at the interface and is consistent with some of
104
er observations concerning Lewis acidity of the surfaces 78' •
These composite and multiphase systems are intrinsically interesting
insight they provide into ion transport mechanisms in solid polymer
lytes. Studies on this behavior also provide valuable information on
e electrolytes behave next to electrode surfaces and how the composite
de structures should be formulated to obtain the best mass transport
ons in the electrolyte in confined spaces. Since similar interfacial
arise with gels and liquids the issues raised here for the solid polymer
s are relevant for gel polymer and liquid lithium ion systems as well as
ll electrodes where the desired power densities and rates are very
igher.
. Gel Polymer Systems

The introduction of small molecule additives or plasticizers into a solid
r electrolyte system leads us into the area of polymer gel electrolytes.
are electrolyte systems that consist of a polymer matrix combined with
nt system. The salt may be dissolved in the solvent or may be attached
polymer framework as is the case in the familiar polyelectrolyte
ane, Nafion®, used in chlor-alkali electrolyzers and in fuel cells. Two
105 106
have recently appeared that describe polymer gel systems • of the
type where polymer matrices selected from PVDF, PAN and PEO are
form a gel with liquid electrolyte systems that are commonly used in
ion systems. Two types of polymer gel battery systems are generally
nvestigation. One is where a polymer gel is sandwiched between
ion battery electrodes where a different polymer (PVDF) from the
107 108
or is used as binder in the electrode. • The second type is often
to as the Bellcore system and refers to a porous polymer gel system
d from copolymers of vinylidene fluoride with hexafluoropropylene,
HFP, and typical lithium ion battery electrolytes, LiPFJEC-DMC. 109 -
Chapter 19
his case the polymer binder in the electrodes is identical to the polymer
separator and the co-polymer formulation is chosen to reduce the
linity and optimize the uptake of solvent into the separator and
des. These gels often contain fumed silica particles to provide
114 116
ved mechanical properties. - In fact, gel electrolytes with excellent
nical properties and liquid-like transport properties have been prepared
nly the presence of fumed silica particles which form a network
re that contains a polyethylene glycol dimethylether (PEGDME)
n of LiTFSI salts. ll7-!2o These systems show great promise as they
ne excellent mechanical properties with high rate capability and the
ial for good chemical and electrochemical stability with the electrodes
providing enhanced safety, lower cost and more flexibility in
ing than a conventional lithium ion cell.
The operation of the gel polymers is dependent upon the presence of
vent molecules that travel with the ions just as occurs in a liquid. Thus,
nductivity and salt diffusion coefficients approach those of liquid
111 121 122
lytes (>10-3S/cm, 10-7cm% at room temperature). ' ' The
ers and ceramic networks provide the mechanical strength by forming
ks through which the liquid is able to migrate. This network formation
nction of the properties of the polymer or particle surface and how it
ts with the solvent cluster as it moves through the matrix. The best
system for this is the presence of clusters and multiplets that form
ar pathways in polyelectrolyte gels as is well known for Nafion® and
3 123
polyelectrolyte membranes. ' As such these solvent-network
tions and gel-electrode interactions are simply extensions of the issues
ered above for the effect of surfaces on dry polymers and consequently
of the considerations are similar. The ability of the network system to
etely contain the liquid and prevent leakage is of great importance as is
tential for the gel to retard reactions, prevent thermal runaway and
ce safety. From a manufacturing point of view polymer gel systems are
rest as the gel precursors can be introduced into the cell as a liquid and
in place, thereby greatly simplifying the manufacturing process. There
wever, a number of drawbacks with this procedure that will be more
nt after discussion of the design needs of practical lithium batteries in
t section.
POLYMER ELECTROLYTES IN LITHIUM

BATTERIES
Since the intense interest in polymer electrolytes is nearly always
d by the use of these materials as separators in lithium batteries, it is
important to bear in mind the requirements of this application. Our

atic of a lithium battery in Figure 19.1 may be elaborated upon as
in Figure 19.12 to provide us with more insight into the conditions
e polymer electrolyte is expected to operate under. Inspection of the
shows a lithium metal anode in contact with a polymer electrolyte
the polymer chains are represented by the thin lines and the salt ions
persed through the electrolyte. The lithium metal thickness and the
lyte separator are each on the order of 30-50JLm thick. The composite
e is represented by the mixture of large cathode particles (vanadium,
nese or cobalt oxides) which are about 1-10JLm in diameter, carbon
anoparticles (10-lOOnm diameter), polymer chains and salt ions in a
hat is about 50-lOOJLm thick. The purpose of the carbon nanoparticles
rovide an electronic conduction path from the current collector to the
e particles. The polymer electrolyte has to function as a binder for the
e particles so that they remain in electronic contact as well as to
e a pathway for ionic conduction through the electrode to the surfaces
cathode particles. In the case of a gel polymer the ions with their
panying solvent molecules must travel through the confined spaces
n the electrode particles while for a dry polymer system, the polymer
Ions plus polymer
= Conducting = cathode particle

carbon particle
9.12. Schematic of a lithium polymer-metal oxide battery showing the environment

peration of the polymer electrolyte.
tal motion must be maintained in the restricted spaces to provide

obility or some other mechanism of ionic transport may operate.
onsideration of these requirements immediately provides some
tives on the approaches that have been used to improve the
Chapter 19
ctivity of polymer electrolytes. The barriers to ionic transport occur

110 111 124 125
in the composite electrodes than in the separator. Hence,
• • •
n immediately see the difficulties for the introduction into a composite
ode of glassy polymers,75 composite systems with more
es78'82 '83 '115 or regular structures such as layered 72'74 or crystalline 30' 76
. It would be unreasonable to expect similar behavior in the composite
ode as is observed in the bulk separator. As we have seen from our
sion of the effects of surfaces on polymer dynamics, this is equally true
olymer electrolytes that depend on segmental motion or vehicular
n of solvent. While battery engineers continue to use composite
des this will be a challenge. However, Sadoway and Mayes have
126
d out the possibilities of thin-film battery systems and these novel
s may become more practical as large-area, thin-film technology is
ped for applications such as organic and polymer light emitting diodes
127
ea lighting for example. In either composite or thin-film electrode
s the electrolyte will continue to have an interfacial region and
ore the study of composite polymer electrolytes will continue to
e invaluable information.
In addition to the physical changes that the electrode surfaces may
e upon the electrolytes one must also consider chemical changes that
due to interactions between the electrodes and the electrolytes. In the
f lithium metal one can expect cleavage of carbon-oxygen bonds to
leading to lithium alkoxides and alkyl-lithium compounds on the
0 26 128
de surfaces.Z If these lithium compounds remain connected to
• '
er chains the resulting strong interaction will lead to immobilization of
lymer chain close to the surface. The complex surface reactions are
l for the formation of effective SEI layers at both lithium metal, carbon
20
loy electrodes and indeed there is evidence that the salt anions may be
129
eactive than the polymers under some circumstances. Corresponding
ion chemistry is expected to occur at the cathode surfaces. Ethers are
132
to be oxidized at potentials less than 4 V vs. Li130' and, while
ate solvents appear to be electrochemically stable to greater than 5
receding chemical reactions may occur to yield more easily oxidized
21 134
s. • In any event, side reactions that involve oxidation of electrolyte
cathode will likely generate hydrogen ions that may initiate further
ns in the organic medium which can lead to further bond formation or
inking. These reactions can drastically change the nature of the
lyte-surface interaction and as we have seen this can have significant
on polymer dynamics close to the electrodes.
The surface effects combine with the mechanical and transport
ties of the bulk polymer electrolyte to provide a complicated system
best approached from a system model point of view. 125 Consideration
w the system behaves in the application can be related back to the
ular models of transport and reactivity to help guide the selection and
of the appropriate materials that can provide the desired improved
68
mance.
1. The Effect of Transport Properties on Cell

Performance
In the previous discussion on lithium ion mobility, the conductivity

imarily used as a measure of the effects of the various structural s.
To properly describe the transport properties of an electrolyte ing
n components requires knowledge of (n/2)(n-1) transport ties.
For a binary salt polymer electrolyte system these properties are tivity
0
(cr), salt diffusion coefficient (Ds) and transference number (t+ the )
transference number is defined as the number of moles of ion uent
crossing a reference plane fixed with respect to the solvent when
135
araday of current is passed. For a gel electrolyte six transport
ies would be required while for a polyelectrolyte only one is required,
nductivity. The impact of the transport properties results from the
Limiting thickness of cathode
.:-'._:·:"!-".
...
I
I
.•
:;_;;:;f \
I
I
I
.·".""•r,_·. ; > V6013ca:thode
ium
) !1
: t?e; p,,---
- '·- ·- >
./:\;:::
9.13. Effects of transference number and salt diffusion coefficient on salt
ation profiles upon Li-polymer cell discharge.
de processes in the cell. In most lithium ion battery systems only the
ion undergoes any reaction at the electrode while the anion does not
o a reaction. Thus, when the lithium ion is removed from solution by
Chapter 19
alation into or deposition on to the electrode, the anion will move

ds the other electrode and tend to accumulate there. This will set up a
ntration gradient next to the electrode which will be relaxed by
on. A system with a large salt diffusion coefficient such as a liquid
lyte will be able to relax the concentration gradient sufficiently to
t concentration polarization from becoming a problem. The higher the t
density or power requirement, the larger this needs to be in order to t
voltage losses due to concentration polarization. The development of
oncentration gradients in the lithium-polymer battery is shown
The figure shows the development of an increase in salt concentration
anode with time of discharge while within the cathode there is a
ion of salt concentration until it reaches zero. This limits the
ibility of the cathode towards the back of the electrode. These
ses are complex and have been exhaustively analyzed by the Newman
25 136 141
• - and the reader is referred to this body of work for detailed
ation. On recharging the battery the salt concentration is depleted next
lithium electrode while the concentration increases occurs within the
site electrode. The only way to avoid these concentration gradients is
mobilize the anion on the polymer to form a polyelectrolyte where
The system modeling uses concentrated electrolyte theory and porous
135
de theory to predict the behavior of the batteries and to successfully
plish this requires knowledge of the transport properties in the bulk of
parator and also within the composite electrodes. Rigorous methods for
rement of transport properties in the bulk of the polymer electrolyte
tor have been developed121.1 22' 142- 144 and these have been used to
142 145 148
e values for a number of important systems. ' - The values of the
rence numbers and salt diffusion coefficients are found to vary with
ncentration and temperature for any particular system and the values
PEMO-LiTFSI system are shown in Figure 19.14. It has been found
e values are affected by phase changes in the electrolyte at high
trations and by side reactions with the lithium metal electrodes at low
trations. However, the values in the concentration range 0.5-2.5 M
to be quite usable as they accurately predict cell behavior (e.g.
g current at various temperatures). These numbers were used to model
124 125
hiurnlpolymerN6013 system ' and to predict the energy and power
y of the system. The analysis demonstrated that the transport properties
PEMO-LiTFSI system at 85°C were adequate to obtain the energy and
density that is desired for electric vehicle use but that performance at
90
'
•••••
10·11
•20 • i ,a
• 4
D D •
<>
IJ
D
C]
•D D /85
D /00
10·12 <)
<> D /40
<>
0 D
<>
"
[J
f t
10·11 0 0.5 15 2 2.5 3 3.5
Conc./M
0.5
0.4
• 1+85 c
.l I+ SOC
[J
{) <> l+40C
0.3
••
0.2
" [J
<}
lijl
j • • •
f
. [J
<}
+...
[J
0.1 [J
" D
1:1
(>
<> <>
-0.1 0.5 1.5 2.5
Conc.IM
9.14. Salt diffusion coefficient (D,) and transference number (t+ 0) for LiTFSI
s a function of salt concentration and temperature (40, 80 and 85°C).
Chapter 19
temperatures was seriously inadequate. It is from analyses such as

3
hat the conductivity goal of polymer electrolytes of 10- S/cm at room
rature has evolved but it is necessary to measure all the relevant
ort properties to correctly predict battery performance. From the earlier
sion on the effects of surfaces and nano-particles on the polymer
ics, it is not unreasonable to expect that the transport properties of
lymer electrolyte may be altered within the composite electrodes and
his may lead to performance that is different from the model
tions, particularly at high rates where the diffusional relaxation may be
cantly altered by the presence of nano-particles. Some hint of this
menon has been noted in the comparison of the performance of
re gel polymer batteries with model predictions. l1l Reduction of the
ffusion coefficient by a factor of two or more was necessary in order to
model predictions to the high rate behavior. A similar comparison of
predictions with discharge and charge behavior has not yet been
eted for lithium-polymer systems. However, examination of the cell
g behavior by current interruption techniques indicates that the mass
ort effects are very significant. Figure 19.15 shows the voltage profiles
harging and charging of a lithium metal/PEO-LiTFSIIV6013 while the
t is interrupted for one hour after ten minutes of current flow at 0.1
2
m • The insets in the figure show the relaxation of the cell at the top of
and at the bottom of discharge. One can observe significantly larger
;
TT
:····:·:;
!•
r
:· ······,·· j 51"/.PE0301-12TFSibypulseSieppolarlzatlon
il?t.m:···, curri.rno!"si':: = cer

,,, , .•....., 35 votiageumit::c-a.25V,.r ture: ·c.
.....
2.6
32.5 33 33.5 34 34.5 35 35.5 36 36.5
; ; ; ; ; · file name:;2y111025
1 1i\m t.i'-5 3 3.5 4 'Tlme(hr)
ng Ohmic drop, Comtant Ohmic drop,

Increasing '11.
creasing '11.
:
. .. . '
..
22 23.5 24 24.5 ··-1--TI···t·····t··..+-···1····+
2.7
22
·
2.1
24 24.5 25 25.5 26 23.5 Zl 27.5 28

ant Ohmic drop, Trre(l'r)
omtant TJ.
Comtant Ohmic drop,
Increasing '11.
9.15. Pulse polarization of aLi metal/PEO-LiTFSI/V 6013 in discharge and charge
The cell is charged or discharged for 10 minutes followed by a 1 hour relaxation.
tration polarizations at the top of charge which is consistent with the

e of salt concentration within the cathode structure that is relaxed
lowly due to the change in transport properties within the composite.
scharge capacity declines rapidly under these cycling conditions. The
tion of capacity fading with poor transport properties has also been
149
strated for a number of electrolytes with manganese oxide cells and
mulation of the composite electrode system has to take into account
otential effects of surface-polymer interactions on the polymer
ics. It has been noted that modification of the carbon particle surfaces
150
EG chains leads to improved cycle life and it is common practice
149
ude surfactants in the cathode mix that improve the cycle behavior
ably by modifying the carbon-polymer interaction.
mpedance measurements taken during the course of the cell cycling
strates steady increase of impedances (see Figure 19.16) that are
lly identified as the interfacial or charge transfer impedance. Modeling
151
impedance response of cells is a complex subject, especially as the
se is a mixture of the two electrodes unless steps are taken to introduce
ence electrode. The response can be controlled by solid-state diffusion
ctronic conductivity within an electrode, ionic diffusion in the
lyte phase or interfacial, charge transfer kinetics. In Figure 19.16 one
serve changes in the impedance as a function of state of charge and the
pment of several impedances which may be related to processes going
ither or both electrodes. Even more interesting is the observation of
ence of the impedance on the time that the measurement is taken after
is turned off. Figure 19.17 shows the impedance plots for a
/PEO-LiTFSI/lithium cell that is polarized five times in the same
2 2
on at 0.2 rnNcm for 2 coulombs of charge/cm • The impedance of the
taken 30 seconds after the current is turned off (a) and then again after
ur (b). The "charge transfer" impedance is observed to change with
ndicating that some of this impedance is due to salt concentration
ts close to the electrode surface. In some cases the impedance actually
es with time while in other cases it increases. It can be observed that
pedance appears to return to a similar value indicating that no
sible process has occurred that would indicate growth of the SEI layer
h chemical reactions or the growth of dendrites. In this case, the
2
density used (0.2 rnNcm ) was well below the limiting current (>2
2
) as calculated from the transport properties and cell geometry. It can
be noted, from the time dependence of the impedance spectrum and
reasing polarization observed on passing current, that one hour is still
ugh time for the concentration gradients to completely relax.
These observations illustrate that the performance of lithium polymer
s is dependant upon the behavior of the electrolyte at the interfaces.
ted previously, the polymer dynamics in composite systems is
Chapter 19
nced by the nature of the electrolyte-surface interaction and by the

tion of movement that the presence of the surface causes. Even in the
of the planar lithium metal electrode we have observed significant
s. The salt concentration gradients that build up on the electrodes can
o depletion on one electrode and very high concentrations that can lead
precipitation on the other. Effects of high concentration may also be
ng before actual precipitation occurs. Figure 19.9 shows that as the salt
ntration increases the glass transition temperature of the polymer
400
• 1imp
1imp:before first charge
350 2imp:after first charge

D 2imp
3imp: after first discharge
300 0 3imp 4imp: after second charge
X 4imp Simp: after first Scycles
E 250 + Simp 6imp: after second S cycles
'!
E /\ 6imp ?imp: after 20cycles
J: 200
0 0 ?imp
N 150 ..loooooo.,
)'" 1....10
\ cxD a. :c.fl
100
50
0
0 100 200 300 400 500 600 700 800
Z'ohm.cm
19.16. The evolution of impedance in Li/PEO-LiTFSI/V 6013 cells cycled at 100

•
'
D
1Z'1
2Z1
0 3Z1 "'su
4Z1
]
X
+ 5Z1
0
N
0 100 150 200 250 300 350 0 50 100 150 200 250 300 350
z Ohmcrn' Z Cllun.cm2
(a) (b)
9.17. Impedance of a Li/PEO-LiTFSI(1:30)/Li cell at 85°C polarized at 0.2 mA/cm 2

ulombs of charge. Impedance recorded after (a) 30 seconds and (b) one hour after
stopped. Five separate polarizations designated 1-SZ in order.
lyte increases. This effect is more severe for EO-containing polymers

even more pronounced for linear PEO and PPO polymers than the
polymers shown in Figure 19.9. The combination of high salt
tration and the interaction with the electrode surface indicates that the
lyte layer next to the electrode is likely to be very restricted in motion
actually a glass. This implies that the mechanism of ion transport h
these surface layers is different from the bulk and that these layers ute
to the interfacial impedance observed in the polarized cells.
tanding of how ions move such layers is of great importance and the
of ion transport in less mobile polymer systems takes on added
75 76
ce in this context. '
Figure 19.9 shows that the dependence of the glass transition
ature upon concentration is much reduced for the TMO-containing
ers and hence they would be expected to generate less interfacial
ance. This is observed at lithium metal electrodes where the TMO
polymers with LiTFSI typically give an interfacial impedance of about
2 2
m-cm compared to 100 ohm-cm for PEO-LiTFSI. As mentioned
although the introduction of the TMO groups was disappointing in
of producing an increase in the bulk conductivity of the polymer
lytes, the tg salt-dependence provides very valuable properties that can
greatly improved cell performance by reduction of the interfacial
i
; 100
0
11.
.II:
ca
Gl
11.
50
0 0.002 0.004 0.006 0.008 0.01 0.012
Anode Film resistance ·Ohm.m2
152
9.18. Dependence of peak power on anode film (SEI) resistance.
Chapter 19
ance. The importance of interfacial impedance on cell performance can

preciated by use of system modeling. In the Li/PEMO-LiTFSI/V 6013
124
calculations referred to above, the interfacial impedances were
nted for by use of a value for the anode film resistance which was taken
2 2
0.01 ohm-m or 100 ohm-cm . It was found that the peak power that
be obtained from the cell is very dependent on the value that is
ed. The results of calculations of peak power that varied this resistance
152
own in Figure 19.18. The decrease in interfacial impedance observed
lacing EO polymers with TMO would lead to a near doubling in the
ower capability. More sobering, however, is the effect of an increase
e interfacial impedance. Since reduction of the cell operating
rature to 25°C typically results in an increase in interfacial impedance
2
ater than 1 ohm-m even for TMO polymers, this indicates that useful
f discharge with a polymer electrolyte will not be possible even with
oom temperature conductivity unless the interfacial impedance problem
solved. Thus, the studies of the effects of surfaces on the transport
ties take on an even greater importance for all types of batteries and
lls.
2. Polyelectrolyte Single-Ion Conductors
The properties demanded of polymer electrolytes for good

mance in lithium batteries that we have discussed now include a
ding list. In addition to chemical and electrochemical stability,
ate transport properties are necessary both in the bulk of the separator
the electrode surfaces. This is difficult to attain if the electrochemical
ses lead to the development of concentration gradients at the surfaces
e transport properties are degraded at extremes of concentration. A way
id this is to use polyelectrolyte single-ion conductor materials which
0
ransference number t+ =1. With these materials no concentration nts
will develop as the anions are unable to move and accumulate on de
of the cell. However, the loss of anionic mobility results in lower
ctivity. The electrolyte portion of the cell is therefore limited by ohmic
nce rather than the concentration polarization. Here again modeling
es invaluable guidance for the design of materials and for targeting
riate performance goals.
The best known polyelectrolyte material is Nafion®, an ionomeric
polymer with a PTFE backbone and fluorinated ethylene oxide side
that are terminated with sulfonate anion groups. A number of similar
ated materials have been examined for gel polymer systems quite
153 155
y - where the ionomeric membrane is swollen with a solvent.
4
ctivities greater than 10- S/cm have been reported. Several attempts
een made to prepare lithium and sodium polyelectrolytes for use in
156 169
t free systems - and the most successful of these have the comb
re with a pendant anion similar to the structures shown in Figure
A typical comb structure will have several comb side chains with no
d anion for every side chain with an attached anion and this allows the
163 167
tration of ions to be controlled. • The best conductivities for these
5
ctrolytes approach 10- S/cm at room temperature. The nature of the
has been shown to play a role through its ability to delocalize the
e charge and hence reduce the strength of ion-pairing to the lithium or
ions. Strong ion-pairing as occurs with the alkylsulfonate anion tends
the cation and slow its motion through the electrolyte. Replacement
sulfonate group with fluoroalkylsulfonates leads to a significant
163
e in conductivity.
<±>
i
I
O c" s-' 's-'CF3

nF2
9.19. Examples of pendant anions on comb structures for single-ion conductors.
Molecular modeling has been carried out on polyelectrolyte systems

170 171
Monte Carlo methods. ' The results of this modeling imply that the
ype systems are the optimal architecture when combined with
riate anions and that the ideal length of the side chains can also be
ed. Comparison with the behavior of the corresponding binary salt
s also allows prediction of the optimum salt concentration for bulk
tivity. It turns out that the optimum ionomer salt concentration is
alf that of the binary salt system. This is an important result as the
salt concentration can lead to lower costs of the electrolyte and
Chapter 19
ly to better interfacial behavior. A controlling factor in the behavior of

lyelectrolyte is the value of the glass transition temperature. As the
oncentration increases so does the glass transition temperature and the
ty of the side chains is accordingly diminished. With higher salt
ntration there are more charge carriers so that conductivity is increased.
an optimum is reached where the ion concentration is maximized
the effect of the rising glass transition temperature reduces the
er mobility. Here again, introduction of more flexible backbones and
hains such as the TMO units should depress the tg while allowing a
concentration of ions to be used.
Unfortunately, little work has been reported on the behavior of single
nductors in real lithium batteries. Most reports refer only to efforts to
se the conductivities usually because the conductivities are so low as to
practical. One report exists on the impedance behavior of a
169
ectrolyte with sodium electrodes but little satisfactory data has been
ed on the interfacial behavior of polyelectrolytes. System modeling has
mployed to estimate the behavior of polyelectrolytes in real cells and
ermine the conductivity that is actually needed. The system modeled e
Li/V6013 cell and the results were compared with the binary salt
using the transport properties shown in Figure 19.14. For both
s an "ideal" polymer material was modeled and this was the binary
4
stem at 85°C and a polyelectrolyte system with conductivity of 10-
An "available" polymer was the binary system at 40°C and a
6
ectrolyte with a conductivity of 4 x 10- s/cm which is the conductivity
C of the fluorinated methide anion system shown in Figure 19.19. The
124
of the modeling show no clear advantage for either system and both
e rather poor power capability due to the effect of the interfacial
nce. It was also noted that increased power and energy density results
ncreasing the voltage of the cell which highlights the need to maximize
ability of the electrolyte to high voltage cathodes. The modeling
se does demonstrate that achievement of lithium polyelectrolyte
4 3
ctivity between 10- to 10- S/cm at ambient temperatures is a
while goal provided that the interfacial impedance is kept below 50
2
m•
3. Chemical and Electrochemical Stability of Polymer

Electrolytes
As we have discussed, the delivery of the lithium ions to the electrode

e is a complicated process. Once there, the dynamics of the layers at
ctrode surface is even more difficult to unravel and yet they are critical
efficiency of the electrochemical reaction. Large surface resistances to

n or ion transfer will lead to slow kinetics, large activation
ations and side reactions. Side reactions lead to loss of cycling
ncy, capacity fading and an increase of interfacial impedance due to
tion of products at the interface or soluble products that may migrate to
urfaces where reactions may occur that reduce efficiency. Given the
nt reactivity of organic materials to 0 V versus Li, it is indeed
able that any lithium battery is feasible. The critical role of the SEI
20
on the electrodes is the subject of intense investigations. These layers
ossess a remarkable set of properties. They must not react with the
de itself and they must separate the reactive electrode from the
ve electrolyte. However, the SEI layer must allow the passage of
ions with a minimum of resistance. To achieve this degree of
vity it is necessary that the layers behave as single-ion conductors and
e mechanism of ion conduction is different from the bulk of the
n. The SEI layers must also accommodate changing geometry of the
de under different states of charge. For intercalation electrodes this
s volume changes as the lithium ion intercalates while for lithium
shape changes and dendrite growth must be accommodated. In the
ase, changes in the surface area of the electrode occur that expose the
lyte to fresh lithium metal. The electrolyte must then participate in the
ion of fresh SEI layer to protect against further reaction.
From the previous discussion concerning the effect of surfaces on
r dynamics, one can appreciate that the passage of the lithium ions
h the surface layers is complex indeed. The presence of resistive layers
ead to depletion of ions at the electrode to levels where their
tration can no longer support the current demanded. In such cases
ns must be injected into or removed from other species present in the
n. These species can include the anions of the salt (C-F bond breaking,
mple) or the solvent molecules themselves. Detection of fluoride ion
strates the reduction of the anion with triflate or TFSI salts. The
on of products from side reactions removes lithium ions from use in
ttery and leads to reduced capacity, particularly if the electrode
ies are balanced. Often, however, one electrode is in large excess (e.g.
metal) and the loss of lithium is not noticed.
Our understanding of the electrode dynamics may be helped by the
al side reactions that do occur. The reactive intermediates which lead
ersible bond cleavage may be related to the mode of electron transfer.
been shown in the past that electrochemical and chemical reactions
volve different reactive intermediates and these lead to different
128
t distributions. The existence of trapped solvated electrons has been
ted during electrochemically driven electron transfer 172 which will
bond cleavage (e.g. carbon-oxygen bonds) once localized on a single
Chapter 19
t molecule. Since a reaction involving an unpaired electron will favor

structures such as tertiary carbons over secondary or primary carbons,
ion of products corresponding to this will support the intermediacy of a
l reactive species. Anionic cleavage will favor secondary over tertiary
ns and this might be expected to occur when two electrons are injected
aneously as may occur during a chemical reaction between lithium
surface and solvent molecules. From an analysis of the products
d from suitable solvent molecules (e.g. 2-MeTHF) one can sometimes
nsight into the electrode processes and any corrosion processes that
educe the capacity of the battery. This understanding can then be used
ign electrolytes that are resistant to the side reactions or which have s
deliberately introduced that lead to the formation of desired SEI
upon reaction.
For linear polymer electrolytes the effect of bond cleavage is to change
olecular weight by breaking the chains and to immobilize the chains on
ectrode, possibly leading to poorer ion transport through the surface
The detection of changes in molecular weight and mechanical
ties are about all that can be gained from linear polymers. Comb
ers are more useful for the investigation of side reactions at electrodes d
cleavage leads to small fragments that can be detected and quantified
onventional chemical analysis techniques (surface and bulk
oscopy, gas and liquid chromatography, mass spectroscopy). Figure
shows the structure of a comb and how bond cleavage reactions at a
ve electrode yield small fragments. Similar reasoning can be applied to
on reactions. Different reactive intermediates will favor certain n
pathways. For example a radical intermediate will favor cleavage of
rtiary carbon in the backbone next to the oxygen while anionic
ediates will favor secondary carbons. Thus electrochemically driven
ns may favor injection of one electron at a time through electron
ng and lead to cleavage at the tertiary carbon and hence the production
ments that are the full side chain length while corrosion reactions may
e injection of two electrons simultaneously through chemical reaction
he metal surface and yield fragments that result from bond cleavage all
the side chains. The product distribution would then contain products
tent with the preferred point of bond cleavage in the side chain. One
lso expect that the cleaved lithium alkoxide fragments will contribute
SEI layer and influence the behavior of the electrode. The structure of
ctrolyte may therefore be designed to provide fragments that promote
nterfacial performance and provide a means for SEI layer repair during
ttery lifetime. The well known but not understood difference in the
cial behavior of ethylene carbonate versus propylene carbonate at
anodes is an excellent example of this.
LowM.Wt.
fragment,
/..........._/ /'--.._ -',,' analyze by
I I I I GC-M:
( ( (\/
Radical cleava o
(electrochemica
here +
) ) \/\;
0 0
) i "> Anionic cleavage

\ \ •:\ : Y
') ')
(chemical - lithium \
(oVo\ 0
metal)here (
\/\f% ) ',
;
'
)
structure affects Li ion transport
through bulk polymer and across interfaces
9.20. Degradation pathways of polymers at lithium electrodes.
The incorporation of functional groups other than ethers in the

lyte structure has been long studied. Substitution of nitrogen or sulfur
61 173 174
oxygen • ' not only alters the dielectric constant but also the
h of the donor atom interaction with the lithium ion leading to
nt solvation energies and different energy barriers to lithium ion
ent. As always, however, the reactivity with the electrodes is critical
essful use of a material. The imine and sulfide functions are more
educed at the anode than the ether groups and hence the presence of a
SEI layer is required. As is well known, however, the highly reactive
carbonates are used successfully in lithium ion batteries due to the
20
on of a remarkable SEI layer. However, other desirable functional
such as nitrile (e.g. acetonitrile) appear to be unable to form a
tory film on lithium metal or on carbon and therefore have very
ed use in lithium batteries.
The ether function reacts with lithium metal also. Figure 19.21(a)
some vigorous cycling at 85°C of a symmetrical lithium/lithium cell
ing the comb branch polymer PE(TMOh (structure shown in Figure
ith LiTFSI salt in a 10:1 oxygen:lithium concentration ratio. The cell
Chapter 19
(a)
:z
...... :.. . . . . ... . . .:... Extract

-:z .......:......... ··f ... 'i"' for GC
..
-s:IJIJ IJ s:m t til" t5 Hi' :z to" :Z5 to" ::l11l" ::!5 ui'
11m' I
I Side-Chain !li
II""
fragments
li!
l:o
I
I
lllo<l
'll:il
IW
1l:il
t:l
!ill
:SOOQ
4:11
.ll:llQ "Ill
IQ Tnr <{ra!)
1
(b)
2
9.2J.(a) Cycling of Li/PE(TMOh-LiTFSI(lO:1)/Li at 85°C and 0.4 mA/cm . Cell
ss= 256 Jlm; (b) Gas chromatogram of extracted membrane from (a).
rane had a thickness of 256J.Lm and the current density was set at 0.4
2
• The cell polarization behavior shows a transition time that indicates
rrent is operating above the limiting current. The delivery of lithium
the electrode is insufficient to support the current and hence a second
n occurs at a different voltage. One may observe that the cell rapidly
ue to a dendritic short circuit. Extraction of the cell components for
is by gas chromatography shows the formation of alkoxy alcohols
ponding to cleavage of the side chains as shown in Figure 19.20 with
M0)3 unit predominating. However, a higher molecular fragment is
en that corresponds to cleavage of the backbone and indicates more
ex behavior than our analysis has so far encompassed.
2
Prior to the cycling at 0.4 rnNcm shown in Figure 19.21(a) the cell
2
en cycled at 0.2 rnNcm for two days to pass a total of 134 coulombs
mAh) with little visible change in the impedance (<5% increase in the
cial impedance). The cell had been cycled during this time at a rate
as below the limiting current where the transport properties were
ent to ensure an adequate supply of lithium ions. The limiting current
135
y be calculated according to the following equation :
Where F = Faraday's constant, Ds = salt diffusion constant in em%, c

3 0
oncentration difference across the cell in moles/cm , t+ =the lithium
nsference number and L = the membrane thickness in em. The salt
on coefficient had been measured for this electrolyte using the
143
ed diffusion method. Estimation of the transference number at about
ows one to calculate the limiting current and it is found that 0.4
2
is above this value, leading to the observation of the transition time
in Figure 19.21(a).
While the cell is cycled at values below the limiting current the
cial impedance does not change appreciably except for concentration
as shown in Figure 19.17. This indicates that the SEI layer that is
y formed (LizO, LiOH, LizC03, LiF, LiOR) is preventing appreciable
ns between the newly plated lithium and the electrolyte components.
er, as soon as the limiting current is reached where insufficient lithium
delivered, electrons are injected into the solvent molecules leading to
ns and dendrite growth is initiated. Dendrite growth increases the
area and reduces the inter-electrode distance therefore increasing the
g current that may be supported. However, the dendrites frequently
ompletely across the gap leading to a short circuit and failure of the
shown in Figure 19.21(a).
. Dendrite Growth
The role of the transport properties in the initiation and growth of

es is well known when the limiting current is exceeded and
ogical advantage is taken of this to electrodeposit metal powders. 175 If
s the only factor in dendrite growth one could simply ensure that the
g current was never exceeded and a major problem for lithium metal
s would be solved. Unfortunately, dendrites also grow at current
es below the calculated limiting current and lead to failure of the cell.
erable attention has been given to the dendrite problem for both liquid
Chapter 19
lymer electrolytes and it remains the most significant issue preventing

16 25 27 176 182
mmercialization of lithium metal polymer batteries. • The
• • -
ictable nature of the dendrite growth leads to unacceptable reliability
ms as well as safety issues due to the formation of "mossy" lithium
ts. A lot of work has been carried out on liquid electrolyte systems to
lithium dendrite growth but until the last few years the published
s on dendrite growth in polymer electrolytes have been much sparser.
tly a number of excellent in situ studies of dendrite growth in PEO
183 193
lytes have appeared - that demonstrate the growth of dendrites in
er electrolytes even at current densities that are well below the
ated limiting current. The topic of dendrite growth is covered in detail
parate chapter in this book so this discussion will be confined to the
the polymer properties in dendrite growth.
During studies of rechargeable lithium metal anodes in liquid
lyte systems it was noted that stack pressure on the cells led to
25 27 194 195
vements in lithium metal cycling. ' ' ' The liquid electrolyte cells
ned a microporous separator and through this the pressure was
itted to the anode surface. These observations led to assumptions that
er electrolytes would also provide better cycling of the lithium metal
de. However, the commonly used polymer system, PEO, is a viscous
under the typical operation conditions and is not a solid. Some of the
in situ studies have amply demonstrated that dendrites grow rapidly in
183 186 188
s liquids ' ' at current densities well below the limiting current. In
ments carried out in the author's laboratory, dendrite growth has been
ed at virtually any current density provide that the polarization
ues for long enough. The rate of initiation and propagation of dendrite
appears to be sensitive to a number of factors including the presence
129 38
purities, the transport properties and concentration of the salt,
68
ty of the charge cycle, the presence of filler particles and the
ular weight which influence the mechanical properties. 100
For these viscous liquid polymer systems pressure cannot be applied
t the aid of a supporting separator network. This can be supplied by
a microporous separator, addition of nano-particle fillers or by cross
33 68 196 197
the polymer itself. Cross-linking is required • ' ' to form a true
lectrolyte with the mechanical properties of a solid to inhibit dendrite
. With an electrolyte with sufficient mechanical strength pressure may
33 198
plied to the anode ' and the dendrite growth further inhibited.
unately, the presence of filler particles and cross-linking tends to raise
ss transition temperature, reduce the mobility of the polymer 33 •68 •100
e transport properties of the electrolyte decline, leading to a lower
g current. A further consequence is that any modification of the
er mechanical properties must be uniform across the membrane. If
iform cross-linking or dispersion of particles occurs the current
ution at the anode will be non-uniform and this will encourage dendrite
. Thus, poorly controlled chemistry such as UV and thermal radical
d cross-linking is likely to lead to non-uniformity in the polymer
ane and cause problems during cycling particularly at high rates. This
reason why in situ curing of polymer gel systems may not be wise.
pment of alternate cross-linking chemistry that is more
35 197
lable • is critical to the production of quality lithium polymer
es.
The previous discussion on the effects of surfaces on the polymer
ics is pertinent to recall at this time. One may recall that the glass
on temperature of polymer electrolytes rises with increasing salt
tration (see Figure 19.9). During discharge of a lithium metal cell with
ry salt system, the concentration gradients at the lithium metal
de rise to levels that induce phase changes (glass formation,
tation) and leads to changes in interfacial impedance (Figure 19.17).
changes will be enhanced by the presence of filler particles and cross n
the polymer and inhomogeneities in the membrane will induce non m
current distribution which may encourage dendrite initiation at the te
electrode in a symmetrical lithium/lithium cell. On charging of the
metal electrode in a lithium /polymer/metal oxide cell, the depletion
concentration at the plating electrode will lead to a lower tg, higher
y at the electrode surface and the ability to sustain a higher current
. During the charging cycle strong mechanical properties are desired
33 68
plating electrode surface to restrain dendrite growth. • However, in
tal oxide electrode, salt concentration rises, leading to a higher tg.
sed mobility and high interfacial impedance. The presence of strongly
ting particles such as carbon black or cross-linking in the polymer
lyte, binder or polymers formed from electrolyte solvents during the
ar or cycle life will lead to decreased transport properties in the
site electrode. This will be the case for dry polymer, gels and liquid s
and will result in high impedance and low rate capability. The
uence of a non-uniform composite electrode composition will be a
neven current distribution which will encourage the initiation of
e growth at the lithium metal counter electrode.
There is a well-developed literature on morphological instability
electrodeposition of a variety of metals, which indicates that even a
199 203
perturbation on the surface will initiate dendrites. - The case of
is complicated by the reactivity of the metal and the need for a
ve film on the surface, commonly known as the SEI. A physical e
of a perturbation would be non-uniformity in the SEI that can give
instability and dendrite growth. It is obvious from our discussion on
r dynamics at surfaces how perturbations may arise from the changing
ies of the polymer electrolyte during cycling and that any design of
Chapter 19
er electrolytes must take these factors into account. There is much

for modeling of the effects of such perturbations and how they can
dendrite growth as cycling is continued. It should be noted that the
f these models assumes that the surface is fundamentally unstable and
ndrite growth is inevitable. The goal will be to understand how to limit
ntrol the dendrite growth sufficiently for useful battery life.
r9t
T PEPE3_Silane·X_Tg_lhird
1 o'
10°
66 6
... ..
6 lo !' o
", ZrTg • '
-,".,'
../r I
)
'
,.
66 D D
0 c 0 0 6
'
:& 6A66 6AA6
0 0 0
00
0 I .o r> 0 o o0
""
0
0 0
o oacaooooP c 0
o
10. 1
·t OO. O
c.\spmdata\06040002.hdf
9.22. Polymer membranes prepared by cross-linking the co-polymer EPE 3 and AGE
35
sing hydrosilylation; (a) DMTA in compression mode at 10Hz. The elastic (E') and
(E") moduli and the tan I> response are shown as a function of temperature. (b) AFM
ct mode.
. Mechanical Properties
The foregoing discussion on dendrite growth highlights the need for
understanding and control of the mechanical properties of the
lytes. The inevitable desire to push the battery to the limits of its
mance will impose great stress on the polymer electrolytes that will
expose any weaknesses. Since the mechanical properties of the
er electrolytes are closely coupled with the transport properties, the
of the polymer systems needs to take close account of whether it will
d next to a lithium metal electrode, in the bulk of the separator or in a
site electrode. Even the seemingly simple process of cross-linking the
r membrane can introduce undesirable properties. For example,
19.22(a) shows the dynamic mechanical thermal analysis (DMTA) of
-linked polymer membrane of co-polymer EPE3 and AGE (10%) with
(20: 1) that shows the presence of a second tg indicating non
204
mity in the material. Examination of the film by contact mode
force microscopy (AFM) is shown in Figure 19.22(b), which shows
sence of non-uniform areas. Phase imaging AFM in tapping mode has
hown to be even more useful for the study of membrane morphology
205
l cell membranes and clearly these are very useful techniques for
control of membranes.
n this case the non-uniformity of the membrane is due to the non
nature of the co-polymer. During the co-polymerization it was noted
e AGE and EPE3 monomers disappeared at different rates indicating
e polymer was a block co-polymer and hence the cross-linking was
trated in the AGE blocks. The membrane showed rapid formation of
es when it was cycled in a symmetrical lithium/lithium cell and this
ascribed to the non-uniformity of the membrane and the fact that the
ane was very thick (400f.Lm) so that the limiting current was low.
unately the requirements for quantitative mechanical measurements
hickness) are difficult to reconcile with the need for thin membranes m)
for lithium batteries. Nevertheless, useful results can still be d if
the membrane can be separated from the electrodes after cycling.
19.23 shows the frequency sweep (a) and dynamic mechanical l
analysis (b) of a composite membrane consisting of the cross-linked
mer of EPE3 and AGE (X-PEPE3), LiTFSI (20:1) and A200 Aerosil
silica before and after cycling at 85°C in a lithium/lithium cell at 0.2
2 204
• The membrane in this case was about 300 !lm thick and the
2
ted limiting current was about 1 mA/cm based on the transport
ies measured in the absence of cross-linking and filler particles. The
cy sweep experiments show a decrease in the moduli after cycling
he DMTA shows an increase in the area under the tan 8 peak,
ng an increase in the proportion of polymer undergoing the transition,
Chapter 19
7
10
o'e
:-
w
f-
N
wE: 1o' - ·
ooooooooo
· o o·
o ooo o 000 o ooooooo
•a• • • •o • • • • • • •• • •• • •••• • • • •
10
5
1-- ----------LH-I'--------------10-' -------------J
10"
F eq (radls]
(a)
1.0 ------ ·-·--' -- --· .. -·--- --·

I ----
·
!I= 0.8
111
!! -!- . -
0.0 20.0 40.0 110.0 80.0 100.0

Temp("CJ
(b)
gure 19.23. Mechanical measurements on X-linked PEPErLiTFSI(20:1)-A200 Aerosil

0%) before and after cycling in a LilLi cell at 85°C. (a) Frequency sweep; (b) dynamic
204
chanical thermal analysis at 10Hz.
ease in the tg. indicating increased mobility and the growth of a

er on the low temperature side that is a P-transition associated with the
ent of the side chains. Extraction of the membrane and analysis by
omatography showed the presence of peaks corresponding to cleavage
side chains and this is consistent with the growth of the P-transition
loss of mechanical strength from cleavage of the cross-links.
These results are consistent with the occurrence of bond cleavage
cycling of the lithium metal. Both cross-linking and the presence of
y interacting filler particles leads to poorer transport and the salt
on coefficient was measured to be reduced by nearly four times by the
ce of the cross-linking and the filler particles. Thus, the current density
ose to the limiting current and the extent of the damage to the
lyte may be explained by injection of electrons directly into the
r under conditions where the supply of lithium ions was insufficient
ntain the current. The loss of mechanical strength observed after
shows that chemical reactions at the electrodes can degrade the
re of the membrane during the life of the cell. In this case the
ation appears to be caused by the unexpectedly large decrease in the
rt properties due to cross-linking and filler particles which are present
bit dendrite growth. This is an apt demonstration of the need to balance
rt properties with mechanical properties in lithium batteries as the
uences of chemical reactions with the high energy electrodes is very
ving.
CONCLUSIONS
The design of polymer electrolytes for lithium batteries is a complex
m that needs to be approached with a combination of sophisticated
ng, diagnostic techniques and significantly greater synthetic effort s
been applied in the past. Molecular modeling is needed to predict
olecular features desired in an electrolyte to attain the transport
ies required to support the desired rate capability and to provide the
ry mechanical strength to act as a separator, inhibit dendrites and to
gether the particles of the composite electrodes. These molecular
s must also be stabilized towards reaction at the electrodes by the
on of an SEI layer with almost magical properties. System modeling
ed to set the performance goals of the materials that need to be d
and rigorous comparison of the model predictions with experiment
ssary. Failure analysis must be accomplished by a combination of
icated diagnostic techniques together with reaction modeling and
is of model materials designed to test the modes of failure. Only then
rates of side reactions be related to lifetime issues.
Chapter 19
It is clear that this task is daunting for dry polymer systems and is
to prove difficult for polymer gels if demands are put upon them
d acting as an adhesive to hold lithium ion electrodes together as is
practiced today. Our survey of polymer electrolyte properties has
strated that the desired properties depend critically upon the
nent of the cell where the electrolyte has to function. The electrolyte o
lithium metal has to have sufficient mechanical strength to inhibit
te growth but the means of attaining this should not reduce the
ort properties. The polymer electrolyte in the composite electrodes
have much more flexibility as the surface effects of the electrodes
o major difficulties with ion transport. This is true in dry polymer, gel
quid systems where polymers are expected to bind the particles
er. These polymer systems also have to resist reactions with these
des over the life of the battery so that these desirable properties do not
e and degrade the performance of the battery.
Given all of these rigorous requirements, it is hardly surprising that the
uction of the lithium metal polymer battery has not yet been successful.
t, it is more surprising that progress has been as successful as it has
given how primitive have been the polymer systems used so far in
rison to the severity of the demands placed on them by battery
ers. This chapter has attempted to illustrate how modeling, diagnostics
nthesis can be used to design better materials for the application. Many
e methods have only become available in recent years so there is much
g work to be done in the corning years. The lithium metal system is
larly difficult due to the dendrite issue but holds the most promise for
high energy density, particularly if ways can be found to increase the
oltage to 4 volts or more. This will require the development of multi
ystems that have a layer resistant to lithium metal on the lithium with
echanical strength and a polymer in the cathode that is stable to four
Since no single material can combine these chemical properties
nt materials will be necessary in different parts of the cell. As we have
it is necessary for the electrolyte to possess different mechanical
ties in different parts of the cell anyway. The nature of these
ayers next to electrodes will be the subject of much research and it is
aging to note that already attempts are being made to design and
ate protective layers on both anodes and cathodes?06" 208 Such efforts are
only the beginning of an exciting new chapter in the development of
batteries which will lead to more realistic demands on the battery
als and better designed materials for the applications.
NOWLEDGEMENT
This work was supported by the Assistant Secretary for Energy

ncy and Renewable Energy, Office of FreedomCAR and Vehicle
ologies of the U.S. Department of Energy under Contract No. DE
76SF00098.
ERENCES
T. Leah, N. P. Brandon, V. Vesovic, G. H. Kelsall, J. Electrochem. Soc. 147 (2000)
3.
. Kerres, J. Membrane Science 185 (2001) 3.
D. Kreuer, J. Membrane Science 185 (2001) 29.
Mazrou, H. Kerdjoudj, A. T. Cherif, A. Elmidaoui, J. Molenat, New J. Chern. 22
98) 355.
Bazinet, F. Lamarche, D. lppersiel, Trends Food Sci. Tech. 9 (1998}107
E. Fenton, J. M. Parker, P. V. Wright, Polymer 14 (1973) 589.
V. Wright, Br. Polymer]. 7 (1975) 319.
Armand, J. Y. Sanchez, M. Gauthier, Y. Choquette, in Electrochemistry of Novel
terials (J. Lipkowski and P.N. Ross, eds.), VCH, New York, NY, (1994) p. 65.
Armand, Solid State Ionics 69 (1994}309.
A. Ratner, (J. R. MacCallum, C. A. Vincent, eds.), Elsevier Applied Science, London
87).
. MacCallum, C. A. Vincent, Polymer electrolyte reviews--2, Elsevier Applied
ence, London; New York, (1989).
M. Gray, Solid polymer electrolytes :fundamentals and technological applications,
H, New York (1991).
M. Gray, Royal Society of Chemistry (Great Britain), Polymer electrolytes, Royal
iety of Chemistry, Cambridge, 1997.
V. Wright, MRS Bulletin 27 (2002) 597.
Scrosati, C. A. Vincent, MRS Bulletin 25 (2000) 28.
Fauteux, A. Massucco, M. McLin, M. Vanburen, J. Shi, Electrochim. Acta 40 (1995)
5.
A. Ratner, P. Johansson, D. F. Shriver, MRS Bulletin 25 (2000) 31.
A. Ratner, D. F. Shriver, MRS Bulletin 39 (1989).
Hooper, M. Gauthier, A. Belanger, Electrochem. Sci. Techno[. Polym. 2 (1990) 237.
Aurbach, in Advances in lithium-ion batteries (W. A. v. Schalkwijk and B. Scrosati,
.), Kluwer Academic/Plenum Publishers, New York, NY, (2002) p. 7.
. Sloop, J. K. Pugh, S. Wang, J. B. Kerr, K. Kinoshita, Electrochem. Solid-State Lett.
001}A42.
Vogdanis, B. Martens, H. Uchtmann, F. Hensel, W. Heitz, Makromol. Chern. 191
02._465 .
. Goldman, L.A. Dominey, V. R. Koch, J. Power Sources 26 (1989) 519.
Gofer, M. Benzion, D. Aurbach, J. Power Sources 39 (1992}163.
M. Abraham, Electrochim. Acta 38 (1993) 1233.
R. Koch, J. Electrochem. Soc. 128 (1981}1293.
M. Abraham, J. S. Foos, J. L. Goldman, J. Electrochem. Soc. 131(1984) 2197.
Chapter 19
J. Wen, J. Kerr, M. Rubin, J. Slack, K. von Rottkay, Solar Energy Materials & Solar
lls. 56 (1999). 299.
Oradd, L. Edman, A. Ferry, Solid State lonics 152 (2002) 131.
G. Andreev, P. G. Bruce, Electrochim. Acta 45 (2000) 1417.
G. Andreev, P. G. Bruce, J. Physics-Condensed Matter 13 (2001) 8245.
J. Wilson, C. V. Nicholas, R. H. Mobbs, C. Booth, Br. Polymer J. 22 (1990) 129.
Vallee, M. Duval, F. Brochu, M. Kono, E. Hayashi, T. Sada,US Patent 5,755,985,
98
R. Allcock, Chern. Mater. 6 (1994). 1476.
R. Allcock, D. E. Smith, Y. B. Kim, J. J. Fitzgerald, Macromolecules 27 (1994). 5206.
M.G. Cowie, V. M. C. Reid, I. J. McEwen, Br. Polymer J. 23 (1990). 353.
M.G. Cowie, A. C. S. Martin, Polymer 32 (1991). 2411.
Buriez, Y. B. Han, J. Hou, J. B. Kerr, J. Qiao, S. E. Sloop, M. M. Tian, S. G. Wang, J.
wer Sources 89 (2000). 149.
Pantaloni, S. Passerini, F. Croce, B. Scrosati, F. Roggero, M. Andrei, Electrochim.
a 34 (1989). 635.
Marchese, M. Andrei, A. Roggero, S. Passerini, P. Prosperi, B. Scrosati, Electrochim.
a 37 (1992). 1559.
Andrei, L. Marchese, S. Passerini, A. Roggero, B. Scrosati,US Patent, 5,162,174,
92).
Andrei, L. Marchese, P. Prosperi, A. Roggero,US Patent 5,173,205, (1992).
Andrei, L. Marchese, A. Roggero, P. Prosperi, Solid State lonics 72 (1994). 140.
M.G. Cowie, K. Sadaghianizadeh, Solid State Ionics 42 (1990) 243.
M. Blonsky, D. F. Shriver, P. Austin, H. R. Allcock, J. Am. Chern. Soc. 106 (1984)
54.
R. Allcock, S. J. M. Oconnor, D. L. Olmeijer, M. E. Napierala, C. G. Cameron,
cromolecules 29 (1996) 7544.
R. Allcock, D. L. Olmeijer, S. J. M. Oconnor, Macromolecules 31 (1998) 753.
C. Zhang, S. B. Fang, Electrochim. Acta 45 (2000) 2131.
Hooper, L. J. Lyons, D. A. Moline, R. West, Organometallics 18 (1999) 3249.
Nishimoto, M. Watanabe, Y. Ikeda, S. Kohjiya, Electrochim. Acta 43 (1998) 1177.
Kono, E. Hayashi, M. Watanabe, J. Electrochem. Soc. 145 (1998) 1521.
Watanabe, T. Endo, A. Nishimoto, K. Miura, M. Yanagida, J. Power Sources 82
99) 786.
Nishimoto, K. Agehara, N. Furuya, T. Watanabe, M. Watanabe, Macromolecules 32
99) 154.
M. Mao, R. F. Perea, W. S. Howells, D. L. Price, M. L. Saboungi, Nature 405 (2000)
3
M. Mao, M. L. Saboungi, D. L. Price, M. B. Armand, W. S. Howells, Phys. Rev. Lett.
(2000) 5536.
L. Saboungi, D. L. Price, G. M. Mao, R. Fernandez-Perea, 0. Borodin, G. D. Smith,
Armand, W. S. Howells, Solid State lonics 147 (2002). 225.
W. Halley, Y. H. Duan, J. Power Sources 89 (2000) 139.
Borodin, G. D. Smith, R. L. Jaffe, J. Comput. Chern. 22 (2001) 641. Aabloo, M.
Klintenberg, J. 0. Thomas, Electrochim. Acta 45 (2000) 1425. Kasemagi, M.
Klintenberg, A. Aabloo, J. 0. Thomas, Solid State lonics 147 (2002)
.
ohansson, Polymer 42 (2001) 4367 .
J. Kearley, P. Johansson, R. G. Delaplane, J. Lindgren, Solid State lonics 147 (2002)
.
C. Redfern, L. A. Curtiss, J. Power Sources 110 (2002) 401.
M.G. Cowie, (J. R. MacCallum, C. A. Vincent, eds.), Elsevier Applied Science,

ndon; New York, (1987).
Sutjianto, L.A. Curtiss, J. Phys. Chern. 102 (1998) 968.
G. Baboul, P. C. Redfern, A. Sutjianto, L.A. Curtiss, J. Am. Chern. Soc. 121 (1999)
0.
. Kerr, G. Liu, L.A. Curtiss, P. C. Redfern, Electrochim. Acta In press (2003).
. Kerr, S. E. Sloop, G. Liu, Y. B. Han, J. Hou, S. Wang, J. Power Sources 110 (2002)
.
V. G. Ruzette, P. P. Soo, D. R. Sadoway, A.M. Mayes, J. Electrochem. Soc. 148
01) A537 .
P. Soo, B. Y. Huang, Y. I. Jang, Y. M. Chiang, D. R. Sadoway, A. M. Mayes, J.
ctrochem. Soc. 146 (1999) 32.
E. Trapa, B. Y. Huang, Y. Y. Won, D. R. Sadoway, A.M. Mayes, Electrochem. Solid
te Lett. 5 (2002) A85.
Zheng, F. Chia, G. Ungar, P. V. Wright, J. Power Sources 97-98 (2001) 641.
G. Zheng, F. S. Chia, G. Ungar, T. H. Richardson, P. V. Wright, Electrochim. Acta 46
01) 1397.
S. Chia, Y. Zheng, J. Liu, G. Ungar, P. V. Wright, Solid State Ionics 147 (2002) 275.
Y. Wei, D. F. Shriver, Chern. Mater. 10 (1998) 2307.
Gadjourova, Y. G. Andreev, D.P. Tunstall, P. G. Bruce, Nature 412 (2001)520.
G. McLin, C. A. Angell, Solid State Ionics 53-56 (1992)1027.
Croce, G. B. Appetecchi, L. Persi, B. Scrosati, Nature V394 (1998) 456.
Persi, F. Croce, B. Scrosati, E. Plichta, M.A. Hendrickson, J. Electrochem. Soc. V149
02) A212.
Golodnitsky, G. Ardel, E. Strauss, E. Peled, Y. Lareah, Y. Rosenberg, J. Electrochem.
. Vl44 (1997) 3484.
Golodnitsky, G. Ardel, E. Peled, Solid State Ionics V147 (2002) 141.
Kumar, L. G. Scanlon, J. Electroceram. V5 (2000) 127.
Kumar, L. G. Scanlon, R. J. Spry, J. Power Sources 96 (2001) 337.
ohansson, M.A. Ratner, D. F. Shriver, J. Phys. Chern. B 105 (2001) 9016
. Weston, B. C. H. Steele, Solid State Ionics 7 (1982) 81.
. Weston, B. C. H. Steele, Solid State Ionics 7 (1982).
Tsagaropoulos, A. Eisenberg, Macromolecules 28 (1995) 396.
Tsagaropoulos, A. Eisenberg, Macromolecules 28 (1995) 6067.
Gagliardi, V. Arrighi, R. Ferguson, M. T. F. Telling, Physica B 301 (2001) 110.
Zhu and S. S. Sternstein, in MRS Proc. Vol. 661 (A. I. Nakatani, R. P. Hjelm, M.
spacher, R. Krishnamoorti, eds.), MRS, Boston, MA, (2000) p. KK4.3.1.
E. Wallace, D. A. Fischer, K. Efimenko, W. L. Wu, J. Genzer, Macromolecules 34
01) 5081.
E. Wallace, J. H. Vanzanten, W. L. Wu, Phys. Rev. A 52 (1995) R3329.
. Vanzanten, W. E. Wallace, W. L. Wu, Phys. Rev. A 53 (1996) R2053.
. Forrest, J. Mattsson, Phys. Rev. E 61 (2000) R53.
. Forrest, J. Mattsson, L. Borjesson, Eur. Phys. J. A 8 (2002) 129.
A. Forrest, K. Dalnokiveress, J. R. Stevens, J. R. Dutcher, Phys. Rev. Lett. 77 (1996)
2.
. Forrest, K. Dalnoki-Veress, Advances in Colloid & Interface Science 94 (2001)
.
L. Vakula, L. M. Pritykin, Polymer Adhesion: Physico-Chemical Principles, E.
wood, London, (1991).
H. Anastasiadis, A. Karim, G. S. Ferguson, Interfaces, Adhesion and Processing in mer
systems: symposium held April 24-27, (2000), San Francisco, California..._U.S.A,
erials Research Society, Warrendale, Penn. (2001).
Chapter 19
B. Kerr, J. Xie, R. Duan, J. Electrochem. Soc., Submitted (2003).

R. Sadoway, B. Y. Huang, P. E. Trapa, P. P. Soo, P. Bannerjee, A.M. Mayes, J.
wer Sources 97-8 (2001) 621 .
Lund, 0. Hammerich, Organic Electrochemistry: an Introduction and a Guidg,
rcel Dekker, New York, 2001.
Hammerich, V. D. Parker, Electrochim. Acta 18 (1973) 537.
B. Appetecchi, F. Croce, L. Persi, F. Ronci, B. Scrosati, Electrochim. Acta V45
00) 1481.
Nishi, in Advances in lithium-ion batteries (W. A. v. Schalkwijk and B. Scrosati,
.), Kluwer Academic/Plenum Publishers, New York, (2002) p. 233.
Scrosati, in Advances in lithium-ion batteries (W. A. v. Schalkwijk and B. Scrosati,
.), Kluwer Academic/Plenum Publishers, New York, (2002) p. 251.
. Kono, E. Hayashi, M. Watanabe, J. Electrochem. Soc. 146 (1999) 1626.
. Kono, E. Hayashi, M. Nishiura, M. Watanabe, J. Electrochem. Soc. 147 (2000) 2517.
S. Gozdz, C. N. Scmutz, J.-M. Tarascon,US Patent (1994).
. Doyle, J. Newman, A. S. Gozdz, C. N. Schmutz, J. M. Tarascon, J- Electrochem.
. 143 (1996) 1890.
Arora, M. Doyle, A. S. Gozdz, R. E. White, J. Newman, J. Power Sources 88 (2000)
9.
Du Pasquier, P. C. Warren, D. Culver, A. S. Gozdz, G. G. Amatucci, J. M. Tarascon,
id State Ionics 135 (2000) 249.
Du Pasquier, T. Zheng, G. G. Amatucci, A. S. Gozdz, J. Power Sources 97-8 (2001)
.
Aihara, G. B. Appetecchi, B. Scrosati, K. Hayamizu, Phys. Chern. Chern. Phys. 4
02) 3443.
Aihara, G. B. Appetecchi, B. Scrosati, J. Electrochem. Soc. 149 (2002) A849.
M. Tarascon, A. S. Gozdz, C. Schmutz, F. Shokoohi, P. C. Warren, Solid State Ionics
8 (1996) 49.
A. Khan, G. L. Baker, S. Colson, Chern. Mater. V6 (1994) 2359.
Fan, S. R. Raghavan, X. Y. Yu, S. A. Khan, P. S. Fedkiw, J. Hou, G. L. Baker, Solid
te Ionics 111 (1998) 117 .
R. Raghavan, M. W. Riley, P. S. Fedkiw, S. A. Khan, Chern. Mater. 10 (1998) 244.
J. Walls, J. Zhou, J. A. Yerian, P. S. Fedkiw, S. A. Khan, M. K. Stowe, G. L. Baker,
ower Sources 89 (2000) 156 .
L. Dai, T. A. Zawodzinski, J. Electrochem. Soc. 143L (1996) 107. L. Dai, T.
A. Zawodzinski, J. Electroanal. Chern. 459 (1998) 111. Eisenberg, J.-S.
Kim, Introduction to Ionomers, Wiley, New York, (1998).
E. Thomas, S. E. Sloop, J. B. Kerr, J. Newman, J. Power Sources 89 (2000) 132
E. Thomas, R. M. Darling, J. Newman, in Advances in lithium-ion batteries (W. A. v.
alkwijk and B. Scrosati, eds.), Kluwer Academic/Plenum Publishers, New York,
02) p. 345.
R. Sadoway, A. M. Mayes, MRS Bulletin 27 (2002) 590 .
N. Powell, Laser Focus World 37 (2001) 41.
B. Kerr, J. Electrochem. Soc. 132 (1985) 2839.
Chusid, Y. Gofer, D. Aurbach, M. Watanabe, T. Momma, T. Osaka, J. Power Sources
98 (2001) 632.
. Schmidt, E. Steckhan, Angew. Chern. 91 (1979) 850.
. Schmidt, E. Steckhan, Angew. Chern. 90 (1978) 717.
W. Boyd, P. Schmazl, L. L. Miller, J. Am. Chern. Soc. 102 (1980) 3856.
R. Zhang, J. K. Pugh, P. N. Ross, J. Electrochem. Soc. 148 (2001) E183 .
Moshkovich, M. Cojocaru, H. E. Gottlieb, D. Aurbach, J. Electroanal. Chern. 497
01) 84.
Newman, Electrochemical Systems, Prentice-Hall, Englewood Cliffs, NJ, (1991).

. Doyle, J. Newman, J. Power Sources 54 (1995)46.
. Doyle, J. Newman, Electrochim. Acta 40 (1995) 2191.
. Doyle, T. F. Fuller, J. Newman, Electrochim. Acta 39 (1994) 2073.
F. Fuller, M. Doyle, J. Newman, J. Electrochem. Soc. 141 (1994) 1 .
F. Fuller, M. Doyle, J. Newman, J. Electrochem. Soc. 141 (1994) 982.
. Doyle, T. F. Fuller, J. Newman, J. Electrochem. Soc. 140 (1993) 1526
Ferry, M. M. Doeff, L. C. Dejonghe, Electrochim. Acta 43 (1998) 1387.
. P. Ma, M. Doyle, T. F. Fuller, M. M. Doeff, L. C. Dejonghe, J. Newman, J.
ctrochem. Soc. 142 (1995) 1859.
. J. Walls, T. A. Zawodzinski, Electrochem. Solid-State Let. 3 (2000)321.
Ferry, M. M. Doeff, L. C. Dejonghe, J. Electrochem. Soc. 145 (1998)1586.
. M. Doeff, P. Georen, J. Qiao, J. Kerr, L. C. De Jonghe, J. Electrochem. Soc. 146
99) 2024.
. M. Doeff, L. Edman, S. E. Sloop, J. Kerr, L. C. De Jonghe, J. Power Sources 89
00) 227.
Edman, M. M. Doeff, A. Ferry, J. Kerr, L. C. De Jonghe, J. Phys. Chern. B 104 (2000)
76.
. M. Doeff, A. Ferry, Y. P. Ma, L. Ding, L. C. Dejonghe, J. Electrochem. Soc. 144
97) L20.
F. Nazar, G. Goward, F. Leroux, M. Duncan, H. Huang, T. Kerr, J. Gaubicher, Int. J.
rg. Mater. 3 (2001) 191 .
. Doyle, J.P. Meyers, J. Newman, J. Electrochem. Soc. 147 (2000) 99.
Rivero, K. 0. Thomas, J. Newman, Unpublished results 2002.
. Doyle, M. E. Lewittes, M. G. Roelofs, S. A. Perusich, J. Phys. Chern. B 105 (2001)
87.
. Doyle, M. E. Lewittes, M.G. Roelofs, S. A. Perusich, R. E. Lowrey, J. Membrane
ence 184 (2001) 257 .
E. Feiring, S. K. Choi, M. Doyle, E. R. Wonchoba, Macromolecules 33 (2000) 9262 .
.P. Siska, D. F. Shriver, Chern. Mater. 13 (2001) 4698 .
Fujinarni, A. Tokimune, M.A. Mehta, D. F. Shriver, G. C. Rawsky, Chern. Mater. 9
97) 2236.
Matsushita, Y. Shimazaki, M.A. Mehta, T. Fujinami, Solid State lonics 133 (2000)
. A. Mehta, T. Fujinami, Chern. Lett. 915 (1997).

C. Rawsky, T. Fujinami, D. F. Shriver, Chern. Mater. 6 (1994) 2208 . Benrabah,
M. Armand, D. Delabouglise,US Patent 5,696,224, (1997). Benrabah, S. Sylla,
F. Alloin, M. Armand, Electrochim. Acta 40 (1995) 2259. M.G. Cowie, G. H.
Spence, Solid State Ionics 123 (1999) 233 .
D. D. Desmarteau, J. Fluorine Chern. 72 (1995) 203.
Q. Hu, D. D. Desmarteau, lnorg. Chern. 32 (1993) 5007.
. Xu, K. S. Siow, Z. Gao, S. Y. Lee, Chern. Mater. 10 (1998) 1951.
S. Zhang, G. X. Wan, J. Appl. Polymer Sci. 48 (1993) 405.
S. Zhang, L. L. Yang, Q. G. Liu, Solid State lonics 76 (1995)121.
S. Zhang, L. L. Yang, Q. G. Liu, Solid State lonics 76 (1995) 127.
F. Snyder, M. A. Ratner, D. F. Shriver, Solid State lonics 147 (2002) 249 .
F. Snyder, M. A. Ratner, D. F. Shriver, J. Electrochem. Soc. 148 (2001) A858.
Rahner, J. Power Sources 82 (1999) 358.
E. A. Dillon, D. F. Shriver, Chern. Mater. 13 (2001) 1369.
York, R. Frech, A. Snow, D. Glatzhofer, Electrochim. Acta 46 (2001) 1533.
Paunovic, M. Schlesinger, Electrochemical Society, Fundamentals of
ctrochemical Deposition, Wiley, New York, 1998.
Chapter 19
. Ambach, E. Zinigrad, H. Teller, P. Dan, J. Electrochem. Soc. 147 (2000) 1274 .

0. Besenhard, J. Guertler, P. Komenda, M. Josowicz, Proc.- Electrochem. Soc. 88-6
988) 618 .
. Fauteux, Electrochim. Acta 38 (1993) 1199 .
D. Fauteux, R. Koksbang, J. Appl. Electrochem. 23 (1993) 1 .
. Kanamura, Erekutoronikusu 43 (1998) 60.
.-i. Takehara, J. Power Sources 68 (1997) 82.
-i. Yamaki, S.-i. Tobishima, K. Hayashi, K. Saito, Y. Nemoto, M. Arakawa, J. Power
urces 74 (1998) 219.
. Rosso, T. Gobron, C. Brissot, J. N. Chazalviel, S. Lascaud, J. Power Sources 97-98
001) 804.
. Brissot, M. Rosso, J. N. Chazalviel, S. Lascaud, J. Power Sources 94 (2001) 212
. Brissot, M. Rosso, J. N. Chazalviel, S. Lascaud, J. Electrochem. Soc. 146 (1999)
93.
. Brissot, M. Rosso, J. N. Chazalviel, S. Lascaud, J. Power Sources 82 (1999) 925.
. Brissot, M. Rosso, J. N. Chazalviel, P. Baudry, S. Lascaud, Electrochim. Acta 43
998) 1569.
. Dolle, L. Sannier, B. Beaudoin, M. Trentin, J. M. Tarascon, Electrochem. Solid-State
tt. 5 (2002) A286 .
Orsini, L. Dupont, B. Beaudoin, S. Grugeon, J. M. Tarascon, International Journal of
rg. Mater. 2 (2000) 701 .
Orsini, M. Dolle, J. M. Tarascon, Solid State lonics 135 (2000) 213.
Orsini, A. du Pasquier, B. Beaudouin, J. M. Tarascon, M. Trentin, N. Langenhuizen,
de Beer, P. Notten, J. Power Sources 82 (1999) 918.
Orsini, A. Dupasquier, B. Beaudoin, J. M. Tarascon, M. Trentin, N. Langenhuizen, E.
beer, P. Notten, J. Power Sources 76 (1998)19.
Osaka, T. Homma, T. Momma, H. Yarimizu, J. Electroanal. Chern. 421 (1997) 153.
A. R. Stiles, K. Brandt, D. S. Wainwright, K. C. Lee,US Patent 4,587,182, (1986).
Eweka, J. R. Owen, A. Ritchie, J. Power Sources 65 (1997) 247.
Tatsuma, M. Taguchi, N. Oyama, Electrochim. Acta 46 (2001)1201 .
. Michot, A. Vallee, P.-E. Harvey, M. Gauthier, M. Armand,US Patent 6,492,449,
02)
. Gauthier, A. Belanger, A. Vallee,US Patent 6,007,935, (1999).
Elezgaray, C. Leger, F. Argoul, J. Electrochem. Soc. 145 (1998) 2016.
S. Lauch, M.D. Pritzker, J. Electroanal. Chern. Interfac. Electrochem. 319 (1991) 33.
S. Louch, M.D. Pritzker, J. Electroanal. Chern. 346 (1993) 211.
G. Sundstrom, F. H. Bark, Electrochim. Acta 40 (1995) 599 .
.D. Pritzker, T. Z. Fahidy, Electrochim. Acta 37 (1992) 103.
Xie, G. Liu, Y. B. Han, K. J.B., Unpublished results (2003).
Wang, M. Rickner, Y. S. Kim, T. A. Zawodzinski, J. E. McGrath, J. Membrane
ence 197 (2002) 231 .
J. Visco, F. Y. Tsang, US Patent 6,432,584, (2002)
J. Visco, M.-Y. Chu, US atent 6,025,094, (2000)
J. Dudney, J. Power Sources 89 (2000) 176.
apter 20
LASS AND CERAMIC ELECTROLYTES FOR

LITHIUM AND LITHIUM-ION BATTERIES
N.J. Dudney
Solid State Division, Oak Ridge National Laboratory, P. 0. Box 2008, Oak Ridge,
Tennessee 37831-6030, U.S.A.
1. INTRODUCTION
In recent years there have been important advances in the stability,
ty, and performance of lithium and lithium-ion batteries. Many of the
trolyte materials being examined are based on organic liquids or
mers, although solid inorganic electrolytes still have an important role a
variety of applications. There are numerous reports of new
positions, advanced synthesis routes, and novel cell architecture for the s
and ceramic electrolytes.
For any rechargeable lithium battery, the electrolyte material must
it the repeated and rapid transfer of Lt ions between the anode and
ode over the expected range of operating conditions (voltage,
perature, and current), without significant deterioration. Additionally,
ideal electrolyte material would be an electronic insulator, ultra-thin,
tweight, free of hazards and inexpensive. The inorganic solid
trolytes offer both advantages and disadvantages over liquid and organic
mer electrolytes. For the required rapid transport of Lt across the
trolyte, the product of the resistivity and electrolyte thickness must be
mized. Typical room temperature conductivities are 10-1 S/cm for
ds, 10-2 S/cm for superionic conductors such as -alumina, 10-3 to 10-6
for various gel and solvent-free (dry) polymers, and 10-4 to 10-8 S/cm
ypical glass and ceramic solid electrolytes. If formed as a very thin film
ess than about 1 f1m, even a rather resistive material may compete
rably when compared to a much thicker cast polymer or liquid-filled
us separator membrane. An ultra-thin electrolyte also provides a
iderable savings in terms of volume and mass for the battery, if not
et by the need for a thick inactive support material. It comes as a
Chapter 20
rise to many that 1-10 )..tm thick glass and ceramic sheets are quite
ible.
One of the principle advantages of the inorganic solid electrolytes is
these materials are generally single ion conductors, meaning that only
Lt ions have an appreciable mobility, while the anions and other cations
a rigid framework. This has many beneficial consequences.
inating the anionic concentration gradient across the electrolyte may
suppress undesirable side reactions or decomposition of the electrolyte.
single ion conductor expands the choice of electrode materials and
its operation at higher voltages. Few polymer electrolytes mimic this
1
le ion behavior and they are usually associated with a higher resistivity.
wise, a negligible electron transport across the electrolyte is necessary to
ent self-discharge of the battery. For a shelf life of months or years, the
trolyte must be an excellent insulator with an electronic conductivity <
2
S/cm. Conductive electron paths may be formed along lithium
rites for liquid or polymer electrolytes, while grain boundaries, pores, or
ks may become electronic paths for glass and ceramic electrolytes.
ificant concentrations of multivalent ions in inorganic electrolytes may
contribute to electronic transport. Ideally, the lithium ion transference
transport numbers should be as close as possible to identically 1.0.
The resistance of the solid electrolyte-electrode interface is just as
ortant and often as large as the bulk resistivity of the electrolyte.
ause of this, reports of an electrolyte's properties are insufficient without
valuation in contact with reversible electrodes comparable to those for
Li-ion battery. One might assume that ion exchange across the solid
rolyte-electrode interface would be more resistive than that with a liquid
oft polymer electrolyte. This is not necessarily the case, due to the
ation of a "solid electrolyte interphase" (SEI) reaction layer, which adds
e interface resistivity for most liquid or polymer electrolytes. The solid
rode-electrolyte interface has the added advantage of providing a dense,
surface that maintains its integrity and inhibits roughening of the
ostructure that leads to mossy or dendritic deposits. Without good
sion however, stresses may cause a rigid interface to form permanent
ks or voids.
The nature of the highly active lithium battery chemistry places severe
rochemical requirements on the electrolyte. Many of the glass and
mic electrolytes proposed in the literature decompose under high
ges or react with lithium, but some are indeed believed to be
modynamically stable. The organic electrolytes, on the other hand,
riably rely on the formation of a protective SEI layer to limit the reaction
composition rate.
As there are hundreds of different electrolyte compositions in the
ture, this chapter will cover work published since 1995 or 1997,
ding newly investigated electrolytes, as well as known electrolytes that
sy and Ceramic Electrolytes for Lithium and Lithium-Ion Batteries 625
inue to be used and studied for experimental batteries. The discussion

also focus on electrolytes for low and ambient temperature conditions,
rimary interest for lithium-ion batteries. In 1995, publication of two
ters summarized the compositions, properties, and battery applications
lassy lithium electrolytes.2• 3 The most recent review of crystalline
4
mic lithium electrolytes appeared in 1997. An earlier overview
essing the motion of the lithium ions through either the amorphous or
talline environment was presented in 1993. 5
The compositions and properties of glassy and ceramic electrolytes to
iscussed in this chapter have been summarized in Tables 20.1-2 at the
of this chapter. Not every composition is included, only representative
mples of each class of electrolyte. The cited literature should be
ulted for a thorough listing. Likewise, many reports cover a range of
positions, but for simplicity, the tabulated compositions are limited to
e giving the maximum observed lithium conductivity. The table includes
synthesis process, the electrolyte thickness, the u+ ion and electronic
port properties, and the electrochemical stability. No doubt, in all cases
lectrolyte could be formed as a thinner layer, but the tabulated values
blish the current art. The stability range is given as reported by the
ors, although in some studies, it appears that currents larger than that
butable to an electronic carrier were observed well before an obvious
kdown voltage. Very few studies evaluate the interface resistance with
rsible lithium electrodes; these results will be noted in the text.
. GLASS ELECTROLYTES
With many glassy lithium conductors already in the literature, the
rity of recent work is tending away from the traditional glass processing
xplore twin roller, vapor deposition, high-energy mechanochemical
ng, or even explosive glass forming processes. These processes expand
normal glass-forming region. A number of studies reveal that
positions with a higher concentration of the Li modifier, e.g. LizO and
give a higher conductivity and lower activation energy for u+
port. For many compositions, the glass is almost fully depolymerized.
Commonly perceived advantages of glassy electrolytes over
alline ceramic materials include: isotropic conductivity, ease of
cation as dense thin films, and lack of grain boundaries. Grain
daries in polycrystalline or granular electrolytes have an inherently r
density, higher defect concentration, space charge polarization, and
gated impurities. These may form resistive barriers for Li+ but fast
for other species, increasing the likelihood of an internal short circuit,
ecipitated Li metal dendrite, or breakdown related to electrochemical
s. The nature and effect of boundaries in glass electrolytes formed as
ed compacts of crushed glass powder has not been determined.
Chapter 20
2.1. Oxide Glasses

Compared to earlier reviews, fewer pure oxide glasses have been
rted in recent years. This is likely due to their unexceptional
ductivity. The big advantage of the lithium oxide glasses is that, although
may be hygroscopic, they are relatively stable in air. This simplifies the
hesis and handling of the material.
As illustrated by the first entry in Table 20.1, glasses with mixed
ers, such as Se and B, have a reasonable conductivity.5 If formed as
thin films, such glass compositions may prove to be promising
trolytes. Unfortunately, this selenate-borate glass has not been evaluated
thin film or with Li-ion electrodes. The second oxide glass is a vitrified
of a crystalline Li electrolyte with a nasicon structure to be discussed
w.6 It was included to illustrate that even when doped with the LiCl
ifier salt, the conductivity of the glass is an order of magnitude lower
that of the polycrystalline material with an open network structure that
itates the Li transport. For other Li compounds, such as Li 3P04 and
iS3, the situation may reverse as the amorphous phase is indeed orders of
nitude more conductive than compact crystalline phases.
2.2. Oxynitride Glasses and Lipon

The phosphorus oxynitride glass, commonly referred to as 'lipon,' was
eered at Oak Ridge National Laboratory (ORNL) in 199C and more
characterized by ORNL researchers in later years.8 Subsequent
ications number roughly 20 from several research institutions. Lipon
s, about 1 J..tm thick, are typically deposited by rf magnetron sputtering a
high purity LhP04 ceramic target in nitrogen plasma. Attempts to icate
glasses of similar compositions from the melt have been
ccessful suggesting that vapor deposition is equivalent to a rapid quench
enting separation of a Li-rich phase. The small amount of N
rporated into the films from the plasma effectively stabilizes the material
ontact with metallic lithium and enhances the ionic transport by 40-fold
pared to amorphous LhP04 films also prepared by sputtering. Lipon has
en highly successful for Li and Li-ion thin film batteries fabricated at
9
L. Lipon is being investigated by a number of groups for a variety of
2
rochemical applications including: thin film batteries, 10 11 12
• • C0
13 14
ors, thin film capacitors, and electrochromic windows.
Deposition rates achieved by magnetron sputtering just exceed 10
m. 15 Because this is rather slow for a practical commercial process,

rchers at Tuffs have reported using a nitrogen ion beam and thermal
oration source in a so-called ion beam directed assembly (IBDA)

10
ess to deposit lipon at >30 nm/m. The IBDA deposited lipon has good
uctivity and electrochemical properties, but appears to be prone to crack
16
to a residual tensile stress. A similarly high deposition rate is achieved
pulsed laser ablation which yields amorphous film of comparable
17
uctivity. Some researchers report a variation of composition and
uctivity as lipon is deposited by sputtering at higher rates, both
18 19
mental and advantageous, although neither has been observed at
s and ORNL. Variation in results from group to group is not surprising
his non-equilibrium synthesis. Film thickness is commonly about 1 J.Lm
eeded for good coverage of rough underlying electrode films. If the
n were deposited onto a smooth and crack free surface, a continuous
ole free film might be obtained with a submicron film, however a great
ovement in cell performance would likely be thwarted without a
parable reduction in the interface resistance. Much is yet to be learned
t control of the vapor deposition processes.
2
Compositions of the lipon films (-O.lmg/cm ) are difficult to
rmine with accuracy greater than about +1- 10%, so a clear analysis of
variation of conductivity and activation energy with composition is
ing. It is certain that a small substitution of N for 0 in the amorphous
04 material greatly enhances the Li mobility, due to either the formation
dditional cross-linking between phosphate groups or a decrease of the
20
rostatic energy when P-0 is replaced by P-N bonds. The maximum
6
rted conductivity is 2-3·10- S/cm for compositions of about
P033N0.46• Variations in the conductivity upon further N substitution or
addition of N + Li are more subtle and difficult to quantify. Equally
ortant for the conductivity is the relaxation of the structure. Increases of
4 times in the conductivity may occur when the lipon film is annealed
ly at temperatures of 150-250°C.
For a successful solid electrolyte, the electrochemical stability and
ronic resistivity is as important as the Li ion conductivity. These
erties are often assessed by current-voltage measurements with lithium a
blocking electrode, but the best evaluation is the long-term stability of gh
voltage lithium cell. As reported at recent meetings, Li/LiCo02
ries with a lipon electrolyte show a negligible self discharge rate at both
charged (4.2V) and fully discharged (3V) conditions when stored for a
21
or more. Based on these results, the estimated electronic resistivity
14
exceed 10 ohm-em. The chemical stability of lipon in contact with Li
l may be attributed to the reducing character of the oxynitride glass
22
ix. Preliminary results for a similar sputter deposited lithium boron
23
itride electrolyte film have also been reported. The electrochemical
lity has not been assessed, but may be comparable to that of the
phate material.
Chapter 20
2.3. Sulfide and Oxysulfide Glasses

Original work on highly conductive Li sulfide glasses based on SiS2,
24
, and B2S3, was reported 15 years ago. '25 The impressive conductivities
low activation energies compared to the oxide glasses were attributed to
weaker bonding of Li with the non-bridging sulfur anions, due to the
er polarizability and softer basicity of the sulfide ion. As for the oxide
ses, the maximum conductivities were identified with large
entrations of the glass modifier allowing only a minimal degree of
merization. These glasses have not been widely used largely because
are quite difficult to handle. Being both highly reactive with air and ly
corrosive with silica containers, great care and a highly efficient e
box are required to synthesize and fabricate the sulfide glasses.
ewed interests by several Japanese groups in recent years have lead to
modified processing and compositions summarized in Table 20.1.
The use of mechanochemical milling was first demonstrated for
26
•SiS2 compositions and then expanded to the LizS•P2SP and various
ulfide compositions. This process uses a high-energy planetary ball
to amorphize the mixed sulfide reactants at room temperature under a
N2 or Ar atmosphere. Achieving complete amorphization, as indicated
-ray diffraction, generally required milling times of about 20 hours, but
otal energy consumed was still less than that used in a traditional melt
ch process. A further advantage of this process is that the resulting
grained powder was ready to fabricate directly as a pressed pellet or as
lectrolyte layer in a compacted solid state battery. In comparison, the
um ion conductivities achieved by the ball-milling were as high or nearly
igh as that of glasses prepared by twin-roller rapid quench of the same
position. There are also examples of an expanded glass forming region
27
g achieved by the mechanochemical milling process.
Addition of a small amount, -5 mole%, of a variety of different
um metal oxides (silicate, phosphate, and germanate) was found to
lize the sulfide glass structure as evidenced by an increase in the
rence between the crystallization temperature and the glass transition
28
erature. Results from NMR and XPS show the oxygen to be primarily
ed as a bridging anion. Larger additions of these lithium metal oxides
me detrimental to the thermal stability and lithium ion conductivity, as
asing amounts of oxygen are found in a non-bridging coordination
h impede the mobility of the lithium ions. Alternatively, additions of
us group IVA, VA, VIA, and VB sulfides greatly enhance the thermal
29
lity of the LizS•SiS2 glass. Even when slow cooled or annealed, many
e compositions remained vitreous with a high lithium conductivity near

S/cm. This allowed for the fabrication of glass plates rather than
tized samples of glass flakes or powders. The electronic conductivity is
y and Ceramic Electrolytes for Lithium and Lithium-Ion Batteries 629
rly small, but the precise effect of the transition metals on the electronic
uctivity has not been assessed.
All solid state batteries, In/LiCo02 and Li/LiCo02, have been
ared using the lithium silicon oxysulfide electrolyte compositions shown
30
able 20.1. '31 '32 The cells were formed by compacting the ground
trolyte powder together with the cathode mixture and In or Li foil in a 1-
diameter insulating tube. After several initial cycles, the cycling
eeded with a 100% coulomb efficiency for 100 cycles indicating good
trochemical stability. Although the electrolyte thickness was not given,
bserved cell resistance (0.7-2 kohm) was higher than anticipated for an
mated 0.5 mm thick layer of the highly conductive electrolyte. This high
nal resistance was attributed to the electrode interfaces, an important
e discussed briefly below. The reversible capacity was only 60 mAh per
LiCo02 when using the In anode, while a higher capacity of 80-90
/g is reported for the Li cell. A variety of cathode materials including
S2 and LizFeC14 were evaluated with this same tubular cell construction
ompacted powders.33 These cathode materials, which had been deemed
ly reversible in liquid cell experiments, were shown to cycle with good
city and efficiency in the cells using the glassy solid electrolyte.
.4. Thionitride Glasses

Highly conductive thionitride glasses are also reported?4 •35 These are
ed by rapid quench or high energy mechanical milling processes as
ribed for several of the sulfide glasses. Although the maximum
uctivity is reported for material milled for 20 hours, a conductivity of
mS/cm is reported after only 20 minutes. Significant oxygen
amination is attributed to the milling media. A single voltammetric
ep (-0.5 to 10 V) indicates a wide stability range; however, the data plots
oo coarse to ascertain the absence of significant tarnishing or interface
ions. It is clear that the electrochemical stability far exceeds that of
.
.5. Lithium Halide Doped Glasses
In recent years, little additional work has been reported for Li halide
d glass compositions, although numerous compositions were reviewed
er.' The most conductive Li electrolyte reported to date, >2 mS/cm, is a
doped sulfide glass of 60Li2S•40SiS2 •2.8TaS2•4Lil.29 One old

position, xLil•(1-x)[0.6Li2S•0.4B2S3] was revisited to evaluate the
ence of the Lii on the lithium motion in light of recent models for the
Chapter 20
6
c transport? The mechanism of ion transport in the glassy structure
inues to be a topic of experimental and theoretical research.
3. CRYSTALLINE CERAMIC ELECTROLYTES
Many ceramic electrolytes have been investigated, although few have

evaluated in actual electrochemical cells. A notable exception is Lil
d electrolytes for Li/12 batteries which are in commercial use to power
iac pace makers. Ceramic electrolyte materials include oxides, sulfides,
des, halides, and a variety of polyanion compositions. Although some
ressive lithium conductivities are reported, fabrication and stability
es require further attention.
The ceramic electrolyte materials were reviewed in 1997 for the
3
ication in high temperature lithium batteries. The materials
irements differ from those for the Li-ion battery only with regard to the
ation energy. For the high temperature applications, high activation
gy offers the advantage of a thermal switch for the cell, whereas aLi-ion
ry electrolyte should ideally have a low activation energy. To avoid
ication, this discussion will highlight only very recent reports,
hasizing the electrolytes formed as films or with valuable insight
erning the electrochemical stability and electronic conduction. These
ummarized in Table 20.2.
Several important issues arise for each class of crystalline electrolyte:
rochemical stability, grain boundaries, and fabrication as thin sheets.
stability is an important issue as many compositions contain transition
l cations which may compensate for lithiation or de-lithiation reactions.
ending on the composition and lattice, such reactions may result in a
ificant electronic conduction through the material. For the superionic
uctive crystal phases, the disorder accompanying grain boundaries
rally impedes the ion transport and decreases the conductivity. The
rolytes are hence very sensitive to processing factors affecting the
ity and grain size. On the other hand, for materials that conduct via
e defects, such as vacancies and interstitials, the ion transport is often
nced by the presence of grain boundaries, a dispersed second phase, or
r extended defects. Most studies report results for cold-pressed and
red ceramic pieces, while few present results for thin layers which
ld be far more appropriate to battery applications. Presumably,
rolling the microstructure in thinner cast, spun or sprayed layers poses
tional difficulties.
3.1. Perovskite Electrolytes
There has been a flurry of research on polycrystalline lithium

trolytes with the perovskite crystalline structure in recent years (Refs.
8 and references therein). In excess of 20 different compositions have
reported. Three of the simplest and most conductive compositions are
rted in Table 20.2. The perovskite structure, with the general formula
03, has a remarkable tolerance for ion substitution, both aliovalent and
alent, on both the A and B lattice sites. This can be exploited to create
lattice structures and compositions with large vacancy concentrations
he A sublattice. For the Li conductors, up to as much as l/6th mole of u+
are substituted into various A-deficient compositions. The lithium
uctivity depends on a variety of factors, including: the lithium and
ncy concentrations on the A sites, the degree of cation ordering, the size
symmetry of the unit cell. Even when all the A sites are filled, there
still be appreciable Li conductivity, suggesting the availability of
tional sites for the lithium.37 The highest reported conductivity is
0-3 S/cm for La213_xLhxTi03, with x-0.11.39 Generally, substitution of
La by other rare-earth cations reduces the conductivity, as does
titution of the Ti with other tetra- or penta-valent cations. There is some
bility in the grain boundary contribution to the total resistance?8• 4° For
e materials the grain boundary contribution to the resistivity is 10 to
fold higher than that of the crystal lattice (bulk), as evaluated by the
uency scans of the complex impedance. In most reports, these materials
prepared as pressed and sintered pellets of powders produced by solid
reaction.
So far, the perovskite electrolytes have not been used in any lithium or
um-ion batteries. One potential difficulty is that additional Li can be
calated into the lattice upon contact with metallic lithium or other
um anode materials with a low electrochemical potential versus Li/Lt.41
may result in a significant electronic conductivity due to the reduction
4
i + cations. The Ta-perovskites with a stable +5 oxidation state,
ever, are expected to be more stable than those with Ti. 42 On the other
, the cyclic voltarnmetry study of a Nb-perovskite40 shows that
ction does indeed commence at voltages below 2.0 versus metallic Li, a
ion that is not fully reversible. This reaction effectively eliminates this
composition as a candidate for any Li-ion battery under current
ideration. There is far too little research on the electrochemical stability
electronic conductivity of these materials to assess whether other
vskites are electrochemically stable towards the lithium intercalation. If
ble composition is identified, a perovskite electrolyte may prove highly
uctive and relatively easy to fabricate and handle in air.
Chapter 20
3.2. Nasicon-Type Phosphates
Another widely studied class is the lithium analogues of the so-called

con phase (Na §.uper!onic conductor). The general composition is
xM2(P04h with 0<x<3. More than a dozen formulas containing different
binations of M cations have been reported. The structure consists of
and M06 polyhedra linked at corners to form open 3D tunnels in the
tal. The original sodium material contained M=Zr, however the smaller
els formed with M=Ti were a better fit for the lithium conductors giving
43
lk (lattice) conductivity of 3 mS/cm. Although controversial in earlier
4
k, it now appears clear that lower conductivities of <10- are attributed
grain boundary contribution to the total impedance. A reasonable
uctivity has also been reported for thin films formed by several cycles
44
ltrasonic spray pyrolysis followed by an 800-1000°C anneal.
As for the perovskites, the potential for lithium insertion compensated
he reduction of the M cation may prohibit their use with most lithium
e materials. Again the Ti 4+ is readily reducible, but the Al,Ta or Al,Ge
45
positions are expected to be electrochemically stable with Li. .4 6
ortunately, this has not been evaluated by either cell fabrication or
ammetry studies. The results in Table 2 for the Al,Ti material show the
experimental evaluation of this Li insertion reaction, which occurs at
47
V versus Li/Li+. Because this lithiation reaction was slow, the
archers where encouraged to add a LiB02•LiF sintering aid which
ked the electronic conductivity while only reducing the lithium
uctivity 2-fold. With this composite electrolyte, they succeeded in
positions
aring
studied,andhave beena Li4Ti5012/LiMn204 battery. Numerous other
.4cycling
48 largely to elucidate the relation between
9
structure and ion mobility. Several with high conductivities are

ented in Table 20.2, including two with aLi-rich composition.
In additional publications, researchers report the properties and
hesis of several glass-ceramic lithium electrolytes for which the ceramic
e has the Nasicon-type crystal structure. The anticipated advantages of a
glass ceramic include: a reduced grain boundary resistance, dense
ostructure, and greater ease of fabrication. Conductivities as high or
er than the sintered ceramic material have been reported for the Al,Ti
50 51
Al,Ge Nasicon grains crystallized from the mother glasses. · As the
al temperature is increased, grain growth leads to enhanced conductivity
reduced activation energy. Using a different approach, grains of
z(P04)3 were mixed with powders of the LizS•SiS2•LhP04 glass
52
ussed above. Small amounts of the ceramic powder, 10 wt.%, had
effect on the conductivity of the sulfide glass.
3.3. Lisicon-Type Materials
Lithium .§.Uperionic conductors have an AB04 composltton and a

cture known as yLhP04 in which the B tetrahedra share comers with
ters of three A04 tetrahedra. Solid solutions are formed with mixed
ns at both the A and B sites, commonly Li and Zn at A sites, and V and
at B sites, but other combinations as well. One of the few thin film
trolytes of this phase is reported in Table 20.2.53 In the bulk material,
composition is Li3.4(V0.6Si0.4)04 with a good (1 mS/cm) lithium
uctivity and low electronic component (1:.,=10-)7 . However, when
sited as a thin film by sputtering, a large excess of LhO is incorporated
the highly disordered material which compromises the electron
sference number. Appreciable self-discharge currents were observed for
film batteries of Li versus Mo, W, or Mn oxide cathodes. The authors
optimistic about improving this performance, however a disordered
cture with transition metal cations will likely always be susceptible to
age currents.
A number of sulfide compounds with a lisicon structure are newly
rted.54• 55 As with the sulfide glasses, care must be taken to protect the
tive materials from the air and oxide containers. The compositions
ng the maximum conductivity for two different solid solutions are given
able 20.2, the Ge,P solid solution being most attractive. A Hebb
ner analysis for the phosphorus compound indicates a low electronic
ference number when polarized up to 2V. The authors report
rochemical stability with lithium metal and for polarization up to -5V
us Li, although the results are difficult to read and the current appears to
10!LAfcm2 at 5V suggesting a significant reaction. Batteries formed of
pacted In/LiCo02 with the more resistive Ge,Ga electrolyte give an
al discharge capacity of -80 mAh/g of LiCo02 at low currents. These
alled 'thio-lisicon' materials deserve further investigation.
.4. Lithium Metal Halides, Nitrides, and Phosphides

Earlier reports of halide compositions were mainly: (1) inverse
els, Li[LiM]Cl4, with M being a wide variety of divalent transition
ls; (2) various combinations of metal nitrides with LhN, e.g. Li 8SiN4;
(3) several mixed Li3P-Li3N compositions? Many of these have good
uctivities at higher temperatures, but are prone to decompose at low
ges versus lithium making them unlikely candidates for Li-ion batteries.
ral different structures identified in recent publications are highlighted
able 2. In all cases, the materials must be carefully dried and prepared in
otective atmosphere. None of the studies have advanced to the point of
ng electrochemical cells or forming the electrolyte as a thin layer. The
Chapter 20
fluorite structure of doped-Li 9N2Cl3, and the unidentified structure of

i 2Ch both have modest low temperature lithium conductivity and
trochemical stability up to 1.5-2.5 V versus Li.56•57 A higher room
perature conductivity
58 is reported for the Li3InBr6, which is considered a
rionic conductor. In the absence of a superion conducting phase,
posite materials may show high conductivity attributed to the disorder
space charge layers formed at phase boundaries, grain boundaries, and r
extended defects. This has been demonstrated for LhP + 3LiCl
9
posite electrolyte.5 The last electrolyte in the table is a soft solid, hence
really a ceramic electrolyte. The combination of a lithium salt and
ng Lewis acid gives a good low temperature conductor that might readily 60
a good interface contact with high surface area electrode materials.
3.5. Other Compositions

Two families
3 of electrolytes that received detailed coverage in the
er review, are not included here. For the electrolytes formed of LizS04
related phases, an adequate lithium transport is limited to the high
erature phase making ambient temperature batteries with these
rolytes impractical. In the second case, no new research has been
rted for electrolytes formed of solid solutions with lithium orthosilicate,
i04. Earlier work reported a large number of these electrolytes doped
a variety of penta-, tri-, and di-valent cations producing large
entrations of mobile lithium defects. Although conductivities are
est, some compositions have proven electrochemically stable and good
61
ronic insulators as was demonstrated for Li9SiAI08 .
A final electrolyte, which eluded the earlier reviewer, is the lithia
62
d boron phosphate xLizO•BP04. These researchers adopted
shingly unique and practical approaches to both forming the electrolyte,
arious soft-chemical syntheses, and a complete solid state battery, by the
63
mic compaction of laminated sheets. Unfortunately, the completed
ries have so far proven more resistive than anticipated from studies of
lectrolyte.
. MODELS FOR ION CONDUCTION

4
Building on earlier models, new work continues to explore ion
64 68
on in a disordered structure. - This includes ion motion in crystalline
rolytes lacking long range order of the lithium, as well as in the glassy
olytes. The description of ion transport as the random walk of
mally activated point defects is not appropriate in these systems. Most
y and Ceramic Electrolytes for Lithium and Lithium-Ion Batteries 635
tion has focused on describing the power-law frequency dispersion of

conductance and NMR studies. Experimental conductivity data span
y orders of magnitude in frequency (de to far infrared) and a wide range
64
mperature. Scaling the conductivity isotherms by a single activation
gy collapses the data onto a single universal curve, good evidence that
harge carrier concentration is not thermally activated.
64
The model emerging to describe the dispersion at moderate
eratures, where the conductivity exceeds the de value, is of an ion
ing backwards and forward many times without contributing to a net de
uctivity. A productive hop occurs only when the neighboring ions relax
vor the new site over the old, before the ion can hop backwards. For
electrolytes, there is experimental evidence for the existence of two
s of target sites, one being more optimally configured for the mobile ion.
e optimum sites form stepping stones for the ion through the network or
64
e. Models of the short-range ordering in lithium glassy electrolytes
65 66
ict such weak structural inhomogenities. ' Although these are
ciated mainly with the distribution of a lithium modifier salt, such as
, similar inhomogeneities likely occur in other glass compositions. The
relaxation model has also been applied to the perovskite lithium
67
rolytes in which lithium is disordered over the vacant A sites. A
rent dispersion is observed at very low temperatures for both glassy
64 68
rolytes and the lithium perovskites. ' The ion conductivity becomes
ortional to the frequency, forming the so-call 'nearly constant loss'
n, which is believed to reflect the ions vibrating within their potential
s.
Ion transport across the electrolyte - electrode interface has received
h less attention, but is critically important for achieving low resistance
state batteries. Simulations and model calculations for the case of a
y electrolyte-crystalline electrode, highlight the importance of the
69
face energy barrier, the mismatched hopping frequencies, and easy
sion paths provided by favorable orientation of the crystal lattice or
70
boundaries.
. CONCLUSION
At the most recent International Meeting on Lithium Batteries

LBll, July 2002), ceramic and glass electrolytes accounted for only a
l percentage of the contributions. Attention is clearly focused on the
rmance of polymer, gel and polymer composite electrolytes, while d
electrolytes continue to be used for most routine electrode studies. he
most part, cells with the organic electrolytes are cheaper and faster to cate
and have acceptable internal resistance and cycle life due to good ess in
controlling the solid electrolyte interphase (SEI) layer formed at
Chapter 20
lectrode- electrolyte interfaces. Although the consumer Li-ion batteries

likely continue to use organic electrolytes, other applications may
fit from the introduction of ceramic or glass electrolytes, including
ications requiring extreme (high or low) operating temperatures,
imum lifetimes, negligible self-discharge, and extremely thin profiles.
Most research on ceramic and glass electrolytes still falls short of what
eeded to address practical application of the materials. Results of the
trochemical stability have been most neglected and this can only be fully
uated by actually making and cycling complete Li or Li-ion cells. Also
ing are reports of fabrication techniques with an emphasis on
mizing the thickness of the electrolyte layer. Of the materials reviewed
his chapter, several standout as particularly promising, including: the
ter deposited Lipon thin films, the slow quenched LizS•SiS2 glasses
ilized with a few percent Ta or Sn sulfide, and the Lit.4Get.6A10.4(P04h
s ceramics. The thio-lisicon also looks interesting, but requires
tional research.
Another valuable approach is to create laminated electrolytes using the
mic or glass electrolytes in conjunction with a liquid or polymer
rolyte. This has been demonstrated using lipon-coated separator
branes, lipon-coated cathode films, and glass-coated metallic lithium
71 72
es. • The addition of such a mechanically hard, single-ion electrolyte
to the liquid or polymer electrolyte layer may eliminate the formation of
um dendrites, inhibit chemical and dissolution reactions of the electrode
rials, and reduce the concentration gradients that form due to anion
sion within organic electrolytes. In addition, ceramic electrolyte
ders are obvious candidates for the filler particles of composite polymer
rolytes.
KNOWLEDGEMENT
I wish to thank Dr. Young-II Jang for assistance in preparation of this

uscript and the Department of Energy for supporting our research on
, composite, and B-alumina lithium electrolytes through contract DE
5-000R22725 with UT-Battelle, LLC.
d Ceramic Electrolvtesfin· Lithium and Lithium-Ion Batteries 637
RENCES
. Lee, K. H. Joo, J. H. Kim, J. H. Woo, H.-J. Sohn, T. Kang, Y. Park, J. Y. Oh,
Stare Ionic.\ 149 (2002) 59.
. Rao, M. Ganguli, "Lithium Ion Conducting Glasses", in Handbook of Solid State
eries and Capacitors, M. Z. A. Munshi, Ed., Singapore, World Scientific, 1995.
. Jones, J. R. Akridge, "Lithium Glassy Electrolyte Batteries", in Handbook of Solid
Batteries and Capacitors, M. Z. A. Munshi, Ed. Singapore, World Scientific, 1995.
. Robertson, A. R. West, A. G. Ritchie, Solid Stare Ionic.\· 104 (1997) 1.
unke, Prog. Solid State Chem. 22 ( 1993) 111.
manaka, T. Shimizu, G. Adachi, Solid State Ionic 62 (1993) 167.
Bates, N.J. Dudney, G. R. Gruzalski, R. A. Zuhr, A. Choudhury, C. F. Luck, J.D.
ertson, 1. Pmn!r Sources 43-44 (1993) 103.
u, J. B. Bates, G. E. Jellison Jr., R. X. Hart, J. Electrochem. Soc. 144 (1997) 524.
. Bates, N. J. Dudney, B. J. Neudecker, A. Ueda, C. D. Evans, Solid State Ionic. ·
(2000) 33.
ereda, R. B. Goldner, T. E. Haas, P. Zerigian, Electrochem. Solid-State Lett. 5
2) A239.
. Park, S.-H. Lee, B.-I. Lee, S-K. Joo, E/ectrochem. Solid- State Lett. 2 (1999) 58.
Jeon, Y. W. Shin, S. C. Nam, W. I. Cho, Y. S. Yoon, J. Electrochem. Soc. 148
1) A318.
ee, S. A. Akbar, C. 0. Park, Sensor Actuat. B-Chem. 80 (2001) 234.
. Yoon, W. L Cho, J. H. Lim, D. J. Choi, 1. Power Sources 101 (2001) 126.
Dudney, B. J. Neudecker, Curr. Opin Solid State Mater. Sci. 4 ( 1999) 479.
ereda, N. Clay, A. Gerouki, R. B. Goldner, T. Haas, P. Zerigian, 1. Power Sources
2000) 20 l.
hao, Z. Fu, Q. Qin, Thin Solid Films 415 (2002) 108.
. Choi, W. I. Cho, B. W. Cho, H. S. Kim, Y. S. Yoon, Y. S. Tak, Electrochem. d-
State Lett. 5 (2002) A 14.
. Roh, S.-D. Lee, H.-S. Kwon, Scripta Mater. 42 (2000) 43.
ang, B.S. Kwak, B. C. Sales, J. B. Bates, J. Non-Cry.1·t. Solids 183 (1995) 297.
Dudney, in preparation.
e Sauze, E. Gueguen, R. Marchand, J. Non-Cry.1"t. Solids 217 (1997) 83.
irkc, W. F. Chu, W. Weppner, Solid State Ionic.\ 93 (1996) l.
. Kennedy, Z. Zhang, J. Electrochem. Soc. 136 ( 1989) 2441.
hang, J. H. Kennedy, Solid State lonics 38 (1990) 217.
Morimoto, H. Yamashita, M. Tatsumisago, T. Minami, J. Am. Ceram. Soc. 82 (1999)
2.
ayashi, S. Hama, H. Morimoto, M. Tatsumisago, T. Minami, 1. Amer. Ceram. Soc.
2001) 477.
inami, A. Hayashi, M. Tatsumisago, Solid State Ionics 136-137 (2000) 1015.
. Bartholomew, D. M. Young, A. J. G. Ellison, J. Non-Cry.1t. Sol. 257 (1999) 242.
Hayashi, H.Yamashita, M. Tatsumisago, T. Minami, Solid State lonics 148 (2002)
omiya, A. Hayashi, H. Morimoto, M. Tatsumisago, T. Minami, Solid State Ionics

(2001) 83.
akada, N. Aotani, K. Iwamoto, S. Kondo, Solid State Ion in· 79 ( 1995) 284.
akada, T. lnada, A. Kajiyama, M. Kouguchi, S. Kondo, M. Watanabe, J. Power
ces 97-98 (2001) 762.
akamoto, M. Tatsumisago, T. Minami, 1. Phys. Chem. B 103 (1999) 4029.
o, A.Hayashi, H. Morimoto, M. Tatsumisago, T. Minami, Chem. Mater. 14 (2002)
4.
inatier, M. Menetrier, A. Levasseur, Solid State !onics 116 ( 1999) 35.
an·a, A.Varez, C. Leon, J. Santamaria, M. Torrcs-Martfnez, J. Sanz, Solid State
n· 134 (2000) 219.
Chapter 20
K Mizumoto, S. Hayashi, Solid State Jonics 127 (2000) 241.

Y. Inaguma, L.Q. Chen, M. Itoh, T. Nakamura, T. Uchida, H. Ikuta, M. Wakihara, Solid
State Commun. 86 (1993) 689.
S. Garcia-Martin, J.M. Rojo, H. Tsukamoto, E. Moran, M.A. Alario-Franco, Solid State
Jonics 116 (1999) 11.
0. Bohnke, C. Bohnke, J.L. Fourquet, Solid State lonics 91 (1996) 21.
H. Watanabe, J. Kuwano, J. Power Sources 68 (1997) 421.
H. Aono, E. Sugimoto, Y. Sadaoka, N. Irnanaka, G. Adachi, J. Electrochem. Soc. 136
(1989) 590.
A.K lvanov-Schitz, A.V. Nistuk, L.N. Dernianets, N.G. Chaban, Solid State /onics 144
(2001) 133.
V. Thangadurai, A.K. Shukla, J. Gopalakrishnan, J. Mater. Chem. 9 (1999) 739.
H. Yamamoto, M. Tabuchi, T. Takeuchi, H. Kageyama, 0. Nakamura, J. Power Sources
68 (1997) 397.
P. Birke, F. Salam, S. Doring, W. Weppner, Solid State /onics 118 (1999) 149.
R. Sobiestianskas, A. Dindune, Z. Kanepe, J. Ranis, A. Kezionis, E. Kazakevicius, A.
Orluikas, Mat. Sci. Eng. B-Solid B76 (2000) 184.
T. Suzuki, K Yoshida, D. Uematsu, T. Kodama, K Toda, Z.-G. Ye, M. Sato, Solid State
onics 104 (1997) 27.
J. Fu, Solid State Jonics 96 (1997) 195.
B. V. R. Chowdari, G. V. S. Rao, G. Y. H. Lee, Solid State Jonics 136-137 (2000) 1067.
K Takada, M. Tansho, I. Yanase, T. Inada, A. Kajiyama, M. Kouguchi, S. Kondo, M.
Watanabe, Solid State Jonics 139 (2001) 241.
J. Yamaki, H. Ohtsuka, T. Shodai, Solid State Ionics 86-88 (1996) 1279.
R. Kanno, T. Hata, Y. Kawamoto, M. Irie, Solid State lonics 130 (2000) 97.
R. Kanno, M. Murayama, J. Electrochem. Soc. 148 (2001) A742.
Y. Jia, J. Yang, Solid State Ionics 96 (1997) 113.
J. Yang, Y. Jia, Z. Yao, Solid State Ionics 96 (1997) 215.
Y. Tomita, H. Ohki, K Yamada, T. Okuda, ZNaturforsch. A 55 (2000) 117.
G.A. Nazri, R.A. Correll, C. Julien, Solid State Ionics 86-88 (1996) 99.
X. Wei, D.F. Shriver, Solid State Jonics 133 (2000) 233.
B.J. Neudecker, W. Weppner, J. Electrochem. Soc. 143 (1996) 2198.
E.M. Kelder, M. J. G. Jak, F. de Lange, J. Schoonman, Solid State lonics 85 (1996) 285.
M. J. G. Jak, F. G. B. Ooms, E.M. Kelder, W.J. Legerstee, J. Schoonman, A.
Weisenburger, J. Power Sources 80 (1999) 83.
K Funke, C. Cramer, B. Rolin, Glastech. Ber. Glass Sci. Techno[. 73 (2000) 244.
. Swenson, C. Karlsson, L. Biirjesson, R.K Heenan, Phys. Rev. B 64 (2001) 134.
P. Vinatier, M. Menetrier, A. Levasseur, Solid State Ionics 136 (1999) 35.
K Mizumoto, S. Hayashi, Solid State /onics 127 (2000) 241.
A. Rivera, A. Varez, J. Sanz, J. Santamaria, C. Leon, J. Alloys Compds. 323-324 (2001)
545.
Deppe, R.F. Wallis, I. Nachev, M. Balkanski, J. Phys. Chem. Solids 55 (1994) 759.
S.H. Garofalini, P. Shadwell, J. Power Sources 89 (2000) 190.
N.J. Dudney, J. Power Sources 89 (2000) 176.
E.S. Nimon, M. Sukeshini, S.J. Visco, Presentation No. 221, International Meeting on
Lithium Batteries, June 23-28, 2002, Monterey, California (to be published within J.
ower Sources).
(mole ratio)
Li20•Se02•B203 (0.5, 0.25, 0.25) 1:>
5.
Q
LiCl•Li 1.4M2(P03)4 M=Al+Ti+Ge
(0.15, 0.85) ;:;·
Li.POyNz "
(x=2.9, y=3.3, z=0.46, typical) (0.24 < z < 1 .2)
"' 'o>
Li. POy Nz ...
t""'
§:
-
Li,BOy Nz
§
:::...
t""'
§:
Li 2S•P2S5 ii:'
(0.75, 0.25) :!
::s
1
Ea, activation energy for Li conduction from simple Arrhenius relation near room temperature. Those shown in ( )
were estimated
from the published graphs.
·
2
t., transference number= transport number for these single ion conductors, generally by Wagner de polarization .
3 ll.V, low/high voltage range of the electrochemical window with respect to Li+/Li.
0\
w
\0
Li2S•SiS2
(0.6, 0.4)
LizS•SiS2•LixM04 M=Si,P,Ge
(0.57, 0.38, 0.05)
LizS•SiS2•LixM04 M=Si,P, Ge
(0.57, 0.38, 0.05)
Li2S•SiS2•Li3P04 (0.63, 0.36, O.OI)
Li2S•SiS2•xMSy M=Sn,Ta,Ti, etc.

(0.6, 0.4, 0.01-0.05)
Li2S•SiS2•Li3N (0.55, 0.40, 0.03)
Li3N•SiS 2
(0.4, 0.6)
9
-€;;:
....
N
0
{mole (kJ/mol) {V)
ratio)
'':"''"
l..a213-xLiJzTi03 solid state reaction & 1 X 10'' 0.7 28±2 1.51>4
s.
Q
cyclic voltamrnetry [37,
0,03 Sx :s; 0.167 pressed x=0.13 10"
& titration 41]
l..a113-xLi3zTa03 solid state reaction & 9x 10"' 2 !'
[38] ;
0.025SxSO.I67 pressed x=0.06 to"' ·
I..a113-xLi 03 solid state reaction & 4x 10·> 2 35±3 1.81?

cyclic voltamrnetry [40] a
'<'
0 SxS0.06 pressed x=0.04
vs. Li
Li1.3Ti1.7Alo.3(P04)] solid state reaction & 3 X Ia:' 0.2-0.3 2.41? .,
[47]
t
Coulometric s
pressed {3 x to· · see :
text titration vs. Li e.
LiAITa(P04)] solid state reaction & 6.5 X 10' 45 ·
[45] :!
pressed
LiAlo.4Ge1.6(P04)] solid state reaction & 1.3 X 10... 42
[46]
pressed
-
Li ,4TiuYo.4(P04)] solid state reaction & 4.5 X 10·> :I
[48]
pressed (1.8 X 10'3)0
-
-
-
-
-
-
-
-
·
-
Lil·2.t(Sc1.,M.J:z(PO,J, M = Zr, Ti, .x=O.l,0.2
Li3Sc ,5Feo.s(P04)]
Li6.1Yo.61Sio.J90s.l6
1.4..rGeJ•.Pa.S4 (.r-:0.25)
L4.rGeJ.J> ..s.
(.x=0.6S)
Li9Bi2Cl3
Li31nBr6
Li3P+3LiCI
composite
LiCF3S •NbF, (0.33, 0.37)
Li,SiAl01
Li_.BPO..,xn [62)
x=0.07
- -
• Indicates the bulk or crystalline component of the total conductivity.
.,
1'\
apter 21
ITHIUM-ION BATTERIES FOR EV, HEV AND

OTHER INDUSTRIAL APPLICATIONS
M. Broussely
Saft BP 1039, 86060 Poitiers, France
1. INTRODUCTION
So called "industrial batteries" are generally sized in the range from
to hundreds of kWh, addressing a market roughly divided in two
ments: transportation and stationary applications. This market is mainly
essed by lead-acid and NiCd systems. Large special batteries for space
defense generally use more advanced and sophisticated chemistries, such
ickel-hydrogen, silver-zinc, silver-aluminum oxide, etc. One of the most
ting questions in this battery business today is: will the "Li-ion
lution" which occurred in portable batteries extend to larger battery
ems? Obviously, that would only occur if two main conditions were
fied: better answer to a real need, and in a cost-effective way. This
ter will review the most important applications in which Li-ion might
g benefits, and will describe the technical status today.
Going back a little over the last 5 to 10 years, the possible use of Li
in large batteries was first considered in electric vehicles, for which high
ific energy is crucial to grant a sufficient range. In parallel with the
lopment of portable Li-ion, very important R&D programs started about
ears ago aiming at demonstrating the feasibility of such batteries in
tric vehicles. These programs were supported by a strong financial effort
over the world: European Commission and french government
nizations in Europe, DOE I USABC in USA, MITI I NEDO I LIBES in
n (addressing first the stationary batteries for home individual power
ces). Saft was one of the few companies involved from the beginning.
n, with the fast developments of REV's, Li-ion also had the opportunity
emonstrate its very high potential in terms of specific power and power
ity.
Chapter 21
Very soon the numerous demonstrated advantages of Li-ion over other

nced systems established it as the best power source for satellites and
r space applications, leading to specific developments and qualified
ry systems. The introduction of Li-ion is now being considered in other
ications, as actual complete battery systems of different kinds are being
.
1.1. Specific Properties Required of Industrial Li-Ion

Batteries
As will be described in more detail in the following sections for each
application, the general specifications are often more demanding than
e for consumer portable electronics. Beside very high power or energy,
nding on the application, longer life is generally required of this high
ed piece of equipment, which means in some cases large number of
es (typically>lOOO), or long shelf life (10 to 15 years), sometimes in an
ntrolled temperature environment. The cost is also an important issue. g
newly introduced, Li-ion is still an expensive technology, largely use
of material costs. Cost reduction, coming with an increase in ufactured
quantities, is the main factor for future expansion of this new nology in
industrial battery markets. As a consequence, once the mum
electrical specifications are met, low material cost is the main meter
that drives the designer's choices.
2. WHICH Li-ION CHEMISTRY FOR A LARGE

BATTERY?
.1. Positive Materials
This is certainly the critical part of aLi-ion cell that is responsible for
differences in cell properties. The use of lithiated transition metal es,
e.g. LiNi02 and LiCo02, as possible positive materials for
1
rochemical power sources was discovered more than 20 years ago.
e then, these compounds were widely investigated as positive materials
2
rechargeable lithium cells. •3 When LiCo02 became the undisputed
4
ode of the Li-ion, it was available in very small quantities, and LiNi02
ly did not exist. Studying the possible preparation processes for
i02,5 the difficulty of preparing a "good" electrochemically active
rial was pointed out. It was later found that the tendency for Ni2+ to
py sites in the u+ layers introduces a structure distortion, which
ically affects its ability to de-insert/insert Li+ ions.6 Therefore, in spite of
Lithium-Ion Batteries for HEVIEV and Other Industrial Applications 647
potential advantage of LiNi02 in energy density and of its greater

ility on overcharge, LiCo02 was chosen by almost all the battery
ufacturers to start production of Li-ion batteries for portable
ications. Due to the huge growth of portable phones and personal
puters, this system is now established as a standard for Li-ion.
However, it was recognized very early that, due to the high and
ewhat unpredictable price of cobalt, it would have been difficult to meet
cost requirements for large systems, particularly for the car industry.
alternatives were consequently explored, using either manganese
el LiMn204) or nickel based oxides. Primarily because of the higher
gy density expected in batteries with the nickel oxide, Saft chose this
78
on early on. • Indeed, lithiated nickel based oxides are generally able to
se up to 0.7 Lt/mol, at 4.1 V vs. LilLi+ without performance
adation on cycling. That represents a potential specific capacity of
nd 200 Ah/kg, compared to 150 Ah/kg for LiCo02 and 120 Ah/kg for
n 204, while maintaining an average voltage of about 3.7 V vs. the
ated carbon anode.
9
Substitution by cobalt (ideally 20%) was early identified as a solution
2
void Ni + formation, allowing easy manufacturing of the appropriate
tural form. In the frame of the LIBES program, the mixed Ni/Co
10
rial LiNi0.7Co0.302 was studied by Sanyo to design a 76 Ah cylindrical
dedicated to home load leveling systems.
For some years, many "doping" metallic ions were tested by
11
erous laboratories, showing interesting specific properties.
12
minum was found to produce a significant stabilization of the
hiated form, inducing a better thermal stability, as shown by differential
13
ning calorimetry (DSC) experiments.
That discovery focused the research work, made jointly with material
ufacturers, towards the definition of a mixed Ni/Co/Al material, which
ars to exhibit excellent electrochemical properties, now recognized by
14
rallaboratories. The exact composition used by Saft is still proprietary,
he ratio of nickel to total metals (except lithium) is greater than 0.75,
wing the expectation of a reasonable cost of the raw material.
Unlike LiNi02, LiNi 1_x-yCoxAly02 does not exhibit monoclinic/
gonal phase transformations during cycling. This may at least partly
ain the excellent stability on cycling, as described later. The crystalline
ture is very similar to that of LiCo02, however the charge discharge
es are different, as shown in Figure 21.1.
Figure 21.2 describes a particular feature of nickel based compounds.
2
to a small remaining quantity of Ni +, the first charge/discharge cycle is
Chapter 21
totally reversible, although better than LiNi02• As a consequence,

use the total excess of lithium is not consumed by the carbon
ivation, the positive electrode limits the cell capacity. At the end of
harge at 2.5V, about 15% of the Li capacity remains in the carbon
trode. If the cell is discharged beyond this voltage limit, a voltage
-- - 1-
-- t--...
1\
-LINIO
-- ueoo,
PC/EC/3DMC+LIPF (1M)
50 100 150 200 (

Capacity (Ahlkg)
e 21.1. Compared voltage profiles Figure 21.2. Deep DOD voltage profile
tal oxides at low rate (C/10, 25°C). of a LiNil-x-yCoxAlyOz/graphite cell at low
rate (C/10). Ambient temperature.
au at about 1.5V is obtained where the lithium is completely consumed,

own in Figure 21.2 for a 40 Ah cell. This interesting feature gives to the
the ability to be deeply discharged even to 0 V without risk of copper
osion unlike other positive materials (in this case the negative electrode
ly at about 1.5V vs. LilLi+). For the same reason, the cell can be stored
he complete discharge state at 2.5V. It should be noted however that
harging the cell at elevated temperature (45°C or more, depending on the
allows the positive material to be totally reduced at high voltage,
lting in about 15% increase of the cell capacity, but corning back then to
gative limited cell.
This particular feature has also a noticeable impact on the cell shelf
When the cell capacity is limited by the negative electrode, any loss of
um directly impacts on the cell capacity, which is generally observed r
on cycling or extended storage at elevated temgerature, because the
ivating layer is not a full electronic insulator. 5 This induces also
anding cell energy stability on cycling or storage, as will be described
.
Finally, the stabilizing property of AI ions on the delithiated form
s the onset of thermal decomposition to a significantly higher
erature (about 100°C), thus reducing the risk of thermal runaway in
of overcharge. In the same manner, the reactivity towards electrolyte is
ced, and that might be also the reason for the good power retention
ng storage. Power loss is indeed generally attributed to electrolyte
ation at the positive electrode interface. 16 This behavior compares very
rably to pure LiNi02 or LiCo02, and is a big advantage over LiMn 204,
which interaction with electrolyte is still a major drawback.
.1.2. LiMnz04
There are two main reasons put forwards to promote the use of the
n204 spinel as the positive for a large Li-ion battery. The first usually
tioned is the cost, expected to be the lowest. However, the cost of Mn02
in primary dry cells is often given as example and goal, ignoring the
ssary lithium salt and added processing costs. In addition, the pure
erial alone fails to provide the sufficient chemical stability to ensure the
ired long life time, and additional processes such as coating are
ssary. Finally, as long as costs of other battery components are not
tically reduced, the final cost/Wh is strongly impaired by the lower
gy density achieved, compared to nickel based materials. In high power
, where the cost/W is more relevant, the lower relative amount of active
erials reduces their influence on final cost. Finally, it is quite difficult
y to predict which one of Mn or Ni based products will lead to the more
effective system, but it can be anticipated that the difference would be
ginal.
Safety is the second mentioned advantage, which is of a particular
ortance for large batteries. This aspect is described in more detail in
her chapter of this book, and is only briefly discussed here. The totally
hiated phase, A.-Mn02, which is the regular full state of charge, is
ally thermodynamically stable. This is not the case of Ni02 or Co02,
h are ultimately formed when LiNiOz or LiCoOz cells are fully
charged. These non-stable forms may decompose via exothermic
tions with the electrolyte and can result in cell burning. This generally
rs when the cell starts to overheat because of electrolyte electrolysis,
the positive material is totally oxidized. The combustion is all the more
tive since lithium metal will have been deposited on carbon electrode,
to the excess of lithium extracted from the positive. Although there are
rent means provided for preventing all Li-ion batteries to go into
charge, the better stability of LiMn2 04 provides an additional safety
l.
That does not mean however that the cell is intrinsically safe and
d be abused in any manner. The oxidized form, A.-Mn02, is still a strong
izing agent and can react violently with organic solvents when
heated to about 300°C. This is usually studied using differential
ning calorimetry (DSC), and can be observed when the experiment
lves a mixture of active material and electrolyte as in a real cell. Figure
describes the DSC heat output from 100% charged materials, in contact
electrolyte in proportions similar to a real cell. It must be pointed out
Chapter 21
the heat power (W) is not related, as is usual, to the weight of the
ple, but to the quantity necessary to bring one Wh to the cell, which is
er to the real case.
The following comments can be made: 1) pure LiNi02 exhibits a low
t temperature for thermal runaway (temperature sharp peak), as low as
t 200°C. 2) Although heating delithiated LixMn2 04 alone would not
uce any thermal effect, the presence of electrolyte leads, as expected, to
ong exothermal reaction at about 300°C. 3) Aluminum doping in the
ed Ni-based material brings a very significant improvement, ranking the
rial to an even better level than LiMn 204 • Therefore, LiMn204-based
e batteries must be protected against overcharging, just like the other
ems. Indeed, a high rate overcharge will produce heat as soon as
trolyte is oxidized, i.e. from the beginning of overcharge, and may reach
point where the positive material reacts with the flammable electrolyte. t
transfer in cell design is also one of the key factors in improving ry
safety, regardless of the electrochemical system.
Beside the advantages of a better safety, a LiMn204-based system still
ers of poor calendar life even at moderately high temperature, which has
attributed to the tendency of Mn 3+ ions to disproportionate into Mn4+
Mn2+. The problem comes from the solubility of Mn 2 + ions in the
rolyte, which reacts on the negative side, and destroys the stability of
passivation layer on the carbon, which is the key point of Li-ion system
ility. Although this can be tolerated for consumer batteries, whose life
ctation is not too high and working temperature normally low, this is a
drawback for industrial batteries. Although improved solutions have
studied in the past years, the problem is not yet satisfactorily solved.
0.04 rc:- ..,--"'CC ,.,.,.., -.-.-.-.-.-,---,-.- --.--, r-:J

Charged positive materials
10°C/min- with electrolyte
0.035 PVDF binder
0.03
0.025
esc
(W/Wh) 0.02
0.015
0.01
0.005
100 200 300 400

500 600
TEMPERATURE ('C)
e 21.3. DSC of different positive materials at full state of charge in the presence of
olyte. Stainless steel crimped sample vessels.
2.2. Negative Materials
There is a wide variety of carbons that can be used as negative

trode to make industrial Li-ion batteries, as with the small portable cells.
ever, because the material costs have a particular importance in the total
ry costs in large systems, low cost graphites are generally preferred.
of the most important properties taken into consideration for the
rial selection is the ability to form a stable passivation layer, which is
amental when long life is needed.
3. GENERAL TECHNOLOGY AND

MANUFACTURING PROCESSES
Basically, there is no difference between the technology implemented
mall portable Li-ion cells and large "industrial" cells. Electrode
ufacturing processes consist of coating metallic foils (Al for positive and
or negative) with a slurry containing the active material, an electronic
uctor (for the positive mix), and a binder dissolved in an appropriate
ent (generally PVDF/NMP). After drying, this coating is calendered to
desired thickness/density. The electrode thickness depends on the
imum power required. This is a very interesting feature of Li-ion battery
nology, allowing a wide range of power/energy ratio designs with the
e basic electrode manufacturing process. However, appropriate current
ction and tabbing, cell shape and design are necessary.
There is also no significant difference in the nature of electrolytes,
6 being the only suitable salt, while cyclic or linear carbonates are the
of the organic solvents. Separators, all in the form of thin (25J.Lm or
microporous membranes, may be either in polypropylene or
ethylene, or mixed polymers.
In most cases, the cell stack is built by rolling the electrodes in spiral,
the case for small cells. Because of their small thickness, this is the
efficient way for assembling electrodes, compared to planar
entional stacking. The coil may be cylindrical, or flattened to make a
matically shaped cell. Both designs have their own advantages and
backs and are currently being produced. Better volume filling in battery
mbly is obtained with prismatic cells, which are however more sensitive
ulging on cycling or aging. The stronger mechanical characteristics of
drical cans give to the cell better dimensional stability, and
ogeneous pressure within the electrode stack. The usual cell range
osed by battery manufacturers is from 3 Ah (high power design) to 100
high energy design).
Chapter 21
3.1. General Considerations for Battery Assembly

Because of the large variety of possible applications, many types of
ry assembly can be designed. These are greatly facilitated by two
erties: the cell SOC/voltage dependence, which allows direct parallel
nection to obtain the required capacity without any electronic
ponent, and the cell ability to work in any position. On the other hand, as
vercharge shuttle mechanism exists in batteries with organic electrolyte,
end of charge and discharge voltages must be monitored for each cell or
llel grouping, which requires specific battery management features,
cially when high voltages are involved (sometimes greater than 300 V).
Progress in high power electronics make this quite achievable,
ever, insuring very precise and stable battery characteristics, with cells
alance, etc. This control works much better than a chemical shuttle
inside cell, as in aqueous systems. Battery management systems may
include isticated intelligent functions, which can communicate
with the ication. The most used information is the battery 'fuel' gauge,
giving a time read out of the state of charge.
Safety is also a major parameter taken into account at the battery
l. The larger size of the battery makes this point even more crucial than
mall batteries. Overcharge is the most critical situation to avoid, which
only happen in case of applying overvoltage to the cells. Redundant
age thresholds are applied to prevent this situation in any circumstance.
rheating resulting from shorts or overloading is also avoided by limiting
maximum current, and by fuses and thermal sensors. Although all types
recaution are taken to avoid hazardous situations, the probability for
ion in a single cell (the non-aqueous electrolyte is flammable) should
be ignored. In such a case, the battery design should insure that fire
ld not propagate to the entire battery.
. EV APPLICATIONS
Reducing the impact of road transportation on the environment is a

datory task for the automotive industry all over the world. With no
ust fumes, the electric vehicle is an optimum solution for urban
portation and goods delivery.
Electric vehicles have been traditionally powered by flooded lead-acid
ickel-cadmium batteries. These two technologies suffer from a low
gy density that makes impractical a long-range vehicle, thus limiting
ic acceptance for such types of vehicles. Thanks to an energy density
is three to four times higher, Li-ion allows carmakers to design electric
cles with very acceptable performance in term of range, top speed and
leration.
Lithium-Ion Batteries for HEV/EV and Other Industrial Applications 653
4.1. EV Battery Requirements and Technology

Comparison
.1.1. Weight
The energy consumption of an electric vehicle is proportional to its

ht. A typical figure for an EV is 120 Wh/ton/km. Since the allowable
ht for the battery onboard a given vehicle is limited, the better the
gy density, the larger will be the available energy and the vehicle range.
mparison of the different technologies is presented in Table 21.1.
e 21.1. Comparison of different technologies of EV batteries for a typical size of 250 kg.
hnology
icle curb weight
tery weight allocation (typ.)
tery structure cooling etc.
dule weight allocation
rgy density (module)
board energy
c. range @ 120 Whlton/km
21.2. Comparison of different technologies of EV batteries for a typical size of 200
chnology
ttery volume allocation (typ.)
ttery structure, cooling etc.
dule volume allocation
lumetric Energy (module)
board energy
nge @ 120 Wh/tonlkm
.1.2. Volume
It is obvious that the volume that can be allocated to a propulsion

ry onboard a vehicle is limited. A typical performance comparison is
ribed in Table 21.2.
.1.3. Power
The vehicle's ability to accelerate and climb hills depends on the

ry power, if it is not limited by the electric motor power. Here again, Li
xhibits superiority over the traditional chemistries, as described in Table
Chapter 21
e 21.3. Power characteristics comparison for a 250 kg battery using different chemistries.
hnology
icle curb weight
ery weight allocation (typ.)
ery structure cooling etc.
ule weight allocation
er density (module)
ery Power
.1.4. Long Term Parking Capability
Aqueous electrolyte systems suffer from moderate to high self

harge that may reach several per cent of the rated energy per day. This
nomenon is reversible (the lost energy is recovered after the next charge)
it limits the long term parking capability of the vehicle. For some of
e systems, a one-month stand at the airport parking for instance cannot
nvisaged. A Li-ion battery, thanks to its low self-discharge rate (a few
cent per month), can be left, even at 20% state of charge, during
nded periods of time without the need of recharge.
4.2. Battery Monitoring and Management
.2.1. Fuel Gauge
Electrochemistries using aqueous electrolytes (PbA, NiCd, NiMH) are

cult to manage since there is no direct way of measuring the battery state
harge. The faradic charge efficiency much lower than 100% (dependent
arameters such as temperature, state of charge, charge rate), the high self
harge rate, etc., make it difficult to assess correctly the battery state of
ge by a coulometric method. When combined with the short-range of the
cles that are equipped with these types of batteries, users suffer from a
city gauge syndrome and experience shows that they barely use 50% of
nboard energy between two consecutive charges.
Li-ion exhibits an almost linear relationship between the open circuit
age and the state of charge of the battery, the faradic charge efficiency is
%, and the self-discharge is low. These facts make an accurate capacity
e easy to implement, typically better than 5%. When associated with the
-range capability of the vehicles fitted with the Li-ion batteries, the fear
flat battery should disappear.
.2.2. Charge Monitoring
An economic drawback of the Li-ion technology is its lack of

charge protection. It requires hardware to insure the monitoring of
age of every group of cells in parallel to avoid any overcharge situation.
ever, the knowledge of every cell voltage (and therefore the state of
ge) allows very precise battery management that helps to improve the
ry life.
4.3. Li-Ion Technology for EV

.3.1. Chemistries
As mentioned earlier, LiCo02 was soon abandoned in large cells for

because of its cost. Manganese-based or nickel-based are the two
ons presently considered for automotive applications. While LiMn204 is
ctive for safety considerations, it has such disadvantages as a low
tical specific capacity of about 110 Ah/kg increasing the potential
Wh of the battery, and a limited calendar life at moderately high
erature (see Section 21.2.1.2.). Most japanese EV battery manufacturers
working on the manganese option 17 while Saft adopted the "nickel
tion". Ongoing research programs on both sides, with the help of
rials manufacturers, aim at finding the best material that would combine
est features of each option.
Each EV battery maker has its own proprietary choice of negative
rial and electrolyte solvent mixture. The use of LiPF6 salt is a standard.
Cell general composition of Saft industrial Li-ion chemistry:
-Positive: LiNixCoyAlz02+ Carbon+ PVDF, on AI substrate (.x>0.75).
-Negative: Graphite blend+ non fluorinated binder on Cu substrate.
-Electrolytes: Carbonate blend, as a function of cell type, 1M LiPF6 +
vc.
-Separator: microporous PE or PP.
.3.2. Cell Design
The general cell design is described in Section 21.3 All battery makers
selected the wound coil technology that is the optimum industrial
ion to achieve a low process cost. These coils are then inserted into
drical or round-edge prismatic metallic cans. The can material may be
el-plated steel or aluminum depending on which of the coil polarities is
ected to the cell structure. A pressure release valve
Chapter 21
21.4. Main characteristics of high-energy Saft cells.
ls
minal Voltage
d Capacity at C/3
ical Power at 80% DOD, 30s pulse
ensions (0 I h)
ical Weight
cific Energy
cific Power at 80% DOD
rgy Density
er Density at 80% DOD
ents cell bursting in case of overpressure.

For pure EV applications, the battery manufacturers have developed
es of cells with individual capacities between 25 and 100 Ah. As an
ple, the main characteristics of high-energy cells manufactured by Saft
escribed in Table 21.4.
Shin Kobe Machinery17 has developed LiMn204-based batteries for
n collaboration with Hitachi Maxell. The performance characteristics of
cylindrical cell used to build 2.7 kWh modules, then assembled into
ries, are summarized in Table 21.5. Nissan has equipped vehicles with
e batteries.
Japan Storage Battery developed a range of high-energy cells of
tical cylindrical shape also using LiMn204 (Table 21.6).
Matsushita Batteries Industrial Co. Ltd designed a cylindrical 123 Ah
18
in the frame of the LIBES program using LiMn 204 and graphite. Its
21.5. Main characteristics of Shin- Kobe high-energy cells.
Capacity
Nominal voltage
Whlkg
Wh/1
W/kg at 85% DOD
21.6. Main characteristics of JSB high-energy cells for EV (from JSB catalogue).
Outer Dimensions Specific

Nominal Nominal Specific
del Voltage Capacity (mm) Weight Power
Energy
(V) (Ah) (kg) 80% DOD
Width Thick. Height (Whlkg)
(W/kg)
V6P 3.75 6 85.5 109 19 0.33 68 >500
V40P 3.75 40 133 170 47 2.1 71 >700
V95P 3.75 95 114 170 94 3.5 102 >700
acteristics are described in Table 21.7. A 30.4 V- 3 kWh module

designed using 8 of these cells connected in series. More recently,
ller sizes were also designed.
e 21.7. Main characteristics of Panasonic Table 21.8. Main characteristics of NGK

energy cell for EV. Insulators cell.
acity Capacity
ight Energy
ensions Energy density
tage (V) Specific energy
19
NGK Insulators designed a 25 Ah cylindrical cell using substituted
ganese spinel, described in Table 21.8. The material and appropriate cell
gn allow to get a an improved safety behavior in severe abuse
ditions, such as nail penetration and overcharge. High power cells are
developed (see 21.5.3.3).
.3.3. Battery Design
The size of an EV battery design is of course dependent on the vehicle

. Typically, battery energy is in the range of 10 to 30 kWh, with a
age of 300 V or more. The batteries are built from modules, each
prising a number of cells with total energy in the range of 1 to 2 kWh.
' s EV module comprises 6 cells that may be configured in any series or
lel combination - see Figure 21.4. The module includes an electronic d
that consolidates cell measurements and transmits data (voltage,
perature) to the battery controller. Because of the lack of an overcharge
tle mechanism, differences in cell voltage in series are re-equilibrated by
tronic means. The charge imbalance, caused by uneven self-discharge, is
lly low in the Li-ion system. Re-equilibration is accomplished simply
automatically by slowly discharging cells with the highest voltages
g resistors. This switching is included at the module level. The module is
designed to control cell temperature. The Li-ion chemistry has a very
entropy coefficient, and the temperature gradient inside a battery is not
itical as it is for some aqueous battery systems (NiMH for instance).
Chapter 21
There is no extra heat on charge,

and heat generation during a drive
cycle is low. Air or liquid cooling
can be used, the selection being
left to the automotive
manufacturer. Air-cooling is
lighter and cheaper but tends to be
noisy and does not allow for
battery heatingduring cold
temperature operation. The liquid
cooling, using the car
infrastructure, is more compact
and quieter but heavier and more
e 21.4. View of Saft EV Li-ion expensive. It allows for battery
le and cells. heating for cold temperature
operation.
In the frame of the LIBES program, the japanese battery makers, in
boration with their automotive industry, selected the air cooling
tion. In the frame of an european program, Saft in collaboration with
AR (european carmakers association) selected the liquid cooling
tion. The modules are then assembled into batteries, to reach the size
ired by the customer. A standard battery system is composed of:
• An assembly of modules electrically connected in series
• A cooling system that may be limited to fans in case of air
cooling and is composed of a pump, hydraulic manifolds, a
thermal regulation valve and a liquid-to-air heat exchanger in
case of liquid cooling
• A fuse to protect the battery against external short circuit
• A contactor to insulate the battery from the vehicle when parked.
This contactor may also be connected to a vehicle shock detector
to insulate the battery in case of an accident
• A ground fault detector which is mandatory for high voltage
batteries
Most of these subsystems are managed by a battery controller that
cts data coming from the module electronic boards, runs the various
ry algorithms (charge, capacity gauge, protection against abuse, etc.)
communicates with the vehicle.
A picture of a battery used in the Daimler Chrysler EPIC van is shown
igure 21.5. The battery characteristics are: 38.5 kWh, 315 V average
ge, >80 kW, 330 kg (excluding battery tray), 212 dm 3• It provides to
a 200-km+ range.
e 21.5. Open view of the Saft Li-ion battery for EPIC Daimler Chrysler van.
.3.4. Typical Performances in Bench Tests
Figure 21.6 describes discharge curves of the Saft "VLE" cell at

rent discharge rates. It can be seen that even this high-energy cell (150
kg) still sustains high discharge rates.
c--.----.--.- 500 -c----------250,0
:t::;_,;,::; :;;;:;;.::;.::..,-:::.:- --..-_- _-+

200,0:
.: 0 350 30
7 - z; i: ;:;# - ---
5 -':'
; g 300 + 150,0
3 d?_ .. 250 -
!8
G
: 200 ;---- - 100,0
9 : -Howe< 50,0
ti SO -+-Energy f
0,0
ti
J....._J....._J...._..L..,_.,!_..J_..J j j .J...._J
0
600 BOO 1000 1200
0 5 10 15 20 25 :D 35 40 45 50 0 200 400
ty(Ph) Cycle number
e 21.6. Discharge profile of Figure 21.7. Specific energy and power of

energy cells LiNil-x-yCoxAlyOz/ Saft "VLE" cells during cycling at 80% DOD
ite at different continuous DST cycle. Every 50 cycles, energy measured
Ambient temperature. at 100% DOD - C/3, power measured at 80%
V= 4.0 V). DOD, 300A - 5mn pulse.
.3.4.1. Cycle Life
The automotive industry considers that an EV battery reaches its end

fe when it has lost 20% of its capacity/energy content or 20% of its
er capability. Figure 21.7 shows the performance of high energy cells r
cycling down to 80% DOD in a "DST" profile, which includes charge
eneration) and discharge peaks at various rates, corresponding to an
age rate of about C/3. For an electric vehicle capable of 150 km per each
e, this corresponds to 150000 km that is comparable to the life achieved
ost conventional vehicles.
Figure 21.7 also describes the energy and power delivered during
nostic 100% DOD cycles at C/3, made every 50 cycles. As can be seen,
Chapter 21
nergy remains remarkably stable during more than 1000 cycles, just like
power ability, measured with a 300 A pulse at 80% DOD. These results
trate the good behavior of the positive material, as well as the stability of
passivating layer on the carbon, preventing lithium loss. As a matter of
Figure 21.2 describing the lithium excess has been recorded after 1000
es, which demonstrates the excellent stability of the layer, which is
20
y due to the patented VC additive.
.3.4.2. Power capability
The nominal power of a Li-ion cell is not really an issue for EV

ications. As described in the next section, much higher power densities
be obtained for HEV applications, through appropriate design. More
ortant to the carmaker is how stable this power capability is, and how it
s with state of charge, temperature, cycling and aging.
Figure 21.8 shows typical specific power variations with SOC at
rent temperatures for the VLE Cell. This power, measured using pulses
50 A for 30s, does not represent the maximum power available.
ations with SOC are derived mainly from the cell voltage.
500 ,---------------r 250
::::.-:::!:
-
·--- ·-
:r= -==- ==--lfl '=ii=:ll_._=_ -=- :::..;_.-! 200

"":::.----- .:c
; g 300
350 t-------------i
150
3:
< f : 25o - lis

l!I 200
!;
150
...-Power,atoraga@40"C f-----------j
100 .l!
·
--
111t ..._... Power,atoraga @ 20"C [ !.
100 -+-Ene<gy,sto•ago@40'Cif--------t 50
5 0 E n- gy ,st o g o @ 2o ·ct=== ====1

0% 20% 40% 00% 00% 100% o +
sx
I+-20'C -e- -1CJ'C .....CJ'C . . ..al"C + - o
140"CI 0 200 400 600 800
Storage time, Days
Figure 21.9. High energy "VLE" cell power

e 21.8. EV cell power as a and energy retention during floating at 4.0 V,
on of SOC and temperature @ 20 or 40°C.
"VLE").
Passivation layer stability on the carbon is an important parameter to

re a good shelf life of the Li-ion batteries. The chemistry selected by
including the choice of electrolyte composition, graphite negative and
xCoyAlz02 positive, ensures an excellent calendar life. This important
erty is illustrated for example in Figure 21.9, showing the energy and
er retention of high-energy cells, stored at a constant float voltage of 4.0
t 40°C. After more than 2 years in these conditions, both energy and
er remain stable, similarly to what is observed on storage at 20°C.
From today's data, the life projection under average EV conditions is

6 years. Accumulating more data will probably extend this value. The
ent life expectancy of manganese-based batteries seems be limited to 3
years, due to the unsolved problem of partial dissolution of the positive
rial in the electrolyte.
.3.4.3. Safety
As discussed in Sections 21.2 and 21.3.1, the main safety concern for
-ion battery is the fact that many of the components are flammable,
ing with the electrolyte. Cell overheating may ultimately lead to battery
ion in the worst case, through active material decomposition. Therefore,
r intrinsic cell stability would come from the discovery of a non- or
flammable electrolyte, which is presently one of the most important
arch tasks for Li-ion throughout the world.
There are, however, means to prevent abusive situations at all levels,
cell to complete battery assembly. They usually include shutdown
rators internal to the cell, fuses to avoid heavy short circuits, and voltage
temperature monitoring to shut down power in case of overvoltage or
heating. Ultimately, no explosion should ever occur, thanks to the case
safety vent design. Considering the worst case, should an individual cell
e, the fire must be contained and/or extinguished, and not propagate to
whole battery. These are the considerations that are taken into account
attery design on a case-by-case basis.
Complete EV batteries have been subjected to abuse tests by
akers in the frame of joint development programs. For example, a
21
re crash test was applied to the Saft battery for EPIC, simulating an
dent where the battery bumps against a tree or a telegraph pole. The
ct speed of the battery was 52 kmlh. Sparks during crash resulted in a
as the electrolyte was burning. It was allowed to burn for about 15 min,
was extinguished without any problem.
The observations were as follows: there was no explosion, the fire was
dramatic and could have been easily extinguished in a short time, there
a very slow transmission to other modules. At the end of the experiment,
the modules of battery in the crash zone were destroyed. It must be
hasized that the crash with the battery system only was really a worst
. It is much worse than for a real crash with the same battery integrated
erly in a vehicle. From this behavior the carmaker considered that the
is acceptable and allows carrying on with the development of this
nology for EV application.
Chapter 21
.3.5. Onboard Vehicle Experimentation
At the time this chapter is being written, the Li-ion for EV is not a
mercially available product, and real life experience with EV Li-ion
ries is limited to fleet operations. More than one hundred vehicles using
on batteries have now been put on the road by various carmakers. These
ries are corning from small production lines set up by the battery
ers. Nissan initially experimented with Li-ion batteries using Sony
alt positive) batteries. After Sony stopped this battery activity, Shin
e (manganese positive) started supplying the batteries. Most european
manufacturers are experimenting with the Saft batteries. Some of these
cles have achieved more than 40000 km and the batteries are still in
operation. A 200-km range on a single charge has been demonstrated on
a small Peugeot 106 and a large Chrysler Voyager minivan. Among he
vehicles running with these Li-ion batteries, not a single problem
ed to the cells has been reported.
.3.6. Cost
This is certainly a key point to consider when speculating about the

ability of Li-ion becoming a very widely used system in EV and more
rally in all industrial applications. Presently, the prices of large Li-ion
batteries correspond to prototypes and small production volumes.
ond the basic product cost, they often include R&D activities. They are
much greater than the targeted volume prices, which are about
/kWh for high energy and $35/kW for high power types.
Cost reduction is a permanent subject in development programs. Li s
still a new technology and a very important cost decrease occurred in
ortable battery business during the last 5 years. The OEM selling price
22
rge quantities is presently in the order of $0.3/Wh and is expected to
er fall down with competition, leading in turn to significant reductions
e cost of large batteries.
The battery assembly cost may vary significantly with the level of
istication of the management system (thermal, electronics). The added
is generally assessed to be in the range of 15 to 20% of the cell costs.
The onboard EV experimentation also demonstrates that Li-ion
nology exhibits the lowest operating cost per kilometer among the four
competitive technologies. The reasons are:
• The energy efficiency (ratio between the discharged energy and the
charged energy) of Li-ion is better than 90% while the energy
efficiencies of aqueous batteries are in the 70-80% range. Therefore,
for the same available energy put into the battery, less energy will be
drained from the charging network.
• For the same onboard energy, a vehicle fitted with Li-ion will have a
longer range thanks to the lighter weight of the battery.
Comparing two identical vehicles, one fitted with aLi-ion battery, and
other with a lead acid battery of the same energy, the lead acid powered
cle consumes from the charging network twice the energy needed to the
n vehicle to achieve the same driving range. Assuming an acceptable
hasing price, and taking into account zero maintenance costs and long
ating life, Li-ion can be a very efficient technology based on life cycle
.
5. HEV APPLICATIONS
The hybrid car concept is based on combining two or more different
er sources in a vehicle, which would complement each other. The goal
nd this approach is making cars more environmentally friendly by
cing emissions as well as improving gas mileage.
Usually the primary power source in the vehicle provides main
ulsion and the second power source is only used when an additional
er boost is needed, for example during vehicle acceleration. The primary
ce has long been considered to be a smaller, more economical internal
bustion engine (ICE). In some more recent scenarios it is foreseen that
ICE could be replaced by a fuel cell. The second power source is a
argeable battery capable of delivering the high power pulses needed for
cle acceleration. In the early days of HEY, other technologies were
idered as this second power source. Supercapacitors are capable of
atedly delivering hundreds of thousands of high rate pulses even at low
eratures. However, the very low energy density of these devices,
lly in the range of 3 to 5 Wh/kg, renders them unsuitable for this
ication. Another energy storage device considered for supplying the
er for the acceleration burst was the flywheel. Although this technology
nteresting potentialities, carmakers were not convinced of its ability to
h maturity due to its complexity, and the issue of containment. As a
lt, the rechargeable battery remains the only viable option that can meet
tringent requirements of the application.
.1. Application Requirements

The battery needs to provide power on demand, normally manifested
hort bursts of a few seconds, accept fast charging from regenerative
ing, and last 15 years. Table 21.9 lists most of the important
23
irements for Power Assist HEY.
Chapter 21
21.9. Technical requirements of Power Assist HEV battery.
haracteristics
ulse discharge power
eak regenerative pulse power
otal available energy
Minimum roundtrip efficiency
old cranking power at -30°C
ycle life at 25Wh cycle energy
alendar life
Maximum weight/volume
perating voltage limits
Maximum allowable self-discharge
emperature range peration
urvival
5.2. Competing Battery Technologies
The competing battery systems for future hybrid electric vehicles are
-acid, Ni-MH and Li-ion technologies. Other technologies such as Ni-Zn
proven deficient in one or more critical performance aspects.
Lead-acid batteries have well established roots and traditions in the
motive industries. The main advantages are very low cost and proven
power capability. Drawbacks include low energy density, short life and
ations in charge acceptance at lower states-of-charge making this
em poorly suited for traditional HEV.
Ni-MH batteries became popular as a replacement for Ni-Cd. They
ide better energy density and their life, longer than that of Lead-acid
ries, is apparently sufficient to meet some of the life requirements in
application. The main disadvantages are their poor performance at
me temperatures and high self-discharge rates.
A number of different types of hybrid vehicles are being considered
he automotive industry. Several japanese carmakers have already
duced power-assist hybrid vehicles. These vehicles presently use NiMH
ries due to the more mature production status of this technology
pared to Li-ion.
.3. High Power Li-Ion Technology Challenges
From the very beginning Li-ion has promised and achieved great
ntage in energy density and specific energy over existing technologies.
ever, due to the use of organic electrolytes with fairly low ionic
ductivity, its possible use in high power applications was not originally
idered. Only in the last several years, R&D conducted for hybrid electric
cles has proven that Li-ion can successfully compete in the high power
argeable battery arena. The success of the development work has
mpted interest from other industries which could use a light but powerful
ce of energy. Why is this possible? The main power limiting factors in
on cells are indeed the low mobility of u+ in the organic electrolyte and
diffusion rate inside the active materials. The most practical way to
ease the power is therefore to increase the electrodes surface area, by
ing them thinner, which also reduces the ionic pathway in the electrolyte
arator and electrode porosity).
Li-ion technology utilizes electrodes that are already much thinner
electrodes used in other rechargeable chemistries. Commercial Li-ion
typically have negative and positive electrodes with thicknesses in the
e of 200 microns. This means that coating, calendering, slitting and
ding processes of thinner electrodes for high power cells must be
rolled much more tightly than traditional mass-produced Li-ion cells.
Another technical challenge is associated with the need to carry much
er currents. Therefore, engineering of the high power cells must address
nal electrode bussing as well as outside connections.
In order to meet the low temperature requirements, which commonly
e by default with most high power applications, electrolyte properties
t be specially tailored. High ionic conductivity, low freezing point and
viscosity enable low temperature discharge. This is accomplished
ugh selection of appropriate solvents and optimizing salt molar content.
ever, selecting solvents that would improve low temperature
ormance may sometimes result in compromised stability at elevated
perature and reduced calendar and cycle life.
24
.3.1. Perl'ormance of Saft Li-Ion HEV Battery Technology
The main feature of HEV batteries is the ability to deliver power on

and as well as to accept charge rapidly. Because of the voltage
ndence on state of charge, the ability to deliver discharge power down
minimum voltage limit decreases with decreasing state-of-charge. On
contrary, the ability to accept rapid charge to a maximum voltage limit
eases with decreasing state-of-charge. In addition the power delivered
a short pulse is greater than the power delivered over a long pulse or
a complete discharge. Therefore, when talking about power it is
ortant to specify both the state of charge and the duration of the
pulse.
Table 21.10 describes the main characteristics of the cells presently
lable.
Chapter 21
.3.1.1. Power
A key parameter for a high power battery system is the point where
battery can deliver as much discharge power as it can accept charge.
point is called the "sweet spot". Usually this point is around 50% SOC is
determined by conducting a series of discharge and charge pulse at rent
states-of-charge. Figure 21.10 shows the sweet spot for one of
s high power cells.
A different way to characterize the power capability of a system is to
ect it to a complete discharge at various rates. In addition to the voltage
it is important to make sure that the cell does not overheat during the
harge because overheating will severely reduce its life. Figure 21.11 s
a set of discharge curves under different discharge loads. Figure 21.12
s the cell temperature rise under 10C (6 min) discharge.
21.10. Main characteristics of Saft Li-ion high power cells.
Cells
Nominal Voltage
Rated Capacity at C/3
Power 50% DOD, ISs pulse
Dimensions (0 I h)
Typical Weight
Specific Energy
Specific Power
Energy Density
Power Density
--
4,--------------------------,
0
0
l ,/'2sa: regen
>
3,6
3,8
3,4
1i: 3,2
f 3
0
/ .......... 3 ---- -z
0
/ "18se:: clscharge g
'iji 2,8 ----------------------'k\--\-1
l r
0 u 2,6 -----------------------\\--1-i
0 2.4 -f-.-:ce'"lNl "'o=m>=ina"c'l -=a=-=pac:;;citycc-.•7."'5A'"h--------------l
2,2 -----------------·
0
2 -r y
0
0% 20% 40% 60% 60'% 100% 0 2 4 6 8
Depthofrlscharge Capacity, Ah
e 21.10. Sweet spot of a high Figure 21.11. Complete discharge at

cell. different discharge rates.
Low temperature performance of high power cells is also critical.

current capability at low temperatures can be achieved through the use
pecial electrolytes, optimized positive and negative active materials, and

oved cell echanical design. Figure 21.13 shows a sequence of three 2-
nd pulses separated by 10-second rest periods, called "cold cranking",
onducted at -40°C.
VU.TAGE "
-- t
80 4,5 ,............................................................................................................................................,
CURRENT
70
35
'
50 !.: .p.....i\1------.>....J. ------'...........;,
80 -- ' f I f I i
40 291Wikg 293W/1<g c2=98.:..:W.:..c/1<::!.g-i
2
30 1,5 t---------------i
TEWERATURE
20 1i-- PNGV Cold Cranking Teat --;
10 0,5 _Three 40Apulo ondo al·40oC
Cal Nominal Copocly.7.5Ah
0 o+-
0 2 4 8 0 2 4 8 8 10 12 14 18 18 20 22 24 26 28
capacily, Ah Time,aec
e 21.12. Temperature rise in a high

r cell under IOC (6 min) load. Figure 21.13. Cold cranking pulses of a
high power cell at -40°C.
.3.1.2. Cycle life
Cycle life of Li-ion like any other rechargeable system varies with
h of discharge, cycle rate and temperature. Figure 21.14 shows that Li s
capable of maintaining power and energy over 1 million shallow high
cycles, representing the HEY utilization profile. Figure 21.15 displays
eep cycles recorded at C/1 rate. 5000 cycles are projected at an average
ient temperature of 20°C. Except for NiH2, which has much lower
gy and power density (see Section 21.7.1.), this performance is
atched by any other battery system. An evaluation of cycle life
ndence on temperature still remains to be performed.
18
--RoomTeffi!>iratiire
--
16
14 ;!. W'k ........... .
12 ln'/o :"""
E kr------ --
10 ln'/o
= --
8 00'/o
6
·-«1'kT---------------
4
i
2 10'k ·----
0 O'k
0 250 000 500000 750 000 1 000 000
o &XJ 1000 1&XJ am 2m 3lD
Cycles
(),des
e 21.14. Deep cycling of a high
cell under HEY load profile Figure 21.15. Cycle life of a high power cells
dic 100% DOD diagnostic). (Crate, 100% DOD).
Chapter 21
.3.1.3. Calendar Life
Last but not least is the calendar life of the system. It has been
ected that Saft HEY cells could maintain the required peak power and
gy necessary for the HEY application over a period of 15 years. These
ates are based on actual data of cells stored at various temperatures and
projections based on the Arrhenius law. It has to be mentioned that the
re of the application, which requires the battery to be at around 50%
-of-charge, is very beneficial for achieving long calendar life in Li-ion.
re 21.16 shows actual capacity and power fading from cells on storage
ree different temperatures.
.3.1.4. Abuse Behavior
Li-ion electrochemistry does not have metallic lithium, which makes it

fer system. However, it is still a high energy density system, and
%
I . ._.
120%,--------------,
A 2s•c
% 2 •c t -- - !:'! ·c
-.o;,c; 1f!.100%
t . ""' "- /-t>---.-...---..,,_._
% eo% t------;;:;=:;=Y.::;:=::::;::=:::::::: ::; :; ;--oo_·cl
]Average Gapaclty flleaau&d at 2s•c1 ]Avaraga Paak Power Maasurod at 2s•q
_J Cella stored at 50%SOC l ': 60% --l Cella stored at 50%SOC _
% 40% t------------, =s-1

0.. 20% +------------1
1: 1
0 5 10 15 20 0 10 15 20
Months Months
e 21.16. Capacity and power evolution of cells stored at 50% SOC at various
ratures.
ains flammable electrolytes. Under some severely abusive conditions

as overcharge, smoke and flame may be observed, as discussed in the
ral section. Because they contain relatively less active materials, and
inactive components (such as current collectors), high power cells are
rally safer than high-energy designs.
Saft high power cells have an integrated fusible link preventing
mal runaway during external short circuit (the 8 Ah cell produces a peak
-circuit current of 1000 A). At the battery level, redundant protections
n place: by monitoring voltage, current and temperature, dangerous
tions resulting from overcharge or overheating can be prevented. Table
1 describes the results of various abuse conditions.
.3.2. Battery Configurations
Many battery configurations can be designed, depending on the

ired characteristics. Electronic systems for battery monitoring obey the
e basic rules valid for high energy batteries (see previous section).
Active cooling is frequently required for these high power systems.
cooling is sufficient for most applications, but liquid cooling might
etimes be preferred. Figure 21.17 shows aLi-ion battery for a Power
st HEV. The battery is 2.8 kWh, 8 Ah at 346 V, providing 50 kW for 18
nds at 50% SOC. This particular battery is air-cooled and has integrated
ery Management System (BMS).
.3.3. Other Li-Ion Battery Systems
As discussed in previous sections, LiMn 204 is the other Li-ion

mistry proposed for HEV applications. Shin Kobe manufactures a high
er cell for HEV. The small cylindrical 3.6 Ah cell demonstrates high
21.11. Abuse test records on high power Li-ion cells.
Test descriPtion
rnal short VL8P
VL16P
Test (ramp to 150°C in 30

Event after >45min) P and VL16P
h VLSP
VL16P
ct Droo Test
rate overcharge
A)
discharge (reversal)
water immersion (8h)
fire (radiant heat 800°C)
25
er discharge capability, as shown in Figure 21.18. A peak power of
W/kg can be obtained at 50% SOC, 20°C.26•27
Japan Storage Battery has also developed a high power design
atteries using LiMn 204. The 3 Ah cell exhibits 1500 W/kg on
Chapter 21
e 21.17. Example of 50 kW high power battery for power assist HEY (2.8 kWh, 8 Ah at
).
ge and discharge at the 40% state of charge "sweet spot". Good deep
D cycle life is obtained at lC rate at 25°C, while some capacity loss is
ved at 60°C (Figure 21.19), as is typical of the Mn-based cells.
Using a substituted manganese spinel, NGK insulators 19 developed
ntly a high power 4 Ah cylindrical cell of enhanced safety, exhibiting a
ific power of 2500 W/kg at ambient temperature and 800 W/kg at -25°C.
edium size 12 Ah cell has been also designed for a 42 V battery.
45'1::
:--.
- 25'1::
60'1::
Charge :lCA ,4.15V(CC/CV)X3h

Discharge: 1CA to 2.75V
5 10 15 0
0 200 400 600 800 1000 1200
discharge energy (Wh) Number of Cycles
e 21.18. High rate discharge of Figure 21.19. Available capacity on

204 Li-ion 3.6 Ah high power cell cycling at different temperature of
Kobe, from Ref. 25). LiMn204 Li-ion 3 Ah high power cell
(from JSB catalogue).
Maintaining the power capability over a long period of time, including

high ambient temperature excursions during normal operation, is the
challenge for the technology in this application.
6. STATIONARY APPLICATIONS
Stationary battery applications can be categorized into broad subsets
d on the way the battery is charged and discharged.
Standby (float) operation. The battery is maintained in a fully
ged condition by a so-called 'float' or constant potential charger. The
ery and the system loads are connected in parallel with the charger
inals. In this mode of operation, the charger supports continuous loads
the battery accepts whatever current is necessary to maintain it at a set
age. If the charger output is lost, the battery immediately assumes the
em loads.
Intermittent charging (cycling) operation. Batteries in these
ications are charged only when the primary energy source is available.
primary energy may come from a solar (photovoltaic, or PV) panel, a
d turbine, an engine-driven generator, the utility grid, or some other
ce. When a load is connected and the primary source is unavailable or is
equate to support the load, the battery is discharged.
Buffer operation. Halfway between the other application groups, the
ary energy source in these applications is generally available but is
uently insufficient to supply the imposed loads. This may be because the
s exceed the rated output of the energy source, or because the rate of
ge in the loads exceeds the capability of the energy source to respond.
n the load level falls below the rated output of the primary source the
ry is recharged.
Each of these application sets imposes its own particular requirements
Li-ion batteries. The following sections describe the applications and
requirements in greater detail, and discuss the advantages and
dvantages of Li-ion batteries in meeting those requirements.
6.1. Standby (Float) Operation

Most applications in this subset incorporate a grid-connected
ger/rectifier that is generally available and maintains the battery in a
state of charge. Examples of standby applications include:
• Telecommunications network switching
• Uninterruptible power supply (UPS) systems
• Power quality systems
• Safety/security systems, e.g., emergency lighting, fire safety
systems, intruder alarms
• Utility switching (transformer protection)
• Turbine control, e.g., lube oil pumps
• Railway signaling
• Generator starting
Chapter 21
The battery duty in these applications can be quite varied. Some duties
ire the battery to support a relatively constant load for a duration of
ral hours, such as telecommunications network switching or railway
aling. Others demand a short, high-rate discharge, such as in UPS
ems or generator starting. A few applications feature a mixture of low l
continuous loads and short high-rate loads, such as utility switching
ems.
Batteries may be located in controlled temperature environments, or
be subjected to extreme temperatures in outdoor cabinets. Battery
gns that minimize high temperature aging and maximize low
perature performance are obviously beneficial in the latter case. The high
metric energy density of Li-ion can be an important factor in some
ications, particularly in cabinetized systems, but may be unimportant in
rs.
0 Although Li-ion batteries are more
D
=
......
... energy efficient than most other
b.
types, with better round-trip
·-...-...
D
m
charge/discharge efficiency and
+3/N 4I'C
-&3/NIIfC lower power consumption on float,
D
+3/Nli!J'C this feature is relatively unimportant
+3/N41'C
D -&-3/NfDC in standby operation. More important
m is that the battery operates reliably
0 :m 40 !IX) 8ll 1<DO over a long lifetime. A general good
FkD1irrn,d¥;
behavior of Li-ion batteries on
e 21.20. Residual capacity evolution of continuous float charging has been
high-energy Li-ion cells under float demonstrated.28 '29
e at various temperatures and voltages.
As an example, Figure 21.20 describes the residual capacity evolution,

ng 823 days at various temperatures and voltages, of Saft high-energy
ules (3 cells in parallel) measured periodically at the floating
perature and at C/8 rate. The capacity retention is excellent, allowing life
ictions of more than 15 years in service. The known failure mode of Li
a slow increase of impedance and consequent energy decline, is a
fit in these applications since it makes battery health monitoring and life
iction easy. Full discharges may be quite rare, so a battery that requires
odic discharge tests or experiences sudden unexpected failure can be a
r liability.
6.2. Intermittent Charging (Cycling) Applications
Stationary applications involving intermittent charging include many

he loads listed in the standby category. These applications are better
gorized by the energy source, which can include:
• Renewables (wind, PV, etc.)
• Engine-driven generators (cyclic operation)
• Hybrid systems
As with standby applications, the battery may support longer energy
loads, shorter power-type loads, or a mixture of the two. Li-ion batteries
particularly well suited for applications involving repeated discharges.
s is not simply an issue of the cycle life of the technology, but also
lves considerations of chargeability. With virtually no side reactions to
pete with the main charging reaction, the process is highly efficient-the
d trip energy efficiency typically exceeds 95%, compared with 60-75%
most other technologies. While this can be important from an economic
t of view, particularly when the total available energy is limited, there is
ven more important practical consideration.
The high efficiency and the voltage dependence of state of charge help
revent severe mismatches in the state of charge of the positive and
ative plates that can occur in aqueous battery systems. Such imbalances
lt from differences in charge efficiency between the plates and typically
sen as a battery is subjected to multiple discharges with partial recharges.
only solution in these cases is to subject the battery to a prolonged
shing' or 'equalizing' charge. Not only does this take the battery out of
ice until the charge can be completed, but in an off-grid system can
lt in prolonged and inefficient operation of generator units. Li-ion
eries are much more resistant to such imbalances and can be charged
kly and efficiently to a known state of charge.
6.3. Buffer Operation
This type of operation is expected to grow in significance in coming

s as a new class of applications emerges. This is in the area of
ibuted generation, in which relatively small generators are sited close to
loads they supply. Distributed generators include microturbines and ll
fuel cell units. The latter are expected to be implemented as 'micro
P' (combined heat and power) systems in individual residences,
ucing both electrical power and heat for hot water or heat exchange
ems. Both generator types are characterized by relatively slow response
oad changes. Batteries can improve the load response for these units
e also providing power for black starting.
Chapter 21
It is worth noting that this type of buffer operation may also require a
ry to absorb power as generator output is ramped down after a load has
abruptly terminated. It is likely that the most efficient way to
mmodate this requirement will be to operate the battery in a less than
charged state. This can be easily achieved with Li-ion batteries, with
relationship between open circuit voltage and state of charge.
.4. Stationary Battery Trials

As the industry moves towards the commercialization of large size Li
atteries, trial sites are being set up to demonstrate the capabilities of this
nology. These include high-energy batteries in telecommunications
ide plant cabinets and high-power batteries for power quality and
ibuted generation support. Such demonstrations are critical to gaining
ptance of Li-ion technology in diverse applications, ready to take
ntage of declining price levels as the batteries move into commercial
uction.
Japanese LIBES R&D program, sponsored by METI and NEDO
18 30
essed the home load leveling application. As for example, Sanyo
gned a 2 kWh-class module (117 Wh/kg, 177 Wh/1) using a 250 Wh cell,
a nickel-based positive, LiNi 0.7Co0.302, and graphite/coke mix as a
tive. This cylindrical cell is described in Table 21.12.
21.12. Main characteristics of Sanyo Li-ion cells (from Ref. 18).
Capacity
Dimensions
Weight
Voltage
Figure 21.21 shows a high-energy prototype Saft battery for

ommunications field trials. This battery is rated 70 Ah at 48 V. For
er power applications, Figure 21.22 shows a 14.5 kWh battery assembly
ble of providing 100 kW for 1 minute.
. SPACE APPLICATIONS
Since the start of the space era, rechargeable batteries were used to
ide power to satellites during eclipse periods. In fact, solar panels are
to power the satellite payload (and recharge the batteries) when the sun
inates them. Batteries take over from the solar panels
Lithium-Ion Batteries for HEV/EV and Other Industrial Applications
675
e 21.2I. Prototype of a Li-ion Figure 21.22. High power lithium ion
ry for Telecoms. battery assembly.
n the satellite enters into the earth's shadow. NiCd in the 60's, NiH 2 in
90's and Li-ion now are the only three technologies for satellite use.
The telecommunication satellites that are presently in geosynchronous t
(GEO), use nickel-hydrogen batteries. The size of telecommunication
llites, and, as a consequence, the electrical power demand during solar
pses, is continuously increasing. In the 80's satellites weighing 0.5 ton,
ered with nickel-cadmium batteries providing 0.5 kW were the standard.
e 90's, the prevailing design was 1 to 3 tons and 2 to 8 kW provided by
el-hydrogen batteries. Now, satellites may weigh 5 tons with power
and in the range of 15 to 30 kW. With the current nickel-hydrogen
nology, 30 kW of power leads to a battery weight of at least 800 kg
esponding to 15-20% of the total satellite weight! That means that the
ry, which was already one of the heaviest pieces of equipment on board
atellite, is now reaching a critical size. Combined with the current price
0-100 k$ to place one kg in orbit, battery weight has become one of the
n issues facing competitive telecommunication satellites.
Thanks to the Stentor project, started in '96, Saft has qualified Li-ion
nology for the first time as the main power source of a satellite. Stentor
e first GEO satellite that will fly this promising new battery technology
pace.
7.1. Battery Requirements for Satellites
Considering the battery constraints, two main satellite families can be

idered.
.1.1. GEO Satellites (Geostationary Earth Orbit)

Chapter 21
Satellites in geosynchronous orbit are positioned over the equator and

el in the same direction as the earth rotates. They appear "fixed" with
ect to a given spot on earth. Satellites in GEO orbit are 35786 km above
earth in the plane of equator. These satellites are mainly used for
ommunications (direct home TV broadcasting, voice and internet).
eorological satellites are also positioned in geostationary orbits
eosat family). The main parameters for GEO are an orbit duration of 24
s, and a number of 90 eclipses per year. As the GEO satellite position is
rom the earth, and due to the inclination of the earth's axis to the ecliptic
e, the satellite goes through the earth's shadow cone only once a day
ng the equinox periods (2 times 45 days per year). Eclipse duration is not
tant but increases during the first 22 days up to 72 minutes and then
eases during the last 22 days of the eclipse period. The expected lifetime
years, and the satellite power ranges from 5 to 25 kW.
.1.2. LEO Satellites (Low Earth Orbit)
LEOs are either elliptical or (more usually) circular orbits at a height

400 to 2000 km above the surface of the earth. The orbit period at these
udes varies between 1.5 and 2 hours. As the altitude of the LEO satellites
ow, their rotation speed is very high (>25000 km per hour). The
imum time during which a satellite in LEO orbit is above the local
zon for an observer on the earth is up to 20 minutes. A global
munications system using this type of orbit (mainly due to the low
agation delay of 20-30ms) requires a large number of satellites in a
ber of different, inclined orbits. Most of the LEO satellites (except the
tellations which are used for direct telecommunication systems: Iridium
Globalstar) are used for Earth or Space observation. The best examples
EO satellites are the Hubble Space Telescope, the Spot family (Earth
ing and survey) and military satellites.
The key parameters are an orbit duration from 1.5 to 2 hours, and a
ber of 20 to 35 minutes eclipses per year from 4500 to 5500 depending
he satellite altitude. The expected life time is 5 to 8 years, and the
lite power vary from 10 to 3000 W. Two main LEO satellite families
be categorized: the "small" LEO (sometimes called nano- or micro
lites) and normal LEO
.2. Competing Technologies
Table 21.13 compares the three main electrochemical systems for

lites: NiCd, NiH2 and Li-ion.
The main advantage offered by Li-ion technology is weight reduction.
e can see in the table, weight benefits are achievable both at the battery
satellite level, because the lower thermal power and higher energy
e 21.13. Performance comparison of NiCd, NiH2 and Li-ion batteries for satellites.
Energy density (Whlkg)

Energy efficiency %
Thermal power (scale
1-10)
Self discharge/day
Energy gauge
Charge management
Modularity
iency of Li-ion compared to NiH2 impacts on solar panel and radiators

s. Moreover, the lower self-discharge allows the launch operations to be
lified, thus avoiding battery recharge when the satellite is installed on
auncher. NiH2 battery self discharge (10 times higher than for Li-ion)
ires recharge of the battery right up to the final countdown. This
arging operation performed under the fuse cap is a critical phase of the
ch mainly because of the thermal management of the battery. In orbit,
efore, management can be simplified as the necessary reconditioning
ation of NiH2 can be eliminated by using Li-ion. As discussed in the
r sections, the direct relationship between voltage and state of charge
ces here again the major benefits of a precise "fuel gauge", and an easy
ctly assembly of the cells in parallel.
These numerous advantages of Li-ion compared to the two current
el technologies lead to an increasing number of satellite projects
ting Li-ion batteries as base-line power source. The only drawback of
Li-ion is the need for more sophisticated electronic systems for battery
agement.
7.3. Saft Cell and Battery Designs

.3.1. Cell Designs
Started at Saft in '95, the development of Li-ion cells for space took
dvantage of the manufacturing line originally built for electric vehicles.
VES140 (140 Wh and 40 Ah) space cell has the same electrochemical
k (electrode assembly) as for the EV cell. The vital parts of the VES 140
are (Figure 21.23): the terminal, which must insure an extremely high
of hermeticity for the mission duration, and the support of the stack in
ontainer, to withstand the high vibration levels during the launch phase.
Chapter 21
The electrodes are made of substituted nickel-based oxide for the

tive and mix of graphite for the negative. The specific properties of this
trochemistry, as described in previous sections, is particularly suitable
satellite needs: very long cycle and calendar life, enhanced by the
ium reserve" of about 15%. The cell shape is cylindrical (250 mm height
54 mm diameter) with a nominal weight of 1116 kg. Its specific energy
25 Wh/kg. This number may appear low compared to other types of Li
cells. The difference mainly comes from the specific design for terminals
margins to insure higher reliability.
This cell design has been qualified either for GEO and LEO
ications in January 2000 and has flown in 2002 on board Stentor. Two r
cell designs are in the qualification process: VES100 (100 Wh and 26
which is similar to the VES140 but shorter (190 mm height) mainly
cated to LEO missions, and MPS15S (15 Wh, 4.5 Ah) which has an
tical prismatic shape and is mainly dedicated to "small" LEO satellites
low power (Figure 21.24).
e 21.23. The VES140 cell. Figure 21.24. The MPS15S cell.
.3.2. Battery Designs
STENTOR is a french technological satellite designed by the CNES

tre National d'Etudes Spatiales) and manufactured by Alcatel Space
stries and Astrium. Its mission is to qualify new satellite technologies
r panels, deployable radiators, electric propulsion, batteries, etc.) under
al conditions. The emerging Li-ion system was proposed in 1995 by
to be included as the new power source, and considered as a strategic
nology. The project objective was to qualify and fly a complete Li-ion
ry design including charge management electronics (Figure 21.25). As
attery power is close to 3 kW, the battery is composed of 2 battery sets
P11S (2 cells mounted in parallel with 11 pairs connected in series). The
tor battery meets all requirements (life time, thermal and mechanical
traints, reliability) of the generic plate-forms Space Bus (Alcatel Space

stries) and Eurostar (Astrium). Stentor is the first GEO satellite to
>
8.
r--
>
·a 1000 1!m
e 2I.25. Stentor battery set. Figure 2I.26. 56 GEO accelerated seasons on

VES140 (82% DOD max).
riment with Li-ion batteries.

From this experience, a range of modules has been designed to cover
lite manufacturers requirements for the near future (battery voltage,
print, and power), suitable for GEO or LEO.
The module design from 3 to 12 parallel cells allows coverage of the
e of satellite power from 5 to 30 kW. Assembling these modules in
s offers the best modularity and simplification of the management
em compared to the string of series cells in parallel design. In each
ule, a by-pass system and an electronic device interface with the satellite
pment. The main function of this electronic interface is to perform the
ncing of the cells to guarantee the battery energy throughout its lifetime.
GEO batteries, delivering up to 108 Wh/kg, are generally 10 to 12
ules in series ("50 volts" bus), or 20 to 24 modules ("100 volts" bus).
satellites require less energy, and bus average voltage is 28 V
esponding to a 9S (9 modules in series).
.3.3. Typical Performances and Life Expectation
Numerous life tests have been performed by ESA, CNES and satellite
ufacturers to measure fading and calendar losses on Saft cells. For GEO,
e than 25 life tests (accelerated or real time) have been conducted (some
still ongoing) with various DOD (from 50 to 90%), various end of
ge voltages (from 3.9 to 4.1V), and with or without electric propulsion
es.3 1.32
More than 56 GEO seasons (corresponding to more than 26 years)
been completed on a 2P3S battery. The test conditions were:
Chapter 21
- accelerated charge with currents from C/3 to C/6 (instead of C/10)

- discharge profile and currents identical to GEO mission, DOD
ranging from 80 to 82%, at 20 °C.
- balancing system used before each season.
Figure 21.26 shows the end-of-discharge voltage during the 56 GEO
ons. Capacity was checked every season, and the energy loss at season
corresponding to the standard life time of GEO satellites (10 years), is
than 2%.
In addition, calendar loss is low, thanks to the system properties
dy described. Considering an average temperature of 20°C over 15
s (25°C during the equinox period to enhance energy and 15°C during
ice as the energy need is low), the calculated energy loss due to the
ge is linked to the increase of the internal resistance and is less than 3%.
the total energy loss for the 3 years of storage and 15 years in a GEO
ion at 80% DOD is less than 5%. Considering an additional margin of
the end-of-life (EOL) energy is 90% of the beginning-of-life (BOL)
gy. VES140 cell LEO life tests are currently at more than 20000 cycles
ivalent to 4 years) at 20% DOD. The energy loss is less than 12%.
sidering 3 years of storage and 8 years in a LEO mission at 20% DOD,
OL energy represents 70% of the BOL energy.
.4. Other Battery Designs

A few other battery manufacturers are addressing the space market
Li-ion batteries, using various designs and electrochemistries.
.4.1. GS- Melco
Melco (Mitsubishi Electronic Corp) has developed the large capacity

n battery shown in Figure 21.27 as an example of the families. The
tic cylindrical 100 Ah cell specific for space has been jointly developed
apan Storage Battery Co, Ltd (JSB) and Melco. The electrochemistry is
d on LiCo02, with other cell components similar to normal Li-ion types.
battery design includes paralleled cells mounted in series (P-S
itecture), one by-pass system per package and electronics for balancing.
Figure 21.27 describes a 100 Ah battery module, using 10 cells in
s. Rated at 3.6 kWh, this 35.9 kg module exhibits 100.3 Whlkg.33 The
m-/on Batteries for HEVIEV and Other Industrial Applications 681
use of a larger cell size increases

the battery specific energy and
reduces the labor cost of
manufacturing and tests. The wide
range of cell capacities (50 to 190
Ah) will support the versatility to
optimize battery size to the
required power. Calendar and
cycle life testing of these cells has
e 21.27. Melco lOOAhxlOS battery been evaluated,34 estimating a
le.
capacity retention over a 15-year GEO nusston + 3 years ground

ge, and 59% retention over a 8-year LEO mission + 3 years ground
ge.
.4.2. AEA Technology pic/COM DEV Ltd
AEA Technology/COM DEV has made the opposite choice using

ially selected small commercial cylindrical 1.5 Ah cells (18650HC
ufactured by Sony).35 LiCo02 is used as the positive material and hard
on as the negative. Cells are connected in series strings to match the
ired voltage and these strings are assembled in parallel to achieve the
ired capacity. No electronics are required at the cell level, provided that
cell properties are uniform and well matched. This cell, originally
gned for multipurpose use, includes a safety current breaker to prevent
vidual cell overcharge. There is no balancing, considering that the failure
ne cell within a string should simply lead to the loss of one string. The
ber of strings is sized taking into account the probability of failures, to
re the required battery life. Thanks to the small capacity increments, this
iguration fits well the various needs for small spacecrafts and
Technology/COM DEV batteries have been selected for a number of
osat scientific projects.
Chapter 21
For example, a 3x(6s-llp)

1.07 kWh construction is
designed for the ESA Rosetta
platform, launched at the
beginning of 2003 towards comet
Wirtanen. Extension of this
concept to larger battery
configurations is being tested, in
order to demonstrate adequate
igure 21.28. Rosetta Lander battery
one of the two modules) (courtesy of
EA Technology pic).
bility. Figure 21.28 shows half of the 151 Wh battery qualified for the
S Rosetta Lander project.
.4.3. Yardney Products/Lithion, Inc.
The Li-ion cell and battery design approach adopted by Yardney

nical Products/Lithion, Inc. utilizes a true parallel-plate prismatic
gn, and includes a mixed cobalt-nickel oxide as positive material (Figure
9). This design yields a highly gravimetrically and volumetrically
ient cell. High packing efficiencies are achievable with prismatic cells,
battery construction is simplified by constraining all cells in a single
k.
CydtNumba
e 21.29. 35 Ah Li-ion Figure 21.30. End of discharge voltage during GEO
ourtesy of Yardney).
cycling of 35 Ah cells (60% DOD max) (courtesy of
Yardney).
um-Ion Batteries for HEVIEV and Other Industrial Applications 683
The 30 Ah cells developed for the 2001 Mars Lander Mission

onstrated over 144 Wh/kg specific energy and 345 Wh/1 energy density.
8-cell (28 V) battery constructed for this mission delivered 111 Wh/kg
238 Wh/1. Life cycling of these cells have demonstrated over 12000
-type cycles to 40% DOD and over 5 years of an accelerated GEO-type
ile to 60% DOD, as shown in Figure 21.30. Upper curves are for charges
1 V, lower curves to 3.9 V.
7.5. Launchers
Launchers also need batteries, mainly based today on NiCd or
/Zn technology. Here again, energy and power density of Li-ion are
ng incentives. A new 1.26 kWh battery has been recently developed by
and is shown in Figure 21.31. A particular feature of this design is to
stand very drastic vibration conditions.
Weight= 18 kg
Capacity =35 Ah
Voltage = 36 V
Operating temperature: -20 to +30°C
Discharge profile is 600 Amp 100 ms +
340 Amp for 5 minutes
e 2I.3I. Saft Li-ion battery for a

her.
Chapter 21
8. CONCLUSIONS
Li-ion electrochemistry has now demonstrated its ability to be used in
e batteries for a very wide range of applications. The main features of
e power sources are high energy or high power, extended cycling ability,
calendar life, and high reliability. Continuous technical improvements
expected, particularly in the field of safety, to enlarge the domain of
zation. Their superiority has already been proven in various applications
ugh extended prototype testing such as EVs. The space industry, where
n brings very substantial benefits, has already adopted and qualified this
nology, which will progressively replace the existing NiH2 batteries. e
is no doubt that this system will very soon find its place in the
strial battery market. Opening the road to wide market segments, such as
motive where HEVs or new 42 V batteries are the biggest stakes, will
tly depend on cost reduction. Material costs and associated electronics
esent the major part of the present costs, and should significantly
ease with increasing production level. Ultimately, Li-ion large batteries
expected to be competitive with the today's common systems on the
of life cycle cost.
KNOWLEDGEMENT
This chapter has been written with the collaboration of Yannick

homieu, Guy Chagnon, Jim Me Dowall, Louis D'Ussel, Kamen Nechev
Guy Sarre (all from Saft).
The data on products presented here cannot be considered as a warranty from the
facturers.
FERENCES
Mizushima, P.C.Jones, P.J. Wiseman and J.B. Goodenough, Mater. Res. Bull. 17 (1980)
.
lichta, M.Salomon, S.Slane and M.Uchiyama, J. Power Sources 21 (1987) 25.
Broussely, F.Perton, J.Labat, R.J.Staniewicz and A.Romero, J. Power Sources 43-44
993) 65.
Yoshino, Asahi Kasei, JP 1989 293 (priority 1985), US Patent 4,668,595 (1987).
ecerf, M.Broussely, and J.P.Gabano, EU Patent 0345 707, US Patent 4,980,080 (1989).
Delmas, J.P. Peres, A. Rougier, A. Demourgues, F. Weill, A. Chadwick, M. Broussely,
erton, Ph. Biensan, and P. Willmann, J. Power Sources 54 (1995) 329.
arre, M.Broussely and R.J.Staniewicz, Proceedings of EVS12, Anaheim, USA (1994).
Broussely, G.Rigobert, J.P.Planchat and G.Sarre, Proceedings ofEVSJ3, Osaka, Japan,

7-32, (1996).
Delmas and I. Saadoune, Solid State Ionics 53-56 (1992) 370.
.Funahashi, Y.Kida, K.Yanagida, T.Nohma and I.Yonezu, ].Power Sources 104
002) 248.
h. Biensan, J.-P. Peres, and F. Perton, ECS Fall Meeting, Abstract 131, Honolulu, Hawai"
999).
. Ohzuku, T. Yanagawa and M. Kouguchi, J. Electrochem. Soc 142 (1995) 4033.
P.Peres, C.Audry, Ph.Biensan, G.Blanc, J.P.Boeuve, V.Dillay, M.Broussely and A.de
uibert, Lithium Battery Discussion, Proceedings, Abstract 32, Arcachon, France (2001).
.H.Chen, J.Liu and K.Arnine, Abstract 176, ECS Fall Meeting, San Francisco (2001).
. Broussely, S.Herreyre, Ph.Biensan, P. Kasztejna, K.Nechev and R.J.Staniewicz, J.
ower Sources 97-98 (2001) 13.
. Broussely, S.Herreyre, F.Bonhomme, Ph.Biensan, Ph.Blanchard, K.Nechev and
.Chagnon, ECS Fall Meeting, San Francisco (2001).
.Nakai, T.Aiba, K.Hironaka, T.Matsumura, and T.Horiba, 41" Battery Symposium in
apan, Nagoya, p. 350 (2000).
.Terada, T.Yanagi, S.Arai, M.Yoshikawa, K.Ohta, N.Nakajima, A.Yanai and N.Arai,
.Kitoh and H.Nemoto, J. Power Sources 81-82 (1999) 887.
.Simon, US Patent 5,626,981, EU Patent 0683537.
.Bitsche, G.Gutmann, A.Schmloz and L.d'Ussel, Proceedings of Jtfh ElectricVehicle
ymposium (EVSJ8), Berlin, Germany (October 2001).
.Takeshita, J9h International Seminar on Primary and Secondary Batteries, Fort
auderdale, USA (2002).
NGV Battery Test Manual, Rev. 3 (February 2001).
. Nechev, M. Saft, G. Chagnon and A. Romero, 2"d AABC, Las Vegas, USA (2002).
Horiba, K.Hironaka, T.Matsumura, T.Kai, M.Koseki, and Y.Muranaka,JJ'h IMLB,
onterey, USA, Abstract 418 (2002).
Horiba, J" AABC, session 5, Las Vegas, USA (2001).
Horiba, K.Hironaka, T.Matsumura, T.Kai, M.Koseki, and Y.Muranaka, J. Power
ources 97-98 (2001) 719.
. Broussely, M. Perelle, J. McDowall, G. Sarre and J. Martaeng, Proceedings of22"d
t. Telecommunication Energy Conference, Phoenix, USA, paper 12.1 (lntelec 2000).
Sasaki, M.lnaba, T.Abe, Z.Ogumi, ECS Fall Meeting, Abstract 271, San Francisco,
SA (2001).
.Yanagida, A. Yanai, A.Funahashi, K.Ohkita, T. Nahma and I.Yonezu, 4rJh Battery
mposium in Japan, Kyoto, p.l (1999).
.Borthornieu, M.Broussely and J.P.Planchat, Proceedings (jh European Space Power
onference, Porto, Portugal, p. 679 (May 2002).
.Dudley, J.P.Planchat and P.Willmann, Proceedings (jh European Space Power
onference, Porto, Portugal, p. 673 (May 2002).
Gonai, T.Okamura and M.Goto, Proceedings J9h Int. Communications Satellite
stems Conference, Toulouse, France, p 147 (April2001).
lnoue, N. Imamura, H.Yoshida, M.Mizutani and M.Goto, Proceedings J9h Int.
ommunications Satellite Systems Conference, Toulouse, France, p 186 (April2001).
.Spurrett, C.Thwaite, M.Slimm and D.Lizius, Proceedings (jh European Space Power
nference, Porto, Portugal, p. 476 (May 2002).
ter 22
LITHIUM BATTERIES FOR MEDICAL

APPLICATIONS
1 1
E. S. Takeuchi\ R. A. Leising , D. M. Spillman ,
2 2
R. Rubino\ H. Gan\ K. J. Takeuchi and A. C. Marschilok
ilson Greatbatch Technologies, 10,000 Wehrle Drive, Clarence, NY 14031, U.S.A.

2
Department of Chemistry, University at Buffalo, Buffalo, NY 14260, U.S.A.
INTRODUCTION
This chapter on lithium batteries for medical applications is not meant

an exhaustive review, but rather a broad overview of some of the
nt types of lithium batteries that power implantable medical devices.
ttery systems described in this chapter fall into two major categories, y
or single use cells containing lithium metal anodes, and secondary or
geable systems utilizing lithium ion chemistry. Primary lithium
es have been used for implantable devices such as cardiac pacemakers,
umps, neurostimulators and cardiac defibrillators. Secondary lithium
teries have been used with left ventricular assist devices, total artificial
and implantable hearing assist devices.
The first human implant of a lithium battery, a lithium/iodine cell that
d an implantable cardiac pacemaker, was conducted thirty years ago. 1
hat time several different lithium anode batteries have been developed
ed successfully in a diverse set of implantable medical devices. The
sed in these devices are typically developed for the application and
sed various combinations of cathode, electrolyte and separator, to meet
cific requirements of a device. Despite the various approaches that
een used, there are several power source characteristics that are le
across all applications.
or all batteries that power implantable medical devices, the important
erations include the following:
um Batteries for Medical Applications 687
• Safety (tolerance to electrical and mechanical abuse)

• Predictability of performance (voltage, current, and time
relationships)
• Reliability (low variability of performance parameters)
• High energy density (also low battery weight and small battery size)
• Low self discharge (loss of capacity due to internal cell processes)
• End-of-life indication (usually loaded or unloaded voltage based).
addition to the above considerations, charging safety and high cycle
very important factors that must be addressed for rechargeable cells. t
of this overview, the above considerations along with other pertinent
and performance information, will be discussed for the following
battery chemistries: iodine, thionyl chloride, CFx, SVO, Mn02, and Li
PRIMARY BATTERIES
The Use of Lithium Metal as an Anode

he principal reason that lithium metal has been utilized as a battery
aterial is the very high energy density that lithium can provide to
. Lithium possesses an electrochemical equivalence of 3860 mAhlg.
parison, the electrochemical equivalences of the cathode materials to
ssed in this paper are listed in Table 22.1. As can be seen in this data
has over 4 times the gravimetric capacity of the next closest
(CFx at 860 mAhlg) and almost 30 times the capacity of LiCo0 2 (at
hlg).
. Gravimetric and volumetric capacities of cathode materials.
e Material Gravimetric Capacity Volumetric Capacity'

(mAh/g) (mAh/cc)
ne
nyl chloride
860 2322
o 315 1521
02 308 1540
o02 131 668
ric capacity based on the true density of the active cathode materials.
Chapter 22
Lithium metal also has a very low standard potential of -3.05 V at

which when combined with any of the cathode materials discussed here
ces an electrochemical couple with high potential. These two factors,
capacity and low reduction potential, make lithium a good anode
al for many batteries. However, there are some limitations. Most
cant is lithium metal reactivity with water, which prohibits the use of
us electrolytes in lithium batteries under ordinary conditions. This
lly limits lithium anode battery systems to the use of organic
lytes or inorganic electrolytes/soluble cathodes. In using these
ueous electrolytes, the conductivity of the electrolyte is limited, leading
eased cell resistance. However, the self-discharge of the system is also
lly low, which is important for long life in implanted devices. In
l, despite the high energy and reactivity of lithium metal, lithium anode
es can be designed to provide high safety and reliability for implantable
al applications, demonstrated by ample field data over the last thirty
2. Lithium/Iodine Systems and their Medical

Applications
Lithium anode batteries have played a key role in making implantable
dical devices viable therapy. The first implantable device to use a
battery was the cardiac pacemaker. The first implantable pacemaker
2
veloped over forty years ago, and a number of different types of power
s were used in early versions of the device. These included
ercury (Mallory) cells, rechargeable NiCd batteries, nuclear batteries,
veral lithium battery systems. However, for the last 30 years the
/iodine battery has been the dominant battery choice for powering
table pacemakers.
The basic cell reaction in a lithium/iodine battery is:
athode contains a mixture of iodine and poly-2-vinylpyridine (PVP),

when reacted together at high temperature forms a conductive, charge r
complex. In the most commonly used approach to the construction of
cell, the molten cathode material is poured into the cell and a layer of
ms at the anode, producing a separator layer in situ. The energy density
lithium/iodine-PVP system is high, due to the high energy density of I2,
strated in Table 22.1, and because separator material and electrolyte
ot need to be added to the cell. During discharge of the Li/12 cell the Lil
grows in thickness, which results in a rise in cell impedance. The
4
ment techniques of microcalorimeter/ and AC impedance have been
ensively to study the discharge process in Li/I2 cells.
ue to the solid electrolyte of the system, Li/I2 cells typically supply
ents, in the J..tA range. This current output has been able to meet the
ent for implantable pacemakers over the last several decades.
nally, the formation of a self-healing separator system, as well as high
nd reliability has made the Li/I2 cell a good choice as a pacemaker
ource. Li/I2 cells have demonstrated very long life under usage
ns, as illustrated by the real time discharge data displayed in Figure a
pacemaker battery on test for -10 years. Other larger implantable
attery designs have demonstrated longevity under similar test
ns for over 14 years. The Li/I2 battery from Figure 22.1 was
5
ted as previously described, using a complex mixture of iodine and
inylpyridine as the cathode material. The cell was tested under
kohm constant resistance load at 37°C. At the beginning of the
e test, this load corresponds to -28 J..tA current drain on the battery.
2.500
2.000
1.500
1.000
500
0 500 1000 1500 2000 2500

Capacity (mAh)
1. Discharge of a lithium/iodine-PVP battery under a 100 kohm load at 37°C.
Lithium/Thionyl Chloride Systems and their Medical

Applications
hile pacemakers have historically required power in the J.!W range,
plantable medical devices, such as drug pumps and neurostimulators,
her power requirements, in the mW range. Thus, the current drains of
vices are too great for Li/I2 batteries, and considerable work has been
develop lithium/thionyl chloride (TC) as an implantable battery
The discharge reaction for the LiffC battery system is defined as:
Chapter 22
the thionyl chloride acts as both the electrolyte (in combination with a
ch as LiA1Cl4) as well as the active cathode material. In the physical
uction of the cell a carbon black cathode is used to catalyze the
chemical reaction of SOCh. The theoretical capacity of thionyl
de as a cathode material is high, as given in Table 22.1, and practical
etric energy densities as high as 1.2 Wh/cc have been reported for small
6
al LiffC batteries. A typical discharge curve for an implantable LiffC
is presented in Figure 22.2.
4.000
3.500
3.000
> 2.500
.§.
8, 2.000
! 1.500
;
! 1.000
g 500
0
0 500 1000 1500 2000 2500 3000
Capacity (mAh)
2.2. Discharge of a lithiurnlthionyl chloride battery under a 49.9 kohm load at 37°C.
LiffC cells can display voltage delay after prolonged storage periods
the voltage of the cell is temporarily depressed, due to a passivation n
the anode. Techniques have been employed to minimize voltage
including careful attention to the purity of the electrolyte, the
7
tration and type of the electrolyte salt, and the use of additives.
As noted in the introduction, an important aspect of battery systems for
table medical devices is a state-of-charge, or end-of-life (EOL)
ion. In practice this means that it is important for the battery to give
g significantly before all of the useable capacity is depleted. This is an
ant limitation of the LiffC chemistry, since cells provide a flat
rge curve with a rapid loss of voltage at the end of the discharge curve,
trated in Figure 22.2. Several methods have been described to provide
dication, including using S02 discharge voltage as an indicator, as well
use of lithium alloys.6
Lithium/CFx Systems and their Medical Applications

everal implantable medical devices require a medium-rate power
such as drug delivery pumps and neurostimulators as mentioned in the
s section. In addition, new advanced pacemakers which combine
al pacing with other therapies, such as multi-site pacing for
ive heart failure (CHF) treatment, as well as telemetry and
mability features, can require a higher (rnA) current output than can
ded by the Li/lz system. In addition to Li!fC cells, Li/CFx cells have
89
eveloped to meet this need. ' CFx is the notation for
bonmonofluoride, a solid state cathode material for lithium batteries.
of fluorinated carbons as cathode materials was first reported by
be and coworkers of Matsushita Electric Industrial Company. 10 The
electrochemical discharge reaction is described as:
Li + CFx ---7 LiF + C
h an intermediate phase consisting of carbon, fluoride ion, and

lithium ion is believed to decompose into the final LiF and C. 11
he construction of the Li/CFx cell includes a solid cathode, consisting
mixed with a polymeric binder. Since CFx is an insulator, an additive
carbon black is mixed with the material to make the cathode
ive. A separator material, such as a film of polypropylene, separates
ode pellet from the lithium anode. A non-aqueous electrolyte
ng of a lithium salt dissolved in an organic solvent (typically LiBF4 in
butyrolactone) is added to complete the cell. CFx has a very high
density, as illustrated in Table 22.1. Li/CFx cells also have a high
of 3.1 V, and are characterized by internal impedance of about 10
An overlay of discharge curves for an implantable grade Li/CFx battery
ayed in Figure 22.3. These curves were collected under constant
e loads of 1, 2.7, 7.5, 32 and 100 kohms at 37°C. These loads
nd to current drains of -2.8 rnA to 0.03 rnA, and illustrate the current
y of the Li/CFx system.
Lithium/Silver-Vanadium-Oxide Systems and their

Medical Applications
he implantable cardiac defibrillator (lCD) detects the onset of
dia (rapid heart beat) and attempts to provide pacing to regulate the
12
te. If the tachycardia condition is not controlled by pacing, but
Chapter 22
3500r--------------------------------------------,
3000
2500
2000
1500
1000
500 32kohm
0+----- ---- ---- r-----r----- -----r----

0 200 400 600 800 1000 1200 1400
Capacity(mAh)
22.3. Discharge of a lithium!CFx battery under constant resistance loads of: 1, 2.7, 7.5,
100 kohm at 37°C.
orates into ventricular fibrillation, the ICD provides a high-energy shock

y to the heart to stop fibrillation. Thus, ICDs are truly life saving
s. However, these devices also have unique power requirements, which e
the need for continuous low current drain over the entire life of the
(several years) to power the monitoring circuitry. In addition, the
also requires very rapid deployment of high current pulses, typically on
der of 2 to 3 amps, when fibrillation is detected. Both of these
ements need to be provided in a small package with high safety and
lity.
The majority of ICDs implanted since the mid 1980s have used
m/silver vanadium oxide (SVO) batteries. The active cathode material
13
oichiometry of Ag2V4011• The cell reaction is defined as:
The addition of 7 equivalents of Li to SVO corresponds to a theoretical

density of 315 mAhlg for the material, as listed in Table 22.1, and
in an open circuit voltage of -2 V at EOL. SVO has been found to
additional equivalents of Li, but at a much lower running voltage,
g this capacity unusable for a practical cathode. However, in this
, metal vanadium oxides have been investigated as possible low voltage
14
materials for lithium ion batteries.
The discharge reaction of Li/SVO batteries has been studied in detail by
bination of physical and wet chemical methods. 15 The reduction of
sulted in the loss of crystallinity of the material over the range O<x<
LixAg2V4011 with a concurrent reduction of Ag+ to Ag0. The
n of metallic silver greatly increased the conductivity of the cathode
, which plays a role in the high current carrying capability of the
5 4
system. At x values >2.4 the reduction of V + to V + and V3+ was
d, and at x>3.8 mixed valent materials were found, containing V 3+,
5
V + in the same sample. The presence of several different oxidation
f vanadium, as well as silver reduction, in the discharge reaction of
cells results in a stepped discharge curve. This step-wise change in
is predictable and provides state-of-charge indication for the battery
ed in the implantable device.
3500 .,----------------------..,
Prepulse Voltage
r------
3000
/
2500
2000
1500 /
Pulse Minirrum Voltage
1000
500
0 --- - ----------- -
0 2 3 4 5 6 7 8 9 10
Tlrre (years)
.4. Discharge of a lithium/silver vanadium oxide battery under long-life test pulse
at 37°C. Four 10 sec, 2 Amp pulses are applied every 30 days, with a background
0 kohm applied continuously.
/SVO batteries typically use electrolytes based on the non-aqueous

solvents such as propylene carbonate and dimethoxyethane. A
salt is dissolved in the electrolyte, and polypropylene or polyethylene r
is used in the system. The results of long term testing of a Li/SVO
ble grade battery are displayed in Figure 22.4. The Li/SVO
ed in this test was constructed in a prismatic cell configuration as
16
ly described, using a multi-plate design. The test conditions
d a constant resistance background load of 100 kohm with sets of
e pulses applied once every 30 days with the entire test conducted at
he high rate pulses of 2 A were each 10 sec in duration and applied in
our, 15 sec apart. As can be seen in Figure 22.4, the cell displayed g
life characteristics, with 9 years of real time data collected.
Chapter 22
The evolution of ICDs has led to a significant decrease in size of these

s, and in turn has put greater demands on the power level provided by
r Li/SVO batteries. Thus, several studies have been carried out to
17 18 19
terize the power capability of Li/SVO cells. In addition, the effect
' '
trolyte additives on minimizing voltage delay and resistance build-up in
20
O cells has been extensively studied. '21'22 Electrolyte additives such as
r organic carbonates like dibenzyl carbonate (DBC) have a significant
on the long term performance of Li/SVO batteries and are believed to
internal resistance by modifying the surface SEI layer at the lithium
6. Lithium/Mn02 Systems and their Medical Applications

In addition to SVO, another cathode system that has been used for ICD
sources is manganese dioxide (Mn02). High rate lithium/manganese e
batteries have been produced where chromium oxide and lead oxide
23
es have been mixed with the Mn02• '24'25 These cells meet the
ements described in the section above. The use of chromium oxides
2.5<x<2.7) and lead oxides (PbCr04, PbMo04 and PbO) in
nation with manganese oxide provided for improved EOL indication for
23
Mn02 system.
For the ICD application, a double cell concept has been developed using
26
/Mn02 system, which provides a 6 V running voltage. The two cells
onnected in series to provide the high voltage, and to avoid external
ell connection, the battery case was used as the electrical connector
n the two cells.
Swelling of batteries during discharge is an important consideration for
table devices, and the swelling of implantable Li/Mn02 cells was
5
d under accelerated test conditions.Z The swelling was found to depend
voltage of the cells, and was attributed to decomposition of the organic
lyte components.
SECONDARY BATTERIES
1. Lithium Ion Systems - General Considerations

While primary lithium batteries have been the subject of extensive
h and enjoy continued commercial success in medical applications,
ary batteries employing lithium ion chemistry have begun to enter the
al power source market. Lithium ion batteries differ from primary
in that the anodes are not lithium metal, which allows for safe
ng of this chemistry under controlled conditions. Lithium ion
also enjoy several distinct advantages over other secondary battery
gies such as nickel-cadmium (NiCd) or nickel metal hydride
, namely: 1) lithium ion cells possess a much higher discharge voltage
which is about three times the running voltage of NiCd and NiMH
ries, 2) lithium ion cells display a much lower self-discharge rate, and
memory effects of NiCd and NiMH systems are not present with
27
ion cells. '30
thium ion cells for implantable medical applications have mainly used
28
cathodes with graphite anodes. The gravimetric energy density of is
listed in Table 22.1, and while the value of 131 mAh/g is the lowest
ble, the rechargeability of the material over several hundreds of cycles
n a much larger useable capacity over the life of the battery. The
fe of a small (40 mAh) lithium ion cell designed for medical
able application is displayed in Figure 22.5. Here the cell was
120%
100%
-....
80%
60%
40%
20%
0%
0 100 200 300 400 500 600
Cycle Number
.5. Discharge capacity for a small (40 mAh) lithium ion cell cycled between voltage
.10 V and 2.75 V using a charge and discharge current of 7 rnA at 37°C.
and discharged at 37°C between voltage limits of 2.75 and 4.10 V,

charge/discharge current of 7 rnA, corresponding to a low rate of
This cell design displayed good cycle life, with >80% of its initial
past 500 cycles. Larger lithium ion cells have also been designed for
ist applications. The cycle life curve for a 1.1 Ah cell charged at a
and discharged at 1 C rate is displayed in Figure 22.6. This cell was
sted at 37°C, between voltage limits of 2.75 and 4.10 V, and displays
Chapter 22
apacity retention. The higher currents used in this test simulate faster
ng (2 h) and high current drain that are a requirement for heart-assist
ations, as outlined below.
Due to the rechargeable nature of lithium ion cells, it has become
ant to develop end of life (EOL) indicators for lithium ion batteries.
ample, impedance spectroscopy has been tested as an EOL indicator
120%
100% .
-·
.
1
80% -
60%
40%
20%
0% 400 600 800 1000 1200

0 200 Cycle Number
2.6. Discharge capacity for a 1.1 Ah lithium ion cell cycled between voltage limits of
and 2.75 V using a charge current of 550 rnA and a discharge current of 1100 rnA at
29
hium ion batteries. Sealed prismatic Sanyo UF653467 cells with a
al capacity of 930 mAh were cycled at a 1C charge-discharge rate.
286 cycles, the battery capacity dropped from 1005 to 700 mAh.
ance measurements of the fully discharged batteries were made and the
st plots as a function of cycle number showed an increase in the size of
w frequency semicircle with the number of cycles, due to the increasing
cial resistance of the cathode and anode. It was proposed therefore that
ance spectroscopy could be a good cycle life predictor for a sealed
ion battery.
The significant difference between lithium metal anodes and graphite
in secondary lithium ion cells is that lithium metal anodes suffer from
dendrite formation, while under controlled cycling conditions,
te anodes do not. Lithium dendrite formation poses significant safety
ns, which ultimately results in a non-viable battery for medical
ations. In addition to lithium dendrite concerns, safety considerations
significant role in the design and testing of lithium-ion cells employed
dical applications2730
' Abuse testing for small lithium ion cells have
efined by the International Electrotechnical Committee (EIC),

riter's Laboratories (UL), and the Japan Storage Battery Association
1
A number of these abuse tests are conducted on lithium ion cells for
applications to evaluate the safety of the design, including:
• Short circuit (on both fused and unfused cells)
• Overcharge
• Overdischarge
• Shock (or drop)
• Crush
• Nail penetration
• Hotbox type tests.
addition, ARC (accelerating rate calorimetry) testing has been
ely studied for lithium ion cells to determine the thermal stability of
32 33
s and design. '
f particular concern for medical cells, the charging of lithium ion
must be carefully controlled, as evidenced by the results of
38
rge tests. Figure 22.7 displays the temperature and voltage curves
during overcharge abuse testing for a prismatic lithium ion cell
d for medical application. This cell was overcharged well past the
4.1 V cutoff, to the point of failure. When all lithium is extracted
e cathode (100% of charge in Figure 22.7), the cell temperature
d rapidly, resulting in thermal runaway and eventual cell venting.
medical applications it is critical that redundant over-voltage control
30
oyed to avoid overcharge conditions.
Lithium Ion Systems - Current and Future Medical

Applications
everal implantable biomedical devices possess power requirements
re so high that they preclude the use of a primary lithium battery. In
ses, rechargeable batteries offer a power source with small size that
o power the device for several years. Two examples of such devices
left ventricular assist device (LVAD) and the total artificial heart
In both examples lithium ion cells have been developed for these
34
ions due to the high energy density of this system. Current versions
devices are powered from external battery packs that can typically be
n the patient's belt. However, there is also a desire to have an
ed back-up battery that will allow the patient to remove the external
ack for a short period of time each day.
Chapter 22
120 12
11
100 10
u
...
.
E 80
!. 8
7
"..'
i
u
60
'5
"' :)_To
i Voltage 6 >
<
40
4
20 3
0 20 40 60 80 100 120
%of Charge
22.7. Temperature and voltage curves for a 1.5 Ah lithium ion cell under overcharge
onditions. The cell was charged at a constant current of 1.5 A (C-rate) at RT where
f charge equaled complete removal of lithium from the LixCo02 cathode material.
The left ventricular assist device is an implantable pumping mechanism

rovides long-term circulatory support to patients with serious heart
35
e. These devices are often used as a "bridge" to a heart transplant,
the LVAD assists the heart in pumping blood, while the heart remains
e in the patient. As a mechanical pump, the device requires high power
ntinuous operation. In one LVAD design the power load is described as
36
for a 12 V battery pack. The total artificial heart is a mechanical
that completely replaces the heart in the patient. Several experimental
37
have been implanted in patients recently. The design of this
ement heart contains two chambers, each capable of pumping over two s
of blood per minute. This device utilizes an implantable Li-ion
pack that is recharged through the skin without any wires or tubes
ating the skin. The implantable batteries for these applications must
ll the requirements defined in the introduction, with additional safety
ements stemming from the charging steps involved with the lithium ion
8
Small button-sized lithium ion cells have been developed for potential
external medical devices such as common hearing aids (HAD),
ermal electrical nerve stimulation devices (TENS), and devices for the
39
ermal application of other drugs (TAD). It has been proposed that
mall lithium ion secondary cells could mitigate the cost, inconvenience,
nvironmental hazard associated with the continual replacement and
al of small primary cells for external medical devices. The 0.45 g, 0.18
iniature cells successfully completed 2700 cycles at 60% DOD and
e rates of 1-3 rnA. The miniature cells had average energy densities
Wh/1 and specific energies of 75 Whlkg, offering approximately 80%
nergy density and 70% of the specific energy of large, commercial
on batteries.
SUMMARY
Lithium batteries for implantable biomedical applications are in

ead use today. Lithium/iodine cells have the longest history, and by
greatest number of implants, numbering into the millions as power
for cardiac pacemakers. Lithium/thionyl chloride and lithium/CFx
have found utility in medium-rate applications, such as
mulators and drug pumps, while lithium/SVO and lithium/manganese
batteries provide the high rate capability needed to power implantable
defibrillators. Rechargeable lithium ion cells have been suggested for r
of future implantable applications, where the high energy density in
geable package lends itself to applications not currently served by
40
cells. • 41 Lithium ion cells have already demonstrated utility in
g left ventricular assist devices as well as total artificial hearts. While
ese applications require small power sources with high energy density
to keep the entire implantable device small, the requirements of
eliability and predictability of performance are the top priority.
ENCES
lmes, J. Power Sources 97-98 (2001) 739.
tbatch, The Making of the Pacemaker (Prometheus Books, Amherst, NY, 2000) p.
tbatch, R. McLean, W. Holmes, C. Holmes, IEEE Trans. On Biomed. Eng.

(1979) 306.
hmidt, P.M. Skarstad, J. Power Sources 65 (1997) 121.
ang, C.F. Holmes, Prog. Batt. & Solar Cells 2 (1979) 50.
arstad, in Batteries for Implantable Biomedical Devices, B.B. Owens, Ed.,
Press, New York, 1986) p. 215.
hlaikjer, in Lithium Batteries, J.-P. Gabano, Ed., (Academic Press, New York,
303.
keuchi, S.A. Smesko, in 1h Annual Battery Conference on Applications and
s, Long Beach, CA 1992.
tbatch, C.F. Holmes, E.S. Takeuchi, S.J. Ebel, in Cardiostim 96, Nice, France 1996.
anabe, M. Fukuda, U.S. Patent No. 3,536,532 (1970).
tanabe, T. Nakajima, R. Hagiwara, J. Power Sources 20 (1987) 87. akeuchi,
W.D.K. Clark, in Implantable Cardioverter-Defibrillators, A hensive
Textbook, N.A.M. Estes, A.S. Manolis and P. Wang, Eds., (Marcel
, New York, 1994) p. 123.
akeuchi, A.C. Marschilok, S.M. Davis, R.A. Leising, E.S. Takeuchi, Coord.
Chapter 22
rn. Rev. 219-221 (2001) 283.

Piffard, F. Leroux, D. Guyomard, J.-L. Mansot, M. Toumoux, J. Power Sources 68
97) 698.
. Leising, W.C. Thiebolt, E.S. Takeuchi, Inorg. Chern. 33 (1994) 5733.
. Takeuchi, W.C. Thiebolt, J. Electrochern. Soc. 135 (1988) 2691.
orton, C. Schmidt in Proceedings of the Symposium on Batteries for Portable
lications and Electric Vehicles, C. F. Holmes and A.R. Landgrebe, Eds.,
97-18, p.389.
Crespi, C. Schmidt, J. Norton, K. Chen, P. Skarstad, J. Electrochern. Soc. 148
01) A30.
chmidt, G. Tam, E. Scott, J. Norton, K. Chen, 1 Jlh International Meeting on
ium Batteries, Monterey, CA, June 23-28, 2002, Abstract 422.
an, E.S. Takeuchi, J. Power Sources 62 (1996) 45.
an, E.S. Takeuchi, ECS Joint International Meeting HI, Oct. 17-22, 1999, Abstract
.
an, E.S. Takeuchi, ECS Proceedings Volume 99-25, p. 417.
ehrmann, R. Fromme!, R. Wolf, U.S. Patent No. 5,587,258 (December 24, 1996).
rews, R. Wolf, G. Fehrmann, R. Staub, J. Power Sources 80 (1999) 107.
rews, G. Fehrmann, R. Staub, R. Wolf, J. Power Sources 97-98 (2001) 747.
rews, R. Wolf, G. Fehrmann, R. Staub, J. Power Sources 65 (1997) 129.
M. Spillman, E. S. Takeuchi, 14th Annual Battery Conference on Applications and
ances, Long Beach, CA, 1999.
Holmes, R.A. Leising, D.M. Spillman, E.S. Takeuchi, ITE Battery Letters l
9) 132.
, E. Murphy, J. Winnick, P.A. Kohl. J. Power Sources 102 (2001) 294.
. Rubino, H. Gan, K. J. Takeuchi, 17th Annual Battery Conference on Applications
Advances, Long Beach, CA, 2002.
potnitz, in Advances in Lithium-Ion Batteries, W.A. van Schalkwijk and B. Scrosati,
., (Kluwer Academic/Plenum Publishers, New York, 2002) 433.
nnerud, J. Shi, R. Chamberlain, S.K. Singh, P. Ralbovsky, B. Barnett, C. Lampe
erud, 20Ft Electrochemical Society Meeting, Philadelphia, PA, May 12-17,2002,
tract No. 109.
. MacNeil, T.D. Hatchard, J. R. Dahn, J. Electrochem. Soc. 148 (2001) A663.
acobs, J. Albright, E. Marlinski, C. Deirmengian, R. McClure, H. Valenta, E.S.
euchi, D. Tamez, J. Conger, O.H. Frazier, R. Jarvik, ASA/0 45 (1999) p.165. amez,
J.L. Conger, G. Jacobs, I. Gregoric, R.W. Inman, B.R. Radovancevic, S.M. re, K.
Eya, H. Eichstaedt, R.K. Jarvik, O.H. Frazier, ASA/0 46 (2000) p. 168.
. MacClean, P.A. Aiken, W.A. Adams, T. Mussivand, J. Power Sources 56 (1995)
more information regarding TAH devices, see the Jewish Hospital, University of
isville, Health Sciences Center. Website: www.heartpioneers.com
. Leising, M.J. Palazzo, E.S. Takeuchi, K.J. Takeuchi, J. Electrochern. Soc. 148
1) A838.
asserini, B.B. Owens, J. Power Sources 97-98 (2001) 750.
Schmidt, P.M.Skarstad, J. Power Sources 97-98 (2001) 742.
51
Takeuchi, R.A. Leising, 201 Electrochemical Society Meeting, Philadelphia, PA,
12-17,2002, Abstract No. 260.
ter 23
RRENT ISSUES AND MARKET TRENDS OF

Li-ION BATTERIES FOR CONSUMER
APPLICATIONS
D. MacArthur
CHEMAC International Corporation, Troy, Michigan 48043, U.S.A.
INTRODUCTION
The focus of this chapter is Li-Ion batteries in consumer applications.
is a long history and a large market of non Li-Ion batteries--many of
rimary lithium batteries--but this other field is too extensive for
on here except to mention the recent apparent success to
rcialize, after a long gestation period, the lithium metal-polymer
lyte technology in the telecommunications backup application. The
should also note that Li-Ion batteries are now finding their way into
non-consumer applications. This, too, is outside the scope of the
chapter.
The subject begins with the lithium metal/molybdenum disulfide
introduced by Moli Energy in the late 1970, early 1980 period. Other
eable lithium systems were introduced for consumer applications
his same time but the Moli cell was the first to reach wide spread use.
ery well known, the product encountered a safety issue. Was it just
construction or was the lithium metal-liquid electrolyte system so e
as to be unacceptable for consumer applications? (And does the
polymer electrolyte solve this problem?) Whatever the technical
ation, the technology gained a perceived image as unsafe and there
need to move on. Sony researchers (and others) saw the opportunity
veloped a high capacity carbon based anode. (Research into lithium
lation in graphite has a much longer history but the capacities needed
ctical use had generally not been observed.) This anode, along with a
oltage metal oxide cathode was key to what came to be called the Li
ttery, first officially described in 1991. The name was intended to
ate the new system from the "unsafe" lithium battery. It is, of course,
imply a "rocking chair" type battery (or the physicist's battery as
Chapter 23
on mentioned). The researchers, themselves, cannot be blamed for the

g confusion from the unfortunate use of "anode" and "cathode" in a
geable system. Not incidental to Sony's development of Li-Ion
logy was a need to leapfrog the nickel-metal hydride technology in the
g portable electronics market.
By early 1980, researchers interested in lithium batteries started
g meetings on just this subject for the exchange of information. Prior
, much of the interest was in very high energy lithium batteries, many
litary interest including molten salt systems. The first of the
ational Meetings on Lithium Batteries was held in Rome in 1982.
t a decade later, the 5th International Meeting on Lithium Batteries was
n 1990 in Beijing, China and that is when the story of Li-Ion really
. A summary of the meeting is contained in the first of the Powers s
on Lithium Batteries. Moli Energy had just discontinued the S2
cell and was proceeding with Li/Mn02 which was considered to be
azardous. The Bell Labs Li/NbSe3 system had been discontinued. l
groups had field tests of Li/V6013 in "AA'' and "C" size cells
way. Developers of LiffiS 2 systems were promoting this technology
e Mead-Hope-Innovision (MHB) group was developing the LiN6013
ith in-situ cross-linked doped polymer electrolyte. Honeywell was
to be working on a 4 volt Li/Co02 system. The Harwell consortium
ium polymer batteries was developing the "dry" polymer electrolyte
s. And Sony EnergyTec had just dropped a highly publicized Li/Mn02
for a new unspecified system called Li-Ion. (Robert Powers, the
of the report, speculated it seemed to be a lithium-carbon fiber/Co02
.) The Powers Reports continued to follow, at two year intervals,
ies and progress in lithium batteries during the next decade.
1. The Market
he growth of the Li-Ion market has been huge, a matter which refutes
mage of battery technology as slow-moving, out-of-date, and
onsive. During the period of 1991 to 1999 growth of not just the Li
arket but the so-called advanced battery market closely paralleled that
semiconductor industry, doubling and redoubling in periods of just a
ars or less. By year 2000, however, growth had greatly diminished
mostly because of the faltering world economy but perhaps also the of
reduction in number and significance of technical advances. The
rowth and following slowdown is shown in the following Table 23.1.
rent Issues and Market Trends of Li-Ion Batteries for Consumer Applications 703
.1. World annual sales ofLi-Ion batteries.
(millions of cells)
illions of year 2000 rs)
ductiOn m 1992 was entirely Sony captive. Value m the table 1s based on estimated factory cost.
cted from the Powers Reports.
t the same time reported manufacturing capacity grew even more

(Tables 23.2-23.3). The number of companies establishing
cturing has been surprising. By year 2002 there were 10 major
cturers located in Japan, Taiwan or Korea and another 4-6 growing
cturers most of which were located in the Republic of China. There is
in manufacturing to China where modem plants producing high
product are rapidly appearing. SAFT, Matsushita, the newly formed
US Lithium Energetics, and several others were located in North
a.
here is a large manufacturing overcapacity which places considerable
n the industry.
.2. Growth in world Li-Ion manufacturing capacity (millions of cells per year at
of the year).
olymer
a from the Powers Reports. Almost all Ll-Ion polymer are of the m-sltu polymenzed type of a
esign intended for wireless telephones.
3. Geographic distribution of Li-Ion manufacturing capacity (2001).
REGION
Japan
China (including Taiwan)
Korea
North America and Europe
e large excess of manufacturing capacity has resulted in a highly

itive environment and falling prices. The opportunity for recovery of
h cost of introduction of technology, the high investment for
ion and high material costs of start-up is now small. An approximate
eakdown of the Li-Ion technology is: cathode 25%, anode 10%,
Chapter 23
ging (includes electrolyte) 20%, electronics 18%, overhead 15%, all

2%. Much of recent development is optimizing the technology for
ic applications.
2. Different Technologies
Early constructions were 18650 (18 mm diameter, 650 mm long) jelly
lls which, over the years, have grown from about 1 Ah capacity in
to 2 Ah capacity in 2000. (The Molicell of 1998 which was "AA" size
r to the 18650 size had about 500 mAh capacity.) Much of the increase
acity is the result of higher density carbon materials for the anode and
packing. By about 1995 there was a shift to 17650 cells of better fit to
nner portable computers. About 300 million cells go to the wireless
one market and about 200 million to the portable computer market.
50% of Li-Ion production is now prismatic. In Table 23.4, the gain in
content over the years is reported.
By 1998, there was growing interest in "flat packs" with superior
ing flexibility for wireless telephones. The introduction of polymer
lytes follows a long history beginning even before the appearance of
batteries. Several technologies have been developed: the "dry" er,
the "solid state" in-situ cross-linked polymer, the gelled in-situ er,
the extracted-activated polymer of the Bellcore technology. Early
vision of the polymer technology was to make batteries like paper is
The expectation of lower cost has not been fulfilled and the critical
has become the form factor as the packages can be thin and shaped to
hin the contours of the telephone. By 2000 Li-Ion chemistry dominated
reless telephone and portable computer markets but with the saturation
ting markets, if Li-Ion production is to increase there is a need to find
pplications. Although the proliferation of portable electronic devices
ation assistants, communicators, tools, and toys-- is growing it does
pear to predict a doubling in battery demand. A new market is needed.
is interest, in smaller sizes such as credit card sizes but Li-Ion
logy is not well suited and many uses are served by primary
logy. Much of the focus is on markets for larger sizes. This includes
backup, power tools and electric vehicles. The prospect of a huge
vehicle market is very enticing. The challenges are cost and safety.
teristics of small and large batteries are reported in Table 23.5
ent issues and Market Trends of Li-lon Batteries for Consumer Applications 705
4. Energy content of Li-lon cells.
Density (Whll)
ic Energy (Whlkg)
numbers are for commercial cells. The change m specific energy, year 2000, 1s related to
n of the plastic envelope package. The change in energy density over the 1992-98 period was f
changes in capacity of the anode and in packaging.
Targets: Small and Large Batteries
he growth of Li-Ion depends on being lighter, smaller and having

harge retention than competitive products. But, the technology uses
tible electrolytes and clearly there is a safety issue here even though
tion inhibitors may be added. The overlap of electrode potentials with
osition reactions of the system makes stability of the system
nt on inherently unstable materials such as the SEI layer, and the
of a recombination system in overcharge forces use of sophisticated
ic controls. Yet, the technology is moving into applications of larger
issan has used Li-Ion of the spinel manganese oxide type in electric
since about 1998. Almost all automotive vehicle manufacturers are
g Li-Ion batteries in hybrid vehicles. Li-Ion technology is well suited
life where batteries are cycled in mid range of charge. JSB has
20K cycles at 20% DOD with only 22% capacity loss (of full
1
) with large Li-Ion cobalt oxide cells for space applications. SAFr
rted on Li-lon (40 Ah jelly roll cells in about 3.5 kWh installations)
2
ommunications backup and has gained up to 5 years of experience.
ated testing is indicating up to 11 year life which is an excellent result
stage in development. As pricing declines into the fifty cents per Wh
nd if no major safety events occur, these products may establish
ant markets.
5. Targets for market growth, Li-Ion type.
TTRIBUTE
E(Whlkg)
D (Whll)
D (W/1) peak/sustained
ife
emperature Range ( C)0
osts ($/Wh)
afety/Robust
ecycle
Chapter 23
Adoption of new materials into the technology is expected to be slow

ecause of the low margins within the industry. This inertia provides an
tunity for startup companies which may be able to establish a base
newer technology. Most manufacturers are using graphite anodes
gh a few may be using a hard carbon. Carbon anodes, although more
, have lower capacity per unit volume. Various pyrolized carbon
als with greater than the 372 mAhlg capacity of the Li-C6 composition
been explored but as far as known are not in use. Carbon coated
te materials may be used by some manufacturers. The higher capacity
y micro alloy materials are not used in Li-Ion cells (although they may
some primary cells) because of lower cycling capability. Each
acturer seems to have a preferred electrolyte composition but all are
on the cyclic carbonates. Choices seem to be influenced by
rature range desired. Shutdown separators are widely used and some
acturers may be using flame inhibitors. Most polymer electrolyte
s in use are of the thermally in-situ cross-linked type; the competing
re process does not appear to widely adopted. The liquid filled, in-situ
inked process has the advantage of compatibility with existing Li-Ion
tion. For the cathodes, most manufacturers are using a cobalt oxide al
often with metal oxide doping; nickel oxide and manganese oxide als,
again with metal oxide doping are in use by a few manufacturers.
livene lithium iron phosphate material is in development. Because of
ial cost advantages it may be used in large cells; however, decrease in e
and potentially lower cycle life may negate any cost advantage.
t continues with manganese oxide based materials for cost,
nmental, and safety reasons.
ERENCES
mamura, T. Inoue, H. Yoshida, M. Mizutani, M. Goto, Power Sources Conference,
ry Hill, NJ, June 10-13, 2002, Paper 26-5.
Dowall, A. Brenier, M. Broussely, P. Lavaur, INTELEC 2002 Conference, Montreal
ec, Sept. 30-0ct 3, 2002, Paper 22.2.
ject Index
LiFeP207 cathode 471
ves for lithium battery 303, 174
materials LiMX04 445
Aluminides Li3Mz(X04h cathodes 457
Antimonides 116 121, 275 LiNi02 cathode 19
Cu-Sn 123 LiVO.33, 345
Graphite anodes 144 Polyanion cathodes 32
Intermetallic anodes 12, 125, 270 Role of bond covalency 64
Lithiated carbons 146 Spinel oxide cathode 25, 30, 362
Metallic lithium anode 297 Spinel manganese oxide 26
Metal oxide anodes 129 Sulfides cathode 350
Mg2Si 127 Synthesis of cathode materials
Nitrides 132, 247 86,317
LiMNx248 Titanium disulfide 5, 90
InN 254 V6013 34
Sn3N4 254 Composite anodes 112
G N4254 Covalent bonding in Li.Co02
Cu3N 257 Crystal field stabilization energy 21
Phospides 134 Electronic properties of LiNi02 , 410
Silicides 127, 247, 259
Li.Si 260 Decrepitation of alloy anodes 290
Li-Si-Mg 261
CrSi2 265 Electrode fabrication 200
NiSi 266 Electron diffraction in crystals 481
Silicon based composite 127 Electrolytes related
Tin anode 116 Additives 174, 303
Tin oxide anode 117 Alkyl carbonates 509, 531
Transition metal oxide 130 Binary electrolytes 522
synthesis 102 Cathodic reactions 545
Ceramic electrolytes 623
Dipole-dipole interactions 513
y buffer operation 673
Electrochemical stability 525
y cost 662
Electron-pair donor (EPD) 516
y design 657
y market 702 Electron-pair acceptor (EPA)
y medical applications 686 Glass electrolytes 625
y safety 661 Gel electrolytes 591
y satellite applications 675 Hydrogen bonding 515
y stationary applications 671 516
y standby operation 671 Ion-dipole forces 511
y space applications 674 Ion transport 575
Liquid electrolytes 509
Mixed electrolyte 509
naceous anodes 114
de materials, 315 Molten salts 567
Cr oxide cathode 345 Physical properties 532
LiNi 1.•Co.02 334, 410 Polar solvents 530
First principle calculations 42 Polyelectrolytes 602
General electronic structure 45 Polymer electrolytes 575
Layered oxide cathodes 11, 24 Selective solvation 518
LiCo02 12, 410, 414 Solvation 517
LiCo.Nit-xOz 333,410 Stability 604
LiFeP04 99, 347, 445 Ternary solvent blends 524
Subject Index
teries 652 Perovskite electrolytes 631

tery requirements 653 Polyanion based cathodes 445
Polyelectrolytes 602
inciples methods 48 Polymer electrolytes 575
Primary batteries for medical applications
analysis 202 687
ctrolytes 591
lized gradient approximation 54 Reactivity of lithiated carbons 167, 195
electrolytes 625 Reversible-irreversible capacity 157
ic carbon 148 Safety issues of lithium battery 113, 661
pplications 663 Saft battery design 677

atteries 645 SEI layer on carbon anode 171, 176
hermal synthesis 93 Infrared analysis 208
Thermal analysis 216
ial Li-ion batteries 646 Solvent co-intercalation 171
ation process 88 Space applications 674
tallic anodes 117 Spin polarization 55
ttent battery charging 673 Spinel cathode 361, 381
hange methods 95 Stabilization of layered oxide cathodes 97
Sulfide glasses 628
ller distortion 27, 493 Synthesis
Hydrothermal 93
ers battery 683 Ion exchange methods 95
LiMn02 cathode 341, 381 Mechanochemistry 103
to spinel transformation 71, 78, 494 Soft chemistry routes 390
2 instability 384
626 TEM study of cathode materials 489
alloys 280, 283 Thionitride glassy electrolytes 629
flat cell201 Tin compound anode 119
-ion battery technology 651 Tin oxide anode 117
athode 100 Titanium disulfide 86
d carbon anode 146
metal anode 112 Vacancy ordering in LiCo02 487
ordering in LiCo02, 489 Vacancy ordering in oxide cathodes 484
storage in carbon 148, 153, 160 Vacancy ordering in 03 LiNi02 485
ensity approximation (LDA) 54
X-ray diffraction of graphite 199
ese oxide cathode 344
ese spinel cathode 335
analysis for Li-ion batteries 701
ochemistry 103
xide anode 130
logy and particle size 321
-type phosphate electrolytes 632

phitic carbon anodes 151
lasses 626
ide Glassy electrolytes 626
ide glassy electrolytes 628
·- 5,00

Lithium Batteries Science and Technology

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Lithium Batteries Science and Technology

Încărcat de

Drepturi de autor:

Formate disponibile

LITHIUM

Stabilization of Layered Oxide Cathode

ISBN: 978-1-4020-7628-2 (hardcover)

Library of Congress Control Number: 2008941279

© Springer Science+Business Media, LLC 2003, First softcover printing 2009

Printed on acid-free paper

erials Aspects: An Overview

Role of Electronic Properties in the Electrochemical Behavior of

thesis of Battery Materials

es and Composite Anodes: An Overview

onaceous and Graphitic Anodes

de and Silicide Negative Electrodes

s and Intermetallic Anodes

9.2.1. Lithium- Carbon Alloys 271

rent Issues of Metallic Lithium Anode

el Cathode Materials for Lithium-Ion Batteries

12.1. Introduction 361

12.2. LiMn204 and Metal-Ion Doped Spinels LiMxMn2.x04 362

yered Manganese Oxides as Cathodes

hodes Based on LiCo02 and LiNi02

14.2.3. Electrochemical Properties and Capacity Fading

yanion-Based Positive Electrode Materials

erstanding Phase Transformations in Lithium Battery Materials by

16.3. Lithium and Vacancy Ordering 484

16.3.2. Lithium and Vacancy Ordering in 03 L,Co02 487

vanced Liquid Electrolyte Solutions

18.1. Introduction 530

18.3.2. Oxidation Reactions in Polar Aprotic Electrolyte 547

ymeric Electrolytes: An Overview

s and Ceramic Electrolytes for Lithium and Lithium-Ion Batteries

20.3.4. Lithium Metal Halides, Nitrides, and Phosphides 633

tery Systems and Applications

ium Batteries for Medical Applications

22.3. Secondary Batteries 694

Chapter Name Chapter

The science and technology of lithium batteries have dominated the

s of the reversible redox process in transition metal oxide with lithium.

erformances of ceramic electrolytes. Physical methods for preparing

Gholam-Abbas Nazri Gianfranco Pistoia

MATERIALS ASPECTS: AN OVERVIEW

aterials Science and Engineering Program, The University of Texas at Austin,

The exponential growth in portable electronic devices such as cellular

BASIC CONCEPTS OF RECHARGEABLE

ode material), called lithium insertion compound, during the

1.1. Comparison of the gravimetric and volumetric energy densities of various

The open-circuit voltage Voc of such a lithium cell is given by the

F is the Faraday constant. Figure 1.3 shows a schematic energy

Although the concept is simple, the lithium insertion compound

e insertion compound LixMyXz (X = anion) should have a high lithium

insertion compound LixMyXz should allow an insertion/extraction of

lithium insertion/extraction process should be reversible with no or

e insertion compound should be chemically stable without undergoing

e redox energy of the cathode in the entire range, x, of lithium

m a commercial point of view, the insertion compound should be

Following the initial demonstration of the concept of rechargeable

Density of State N(E) Density of State N(E)

CELLS WITH INSERTION COMPOUNDS AS

Despite the development of several lithium insertion compounds

ivating film on the metallic lithium anode. Although the passivating

1.5. Electrochemical potential ranges of some lithium insertion compounds in

Such cells consisting of lithium insertion compounds as both cathodes

unds, LiCo02, LiNi02 and LiMn204 oxides having a high electrode

1.6. Schematic illustration of the charge/discharge process in a lithium-ion cell

rnAh!g causes an important sacrifice in energy density, the lithium