Applied Physics in The 21

HORIZONS IN WORLD PHYSICS SERIES
APPLIED PHYSICS IN THE 21ST

CENTURY, (HORIZONS IN WORLD
PHYSICS, VOLUME 266)
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or
by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no
expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of information
contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in
rendering legal, medical or any other professional services.
Volume 266 Applied Physics in the 21st Century, Edited by Raymond P. Valencia
Volume 265 The Physics of Quarks: New Research, Edited by Nicolas L. Watson and Theo M. Grant
Volume 264 Lev Davidovich Landau and his Impact on Contemporary Theoretical Physics
Edited by Ammar Sakaji and Ignazio Licata
Volume 263 New Trends in Quantum Coherence and Nonlinear Optics Edited by Rafael Drampyan
Volume 262 Astrophysics and Condensed Matter Edited Thomas G. Hardwell
Volume 261 Superconductivity Research Advances Edited by James E. Nolan
Volume 260 Instabilities of Relativistic Electron Beam in Plasma, Edited by Valery B. Krasovitskii
Volume 259 Self-Focusing of Relativistic Electron Bunches in Plasma, Edited by Valery B. Krasovitskii
Volume 258 New Topics in Theoretical Physics, Edited by Henk F. Arnoldus and Thomas George
Volume 257 Boson’s Ferromagnetism and Crystal Growth Research, Edited by Emerson D. Seifer
Volume 256 Horizons in World Physics, Edited by Victor H. Marselle
Volume 255 Frontiers in General Relativity and Quantum Cosmology Research, Edited by Victor H. Marselle
Volume 254 Optics and Electro-Optics Research, Edited by Albert V. Berzilla
Volume 253 Trends in General Relativity and Quantum Cosmology, Edited by Albert Reimer
Volume 252 Chemical Physics Research Trends, Edited by Benjamin V. Arnold
Volume 251 Quantum Dots: Research Developments, Edited by Peter A. Ling
Volume 250 Quantum Gravity Research Trends, Edited by Albert Reimer
Volume 249 General Relativity Research Trends, Edited by Albert Reimer
Volume 248 Spacetime Physics Research Trends, Edited by Albert Reimer
Volume 247 New Developments in Quantum Cosmology Research, Edited by Albert Reimer
Volume 246 Quantum Cosmology Research Trends, Edited by Albert Reimer
Volume 245 Horizons in World Physics, Edited by Tori V. Lynch
Volume 244 Horizons in World Physics, Edited by Albert Reimer
Volume 243 Theoretical Physics 2002, Part 2, Edited by Henk F. Arnoldus and Thomas F. George
Volume 242 Models and Methods of High-Tc Superconductivity: Some Frontal Aspects, Volume 2,
Edited by J. K. Srivastava and S. M. Rao
Volume 241 Models and Methods of High-Tc Superconductivity: Some Frontal Aspects, Volume 1,
Edited by J. K. Srivastava and S. M. Rao
Volume 240 Horizons in World Physics, Edited by Albert Reimer
Volume 239 Theoretical Physics 2002, Part 1, Edited by Henk F. Arnoldus and Thomas F. George
Volume 238 Theoretical Physics 2001, Edited by Henk F. Arnoldus and Thomas F. George
Volume 237 Generalized functions in mathematical physics: Main ideas and concepts,
Edited by A.S. Demidov
Volume 236 Edge Excitations of Low-Dimensional Charged Systems, Edited by Oleg Kirichek
Volume 235 On the Structure of the Physical Vacuum and a New Interaction in Nature (Theory,
Experiment, Applications), Edited by Yu. A. Baurov
Volume 234 Electron Scattering on Complex Atoms (Ions), Edited by V. Lengyel, O. Zatsarinny and E. Remeta
Volume 233 An Introduction to Hot Laser Plasma Physics,
Edited by A.A. Andreev, A.A. Mak and N.A. Solovyev
Volume 232 Electron-Atom Scattering: An Introduction, Maurizio Dapor
Volume 231 Theory of Structure Transformations in Non-equilibrium Condensed Matter,
Edited by Alexander L. Olemskoi
Volume 230 Topics in Cosmic-Ray Astrophysics, Edited by M. A. DuVernois
Volume 229 Interactions of Ions in Condensed Matter, Edited by A. Galdikas and L. Pranevièius
Volume 228 Optical Vortices, Edited by M. Vasnetsov and K. Staliunas
Volume 227 Frontiers in Field Theory,
Edited by Quantum Gravity and Strings, R.K. Kaul, J. Maharana, S. Mukhi and S. K. Rama
Volume 226 Quantum Groups, Noncommutative Geometry and Fundamental Physical Interactions,
Edited by D. Kastler, M. Russo and T. Schücker
Volume 225 Aspects of Gravitational Interactions, Edited by S.K. Srivastava and K.P. Sinha
Volume 224 Dynamics of Transition Metals and Alloys, Edited by S. Prakash
Volume 223 Fields and Transients in Superhigh Pulse Current Devices Edited by G.A. Shneerson
Volume 222 Gaussian Beams and Optical Resonators, Edited by A. N. Oraevsky
Volume 221 Laser Cathode-Ray Tubes, Edited by Yu. M. Popov
Volume 220 Quantum Electrodynamics with Unstable Vacuum, Edited by V.L. Ginzburg
Volume 219 The Muon Method in Science, Edited by V.P. Smilga and Yu.M. Belousov
Volume 218 Radiative and Collisional Characteristics of Ions in Hot Plasmas, Edited by I.I. Sobel'man
Volume 217 Nonlinear and Quantum Optical Phenomena in Nonequilibrium Media ,
Edited by V.A. Shcheglov
Volume 216 Semiconductors and Insulators: Optical and Spectroscopic …, Edited by Yu.I. Koptev
Volume 215 Relativistic Theory of Gravity, Edited by A.A. Logunov
Volume 214 Research in Quantum Field Theory, Edited by V.I. Man’ko
Volume 213 Nonlinear Theory of Strong Electromagnetic Wave-plasma…, Edited by O.N. Krokhin
Volume 212 Dynamics of Elementary Atomic-Molecular…, Edited by O. S. Scheglov
Volume 211 Optics and Lasers, Edited by G.G. Petrash
Volume 210 Theory and Technology of High Temperature Superconductors, Edited by Yu.I. Koptev
Volume 209 Theory of Interaction of Multilevel Systems…, Edited by V.I. Man’ko
Volume 208 Physical Properties of High Temperature Superconductors, Edited by Yu.I. Koptev
Volume 207 Generalized Mayer Series in Classical Statistical Mechanics, Edited by Igor J. Ivanchik
Volume 206 Mathematical Physics, Applied Mathematics and Informatics, Edited by Yu.I. Koptev
Volume 205 Squeezed and Correlated States of Quantum Systems, Edited by M.A. Markov
Volume 204 Sakharov Memorial Lectures in Physics, Edited by L.V. Keldysh and V.Ya. Fainberg
Volume 203 Fiber Optics: Research and Development, Edited by Ye.M. Dianov
Volume 202 Gas Dynamics, Edited by Yu.I. Koptev
Volume 201 Generation of Nonlinear Waves and Quasistationary …, Edited by L.M. Kovrizhnykh
Volume 200 Laser Thermonuclear Targets and Superdurable Microballoons, Edited by A.I. Isakov
Volume 199 Spatiotemporal Characteristics of Laser Emission, Edited by M.V. Pyatakhin and A.F. Suchkov
Volume 198 Theory of Nonstationary Quantum Oscillators, Edited by M.A. Markov
Volume 197 Reliability Problems of Semiconductor Lasers, P.G. Eliseev
Volume 196 Pulsars, Edited by A.D. Kuzmin
Volume 195 Physical Characteristics and Critical Temperature of High Temperature Superconductors,
Edited by M.M. Sushchinskiy
Volume 194 Biophysical Approach to Complex Biological Phenomena, Edited by E. Volkov
Volume 193 Research on Chemical Lasers Edited by A.N. Orayevskiy
Volume 192 Atomic and Ionic Spectra and Elementary Processes in Plasma , Edited by I.I. Sobelman
Volume 191 Polarization of the Vacuum and a Quantum Relativistic Gas …, A.Ye. Shabad
Volume 190, Superconductors with A15 Lattice and Bridge Contacts Based …,
(Supp. Vol.) Edited by M. M. Sushchinskiy
Volume 190 Metal-Optics and Superconductivity, Edited by A.I. Golovashkin
Volume 189 Clusters of Galaxies and Extragalactic Radio Sources, Edited by A.D. Kuz'min
Volume 188 Integral Systems, Solid State Physics and Theory of Phase Transitions. Pt. 2. Symm. and Alg.
Structures in Physics, Edited by V.V. Dodonov and V.I. Man'ko
Volume 187 Quantum Field Theory, Quantum Mechanics and Quantum Optics: Pt. 1. Symm. and Alg.
Structures in Physics, Edited by V.V. Dodonov and V.I. Man'ko
Volume 186 Photoproduction of Pions on Nucleons and Nuclei, Edited by A.A. Komar
Volume 185 Injection Lasers in Optical Communication and Information Processing Systems,
Edited by Yu. M. Popov
Volume 184 Electron Processes in MIS-Structure Memories, Edited by A.F. Plotnikov
Volume 183 Invariants and the Evolution of Nonstationary Quantum Systems, Edited by M.A. Markov
Volume 182 Luminescence of Wideband Semiconductors, Edited by M.D. Galanin
Volume 181 Pulsed Metal and Metal Halogenide Vapor Lasers, Edited by G.G. Petrash
Volume 180 Inelastic Light Scattering in Crystals, Edited by M.M. Sushchinskiy
Volume 179, Electron-Excited Molecules in Nonequilibrium Plasma,
(Supp. Vol. 2) Edited by N.N. Sobolev
Volume 179, Interaction of Ultrashort Pulses with Matter ,
(Supp. Vol. 1) Edited by M.D. Galanin
Volume 179 X-Ray Plasma Spectroscopy and the Properties of Multiply-Charged Ions,
Edited by I.I. Sobel'man
Volume 178 The Delfin Laser Thermonuclear Installation: Operational Complex …, Edited by G.V. Sklizkov
Volume 177 Stoichiometry in Crystal Compounds and Its Influence on Their Physical Properties,
Edited by Yu. M. Popov
Volume 176, Quantization, Gravitation and Group Methods in Physics
(Supp. Vol) Edited by A.A. Komar
Volume 176 Classical and Quantum Effects in Electrodynamics Edited by A.A. Komar
Volume 175, Luminescence and Anisotropy of Zinc Sulfide Crystals
(Supp. Vol.) Edited by M.D. Galanin
Volume 175 Luminescence Centers of Rare Earth Ions in Crystal Phosphors, Edited by M.D. Galanin
Volume 174, Electron Liquid Theory of Normal Metals
(Supp. Vol. 2) Edited by V.P. Silin
Volume 174, Nonequilibrium Superconductivity,
(Supp. Vol. 1) Edited by V.L. Ginzburg
Volume 174 Thermodynamics and Electrodynamics of Superconductors, Edited by V.L. Ginzburg
Volume 173 Quantum Mechanics and Statistical Methods , Edited by M.M. Sushchinskiy
Volume 172 Phase Conjugation of Laser Emission , Edited by N.G. Basov
Volume 171 Research on Laser Theory , Edited by A.N. Orayevskiy
Volume 170 The Theory of Target Compression by Longwave Laser Emission , Edited by G.V. Sklizkov
Volume 169 The Physical Effects in the Gravitational Field Black Holes, Edited by M.A. Markov
Volume 168 Issues in Intense-Field Quantum Electrodynamics, Edited by V.L. Ginzburg
Volume 167 Group Theory, Gravitation and Elementary Particle Physics, Edited by A.A. Komar
Volume 166 The Nonlinear Optics of Semiconductor Lasers, Edited by N.G. Basov
Volume 165 Solitons and Instantons, Operator Quantization, Edited by V.L. Ginzburg
APPLIED PHYSICS IN THE 21ST

CENTURY, (HORIZONS IN WORLD
PHYSICS, VOLUME 266)
RAYMOND P. VALENCIA
EDITOR
Nova Science Publishers, Inc.

New York
Copyright © 2010 by Nova Science Publishers, Inc.
All rights reserved. No part of this book may be reproduced, stored in a retrieval system or
transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical
photocopying, recording or otherwise without the written permission of the Publisher.
For permission to use material from this book please contact us:
Telephone 631-231-7269; Fax 631-231-8175
Web Site: http://www.novapublishers.com
NOTICE TO THE READER

The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or
implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of
information contained in this book. The Publisher shall not be liable for any special,
consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or
reliance upon, this material.
Independent verification should be sought for any data, advice or recommendations contained in
this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage
to persons or property arising from any methods, products, instructions, ideas or otherwise
contained in this publication.
This publication is designed to provide accurate and authoritative information with regard to the
subject matter covered herein. It is sold with the clear understanding that the Publisher is not
engaged in rendering legal or any other professional services. If legal or any other expert
assistance is required, the services of a competent person should be sought. FROM A
DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE
AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS.
LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA
Available upon request.
ISBN: 978-1-61728-139-6 (E-Book)
Published by Nova Science Publishers, Inc.

  New York
CONTENTS
Preface ix
Chapter 1 Emerging Concepts and Challenges in Nano
Metal Oxide Thin Films 1
A. Subrahmanyam,, T. P. J. Ramesh,
A. Karuppasamy, U. K. Barik, K. Jagadeesh Kumar,
N. Ravichandra Raju and R. V. Muniswami Naidu
Chapter 2 Spectroscopic Analysis of Chemical Species in Carbon
Plasmas Induced by High-Power IR CO2 Laser 63
J. J. Camacho, J. M. L. Poyato,
L. Díaz and M. Santos
Chapter 3 Induction Transformer Coupled Discharges:
Investigation and Application 113
I.M. Ulanov and M.V. Isupov
Chapter 4 Features on the High Frequency Dielectric
Response in Ferroelectric Materials 169
J. D. S. Guerra
Chapter 5 The Principle that Generates Configuration in
Animate and Inanimate Systems – A Unified View 195
Antonio F. Miguel
Chapter 6 Computational Studies on Drag Reduction
Effect by Surface Grooves 217
Haosheng Chen and Yongjian Li
Chapter 7 Current Limiting in Oxide Ceramic Structures 245
Alexander Bondarchuk
Chapter 8 Characterisation of Silicide Thin Films for
Semiconductor and Nanotechnology Electronics 273
Madhu Bhaskaran, Sharath Sriram
and David R. G. Mitchell
viii Contents
Chapter 9 Magnetically Modified Biological Materials as Perspective

Adsorbents for Large-Scale Magnetic Separation Processes 301
Ewa Mosiniewicz-Szablewska, Mirka Safarikova,
and Ivo Safarik
Chapter 10 Optimisation of Deposition Conditions for
Functional Oxide Thin Films 319
Sharath Sriram and Madhu Bhaskaran
Chapter 11 Optical Waveguides Produced by Ion
Implantation in Oxide Glasses 341
Feng Chen, Xue-Lin Wang,
Lei Wang and Ke-Ming Wang
Chapter 12 Thin Film Piezoelectric Response
Coefficient Estimation Techniques 353
Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell
Chapter 13 Plasma Technology: An Alternative to Conventional
Chemical Processes for Hydrogen Production 365
María Dolores Calzada
Chapter 14 Target-Plasma-Film Interactions in High
Power Pulsed Magnetron Sputtering (HPPMS) 373
K. Sarakinos
Index 391
PREFACE
Applied physics is rooted in the fundamental truths and basic concepts of the physical
sciences but is concerned with the utilization of these scientific principles in practical devices
and systems. This new and important book gathers the latest research from around the globe
in this dynamic field.
Chapter 1 - The metal oxides thin films have created a revolution in technology. The
nano metal oxide thin films continue in giving a new dimension to the understanding pushing
the frontiers of science with new concepts and advanced applications. The present article is an
attempt to compile the emerging concepts and the novel applications in four metal oxide
systems; the article also gives the details of the complexities in the understanding of the
physical phenomena. The four nano metal oxide systems coverd in this article are : (i) nano
titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices, (ii) nano Silver
oxide thin film fluorescence in non-volatile optical memories, (iii) Nano Tungsten oxide thin
film electro-chromics in efficient and smart windows and (iv) the p- type transparent
conducting oxides (TCO) thin films for transparent electronics with a special reference to
Zinc oxide thin films. It is interesting to note that the materials described in the four systems
even though possessing a good knowledge base, still offer stiff challenges in the basic
understanding. The emphasis in the present article is laid on the new applications, the related
physics and chemistry. The preparation methods described are limited to those which have a
scope for up-scaling (and for possible industrial adoption). This article is not a review of the
state of the art for these metal oxide systems. The aim of this article is to introduce the new
and emerging applications with the existing concepts.
Chapter 2 - This chapter describes some fundamentals of laser-induced breakdown
spectroscopy (LIBS) and experimental results obtained from ultraviolet-visible-near infrared
(UV-Vis-NIR) spectra induced by laser ablation of a graphite target, developed in our
laboratory. Ablation was produced by a high-power IR CO2 pulsed laser using several
wavelengths (λ=9.621 and 10.591 µm), power density ranging from 0.22 to 6.31 GW cm-2
and medium-vacuum conditions (typically at 4 Pa). Spatially and time resolved analysis were
carried out for the plasma plume. Wavelength-dispersed spectra of the plume reveal the
emission of C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and molecular features of C2, CN, OH, CH,
N2, N2+ and NH. For the assignment of molecular bands a comparison with conventional
emission sources was made. The characteristics of the spectral emission intensities from the
different species have been investigated as functions of the ambient pressure, laser irradiance,
delay time, and distance from the target. Excitation, vibrational and rotational temperatures,
x Raymond P. Valencia
ionization degree and electron number density for some species were estimated. Time-gated
spectroscopic studies have allowed estimation of time-of-flight (TOF) and propagation
velocities for various emission species.
Chapter 3 - Researches in the field of low-temperature plasma provide development of
devices, which are widely used by the advanced and high-tech industries. Low-temperature
plasma is applied by such major industries as microelectronics, semiconductor industry, solar
cell production, plasma chemistry, metallurgy, lighting engineering, etc. Among the known
methods of production and use of low-temperature plasma (DC and AC arc discharges, RF
plasmatrons, microwave plasmatrons) the devices based on application of induction
transformer coupled toroidal discharges (TCTD) are the least studied and, thus, rarely used.
Simultaneously, these discharges can be used for development of electrodeless generators of
low-temperature plasma: transformer plasmatrons and new induction sources of light.
The chapter deals with investigation of properties of TCTD, transformer plasmatrons and
induction light sources on the basis of TCTD. Investigation results on electrophysical
properties of TCTD aimed at development of transformer plasmatrons are presented in the
current paper. Dependences between the strengths of TCTD electric field, discharge current
and gas flow are obtained for different gases within the pressure range of 10÷105 Pa. The
thermophysical characteristics of TCTD were determined: device efficiency, energy balance
of a discharge (heat losses to the discharge chamber wall, plasma jet power). The stable
TCTD of the atmospheric pressure in argon and in air was firstly obtained and studied by the
authors of this chapter. The process of plasmachemical synthesis of NO in air plasma of
TCTD was studied. The abnormally high percentage of NO ~7 % was obtained without
product quenching. The transformer plasmatrons of the 10÷200 kW power, operating under
the pressures of 10÷105 Pa on argon, air and argon+hydrogen, argon+oxygen mixtures, were
developed on the basis of the studies performed. The schemes and constructions of these
plasmatrons are presented.
The electrophysical and spectral characteristics of TCTD were studied in vapors of
mercury and neon. The electrodeless sources of visible and UV radiation with the power of
100 W ÷ 100 kW were developed on the basis of this research.
Chapter 4 - Single crystal and/or polycrystalline ferroelectric materials show a high
frequency dielectric dispersion, which has been attributed as well to a dispersive (relaxation
like) as a resonant mechanism. Physical properties such as relaxation and/or resonant motion
mechanisms can be investigated by analyzing the complex dielectric permittivity (real, ε’ and
imaginary component, ε’’) in a broad spectral frequency range (100 MHz–13 GHz).
Especially, for classical (or ‘normal’) and relaxor ferroelectric systems a dielectric response
indistinguishable of dispersion or a resonance mechanism has been found in the literature.
The occurrence of such common dispersion process in so different kinds of ferroelectric
systems has encouraged the development of several mutually excluding models to explain this
physical phenomenon. Nevertheless, the reported results are not conclusive enough to clearly
distinguish each mechanism. In this work, a detailed study of the dielectric dispersion
phenomenon, including the microwave frequencies, carried out in perovskite structure-type
ferroelectric systems, for ‘normal’ and/or relaxor compositions, is presented. The dielectric
response in “virgin” and poled state have been investigated taking into account the relative
direction between the measuring direction and the orientation of the macroscopic polarization.
Results revealed that the dielectric response in ferroelectric systems may be described as a
Preface xi
general mechanism related to an “over-damped” resonant process rather than a simple

relaxation-like dielectric behavior.
Chapter 5 - The generation of flow configuration (shape, structure, patterns) is a
phenomenon that occurs across the board, in animate and inanimate flow systems. Scientists
have struggled to understand the origins of this phenomenon. What determines the geometry
of natural flow systems? Is geometry a characteristic of natural flow systems? Are they
following the rule of any law? Here we show that the emergence of configuration in animate
flow systems is analogous to the emergence of configuration in inanimate flow systems, and
that features can be put on a unifying theoretical (physics) basis, which is provided by the
constructal law.
All scientific endeavors are based on the existence of universality, which manifests itself
in diverse ways. Here the authors also explore the idea that complex flow systems with
similar functions have a propensity to exhibit similar behavior. Based on this thought
relationships that characterize animate systems are tested in relation to cities and countries,
and some conclusions are drawn.
Chapter 6 - The drag reduction effect caused by the periodic surface grooves were studied
using computational fluid dynamic method in two different flow conditions: laminar flow in a
slide-disk interface, and turbulent flow on the grooved surface immerged in water.
In the first part, the drag results by Reynolds equation that is commonly used in
lubrication calculation is compared with the CFD result based on Navier-Stokes equation. It
was validated that the Reynolds equation is not suitable when the groove depth is higher than
10% of the interface distance. Then, the drag forces on the surface with transverse and
longitudinal grooves are calculated using CFD method. It was found that the ‘side wall’ effect
causes the drag reduction, which means the drag reduction would appear when the loss of the
drag on the groove’s bottom can not be compensated by the pressure drag or the viscous drag
on the side walls of the grooves.
In the second part, the drag force on the transverse rectangular grooves in turbulent flow
is analyzed based on the RANS equations coupled with the RNG k-ε turbulent model. It was
found that the pressure drag force makes up a large proportion of the total drag, and the
turbulent vortex structure in the grooves affects the drag characteristic of the surface. The
‘side wall’ effect also functions in turbulent flows, and the drag reduction is determined by
the synthesized effect of the reduction of viscous drag and the increment of pressure drag.
Chapter 7 - The review of physical phenomena which lead to current limiting behavior
(current is increased weaker than voltage, saturated and even decrease) in oxide ceramic
structures are presented. Particular attention is given to the mechanisms of current limiting in
materials whose electrical conductivity is controlled by potential barriers at the grain
boundaries. In particular, the current saturation effect in the nano-grained ceramic films is
examined. The consideration of physical models describing the current limiting in specific
material is followed by the short posing of main experimental data. For some models, the
computational modeling has been performed.
Chapter 8 - Nanotechnology devices require low resistance contacts, which can be
fabricated by the incorporation of silicide thin films This chapter discusses in detail the study
of silicide thin films using a suite of materials characterisation tools. The silicides of interest
in this study were titanium silicide (TiSi2) and nickel silicide (NiSi), given their low
resistivity and low barrier heights to both n-type and p-type silicon. The silicide thin films
were formed by vacuum annealing metal thin films on silicon substrates. Silicide thin films
xii Raymond P. Valencia
formed from metal films deposited by DC magnetron sputtering and electron beam
evaporation were compared. The composition, crystallographic orientation, and morphology
of these thin films were studied using spectroscopy (AES, SIMS, RBS, in situ Raman
spectroscopy), diffraction (Bragg-Brentano and glancing angle XRD, RHEED), and
microscopy techniques (TEM, SEM, and AFM).
Chapter 9 - Novel magnetically modified biological materials, containing magnetic iron
oxides nanoparticles as labels, have been successfully developed and applied as magnetic
affinity adsorbents for the magnetic separation of various biologically active compounds and
xenobiotics.
The main attention was focused on cheap and easy to get magnetic adsorbents which
could be applied for large-scale processes. Among them magnetically modified plant-based
materials (sawdust) and microbial cells (yeast and algae) were taken into consideration. An
inexpensive, extremely simple procedure was proposed for the preparation of such magnetic
adsorbents using standard water-based ferrofluids containing maghemite nanoparticles with
the diameter of about 12 nm. Such ferrofluids can be prepared in a simple way (almost in any
lab) and such nanoparticles can be used to prepare biocomposite materials enabling their
simple magnetic separation with standard permanent magnets. Both of these properties are
important for possible large-scale applications.
The structural, adsorption and magnetic properties of the developed materials were
studied in detail by means of scanning electron microscopy, transmission electron
microscopy, spectrophotometric measurements, ESR spectroscopy and conventional magnetic
methods (DC magnetization and AC susceptibility measurements). The prepared materials
efficiently adsorbed selected biologically active compounds and xenobiotics (mainly different
enzymes, water-soluble organic dyes and heavy metal ions). Their magnetic behavior was
dominated by the superparamagnetic relaxation of isolated single domain maghemite
nanoparticles, although a little amount of agglomerates was also present. However, these
agglomerates were sufficiently small to show at static conditions the superparamagnetic
behavior at room temperature which allows to use the developed materials as magnetic
adsorbents in the magnetic separation techniques. Moreover, the prepared materials exhibit
the peculiar features enabling their rapid and efficient removal not only from solutions, but
also from suspensions. Such materials could be efficiently used to isolate rare biologically
active compounds from difficult-to-handle materials including raw extracts, blood and other
body fluids, cultivation media, environmental samples, etc.
Inexpensive raw materials, extremely simple preparation method, affinity to various
biologically active compounds and both organic and inorganic xenobiotics, and distinctive
magnetic properties make the developed materials greatly suitable as magnetic adsorbents for
large-scale magnetic separation processes.
Chapter 10 - This chapter describes a systematic approach to determining the optimal
conditions required to deposit thin films of complex functional oxides using RF magnetron
sputtering. The motivation of this study was to attain films of designed stoichiometry and of
preferentially oriented perovskite crystal structure, as the composition and structure of
complex oxides determines their functionality (e.g. ferroelectricity, piezoelectricity, etc.)
which is exploited by a variety of applications. The process has been demonstrated with
strontium-doped lead zirconate titanate (PSZT) thin films using a combination of materials
characterisation techniques – X-ray diffraction, X-ray photoelectron spectroscopy, atomic
force microscopy, and crystal structure calculations. The major variables in the deposition
Preface xiii
process using RF magnetron sputtering were identified as the deposition substrate

temperature, post-deposition thermal processing, oxygen partial pressure during deposition,
and the choice of metallisation on silicon substrates. Starting with these variables the
influence of each was systematically determined. The results highlighted the advantages of
slow cooling to promote perovskite growth, the influence of oxygen on the composition and
crystal structure of the thin films, and the presence of modified unit cell structure for the
PSZT thin films on gold- and platinum-coated silicon substrates. The combination of these
results led to the limiting the variables in deposition to finite values and arriving at two
definite sets of deposition conditions for deposition based on the dependence of
microstructure on thermal conditions and the choice of substrate. The validity of the
conditions chosen is then demonstrated by deposition of PSZT thin films on thermally-grown
silicon dioxide and attaining nanocolumnar preferentially oriented thin films.
Chapter 11 - Optical waveguides can restrict light propagation in very small size of order
of several microns, reaching high optical intensities even at low pumps; consequently, some
properties of the bulk materials may be considerably improved in waveguide structures, such
as non-linear responses, laser actions and optical signal amplifications. As one mature
technique for material property modifications, ion implantation has been applied to construct
optical waveguides in many optical materials, including insulating crystals and glasses,
semiconductors and organic materials. Oxide glasses receive much attention for various
photonic and telecommunication applications for its competitive costs and excellent optical
features. In this chapter, the authors reviewed the research results obtained for optical
waveguides in oxide glasses produced by ion implantation techniques, by giving a brief
introduction of basic fabrication principles and methods and a summary of interesting results
obtained in this topic. The prospects of possible practical applications of ion-implanted oxide
glass waveguides in photonics are also demonstrated.
Chapter 12 - The response of piezoelectric materials is quantified using charge and
voltage coefficients. One such coefficient is d33, which numerically describes the resulting
effect for an applied cause along the same direction. There is a lack of established techniques
for quantitative estimation of d33 for piezoelectric thin films. This initiated an investigation
into the development of such techniques, as a consequence of which two new techniques for
piezoelectric coefficient estimation, under the inverse piezoelectric effect, have been
developed. One technique capitalises on the measurement accuracy of the nanoindenter in
following thin film displacement, while the other uses a standard atomic force microscope in
contact imaging mode to estimate d33. Both techniques were developed by rigorously testing
them against standard materials and avoiding commonly reported sources of error. Full details
on the development, scope, and limitations of both techniques are presented in this chapter.
Chapter 13 - There is currently a strong debate on the need to look for new alternative
energy sources to replace fossil fuels (particularly oil), due to atmospheric pollution derived
from their use, probable reserve exhaustion and productive countries’ excessive dependence
on future political events. Therefore, many countries are currently giving an incentive to
research aimed at the development of new energy sources, among which hydrogen —and its
use by means of fuel-cells— is to be included. Within the so-called hydrogen economy,
hydrogen production —from different raw materials such as hydrocarbons, alcohols and some
others— becomes an important aspect, since hydrogen is not a natural product but is obtained
by means of water vapour. However, such process leads to the production of high amounts of
CO2. The use of plasma technology allows generating hydrogen by decomposing organic
xiv Raymond P. Valencia
compounds (alcohols and methane), so plasma acts as a reactive environment; this way,
biogas —which can be easily found in city landfills— can also be used for hydrogen
obtention. Thus, the use of plasma with this purpose arises as an alternative process not only
to reforming of hydrocarbons and other substances with water vapour but also to other purely
chemical processes, avoiding or minimizing CO2 emissions (involving greenhouse effects). A
subproduct —obtained with H2 and which also deserves special attention— is solid carbon,
which has important added value, since it may work as a base element for chemical industry
and other technological fields. Given the foregoing, research in the use of plasma technology
for hydrogen production is one of the research lines within Applied Physics with higher
applicability potential within the energy sector.
Chapter 14 - Growth of films by plasma-assisted physical vapor deposition (PVD)
techniques provides means for tailoring their properties and improving their functionality in
technological applications. State of the art plasma-assisted PVD techniques, like direct
current magnetron sputtering (dcMS), suffer from low degree of ionization and thus, low ion-
to-neutral ratios in the flux of the deposited material. The implementation of external sources
for the enhancement of the ionization is in many cases technically complicated and increases
the end-product cost. High power pulsed magnetron sputtering (HPPMS) is a novel sputtering
technique that elegantly enables the conversion of conventional sputtering source into an ion
source. By applying the power in unipolar pulses of low frequency (<2 kHz) and low duty
cycle (<10%) high values of peak target current density up to ~ 5 Acm-2 are achieved. This
leads to the generation of ultra dense plasmas with electron densities of ~1019 m-3. These
values are up to four orders of magnitude higher than those obtained by dcMS and result in a
high degree of ionization of the sputtered material (up to ~90%). Moreover, due to the low
duty cycles, the average target current densities are maintained at values of ~0.5 Acm-2 which
are comparable to those in dcMS processes. Therefore, HPPMS allows for highly ionized
deposition fluxes without excessive thermal load at the target.
In the current short communication the target-plasma interactions and their effect on the
deposition rate in non-reactive and reactive HPPMS are discussed. Furthermore, the
interactions of the plasma species in HPPMS with the growing film and their implications on
the microstructure and the phase formation of metal nitride and oxide films are demonstrated.
In: Applied Physics in the 21st Century… ISBN: 978-1-60876-074-9
Editor: Raymond P. Valencia, pp. 1-62 © 2010 Nova Science Publishers, Inc.
Chapter 1
EMERGING CONCEPTS AND CHALLENGES IN NANO

METAL OXIDE THIN FILMS
A. Subrahmanyam1,2, T. P. J. Ramesh3, A. Karuppasamy4,

U. K. Barik5, K. Jagadeesh Kumar2, N. Ravichandra Raju2
and R. V. Muniswami Naidu2
1
DAAD Professor, Department of Electrical and Computer Engineering
Technical University, 01062 Dresden, Germany
2
Department of Physics, Indian Institute of Technology Madras, Chennai 600036, India
3
Department of Cardiothoracic Surgery, Apollo Hospital, Chennai, India
4
Fraunhofer Institut für Elektronenstrahl und Plasmatechnik (FEP),
Winterbergstraße 28, 01277 Dresden, Germany.
5
Samtel, Ghaziabad, India
SUMMARY OF THE ARTICLE

The metal oxides thin films have created a revolution in technology. The nano metal
oxide thin films continue in giving a new dimension to the understanding pushing the
frontiers of science with new concepts and advanced applications. The present article is
an attempt to compile the emerging concepts and the novel applications in four metal
oxide systems; the article also gives the details of the complexities in the understanding
of the physical phenomena. The four nano metal oxide systems coverd in this article are :
(i) nano titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices,
(ii) nano Silver oxide thin film fluorescence in non-volatile optical memories, (iii) Nano
Tungsten oxide thin film electro-chromics in efficient and smart windows and (iv) the p-
type transparent conducting oxides (TCO) thin films for transparent electronics with a
special reference to Zinc oxide thin films. It is interesting to note that the materials
described in the four systems even though possessing a good knowledge base, still offer
stiff challenges in the basic understanding. The emphasis in the present article is laid on
the new applications, the related physics and chemistry. The preparation methods
1
Corresponding author: email : manu@iitm.ac.in
2 A. Subrahmanyam, T. P. J. Ramesh, A. Karuppasamy et al.
described are limited to those which have a scope for up-scaling (and for possible
industrial adoption). This article is not a review of the state of the art for these metal
oxide systems. The aim of this article is to introduce the new and emerging applications
with the existing concepts.
A. INTRODUCTION TO METAL OXIDES

The metal oxide thin film is a wide subject where one can find a vast amount of
literature: books, monographs, review articles and innumerable number of research papers
[A1-A3]. The metal oxides have challenges encompassing the scientist (on basic issues and
understanding), the engineer (for possible applications and application oriented problems) and
the technologist (for the cost effective process developments and up-scaling). The metal
oxides, including the nano-metal oxides, have a variety of applications almost spanning the
entire gamut of strategic to the day to day devices and absolutely mundane to novel functional
materials and devices. To name a few areas of the metal oxides are: microelectronics, non-
volatile memories, spintronic devices, bio-medical devices, sensors, piezoelectric devices,
magnetic devices, super conducting magnets and magnetic materials, thermal barrier and anti-
corrosion coatings, heterogeneous catalysis, energy conversion and storage devices etc. The
metal oxide is presently a few Billion dollars industry with a consistent and rapid growth.
Most metals are oxidized immediately when exposed to the ambient. One can also grow metal
oxides with controlled oxygen both on the surface and in the bulk. Thus, one may address the
understanding of metal oxides from the surface and bulk points of view: each having its
implications, complexities and applications. Most of the applications, the bulk control the surface
properties. The nature and strength of the bond between a metal and the oxygen varies with the
constituent metal and the preparation or growth conditions; the oxidation process is quite complex.
A number of different bonding mechanisms between the metal and the oxygen have been
reported, including the weak van der Waals dispersion forces between induced surface dipoles,
electrostatic polarization binding, and strong ionic and covalent bonding (the latter being more
directional). There are several questions still to be answered in the understanding of the oxidation
mechanism and oxidation kinetics in metal oxides.
The metal oxides can adopt a vast number of structural geometries and exhibit insulating,
semiconducting, metallic and even superconducting behavior. While analysing the magnetic
properties, metal oxides made a significant progress. Recent years have witnessed a
tremendous growth of research in the field of magneto-electronics, in view of its obvious
potential for novel devices with entirely new capabilities. In this context, phenomena such as
giant magneto-resistance (GMR), colossal magneto-resistance (CMR), spin-tunneling in
junctions (STJ) have attracted significant attention [A4]. The oxides of transition metals and
their interfaces with metals and semiconductors represent one of the most exciting and
challenging areas currently studied in magnetism.
The surface science of metal oxides is a relatively very young filed and there are a lot of
challenges and opportunities. Single metal atoms tend to form relatively strong chemical
bonds to most oxide surfaces. This is mainly the result of a severe under-coordination of the
metal atom (compared to its favored bulk state), which is therefore reactive with the surface.
These ionic or covalent metal–oxide bonds are correspondingly quite short, normally around
1–2 Å. Typical bond strengths range from 1 to several eV, and are normally referred to as
Emerging Concepts and Challenges in Nano Metal Oxide Thin Films 3
adsorption or desorption energies—defined as the energy required to remove a metal atom to

the vacuum level. The metal–oxide bonds tend to further strengthen at various oxide surface
defects, e.g., anion/cation vacancies, which can trap and immobilize diffusing metal atoms.
Many of the properties of metal oxides are related to the surface and interface structure. A
microscopic understanding at the atomistic level of the oxide surfaces and interfaces is of
fundamental importance for the development of new materials, nano-structures and novel
devices. In the emerging field of nanotechnology, one of the goals is to make the
nanostructures with special or tailored properties with respect to those of the bulk materials.
The oxide nanostructures or nanoparticles can exhibit unique properties owing to their limited
size and high density of low-coordinated sites or can provide new extended systems with
unprecedented structure, morphology and/or composition. In order to facilitate tailored
nanoparticle production, the study of the largely unknown mechanisms of metal oxide
nucleation and subsequent growth (which mainly depend upon the surface energy adhesion
kinetics) is of pivotal importance. A key aspect in this respect is the control of defects and
impurities, a field closely related to the defect engineering of materials.
It is rather difficult to present a concise review of the metal oxide thin films. With all these
un-limitations, the present paper attempts to describe the emerging concepts in four metal oxide
thin film systems: (i) titanium, (ii) silver, (iii) tungsten and (iv) zinc. These four thin film systems
have vast applications and pose a variety of complexities; the present article confines to:
(i) nano Titanium oxide thin film photo catalysis in bio-medical (lung assisted) devices,
(ii) nano Silver oxide thin film fluorescence in non-volatile optical memories,
(iii) Nano Tungsten oxide thin film electro-chromics in efficient and smart windows and
electro-chromic displays and memories and
(iv) nano Zinc oxide for p- type transparent conducting oxide (TCO) thin film for
transparent electronics.
This paper is a collection of ideas, and truly emerging, with potential applications. A vast
amount of work is needed to understand these metal oxide systems before one realizes these
devices. The aim of this paper is to introduce the problem from its fundamentals. The fond
hope of the authors is to invite the researchers towards the challenges posed by these four
oxide systems, ultimately leading to a better understanding of the phenomena and to develop
new technologies with new and novel devices.
1. TITANIUM OXIDE THIN FILM FOR LUNG ASSISTED DEVICES

1.1. Summary
Titanium oxide is a well known photocatalyst. One of the important applications of the
photocatalysts is water splitting for generation of alternate fuel : hydrogen. The same concept
is being used in oxygenating human blood. The water available in the blood can be the source
of oxygen. Since this oxygen is generated in the immediate vicinity of the haemoglobin, this
seems to be a most efficient method of oxygenating human blood. The efficacy levels are
quite encouraging.
1.2. Introduction
Human health care is one among the top priority in materials and bio-medical research.
Nano materials, in particular, the nano metal oxides is the focus of attention for several
advanced applications. Among the several ailments, respiratory failure is one of the leading
causes of both mortality and morbidity world wide. With ever increasing environmental
pollution, the first and most important organ being affected and afflicted is the lungs.
Advances in the management of end stage lung disease have not paralled the management of
end stage heart, liver or renal disease.
Lungs is the organ which supplies the oxygen to the blood and removes carbon dioxide
from the blood. The oxygen and carbon dioxide are attached to the Haem group of the red
blood cells which is the transport carrier. Any lung damage severely affects the oxygen
infusion to the haem group. The development of lung assisted devices so far are based on the
concept of gas diffusion through hollow fibre membranes; this technology is constrained for
long term usage. There are excellent reviews on the efforts on lung assisted devices [1.1]. The
primary function of any lung assisted device (either in vitro or in vivo) is to oxygenate the
blood and to remove the carbon dioxide.
An emerging concept in the lung assisted devices is to split the water (by photocatalytic
process) available in the blood to produce oxygen; since this oxygen is generated in the close
proximity of the haemoglobin, the oxygenation process is very effective and efficient. Also,
the photocatalysts are made of metal oxides, they are bio-compatible and a long term usage is
feasible with almost little maintenance. The results from the preliminary experiments are very
encouraging. With this application, several millions of population suffering from lung
ailments are anticipated to get relief in a most cost effective way.
1.3. Introduction to Photo-Catalysis
Photocatalysis is an established phenomenon shown by several metal oxide thin films:

oxides of titanium, tungsten, vanadium, zinc and iron etc. This oxidation photocatalytic
phenomenon is widely being used for addressing the problems of environmental pollution.
The titanium dioxide (TiO2) is studied rather extensively and is ascertained to be bio-
compatible and it is an efficient photocatalyst.
Titanium metal has been proven to be a bio-compatible material and a large number of
implants and substitutes are being manufactured on commercial basis [1.2] Titanium is a
highly reactive metal which forms the oxides and the oxidation kinetics are extremely fast, of
the order of a few nano-seconds. Thus titanium is always covered with a thin oxidizing layer
all the time (unless one works in ultra high vacuum).
Titanium dioxide is a stable compound which is resistant to chemical attack (this property
should not be confused with the chemical activity) and it is corrosion resistant. From among
the several applications of titanium dioxide, the photocatalytic activity is a field of great
technological and industrial importance. The initial work of Fujishima and Honda [1.3] on the
photocatalytic splitting of water with titania without an external bias has evoked considerable
interest among many researchers. Most of the efforts in the early days have been focussed to
generate hydrogen as a clean fuel, but, it is still far from realization. However, the water
splitting by photocatalysis, has opened up novel applications and new challenges; one of the
challenges is to understand the surface of titania. Photo active TiO2 surfaces react with
aqueous solutions and they will dissociate organic molecules, can oxidize water into oxygen
and hydrogen, they also show anti-bacterial activity. TiO2 is an age old (proven) industrial
material used as pigment in paints. High refractive index titania is widely used in optical
coatings. TiO2 is the semiconductor that facilitates charge transfer mechanism in dye
sensitized solar cells. Recently, the nano-titania is emerging as a self cleaning coating and
also as environmental protection and it is termed as green chemistry material. There are many
applications of pure, functionalized or doped titania films or nanoparticles [1.4] (e.g. titania).
Many of these applications have potential impact on our every-day life, like in solar cells, as
self-cleaning surface film on glass windows [1.5], as photocatalyst, including the removal of
toxic pollutants. There are excellent reviews on the photocatalytic property of titanium
dioxide [1.6]
The titanium dioxide (TiO2) is being explored for medical purposes only recently. The
nano-phase titanium dioxide to medicine or bio-medical engineering has been developed
during the past fifteen years. Compared to the bulk TiO2, the nano-TiO2 exhibits higher
photocatalytic efficiency leading to pronounced phototoxic effects on bacteria, cancer cells
and tumours.
TiO2 is a wide band gap semiconductor; when a mono-chromatic wavelength
corresponding to the band gap is incident, an electron – hole pair will be generated. The
electron moves to the conduction band and the hole to the valence band. If the two charges
can be separated from recombination or trapping, the electron reduces oxygen to O2- and the
hole oxidizes (the harmful carbon monoxide to harmless carbon dioxide). Both oxidation and
reduction cannot take place simultaneously as these chemical reactions are rate controlled.
The high efficiency of the photocatalytic action depends upon (i) the quantum yield of the
photogenerated carriers, (ii) the effective removal of one the photogenerated carriers, (iii) the
recombination or trapping centres and (iv) the interaction of the photogenerated carrier with
the lattice or surface defects.
TiO2 (Eg = 3.2 eV)

Band Gap O2 -
Conduction Band
hν(350nm) e- Reduction
O2
hν
h+ Oxidation
UV light Valence Band
CO2, H2O
Figure 1. Photocatalytic mechanism operative in Titanium oxide
It may be noted that the nano-TiO2 is more effective and efficient in photocatalysis than
the bulk because the nano TiO2 has no charge present inside the particle, but do have a strong
electric field, where as the larger particles contain a layer of charge underneath the surface at
a distance of the space charge. For the nano TiO2, large number of photogenerated charges
move on to the surface and thus more active centres exist on the surface. Moreover, due to
high surface to volume ratio and large interfacial surface area, photogenerated electrons do
not recombine easily with the photo generated holes. Thus the photocatalytic activity of nano-
TiO2 is much higher compared to that of the bulk TiO2.
1.4. Basics of Water Splitting by Photo-Catalysis
The concept of water splitting by photocatalysis [1.7] is being adopted in the

development of lung assisted devices. The photo generated holes in the presence of water can
oxidize water into oxygen and hydrogen. The technological problem is to remove the photo
generated electron before it combines with the hole. The catalytic process is a liquid / solid
interface phenomenon and is generally categorized as “heterogeneous photocatalysis”. The
quantum efficiency of the photocatalytic process (or more precisely, the electron – hole
generation) is quite low in TiO2. The main issues presently being focussed are the
development of visible wave length photocatalyst and enhancement of the photocatalytic
efficiency.
The basic reactions of TiO2 surface in the presence of light of suitable wavelength (that
generated electron – hole pairs) and water (or humidity) may be summarized as :
TiO 2 + hν → h + + e − (h + = hole vacated by e- )

h + + H 2 Oabs → OH + H +
h + + OH − abs → OH
e − + O 2 → O•−
2 (superoxide)
1.5. Oxygenation of Blood by Photo-Catalysis
There are two important issues in the oxygenation of blood: first,the supply of oxygen to
the haemoglobin and the second, the mechanism of intake of oxygen by the haem group. As is
well known, lungs are the organ that supply oxygen from atmosphere into the alveoli; the gas
exchange/ diffusion mechanism across the alveoli is facilitated by a thin membrane (the same
membrane is responsible for removing carbon dioxide from the blood). This membrane, if
damaged or clogged by the particles, the efficiency of the lungs decrease significantly. With
the ever increasing levels of pollution, the lung functions get diminished or deteriorated in its
full activity even in a healthy human. The lung diseases are the second largest among the
population in the world and very unfortunately the lung therapy has not progressed as much
as the heart diseases.
The basic reactions of the oxygen with the haemoglobin and the basic structure of
haemoglobin are given in Notes 1 at the end of this article.
Preliminary experiments conducted on the photocatalytic oxygenation of deoxygenated
human blood have given positive results [1.8 – 1.11]. There are also Patents available on the
subject [1.12 – 1.16]. The experiment basically consists of a photocatalytic cell (Figure 2)
consisting of the photocatalytic coating of pure or doped TiO2 on the outer wall of the inner
quartz tube; the inner tube is enclosed by a bigger diameter outer quartz tube with a provision
of inlet and out let for the blood. The UV light is incident on the photocatalytic coating while
the blood is flowing on the surface of the inner tube and the photocatalytic action takes place.
The oxygen deficient venous blood (looks blackish in color) is introduced and the
photocatalytic surface is activated with suitable light to generate the electron – hole pairs. The
oxidation mechanism of water available in the blood with the photogenerated holes produces
dissolved oxygen and this dissolved oxygen is attached to the haem group. The blood gases
are measured by a blood gas analyzer and also by an optical spectrophotometer.
One of the important and critical points of this concept is that the photocatalytic action
should produce only dissolved oxygen and not the reactive species oxygen.
The pH, Po2 (partial pressure of dissolved oxygen), PCo2 (partial pressure of dissolved
of carbon dioxide), Sao2 (percent saturation of haemoglobin) can be measured using a
conventional blood gas machine (I—Stat). The haemoglobin of the subject was measured
with routine lab analysis and found to be 16gm/dl. The increase in whole blood oxygen
content (CaO2) in mls of oxygen/100ml of blood is calculated using the formula [1.17]
[1.34 X Hb gm/dl X SaO2 (expressed as fraction)] + (0.003 X Po2 mmHg) (1)
Figure 2. Basic concept of a photolytic lung assisted device (in vitro)
The photocatalyst prepared in the form of thin films have oxygenated the human blood
with good efficacy (about 10 ml of oxygen in 100 ml of blood in 120 minutes with a surface
area of 14 cm2) as shown in the Figure 3 [1.9]. The oxygenation depends upon the
photocatalytic area and photocatalytic efficiency. One precaution to be exercised in prolonged
use of the photocatalytic process is that the photocatalyst should not undergo photocorrosion
or modification of the catalytic surface.
Since the oxygen is generated within the blood, the process is highly efficient and thus
this process does not demand large areas (as in the case of membrane oxygenators), means the
device can be really portable. The efficacy of oxygenation shows a very high promise to
develop into a small portable device which can come to the rescue of patients with various
pulmonary disorders. The important feature of this new concept is that the materials
employed are bio-compatible (nano- Titanium dioxide), the oxygenation is instantaneous,
quantum efficiency of oxygen production is simple and the device seems to have no
limitations on prolonged usage, and the UV light does not seem to damage any of the blood
constituents (like the white blood cells and the DNA).
The device based on this concept requires sensors that measure the oxygen, carbon
dioxide (attached to the haem group) and other blood constituents which need to be
developed. As a first step, to estimate the oxygen attached to the haem group, the optical
absorption measurements have been performed and the results show that the optical
techniques have the potential to be used as sensors/ monitors. More details on the optical
measurements of the blood can be found in the book: Visible and near infra red absorption
spectra of human and animal haemoglobin [1.18].
The dissolved oxygen generated by photocatalysis can also be measured by “oxy probes”
(M/s Ocean Optic Inc). The Oxy probe is based on the principle that when a fluorophor sees
dissolved oxygen, the fluorescence signal decreases correspondingly(quenching). Presently,
correlation is being attempted between the measurements on oxy-probes, optical techniques
and the blood gas machine results.
22
CaO2 (ml of oxygen per 100 ml of blood)
20
TEST 1
TEST 2
18 CONTROL 1
CONTROL 2
16
14
12
10
6
0 20 40 60 80 100 120
Time (min)
Figure 3. Photolytic oxygenation of whole human blood
1.6. Challenges in the Process of Oxygenation by Photocatalysis
The challenges of the present concept are to evaluate the efficacy of oxygenation of blood
(or equivalently, the measurement and quantification of oxygen produced), the effect of UV
light on the constituents of the blood, the quantification of the amount of water in the
participating photocatalytic reaction, any change in the viscosity and pH of the blood during
the oxygenation process etc. The challenge related to the photocatalytic surface is to
understand the photo-corrosion that may degrade the performance over long periods of usage.
One of the questions to be answered is what happens to the hydrogen produced (or hydrogen
is produced at all) and how to remove the carbon dioxide from the hem group. The
fundamental challenge is to set the protocols for the experiment.
1.80
414.59
414.61
1.6
1.4
414.55
1.2
414.57
1.0
A
0.8
0.6
414.35
0.4 576.56
576.51
0.2
0.00
200.0 250 300 350 400 450 500 550 600 650 700.0
nm
Figure 4. Optical absorption of photocatalytic oxygenated whole human blood
2. NANO SILVER – SILVER OXIDE THIN FILMS

2.1. Summary
Two important observations have been made in nano-silver oxide: (i) p- type (hole)
conductivity and (ii) the fluorescence. The fluorescence phenomenon has got potential
applications in the development of ultra high density optical non-volatile memories and in
Medical/Molecular imaging. The nano silver gives raise to localized surface plasmons (LSP).
The LSPs have potential applications in nanoscale optics and photonics as well as in the
construction of sensors and biosensors. These two observations are manifestations of the bond
between the metallic silver and oxygen. .
2.2. Introduction
Silver has long been used in photo-activated image processing (photography), high
density alkaline batteries, household mirrors and tableware. Because of its high reflectivity,
silver is an important component in optical coatings; it is used in optical solar reflectors
(OSRs) in satellites. Recently, silver is used to generate localized electric fields for molecular
detection and electrical field enhancement. Silver is very reactive noble metal and forms
silver oxide by interacting with oxygen in the ambient. Nanosize structures of silver have, as
anticipated, very different properties compared with that of the bulk silver. Some of the
emerging applications of nano-silver oxide thin films are: the anti-bacterial activity [2.1],
possible use in super -Resolution Near-Field Structure (RENS) for optical read – write ultra
high density non-volatile memories [2.2], fluorescence imaging [2.3] and the property of
Surface Enhanced Raman Scattering (SERS) [2.4, 2.5]. The decomposition of nano-silver
oxide (into oxygen and small metallic silver clusters at a relatively lower temperatures and
high energy optical beams) can create a strong light-scattering centre that resolves small pits
or marks beyond the diffraction limit is a useful property in ultrahigh-density optical data
storage applications [2.2].
Silver, because of its d-shell electrons, exists in different oxidation states and forms
several oxides: AgO, Ag2O, Ag3O, Ag2O3. The formation of these oxides depends upon the
growth conditions/ reaction kinetics: availability of oxygen in the growth chamber and the
energy required for the oxidation. The surface morphology and the nucleation kinetics depend
upon the kinetic energy of the silver oxide molecules reaching the substrate. The heat of
formation of Ag2O and AgO are -31.1 kJ/mol and 220 kJ /mol respectively. It is known that
the AgO phase is relatively stable at high oxygen pressures and at low temperatures. The AgO
exists in different crystal systems like cubic, monoclinic and tetragonal [2.6 – 2.8].
As it is well known that the noble metal clusters have strong electromagnetic field due to
the electron gas. The local oscillations of the electron gas in the nano metal clusters are called
Plasmons. The localized surface plasmons (LSPs) are charge density oscillations confined to
nano metallic clusters. Excitation of LSPs by an electric field with suitable monochromatic
wavelengths results in strong light scattering, appears as intense surface Plasmon absorption
bands; consequently, at resonance, there will be an enhancement of the local electromagnetic
field. The frequency and intensity of the surface Plasmon (SP) absorption bands are
characteristic of the material. The example of SP resonance is the red colour of aqueous
dispersion of colloidal gold particles (the colloidal gold is used as bio-markers).
When the metal nano clusters are in the immediate vicinity of an insulator, the surface
plasmons can induce defect levels in the insulator. These defect levels when photo-activated
will give fluorescence emission. One example is : silver – silver oxide nano clusters. The
fluorescence intensity is significantly high in these photo-activated nano clusters. The SP
absorption, consequently, the fluorescence intensity, depends upon the size, size distribution
and shape of the nano-structures.
2.2. Surface Plasmons and Fluorescence in Silver – Silver Oxide Nano

Clusters
The silver nanoparticles have the property to exhibit surface plasmons as they can sustain
resonant collective electron oscillations [2.9]. These surface plasmons can be optically
excited either in the near ultraviolet or visible range of wavelengths. As it is well known that
the properties of the surface plasmon resonance are critically dependent on the shape, size,
and spatial arrangement / close proximity of the metal and insulator particles [2.10]. The
Plasmon resonance has a vast potential in nano-optics [2.10] and in bio-medical engineering
[2.11]. The size-dependent physical properties in nanoparticles have a special advantage to
study the evolution from small-molecule to bulk-systems. A good example is the fluorescence
of noble metals. It is well-known that the quantum yield for fluorescence from noble metal
films is extremely small, most of the times, below the detection levels, of the order of 10-10
[2.12]. The fluorescence mechanism is because of radiative recombination of sp-conduction-
band electrons with d-band holes [2.12, 2.13]. This process is explained by the plasmon
enhancement of the inter-band transitions.
Interestingly, Dickson et al. reported recently on nano silver clusters showing a structured
luminescence band [2.3] thus exhibiting properties characteristic of distinct molecular
clusters. Studies in this area are complicated by the difficulty of generating and stabilizing
very small clusters. The silver oxide nano clusters can be stabilized by encapsulation in
dendrimers, by thiol - ligands as well as by peptide-capping agents and by embedding in glass
matrixes.
Thus, the main focus for the search in silver / silver oxide is to grow nano sized metal /
metal oxide clusters and on the preparation of a single stable phase oxide of silver. The
present section confines to the growth of nano sized silver oxide clusters in thin film form.
There are several techniques followed to grow these silver oxide thin films: RF [2.14] and DC
reactive sputtering [2.5, 2.15], reactive thermal evaporation [2.16], chemical [2.17] and
electro-chemical synthesis [2.18a, 2.18] and by reactive pulsed laser deposition (RPLD)
[2.19].
The silver oxide thin films prepared by reactive DC Magnetron sputtering technique
[2.15] have shown that (i) the oxygen pressure in the growth chamber ( ~10-3 mbar)
influences the formation of different oxides of silver, (ii) at high oxygen flow rate in the
growth chamber (2.01 sccm, corresponds to a chamber pressure of 7 x 10 -3 mbar) and at a
sputter power density of 1.0 watts /cm2, stiochiometric Ag2O is formed, (iii) the mixed silver
oxide films show P- type conductivity.
The silver films are prepared at three magnetron powers: 20, 30 and 40 watts. The
distance between the target and the substrate is kept at 7.5 cm. The sputter time is varied from
30s to 120s. A thickness monitor (Hind Hivac Model DTM 101) displayed the thickness of
the films during the deposition. Thickness of the films grown is in the range 5.0- 40.0 nm.
The oxidation of these metallic films is carried out with pure oxygen gas (99.9%) and at 520
± 2 K for 30 minutes in a chamber (evacuated initially to a pressure of 1x10-5mbar; during
oxydation the chamber pressure is at 5x10-4mbar).
For plasmonic devices, it is desirable to have a single low energy phase of silver oxide
(where photo-activation can be achieved more easily). The pulsed laser deposition (PLD) and
the chemical modulation techniques have the potential to produce a single phase
photoactivated silver oxide nano clusters.
Dellasaga et al., recently reported the growth of silver oxide thin films by PLD technique,
using the fourth harmonic beam (λ = 266 nm) of Nd: YAG Laser with a fluence of 1.6 J/cm2
and varying the oxygen pressure in the chamber between 4- 150 Pa range (they have not
reported the electrical and optical properties) [2.19]. These experimental conditions have
produced varying visible plume length and large droplets (~ 2µm to 200 nm) on the grown
film. The varying visible plum length has a significant influence on the reaction and growth
kinetics. In the recent study conducted in the author’s laboratory [2.20], it is reported that in
the oxygen pressure range 9 Pa to 50 Pa, with the third harmonic of the laser pulse (1.06
J/cm2), it is observed that with increasing oxygen pressure in the growth chamber: (i) the
hexagonal Ag2O transforms to monoclinic AgO, (ii) the grain size in the film increases from
59 nm to 200 nm, (iii) the surface roughness of the film increases from 09 nm to 42 nm, (iv)
the resistivity of the films increases, (v) the activation energy (from electrical conductivity
measurements) increases from 0.64 eV to 0.75 eV, (vi) the surface work function of the films
increases from 5.47 eV to 5.61 eV and (vii) the optical band gap of AgO thin films decreases
from 1.01 eV to 0.93 eV. The Raman spectroscopy on AgO thin films show low wave
number peaks corresponding to the stretching vibration of Ag – O bonds. This study shows
that single phase AgO thin films, a requirement for plasmonic devices, can be prepared at
room temperature by PLD technique with an oxygen pressure of 20 Pa.
From the earlier work, two important properties (from among the several interesting
properties) exhibited by silver oxide thin films are discussed in this section. They are: (i) p-
type conductivity and (ii) fluorescence emission from photoactivated silver oxide.
2.3. P- type Conductivity in Mixed Phase Silver Oxide Thin Films
The mixed silver oxide thin films (AgO and Ag2O) prepared by reactive DC Magnetron
sputtering technique have shown P- type (hole) conductivity [2.5]. From the concepts of
silicon physics, it is well known that the holes are formed if there is localization of covalent
bonds in the material. How these covalent bonds are formed in the silver oxide films is the
question. The p-type conductivity in these films has been explained on the basis of
Sanderson’s theory of partial ionic charge; that is, the bond between the oxygen and silver
contains both ionic and covalent. The mixed bonding is also supported by the theoretical
calculations of electronic structure and bonding mechanism [2.21].
From the theory and empirical formulation proposed by Sanderson [2.22], it is possible to
calculate partial ionic charge on the various oxides of silver. With the data given in Sanderson
[2.22], calculations of the partial ionic charge of silver in Ag2O, Ag2O3, AgO and Ag3O4
yields 41, 26.7, 32 and 28.4%, respectively [2.5]. The partial ionic charge indicates the
presence of covalent bonds in the material. Since the bonding between silver and oxygen is
partially ionic and also since the electronic configuration after molecule formation reduces
towards the non-bonding nature of the oxide ions (which is the basic assumption in
Sanderson’s theory), there is a possibility of localization of the covalent bonds in these silver
sub oxide thin films. The localization of the oxide ions gives raise to the p-type nature of the
conduction (this concept is equivalent to the chemical modulation of the valence bond).If the
density of these localized states is considerably large in the film, the holes appear as majority
carriers showing p-type conduction; the p- type conduction can be confirmed with Seebeck
Coefficient and in Hall Effect measurements. Thus, the silver sub oxide film possesses both
n-type (due to electrons from metallic silver) and p-type (the oxide bond with silver)
conduction. One possibility for understanding the covalence is through the bond energy;
when the silver metal reacts with oxygen, if the bond energy for the sub-oxides (for example
Ag2O3, AgO, Ag2O, Ag3O4) is less, this weak bond may give rise to the presence of an
acceptor level leading to hole conduction in the film. It is known that in the sputtering
process, the energy to form the molecule is taken from the plasma. More details of the p- type
conduction and the complexities in the measurement are discussed in the section 4 (p-type
TCOs).
2.4. Fluorescence from Photo-Activated Nano-Silver Oxide
The research group of Dickson at the Georgia Institute of Technology observed photo
activated fluorescence from sol- gel prepared nano scale silver oxide (Ag2O) films, exhibiting
properties characteristic of distinct molecular clusters. The photoactivated fluorescence was
produced from non-luminescent silver oxide thin films by photoexcitation [2.23, 2.24, 2.25,
2.26]: silver oxide decomposes into oxygen and small metallic silver clusters of several
nanometer size. This dissociation of silver oxide into metallic silver and partial silver oxide is
the basis for a plasmonic structure. After local photoactivation of silver oxide films with UV
or blue light, the possibility for exciting the fluorescence into the visible spectra represents an
attractive prospect for advanced optical data storage [2.14, 2.15]. However, several studies
report that metallic interfaces play complex roles in the basic interactions of an
electromagnetic field with optically active materials. For example, smooth metallic surfaces
reduce the radiation from nearby organic dyes through non-radiative energy transfer, while
molecules absorbed onto metallic electrodes show several orders of magnitude increase in
surface-enhanced Raman signals [2.27]. Furthermore, quantum dots (QDs) located within the
surface plasmon polariton (SPP) field of the metal also shows a large photoluminescence (PL)
enhancement [2.28]. Then the question is: how to obtain an optimized silver oxide – silver
cluster size to obtain the photoexcited fluorescence or how to control the sizes of the silver
clusters. One requirement is that the silver oxide should be in a stable single phase (either
Ag2O or AgO) and the phase should be such that it can be photoactivated (or the energy
required to dissociate oxygen from the silver should be low).
Figure 2.1. Basic plasmonic structure of Silver – silver oxide nano clusters
Green
525-532 nm light Red emission
Fluorescence
Blue 488-
514.5 nm
Filter for 660 nm
Filter for
514.5 nm Scattered and
fluorescent
emissions
Ag2O film
Figure 2.2. Basic concept of optical ultra high density read – write non-volatile memory
0.91
fluoroscence from silver oxide film
0.90
0.89
Normalised Intensity (nm)
0.88
0.87
0.86
0.85
0.84
630 640 650 660 670 680

Wave length (nm)
Figure 2.3. Fluorescence observed in DC Magnetron sputtered silver oxide thin films
It may be mentioned here that the photoactivation of silver oxide into silver nano clusters
probably depends upon the bond energy of oxygen to silver. Thus, it is quite likely that the
nature of the bond : electrovalent or covalent, the oxidation energy and kinetics of nano silver
are extremely important in the basic understanding of photoactivation. The mechanism for
photoactivation is postulated to be initiated by absorption of a visible photon by Ag2O, a
semiconductor material with a band gap of 2.25 eV . Photoexcitation of Ag2O is known to
produce silver peroxide, AgO, which may photolytically form AgO. Further evidence for this
mechanism was obtained from the photoactivated generation of fluorescent silver particles
from AgO films. Such a light-induced reaction sequence may thus supply a nucleation site for
the growth of silver clusters.
1.01
400 Å
1.00 250 Å
100 Å
0.99
Normalised Intensity
0.98
0.97
0.96
0.95
0.94
0.93
0.92
600 620 640 660 680

Wavelength (nm)
Figure 2.4. The fluorescent intensity in silver oxide thin films with different thicknesses
1.00 20 W
30 W
0.95 40 W
Normalised Intensity
0.90
0.85
0.80
0.75
0.70
600 620 640 660 680 700

Wave length (nm)
Figure 2.5. The fluorescence spectrum for silver oxide thin films grown with varying DC Magnetron
power densities
More recently, it has been found that silver film can be cheaply and easily transformed
into Ag nanoparticles and nanowires in a gas mixture of hydrogen and oxygen in a vacuum
chamber. The particle and wire diameters are very uniform and approximately 20 -50 nm, and
they can be three-dimensionally fabricated on almost all material surfaces without the need
for thermal annealing. Silver nanoparticles and wires will soon appear in small and cheap
molecular detection sensors by small precise adjustments for generating surface plasmon
conditions.
A periodic array of silver oxide may be the first step in realizing a ultra high density (~
tera byte per square inch) [2.18]. The composite film is made from silver coated polymeric
arrays on an indium tin oxide (ITO) coated glass substrate. By adjusting column diameters
and lattice constants of the array to coincide with the excitation wavelength, the fluorescence
was markedly enhanced. The increase is due to the efficient energy transfer from silver
surface plasmon to silver oxide that is formed during the preparation of the composite film.
3. NANO TUNGSTEN OXIDE THIN FILM FOR ELECTRO-CHROMICS

3.1. Summary
Electro-chromics is an emerging area for smart and energy efficient windows and
displays. The well researched and reasonably understood material is tungsten oxide.
However, the coloration properties of the tungsten oxide are not understood in detail yet. The
nano tungsten oxides wires and rods have a vast potential not only in enhancing the coloration
properties but also in giving new dimensions in display technology. The present section
summarizes the efforts to understand the coloration / optical process based on the “site
saturation” model. The basic experiments to prepare the nano-tungsten wires and rods also
have been mentioned. The other important observation is that when tungsten oxide is
irradiated with electrons, a strong coloration is produced and the coloration is retained for
almost infinite time. This electron beam induced coloration can be applicable for optical
memories. The details of the interaction of electrons with the tungsten oxide are described in
this section.
3.2. Introduction
Electrochromism is defined as the ability of a material to change its optical states in

response to an applied DC voltage. In detail, the optical transmittance of the electrochromic
materials can be changed in a controlled and reversible manner from a complete transparent
state to a fully dark colored state by applying an electric potential across the material; this
electric potential enables the charges from external circuit to intercalate with the
electrochromic lattice. Tungsten oxide is the well known electrochromic material which is
extensively studied for its potential application in several fields like electrochromic smart
windows, displays, solar cells, fuel cells, gas sensors, automotive rear-view mirrors, sun
roofs, and organic/inorganic hybrid memory devices etc, [3.1]. Tungsten oxide is also a
photocatalytic material. The transmittance of WO3 films can be altered in a reversible and
persistent manner by the intercalation/de-intercalation of small cations (H+, Li+, Na+) and
electrons into the film. The reaction mechanism responsible for coloration in tungsten oxide
thin films is widely accepted as the double injection/ extraction of electrons and ions in
tungsten oxide,
WO3 + xe − + xM + ↔ M xWO3 ( M = Li, Na, H ) 3.1

(Transparent) (Blue)
There are two aspects to the electro-chromics: the material aspect (a transition metal
oxide) and the device aspect consisting of the ion intercalation and de-intercalation. The
electrochromic properties like coloration efficiency, cyclic durability and kinetics of
coloration-bleaching process of the electro-chromic material strongly depend on its structural,
morphological, and compositional characteristics, and therefore on the deposition techniques
and growth parameters. The commonly employed deposition techniques are thermal
evaporation, electron beam evaporation, sputtering, pulsed laser ablation, chemical vapor
deposition, sol-gel coating, and electro deposition.
Figure 3.1 Schematic of an Electro-Chromic Device
Electrochromic device consists, in general, of two (transparent) electrodes, an electrolyte,

and a minimum of one electrochromic active layer (Fig 3.1). A transparent glass, usually used
as a substrate, is coated with a transparent and conductive electrode (ITO), which is covered
with the electrochromic active material. These layers are part of an electrochemical cell
completed by an electrolyte. In all-solid-state devices, an ion conductor is used instead of an
electrolyte, and an ion storage medium (counter electrode) is necessary. Traditionally, current
or charge is recorded as a function of applied voltage, and transmittance or reflectance is
measured as a function of the potential to characterize electrochromic materials. Most
investigations for the electrochromism of single film are done in liquid electrolyte
experiments. In principle, only two electrodes are necessary for an electrochromic
experiment, but for fundamental research a three electrode configuration is mostly used to
control the potential at the electrochromic film (working electrode), relative to a reference
electrode. The current -voltage graphs and transmittance-voltage graphs are typically
recorded, but also the response in the electrical and optical data to a constant voltage is
measured to determine the switching speed of an electrochromic film. An important property
of electrochromic thin films for evaluation is the coloration efficiency (CE), expressed as cm2
C-1. In order to obtain this value, the total injected/ intercalated charge, as a function of unit
area, and the change in the optical density (δOD) must be known. Both CE and OD are
wavelength dependent quantities. The change in the optical density is correlated with the
transported charge Q (per unit area). From the two quantities, the coloration efficiency can be
determined by the equation: CE (λ) = δOD/Q. CE is a number which is independent of the
film thickness, because δOD and Q depend in the same way on the film thickness.
One of the desired applications of electrochromic systems is their use in the ‘smart
windows’. Compared with other window/glass constructions, up to 50 % of the energy costs
for heating and cooling buildings can be saved with `smart windows [3.2]. Today, these
advantages are still offset by the heavy currents the window requires for optimal operation.
For example, the charge required for the color change has an order of magnitude of 100.0
Coulomb m-2, even for the best systems known today. Since the system operating voltage
normally is of the order of 2.0 V, a current of several amperes must be applied for a fast-
switching system via the surface electrodes at this relatively low-operating voltage. In order
to obtain color-changing times in the order of minutes range for a window having a typical
dimension of 2.0 m x 2.0 m, a current of approximately 10.0 A (of 2.0 V) is necessary. If
there is any inhomogenity or non-uniformity, the coloration shows up: this leads to partly
darkened parts of the window (near the electrodes), which may cause irreversible material
deteorations. A second significant problem is the heating-up of the window by use of
absorptive materials leading to impairment of the long term stability. These problems are still
awaiting solutions.
Two questions are addressed in the present section on electro-chromics: (i) what is the
optical process of coloration in the tungsten oxide thin films and the testing of “site
saturation” model for stiochiometric tungsten oxide and (ii) how to prepare the nanowires
and nanorods of tungsten oxide for enhanced electro-chromic performance.
3.3. The Optical Absorption Process in Tungsten Oxide Thin Films
The optical absorption process, in electro-chromic materials in general, and tungsten

oxide in particular, is not yet clearly understood. Faughnan et al [3.3] reported that the
6+ 5+
coloration process is primarily due to inter-valence charge transfer between W and W
sites; Lee et al [3.4] proposed that the sub-stoichiometric tungsten oxide thin films rich in
W 4+ states exhibit better electrochromic properties where the optical absorption/ modulation
4+ 5+
during intercalation of charges is primarily due to the transitions between W and W sites.
Several researchers have supported the validation of this theory and various attempts have
been made to deposit sub-stoichiometric tungsten oxide thin films for optimum
electrochromic performance. Recently, Berggren et al [3.5]have reported (on Li+ intercalation
in tungsten oxide) that the coloration process is independent of the stoichiometry of the films
and a statistical theory has been developed based on “site saturation” model:
The essential features of the model (based on Li intercalation) are (i) the relative number
6+ 5+ 4+
of electronic transitions (P) between the states W , W , W are determined by the
amount of charge intercalated in the sample i.e., x = Li / W ratio (ii) the analytical
expressions for the number of possible electronic transitions are given by,
W 6 + ↔ W 5+ : P = 2 x ( 2 − x ) 3 , 3.2
W 5+ ↔ W 4+ : P = 2 x 3 (2 − x), 3.3
W 6+ ↔ W 4+ : P = x 2 (2 − x) 2 3.4
and (iii) the absorption strength per transition depends on the type of transitions (higher for
W 6+ → W 5+ ).
In the “site saturation” model, the inter-valence charge transfer between different states
6+ 5+ 4+
of tungsten i.e., W , W , W is governed by the amount of charge intercalated in the
sample. According to Berggren et al., if the ratio of Li / W is less than 1.0, then the transition
6+ 5+
occurs primarily between W and W states and with further intercalation of charges i.e.,
1.0 < Li / W < 2.0, the transition between W 5+ and W 4+ states becomes predominant.
From this theory, one may infer that at low intercalated charges, the stoichiometric or
super stoichiometric tungsten oxide thin films undergo higher optical absorption (due to the
6+
larger number of available W sites) compared to sub-stoichiometric ones. From an
application point of view, this is an interesting result, since the realization of larger optical
modulation with small charge insertion or extraction would enhance the speed and the long -
term cycling stability of the electrochromic device. Then the question is how to prepare a
super stoichiometric tungsten oxide thin films. From research point of view, varying the
oxygen content in the tungsten oxide thin films while achieving the stoichiometry is an
interesting problem. One of the techniques to grow stoichiometric or super stoichiometri
tungsten oxide (WO3) thin films is by sputtering metallic tungsten target using pure oxygen
as a sputter gas [3.6]. Using oxygen as sputter gas is not uncommon; it has been employed to
deposit transition metal oxides like TiO2 [3.7]. The growth conditions and the properties of
oxygen sputtered tungsten oxide thin films are summarized in Table 3.1. All the experimental
data on oxygen sputtered WO3 thin films like the optical modulation, coloration efficiency,
switching speed and cycling durability are in support of the recently reported statistical theory
based on “extended site saturation model”.
Table 3.1 Oxygen sputtered tungsten oxide thin films : growth conditions and properties
The x = Li / W is replaced by H / W in the present paper. In the present case, the value
of x was found to be ~ 0.6 as determined from the expression [3.5],
Qe
x= , 3.5
ρel
where Qe is the total effective charge passed through the device; e is the electron charge; l is
the film thickness; and ρ is the density of tungsten sites present (the values of Qe and ρ are
evaluated with the details given in reference [3.8]). The relative number of transitions
plausible at x = 0.6 have been plotted in Figure 3.2 (inset) following equations 3.2 to 3.4.
Since coloration efficiency is a thickness-dependent quantity, the coloration behavior of the
films is presented in terms of thickness independent quantity ‘α ‘[Figure 3.2]. The value of x
~ 0.6 is also confirmed by the pronounced optical absorption coefficient at E ~ 1.5 eV. It is
also found that the absorption coefficient is highest for the S2 samples whereas the coloration
efficiency is highest for the S3 samples. Such a variation is attributed to the difference in the
thickness of the samples; 553.0 nm for S2 and 621.0 nm for S3. All the above results clearly
6+ 5+
indicate the predominance of W → W transitions in the oxygen sputtered stoichiometric
tungsten trioxide thin films at low intercalated charge levels (x ~ 0.6). Since the relative
6+
density of W states available in the oxygen sputtered films is higher, the optical absorption
6+ 5+
strength for W → W transitions per unit-inserted charge is much larger than that of the
sub-stoichiometric films. Generally, the better coloration observed in the oxygen sputtered
WO3 thin films is in support of the ‘extended site saturation model’. However, the large
differences between the optical properties of the films S1-S3 have various reasons like ‘site
saturation’, thickness variation and the slight differences in the amount of proton
intercalation.
Figure 3.2. The absorption coefficient of WO3 films sputtered at different oxygen pressures as a
function of energy. (Inset): The inset shows the relative number of transitions between different states
of tungsten as a function of H/W ratio.
The effect of temperature on the properties of WO3 films by intentionally heating the
substrates during deposition has also been investigated. At lower oxygen chamber pressure,
substrate heating has resulted in the formation of WO3 nanowires without any catalyst.
3.3. Kelvin Probe Studies
The charge intercalation brings about a change in the surface work function of these
electro chromic thin films. It is possible to estimate the surface work function of these thin
films in the colored and the bleached states using Kelvin probe measurements [3.9]. The
surface work function may be estimated, as an approximation, from the concepts of bands
(generally the band theory is for crystalline solids with periodicity). This approximation is
mainly to understand the differences between the surface work function of the thin film in the
colored state (charge intercalated) and in the bleached state (the charge is de-intercalated).
The concept of bands is invoked mainly to obtain a reasonable understanding of the surface
work function of these amorphous thin films; equivalently, one can also understand these
changes in terms of surface electrode potential. In as deposited WO3 thin films, the Fermi
level lies in the band gap between the conduction band and the valence bands (wide band gap
metal oxides). In principle, during the intercalation process, electrons occupy the conduction
band and there will be a shift in the Fermi level position [3.10]. The coloration being a double
injection: electrons flow into the tungsten oxide (from the transparent and conducting ITO)
and the protons (or positive ions) from the electrolyte. The intercalated charges move the
Fermi level towards the conduction band. Hashimoto et al. [3.11] have studied Li intercalated
amorphous WO3 thin films by X ray Photoelectron Spectroscopy (XPS) and has shown that
the movement of the Fermi level is in proportion to the amount of intercalated charge. The
surface work function is the measure of the energy from the Fermi level to the vacuum level.
The measurement on work function difference of these films before and after ion intercalation
may be directly correlated to the movement of the Fermi level. The work function may be
evaluated from the Contact Potential Difference (CPD) measured by the Kelvin probe
technique. The relation between the CPD and surface work function can be given by
qVCPD= ΦR– ΦF 3.6
where q is the electronic charge, VCPD is the contact potential difference measured, ΦR = 4.83
eV is the work function of the gold coated reference electrode and ΦF is the work function of
the film under study.
For the Kelvin probe measurements, the intercalation is done on selective area of the
WO3 film by forming a hemispherical drop of electrolyte solution (0.5 M HCl dissolved in
distilled water) on the film and using platinum as the counter electrode. Figure 3.3 shows the
variation in CPD of the tungsten oxide electrochromic layers scanned over a selectively
colored portion (0.5 cm) of the films.
The counter electrode tip was placed at the centre of the hemispherical drop such that the
maximum CPD value at the centre of the colored portion could be observed. The figure
clearly presents the difference in the amount of charge intercalated in each sample. The
maximum CPD values of colored samples S1, S2 and S3 are found to be 400.0 mV, 449.0
mV and 511.0 mV (the accuracy of the measurement is ±1.0 mV); following equation 3.6, the
work functions are: 4. 41 eV, 4.36 eV and 4.30 eV respectively. The work function of the
uncolored samples is found to be ~ 4.86 eV. The difference in the CPD values indicates a
change in the Fermi level in tungsten oxide with intercalation [3.12]. These values support the
view that the decreasing work function is due to the enhancement of conductivity of the
tungsten oxide film with intercalation.
500 colored WO3 -2

1.5 x 10 mbar
-2
3.1 x 10 mbar
-2
5.2 x 10 mbar
400
300
CPD (mV)
200
100
Decolored WO3
-100
0 1000 2000 3000 4000 5000
Distance (μm)
Figure 3.3. Contact potential difference measurements (CPD) on a selective colored portion (0.5 cm) of
oxygen sputtered tungsten oxide thin films
Figure 3.4. Topography of Contact Potential Difference (CPD) for the oxygen sputtered WO3 film (S3)
scanned over an area of 2.0 cm x 1.0 cm
Lower work function of S3 in comparison with that of S1 and S2 can also be explained as
due to higher electronic conductivity of S3. Similar observation on the fall in electrical
resistivity with coloration has been reported earlier [3.13]. Figure 3.4 shows the topography
of CPD for the tungsten oxide electrochromic layer of S3 sample (scanned over an area of 2.0
cm x 1.0 cm) depicting the variation in CPD of selectively colored WO3 film. The stability of
an electrochromic device with repeated color / bleach cycles depends on the amount of
residual charge left on the film in each cycle after de-intercalation. The cycling durability of
the films has been studied by measuring the amount of residual charge left in the sample
using Kelvin probe. The bleached portion of the film has a CPD of -120.0 mVolts, which is
even less than the CPD of rest of the sample. The dip in the plot clearly indicates that there
are no residual charges left in the sample after bleaching; ensuring a better cycling durability
of the electrochromic device based on oxygen sputtered WO3 films.
3.4. A Brief Outline on Nano-Electro-Chromics Materials and Processes
In the recent days, one-dimensional (1-D) tungsten oxide nanostructures like nanorods;
nanowires and nanotubes have attracted much attention due to their potential applications in
various fields [3.14-3.16]. Due to large surface to volume ratio, the one-dimensional
nanostructures of tungsten oxide has been used in wide range of areas like field emission,
electrochromism, luminescence and gas sensing [3.17-3.20]. The significant application of 1-
D nanostructured tungsten oxide is its use as a structure-directing precursor for WS2 nanotube
production [3.21], a useful material in tribological applications and catalyses. Many
researchers have grown different forms of tungsten oxide nanostructures mostly by the
oxidation of metallic tungsten in the form of powders, foils and filaments [3.22 – 3.25]. It is
found that the resulting compositions, crystal structures, and shapes vary significantly with
preparation techniques. Nonetheless, studies on nanoscale tungsten oxide materials are still
limited by the lack of an appropriate preparation method. Therefore efforts to develop a
simple, economic and efficient method for the synthesis of tungsten oxide 1-D nanostructures
for device applications are still required. Nano crystalline and one dimensional WO3
nanowires have been studied for electrochromic devices due to its fast diffusion kinetics and
improved reliability. Meda et al. (2002) prepared nano crystalline WO3 by organometallic
chemical vapor deposition, and reported coloration efficiency of 22 cm2C-1 [3.26]. Particle
size in their studies observed was about 20 – 40 nm. Deshponde et al. (2007) have reported a
higher coloration efficiency of 44 cm2 C-1 for their nano structured thin films compared to 24
cm2 C-1 of amorphous thin films [3.27]. Durability of nano structured thin films tested in
acidic electrolyte was superior to that of the amorphous thin films. Both studies discussed
above have supported their observation of increased coloration efficiency and durability by
the porous and increased surface area characteristics of the nanocrytalline materials. One
dimensional nanowire has been synthesized by solvothermal technique [3.28], solution route
[3.29] and thermal evaporation of WO3 powders [3.30- 3.32]. Yoo et al. (2007) investigated
WO2.72 nanowires synthesized using solvothermal method [3.28]. They reported higher Li+
diffusion coefficient of 5.2×10-11 cm2S-1 and coloration and bleaching time of 3.5 s and 1.1 s
respectively. Coloration efficiency was 55 cm2 C-1. They attributed this higher coloration
efficiency and elevated Li diffusion coefficient to increased surface area and short diffusion
path length for Li+. Similarly a fast response time has been reported for the single crystalline
WO3.H2O structure produced through solution route, using PVA-124 as a structure directing
agent and glacial acetone as a stabilizer [ 3.29]. They reported a wider lattice spacing of d =
5.36 Å towards the electrolyte side which responsible for the faster kinetics. WO3.H2O was
formed by the layered build up of corner sharing octahedral with water molecule in between
the planes. This structure leads to an increased d spacing in this single crystalline WO3.H2O
system.
It is found feasible to prepare tungsten oxide (W18O49) nanorods and nanowires by the
conventional dc magnetron sputtering technique in active arc suppression mode with metallic
tungsten target in Ar + O2 atmosphere (at a sputtering power density of 3.0 Wcm-2 ) with
substrate temperature. During deposition, the argon chamber pressure was kept fixed at 1.2 x
10-2 mbar whereas the oxygen chamber pressure was varied in the range 1.0 x 10-3 mbar to 5
.0 x 10-3 mbar. The deposition was made on glass, silicon and ITO coated glass substrates.
The following results describe the nature of the dc magnetron sputtered tungsten oxide
nano-wires and their dependence on the oxygen chamber pressure. The surface morphology
of the substrate heated WO3 films was studied by scanning electron microscopy (SEM).
Figure 3.5 shows the SEM image of tungsten oxide thin film deposited at an oxygen chamber
pressure of 1.0 x 10-3 mbar. The micrograph shows the formation of randomly dispersed
nanowires having a diameter of 60– 80 nm and length of ~ 10.0 µm. However, the film
deposited at an oxygen chamber pressure of 2.0 x 10-3 mbar shows a nonuniform
distribution of nanowires i.e., in a matrix of lengthy and horizontally dispersed nanowires. An
island of quasialigned nanowires with larger diameters and shorter lengths is also found. The
region surrounding the island is filled with nanowires of average length 10.0 μm and average
diameter 50 – 80 nm. The maximum length of these wires extends up to 30.0 μm.
Figure 3.5. SEM image of the WO3 film deposited at an oxygen chamber pressure of 1 x 10-3 mbar with
a substrate temperature of 450 °C
The nanowires in the island region have an average diameter of 60.0 – 100.0 nm and
length of 3.0 μm. Despite the difference in the dimension and orientation of the two types of
nanowires, it is found that the diameter of the nanowires is uniform throughout their length
(Figure 3.6). With further increase in the oxygen chamber pressure to 3.0 x 10-3 mbar, it is
found that the morphology to change from nanowires to that of mixed features with larger
particulates and rectangular rods of length 1.0 to 10.0 µm [Figure 3.7]. The morphology of
the films deposited at an oxygen chamber pressure of 4.0 x 10-3 mbar shows a uniform
distribution of particles with an average particle size of about 500.0 nm. The surface is found
to be dense, compact and is free from any pores [Figure 3.8]. However, the films deposited at
an oxygen chamber pressure of 5.0 x 10-3 mbar shows the presence of pores of diameter 50.0
nm to 500.0 nm.
There are also clusters of particles distributed through out the film [Figure 3.9]. The
above results clearly show the pressure dependence on the properties of substrate heated WO3
thin films.
3.5. Structural Properties of WO3 Nanowires
Figure 3.10 shows the X-ray diffraction pattern of the WO3 nanowire thin films deposited
by DC magnetron sputtering in active arc suppression mode. The argon chamber pressure was
kept constant at 1.2 x 10-2 mbar whereas the oxygen chamber pressures were maintained at
1.0 x 10-3 mbar and 2.0 x 10-3 mbar. Except intensity, we could not observe much difference
in the peak positions between the two samples.
Figure 3.10. X-ray diffraction pattern of the WO3 nanowires deposited at an oxygen chamber pressure
of 1 x 10-3 mbar and 2 x 10-3 mbar with a substrate temperature of 450 °C
All the main peaks can be indexed with the reflections of monoclinic tungsten oxide
(W18O49) system (JCPDS card No: 05-0392). Consistent with general features of
nanomaterials, the overall diffraction intensity is weak, and peak broadening is pronounced.
Diffraction peaks of (010) are stronger than the rest, indicating that the major growth
direction is [010]. The above result shows that these nanowires grow along the [010]
direction, which is reasonable because the close-packed planes of monoclinic W18O49 crystal
are (010) [3.18].
3.6. Morphology of WO3 Nanowire Thin Films
Figure 3.11 shows the scanning electron microscopy image of tungsten oxide thin film
deposited at an oxygen chamber pressure of 1.0 x 10-3 mbar. The micrograph shows the
formation of randomly dispersed nanowires having a diameter of 60.0 – 80.0 nm and length
of ~ 10.0 µm.
Figure 3.11. SEM image of WO3 thin films deposited at an O2 chamber pressure of 1 x 10-3 mbar
showing nanowire formation over a large area of ~ 300 µm2
Figure 3.12. Low magnification SEM image of WO3 thin films deposited at an O2 chamber pressure of
1 x 10-3 mbar showing nanowires and cracks all along the surface.
As could be seen from the figure, the maximum length of the nanowires extends up to
30.0 µm. A lower magnification SEM image of the surface shows cracks all along the film
[Figure 3.12]. It is evident that the surface is uniformly filled with nanowires over a large area
of ~ 50000.0 µm2. Large area coatings of nanowire thin films have not been reported earlier
because of the limitations in the method of deposition. Since sputtering is an industrially
viable technique used for large area coatings, we have been able to achieve such a large
coating of nanowire thin film. However, the cracks found all over the films gives a clue that
nanowire growth could be associated with thermal induced strain in the film.
showing vertically grown nanowires
showing cracks and nanowires
To confirm this proposition, we deposited WO3 thin films at 2.0 x10-3 mbar with a
substrate heating temperature of 450 ºC. Fig 3.13 shows the surface morphology of WO3
films deposited at 2.0 x 10-3 mbar. The figure clearly shows the presence of vertically aligned
nanowires along with horizontally dispersed nanowires. In fact, Figure 3.13 is the low
magnification image of Figure 3.12 which clearly shows the formation of vertically aligned
nanowires. The lower magnification SEM image of these samples also show cracks all over
the film [Figure 3.14]. The above results give an indication that the catalyst free nanowire
formation could be due to thermally induced strain in the film. In addition, the absence of
nanowires at higher oxygen chamber pressures indicates that the essential conditions for the
growth of nanowires are the oxygen to tungsten ratio and thermally induced strain in the film.
Figure 3.15. SEM image of WO3 thin films deposited at an O2 chamber pressure of 0.5 x 10-3 mbar
showing the formation of nanowires within the cracks
To verify this proposition, WO3 thin films have been deposited at an oxygen chamber
pressure of 0.5 x 10-3 mbar. The morphology of these films studied by SEM shows cracks
along the film and within the cracks we could observe the formation of nanorods of length ~
500.0 nm [Figure 3.15]. With decrease in O2 chamber pressure (0.5 x 10-3 mbar) we could
observe significant changes in the surface morphology like (i) nanowires are formed inside
the cracks and no features are observed on the top surface whereas in the other films (1.0-2.0
x 10-3 mbar) the nanowires are formed only on the top surface (ii) the density of cracks
decreases with increase in oxygen chamber pressure. The difference in the location of
nanowire formation can be attributed to the difference in oxygen to tungsten ratio. Similar
observation of tungsten oxide nanowires grown inside the crack surface has been reported
earlier. It has been explained on the basis that the surface within the crack has oxygen to
tungsten ratio adequate for nanowire growth [3.33]. The importance of oxygen to tungsten
ratio in catalyst free formation of nanowires has also been reported for sputtered WO3 thin
films [3.34]. The pressure dependence of the density of cracks can be attributed to the
difference in deposition time of the films. To maintain same thickness in films deposited at
various O2 chamber pressures (Po2), the deposition time is varied between 6.0 min (Po2 – 1.0
x 10-3 mbar) to 18 min (Po2 –5.0 x 10-3 mbar).
The samples deposited at Po2 of 1.0 x 10-3 mbar have a thickness of 350.0 nm as
estimated from the reflectance measurements. Due to technical difficulties involved in TEM
measurement of thicker films, we have evaluated the surface crystallite size of nanowires by
atomic force microscopy (AFM). Figure 3.16 shows 2D and 3D representations of the surface
morphology of WO3 samples deposited at Po2 of 1.0 x 10-3 mbar. The figures clearly show the
formation of well-defined nanowires. In addition, it is clearly evident that the surface is
smooth and dense with randomly dispersed nanowires spread throughout the film. The
average diameter of the nanowire determined from the height image was found to be 70.0 nm.
The detailed microstructures of tungsten oxide thin films were characterized using
transmission electron microscopy (TEM). TEM measurements were made on WO3 films
deposited at 1.0 x 10-3 mbar. The thickness of the films used for TEM characterization was ~
100.0 nm. Figure 3.17 shows transmission electron microscopy image of WO3 thin films
deposited at Po2 of 1.0 x 10-3 mbar and with a substrate heating temperature of 450 ºC.
Figure 3.16. 2D and 3D representations of the surface morphology of WO3 nanowire thin films
deposited at Po2 of 1 x 10-3 mbar and substrate temperature of 450 ºC
Figure 3.17. Transmission electron microscopy (TEM) image of WO3 thin films deposited at Po2 of 1 x
10-3 mbar and substrate temperature of 450 ºC
Figure 3.18. Higher magnification TEM image of WO3 thin films deposited at Po2 of 1 x 10-3 mbar and
substrate temperature of 450 ºC
Figure 3.19. Selected Area Electron Diffraction Pattern (SAED) of an individual WO3 nanorod
deposited at Po2 of 1 x 10-3 mbar
Figure 3.20. SAED pattern taken along the edge of a nanorod showing a superposition of single
crystalline and nanoparticle pattern
The micrograph shows the formation of tungsten oxide nanorods with an average length
of 150.0 nm and diameter in the range 40.0 – 100.0 nm. The nanorods are randomly dispersed
and are closely packed. From the higher magnification TEM image we find that the individual
nanorods are tapered along the edges [Figure 3.18]. In addition, between the nanorods we
could find a distribution of nanoparticles with an average particle size of ~ 7.0 nm.The
selected area electron diffraction pattern (SAED) of an individual nanorod exhibits a single
crystalline dot pattern [Figure 3.19] with lattice spacings of 3.76 Å corresponding to [010]
growth directions in W18O49 system [3.35]. Interestingly, the SAED pattern taken along the
edge of a nanorod shows a single crystalline pattern superposed on the ring pattern of
nanoparticles [Figure 3.20]. The above results suggest that well-crystallized W18O49
nanowires grow along the [010] direction.
From the SEM, TEM and AFM measurements, we did not find any liquid droplet features
at the end of tungsten oxide nanowires and nanorods and in fact there is no catalyst involved
during the deposition. Thus the well-known vapor-liquid-solid (VLS) growth mechanism
[3.36] cannot be applied for our nanowire growth, because a metal particle is necessary at the
growth front of nanowire to act as the catalytic site. Therefore, in our samples, the vapor-solid
growth process is presumed to control the formation of one-dimensional nanorods or
nanowires. Sputtering of metallic tunsgten in an appropriate oxygen atmosphere at suitable
substrate temperature is presumed to initiate the nucleation of nanowires. In our case, the
WO3 nanowires were formed at an oxygen chamber pressure of 1.0 and 2.0 x 10-3 mbar at ~
450 °C.
In the literature [3.37], the interfacial strain between substrate and film has been reported
to initiate whisker growth in tungsten oxide. In the present study, strain induced by difference
in thermal expansion between the different layers of WO3 during deposition is presumed to
relax as nanowires. The above results indicate that the key factors responsible for the catalyst
free growth of sputtered WO3 nanowires could be (i) the appropriate oxygen chamber
pressure (or O/W in film) during deposition and the (ii) thermally induced strain in the
substrate heated WO3 films. The properties of the substrate heated WO3 films deposited at
various oxygen chamber pressures (1.0 – 5.0 x 10-3 mbar) are presented in table 3.4.4.3. From
the table, we find the nanowires deposited at Po2 of 1.0 – 2.0 x 10-3 mbar to exhibit better
electrochromic properties.
3.7. Growth of Layered Tungsten Oxide Thin Films
The layer structured tungsten oxide thin films were deposited on glass substrates by
reactive dc magnetron sputtering (in active arc suppression mode) of metallic tungsten target
in argon and oxygen atmosphere. The sputtering parameters used for the growth of layered
structures are; sputtering power of 2.0 W/cm2, working pressure (argon + oxygen) of 1.4 x 10-
2
mbar (the chamber pressure with argon gas alone is 1.0 x 10-2 mbar) and deposition time of
10 minutes. The sputtered WO3 thin films have been irradiated with electron beam in a
commercial electron beam evaporation system at a vacuum of 5.0 x 10-5 mbar. For electron
bombardment, the samples were placed right on the top of the slot for crucible holder in the
electron beam evaporation system. In the present study, three samples of size 1.0 cm x 1.0 cm
have been grown in one run of sputtering. Of the three samples, one is characterized as such
and is labeled as S1. The other two samples have been subjected to electron irradiation:
sample S2 (bombarded with electron beam of energy 3.5 KeV and current of 3.0 mA for 8s)
and sample S3 (bombarded with electron beam of energy 3.5 KeV and current of 6.0 mA for
4s); the sample S3 is shown in the Figure 3.21.
Table 3.2. Deposition conditions and the experimentally determined optical and
electrochromic parameters of tungsten oxide thin films grown with a
substrate heating temperature of 450 °C
3.8. Effect of Electron Beam Irradiation on the Properties of WO3 Thin Films
The tungsten oxide can switch between two optical states in response to an applied dc
voltage (electrochromism), UV irradiation (photochromism) and thermal energy
(thermochromism). Apart from the above-mentioned conventional methods, WO3 can also be
colored by irradiation methods using electrons [3.38], ions [3.39] and lasers [3.40 ]. The
chromic behavior of tungsten oxide also has a vast potential in the emerging field of ultra
high density optical data storage. Once colored, the tungsten oxide has an inbuilt memory to
retain the coloration information even after the source is removed. This property of tungsten
oxide, as indicated in the literature [3.41, 3.42], makes it a candidate suitable for possible
ultra high density data storage application. Recently, the optimization of nanoscale electron
emitters (nanotube, nanowire and nanowires) gives feasibility for high-density data storage
based on the electron bombardment of tungsten oxide thin films. The effect of electron
bombardment on the structural, morphological, coloration and luminescent properties of dc
magnetron sputtered WO3 thin films is presented. For these studies comparatively thicker
films (1061.0 ± 5.0 nm) with layered structures have been grown. Due to electron
bombardment, significant changes in the physical properties of WO3 have been observed. In
addition, by varying the energy and flux of electrons we observed coloration and
luminescence in WO3 films. Based on these results, a concept for data storage application due
to electron beam induced coloration/ luminescence in tungsten oxide thin films has been
proposed [3.43].
Figure 3.21 Photograph showing electron bombarded tungsten oxide thin film (S3)
3.8.1. Structural properties of electron bombarded WO3 films
Figure 3.21 shows the X-ray diffraction pattern of the as- deposited (S1) and electron
bombarded tungsten oxide thin films (S2, S3). The broad featureless pattern at lower angles
clearly indicates the amorphous nature of as-deposited tungsten oxide thin films. However,
the XRD pattern of electron bombarded samples (S2, S3) shows high intensity peaks
corresponding to the tetragonal WO3 system (JCPDS card No: 89-1287). The samples S2 are
found to be highly crystalline with sharp peaks. The highest intensity peak at 24.0° shows
preferential orientation along the (100) plane. But the S3 sample shows broad and less intense
peaks, characteristic of nanoscale materials. The average crystallite sizes determined by
Scherrer’s formula are found to be 40.0 nm and 15.0 nm for S2 and S3 samples respectively.
Figure 3.21. X-ray diffraction pattern of the as-deposited (S1) and electron bombarded (S2, S3) WO3
thin films.
3.8.2. Optical properties of electron bombarded WO3 films
The transmittance spectrum of the tungsten oxide thin films (S1 - S3) in the wavelength
range 250.0 nm – 850.0 nm is shown in Figure 3.22. The spectrum of glass substrate is also
presented for reference. The as-deposited WO3 films have a high transmittance of ~ 85.0 % in
the entire visible range of 400.0 nm – 850.0 nm. The thickness of the film is 1061.0 ± 5.0 nm.
The sharp fall in transmittance at λ = 340.0 nm corresponds to the fundamental absorption
edge of amorphous tungsten oxide. When these samples are bombarded with electron beams,
the entire film (1.0 cm x 1.0 cm) turns deep blue in S2 sample, showing a maximum
transmittance of 15.0 % at λ = 503.0 nm. In addition, the band edge is also red shifted by 32.0
nm. But in the case of S3 samples, only a portion of the film in and around the point of
impingement turns blue showing a maximum transmittance of 21.0 % at λ = 477.0 nm.
Interestingly, the band edge of S3 sample is blue shifted by 60.0 nm. Coloration of the film
by electron impingement is attributed to the reduction of WO3 subsequently leading to the
release of oxygen atoms from the sample [3.44]. Each oxygen vacancy contributes a
maximum of two free electrons [3.45] and hence the extraction of oxygen atoms by electron
bombardment increases the carrier concentration considerably. Thus, the blue shift in S3
sample can be ascribed to the conventional Burstein - Moss effect [3.46]. Similar results have
been reported for the WO3 thin films colored by proton/ lithium intercalation [3.47, 3.1].
However, understanding of the red shift observed in S2 sample requires more investigations.
(Figure 3.23 ) shows the refractive index of as-deposited WO3 thin film as a function of
wavelength [400.0 – 850.0 nm]. The experimental values of n are in fair agreement with the
reported values [3.39] for WO3 thin films. But, interestingly, the wavelength dependence of
extinction coefficient k shows step like features as shown in Figure 3.24. The reason for such
a behaviour is not clear yet; but it is presumed to be associated with the layered structure.
Figure 3.22. Spectral dependence of transmittance in the wavelength range 250-850 nm of the as-
deposited (S1) and electron bombarded (S2, S3) WO3 thin films
Figure 3.23. Refractive index of the as-deposited sample (S1) as a function of Wavelength
Figure 3. 24. Spectral dependence of extinction coefficient for the as-deposited tungsten oxide thin
films
3.8.3. Compositional studies of electron bombarded WO3 films
The effect of electron bombardment on the stoiciometry of the samples has been studied
by evaluating the chemical composition of the films by XPS measurements. As shown in
Figure 3.25, the XPS spectrum of the as-deposited sample (S1) shows two peaks with binding
energies of 35.8 and 37.9 eV, corresponding to W4f7/2 and W4f5/2, respectively. They are the
valence state peaks of W6+ corresponding to the standard spectra of WO3 [3.48]. But when the
sample is bombarded with electron beam, the spectrum becomes broader showing additional
peaks, which create a shoulder at the lower binding energy side. In samples S2 and S3, the
shoulder is found approximately at 32.3 and 33.4 eV respectively. In addition, the valley of
the original doublet is also diminished considerably.
Figure 3.25. XPS spectrum of the as-deposited (S1) and electron bombarded (S2, S3) WO3 films in the
region of W 4f levels
These changes in the spectra are interpreted in earlier reports as a consequence of

nonstoichiometric conditions on the film surface, may be due to the formation of surface
oxygen vacancies [3.49]. All these results suggest the occurrence of compounds with less
oxygen to tungsten ratio by electron bombardment.
3.8.4. Surface morphology of electron bombarded WO3 films
The surface morphology of WO3 thin films has been studied by scanning electron
microscopy (SEM) and atomic force microscopy (AFM) measurements. The surface of the
as-deposited sample S1 is found to be smooth at the center but along the edges, it shows a
layered structure [Figure 3.26]. Several platelets are found to be stacked uniformly one over
the other. The platelets are found to have different sizes and shapes, but interestingly, all the
platelets have nearly same thickness as shown in [Fig 3.27]. The possible mechanism of
layered growth may be understood from the active arc suppression mechanism of the pulsing
unit: Sparc-LE 20. When charges build on the target due to surface oxidation, the pulsing unit
reverses the target voltage for ~10.0 μs. However, reversal of the target voltage attracts more
electrons subsequently leading to the discharge of positively charged insulative layer. This
action lowers the surface potential of the insulative layer and thus prevents the breakdown of
an arc. Once the arc is suppressed, the Sparc-LE unit reverses the target voltage back to
negative potential and normal sputtering is resumed. Thus the reduction of sputtering rate by
target poisoning along with the reversal of target voltage for few microseconds is thought to
initiate layer growth in the present investigation. The uniform thickness of the platelets
indicates that the target is charged up at regular intervals.
Figure 3.26. Scanning electron microscopy (SEM) image of the as-deposited tungsten oxide thin film
showing layered structure along the edge of the film
Figure 3.27. High magnification SEM image of the layered structure showing uniform thickness of the
individual platelets
Figure 3.28. SEM image of the electron bombarded sample S3
Figure 3.29. High magnification SEM image of sample S3 showing cracks at the point of electron
impingement
The surface morphology of electron bombarded samples S2 and S3 is alike and so a

detailed analysis of S3 alone is presented here. When electron beam is irradiated on the film,
cracks start developing at the point of impingement and spreads all along the film as shown in
Figure 3.28. The high magnification SEM image taken at the center of the film shows flakes
of different dimensions [Figure 3.29]. The cracks probably originate from the strain induced
by the difference in thermal expansion between the substrate and the film. A detailed
investigation on the crack surface of S3 was done by atomic force microscopy.
Figure 3.30. The Atomic force microscopy images (AFM) of electron bombarded samples S3 taken at
different magnifications
At the point of impingement, the surface is found to be smooth with large flakes
separated by fine cracks [Figure 3.30 (e)]. However, the AFM image of a single flake shows a
dip at the center with slight elevation along the edges [Figure 3.30 (f)]. In addition, high
magnification AFM image of individual flakes shows randomly dispersed nanoparticles
within the flake [Figure 3.30 (g)]. The surface crystallite size of the nanoparticles determined
from the height image was found to be ~ 25.0 nm [Figure 3.30 (h)].
3.8.5. Luminescence in electron bombarded samples
Apart from the changes in the physical properties, the electron bombardment also gives
rise to luminescence. The electron bombarded sample S3 shows intense red emission at λ =
619.0 nm. The photographs of the as -deposited (S1), colored (S2), and light emitting (S3)
samples are shown in Figure 3.31 (a) - (c). Interestingly, the light emission could not be
ascertained to the conventional cathodoluminescence expected from electron bombardment.
Rather the emission is found to be photoluminescence (PL) from the structural modifications
(tungsten oxide nanoparticles) and defect sites induced by electron bombardment. Figure 3.31
(d) shows the PL spectrum of S3 sample excited at 325.0 nm. The band to band emission was
observed at 357.0 nm; for the sake of clarity, the spectrum showing red emission alone is
presented here. The intense red emission is attributed to the transitions due to localized sites
induced by the presence of oxygen vacancies and defects in WO3 film [3.50]. All the above
results clearly indicate that the electron bombardment of WO3 thin films leads to significant
changes in the structural, optical and morphological properties.
Figure 3.31. The Photographs of the (a) as-deposited sample S1; b) electron bombarded sample S2; and
(c) electron bombarded sample S3. (d) PL spectrum of sample S3 excited at λ = 325 nm. (e) Outline for
the data storage application based on the coloration/luminescence of WO3 films by electron
bombardment
With the optimization of nanostructured electron emitters (nanorods, nanowires and

nanotubes), the coloration and luminescence induced by electron bombardment can be used
for data storage. The concept for data storage using electron bombardment is shown in Figure
3.31 (e). The colored /luminescent portion of the film can be treated as binary ‘1’ and the as-
deposited portion as ‘0’ or vice versa. As the electrons can be focused to a fine beam (~ a few
tens of nanometers), it is possible to achieve ultra high-density data storage using this
method, if optimized.
4. ZINC OXIDE FOR TRANSPARENT ELECTRONICS

(p-type CONDUCTING)
4.1. Summary
Zinc oxide (ZnO) thin films have attracted attention mainly due to their stability and cost
compared to the conventional tin doped indium oxide (ITO) thin films, it is contemplated that
ZnO may replace ITO. The basic physics for all these transparent conducting oxide (TCO)
remains the same. However, the difficulties encountered in producing p type transparent
conducting oxides (TCOs) in general, and ZnO, in particular, is posing a challenge to the
research community. There are several approaches followed to dope the ZnO for p- type
conduction. The present section summarizes the basic physics of p- type TCOs and the efforts
on the p- type doping in ZnO thin films.
4.2. Introduction
Transparent Conducting Oxide (TCO) thin films have a history of over sixty years and a
commercial production exceeding a few Billion US Dollars; and it is still a topic of intense
study in several research laboratories [4.1]. The present day industry mainly concentrates on
tin doped indium oxide (ITO) for TCO. But Zinc Oxide (ZnO), with abundance in nature
(which makes it cost effective) and higher motilities than ITO is a possible replacement for it
in near future [4.2].
It is well known that, TCO thin films are direct and wide band gap degenerate
semiconductors. The stiochiometric metal oxides are all insulators; the conductivity in TCOs
arise because of the oxygen vacancies (metal rich). Thus all TCOs are n- type
semiconductors. The conduction carrier concentration can be enhanced by replacing the metal
with a high valence element (for example, tin can replace indium); the replacement can be
achieved more efficiently if the sizes of the dopant element matches with the metal. The
transport mechanism in these TCO films, in general, is dominated by ionized impurity
scattering. More details of the carrier generation and transport mechanism have been
reviewed by Bel Hadj Tahar et al [4.3].
If p- type TCOs are prepared, a transparent p-n homojunction can be fabricated. A
transparent homojunction is a good building block for efficient and new class of
optoelectronic devices. There are a few reports on p type TCOs; however, the questions
related to reliable and repeatable process and stability are still to be worked out [4.4, 4.5].
The requirements for a p-type transparent conducting oxide (TCO), as may be found in
the literature, are: (i) chemical modulation of valence band (CMVB); and (ii) the electronic
configuration of the cationic species should preferably have a d10 s0 (d10 s2 systems did not
yield positive results so far) [4.6]. It may be noted that for metal oxides, the valence band is
commonly the oxygen 2p band. Even though doping in this band is not difficult, the mobility
of such holes is found to be quite low [4.6]. One way to obtain reasonable motilities in a p-
type TCO would be to dope the valence band of the cation d or s band. The cation s band may
offer the possibility of higher mobility than cation d bands and the 4d and 5d bands are still
better than the 3d bands. The question is how to achieve this selective doping in the
preparation of the material, particularly, in the thin film form.
The possible systems suggested for p- type TCOs are the materials with crystal structures
of Delaffosite (CuAlO2 and AgInO2) and of the Spinels (ZnO). It may be noted that it is very
difficult to grow thin films with these crystal structures. The fundamental issue in these p-
type TCO thin films is to realize a large concentration of holes in the structure. If the hole is
visualized as a localization of the covalent bond, then one has to control the oxygen reactivity
with the metal ion to support the localization. In Delaffosites, it is rather difficult as the
oxygen vacancies are inherent and contribute to electrons. However, holes may be realized in
these materials through the multivalence of the copper and indium in the Delaffosite
structures. The experimental results support this view on silver oxide [4.7] and on AgInO2
thin films prepared by reactive electron beam and DC Magnetron sputtering techniques [4.8].
The important result of the investigation is that the oxygen bonding to the indium and silver
(in both indium doped silver oxide and silver doped indium oxide) is very sensitive and the
localization of the valence bonds occur in a very narrow range of the growth conditions [4.9].
Thus the window for obtaining p- type behaviour seems to be very small and very sensitive to
the experimental parameters.
There are many questions related to the concepts of p – type TCO thin films. for
example: (i) a good and confirmed recipe to prepare a p-type TCO thin films, (ii) the origin of
the holes, (iii) the electrical transport mechanism (iv) the noise in the fabricated p-n junction
and (v) the intricacies of measurement technique to estimate the carrier (hole) concentration
and many more [4.10]. In these p- type TCO thin films there exists a vast potential for
condensed matter theorist: in identifying possible material systems, theoretical modelling of
the practically grown material with defects, calculation of band gaps and the nature of the
band gap (direct or indirect), estimation of maximum limit for hole density, mobility and high
optical transparency (in the visible spectrum) in the material systems, optimization methods
to enhance the performance (the density and mobility of holes) and possible operative
mechanism for these materials to possess radiation resistance property etc, [4.11].
4.3. Measurements on p- type TCOs
In all the TCO thin films, particularly in the multivalent metal ion systems (Silver,
Indium and copper) oxygen plays a very important role in realizing both n and p-type as
shown in our earlier experiments with Silver oxide AgO [4.7]. In order to achieve the p- type
conductivity in TCO films, one has to modulate the bond energy of the oxygen to the metal
ion, while doing so the oxygen vacancies may still remain in the lattice. Thus, the intrinsic
carrier concentration (ni) remains almost the same in the wide band gap oxide semiconductor
and the electron (n) and hole (p) product remains at (thermodynamic equilibrium)
n p = ni2 4.1
The holes are incorporated into the system at the expense of the electrons. At any given
instant of thermodynamic equilibrium, both electrons and holes will be present in the system.
For such a system, the measurement of Hall effect is rather tricky.
The recent technique: Quantitative Mobility Spectrum Analysis (QMSA)[4.12] is a

powerful method for analyzing bi-polar conduction (conduction by both electrons and holes
simultaneously). The basic principle of the QMSA measurement is that the mobilities of
electrons and holes depend upon the magnetic fields. It may be noted that even though the
contribution of electrons and holes for conductivity is independent of the applied magnetic
fields (magneto resistance), the resultant combined effects, due to velocity distribution, may
be dependant on the magnetic fields, provided, the magnetic field intensity is very high. Thus,
the magnetic field dependant Hall measurements can effectively separate the electron and
hole contributions. . The measurements done with QMSA are very well suited (and in fact
QMSA gives a vast amount of information on transport) for all semiconductors with bi-polar
conduction of carriers having reasonably high mobilities.
4.4. Progress in the Understanding of the Origin of Holes and p- type

Conduction
The origin of holes in these p- type TCO films is yet to be understood in detail. In the
opinion of the authors, Sanderson’s theory of partial ionic charge may be one of the
approaches to account for the holes in these p- type TCO thin films. As it is well known, the
existence of holes is due to the localization of valence bands in the material. Considering the
material systems: elemental metal oxides, delaffosites and spinels, it may be rather difficult,
from the conventional understanding, to visualize the presence of covalent bonds.
The feature of an atom which control its physical and chemical properties are: its
electronic configuration and the “effective nuclear charge” felt by the valence electrons. The
effective nuclear charge is the positive charge that would be felt by a foreign electron on
arriving at the periphery of the atom. Many atomic properties may be correlated with
effective nuclear charge; for example, (i) the ionization potentials gradually increase from left
to right across the periodic table, (ii) electron affinities become increasingly negative in the
same direction and (iii) electro negativities increase from left to right.
An important contribution to the understanding of bond formation is the principle of
electro-negativity equalization (proposed by Sanderson) which states that “when two or more
atoms initially different combine chemically, they become adjusted to the same intermediate
electro-negativity within the compound”. The partial charge is defined by the change in
electro-negativity undergone by an atom on bond formation to the change it would have
undergone on becoming completely ionic with charge + or – 1. This theory of coordinated
polymeric model of Sanderson is a bridge between the ionic and covalent extremes of bond
types. Though the proposed theory is empirical, experimental measurements have confirmed
the existence of partial ionic charges in many compounds (examples of calculations have
been given in reference 4.7).
With these ideas, one may realize the p- type conduction by controlling of oxygen
reactivity and the valence state of silver in the silver indium oxide thin films. The Ag: InO
thin films have been prepared by two techniques: electron beam evaporation and reactive DC
magnetron sputtering [4.8, 4.9, 4.13]. In the electron beam evaporation, the electron filament
current is varied keeping the oxygen pressure in the growth chamber constant. In DC
magnetron sputtering, the oxygen flow rate is varied keeping the magnetron power density
constant. In both the techniques, the p - type conductivity with reasonable transmission in the
visible region (400 -900 nm) has been achieved. In order to notice the narrow window of the
growth / process parameters, nu*merical data are presented in Tables 1 and Table 2
respectively for the reactive electron beam evaporated and reactive DC Magnetron sputtered
silver indium oxide thin films.
Table 4.1. Electrical and optical properties of silver doped indium oxide thin films
prepared by reactive electron beam evaporation [4.8]
current (mA)
(cm2 v−1 s−1)

Evaporation
(cm3/C) and
carrier from
rate (nm/s)
coefficient
Eg (eV) hν
Refractive
632.8 nm)
Thickness
temp.(°C)
hot probe
Substrate
Filament
index (at
mobility
function
Type of
type Of
th d
(αhν)2
charge
charge
carrier
Work
(nm)
(eV)
Hall
Hall
50 10 0.05 192.0 1.208 1.49 1.69 −1.47×10−3 P 4.686
15 0.89 186.0 1.201 2.00 0.85 −3.95×10−3( N 4.395
20 2.67 157.0 1.192 − 5.60 −1.2210−3 P 4.672
−4(N)
30 5.33 154.0 1.196 – 7.45 −0.97×10 P 4.763
−3(N)
40 16.0 156.5 1.188 1.73 0.89 −6.13×10 N 4.500
200 10 0.05 155.5 1.217 0.96 6.03 −1.69×10−4(N) P 4.720
−3(N)
15 0.89 160.0 1.177 1.69 1.29 −2.28×10 N 4.545
−4(N)
20 2.67 164.0 1.164 – 3.13 −0.74×10 P 4.623
−4(P)
30 5.33 147.0 1.205 – 1.52 0.93×10 P 4.593
40 16.0 163.5 1.235 1.21 1.59 −1.2510−4(N) P 4.541
250 10 0.05 166.5 1.159 0.76 1.63 5.99(P) P 4.828
−3(N)
15 0.89 155.0 1.192 1.98 3.02 −1.38×10 N 4.683
−4(N)
20 2.67 146.5 1.184 – 14.77 −0.78×10 P 4.740
30 5.33 138.5 1.185 – 6.36 −1.59×10−4(N) P 4.651
−4(N)
40 16.0 161.5 1.191 0.81 4.58 −0.87×10 P 4.633
4.5. Systems with Spinel Structure
As it is well known, the Spinel belongs to cubic system; the unit cell contains 32 cubic
close packed Oxygen ions, 08 divalent cations of A and 16 trivalent cations of B. There are
equivalent positions in this cell for eight tetrahedral cation sites and for sixteen cation sites.
The total number of tetrahedral and octahedral vacancies between the oxygen layers in the
unit cell are 64 and 32 respectively. Thus, 7/8 of the tetrahedral and ½ of the octahedral holes
remain vacant. The Spinel structure contains rutile chains running along <110> directions
giving a free path to electrons (high electronic motilities). The spinel lattice offers excellent
possibilities for electron doping. The material systems belong to the spinel are : ZnAl2O4,
ZnGa2O4, ZnIn2O4, CdAl2O4, CdGa2O4 and CdIn2O4. For a typical ZnAl2O4 the composition
of Zn and Al are : 29.4 % and 55.6 % respectively and the oxygen concentration is about 30 -
35%; the crystal system is Isometric Hexoctahedral with Fd3m Space Group and has eight
formula units per unit cell. These materials are nonmagnetic.
Table 4.2. Electrical and optical properties of reactive DC Magnetron sputterd Silver
indium oxide thin films.
Refrac-tive
Resistivity
(αhν)2~hν
(cm2v-1s-1)
α2~hν plot
flow rates
at 40 watt
cm3/coul)
Ohm.cm)
index (at
mobility
function
Eg (eV)
Oxygen
coeffic-
Eg(eV)
(sccm)
Thick
Work
632.8
(nm)
(10-2
(10-3
(eV)
Nm)
Hall
Hall
ness
plot
ient
0.00 215.0 0.28 -9.137 3.256 4.552 1.133 - 1.79
0.54 169.2 0.21 -6.345 2.98 4.478 1.194 0.85 1.78
1.09 199.2 7.25 -43.70 0.60 4.644 1.159 1.39 1.83
1.71 149.6 125.52 +25.86 0.206 4.694 1.220 1.36 1.83
2.36 186.5 - - - 4.767 1.132 0.50 1.32
2.86 214.2 - - - 4.783 1.151 0.55 2.92
Magnetron
power
(Watts) at
1.09sccm
40 199.2 7.25 -43.7 0.60 4.644 1.159 1.39 1.83
50 141.7 5.14 -37.9 0.73 4.635 1.188 1.32 2.34
60 206.7 3.39 -21.1 0.62 4.577 1.175 1.37 2.47
70 144.0 0.14 -5.6 4.07 4.575 1.205 1.23 1.70
80 200.6 0.12 -1.2 1.04 4.549 1.181 1.11 1.12
4.6. Physical Properties of ZnO
Zinc oxide (ZnO) is a II-VI transparent conducting oxide (TCO), with a wide band-gap of
~3.3 eV at room temperature. ZnO has a very high excitonic binding energy of 60 meV (25
meV for GaN), this property makes it so important with respect to the optoelectronic devices
performance, even well above the room temperature (RT). A stoichometric ZnO has an empty
zinc 4s-band and a filled valence 2p-band resulting in a band gap of 3.3 eV (the two valence
4s electrons of zinc transfer to the oxygen 2p-band). The electrical conductivity of ZnO is due
to the intrinsic defects and / or external doping. In the defect structure, the ZnO is not in a
stoichometric ratio and there is excess Zn or low amount of O. One of the two valence
electrons of the Zn interstitial atom can be easily ionized and act as donors. Usually in a
defect structure of ZnO, an increase in O does not give rise to free holes: the acceptor level
corresponding to oxygen defect centres is too far from the valence band to be ionized at room
temperature or the number of interstitial Zn is always higher than O. The characterization of
native point defects in ZnO is still in detailed investigation. For example, experimental
evidence for ZnO with an excess of Zn is inconclusive as to whether the dominant defects are
metal interstitials or oxygen vacancies. This information is essential to understand the
behaviour of the material and to tailor its numerous applications. The first-principles pseudo-
potential method was used to determine the electronic structure, atomic geometry, and
formation energy of native point defects in ZnO [4.11]. The results show that both the Zn and
O vacancies are the relevant defects in ZnO. It is found that the most abundant native defects
to be Zn and O vacancies depending on the Zn partial pressure.
In its non-stoichometric form, it is an n-type semiconductor, and could be grown with

resistivity as small as ~10-6 ohm m, i.e., a resistivity only 1-2 orders of magnitude larger than
that of metals. Due to its large (3.37 eV) and direct band gap, ZnO is a potential competitor
for GaN-based light-emitting devices in the blue spectral range. However, like for other wide
band gap semiconductors controlled p-type doping is the challenge. As grown and undoped
ZnO is n-type conducting. To achieve the desired p-type conduction, one requirement is to
suppress the residual donors and to avoid any deep level defects which hinder the activity of
the potential p-type dopants. On this way a clear atomistic identification of the electrical
active species in the material is helpful. These features make ZnO a potential candidate
material for use in optoelectronic devices, such as blue LED's and laser diodes (LD's)[4.14].
ZnO has interesting luminescent properties [4.10].The reason for the high stability of
ZnO in the presence of reducing plasma and high temperature has been attributed to its high
surface binding energy. The important physical properties of ZnO are summarized in these
two tables [4.4, 4.14]
Table 4.3. Optical properties of ZnO
ZnO Optical Properties

Bandgap (eV) Room T: ~3.3eV @ 300K Low T: 3.437eV @ 2K
Indices of n-=
2.008 (IR)
Refraction
n|| = 2.029 (IR)
Relative Static E-= 7.77] E||= 8.91
Dielectric
Constants
(measured on Optical E-= 3.70] E||= 3.78
bulk)
Table 4.4. Physical properties of ZnO
Parameter Symbol Literature value

Hole effective mass m*p 0.64 × (9.1095 × 10–31) kg
High-frequency dielectric constant ε∞ 3.72 × (8.8542 × 10–12) (F m–1)
Static dielectric constant ε0 8.12 × (8.8542 × 10–12) (F m–1)
Longitudinal lattice constant C 5.207 Å
Piezoelectric coefficient P 0.21
Longitudinal optical phonon energy ω 72 meV
Acoustic deformation potential Ec 15 eV
Debye temperature T 837 K
Longitudinal elastic constant cl 1.6 × 1011 Pa
Thin transparent and electrically conducting ZnO films were deposited using different
deposition techniques, like sputtering techniques [4.15], chemical vapour deposition (CVD)
[14.16], molecular beam epitaxy (MBE) [4.17], filtered vacuum arc deposition (FVAD)
[4.18], and more [4.19]. In contrast to n-type ZnO, p-type ZnO films can be produced only by
doping. The major challenges in realizing p type ZnO are compensation of donor level defects
in system by the acceptors and the activation of the acceptor dopant. In some specific cases,
even if the acceptor level is formed, it can’t be activated if it is deep in nature.
The various methods used in growing p-type doped ZnO include standard doping, usually
with nitrogen [4.20], co-doping [4.21, 4.22] or diffusion doping from the substrate material or
an interface layer of the dopant, using an excimer laser [4.23]. Generally, the grown p-type
films have resistivities of (1-400) x 10-3 Ω m, mobility of the order of (0.1-50) x 10-4
cm2/(V·s), and hole concentration of ~1021-1027 m-3. Ohara et al. reported on p-type ZnO thin
films obtained by Sb doping, using the excimer laser diffusion method [4.24]. The ZnO layers
were epitaxially grown on Si substrates (thickness ~50 nm), and p-type films with resistivity
of 8·10-5 Ω m, mobility of 1.5x10-4 m2/ (V·s), and hole concentration of 5x1026 m-3 were
reported, using gold electrodes as contacts. No data on the composition or microstructure of
the doped films were reported. Kobayashi et al [4.25] have predicted that nitrogen is a
candidate for producing a shallow p- type dopant in ZnO. The ab initio calculations of
electronic band structure of ZnO by Yamamoto et al [4.26] have shown that simultaneous co-
doping using reactive donor co-dopants: Al, Ga and In, not only enhances the incorporation of
N acceptors but also gives raise to a shallow Nitrogen acceptor levels in a band gap in p- type
doped ZnO crystals. Joseph et al [4.27] found that the N doping is effective only with N2O
through an ECR plasma source (but not with N2 gas). p- type ZnO films have been prepared
at 650o C by chemical vapour deposition by Minegishi et al [4.28] the hole concentration
obtained is 1.5 x 1016 /cm3 with a mobility of 12 cm2/V.s. By using As a dopant and
employing pulsed laser deposition (PLD), Ryu et al [4.29] have synthesized p – type ZnO.
Elina Kaminska et al [4.30] reported the preparation of p-type ZnO by oxidation of thin
films zinc nitride. Nitrogen incorporation is seen as the most prospective way to obtain p-type
conductivity. The starting material was grown by rf reactive magnetron sputtering from high
purity Zn target, in Ar-N2 plasma. Crucial point during the deposition was to adjust both the
N2/Ar ratio and the working pressure so as to deposit Zn3N2 without any metallic Zn
inclusions. For oxidation, samples were furnace annealed in O2 flow Quartz and sapphire as
well as lattice-matched GaN, ZnO and ZrB2 were used as substrates. The effect of the growth
direction - determined by growth conditions - on the structural perfection of the resultant ZnO
films as well as on efficiency of nitrogen incorporation was studied. Results of x-ray
diffraction, secondary ion mass spectrometry and optical transmission measurements of the
oxidized layers, indicating that formation of ZnO did take place. This work reports on the
appropriate conditions for the growth of p-type ZnO with Hall effect mobility of 20 cm2/Vs
and carrier concentration of 2x1015 cm-3. Further increase of p-type conductivity up to 5x1017
cm-3 with mobility of 5 cm2/Vs was obtained by additional doping the starting material with
Cr. The window for obtaining p- type behaviour in ZnO seems to be very small and very
sensitive to the experimental parameters.
4.7. ZnO Homojunction
In ZnO, the n-type doping is relatively easy to accomplish with elements such as Al, Ga
or In. On the other hand, reliable p-type doping is very difficult to achieve. Usually group I
and group V elements are used as acceptors. So far, the most common acceptor has been
nitrogen because it readily substitutes for oxygen and may not distort the lattice significantly.
Besides nitrogen, phosphorous, lithium and arsenic have been successfully used to fabricate
p-type ZnO. In addition, there are a few reports on intrinsic p-type behaviour in intentionally
undoped ZnO, in which the oxygen pressures were optimized. For example, intrinsic p-type
ZnO thin films have been deposited by plasma-assisted metal-organic chemical vapour
deposition. The optimal results reported are, resistivity of 12.7 ohm cm, Hall mobility of 2.6
cm2 /V s, and a hole concentration of 1.88 x 1017 cm−3. The origin of intrinsic p-type
behaviour has been ascribed to the formation of zinc vacancy and some complex acceptor
[4.31]. It is also conjectured that the oxygen chemical potential is enhanced by virtue of
oxygen plasma, which can lower the formation energy of some acceptor defect, such as zinc
vacancy, and thus accounts for the p-type conductivity. ZnO homojunctions deposited by
three different deposition techniques are referred here.
Xu et al fabricated [4.32] ZnO homojunction light-emitting diode by metal organic
chemical vapour deposition. Using NO plasma, p-type ZnO thin films were grown on n-type
bulk ZnO substrates. The as-grown films on glass substrates showed hole concentration of
1016–1017 cm−3 and mobility of 1–10 cm2 V−1 s−1. The typical ZnO homojunction showed
rectifying behavior with a turn-on voltage of about 2.3 V and electroluminescence at room
temperature with band-to-band emission at I=40 mA and defect-related emissions in the blue-
yellow spectrum range.
Jiao et al [4.33] fabricated ZnO p-n junction light-emitting diode on a-plane Al2O3
substrate by plasma-assisted molecular-beam epitaxy. NO plasma activated by a radio
frequency atomic source was used to grow the p-type ZnO layer of the LED. The current-
voltage measurements at low temperatures showed a typical diode characteristic with a
threshold voltage of about 4.0 V under forward bias. With increasing temperature, the
rectification characteristic was degraded gradually, and faded away at room temperature.
Electroluminescence band of the ZnO p-n junction LED was located at the blue-violet region
and was weakened significantly with increase of temperature.
Cao et al [4.34] fabricated ZnO p-n junctions by a simple method of sol-gel spin coating.
N-doped and In-N-co-doped ZnO films have been fabricated on quartz glass substrate by sol-
gel spin coating. Their p-type conductivities were characterized by the Hall measurements,
revealing low resistivities of the order of 10−1 Ω cm. Thin-film junctions comprising an
undoped ZnO layer and a N-doped ZnO layer displayed the typical rectifying characteristics,
suggesting formation of p-n homojunctions at the interfaces. However, it is the inexplicit p-
type doping mechanism as well as the stability and reproducibility problems that become the
constraint factor (as already indicated) in the development of ZnO devices.
4.8. Measurement of Charge Density by Hall Effect in p- type TCO Thin

Films
The conventional Hall effect experiment, based on the basic principle of the deflection of
free carriers in a magnetic field produces the Hall field in a direction perpendicular to the
applied magnetic field and current (passing through the sample). The Hall field (Ez) produced
account for the majority carriers in the sample
Ez = R Ix Bz 4.2
where Ix is the current passing through the sample, Bz is the magnetic field and Ey is the Hall
field produced (the subscripts indicate the Cartesian coordinate directions) and R is the Hall
coefficient related to the majority carrier concentration as
R = ± (1/ nq) 4.3
n is the density of majority carriers; +ve sign is for holes and –ve sign is for electrons.
While deriving the expressions (4.2) and (4.3):
(i) the simultaneous contribution of electrons and holes is not considered

(ii) the effect of distribution of velocities is not taken into account
(iii) it is inherently assumed that the constant energy surfaces (in momentum space) are
assumed to be spherical and hence no anisotropy in the effective mass of charge
carriers
(iv) if there is any anisotropy in the effective masses of charge carriers, the density of
states (DOS) function and averaging transport properties over velocities are to be
modified.
When both electrons and holes are contributing to conductivity in almost competing
concentrations, both the type of carriers deflect on to opposite faces of the sample in the
conventional Hall experiment; under such experimental conditions, the results of the Hall
effect are to be measured with great care and interpretation is to be made with utmost caution.
In most of the cases of TCO thin films, the resistivity of the sample is high, then the
measurement of Hall coefficient may give erroneous results (See Table 4.1 and 4.2). It may
be recalled that the main difference between the well understood p-type doping in Silicon and
p- type doping in TCO is that the TCO contains inherently a large number of free electrons
due to oxygen deficiency . The other difference is that the Hall mobilities in TCO films are
one (or two) order of magnitude less than that of Silicon.
The fundamental equation governing the Hall Coefficient when both electron and holes
(non extrinsic case) are present in the sample and taking into consideration the distribution of
velocities (involving the solution of Boltzmann equation) and if the dominant scattering
process is acoustical mode phonon scattering , is given by [4.35].
RH = ( -3π/8ce)[ (p μp2 – n μn2 ) / (p μp + n μn )2 ] 4.4
where μn and μp are the mobilities of the electrons and holes respectively, to be determined
independently from other experiments.
If the number densities for electrons and holes are comparable, assuming the hole
mobility being smaller (because of higher effective mass), the Hall coefficient will be
considerably affecting the carrier concentration values, most of the times leading to intrinsic
regime, while measuring the Hall coefficient as a function of temperature in mixed
conduction process, the intrinsic regime appears more often. The intrinsic property has been
reported by Subrahmanyam et al. for In + Ag oxides [4.13]. This is one of the reasons (apart
from the contacts to measure the Hall voltage) for several workers to encounter difficulties in
the accurate measurement of Hall coefficient in the p – type TCO films.
While measuring the Hall coefficient at low temperatures on the bipolar conduction
semiconductors, the behaviour of mobility and the impurity scattering add to the complexity
of the measurement.
Some investigators followed SIMS (Secondary Ion Mass Spectrometry) to estimate the
dopant concentration; however, SIMS gives the dopant (acceptor or donor) concentration in
the layers of the materials, but not the active centres that effectively contribute to electrical
transport.
4.9. The Future Trends of the TCO Thin Films
The flexible electronics is required in several applications where weight and portability
are the demands. Presently, the concept of flexible electronics is to develop the conventional
circuit on a flexible plastic substrate. The recent advancement in flexible electronic displays
has been revealed at Philips [4.36]. There are many projects aiming to develop "electronic
paper". Such a display could, for example, be used create a fully updatable newspaper which
could rolled up into a coat pocket. Flexible displays could also be used to create new mobile
phones and other easily collapsible gadgets.
Comparatively little work has been reported on the properties of ITO and AZO (Al doped
ZnO)thin films on plastic substrates even though roll-to-roll vacuum deposited ITO films are
commercially available. With the recent developments in the high refractive index plastic
ophthalmic lenses, the mechanically hard ITO and AZO sputter coatings (5-7 GPa) may have
a vast potential in scratch resistant and anti reflection coatings. Process temperatures with
common plastic substrates are < 100 ºC, typically near room temperature (RT), which results
in amorphous and poor quality of ITO. Typically, the resistivity, ρ, for ITO thin films made
with a RT deposition processes is ~ 6 x 10–4 Ω–cm or higher. This high resistivity (low
conductivity) is not suitable for many applications such as a transparent electrode for Organic
Light Emitting Diode (OLED) displays deposited on flexible plastic film substrates. For
several reasons, including performance, cost and availability, finding a TCO replacement for
ITO is currently an area of very active research. However, replacement candidate TCO, e.g.,
Al doped ZnO (AZO) and Ga doped ZnO (GZO), exhibit similar degraded electrical
performance when deposited near RT. Additionally, mechanical issues with TCO thin films
on plastic substrates also are of key importance. For example, the critical strain for ITO is ~
1.5%, which limits the useful film thickness and other processing and application parameters.
The effect of the deposition method selected, and the choice of process conditions, on the ITO
resistivity are to be evaluated. The efficacy of various choices in improving the electrical
performance of RT deposited ITO (TCO) will be the focal theme of many investigations to
come.
For gas sensing applications, the nanostructured materials are recognized as essential for
achieving high gas sensitivity. Numerous processing schemes have been tested successfully,
albeit only on a laboratory scale. Processing techniques should be able to provide the desired
oxide composition with specific dopant and requiring the least number of processing steps. In
the processing of nanostructured oxides, more fundamental work is needed to understand the
role of nanostructured oxide materials on gas adsorption and conductivity. In general, a
detailed understanding is required on the nano structured TCO thin films for not only gas
sensing but also in electromagnetic shielding and radiation resistance applications.
ACKNOWLEDGEMENTS
One of the authors (AS) gratefully acknowledges the support received from Professor
Eberhard Schultheiss, Director, Fraunhofer Institute fur Elektronenstrahl und Plasmatechnik
(FEP), Dresden, Germany and DAAD, Germany.
Notes 1
Present Day Technology for Lung Assisted Devices
Four principal development efforts have tackled the challenge of intravascular artificial
lungs since the 1980s. Mortensen and colleagues at CardioPulmonics, Inc. (Salt Lake City,
UT) developed the IVOX, the only intravascular artificial lung that has undergone human
clinical trials.The IVOX consisted of a bundle of crimped hollow fiber membranes joined at
the distal end to the inner lumen of a dual-lumen gas conduit, and at the proximal end to the
outer lumen of the gas conduit, which led outside the body to a console for providing sweep
gas flow through the fibers. The crimped fibers of the IVOX helped minimize fiber clumping
in the vena cava and also helped disturb blood flow and diffusional boundary layers on fiber
surfaces to improve overall gas exchange permeance. A total of 160 patients with severe
acute respiratory distress were studied in applications that lasted up to 28 days of support. The
clinically tested IVOX ranged from 0.21 m2 to 0.51 m2 in membrane area, and the average
rates of O2 and CO2 transfer accomplished in the trials ranged from 40 to 70 ml/min, or about
20–30% of baseline metabolic needs. The IVOX demonstrated that intravascular artificial
lungs can be implanted within the vena cava and perform for extended periods without
significant complications in situ (for example, from blood thrombosis).
The implantable Intrathoracic Artificial Lung (ITAL) under development at Northwestern
University focuses on resting the lung in acute respiratory failure and as a bridge-to-lung
transplantation in chronic lung failure. Mathematical models were developed to estimate the
required surface area for 200 ml/min of oxygen transfer at a blood flow rate of 5 l/min with a
pressure drop of less than 15 mm Hg.
The BioLungTM total artificial lung under development at MC3, Inc. (Ann Arbor, MI)
and the University of Michigan is intended for complete respiratory support as a bridge to
transplant for 1–6 months.
A paracorporeal total artificial lung for chronic respiratory support (Chronic Artificial
Lung, or CAL) is under development at the University of Maryland. The CAL is intended as a
bridge-to- transplant device with the goal of 21-day support of basal metabolic needs using a
device less than 0.5 m2 in fiber membrane area. The CAL uses active mixing from a rapidly
rotating disc made of microporous hollow fiber membranes that enhance gas exchange by
increasing blood flow velocity past fiber surfaces and reducing diffusional boundary layers.
Hb + O2 <-> HbO2
HbO2 + O2 <-> Hb(O2)2
Hb(O2)2 + O2 <-> Hb(O2)3

Hb(O2)3 + O2 <-> Hb(O2)4
Summary Reaction
Hb + 4O2 -> Hb(O2)4
REFERENCES
[A1] Metal Oxides: Chemistry and Applications (Chemical Industries), Fierro, J. L. G. (CRC
Press, 2005).
[A2] The Surface Science of Metal Oxides, Victor, E; Henrich, PA. Cox, Cambridge
University Press, 1996.
[A3] The Defect Chemistry of Metal Oxides, Smyth, DM, Oxford University Press, 2000.
[A4] Parkin SP; et al. Giant tunnelling magnetoresistance at room temperature with MgO
(100) tunnel barriers, Nature Materials, 2004, 3, 862.
[1.1] Federspeil, WJ; Svitek, RG. Lung, Artificial: current research and future directions,
Encyclopaedia of Biomaterials and Biomedical Engineering, Ed: Wnek, G E and.
Bowling, G.L. 2008, pp 1673-1682.
[1.2] Titanium in Medicine, Ed by Bruneyye, DM; Tengvall, P; Textor, M; Thomas, P.
Springer, 2001.
[1.3] Fujishima, A; Honda, K. Electrochemical Photolysis of Water at a Semiconductor
Electrode, Nature, 1972, 238, 37.
[1.4] Diebold, U. The surface science of titanium dioxide, Surf. Sci. Reports, 2003, 48, 53.
[1.5] Zubkov, T; Stahl, D; Thompson, TL; Panayotov, D; Diwald, O; Yates JT. Jr.,
Ultraviolet light-induced hydrophilicity effect on TiO2(110)(1x1). Dominant role of
the photooxidation of adsorbed hydrocarbons causing wetting by water droplets,
J. Phys. Chem. B, 2005, 109, 15 454.
[1.6] Hashimoto, K; Irie, H; Fujishima, A. TiO2 Photocatalysis: A Historical Overview and
Future Prospects, Jpn J Appl Phys Part 1, 2005, vol 44, 8269.
[1.7] Akihiko, K; Yugo, M. Heterogeneous photocatalyst materials for water splitting, Chem.
Soc. Rev., 2009, Vol 38, 253.
[1.8] Subrahmanyam, A; Arockia Doss, T; Paul Ramesh, T. Studies on the Oxygenation of
human blood by Photocatalytic action, Artificial Organs, 2007, vol 31, pp 819.
[1.9] Subrahmanyam, A; Ramesh, TPJ; Ramakrishnan, N. Oxygenation of human blood
using photocatalytic reaction, ASAIO Journal, 2007, vol 53 (4), pp 434.
[1.10] Martin, PM; Monzyk, BF; Burckle, EC; Bbusch, JR; Gilbert, RJ; Dasse, KA. Progress
towards development of a photolytic artificial lung, Materials science & engineering.
B, Solid-state materials for advanced technology, 2005, vol. 119, 246.
[1.11] K.A. Dasse, B.F.Monzyk, E.C.Burckle, J.R.Busch, R.J.Gilbert, Dvelopment of a
photolytic artificial lung: preliminary concept validation,., ASAIO J. vol 49, 556
(2003).
[1.12] Artificial Infusion of Oxygen into human blood by photocatalytic reaction” Indian patent
Application No: 427 CHE, 2006 dated 10 March 2006.
[1.13] Photolytic artificial lung, US Patent 6866755B2 (2005).
[1.14] Photolytic artificial lung, US Patent 7485261 B2 (2009).
[1.15] Oxygen generation in whole blood by photolytic activation, US Patent 7399717B2
(2008).
[1.16] Artificial Pulmonary capillary, US Patent 2004/ 0243051A1 (2004).
[1.17] Marino, PL. The ICU Book, 2 Edition, Williams & Wilkins, Baltimore MD, 1998, pp
20-21.
[1.18] Visible and near infrared absorption spectra of human and animal haemoglobin,
W.G.Zijlstra, A. Buursma and O.W.van Assendelft (published by VSP Nederland,
2000.
[2.1] Antelman, MS. US Patent, 1993, 5.211.855.
[2.2] Tominaga, J., “The application of silver oxide thin films to plasmon photonic devices”
J. Phys.: Condens. Matter., 2003, 15, R1101-R1122.
[2.3] Peyser, LA; Vinson, AE; Bartko, AP; Dickson, RM. Photoactivated Fluorescence
from Individual Silver Nanoclusters, Science, 2001, 291, 103.
[2.4] Fujimaki, M; Awazu, K; Tominaga, J. Surface-enhanced Raman scattering from Ag
nanoparticles formed by visible laser irradiation of thermally annealed AgOx thin films,
J. Appl. Phys., 2006, 100, 074303-1.
[2.5] Barik U K, Srinivasan S, Nagendra C L, and Subrahmanyam A, “Electrical and optical
properties of reactive DC magnetron sputtered silver oxide thin films: role of oxygen”,
Thin Solid Films, 2003, 429, 129 -134.
[2.6] JCPDS card 76-1489.
[2.7] JCPDS card 75-0969.
[2.8] JCPDS card 84-1108.
[2.9] Bohern, CF; Huffman, DR. Absorption and scattering of light by small particles, John
Wiley, New York, 1983.
[2.10] Anderson, PC; Rowlen KL. “illiant Optical Properties of Nanometric Noble Metal
Spheres, Rods, and Aperture Arrays”, Appl. Spectrosc., 2002, 56, 24A-135A.
[2.11] Nicewarner-Pena, SR; Freeman, RG; Reiss, BD; He, L; Pena, DJ; Walton, ID; Cromer,
R; Keating, CD; Natan, MJ. “Submicrometer Metallic Barcodes”, Science, 2001, 294,
137-141.
[2.12] Mooradian, A. “Photoluminescence of metals”, Phys. Rev. Lett., 1969, 22, 185-187.
[2.13] Apell, P; Monreal, R; Lundqvist, S. “Photoluminescence of Noble Metals”, Phys. Scr.,

1988, 38, 174-179.
[2.14] Newman, DM; Panchmatia, P. “Nano silver oxide: a rewritable optical recording
medium”, IEE Proc.-Sci. Meas. Technol., 2003, 150, 214-217.
[2.15] Subrahmanyam, A; Suman Kumar, P. “Nano silver oxide thin films for optical
memories : New results”, IETE Journal of Research, 2006, 55, 365-368.
[2.16] Al-kuhaili, MF., “Characterization of thin films produced by the thermal evaporation of
silver oxide” J. Phys. D: Appl. Phys., 2007, 40, 2847-2853.
[2.17] Peyser, LA; Lee, TH; Dickson, RM. “Mechanism of Agn Nanocluster Photoproduction
from Silver Oxide Films”, J. Phys. Chem. B, 2002, 106, 7725-7728.
[2.18a]Varkey, AJ; Fort, AF. “Some optical properties of silver peroxide (AgO) and silver
oxide (Ag2O) films produced by chemical-bath deposition”, Sol Energy Mater.Sol.
Cells, 1993, 29, 253-259.
[2.18] Chih Min Chuang, Wu, Wei Fang Su, MC; Cheng, KC; Chen, YF., “High intensity
fluorescence of photoactivated silver oxide from composite thin film with periodic
array structure”, Appl. Phys. Lett., 2006, 89, 061912-1- 061912-3.
[2.19] Dellasega, D; Facibeni, A; Fonzo, FD; Russo, V, Conti, C; Ducati, C; Bassi, AL;
Bottani, CE.,” Nanostructured high valence silver oxide produced by pulsed laser
deposition”, Appl. Surf. Sci., 2009, 255, 5248-5251.
[2.20] Ravi, N. Chandra Raju, Jagadeesh Kumar, K; Subrahmanyaml, A. “Physical Properties
of silver oxide thin films by pulsed laser deposition: effect of oxygen pressure during
growth”, J. Phys. D: Appl. Phys. (Manuscript under publication).
[2.21] Deb, A; Chatterjee, AK, “The electronic structure and chemical bonding mechanism of
silver oxide”, 1998 J. Phys.: Condens. Matter, 1998, 10, 11719- 11729.
[2.22] West, AR. Solid State Chemistry and its Applications, Revised Edition, John Wiley,
New York, 1984, p.290.
[2.23] Abdelhamid Maali, Thierry Cardinal, Mona Treguer-Delapierre, “Intrinsic
Fluorescence from individual silver nanoparticles”, Physica E, 2003, 17, 559 – 560.
[2.24] Burgel, C; Mitric, R; Bonacic –Koutecky, V., “Emissive properties of silver particles at
silver oxide surface defects”, Appl. Phys. A, 2004, 82, 117-123.
[2.25] Gleitsmann, T; Stegemann, B; Bernhardt, TM., “Femto second- laser –activated
fluorescence from silver oxide nanoparticles”, Appl. Phys. Lett., 2004, 84, 4050-4052.
[2.26] Neal, T D; Okamoto, K; Scherer, A., “Surface plasmon enhanced emission from dye
doped polymer layers”, Opt. Express, 2005, 13, 5522-5527.
[2.27] Lynn Anne Capadona, “Photoactivated Fluorescence of small silver nano clusters and
their relation to Raman spectroscopy” A Thesis, Georgia institute of Technology, July,
2004, [http://etd.gatech.edu/theses/available/etd-07022004-120153/].
[2.28] Gryczynski, I; Malicka, J; Jiang, W; Fischer, H; Chan, WCW; Gryczynski, Z;
Grudzinski, W; Lakowicz, JR., “surface-plasmon-coupled emission of quantum dots”,
J. Phys. Chem. B, 2005, 109(3), 1088-1093.
[3.1] Granqvist, CG. Handbook of Inorganic Electrochromic Materials, Elsevier,
Amsterdam, 1995, pp. 139.
[3.1a] Granqvist CG. Handbook of Inorganic Electrochromic Materials, Elsevier, Amsterdam,
1995, pp. 146
[3.2] Granqvist, CG; Azens, A; Heszler, P; Kish, LB; Österlund, L. Nanomaterials for
benign indoor environments: Electrochromics for “smart windows”, sensors for air
quality, and photo-catalysts for air cleaning, Sol. Energy Mater. Sol. Cells, 2007, 91,
355-365.
[3.3] Faughnan, BW; Crandall, RS; Heyman, PM. The electrochromic properties of tungsten
trioxide film RCA Rev., 1975, 36, 177.
[3.4 Lee, SH; Cheong, HM; Tracy, CD; Mascarenhas, A; Benson, DK; Deb, SK. Raman
spectroscopic studies of electrochromic a-WO3, Electrochim. Acta, 1999, 44,
3111-3115.
[3.4a] Lee, SH; Cheong, HM; Tracy, CE; Mascarenhas, A; Czanderna, AW; Deb, SK.
Electrochromic coloration efficiency of a-WO3-y thin films as a function of oxygen
deficiency, Appl. Phys. Lett., 1999, 75, 1541-1543.
[3.5] Berggren, L; Niklasson, GA. Optical charge transfer absorption in lithium-intercalated
tungsten oxide thin films, Appl. Phys. Lett., 2006, 88, 081906-1-081906-3.
[3.6] Subrahmanyam, A; Karuppasamy, A; Suresh kumar, C. Oxygen-sputtered Tungsten
oxide Thin Films for Enhanced Electrochromic Properties, Electrochem.Solid-State
Lett., 2006, 9, H111-H114.
[3.6a] Subrahmanyam, A; Karuppasamy, A. Optical and electrochromic properties of oxygen
sputtered tungsten oxide (WO3) thin films, Sol. Energy Mater. Sol. Cells, 2007, 91,
266-274.
[3.7] Asanuma, T; Matsutani, T; Liu, C; Mihara, T; Kiuchi, M. Structural and optical
properties of titanium dioxide films deposited by reactive magnetron sputtering in pure
oxygen plasma, J. Appl. Phys., 2004, 95, 6011-6016.
[3.8] Denesuk, M; Uhlmann, DR. Site-Saturation Model for the Optical Efficiency of
Tungsten Oxide-Based Devices J. Electrochem. Soc., 1996, 143, L186-L188.
[3.9] Subrahmanyam, A; SureshKumar, C; Muthu Karuppasamy, K. Sol. Energy Mater. Sol.
Cells, 2007, 91, 62.
[3.10] Granqvist, CG. Electrochromic tungsten oxide films: Review of progress 1993- 1998,
Sol. Energy Mater. Sol. Cells, 2000, 60, 201.
[3.11] Hashimoto, S; Matsuoka, H. Mechanism of electrochromism for amorphous WO3 thin
films, J. Appl. Phys., 1991, 69, 933.
[3.12] Hashimoto, S; Matsuoka, H. Mechanism of electrochromism for amorphous WO3 thin
films, J. Appl. Phys., 1991, 69, 933-937.
[3.13] Berggren, L; Ederth, J; Niklasson, GA. Electrical conductivity as a function of
temperature in amorphous lithium tungsten oxide, Sol. Energy Mater. Sol. Cells, 2004,
84, 329-336.
[3.14] Lee, K; Seo, WS; Park, JT. Synthesis and optical properties of colloidal tungsten oxide
nanorods, J. Am. Chem. Soc., 2003, 125, 3408-3409.
[3.15] Li, Y; Bando, Y; Golberg, D. Quasi-Aligned Single-Crystalline W18O49 Nanotubes and
Nanowires Adv. Mater., 2003, 15, 1294-1296.
[3.16] Galléa, F; Li, Z; Zhang, Z. Growth control of tungsten oxide nanostructures on planar
silicon substrates Appl. Phys. Lett., 2006, 89, 193111-1 – 193111-3.
[3.17] Zhou, J; Gong, L; Deng, SZ; Chen, J; She, JC; Xu, NS; Yang, R; Wang, ZL. Growth
and field-emission property of tungsten oxide nanotip arrays, Appl. Phys. Lett., 2005,
87, 223108-1 – 223108-3.
[3.18] Liao, CC; Chen, FR; Kai, JJ. WO3-x nanowires based electrochromic devices, Sol.
Energy Mat. Sol. Cells, 2006, 90, 1147-1155.
[3.19] Feng, M; Pan, AL; Zhang, HR; Li, ZA; Liu, F; Liu, HW; Shi, DX; Zou, BS; Gao, HJ.
Strong photoluminescence of nanostructured crystalline tungetn oxide thin films, Appl.
Phys. Lett., 2005, 86, 141901-1 – 141901-3.
[3.20] Ponzoni, A; Comini, E; Sberveglieri, G; Zhou, J; Deng, SZ; Xu, NS; Ding, Y; Wang,
ZL. Ultrasensitive and highly selective gas sensors using three-dimensional tungsten
oxide nanowire networks, Appl. Phys. Lett., 2006, 88, 203101-1 – 203101-3.
[3.21] Rothschild, A; Sloan, J; Tenne, R. Growth of WS2 Nanotubes Phases, J. Am. Chem.
Soc., 2000, 122, 5169-5179.
[3.22] He, Y; Wang, S; Huang, B; Liu, CT; Liaw, PK. Novel tungsten oxide microneedles
with nanosized tips, Appl. Phys. Lett., 2006, 88, 223107-1 – 223107-2.
[3.23] Zhao, YM; Li, YH; Ahmad, I; McCartney, DG; Zhu, YQ; Hu, WB. Two-dimensional
tungsten oxide nanowire networks, Appl. Phys. Lett., 2006, 89, 133116-1 – 133116-3
[3.24] Gu, G; Zheng, B; Han, WQ; Roth, S; Liu, J. Tungsten oxide nanowires on tungsten
substrates, Nano Lett., 2002, 2, 849-851.
[3.25] Wang, SJ; Chen, CH; Ko, RM; Kuo, YC; Wong, CH; Wu, CH; Uang, KM; Chen, TM;
Liou, BW. Preparation of tungsten oxide nanowires from sputter-deposited WCx films
using an annealing/oxidation process, Appl. Phys. Lett., 2005, 86, 263103-1 -263103-3.
[3.26] Meda, L; Breitkopf, RC; Haas, TE; Kirss, RU. Investigation of electrochromic
properties of nanocrystalline tungsten oxide thin film, Thin Solid Films, 2002, 402,
126-130.
[3.27] Deshpande, R; Lee, SH; Parilla, PA; Jones, KM; To, B; Mahan, AH; Perkins, JD;
Mitra, S; Dillon, AC. “Optimization of crystalline tungsten-oxide nanoparticles for
improved electrochromic applications, Solid State Ionics, 2007, 178, 895.
[3.28] Yoo, SJ; Lim, JW; Sung, YE; Jung, YH; Choi, HG; Kim, DK. Fast switchable
electrochromic properties of tungsten oxide nanowire bundles, Appl. Phys. Lett., 2007,
90, 173126-1-173126-3.
[3.29] Seung-Wan Song and In-Su Kang, A facile fabrication route to tungsten oxide films on
ceramic substrates, Sensors and Actuators B: Chemical, 2008, Vol 129, 971-976.
[3.30] Baek Yunho; Kijung Yong, Controlled growth and characterization of tungsten oxide
nanowires using thermal evaporation of WO3 powder, Journal of physical chemistry.
C, 2007, vol. 111, pp. 1213-1218.
[3.31] Gang Gu,† Bo Zheng,† WQ; Han,‡ Siegmar Roth,§ and Jie Liu*†, Tungsten Oxide
Nanowires on Tungsten Substrates, Nano Letters, 2002, 2 (8), pp 849–851
[3.32] Papaefthimiou1, S; Leftheriotis1 G; Yianoulis, P. Electrochromic phenomena in
transition metal oxide thin films prepared by thermal evaporation, Ionics, 1998, Vol 4,
pp 321-329.
[3.33] Hong, K; Xie, M; Wu, H. Tungsten oxide nanowires synthesized by a catalyst-free
method at low temperature, Nanotechnology, 2006, 17, 4830-4833.
[3.34] Chen, CH; Wang, SJ; Ko, RM; Kuo, YC; Uang, KM; Chen, TM; Liou, BW; Tsai, HY.
The influence of oxygen content in the sputtering gas on the self-synthesis of tungsten
oxide nanowires on sputter-deposited tungsten films, Nanotechnology, 2006, 17,
217-223.
[3.35] Lou, XW; Zeng, HC. An Inorganic Route for Controlled Synthesis of W18O49
Nanorods and Nanofibers in Solution, Inorg. Chem., 2003, 42, 6169-6171.
[3.36] Wagner, RS; Ellis, WC. Vapor-Liquid-Solid Mechanism of Single Crystal Growth,
Appl. Phys. Lett., 1964, 4, 89-90.
[3.37] Klinke, C; Hannon, JB; Gignae, L; Reuter, K; Avouris, P. Tunsgten Oxide Nanowire
Growth by Chemically Induced Strain, J. Phys. Chem. B, 2005, 109, 17787-17790.
[3.38] Lin, TT; Lichtman, D. The effects of electron bombardment on the surface composition
of WO3, oxidized W and Ta foils, J. Appl. Phys., 1979, 50, 1298-1303.
[3.39] Zayim, EO; Baydogan, ND. Irradiation effect in WO3 thin films, Sol. Energy Mater.
Sol. Cells, 2006, 90, 402-413.
[3.40] Zhao, Y; Feng, ZC; Liang, Y; Sheng, HW. Laser-induced coloration of WO3, Appl.
Phys. Lett., 1997, 71, 2227-2229.
[3.41] Qiu, H; Lu, YF ; Mai, ZH. Electrochromic writing and erasing on tungsten oxide films
in air by scanning tunneling microscopy, J. Appl. Phys., 2002, 91, 440-443.
[3.42] Aoki, T; Matsushita, T; Suzuki, A; Tanabe, K; Okuda, M. Write-once optical recording
using WO2 films prepared by pulsed laser deposition, Thin Solid Films, 2006, 509,
107-112.
[3.43] Karuppasamy, A; Subrahmanyam, A. Electron beam induced coloration and
luminescence in layered structure of WO3 thin films grown by pulsed dc magnetron
sputtering, J. Appl. Phys., 2007, 101, 113522-113526.
[3.44] Morita, H; Miura, T; Washida, H. Coloration of Tungsten Oxide Films Induced by Ion
or Electron Irradiation, Jpn. J. Appl. Phys., 1981, 20, L323-L325.
[3.45] Niklasson, GA; Berggren, L; Larsson, AL. Electrochromic tungsten oxide: the role of
defects, Sol. Energy Mater. Sol. Cells, 2004, 84, 315-328.
[3.46] Burstein, E. Anomalous Optical Absorption Limit in InSb, Phys. Rev., 1954, 93,
632-633. Moss, TS. Proc. Phys. Soc. London, 1954, Sect. B 67, 775.
[3.47] Deepa, M; Singh, P; Sharma, SN; Agnihotry, SA. Effect of humidity on structure and
electrochromic properties of sol-gel-derived tungsten oxide films,Sol. Energy Mater.
Sol. Cells, 2006, 90, 2665-2682.
[3.48] Scarminio, J; Moraes, MAB; Dias, RCE; Rouxinol, FP; Durrant, SF. Tungsten Oxide
Films of High Electrochromic Efficiencies Obtained by Deposition, Electro-chem.
Solid-State Lett., 2003, 6, H9-H12.
[3.49] Santucci, S; Lozzi, L; Passacantando, M; Nardo, SD; Phani, AR; Cantalini, C; Pelino,
M. Study of the surface morphology and gas sensing properties of WO3 thin films
deposited by vacuum thermal evaporation, J. Vac. Sci. Technol. A, 1999, 17, 644-649.
[3. 50]Feng, M; Pan, AL; Zhang, HR; Li, ZA; Liu, F; Liu, HW; Shi, DX; Zou, BS; Gao, HJ.
Strong photoluminescence of nanostructured crystalline tungetn oxide thin films, Appl.
Phys. Lett., 2005, 86, 141901-1 – 141901-3.
[4.1] First world Conference on TCO, Greece, September 2006.
[4.2] Second International Symposium on Transparent Oxide Thin Films for Electronics and
Optics (TOEO-2), Japan, November 2001, Thin Solid Films, (2002) Vol. 411.
[4.3] Bel Hadj Tahar, R; Ban, T; Ohya, Y; Takahashi, Y. Tin doped indium oxide thin films:
Electrical properties, J.Appl.Phys., 1998, 83, 2631.
[4.4] Look, DC. Progress in ZnO materials and devices, Journal of electronic materials,
2005, Vol. 35, pp 1299-1305.
[4.5] Banerjee, AN; Chattopadhyay, K.K. Progress in Crystal Growth and Characterization
of Materials, 2005, Vol.50.
[4.6] Transparent Conducting Oxides, M R S Bull. 2000, Vol. 25, No 8.
[4.7] Barik, UK; Srinivasan, S; Nagendra, CL; Subrahmanyam, A. Electrical and optical
properties of reactive DC magnetron sputtered silver oxide thin films: role of oxygen
Thin Solid Films, 2003, Vol. 429, pp 129-134.
[4.8] Subrahmanyam, A; Baraik, UK. Synthesis of P-type transparent conducting silver:
indium oxide (AIO) thin films by reactive electron beam evaporation technique,
J.Phys. Chem. Solids, 2005, Vol. 66 pp 817-822.
[4.9] Studies on electrical and optical properties of Silver : Indium Oxide (AIO) thin films
by reactive electron beam evaporation and reactive DC Magnetron sputtering
technique: a Ph D Thesis by Ullash Kumar Barik, submitted to Indian Institute of
Technology Madras, Chennai, India, 2005.
[4.10] A study of Transparent Zinc Oxide Thin Films, a PhD Thesis by Ou Yi submitted to
the Colorado School of Mines, 1993.
[4.11] Kohan, AF; Ceder, G; Morgan, D; Van de Walle, CG. First-principles study of native
point defects in ZnO, Phys. Rev. B, 2000, 61, 15019.
[4.12] Meyer, JR; Hoffman, CA; Antoszewski, J; Faraone, L. Quantitative mobility spectrum
analysis of multicarrier conduction in semiconductors Journal of Applied Physics,
1997, Vol. 812, pp709-713.
[4.13] Subrahmanyam, A; Baraik, UK. Electrical and optical properties of reactive dc
magnetron sputtered silver-doped indium oxide thin films: role of oxygen, Appl. Phys.
A, 2006, Vol.84.
[4.14] Özgür, Ü; Alivov, YAI; Liu, C; Teke, A; Reshchikov, MA; Doan, S; Avrutin, V; Cho,
SJ; Morkoç, H. A comprehensive review of ZnO materials and devices, J .App.
Physics, 2005, Vol. 98, 041301.
[4.15] Soo Kim, H. Sang-Hun Jeong, Sang Sub Kim and Byung-Teak Lee, Magnetron
sputtering growth and characterization of high quality single crystal ZnO thin films on
sapphire substrates, Semicond. Sci. Technol., 2004, Vol.19 pp L29–L31.
[4.16] Ataev, BM; Mamedov, VV; Omaev, AK; Magomedov, BA. Epitaxial ZnO films grown
by RF-assisted low-temperature CVD method, Materials Science in Semiconductor
Processing, 2003,Vol. 6, pp 535-537.
[4.17] Look, DC; Reynolds, DC; Litton, CW; Jones, RL; Eason, DB; Cantwell, G.
Characterization of homo epitaxial p-type ZnO thin films by molecular beam epitaxy,
Appl. Phys. Lett., 2002, Vol. 81(10), pp 1830-1832.
[4.18] Li, X; Yan, Y; Gessert, TA; Perkins, CL; Young, D; DeHart, C; Young, M; Coutts, TJ.
Dependence of Zinc Oxide Thin Films Properties on Filtered Vacuum Arc Deposition
Parameters, J. Vac. Sci. Technol. A, 2003, Vol. 21(3), pp 1342-1346.
[4.19] Alam, MJ; Cameron, DC. Preparation and properties of transparent conductive
aluminium-doped zinc oxide thin films by sol–gel process, J. Vac. Sci. Technol. A
2001, Vol.19 (4), pp 1642-1646.
[4.20] Li, X, Yan, Y, Gessert, TA; Perkins, CL; Young, D; DeHart, C; Young, M; Coutts, TJ.
p-Type ZnO Thin Films Formed by CVD Reaction of Diethyl zinc and NO Gas , J.
Vac. Sci. Technol. A, 2003, Vol. 21(4), pp 1342-1346.
[4.21]Joseph, M; Tabata, H; Saeki, H; Ueda, K; Kawai, T. Control of the electric and
magnetic properties of ZnO films Physica B, 2001,Vol. 302-303, pp 140-148 .
[4.22] Yamamoto, T. "Codoping method for solutions of doping problems in wide-band-gap
semiconductors," Phys. Stat. Sol., (A) ,Vol. 193, pp.423–433.
[4.23] Hatanaka, Y; Niraula, M; Nakamura, A; Aoki, T. Excimer laser doping techniques for
II–VI semiconductors, Appl. Surf. Sci., 2001, Vol.175-176, pp 462-467.
[4.24] Ohara, K; Seino, T; Nakamura, A; Aoki, T; Kominami, H; Nakanishi, Y; Hatanaka, Y.
Sb excimer-laser doping in ZnO films prepared by oxidation of sulfide on Si, Applied
Surface Science, 2005, Vol.244, pp 369-372.
[4.25] Kobayashi, A; Sankey, OF; Dow, JD. Deep energy levels of defects in the wurtzite
semiconductors AIN, CdS, CdSe, ZnS, and ZnO, Phys Rev B, 1983, Vol. 28, pp
946 – 950.
[4.26]Yamamoto, T and H.K-Yoshida, H.K, Solution Using a Codoping Method to
Unipolarity for the Fabrication of p-Type ZnO, Jap. J. Appl.Phys. 1999, Vol. 38, L
166 -169.
[4.27] Joseph, M; Tabata, H; Kawai, T. p-Type electrical conduction in ZnO thin films by Ga
and N codoping, Jap. J. Appl.Phys., 1999, Vol. 38, pp L 1205 –L1207.
[4.28] Minegishi, K; Koiwai, Y; Kikuchi, Y; Yano, K. Growth of p-type Zinc Oxide Films
by Chemical Vapour Deposition, Jpn.J.Appl.Phys, 1997,Vol. 36, L 1453 -1455.
[4.29] Ryu, YR; Zhu, S; Look, DC; Wrobel, JM; Jeong, HM; White, HW. Synthesis of p-
type ZnO films, J.Cryst Growth, 2000, Vol.216, pp 330-334.
[4.30] Kaminska, E; Piotrowska, A; Kossut, J; Butke, R; Daniel, W; Golasewska, K; Barcz,
A; Jakiela, R; Dynowska, E; Przezdziecka, E.. P-type conducting ZnO: Fabrication and
characterization, E-MRS Fall meeting, 2004.
[4.31] Zeng, YJ; Ye, ZZ; Xu, WZ; Lu, JG; He, HP; Zhu, LP; Zhao, BH; Che, Y; Zhang,
SB. p-type behavior in nominally undoped ZnO thin films by oxygen plasma growth ,
Appl. Phys. Lett., 2006, Vol. 88, 262103.
[4.32] Xu, WZ; Ye, ZZ; Zeng, YJ; Zhu, LP; Zhao, BH; Jiang, L; Lu, JG; He, HP; Zhang,
SB. ZnO light-emitting diode grown by plasma-assisted metal organic chemical vapor
deposition, Appl. Phys. Lett., 2006, Vol. 88,173506.
[4.33] Jiao, SJ; Zhang, ZZ; Lu, YM; Shen, DZ; Yao, B; Zhang, JY; Li, BH; Zhao, DX; Fan,
XW; Tang, ZK. ZnO p-n junction light-emitting diodes fabricated on sapphire
substratesAppl. Phys. Lett., 2006, Vol. 88, 03191.
[4.34] Cao, Y; Miao, L; Tanemura, S; Tanemura, M; Kuno, Y; Hayashi, Y. Low resistivity
p-ZnO films fabricated by sol-gel spin coating, Appl. Phys. Lett., 2006, Vol. 88,
251116.
[4.35] Putley, EH. Hall effect and related Phenomenon, Butterworths, 1960.
[4.36] http://www.newscientist.com/article.ns?id=dn4602.
Chapter 2
SPECTROSCOPIC ANALYSIS OF CHEMICAL SPECIES

IN CARBON PLASMAS INDUCED BY HIGH-POWER IR
CO2 LASER
J. J. Camacho1*, J. M. L. Poyato1, L. Díaz2 and M. Santos2

1
Departamento de Química-Física Aplicada. Facultad de Ciencias. Universidad
Autónoma de Madrid. Cantoblanco. 28049-Madrid. Spain.
2
Instituto de Estructura de la Materia, CFMAC, CSIC, Serrano 121.
28006-Madrid, Spain
ABSTRACT
This chapter describes some fundamentals of laser-induced breakdown spectroscopy
(LIBS) and experimental results obtained from ultraviolet-visible-near infrared (UV-Vis-
NIR) spectra induced by laser ablation of a graphite target, developed in our laboratory.
Ablation was produced by a high-power IR CO2 pulsed laser using several wavelengths
(λ=9.621 and 10.591 µm), power density ranging from 0.22 to 6.31 GW cm-2 and
medium-vacuum conditions (typically at 4 Pa). Spatially and time resolved analysis were
carried out for the plasma plume. Wavelength-dispersed spectra of the plume reveal the
emission of C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and molecular features of C2, CN, OH,
CH, N2, N2+ and NH. For the assignment of molecular bands a comparison with
conventional emission sources was made. The characteristics of the spectral emission
intensities from the different species have been investigated as functions of the ambient
pressure, laser irradiance, delay time, and distance from the target. Excitation, vibrational
and rotational temperatures, ionization degree and electron number density for some
species were estimated. Time-gated spectroscopic studies have allowed estimation of
time-of-flight (TOF) and propagation velocities for various emission species.
*
Corresponding author: Email: j.j.camacho@uam.es
64 J. J. Camacho, J.M.L. Poyato, L. Díaz et al.
1. INTRODUCTION
The interaction of a high-energy infrared (IR) pulsed laser beam with solid materials has
been investigated extensively over the past several years, due to its significance in
technologies such as laser sampling for chemical analysis and pulsed laser deposition of thin
film [1-5]. Laser–material interactions involve non-linear complex and collective processes
that are not fully understood. Among the many factors the laser irradiance is one of the most
important in controlling the mechanism of the laser–material interaction. For example, at high
irradiance, a laser induced plasma can be formed above the sample surface which may absorb
the incident laser energy, thereby shielding the sample and decreasing the efficiency of laser
energy available for mass ablation. The properties and composition of the resulting ablation
plume may evolve, both as a result of collisions between particles in the plume and through
plume-laser radiation interactions. The laser-target interactions will be sensitively dependent
both on the nature and condition of the target material, and on the laser pulse parameters.
Subsequent laser-plume interactions will also be dependent on the properties of the laser
radiation, while the evolution and propagation of the plume will also be sensitive to collisions
and thus to the quality of the vacuum under which the ablation is conducted and/or the
presence of any background gas.
Optical emission spectroscopy (OES) is a powerful tool to get information on the laser-
ablated species. For laser ablation of carbon, OES studies in different atmospheres are
reported and these studies have yielded many interesting results [6-17]. The major parts of
this work are already published by us in different journals [18-20].
The objectives of this work are: (1) to show some fundamentals of laser-induced
breakdown spectroscopy (LIBS) and, (2) to review of our recent results on LIBS analysis of
chemical species in carbon plasmas induced by high-power IR CO2 laser, adding some new
results.
OES has been used to investigate thermal and dynamical properties of a plume produced
by laser ablation of a graphite target at air pressures around 4 Pa. Ablation is performed using
a high-power IR CO2 pulsed laser. The emission generated by the plasma in the spectral
region 200-1100 nm is due to electronic relaxation of excited C, N, H, O, ionic fragments C+,
C2+, C3+, C4+, and molecular features of C2, CN, OH, NH, CH, N2+ and N2. As far as we
know, a spectrum so rich in atomic lines belonging to ionized species and molecular features
has not been observed previously in similar experiments. Also we analyzed these spectra at
different distances from the target along the plasma expansion direction. Finally we present
some new results obtained from the time resolved spectroscopic analysis of the laser ablation
of a graphite target. From these results temperature, electron densities and ionization degree
are obtained. Also we have studied here the spectral emission intensities from different
species as functions of the ambient pressure and laser irradiance.
Although OES gives only partial information about the plasma particles, this diagnostic
technique helped us to draw a picture of the plasma in terms of the emitting chemical species,
to evaluate their possible mechanisms of excitation and formation and to study the role of
gas-phase reactions in the plasma expansion process.
Spectroscopic Analysis of Chemical Species in Carbon Plasmas Induced... 65
2. FUNDAMENTALS OF LASER INDUCED BREAKDOWN SPECTROSCOPY

(LIBS)
Excellent textbooks and reviews about the fundamentals of laser-induced breakdown
spectroscopy (LIBS) and examples of various processes are readily available today [21-24].
LIBS, also sometimes called laser-induced plasma spectroscopy (LIPS), is a technique of
atomic-molecular emission spectroscopy which utilizes a highly-power laser pulse as the
excitation source. LIBS can analyze any matter regardless of its physical state, being it solid,
liquid or gas. Because all elements emit light when excited to sufficiently high energy, LIBS
can detect different species (atomic, ionic and molecular) and limited only by the power of
the laser as well as the sensitivity and wavelength range of the spectrograph/detector.
Basically LIBS makes use of OES and is to this extent very similar to arc/spark emission
spectroscopy. LIBS operates by focusing the laser onto a small area at the surface of the
sample; when the laser is triggered it ablates a very small amount of material which
instantaneously generates a plasma plume with temperatures of about 10000–30000 K. At
these temperatures, the ablated material dissociates (breakdown) into excited ionic and atomic
species. At the early time, the plasma emits a continuum of radiation which does not contain
any known information about the species present in the plume and within a very small
timeframe the plasma expands at supersonic velocities and cools. At this point the
characteristic atomic/ionic and molecular emission lines of the species can be observed. The
delay between the emission of the continuum and characteristic radiation is of the order of 1
µs, this is one of the reasons for temporally gating the detector. LIBS is technically similar
and complementary to a number of other laser-based techniques (Raman spectroscopy, laser-
induced fluorescence etc). In fact devices are now being manufactured which combine these
techniques in a single instrument, allowing the atomic, molecular and structural
characterization of a sample as well as giving a deeper insight into physical properties. A
typical LIBS system consists of a pulsed laser and a spectrometer with a wide spectral range
and a high sensitivity, fast response rate and time gated detector. The principal advantages of
LIBS over the conventional analytical spectroscopic techniques are its simplicity and the
sampling speed.
2.1. Nature of the Plasmas
Plasma is an ionized gas, a distinct fourth state of matter. The free electric charges
(electrons and ions) make plasma electrically conductive (sometimes more than gold and
copper), internally interactive, and strongly responsive to electromagnetic fields. Ionized gas
is usually called plasma when it is electrically neutral (i.e., electron density is balanced by
that of positive ions) and contains a significant number of the electrically charged particles,
sufficient to affect its electrical properties and behaviour. Plasmas occur naturally but also can
be effectively man-made in laboratory and in industry, which provides opportunities for
numerous applications, including thermonuclear synthesis, electronics, lasers, fluorescent
lamps, and many others.
Plasma offers three major features that are attractive for applications in chemical-physics:
(1) temperatures and energy density of at least some plasma components can significantly
exceed those in conventional technologies, (2) plasmas are able to produce very high
concentrations of energetic and chemically active species (e.g., electrons, ions, atoms,
molecules and radicals, excited states, and different wavelength photons), and (3) plasma
systems can essentially be far from thermodynamic equilibrium, providing extremely high
concentrations of chemically active species and keeping bulk temperature as low as room
temperature. These plasma features permit significant intensification of traditional chemical
processes, essential increase of their efficiency, and often successful stimulation of chemical
reactions impossible in conventional chemistry. Plasmas are found in nature in various forms
and are characterized normally by their electron density ne and electron temperature Te. On
earth they exist in the ionosphere at height of 70-500 km (density ne = 106 cm-3, Te = 2300 K).
Solar wind is another natural plasma originating from the sun with ne = 10 cm-3 and Te = 105
K. The corona which extends around the sun has an electron density ne = 108 cm-3 and its
electron temperature is Te = 106 K. Finally, white dwarfs, the final state of stellar evolution,
have a ne of 1030 cm-3. In plasma formation, as the temperature of material is raised, its state
changes from solid to liquid and then to gas. If the temperature is elevated further, an
appreciable number of gas atoms are ionized and go into a high temperature gaseous state in
which the charge numbers of ions and electrons are almost the same and charge neutrality is
satisfied at a macroscopic scale. When the ions and electrons move collectively, these
charged particles interact via Coulomb forces which are long-range forces and decay with the
inverse square of the distance between charged particles. Therefore, many charged particles
interact with each other by long range forces rather then through short range collision process
like in a common gas. This results in different kinds of collective phenomena such as plasma
instabilities and wave phenomena [25].
2.2. Local Thermodynamic Equilibrium (LTE). Model for the Plasma
Plasma description starts by trying to characterize properties of the assembly of atoms,

molecules, ions and electrons rather than individual species. If thermodynamic equilibrium
exits, then plasma properties can be described through the concept of temperature.
Thermodynamic equilibrium is rarely complete, so physicists have settled for a useful
approximation, local thermodynamic equilibrium (LTE). In LTE model it is assumed that the
distribution of population densities of the electrons is determined exclusively through
collisional processes and that they have sufficient rate constants so that the distribution
responds instantaneously to any change in the plasma conditions. In such circumstances each
process is accompanied by its inverse and these pairs of processes occur at equal rates by the
principle of detailed balance. Thus, the distribution of population densities of the electrons
energy levels is the same as it would be in a system in complete thermodynamic equilibrium.
The population distribution is determined by the statistical mechanical law of equipartition
among energy levels and does not require knowledge of atomic cross sections for its
calculation. Thus, although the plasma density and temperature may vary in space and time,
the distribution of population densities at any instant and point in space depends entirely on
local values of density, temperature, and chemical composition of plasma. If the free electrons
are distributed among the energy levels available for them, their velocities have a Maxwellian
distribution
3/ 2
⎛ m ⎞ ⎛ m v2 ⎞ 2
dn v = ne 4π ⎜⎜ ⎟⎟ exp⎜ − ⎟ v dv , (2.1)
⎝ 2πk B T e ⎠ ⎜ 2k B T e ⎟
⎝ ⎠
where m is the electron mass, v is the electron velocity and kB is the Boltzmann constant. For
the bound levels the distributions of population densities of neutrals and ions are given by the
Boltzmann (2.2) and Saha (2.3) equations
Nj gj ⎛ ( E − Ei ) ⎞
= exp⎜⎜ − j ⎟⎟ , (2.2)
Ni gi ⎝ k T
B e ⎠
3/ 2
N z +1, k ne g z +1, k ⎛ 2π m k B Te ⎞ ⎛ Ip ⎞
= 2⎜⎜ 2
⎟
⎟ exp⎜⎜ − z, k ⎟⎟ , (2.3)
N z,k g z,k ⎝ h ⎠ ⎝ k BT e ⎠
where Ni, Nj, Nz+1,k and Nz,k are the population densities of various levels designated by their
quantum numbers j (upper), i (lower) and k (the last for the ground level) and ionic charge z
and z+1. The term gz,i is the statistical weight of the designated level, Ej and Ei are the energy
of the levels j and i and Ipz,k is the ionization potential of the ion of charge z in its ground
level k. Equations (2.1)-(2.3) describe the state of the electrons in an LTE plasma. For
complete LTE of the populations of all levels, including the ground state, a necessary
condition is that electron collisional rates for a given transition exceed the corresponding
radiative rates by about an order of magnitude [26]. This condition gives a criterion [27] for
the critical electron density of the level with energy ΔE
3 3
5 ⎛α ⎞ 7 ⎛ ΔE ⎞ ⎛ k BT e ⎞
ne ,crit ≥ ⎜ ⎟⎟ z ⎜⎜ 2 ⎟⎟ ⎜⎜ 2 ⎟⎟ ≅ 1.6 × 1012 Te1 / 2 (ΔE ) 3 , (2.4)
8 π ⎜⎝ a0 ⎠ ⎝ z EH ⎠ ⎝ z EH ⎠
where α is fine-structure parameter, a0 is Bohr radius, and EH is the hydrogen ionization

potential. In the final form of Eq. (2.4), ne,crit is given in cm-3, Te in K and ΔE in eV. Many
plasmas of particular interest do not come close to complete LTE, but can be considered to be
only in partial thermodynamic equilibrium in the sense that the population of sufficiently
highly excited levels are related to the next ion’s ground state population by Saha-Boltzmann
relations, respective to the total population in all fine-structure levels of the ground state
configuration [26]. For any atom or ion with simple Rydberg level structure, various criteria
were advanced for the minimum principal quantum number ncrit for the lowest level, often
called thermal or collision limit, for which partial thermodynamic equilibrium remains valid
to within 10%. One criterion with quite general validity is given by Griem [27]:
2 / 17
⎡ 10 z 7 ⎛α ⎞
3
⎤ ⎛ k BT e ⎞
1 / 17
ncrit ≈⎢ ⎜⎜ ⎟⎟ ⎥ ⎜⎜ 2 ⎟⎟ . (2.5)
⎢⎣ 2 π ne ⎝ a0 ⎠ ⎥⎦ ⎝ z EH ⎠
2.3. LIB Plasma
As we mentioned previously, plasma is a local assembly of atoms, molecules, ions and

free electrons, overall electrically neutral, in which the charged species often act collectively.
The LIB plasma is initiated by a single laser pulse. If we consider the temporal evolution of
LIB plasma, at early times the ionization grade is high. As electron-ion recombination
proceeds, neutral atoms and molecules form. A recombination occurs when a free electron is
captured into an ionic or atomic energy level and gives up its excess kinetic energy in the
form of a photon.
LIB plasmas are characterized by a variety of parameters, the most basic being the degree
of ionization. A weakly ionized plasma is one in which the ratio of electrons to other species
is less than 10%. At the other extreme, high ionized plasmas may have atoms stripped of
many of their electrons, resulting in very high electron to atom/ion ratios. LIB plasmas
typically, for low power laser intensities, fall in the category of weak ionized plasmas. At
high laser power densities, LIB plasmas correspond to strong ionized plasmas.
2.3.1. Initiation mechanism: Multiphoton ionization (MPI) and electron impact

ionization (EII)
Plasma is initiated by electron generation and electron density growth. The conventional
LIB plasma can be initiated in two ways: multiphoton ionization (MPI) and electron impact
ionization (EII) both followed by electron cascade. MPI involves the simultaneous absorption
of a number of photons n, required to equal the ionization potential IP(A) of an atom or
molecule A
nhν + A → A+ + e + IP(A); nhν ≥ IP(A), (2.6)
where n is the number of photons needed to strip off an electron, which corresponds to the
integer part of the quantity:
I P + ε osc
n= + 1. (2.7)
hν
Here εosc is the oscillation energy of a free electron in the alternating electric field. Within
the classical microwave breakdown theory [28], a free electron oscillates in the alternating
electric field E of the laser electromagnetic wave (with frequency ω and wavelength λ (µm)),
and its oscillation energy,
e2 E 2 e2
ε osc [eV ] = = I λ2 = 4.67 × 10 −14 I W λ2 ,
3 W
(2.8)
4mω 2
4mπc
remains constant. In Eq. (2.8) e is the electron charge and Iw is the laser intensity (irradiance,
power density or flux density in W cm-2). The probability of MPI WMPI, by absorbing
simultaneously n laser photons to strip off an electron, is expressed by the classical formula
[29]
n n
⎛ ε ⎞ ⎧ 6.35 × 10 −14 I W ⎫
WMPI s [ ]≅ ω n
−1 3/ 2
⎜⎜1.36 osc
IP
⎟⎟ = 1.88 × 1015 λ2 n −1 n 3 / 2 ⎨
IP
⎬ , (2.9)
⎝ ⎠ ⎩ ⎭
where IP is in eV. Besides, the probability of simultaneous absorption of photons decreases

with the number of photons n necessary to cause ionization.
EII process consists on the absorption of light photon by free or quasifree electrons,
producing electrons with enough kinetic energy e* to ionize one atom or molecule
e + nhν + A → e* + A → 2e + A+. (2.10)
Two conditions must co-exist for EII to initiate: (i) an initial electron must reside in the
focal volume; and (ii) the initial electron must acquire energy which exceeds the ionization
energy of the material in the focus. These free or quasifree electrons can be produced by the
effect of cosmic ray ionization (natural ionization), by means of MPI, or by a breakdown
induced in some impurity. In air at atmospheric pressure, the natural electron density is ~103
cm-3 [30]. These electrons in the focal volume gain sufficient energy, from the laser field
through inverse bremsstrahlung collision with neutrals, to ionize atoms, molecules or ions by
inelastic electron-particle collision resulting in two electrons of lower energy being available
to start the process again
e*[ε ≥IP(A)] + A → A+ + 2e; e*[ε ≥IP(A+)] + A+ → A2+ + 2e. (2.11)
The MPI mechanism dominates electron generation only for low exciting wavelengths.
Therefore initial EII becomes a problem at a higher wavelength because neither cascade nor
MPI can furnish sufficient number of electrons. At higher laser intensities, electric field of the
laser is able to pull an outer shell electron out of its orbit. After the initial electron ejection the
LIB plasma is commonly maintained by the absorption of optical energy and the EII.
Electrons in the laser field will gain energy through electron-neutral inverse bremsstrahlung
collisions and will lose energy by elastic and inelastic collisions with the neutral species
through excitation of rotational and vibrational degree of freedom of molecules and excitation
of electronic states. While some electrons will be lost by attachment, new electrons will be
produced by ionizing collisions. At high laser intensity, few electrons will be generated with
energy larger than the ionization energy. The wavelength-resolved emission spectra from the
laser plasma are not expected to vary due to the plasma origin. However plasma origin may
be relevant, if the enhancement is observed between UV, visible and IR excitation
wavelengths. Once that LIB plasma is formed, its growth is governed by the continuity rate
equation for the electron density [31]
dne
= ν i ne + Wn IWn N − ν a ne − ν R ne + De∇ 2 ne ,
2
(2.12)
dt
where νi is the impact ionization rate, Wn is the multiphoton ionization rate coefficient, Iw is
the intensity of the laser beam, n is the number of photons required for MPI, N is the number
of atoms/molecules per unit volume, νa is the attachment rate, νR is the recombination rate
and De is the electron diffusion coefficient. The term dne/dt is the net rate of change in
electron concentration at a point in the focal volume at a time t after the release of initiatory
electrons. On the right side of the equation (2.12), the first term is the electron generation due
to impact ionization. The second term on the right is MPI rate. The third, fourth and fifth
terms are sink terms which represent electron attachment, recombination and diffusion,
respectively. Impact ionization is defined by multiplying the number of electrons per unit
volume to the impact ionization rate νi. The impact ionization rate refers to the rate at which
electrons are generated as a result of ionizing collisions. At high laser intensity, a few new
electrons can be generated and gain energy larger than their ionization energy which leads to
the generation of new electrons by impact ionization, thereby leading to the cascade growth.
2.3.2. Electron attachment, recombination and diffusion
Electron attachment is the rate of electron attachment νa multiplied by the number of

electrons per unit volume. The LIB plasma typically loose electrons to the neutral species via
the attachment mechanism in the form of three-body attachment or two-body dissociative
attachment. Three-body attachment is: e + AB + X → AB- + X, where X appears to be a
facilitator that allows the electrons to be gained by AB even through X remains unchanged
throughout the process. Two-body dissociative attachment is: e + AB → A- + B. In this
mechanism the electrons must exhibit a threshold electron energy that is equal to the
difference between the dissociative energy of AB and the attachment energy of A, which
results in the separation of A and B.
Electron recombination is the rate of electron recombination νR multiplied to the number
of electrons per unit volume. When the electron density is high, such as during the last stage
of cascade breakdown, the LIB plasma can lose electrons to ions through electron-ion
recombination. Similar to the electron attachment, three-body recombination and two-body
recombination occurs as: e + AB+ + X → AB + X, e + AB+ → A + B. The electron-ion
recombination rate has been studied theoretically for a three-body recombination by Gurevich
and Pitaevskii [32]
ne2
ν R = 8.8 × 10 − 27
3.5
[ s −1 ] , (2.13)
Te
where ne is the electron density in cm-3 and Te is the electron temperature in eV. The electron
diffusion term is expressed as D e ∇ 2 n e (Eq. (2.12)). This loss mechanism, more important
for a small diameter laser beam, is the diffusion of electrons out of the focal volume. Morgan
[33] referred to the combined effect of diffusion and cascade ionization as the responsible for
top-hat intensity profile. By imposing an electron skin at the edge of the intensity profile, they
found that the electron density grows exponentially as
2.408 De
ve = , (2.14)
a2
where v e is the average electron velocity, De is electron diffusion coefficient and a is the
radius of the beam. The equation (2.14) is intended to be an upper boundary for diffusion
losses only because laser beams typically have a radial distribution closer to the gaussian
rather than top-hat distribution.
In summary, two mechanisms MPI and EII can initiate a conventional LIB plasma
formation. After the LIB plasma formation the temporal growth is governed by the equation
(2.12). The recombination of these two source terms (MPI and EII) and three sink terms
(electron attachment, electron recombination and electron diffusion) controls the development
of the conventional LIB plasma. These mechanisms that directly affect the temporal
development of the LIB plasma, determine the necessary spectroscopic techniques required to
spectrally resolve elemental species inside the LIB plasma.
2.4. Elements of LIBS
In contrast to conventional spectroscopy, where one is mainly concerned with the

structure of an isolated atom and molecule, the radiation from the plasma also depends on the
properties of the plasma in the intermediate environment of the atomic or molecular radiator.
This dependence is a consequence of the long-range Coulomb potential effects which
dominate the interactions of ions and electrons with each other and with existing neutral
particles. These interactions are reflected in the characteristic radiations in several ways. They
can control population densities of the discrete atomic states, spectral shift and broadening by
Stark effect, decrease of ionization potentials of the atomic species, cause continuum
radiation emissions and emission of normally forbidden lines. Generally, the radiation emitted
from a self-luminous plasma can be divided into bound-bound, bound-free, and free-free
transitions.
2.4.1. Line radiation
Line radiation from plasma occurs for electron transitions between the discrete or bound
energy levels in atoms, molecules or ions. In an optically thin plasma of length l along the
line of sight [34], the integrated emission intensity Iji of a spectral line arising from a
transition between bound levels j and i is given by
A ji hν ji
4π ∫
I ji = N j ds = hν ji A ji N j l , (2.15)
where Nj is the population density of the upper level j, hνji is the photon energy (energy
difference between levels j and i), and Aji is the spontaneous transition probability or Einstein
A coefficient. The integration is taken over a depth of plasma viewed by the detector, and the
intensity of radiation is measured in units of power per unit area per unit solid angle.
Transition probabilities can be sometimes expressed via the oscillator strength fji. This is
defined as the ratio of the number of classical oscillators to the number of lower state atoms
required to give the same line-integrated absorption [35]. Its relationship to the Einstein
coefficient is
4πε 0 mc 3 g j
f ji = A ji . (2.16)
e 2 8π 2ν ji gi
The usefulness of fji is that it is dimensionless, describing just the relative strength of the
transition. The detailed values of Aji, gi, and gj can be obtained from reference compilations or
from electronic databases, i.e by NIST [36].
2.4.2. Continuum radiation
The origins of continuum radiation are both bound-free and free-free transitions. The
absorption of radiation from a discrete atomic state, such that the photon has enough energy
to extend above the next ionization threshold, results in a release of an electron and gives rise
to the process of photoionization. The reverse process of recombination occurs when an ion
and an electron recombine with emission of a photon to form an ion in the next lowest ionic
state (or in the neutral atomic state). Since the upper state is continuous, the emitted or
absorbed radiation in both processes is also continuous. Transitions between two free energy
levels can occur in plasmas increasing the energy exchanges of charged particles. Classically,
this takes place because a moving charge radiates when it is accelerated or retarded. For most
cases of practical importance, these free-free transitions are classified as bremsstrahlung or
cyclotron spectra. In bremsstrahlung, the acceleration of charged particle takes place via the
Coulomb field of charged particles. In cyclotron radiation, the acceleration is due to the
gyration of charged particles in a magnetic field. The total continuum radiation at any
particular frequency I(ν) is the sum of the contributions from all such processes having
components at the specified frequency. Thus
1 ⎡ ⎤
I (ν ) d ν =
4π ∫n ∑ N
e
i
i ⎢γ (i , Te ,ν ) + ∑ α (i , p , Te ,ν ) ⎥h ν ds d ν ,
⎣ p ⎦
(2.17)
where γ(i, Te, ν) is the atomic probability of a photon of frequency ν being produced in the
field of an atom or ion (specified by i) by an electron of mean kinetic temperature Te making
free-free transition; α(i, p, Te, ν) is the corresponding probability where the electron makes a
free-bond transition into a level p. As before, the integration is taken over the plasma depth s.
2.4.3. Line Broadening; Determination of electron number density from Stark

broadening of spectral lines
The shape of the spectral lines in the LIB has been studied since the first observation of
the laser-induced breakdown in early 1960s. It plays an important role for the spectrochemical
analysis and quantification of the plasma parameters. The observed spectral lines are always
broadened, partly due to the finite resolution of the spectrometers and partly to intrinsic
physical causes. In addition, the center of the spectral lines may be shifted from its nominal
central wavelength. The principal physical causes of spectral line broadening are the Doppler,
resonance pressure, and Stark broadening. There are several reasons for this broadening and
shift. These reasons may be divided into two broad categories: broadening due to local
conditions and broadening due to extended conditions. Broadening due to local conditions is
due to effects which hold in a small region around the emitting element, usually small enough
to assure local thermodynamic equilibrium. Broadening due to extended conditions may
result from changes to the spectral distribution of the radiation as it traverses its path to the
observer. It also may result from the combining of radiation from a number of regions which
are far from each other.
Natural broadening: The uncertainty principle relates the lifetime of an excited state
(due to the spontaneous radiative decay) with the uncertainty of its energy. This broadening
effect results in an unshifted Lorentzian profile. The full width at half maximum (FWHM) of
natural broadening for a transition with a natural lifetime of τji is: ΔλNFWHM=λ2/πcτji. The
natural lifetime τji is dependent on the probability of spontaneous decay: τji=1/Aji. Natural
broadening is usually very small compared with other causes of broadening.
Doppler broadening: The Doppler broadening is due to the thermal motion of the
emitting atoms, molecules or ions. The atoms in a gas which are emitting radiation will have a
distribution of velocities. Each photon emitted will be "red" or "blue" shifted by the Doppler
effect depending on the velocity of the atom relative to the observer. The higher the
temperature of the gas, the wider the distribution of velocities in the gas. Since the spectral
line is a combination of all of the emitted radiation, the higher the temperature of the gas, the
broader will be the spectral line emitted from that gas. This broadening effect is described by
a Gaussian profile and there is no associated shift. For a Maxwellian velocity distribution the
line shape is Gaussian, and the FWHM may be estimated as (in Å):
ΔλDFWHM = 7.16 × 10 −7 ⋅ λ ⋅ T / M , (2.18)
being λ the wavelength in Å, T the temperature of the emitters in K, and M the atomic mass in
amu.
Pressure broadening: The presence of nearby particles will affect the radiation emitted
by an individual particle. There are two limiting cases by which this occurs: (i) Impact
pressure broadening: The collision of other particles with the emitting particle interrupts the
emission process. The duration of the collision is much shorter than the lifetime of the
emission process. This effect depends on both the density and the temperature of the gas. The
broadening effect is described by a Lorentzian profile and there may be an associated shift.
(ii) Quasistatic pressure broadening: The presence of other particles shifts the energy levels in
the emitting particle, thereby altering the frequency of the emitted radiation. The duration of
the influence is much longer than the lifetime of the emission process. This effect depends on
the density of the gas, but is rather insensitive to temperature. The form of the line profile is
determined by the functional form of the perturbing force with respect to distance from the
perturbing particle. There may also be a shift in the line center. Pressure broadening may also
be classified by the nature of the perturbing force as follows: (i) Linear Stark broadening
occurs via the linear Stark effect which results from the interaction of an emitter with an
electric field, which causes a shift in energy which is linear in the field strength (∼E and
∼1/r2); (ii) Resonance broadening occurs when the perturbing particle is of the same type as
the emitting particle, which introduces the possibility of an energy exchange process (∼E and
∼1/r3); (iii) Quadratic Stark broadening occurs via the quadratic Stark effect which results
from the interaction of an emitter with an electric field, which causes a shift in energy which
is quadratic in the field strength (∼E and ∼1/r4); (iv) Van der Waals broadening occurs when
the emitting particle is being perturbed by Van der Waals forces. For the quasistatic case, a
Van der Waals profile is often useful in describing the profile. The energy shift as a function
of distance is given in the wings by e.g. the Lennard-Jones potential (∼E and ∼1/r6).
Stark broadening of spectral lines in the plasma occurs when an emitting species at a
distance r from an ion or electron is perturbed by the electric field. This interaction is
described by the Stark effect. The linear Stark effect exists for hydrogen and for all other
atoms. Stark broadening from collisions of charged species is the primary mechanism
influencing the emission spectra in LIBS. Stark broadening of well-isolated lines in the
plasma can be used to determine the electron number density ne(cm-3). An estimation of the
Stark width (FWHM) and line shift of the Stark broadened lines is given as [26-27,34-35,37-
39]:
1/ 4
Δλ Stark
FWHM
⎛ n ⎞ ⎛ n ⎞
= 2W ⎜ e16 ⎟ + 3.5 A⎜ e16 ⎟ (1 − BN ) W ⎛⎜ 10n
−1 / 3
D
e
16
⎞
⎟, (2.19)
⎝ 10 ⎠ ⎝ 10 ⎠ ⎝ ⎠
1/ 4
⎛ n
Δλ Shift = D ⎜ e16
⎞ ⎛ n ⎞
⎟ ± 2 A⎜ e16 ⎟ (1 − BN ) W ⎛⎜ 10n
−1 / 3
D
e
16
⎞
⎟, (2.20)
⎝ 10 ⎠ ⎝ 10 ⎠ ⎝ ⎠
where W is the electron impact parameter or half-width, A is the ion impact parameter both in
Å, B is a coefficient equal to 1.2 or 0.75 for ionic or neutral lines, respectively, D (in Å) is the
electron shift parameter and ND is the number of particles in the Debye sphere
1/ 2
N D = 1.72 × 109 T 3 / 2 / ne .
The electron and the ion impact parameters are functions of temperature. The first term
on the right side of Eq. (2.19) refers to the broadening due to the electron contribution,
whereas the second one is the ion broadening. Since for LIB conditions Stark broadening is
predominantly by electron impact, the ion correction factor can safely be neglected, and Eq.
(2.19) becomes
⎛ ne ⎞
Δλ Stark
FWHM ≈ 2W ⎜ 16
⎟. (2.21)
⎝ 10 ⎠
The coefficients W are independent of ne and slowly varying functions of electron

temperature. The minus sign in Eq. (2.20) applies to the high-temperature range of those few
lines that have a negative value of D/W at low temperatures. A comprehensive list of width
and shift parameters W, A and D is given by Griem [27, 39].
2.4.4. Determination of excitation, vibrational and rotational temperatures
The excitation temperature Texc can be calculated according to the Boltzmann equation
under the assumption of LTE. The significance of this temperature depends on the degree of
equilibrium within the plasma. For plasma in LTE, any point can be described by its local
values of temperature, density, and chemical composition. By considering two lines λji and
λnm of the same species, characterized by different values of the upper energy level (Ej≠En),
the relative intensity ratio can be used to calculate the plasma excitation temperature
E j − En
Texc = . (2.22)
⎡I ⋅λ ⋅ g ⋅ A ⎤
k B ln ⎢ nm nm j ji ⎥
⎣⎢ I ji ⋅ λ ji ⋅ g n ⋅ Anm ⎦⎥
When selecting a line pair, it is advisable to choose two lines as close as possible in
wavelength and as far apart as possible in excitation energy. This is to limit the effect of
varying the spectral response of the detection system. The use of several lines instead of just
one pair leads to greater precision of the plasma excitation temperature estimation. In fact,
though the precision of the intensity values can be improved by increasing the signal
intensity, the transition probabilities Aji reported in the literature exhibit significance degree
of uncertainty (5-50%). The excitation temperature can be calculated from the relative
intensities of a series of lines from different excitation states of the same atomic or ionic
species from the slope of the Bolztmann plot ln[Iji·λji/gj·Aji] versus Ej/kB
⎡ I ⋅λ ⎤ Ej
ln ⎢ ji ji ⎥ = C − , (2.23)
⎢⎣ g j ⋅ A ji ⎥⎦ k B ⋅ Texc
where Iji is the emissivity (W m-3 sr-1) of the emitted j→i spectral line, λji is the wavelength,
gj=2Jj+1 is the statistical weight, Aji is the Einstein transition probability of spontaneous
emission, Ej/kB is the normalized energy of the upper electronic level and C=ln(hcNj/4πQ(T))
(Q(T) is the partition function). The values of the λji, gj, Aji and Ei for selected atomic or ionic
lines can be obtained from the NIST Atomic Spectral Database. A set of selected spectral
lines can be chosen based on their relative strengths, accuracies and transition probabilities.
The emission spectra of the diatomic species reveal a relatively complex structure which
is due to the combination of the electronic transitions from the different rotational and
vibrational states [40-42]. The emission intensities of the molecular bands can be analyzed in
order to calculate the molecular vibrational temperature Tvib. For a plasma in LTE, the
intensity of an individual vibrational v’-v” band Iv’-v” is given by
⎛I ⋅ λ4 ⎞ G ( v' ) h c
ln⎜⎜ v ' − v" v ' − v" ⎟⎟ = A − , (2.24)
⎝ q v ' − v" ⎠ k B ⋅ Tvib
where A is a constant, λv’-v” is the wavelength corresponding to the band head,

∞ 2
q
v'− v"
= ∫ Ψv' ( R)Ψv" ( R)dR is the Franck-Condon factor and G ( v' )h c / k B is the normalized
0
energy of the upper vibrational level. A line fit to ( )

ln I v '− v" ⋅ λ4v '− v" / q v '− v" as a function of
the upper normalized electronic-vibrational energies has a slope equal to -1/Tvib.
On the other hand, the emission intensities of the rotational lines of a vibrational band can
be analyzed in order to estimate the effective rotational temperature Trot. In this case it is
necessary to consider the Hund´s coupling case for the both electronic states implied in the
transition. From the assignment of the rotational spectrum it is possible to estimate the
effective rotational temperature by considering the J value for the maximum of the band
Trot=(2 Bv h c/kB)(Jmax+1/2)2, being Bv the rotational constant for upper v vibrational level and
Jmax the total angular momentum at the maximum.
Another method for estimating the vibrational and rotational temperatures is based on a
simulation program of the spectra. Software developed in our laboratory [43] calculated the
spectra of a diatomic molecule by summing the intensity of all rovibrational levels and
convoluting the results with the instrumental line shape of the optical system. The emission
intensity Iv’,J’-v”,J” of a molecular line can be approximated by
64π 4ν~ 4 v ', J ' − v ", J " 2

I v',J' − v",J" ≈ N v ', J ' Re qv ', v " S J ', J " , (2.25)
3( 2 J '+1)
where ν~v ', J ' − v ", J " is the wavenumber of the transition, 2J’+1 is the rotational degeneracy of
the upper state, Nv’,J’ is the population in the initial (upper) state, R e is the average electronic
transition moment, qv’,v” is the Franck-Condon factor and SJ’,J” is the Hönl-London factor
[44]. Spectrum simulations are based on comparison of experimental and calculated spectra
for different rotational and vibrational population distributions which depend on temperature.
2.4.5. Ionization degree of the plasmas: Saha equation
In plasma there is a continuous transition from gases with neutral atoms to a state with
ionized atoms, which is determined by an ionization equation. The transition between a gas
and a plasma is essentially a chemical equilibrium, which shifts from the gas to the plasma
side with increasing temperature. Let us consider the first three different ionization equilibria
of an atom A:
A ↔ A+ + e + IP(A-I),
A+ ↔ A2+ + e + IP(A-II),
A2+ ↔ A3+ + e + IP(A-III).

For each ionization equilibrium, considering the atoms and ions in their ground electronic
state, the LTE between ionization and recombination reactions at temperature T is described
by the Saha equation (2.3)
(
ne ⋅ N i g e ⋅ g i 2π m k B T
=
)
3/ 2
e − Ei / k B T , (2.26)
N0 g0 h3
where ne = Ni are the electron and ion densities in the different ionization equilibria in the
second member of ionization equilibria. From this equation, ionization degree ne·Ni/N0 can be
estimated.
2.5. Effects of Physical Variables in LIBS
The variables that can influence the LIBS measurements are mainly the laser properties
i.e. wavelength, energy, pulse duration, focusing spot size, shot-to-shot energy fluctuations,
ambient conditions, physical properties of the sample and the detection window (delay time
and gate width). How these parameters affect the precision and accuracy of LIBS are
addressed below.
2.5.1. Laser parameters
In LIBS a high-power laser is used to ablate or to breakdown a gaseous sample in the

form of plasma. The primary energy related parameters influencing the laser-gas interaction
are the laser peak power PW (or radiant pulse energy per time, in W) and the laser peak
intensity IW (power density or irradiance; energy per unit area and time, W cm-2) given by
PW = EW / τ FWHM , (2.27)
I W = PW / πr 2 , (2.28)
where EW (in J) is the pulse energy, τFWHM (in s) is the pulse duration at the FWHM and πr2 is
the focal spot area (cm2). The fluence ΦW (in J cm-2) on the focused spot area, the photon flux
density Fph (photon cm-2 s-1) and electric field FE (V cm-1) are given by
ΦW = EW / πr 2 , (2.29)
Fph = IW λ / hc , (2.30)
FE = IW / cε 0 , (2.31)
where λ is the laser wavelength, h is the Planck constant, c is the speed of light, and ε0 is the
electric constant. The laser peak intensity IW, fluence, photon flux and electric field are
inversely proportional to the focused spot area. For LIBS, the peak intensity IW (and the other
properties ΦW, Fph, FE and PR) that can be delivered to the sample is more important than the
absolute value of the laser power. For the formation of plasma, the laser fluence needs to
exceed the threshold value, typically of the order of several J cm-2 for a nanosecond laser
pulse [45]. If the laser energy is very close to the breakdown threshold, the pulse-to-pulse
fluctuations can cause the plasma condition to be irreproducible, which results in poor
measurement precision. The intensities of the emission lines are proportional to the laser
energy while the laser plasma is in the optical thin region. When the laser energy increases
further, it produces very dense and hot plasma that can absorb laser energy. This will lead to
an increase in the continuum emission and a decrease in the signal intensity. Besides, the laser
pulse duration and the shot-to-shot fluctuations can also affect the signal reproducibility and
hence LIBS precision.
2.5.2. Focal properties
The laser power density at the focal volume is inversely proportional to the focused spot
size. For a laser beam with a Gaussian profile, the focused beam waist w0 is given by [46]
λ f
w0 = , (2.32)
π ws
where f is the focal length of the lens and ws is the radius of the unfocused beam. The higher
laser power density at the focal point can be achieved by reducing the focused beam waist
using a shorter focal length lens.
On the other hand, the angular spread in laser light generated by the diffraction of
plane waves passing through a circular aperture consists of a central, bright circular
spot (the Airy disk) surrounded by a series of bright rings. The beam divergence angle
θ, measured to edges of Airy disk, is given by θ=2.44 λ/d, where λ is the laser
wavelength and d is the diameter of the circular aperture. It can be shown that a laser
beam, with beam divergence θi , incident on a lens of focal length f, whose diameter is
several times larger than the width of the incident beam, is focused to a diffraction-
limited spot of diameter approximately equal to f θi . If the focal region of the laser
beam is assumed to be cylindrical in shape, the spot size in terms of length l, can be
approximated as
l = ( 2 − 1)θ i f 2 / d . (2.33)
2.5.3. Laser absorption in the plasma
In LIBS the evaporation of the material begins just after the impact of the leading edge of
the laser pulse on the surface. The time required for the removal of the material is less than a
nanosecond, thus once the plasma is formed, a part of the laser beam will be absorbed in the
plasma by the electron-neutral or electron-ion inverse bremstrahlung (e-n IB and e-i IB), or
by photoionization (PI) of the excited states. Consequently not the full laser irradiance will be
able to reach the target, which is called plasma shielding. In the case of IB absorption the free
electrons gain kinetic energy from the laser beam thus promoting plume ionization and
excitation through collisions with excited and ground state neutrals. The IB process is usually
described by the inverse absorption length αIB (cm-1) [31, 47]
α IB , e − n = σ e − n [1 − e− hν / kT ]ne N 0 , (2.34)
4e6λ3 2π
α IB, e −i = neσ e −i = [1 − e− hν / kT ] ne Ni , (2.35)
3hc 4 m 3mkT
where N0 is the neutral atomic density, Ni is the ionic density, σe-n is the electron-neutral cross
section of photon absorption and σe-i is the electron-ion cross section of photon absorption.
The term [1 − e− hν / kT ] represents the losses due to stimulated emission. The electron number
density in the plasma depends on the degree of ionization, evaporation rate and the plasma
expansion velocity. Moreover, the absorption coefficient shows different temperature
dependence for different energy density regimes. In the case of short wavelength lasers the
photoionization of the excited atoms can play significant role. In fact, the absorption
coefficient of this process σPI (cm-1) is given by [48]
( I P )5 / 2
α PI = N nσ PI ≈ ∑ 2.9 × 10 −17
Nn , (2.36)
n ( hν ) 3
being Nn is the number density (in cm-3) of the excited state n, Ip is the ionization potential in
eV and hν is the photon energy in eV. In this equation the summation is performed over the
energy levels which satisfy the condition hν>En. Equation (2.36), although derived for
hydrogen like atoms, can be applied to complex atomic systems.
3. EXPERIMENTAL
LIBS is a plasma based method that uses instrumentation similar to that used by other
spectroscopic methods (atomic emission spectroscopy, laser-induced fluorescence etc). A
typical LIBS apparatus utilizes a pulsed laser that generates the powerful optical pulses used
to form the plasma. Principles of laser operation in general and the operation of specific lasers
are described in detail in numerous books. The discussion here will be limited to the
fundamentals of the operation of the transversely excited atmospheric (TEA) carbon dioxide
laser used in this work. On the other hand, it is necessary a convenient detection system. We
described here the detection system used in this work.
3.1. Pulsed TEA CO2 Laser
The CO2 laser is a near-infrared gas laser capable of very high power and with the highest
efficiency of all gas lasers (≈10-20%) and for cw operation the highest output power. Although
CO2 lasers have found many applications including surgical procedure, their popular image is as
powerful devices for cutting, drilling, welding or as weapons for military applications. The
linear CO2 molecule has three normal modes of vibration, labelled ν1 (symmetry stretch), ν2
(bending vibration) and ν3 (asymmetric stretch), and plotted in the upper part of Figure 1. The
fundamental vibration wavenumbers are 1354, 673 and 2396 cm-1, respectively. The vibrational
state of the molecule is described by the number of vibrational quanta in these modes. The
bending vibrational mode is twofold degenerate and can have a vibrational angular momentum
along the CO2 axis. The number of quanta of this vibrational angular momentum is stated as an
upper index to the vibrational ν2 quanta. The upper laser level (0001) denotes the ground
vibrational state for the mode ν1, the ground vibrational state for the mode ν2 which is doubly
degenerate, and the first excited vibrational state for the mode ν3. The active medium is a gas
discharge in a mixture of He, N2 and CO2. By electron impact in the discharge excited
vibrational levels in the electronic ground states of N2 and CO2 are populated (Figure 1). The
vibrational levels v = 1 in the N2 molecule and (ν1, ν2, ν3) = (0001) in the CO2 molecule are near-
resonant and energy transfer from the N2 molecule to the CO2 molecule becomes very efficient.
This populates the (0001) level in CO2 preferentially, creates inversion between the (0001) and
the (0200) levels, and allows laser oscillations on many rotational transitions between these two
vibrational states in the wavelength range 9.6-10.6 µm. The main laser transitions in CO2 occur
between the excited states of the mode ν3(0001) and the symmetric stretching mode ν1(1000)
(10.6 μm) or the bending mode ν2(0200) (9.6 μm). A single line can be selected by a Littrow-
grating, forming one of the resonator end mirrors.
Figure 1. Level scheme and the three normal vibrational modes of the CO2 molecule.
Helium atoms do not take part directly in the excitation of CO2 molecules but do play an
important role in heat-transfer from the gas mixture to the tube walls, as well as facilitating the
depopulation of the lower vibrational levels in CO2; contributing in this way to the ,maintenance
of the population inversion. The power of CO2 lasers depends on their configuration. The laser
used in these experiments was a transversely excited atmospheric (TEA) CO2 laser in which the
electric discharge is transverse to the laser cavity’s axis. The pressure in the tube is close to
atmospheric pressure. The CO2:N2:He mixture is exchange in a continuous way, enhancing the
output power of the laser. The laser can achieve a power of 50 MW. The optical materials used in
lasers emitting radiation in the infrared range are obviously different than those used in the visible
range. For example, materials such as germanium (Ge) or gallium arsenide (GaAs) are completely
opaque in the visible range, while being transparent in the infrared range. Some materials, such as
zinc selenide (ZnSe), are transparent in both spectral ranges. Typical materials transparent in the
IR range are: NaCl or CsI. Metal mirrors (copper, molybdenum, gold) are used in the IR range,
owing to their small absorption (and large reflectivity) as well as their large heat capacity which
enables removal of heat from the active medium.
3.2. Spectrographs and Detectors
A detection system consists in a wavelength dispersing element and an electronic device as

detector. Wavelength dispersing elements used for LIBS generally have to fulfilled two opposite
requests: (i) need for high resolving power λ/∆λ because of highly pronounced spectral interferences
in atomic-molecular emission spectroscopy; (ii) For an analysis with a single laser pulse the wide
spectral region should be covered. Today, usually, four different spectrograph mountings are mainly
in use for OES: Czerny-Turner, Echelle, Rowland and Paschen-Runge. Here, only the first system
used by us will be described. A Czerny-Turner mount uses a plane grating. The incident radiation
passes through the entrance slit and strikes a parabolic collimating mirror. This mirror produces a
collimated light beam reflected onto the grating, which spatially disperses the spectral components of
the incident radiation. Collimated rays of diffracted radiation strikes a second parabolic focusing
mirror. The dispersed radiation is focused in the focal plane producing the entrance slit images in that
plane. Because the parallel rays of a given wavelength are incident on the focusing element at a
specific angle, each wavelength is focused to a slit image at a different center position on the focal
plane. In this focal plane the detector is placed (see Figure 2).
For the detection of plasma emissions in LIBS experiments, photomultipliers (PMT),
charge-coupled and intensified charge coupled devices (CCD and ICCD) and optical
multichannel analyzers (OMA) can be utilized depending on the type of spectrometer. A PMT
consists of the photo-emissive cathode, several dynodes, and an anode. When the radiation
hits the chatted electrons are generated. A high voltage across the dynodes leads to further
multiplication of these photoelectrons. A PMT is relatively inexpensive and very sensitive
detector, which allows recording of time-integrated plasma emission signals. It requires,
however, a scanning mode monochromator to obtain full spectral information. Therefore, the
PMT is not convenient for multielement analysis with a single laser pulse or when fast scan
results are required. The charge coupled devices (CCDs) are made up of a two-dimensional
array of semiconductor capacitors (pixels) that have been formed on a single silicon chip.
Commercially available CCD chips usually consisted of 512×512 or 1024×1024 pixels.
Performance characteristics of these instruments with respect to sensitivity, dynamic range,
and signal-to-noise ratio appear to approach those of PMT’s in addition of having the
multichannel advantage. A CCD system can be equipped with an intensifier (microchannel
plate) which allows high photonics gain and possibility of electronic gating which is of
essential importance for time resolved LIBS measurements.
Figure 2. Schematic diagram of the experimental set-up for pulsed laser ablation diagnostics.
3.3. Schematic Diagram for LIBS
The experimental configuration used to study carbon by LIBS is shown in Figure 2. The
laser-induced plasma was generated using a TEA CO2 laser (Lumonics model K-103)
operating on an 8:8:84 mixture of CO2:N2:He, respectively. The laser is equipped with frontal
Ge multimode optics (35 % reflectivity) and a rear diffraction grating with 135 lines mm-1
blazed at 10.6 μm. The CO2 laser irradiation of the target was carried out using the 9P(28)
line at λ=9.621 µm (ΔE=0.1289 eV) and the 10P(20) line at λ=10.591 µm (ΔE=0.1171 eV).
The temporal shape of the TEA-CO2 laser pulse, monitored with a photon drag detector
(Rofin Sinar 7415), consisted in both cases in a prominent spike of a full width at half
maximum (FWHM) of around 64 ns carrying ∼90% of the laser energy, followed by a long
lasting tail of lower energy and about 3 μs duration. The laser pulse repetition rate was
usually 1 Hz. The divergence of the emitted laser beam is 3 mrad. The pulsed CO2 laser beam
was focused with a NaCl lens of 24 cm focal length onto the target. The primary laser beam
was angulary defined and attenuated by a diaphragm of 17.5 mm diameter before entering to
the cell. The CO2 laser average energy was measured in front of the lens with a Lumonics
20D pyroelectric detectors through a Tektronix TDS 540 digital oscilloscope. Energy losses
were estimated by making pulse energy measurements with and without the NaCl window in
place. The focused radius of the laser beam (0.05 cm) was measured at the target position
with a pyroelectric array Delta Development Mark IV. The laser intensity (power density or
irradiance) on the focal volume was varied using several calibrated CaF2 attenuators and
range from 0.29 to 6.31 GW×cm-2. The high purity graphite target (~99.99 %) was placed in a
low-vacuum cell equipped with a NaCl window for the laser beam and two quartz windows
for optical access. The graphite target is initially at ambient temperature (298 K) and it is not
water-cooled. The cell was evacuated with the aid of a rotary pump, to a base pressure of 4 Pa
that was measured by a mechanical gauge. Optical emission from the plume was imaged by a
collecting optical system onto the entrance slit of different spectrometers. The light emitted
from the laser-induced plasma was optically imaged 1:1, at right angles to the normal of the
target surface, by a quartz lens (focal length 4 cm, f-number = f/2.3) onto the entrance slit of
the spectrometer. The distance between plasma axis and entrance slit was 16 cm. The lens
causes a bit chromatic aberration for OES measurements, although the geometric efficiency is
barely affected. Two spectrometers were used: 1/8 m Oriel spectrometer (10 and 25 μm slits)
with two different gratings (1200 and 2400 grooves mm-1) in the spectral region 2000-11000
Å at a resolution of ~1.3 Å in first-order (1200 grooves mm-1 grating), and an ISA Jobin
Yvon Spex (Model HR320) 0.32 m equipped with a plane holographic grating (2400 grooves
mm-1) in the spectral region 2000-7500 Å at a resolution of ~0.12 Å in first-order.
Two detectors were attached to the exit focal plane of the spectrographs and used to detect
the optical emissions from the laser-induced plasma: an Andor DU420-OE (open electrode)
CCD camera (1024 256 matrix of 26 26 μm2 individual pixels) with thermoelectric cooling
working at –30 ºC; A 1024 1024 matrix of 13 13 μm2 individual pixels ICCD (Andor iStar
DH-734), with thermoelectric cooling working at –20 ºC. The low noise level of the CCD
allows long integration times and therefore the detection of very low emission intensities. The
intensity response of the detection system was calibrated with a standard (Osram No.6438, 6.6-
A, 200-W) halogen lamp and Hg/Ar pencil lamp. Several (Cu/Ne, Fe/Ne and Cr/Ar) hollow
cathode lamps (HCL) were used for the spectral wavelength calibration of the spectrometers.
3.4. Timing Considerations
For synchronization, the CO2 laser was operated at the internal trigger mode and the ICCD
detector in external and gate modes. The external trigger signal generated by the laser is fed
directly into the back of the ICCD detector head. The total insertion delay (or propagation
delay) is the total length of time taken for the external trigger pulse to travel through the digital
delay generator and gater so that the ICCD will switch on. This insertion delay time is 45 ± 2 ns.
The time jitter between the laser and the fast ICCD detector gate was about ± 2 ns. The delay
time td is the time interval between the arrival of the laser pulse on the target and the activation
of the ICCD detector. The gate width delay time tw is the time interval during which the plasma
emission is monitored by the ICCD. Both parameters were adjusted by the digital delay
generator of the ICCD detector. The resolution of the gate pulse delay time and the gate pulse
width time are 25 ps. The CO2 laser pulse picked up with the photon drag detector triggers a
Stanford DG 535 pulse generator which is used as external trigger in the ICCD camera. The
laser pulse and the gate monitor output were displayed in a Tektronix TDS 540 digital
oscilloscope, allowing to control td eliminating the insertion time of the camera.
4. RESULTS AND DISCUSSION

When a high-power laser pulse is focused on a solid surface the target becomes ablated.
If the laser irradiance in the focal volume surpasses the breakdown threshold of the system
formed by the vaporized atoms and residual gas, a breakdown, characterized by a brilliant
flash of light accompanied by a distinctive cracking noise, is produced. At the top of figure 2
we show an image of laser-induced breakdown (LIB) plasma in graphite induced by a single
CO2 laser pulse. The plume size is around 14 cm. The laser was focused on a point at the
centre of the target. The observations of the LIB geometry during the experiments indicate
that the actual plasma region is not entirely spherical, but lightly elongated in the direction of
the laser beam propagation. The CO2 laser pulse remains in the focal volume after the plasma
formation for some significant fraction of its duration and the plasma formed can be heated to
very high temperatures and pressures by inverse bremsstrahlung absorption. Since plasmas
absorb radiation much more strongly than ordinary mater, plasmas can block transmission of
incoming laser light to a significant degree; a phenomenon known as plasma shielding [49].
The high temperatures and pressures produced by plasma absorption can lead to thermal
expansion of the plasma at high velocities, producing an audible acoustic signature, shock
waves, and cavitation effects. The plasma also tends to expand back along the beam path
toward the laser, a phenomenon known as moving breakdown. The shock wave heats up the
surrounding gas which is instantaneously transformed in strongly ionized plasma.
For the present experiments the measured focused-spot area was 7.85×10-3 cm2. This
value is higher than the calculated area (2.2×10-4 cm2) obtained from the beam waist (Eq.
2.32). This fact is due to the non-gaussian profile of the CO2 laser beam. Moreover the CO2
laser beam passes through a circular aperture of diameter 17.5 mm. For this diaphragm the
calculated divergence angle for the laser beams at 9.621 and 10.591 µm are 1.3 and 1.5 mrad,
respectively. Thus, considering the total beam divergence (∼4.4 mrad), the calculated
diameter of the focused TEA-CO2 laser (beam waist) is 1.06 mm, which is very similar to the
measured value (∼1 mm). If the focal region of the laser beam is assumed to be cylindrical in
shape, the spot size in terms of length l (Eq. 2.33) of the focused TEA-CO2 laser is 6.0 mm,
which is similar to the measured value (∼7 mm).
Table 1. Laser parameters for some LIBS experiments.
Energy Power Intensity Fluence Photon Flux, Fph Electric Field FE

EW (mJ) PW (MW) IW (GW cm-2) ΦW (J cm-2) (photon cm-2 s-1) (MV cm-1)
3161 49.5 6.31 403 3.07×1029 1.54
2685 42.1 5.36 342 2.60×1029 1.42
2256 35.4 4.50 287 2.18×1029 1.30
1732 27.1 3.46 220 1.67×1029 1.14
1209 19.0 2.41 154 1.17×1029 0.985
503 7.88 1.00 64.0 4.86×1028 0.615
324 5.08 0.648 41.3 3.14×1028 0.494
273 4.27 0.544 34.7 2.64×1028 0.453
242 3.79 0.483 30.8 2.34×1028 0.427
203 3.19 0.406 25.9 1.97×1028 0.391
171 2.68 0.341 21.8 1.65×1028 0.358
149 2.33 0.297 19.0 1.44×1028 0.335
131 2.08 0.262 16.7 1.26×1028 0.314
110 1.73 0.220 14.0 1.07×1028 0.288
2 +
50000
7-4 CN: B Σ -X Σ
Δv=+2
(b)
OH: A2Σ+-X2Πi Δv=0

C2+
400000
C3+
C+
Δv=+3
C+
2 +
9-7
11-9
(a) 45000
C+
C+
5-3 6-4
C2: D1Σ+u-X1Σ+g Δv=0
C+
350000
C
40000
5-2 6-3
Relative Intensity / a. u.
10-8
2+
300000
C+
35000
C+ CN: B2Σ+-X2Σ+
C2+ C2+
10-7
1 + 1
C2: E Σ g-A Πu
Δv=-1
250000 30000 2 +
CN: B Σ -X Σ
2 +
C+
Δv=+2, 1, 0, -1 3 3
Δv=0 C2: d Πg-a Πu
CH: B2Σ--X2Πr v'=0-v"=0

C4+
C2+
25000
CH: A2Δ-X2Πr
3 3
C2: e Πg-a Πu
C2+
200000 Δv=+1
0-0
C
2 2
CN: D Πi-A Πi
v'=0-v"=0
20000
CN: B2Σ+-X2Σ+
C3+
1-0 C : C1Π -A1Π
150000 2000 2200 2400 2600 2800 3000 3200 3400
Δv=+1
NH: A3Πi-X3Σ-
v'-v"
C+
15000
CH: C2Σ+-X2Πr
0-0
C+
2+
C+
C
3-2
C
100000 C Hγ
2-2 1-1
4-3
C3+
C2+ C2+
10000
10-9 11-10
1-02-1
1-1
Hβ
C2+
50000
0-0
CC2+2+ +
12-11
5000
C
C+
2+
4+
C2+
C
C+
C
0 0
2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000 4200 4400 4600 4800
Air Wavelength / Å Air Wavelength / Å
C C+
100000 N
160000
+
v'-v"
(d)
0-0
N
(c)
C+
+
C
2 2 +
140000 CN:AΠ- X Σ 2 2 +
3
C2: d Πg-a Πu
3
i CN: AΠi- X Σ
80000 Δv=+4
6-3
Δv=0 Δv=+3
5-2
120000
4-1
C2+
6-2
N
3-0
C2+ 8-4
5-1
1-1
100000
7-3
60000
+
C 3 3 80000
C2: d Πg-a Πu
C+
Δv=-1
40000
2-2
+
60000
C 6300 6400 6500 6600 6700 6800 6900 7000 7100 7200 7300 7400 7500 7600
1-2
2-3
C C C+
3-4
3 3
C C2: d Πg-a Πu
0-1
40000
2+
4-5
20000 C Δv=-2
C3+
C2+
C4+
C+
Hβ C 20000
C2+ +
4-6
3-5
C2+
2-4
Hα
C
+
C
1-3
C
0-2
0 C 0
5000 5200 5400 5600 5800 6000 6200 6400 6600 6800 7000 7200 7400 7600
1 1 +
C2: AΠu-X Σg 2 2 +
CN: AΠi-X Σ N 2 +
O CN: A Πi- X Σ
2
2-0
14000 Δv=+2 10000

N
(e)
3-0
(f) 9000 Δv=+1 C

1-0
O C
NN
3-1
O C
12000 N C
N C
8000
C N
10000 C
7000 C
4-2
6000
N
8000 N
2-1
5000 C
N
6000
5-3
4000
N
4000 N 3000 C
N
2000
2000
1000
0 0
7600 7800 8000 8200 8400 8600 8800 8800 8900 9000 9100 9200 9300 9400 9500 9600
Figure 3 (a)-(f). Low-resolution PLA of carbon emission spectrum observed in the 1920-9680 Å region
at an air pressure of 4 Pa, excited by the 10P(20) line at 944.20 cm-1 of the CO2 laser, and assignment of
the atomic lines of C, C+, C2+, C3+, C4+, N, O and molecular bands of C2(E1Σ+g–A1Πu; Freymark system),
C2(D1Σu+– X1Σg+; Mulliken system), C2(e3Πg – a3Πu; Fox-Herzberg system), CN(D2Πi–A2Πi), OH(A2Σ+–
X2Π), CH(C2Σ+–X2Π), CN(B2Σ+–X2Σ+; Violet system), NH(A3Π–X3Σ-), C2(C1Πg– A1Πu; Deslandres-
d’Azambuja system), C2(d3Πg–a3Πu; Swan band system), CH(B2Σ-–X2Π), CH(A2Δ–X2Π), C2(A1Πu–
X1Σg+; Phillips system), and CN(A2Π–X2Σ+; red system).
4.1. Identification of the Chemical Species in the Pulse Laser Ablation

Plasma Plume
For the different pulse laser energies measured in this work, the calculated laser peak
power (Eq. 2.27), intensity (Eq. 2.28), fluence (Eq. 2.29), photon flux (Eq. 2.30), and electric
field (Eq. 2.31) are tabulated in Table 1. In the scanned spectral region, from UV to NIR,
OES reproduce particular emission of carbon plasmas in a low-vacuum air atmosphere (Pair=4
Pa). Typical time-integrated and spectral-resolved low-resolution OES from LIB of graphite
is shown in Figure 3(a)-(f). In the recording of the spectra of Figure 3(c-f) a cutoff filter was
used in order to suppress high diffraction orders. In general, the spectra of the PLA plume are
dominated by emission of strong electronic relaxation of excited atomic C, ionic fragments
C+, C2+ and C3+, and molecular features of C2(d3Πg–a3Πu; triplet Swan band system). The
medium-weak emission is mainly due to excited atomic N, H, O, ionic fragment C4+ and
molecular features of C2(E1Σ+g–A1Πu; Freymark system), C2(D1Σu+– X1Σg+; Mulliken system),
CN(D2Π–A2Π), C2(e3Πg–a3Πu; Fox-Herzberg system), C2(C1Πg–A1Πu; Deslandres-
d’Azambuja system), OH(A2Σ+–X2Π), CH(C2Σ+–X2Π), NH(A3Π–X3Σ-), CN(B2Σ+–X2Σ+; violet
system), CH(B2Σ-–X2Π), CH(A2Δ–X2Π), C2(A1Πu–X1Σg+; Phillips system) and CN(A2Π– X2Σ+;
Red system).
In the spectrum of Figure 3(a) in the 1920-3480 Å region, very strong atomic C, C+, C2+
and C3+ lines dominate, but also weak C4+ and molecular bands of C2(E-A; Δv=v’-v”=+2, +1,
0, -1 sequence from 200 to 222 nm), C2(D-X; Δv=0 sequence near 231.4 nm), CN(D-A; in the
spectral range 223 to 260 nm), C2(e-a; in the spectral range 240 to 290 nm), CN(B–X; Δv=3
sequence from 306 to 326 nm), OH(A–X; Δv=0 sequence from 306 to 318 nm), CH(C–X;
Δv=0 sequence from 314 to 317 nm), NH(A–X; Δv=0 sequence near 336 nm) and CN(B–X;
Δv=2 sequence from 326 to 348 nm) are observed. In this spectrum the predominant emitting
species are the C2+ 2p2 1D2 → 2s2p 1P01 atomic line at 2296.87 Å, C
1 1 3+
2 p ( 2 P0 )3s P10 → 2 p 2 S 0 atomic line at 2478.56 Å, two lines of C at 2524.41 and 2529.98
Å, several lines of C+ at 2836.71 and 2992.62 Å and the v’=0-v”=0 band of NH(A–X) at 3360
Å. In the spectrum of Figure 3(b), the predominant emitting species are C+ (doublet
2s 2 4s S1/ 2 → 2s 2 3 p P10/ 2,3 / 2 at 3918.98 and 3920.69 Å, respectively, and multiplet
2 2
2s 2 4 f FJ0' → 2s 2 3d DJ " around 4267 Å), and the molecular bands of CN(B–X; Δv=0
2 2
sequence). Many medium intensity atomic lines of C+, C2+ and C3+, weak hydrogen lines of
the Balmer series (Hβ, Hγ etc), and several molecular bands of CN, C2, and CH are also
present. In the spectrum of Figure 3(c), the predominant emitting species are C+ and C2
(molecular bands: d-a; Δv=0, -1, and -2 sequences from 480 to 630 nm). Many weak lines of
C, C+, C2+ and C3+ are also present. In the spectrum of Figure 3(d), the most intense lines are
the doublet structure of C+ 2 s 2 3 p 2 P30/ 2,1 / 2 → 2 s 2 3s 2 S1 / 2 at 6578.05 and 6582.88 Å,
respectively, C 2 s 2 2 p ( 2 P0 ) 4d 1 P10 → 2 s 2 2 p ( 2 P0 )3 p 1 P1 atomic line at 6587.61 Å, C+
+
2s 2 3d D3 / 2 → 2s 2 3 p P10/ 2 at 7231.32 Å and C 2s 2 3d D5 / 2 → 2s 2 3 p P30/ 2 at 7236.42 Å.
2 2 2 2
Also many weak lines of C, C+, C2+, Hα, N, and several bands v’-v” (5-1, 6-2, 7-3, 8-4, 3-0, 4-
1, 5-2 and 6-3) corresponding to CN(A–X) are also present. The spectrum of Figure 3(e),
shows the emission of many atomic lines of C, O, and N, the 3-0 band of C2(A-X) and several
bands (2-0, 3-1, 4-2, and 5-3) of CN(A–X). Finally, in the spectrum of Figure 3(f), the
emission of many atomic lines of C and N and mainly the 1-0 and 2-1 bands of CN(A–X) can
be appreciated.
a) Pulsed laser ablation of Carbon
b) Acetylene/oxygen flame 1 1 +
C3: Ã Πu-X Σg
3 3 -
(000)−(000)
NH: A Πi-X Σ
Δv=0
2 2
CH: A Δ-X Πr
c) Propane-butane/air flame 3 3
2 - 2 0-0 C2: d Πg-a Πu
CH: B Σ -X Πr
0-0 Δv=+1
1-1
d) Free-burning carbon arc 2 +

CN: B Σ -X Σ
2 +
Δv=+2
Δv=+1
Δv=-1
Δv=-2
Δv=0 4400 4600 4800
3400 3600 3800 4000 4200 4400 4600 4800

Air Wavelength / Å
Figure 4. Low-resolution emission spectra from: a) PLA of carbon at an air pressure of 4 Pa, excited by
the 9P(28) line at 1039.36 cm-1 of the CO2 laser; b) Acetylene/oxygen flame; c) Propane-butane/air
flame; d) Free-burning carbon arc.
For the assignment of the atomic lines of C, C+, C2+, C3+, C4+, H, N and O we used the
information tabulated in NIST Atomic Spectral Database [36]. The observed emission
molecular bands are identified using the spectroscopic information available in Refs. [50].
Moreover, these molecular bands were compared with the spectra obtained in our laboratory
by conventional sources (free-burning carbon arc, propane-butane/air flame and
acetylene/oxygen flame). As example figure 4 shows several time-integrated OES at low-
resolution from: (a) PLA of carbon (air pressure of 4 Pa and CO2 laser power density IW=1.00
GW cm-2); (b) Acetylene/oxygen flame; (c) Propane-butane/air flame; (d) Free-burning
carbon arc. In the Acetylene/oxygen flame around 405 nm, several bands of the
~1
C3 ( A ~
Π u − X 1Σ +g ) comet head group are observed which were not detected in the PLA of
carbon. As shown in Figure 4, ionic carbon lines C+, C2+, C3+ and C4+ cannot be observed in
flames [Figure 4(b,c)] or carbon electric arcs [Figure 4(d)].
2 + 2
CH: C Σ -X Πr
(a) C+ (b) 3
NH: A Πi-X Σ
3 -
0-0
Δv=0
1-1
2 + 2 +
CN: B Σ -X Σ
3-3
2-2
1-1
Δv=0
0-0
6-3
5-2
Δv=+3
7-4
2 + 2
OH: A Σ -X Πi
1-1
0-0
Δv=0 2 + 2 +
CN: B Σ -X Σ
C2+: 2p( P )3s P 1-2s3d D2
P1 P2
1
9-7
10-8
8-6
Δv=+2
5-3
7-5
6-4
R2 Q2
Q1
R1
0
1
0
2
3000 3025 3050 3075 3100 3125 3150
2975 3000 3025 3050 3075 3100 3125 3150

3275 3300 3325 3350 3375 3400 3425
C+
2 + 2 +
CN: B Σ -X Σ
(d)
3/2
2-1
3-2
1-0
Δv=+1
4-3
C : 2s24s S1/2-2s23p P1/2

0
(c) C2: C1Πg-A1Πu
2 +
CN: B Σ -X Σ
2 +
2
0-0
2-2
3-3
1-1
Δv=0
4-4
1-0
2-1
4-3
3-2
Δv=+1
C+
2
2 +
CN: BΣ -X Σ
2 + C2: C1Πg-A1Πu
0-0
12-11
13-12
11-10
10-9
9-8
Δv=0
6-5
7-6
8-7
Δv=+1 (tail bands)

C+
+
2 - 2
C+C+
CH: B Σ -X Πr
0-0
3750 3775 3800 3825 3850 3875 3900 3925

3525 3550 3575 3600 3625 3650
Air Wavelength / Å
Air Wavelength/ Å
3 3
C2: d Πg-a Πu
C2+
C+
C+: 2s2p(3P0)3p 4S3/2-2s2p(3P0)3s 4P05/2

C+ C2+
(e) (f)
C2+
3 3
C2: d Πg-a Πu
C+: 2s2p(3P0)3p 4S3/2-2s2p(3P0)3s 4P03/2
0-1
2-3
1-2
3-4
2 2 Δv=-1
CH: A Δ-X Πr Δv=+2
3-1
4-2
2-0
R2 R1 0-0 P1 P2
C+
Q2 Q1
2 + 2 +
CN: B Σ -X Σ
Hγ
0-1
4-5
2-3
1-2
Δv=-1
3-4
4150 4175 4200 4225 4250 4275 4300 4325 4350 4375 5500 5525 5550 5575 5600 5625 5650
Figure 5 (a)-(f). Measured high-resolution PLA of carbon emission spectra observed in different
regions at an air pressure of 4 Pa, excited by the 9P(28) line of the CO2 laser with a laser intensity of
5.36 GW cm-2 , and assignment of some atomic lines and molecular band heads.
In order to get more insight into PLA of graphite and to obtain an unambiguous
assignment of the emission lines and molecular bands, we have scanned the corresponding
wavelength regions with higher resolution (~0.12 Å in first-order). The spectra have been
obtained with twenty-four successive exposures on the CCD camera in the spectral region
200-750 nm by a ISA Jobin Yvon Spex 0.32 m spectrometer. As examples, Figure 5(a-f)
shows several spectra recorded in the PLA of carbon experiment at high-resolution. These
spectra were recorded in the following experimental conditions: air pressure 4 Pa, CO2 laser
excitation line 9P(28) at 9.621 μm and laser intensity 5.36 GW cm-2. The relative intensities
of the observed emission lines reasonably agree with tabulated values in NIST Atomic
Spectral Database [36]. In Figure 5(a-f) we have indicated with italic the position of the band
heads v’-v” of violet system of CN while in regular typeface the bands of the other molecular
systems. In Figure 5(a-f), a rather complex structure is observed, in consequence of the
overlapping between rotational lines of different molecular band systems. Figure 5(a) displays
the overlapping between CH(C-X; Δv=0 sequence), CN(B-X; Δv=3 sequence), and OH(A-X;
Δv=0 sequence). The relative position of the main branches for the OH(A-X) 0-0 band is
indicated. In Figure 5(b), the high intensity of the 0-0 band for NH(A-X) is observed. This fact
is in agreement with the high Franck-Condon factor (q00=0.9998) for this transition. In Figure
5(c) a partial overlapping among CN(B-X; Δv=1) and C2(C-A; Δv=1) is observed. This
spectrum clearly shows the reversal of the bands from v”=5, which is due to the overlap
between high vibrational quantum number bands with low vibrational quantum number
bands. So, the first vibrational bands (1-0, 2-1, 3-2, 4-3 and 5-4) are shaded to the violet and
after reversal (6-5, 7-6, …) are shaded to the red. Figure 5(d) shows a portion of the rotational
lines for the CH(B-X) 0-0 band with several single ionized carbon lines. A coincidence in the
position among the CN(B-X) 4-4 and C2(C-A) 0-0 band heads is observed. Very weak
emission attributable to the N2+(B2Σu+–X2Σg+) system (the most prominent v’=0-v”=0
transition appears at ∼ 391 nm) is also identifiable. In the spectrum of Figure 5(e) the CN(B-
X) Δv=-1 sequence, CH(A-X) 0-0 band, and C2(d-a) Δv=2 sequence were identified. Also,
several C+, C2+, and atomic hydrogen lines are observed. Finally, Figure 5(f) displays the
rotational structure of C2(d-a) 0-1, 1-2, and 2-3 bands. The spectral features clearly show the
complexity of the relaxation process and bring out the possibility of cascading processes.
4.2. Plasma Excitation, Vibrational and Rotational Temperature

Measurements
The excitation temperature Texc was calculated from the relative intensities of some C+
atomic lines (250-470 nm spectral region) according to the Boltzmann equation (2.23). The
estimated excitation temperature was Texc= 23000 ± 1900 K (Figure 6). The relevant
spectroscopic parameters for the C+ transitions have been listed in Table 2.
0.5
0.0 C+
ln(Iki λki/gk Aki) (a.u.)

-0.5 Texc=23000 ± 1900 K
-1.0
-1.5
-2.0
190000 200000 210000 220000 230000 240000
Ek/kB (K)
Figure. 6. Linear Boltzmann plot for several C+ transition lines used to calculate plasma temperature,
Texc. Plot also shows linear fit to the data with a regression coefficient of R2~0.98.
Table 2. List of C+ transition lines and their spectral database

(NIST Atomic Spectra Database, 2006) used for plasma temperature calculation.
Transition array Air λ (Å) gi gj Aji (s-1) Ei (cm-1) Ej (cm-1) Rel. Int.
(Arb. Uni.)
2s2p2 2P1/2-2p3 2D03/2 2509.12 2 4 4.53×107 110624.17 150466.69 20795
2s2p2 2P3/2-2p3 2D05/2 2512.06 4 6 5.42×107 110665.56 150461.58 41490
2s23p 2P01/2-2s2 4d 2746.49 2 4 4.36×107 131724.37 168123.74 8500
2
D3/2
2s2p2 2S1/2-2s23p 2P03/2 2836.71 2 4 3.98×107 96493.74 131735.52 76920
2s2p2 2S1/2-2s23p 2P01/2 2837.60 2 2 3.97×107 96493.74 131724.37 44700
2s23p 2P01/2-2s2 4s 2S1/2 3918.98 2 2 6.36×107 131724.37 157234.07 7500
2s23p 2P03/2-2s2 4s 2S1/2 3920.69 4 2 1.27×108 131735.52 157234.07 16000
2s23d 2D3/2-2s2 4f 4267.00 4 6 2.23×108 145549.27 168978.34 45000
2 0
F 5/2
2s23d 2D5/2-2s2 4f 4267.26 6 8 2.38×108 145550.70 168978.34 70000
2 0
F 7/2
The detection of the C2(d–a) Swan and CN(B-X) bands is of particular interest since it
provides an estimation of the plasma vibrational temperature. The emission intensities of the
C2 Swan Δv=-1 and CN Δv=0 band sequences were analyzed in order to calculate the
molecular vibrational temperature Tvib. For a plasma in LTE, the intensity of an individual
vibrational v’-v” band Iv’-v” is given by Eq. (2.24). Two Boltzmann plots of the band
intensities against the vibrational energy are given in Figure 7, along with the corresponding
Franck-Condon factors. For C2 and CN the estimated vibrational temperatures were
Tvib=18800 ± 860 K [Figure 7(a)] and 21400 ± 900 K [Figure 7(b)], respectively. Figure 8
shows the variation of vibrational temperature at 4 Pa of air pressure with laser fluence. The
vibrational temperature is maximum at a most efficient laser fluence of 287 J cm-2. These
results are consistent with earlier reports on vibrational temperature by different authors
[6,11,12,16].
(a)
0.30 v'=1
v'=2
40.7
Franck-Condon factor,qv'-v"
0.25
v'=0 v'=3
0.20
40.6
ln(Iv'-v" λ4v'-v"/qv'-v") (a. u.)

v'=4
0.15
v'=5
40.5 0.10
v'=6
v'=7
0.05 v'=8
v'=9
40.4
0.00
1 2 3 4 5 6 7 8 9 10
40.3 v"
40.2 3 3
C2: d Πg-a Πu; Δv=−1
40.1
40.0 Tvib=18800 ± 860 K

39.9
30000 32000 34000 36000 38000 40000 42000

G(v')hc/kB (K)
(b) 34.3
v'=8
v'=11
0.35
v'=5
v'=13
34.2 0.30
Franck-Condon factor,qv'-v"
v'=3
0.25
ln(Iv'-v" λ4v'-v"/qv'-v") (a. u.)
34.1 v'=15
0.20
v'=1
0.15
v'=0
34.0 0.10
0.05
33.9 0.00
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
v"
33.8
2 + 2 +
33.7 CN: B Σ -X Σ ; Δv=0
33.6
Tvib=21400 ± 870 K
33.5
33.4
2000 4000 6000 8000 10000 12000 14000 16000

G(v')hc/kB (K)
Figure 7. Left (a) panel: Linear Boltzmann plot of the C2 Swan Δv=-1 band sequence intensity versus
the normalized energy of the upper vibrational level; Right (b) panel: Linear Boltzmann plot of the CN
violet Δv=0 band sequence intensity versus the normalized energy of the upper vibrational level;
Experimental conditions: laser power density of 4.5 GW cm-2 and vacuum pressure 4 Pa. Plots also
show linear fit to the data and the corresponding Franck-Condon factors.
21000
20000
19000
18000
Tvib (K)
17000
16000
15000
14000
13000
12000
50 100 150 200 250 300 350
-2
Laser Fluence (J×cm )
Figure 8. The vibrational temperature Tvib calculated from the C2 Swan Δv=-1 sequence bands as a
function of the CO2 laser fluence.
2 2
33/2
25/2
17/2
CH: A Δ -X Π r; v'=0-v"=0
11/2
3/2
R2e,R2f
Q2ef,Q 2fe J"=3/2, ...
P2e,P2f
J"=3/2, ... R12e,R12f
15/2
3/2
9/2
+ Q12ef,Q12fe
C
J"
P12e,P12f
17/2
13/2
2+
C
9/2
1/2
5/2
R21e,R21f
13/2
5/2
Q21ef,Q21fe J"=3/2, ... 3 3

C 2: d Π g-a Π u Δ v=+2
4-2
P21e,P21f
27/2
21/2
33/2
3-1
5-3
R1e,R1f
2-0
J"=3/2, ... Q1ef,Q1fe J"=7/2, ...

P1e,P1f
25/2
15/2
7/2
J"
4200 4225 4250 4275 4300 4325 4350 4375 4400
Wavelength / Å
Figure 9. High-resolution PLA of carbon emission spectra at an air pressure of 4 Pa, excited by the
10P(20) line of the CO2 laser with a laser intensity of 6.31 GW cm-2, and assignment of some ionic
carbon lines, the band heads of C2(d-a) Δv=+2 sequence and the rotational structure of the 0-0 A2Δ-X2∏
band of CH.
Figure 9 presents a typical resolved LIB emission spectrum of graphite and its rotational
assignment for the v’=0-v”=0 band of the A2Δ-X2Π system of CH. This electronic band is the
strongest visible feature of the air, oxygen/acetylene flame and a dominant feature of all
hydrocarbon combustion (see figure 4). This spectrum consists in a doublet due to a transition
between a Δ upper state (Hund case (b), Ae=-1.11 cm-1) and a Π ground state intermediate
between Hund cases (a) and (b) (Ae=+28.1 cm-1) depending on the J value. The fine-structure
components are indicated on the branch designations by subscript: 1 ≡ 2Δ5/2-2Π3/2 and
subscript 2 ≡ 2Δ3/2-2Π1/2. The upper and lower states, Λ doublets are labelled by e and f; when
there are the same for both rotational levels ee is abbreviated by e and ff is abbreviated by f.
The selection rules involving the parity levels (e or f) are: e ∀# f, e ΠΤ e, f ΠΤ f for ΔJ=±1 (R
and P branches) and e ΠΤ f, e ∀# e, f ∀# f for ΔJ=0 (Q branches) [42]. For each value of the
quantum number N (N=J-S), there are four nearly degenerate energy levels, e or f, J=N±1/2.
Based on these assumptions we can expect 12 main branches (ΔJ=ΔN corresponding to R, P
and Q branches) and 12 satellite branches (ΔJ≠ΔN). A partial overlapping of the 0-0 band in
the region of P1 and P2 branches of the Δv=+2 of the C2 Swan band, whereas two lines of C+
and C2+ are also present. To estimate the effective rotational temperature, we consider the J
value for the maximum of the 0-0 band (A-X) of CH. This effective rotational temperature is
found to be Trot=2060 ± 50 K for Jmax=13/2 of the R1 branch. Figure 10 shows the simulated
spectra for the 0-0 A2Δ-X2∏ band of CH calculated (Eq. 2.25) at different temperatures. A
good agreement between simulated and observed spectra at a temperature of about 2000 K
over the entire range 4200-4400 Å proves that self-absorption is negligible and the rotational
levels follow a Boltzmann distribution.
C H : A 2 Δ -X 2 Π v '= 0 -v " = 0
T ro t = 1 0 0 0 K T ro t = 2 0 0 0 K
4200 4250 4300 4350 4 4 04

02 0 0 4250 4300 4350 4400
T ro t = 3 0 0 0 K T ro t = 4 0 0 0 K
4200 4250 4300 4350 4 4 04

02 0 0 4250 4300 4350 4400
W a v e le n g th / Å W a v e le n g th / Å
Figure 10. Simulated spectra for the 0-0 A2Δ-X2∏ band of CH calculated at different temperatures.
4.3. Ionization Degree of the Plasma
As the CO2 laser beam is focused on the graphite surface, the carbon material absorbs the
laser energy to melt, vaporize, and excite the target material. The carbon vapor absorbs more
energy and forms high temperature plasma near the surface. The plasma expands into the
low-vacuum atmosphere (N2, O2, H2O, etc) and transfers its energy to it. If the pressure
around the target is bigger than ~1000 Pa, the breakdown of the air takes place in a significant
way. Neutral, single and highly ionized carbon emission lines are found close to the target
graphite surface. The carbon clusters and the molecules of the atmosphere obtain an energy
that exceeds the binding energy. In these conditions the plasma becomes a mixture of
electrons, positive ions such as C+, C2+, C3+, C3+, C4+, neutral atoms such as C, N, O and H,
and molecules such as C2, CN, CH, NH, and OH in excited electronic states. Let us consider
the first three different ionization equilibria of carbon:
3 2 0
C(2s22p2 P0 ) ↔ C+(2s22p P1 / 2 ) + e + IP(C-I),
2 0 1
C+(2s22p P1 / 2 ) ↔ C2+(2s2 S 0 ) + e + IP(C-II),
1 2
C2+(2s2 S 0 ) ↔ C3+(2s1 S1 / 2 ) + e + IP(C-III),
where the first three ionization potentials (IPs) for carbon are IP(C-I)=11.2603 eV, IP(C-
II)=24.3833 eV and IP(C-III)=47.8878 eV [51]. Taking into account the consideration of
section 2.4.5, we can obtain the ionization degree. Figure 11 shows the ionization degree
Ni/(N0+Ni) of C, C+ and C2+, plotted as a function of the gas temperature T at a constant total
pressure P=(N0+ne+Ni)kBT. The graph shows that carbon is already fully ionized at thermal
energies well below the first ionization-energy of 11.2603 eV (equivalent to 130670 K). If we
consider a temperature of 23000 K, the ionization degrees of C, C+ and C2+ obtained by
means of the Saha equation are 0.999, 0.999 and 0.28, respectively. These so high values of
the ionization degrees justify the observed emission spectra. Keeping in mind these results,
the temperature obtained from the relative intensity of C+ lines was chosen as the first
approximation for the average excitation temperature.
4.4. Electron Number Density
The electron number density was obtained by considering the discussion reported in
section 2.4.3. In our experiments, for C+ lines, the Doppler line widths are 0.08-0.13 Å at
23000 K (Eq. 2.18). Stark line broadening from collisions of charged species is the primary
mechanism influencing the emission spectra in these experiments. In our case, the estimation
of electron density ne has been carried out by measuring the broadening of the spectral profiles
of isolated lines of C+ (2174, 2747, 2837, 2993, 3877, 3920, 4267, and 5890 Å) from the
time-integrated high-resolution spectra. The electron number densities of the laser-induced
plasma were determined at a laser power density of IW=1 GW cm-2 and air pressure of 4 Pa. A
Lorentz functions were used to fit the spectra. In order to extract the Stark broadening from
the total experimentally measured line broadening, we have to previously deconvolute the
main effects that contribute to the broadening of the spectral line. Values of the electron
impact half-width W were taken from the extensive tables given by Griem [27]. Electron
densities in the range (0.69-5.6)×1016 cm-3, with an estimated uncertainty of 10%, were
determined. At the evaluated temperature of 23000 ± 1900 K, Eq. (2.4) yields ne≈(0.39-
2.2)×1016 cm–3. These electron densities are close to measured values. Based on these
calculations, it is difficult to tell whether the plasma is in LTE or not. A possible reason for
non-thermal equilibrium could be the large integration time used in the experiments.
The formation of ionic species is a usual phenomenon in LIB technique. The interaction
between the laser and the ablation plume is governed by EII and/or by multiphoton ionization,
both followed by electron cascade. EII is the most important for the longer wavelengths used
in this work. MPI (Eq. 2.10) on the other hand is relatively improbable for carbon atoms in
the ground state C(2s22p2 3P0), since its high ionisation potential (11.2603 eV [51]), means
that 88 photons are required for this process. The observed ionic emissions are best explained
by an EII mechanism (Eq. 2.11). The free or quasifree electrons are produced by the high-
power laser pulse at the target surface. These electrons gain sufficient energy from the laser
field through inverse bremsstrahlung collisions with neutrals, to ionize carbon atoms or ions
by inelastic electron-particle collisions resulting in two electrons of lower energy being
available to start the process again (Eq. 2.11). In general, the probability of MPI is
WMPI ∝ ΦWn ∝ FE2 n . Calculations of PMI probability for carbon give a negligible value of WMPI
for the CO2 laser at λ=10.591 µm and IW=6.31 GW×cm-2 (n=88) (Eq. 2.9).
1.0
0.9
0.8
2+
Ionization degree
C
0.7
+
C
0.6
C
0.5
2 3 + 2 0
0.4 C(2p P0) ↔C (2p P1/2)+1e
+ 2 0 2+ 2 1
C (2p P1/2)↔C (2s S0)+1e
0.3
2+ 2 1 3+ 2
C (2s S0)↔C (2s S1/2)+1e
0.2
0.1
0.0
0 5000 10000 15000 20000 25000 30000
T/K
Figure 11. Temperature dependence of the ionization degree Ni/(N0 + Ni) of carbon C, carbon singly
ionized C+ and carbon doubly ionized C2+ at a constant pressure of 4 Pa.
4.5. Effect of Laser Irradiance
Laser-sample and laser-plasma interactions are strongly dependent on the laser beam
irradiance on the target. To see the effect laser irradiance the measurements were also carried
out at different laser fluences. Optical emission spectra of the carbon plasma plume in
medium-vacuum (~4 Pa) as a function of the laser intensity are shown in Figs. 12(a) and
12(b). These spectra were recorded at a constant distance of 1.5 cm from the target surface
along the plasma expansion direction. An increase of atomic and molecular emission intensity
with increasing the laser fluence was observed. Figure 13(a) shows the emission intensity
change of C(247.856 nm), C+(251.206 nm), C2+(269.775 nm), C3+(252.998 nm), C4+(227.792
nm), OH 0-0 band head (306.35 nm), and NH 1-0 band head (336.00 nm) as a function of the
carbon dioxide laser fluence. The C3+, C+ and C emission intensity increases drastically with
the laser fluence. Beyond ~100 J cm-2, a sharp increase of atomic (especially for C3+, C+ and
C) and molecular line intensities was observed. The C2+, C4+, OH 0-0 band head, and NH 1-0
band head emission intensity increases lightly with the laser fluence. Figure 13(b) shows the
emission intensity change of C+(392.07 nm), C2+(418.69 nm), NH 1-0 band head of the A-X
system, CN 1-0, 2-1, 0-0, 1-1 band heads of the B-X violet system, C2 1-0 band head of the
C1Πg-A1Πu Deslandres-d’Azambuja system, CH 0-0 band head of the A-X system, C2 1-0
band head of the d-a Swan system, and Hβ line as a function of the laser fluence. An increase
of atomic and molecular emission intensity with increasing the laser fluence was observed.
Also the background increases with the laser power. At higher laser fluences (154-342 J cm-
2
), the spectral lines and molecular bands are considerably more broadened than at lower
fluences as a result of the high pressure associated with the plasma. It is assumed that at
higher laser fluence the PLA plasma is more energetic and more ionized so that the
surrounding air can better confine the plasma; the plasma also cools down more rapidly due
to the confinement.
C3+
C2+
C+
-2
C+
5.36 GW cm -2
C+
0-0
5.36 GWcm
C
C3+
C2+
(a)
C+
C2+
C+
4.50
0-0
4.50
2º Diffraction order C2+
C+
3.46
C2+
3.46
2.41 2.41
1.00 1.00
4-3
C2: d3Πg-a3Πu
CH: A2Δ-X2Πr
-2
0.65 0.65 GWcm
v'=0-v"=0
C+
Δv=+1
0.54
3-3
C2+
-2
(b)
C+
0.48 GW cm
3-2
2+
2+
CN: B2Σ+-X2Σ+
C2+ C
Δv=0
2-1
10-9 1-0
C2+
12-11
2000 2200 2400 2600 2800 3000 3200 3400 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 4600 4700 4800
Air Wavelength / Å Air Wavelength/ Å
Figure 12. Low-resolution PLA of carbon emission spectrum observed in the (a) 2000-3640 Å and (b)
3660-4800 Å regions, at an air pressure of 4 Pa, excited by the 9P(28) line at 1039.36 cm-1 of the CO2
laser, as a function of the laser power density.
(a)220000 C (247.856 nm)

(b) 120000
NH1-0A-X
200000 +
C (251.206nm)
100000
CN1-0B-X
2+
180000 C (269.775nm) CN2-1B-X
3+ C21-0C-A
160000 C (252.998nm)
80000 CN0-0B-X
Intensity / a.u.
4+
C (227.792nm)
140000 CN1-1B-X
Intensity / a.u.
OH0-0band(306.35nm) +
C (392.07nm)
120000 NH1-0band(336.00nm) 60000 2+
C (418.69nm)
100000 CH0-0A-X
C21-0d-a(Swan)
80000 40000
Hβ
60000
40000 20000
20 30 40 50 60
20000
0
0
50 100 150 200 250 300 350 50 100 150 200 250 300 350
-2
Laser Fluence/ J×cm
-2
Laser Fluence/ J×cm
Figure 13 (a)-(b). Emission intensity change of: (a) C(247.856 nm), C+(251.206 nm), C2+(269.775 nm),
C3+(252.998 nm), C4+(227.792 nm), OH 0-0 band head (306.35 nm), and NH 1-0 band head (336.00
nm); (b) C+(392.07 nm), C2+(418.69 nm), NH 1-0 band head of the A3Π-X3Σ- system, CN 1-0, 2-1, 0-0,
1-1 band heads of the B2Σ+-X2Σ+ violet system, C2 1-0 band head of the C1Πg-A1Πu, CH 0-0 band head
of the A2Δ-X2Π, C2 1-0 band head of the Swan system d3Πg-a3Πu, and Hβ line as a function of the
carbon dioxide laser fluence.
4.6. Effect of Ambient Pressure on the Plasma
The emission characteristics of the laser-induced plasma are influenced to a large extent
by the nature and composition of the surrounding atmosphere. The pressure of the air ambient
atmosphere is one of the controlling parameters of the plasma characteristics, as well as the
factors related to the laser energy absorption. An interesting observation was the effect of the
air pressure, studied in the range 4.6 to 63500 Pa. Figs. 14(a)-(b) show typical OES from a
carbon plasma plume at different air pressures. These plasma plumes were generated by the
CO2 laser intensity of 1.00 GW cm-2. In general, the spectra of the PLA plume at low
pressures (P<1500 Pa) are dominated by emission of electronic relaxation of excited atomic
C, N, H, O, ionic fragments C+, C2+ C3+ and C4+, and molecular features of C2(E–A), C2(D–
X), C2(d–a), C2(D–X), C2(e–a), C2(C–A), C2(A-X), CN(D–A), CN(B–X), CN(A–X), OH(A–X),
NH(A–X), CH(C–X), CH(B–X) and CH(A–X). The spectra of the PLA plume at high pressures
(P>10000 Pa) are dominated by emission of electronic relaxation of excited atomic N, O, H,
ionic fragments N+ and O+, and molecular features of CN(B–X) and CN(A–X). The intensities
of the C2 1-0 band head of the C1Πg-A1Πu (3607 Å), C2 1-0 band head of the D1Σ+u–X1Σ+g
(4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+-X2Σ+ violet system, CH 0-0 band head of
the A2Δ-X2Π (4307 Å), C+(3919 Å), C+(4267 Å), C2+(4593 Å), C3+(4657 Å), and Hβ spectral
lines increase with increasing pressure, reach a maximum at about 200 Pa, and then decrease
with higher pressures. Similar results were reported in the literature [52, 53]. Figure 15 shows
the evolution of the emission intensity of C2 1-0 band head of the C1Πg–A1Πu (3607 Å), C2 1-
0 band head of the D1Σ+u–X1Σ+g (4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+–X2Σ+
violet system, CH 0-0 band head of the A2Δ–X2Π (4307 Å), N+(3437 Å), O+(3410 Å),
C+(3919 Å), C+(4267 Å), C2+(4593 Å), C3+(4657 Å), and Hβ atomic lines as a function of air
pressure. From figure 15, the intensity of the C2+(4593 Å), C+(4267 Å), and C3+(4657 Å)
spectral lines is found to be more sensitive to the pressure that the CH 0-0 band head of the
A–X (4307 Å), C+(3919 Å), and Hβ atomic lines. The lines N+(3437 Å) and O+(3410 Å)
produced in the breakdown of the air, are not observed in the PLA of carbon at low-air
pressures. The intensity of CN (Δv=0 sequence) increases with increasing air pressure, reach
a maximum at about 200 Pa, and then stays constant as the pressure is increased further.
Beyond 200 Pa (see Figure 15), a decrease in the time-integrated emission intensities of C+,
C2+, C3+, CN, C2, CH, NH and H was found. However, an increase in the emission intensities
of the N+ and O+ lines was observed. We suggest that these effects are related to shielding by
the air plasma, where a part of the laser energy is absorbed by the air plasma during its
expansion. This result in a reduction of the atomic and ionic emission intensity of species
formed from the carbon target. At low pressures (P<200 Pa), the C, C+, C2+, C3+, and CN, C2,
CH, NH emissions are produced nearer to the carbon target than the N+ and O+ emissions
produced nearer to the air plasma position. In general, the air ambient gas will confine the
plasma near the target (produced mainly by C, C+, C2+, C3+, and CN, C2, CH) and prevent the
electrons and species produced near the target escaping quickly from the laser focal volume
(observation region). Therefore, the emission intensity increases with increasing pressure.
However, at higher pressures (more than 200 Pa in our case), the ambient gas will hinder the
plasma from penetrating the atmosphere and predictably cause a higher plasma temperature.
The emission intensity of H, C, C+, C2+, C3+, and CN, C2, CH decreases because of the fact
that the laser energy is absorbed by air, producing air breakdown and increasing the N, O, N+,
and O+ emission intensity, in agreement with our observation in Figure 15. At lower air
pressures, the absence of the shielding air plasma results in a strong increase in the intensity
of the C, C+, C2+, C3+ emission from the carbon target plasma. At such lower air pressures the
relative contribution of the N+, O and O+ emission diminishes, and the emission from carbon
surface component becomes dominant.
3+
C C 2º order
N
+ + +
NN O
+ +
N O N O
+ N N
+ + + + (b) 1150000000 +
C C+ Pair=4.6Pa
C C+ 3+
(a) 150000 Pair=55000Pa
N
+ +
N O+ + 5000 Hβ 2+
C
100000 +
O O
+
+ O O+
50000 O
+ O 0
120000 Δv=0
0
C2: dΠg-aΠu Δv=-1
0-0
3 3
Δv=0
1-1
120000 80000 Pair=39.9Pa

Pair=10100Pa
2 + 2 +
CN: BΣ -XΣ
2-3
2-2
3-4
1-2
0-1
80000 Δv=+1 40000

Δv=-1
40000 0
0 180000
120000
140000 Pair=140Pa 2 + 2 +
CN: BΣ -XΣ
Pair=664Pa
60000
70000
0
0 +
140000 + + N P =9040Pa
C2+ C2+ 2º order
100000 3
C2: dΠg-aΠu
3
N N air
Pair=35.5Pa Δv=+1 70000 N
+ +
N
50000
N
0 0
C3+
20000 Pair=65300Pa
C+
+
C+
C+
Pair=4.6Pa 100000 O
O O
2+
+
O
C2+
10000
C
50000
0
O O
0
3400 3600 3800 4000 4200 4400 4600 4800 5000 5200 5400 5600 5800 6000 6200
Air Wavelength/ Å AirWavelength/ Å
Figure 14 (a)-(b). Low-resolution PLA of carbon OES observed at various air pressures in: (a) 3400-
4880 Å region and (b) 4800-6300 Å region.
C2 C-A 1-0 3607 Å

C2 d-a 1-0 4737 Å
CN B-X 1-0 3596 Å
CN B-X 0-0 3882 Å
+
N 3437 Å
+
O 3410 Å
+
C 3919 Å
+
C 4267 Å
3+
C 4657 Å
2+
C 4593 Å
Hβ 4861 Å
CH A-X 0-0 4307 Å
0.1 1 10 100
Air Pressure / mbar
Figure 15. Emission intensity change of C2 1-0 band head of the C1Πg-A1Πu (3607 Å), C2 1-0 band head
of the d3Πg-a3Πu (4737 Å), CN 1-0 and 0-0 band heads of the B2Σ+-X2Σ+ violet system, CH 0-0 band
head of the A2Δ-X2Π (4307 Å), N+ (3437 Å), O+ (3410 Å), C+(3919 Å), C+(4267 Å), C2+(4593 Å),
C3+(4657 Å), and Hβ line as a function of the air pressure around the carbon target.
0-0
3 -
NH: A Πi-X Σ
+
C
z=5 cm 2 +
C N: B Σ -X Σ
2 + 1-1
3
0-0
Δ v=+3
z=8 cm
6-3
5-2
z=14 cm 2
C N : B Σ -X Σ
+ 2 +
OH: A Σ -X Πi Δv=0
Δ v=+2
9-7
7-4
11-10
10-8
z=18 cm
6-4
5-3
2
2 +
+
C
2800 2900 3000 3100 3200 3300 3400

Air W avelength / Å
Figure 16. Optical emission spectra of the graphite ablation plume at: 5, 8, 14, and 18 cm from the
target.
4.7. Spatial Characterization
In this section we present experimental results on the laser ablation of a graphite target at
Pair=4 Pa by using a high-power IR CO2 pulsed laser (λ=9.621 µm and laser fluence of 342 J
cm-2) and distances from 0.2 up to 20 cm, from the target along the plasma expansion
direction. We discuss the dynamics of the plume expansion and formation of different atomic
(C, N, H and O), ionic (C+, C2+, C3+ and C4+) and molecular (C2, OH, CN, CH and NH)
species. Although OES gives only partial information about the plasma particles, this
diagnostic technique helped us to draw a picture of the plasma in terms of the emitting
chemical species, to evaluate their possible mechanisms of excitation and formation and to
study the role of gas-phase reactions in the plasma expansion process, allowing a discussion
of the probable OH, NH, CH, CN, H, O and N formation by gas phase reactions during the
propagation of the plasma plume.
The plasma emission was recorded at several distances along the plasma expansion
direction (Z axis of Figure 2) at a constant distance y=10 cm with respect to the focusing lens
onto the entrance slit of the monochomator and parallel to the target surface. Figures 16-18
show time-integrated OES following nanosecond pulsed laser ablation of graphite monitored
at several distances from the target in different spectral regions. One can see from these
figures that the emission along the plasma Z-axis is, in the conditions of sensitivity of our
detection system, about 20 cm. Also we can observe that the intensity of the C2+, C3+ and C4+
ionic emission lines decays rapidly at distances higher to 1.5 cm. On the other hand it can be
see that at high distances from the graphite target (z=18 cm) atomic lines from C+, C, H, O,
and N are still observed. The intensities of the molecular bands for different species (OH, NH,
CH, CN and C2) generally decay with distance, being observed up to 18 cm.
C+
C2+
C2: d Πg-a Πu
C+
C2+
z=0.2cm (b)
4-3 3 C 3
3+
Δv=+1
0-0
2º Diffraction order C2+

1-0 C2: C1Πg-A1Πu
z=5cm z=0.2cm
C+
CH: A2Δ-X2Πr
CN: B2Σ+-X2Σ+
z=8cm z=5cm
C+
Δv=0
z=14cm
C2+
z=8cm
0-0
C2+
z=18cm
CH: B2Σ--X2Πr
z=14cm
CN: B2Σ+-X2Σ+
2-1 3-2
(a)
C2+
Δv=+1
C+
z=18cm
0-0
C2+
1-0
10-9 1-0
Hβ
12-10
3500 3600 3700 3800 3900 4000 4100 4200

Air Wavelength/ Å 4200 4300 4400 4500 4600 4700 4800 4900
AirWavelength/ Å
Figure 17. Optical emission spectra of the graphite ablation plume monitored at: 0.2, 5, 8, 14, and 18
cm from the target: (a) 3500-4200 Å; (b) 4200-4950 Å regions.
z=18 cm O
(a) 200 (b) 400 z=18 cm CC
100 N C
200 C O
0
600
900
400 z=14 cm
z=14 cm
Emission Intensity / a. u.
600
200
300
Emission Intensity / a. u.
0
2000 3 3
1500 C2: d Πg-a Πu z=8 cm 60000
1000
z=8 cm
Δv=-2 30000
500 Δv=0
0 Δv=-1 0
10000
60000 C
z=5 cm 40000 z=5 cm
5000
20000
0 0
1-1
0-0
60000
z=0.2 cm 60000 O C
2-2
z=0.2 cm
2-3
40000
3-4
C+
1-2
C2+
+0-1 40000 N
C3+
4-5
Hβ
v'-v"
20000 20000
3-5
2-4
4-6
0-2
1-3
N CCC
C
O
0 0
4800 5000 5200 5400 5600 5800 6000 6200 8600 8800 9000 9200 9400 9600
Figure 18 (a)-(b). Optical emission spectra of the graphite ablation plume monitored at: 0.2, 5, 8, 14,
and 18 cm from the target in the spectral region: (a) 4800-6250 Å and (b) 8430-9700 Å.
Figure 19(a) shows the emission intensity change of C+(3920.7 Å), C2+(4186.9 Å),
Hβ(4861.36 Å), NH 1-0 band head (3365 Å), CN B-X 1-0, 0-0, and 1-1 band heads (3591,
3887, 3875 Å, respectively), C2 C-A 1-0 band head (3615 Å), CH A-X 0-0 band head (4317
Å), and C2 d-a 1-0 band head (4743 Å), as a function of the distance. The C2+ emissions are
only observable close to the carbon target (z<4 cm). The intensity of CN, NH, CH bands, and
Hβ line increase lightly with increasing the distance, reach a maximum at about 5 cm, and
then stay constant as the distance is further increased. Beyond 8 cm, a decrease in the time-
integrated emission intensities of these species was found. The intensity of the C2 band falls
continuously with the distance to the target. The continuous decrease of the C+, C2+, and C2
emission intensities with the distance indicates that these species are mainly formed from the
carbon surface. This decrease is more drastic for the C2+ emission intensity probably due to
the fact that the recombination with free electron is faster than for C+ species. The slight rise
of the NH, CN, CH and Hβ emission intensities as a function of the distance up to zmax ≈ 5 cm
indicates that these electronically excited species are mainly formed by gas phase reactions
during the propagation of the plasma plume through the air gas. The NH, CN, CH and Hβ
emission intensities diminish for z > 8 cm due to the decrease in density during the expansion
of the plasma plume. In figure 19(b), the emission intensity change of O(8446.359 Å),
O(9262.776 Å), N(8680.28 Å), C(9061.43 Å), C(9078.28 Å), C(9088.51 Å), C(9094.83 Å),
C(9111.80 Å), and CN A-X 1-0 band head at 9192 Å as a function of the distance from the
target are shown. The intensity of O(8446.359 Å), O(9262.776 Å), N(8680.28 Å) and CN A-X
1-0 band head increases lightly with increasing the distance from the target expansion
direction, reach a maximum at about 5 cm, and then stays constant as the distance is further
increased. Beyond 8 cm a decrease in the time-integrated emission intensities of these species
was found. However, all the carbon emission lines decrease with increasing the distance from
the target. As previously discussed in figure 19(a) the behaviour of carbon species are mainly
formed from the surface target but, molecular species such as NH, CN and CH are produced
in gas phase.
(a)
NHA-X1-0
100000 100000
CNB-X1-0
C2 C-A1-0 OI 8446.359Å
CNB-X0-0 OI 9262.776Å
NI 8680.280Å
Relative Intensity / a.u.

Relative Intensity / a.u.
CNB-X1-1
+ CI 9061.43Å
C 3920.7Å CI 9078.28Å
10000 2+ 10000
C 4186.9Å CI 9088.51Å
CHA-X0-0 CI 9094.83Å
C2 d-a 1-0 CI 9111.80Å
Hβ CNA-X1-0
1000 1000
(b)
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Distancefromthetarget, z/ cm Distancefromthetarget, z/ cm
Figure 19 (a)-(b). Emission intensity change of: (a) C+, C2+, Hβ, NH, CN, C2 (C-A 1-0 band head), CH,
and C2 (d-a 1-0 band head) and (b) O (two lines), N, C(five lines) and CN (A-X 1-0 band head), as a
function of the z distance.
I order to further identify properties of the ablation plasma plumes originated from
graphite targets, we have estimated the vibrational temperatures (Eq. 2.24) of C2 molecule (d-
a Swan Δv=-1 band sequence) as function of z-distance. The estimated vibrational
temperatures were Tvib=23000±1000, 16200±900, 10800±600, 7700±500 K at 0.2, 1.5, 5 and
9 cm from the target along the plasma expansion direction, respectively compatible with a
cooling stage.
4.8. Temporal Evolution of the Plasma
In this section time-resolved OES analysis for the plasma plume, produced by high-
power tunable IR CO2 pulsed laser ablation of graphite, at λ=10.591 µm and a laser fluence of
402 J cm-2 is presented. We focus our attention on the temporal evolution of different
atomic/ionic and molecular species over a broad spectral range from 190 to 1000 nm.
Excitation temperature, electron density and vibrational temperature in the laser-induced
plasma were estimated from the analysis of spectral data at various times from the laser pulse
incidence.
In time-resolved measurements, the delay td and width tw times were varied. It was
verified that the plasma was reproducible over more than 7 ablation events by recording the
same spectrum several times. The temporal history of laser-induced breakdown carbon
plasma is illustrated schematically in Figure 20. The time when beginning of the CO2 laser
pulse is triggered is considered as the origin of the time scale (t=0). Inserts illustrate some
emission spectra recorded at different delay and width times. The temporal shape of the CO2
laser pulse is also shown. Because the LIB plasma is a pulsed source the resulting spectrum
evolves rapidly in time.
The LIBS spectra of carbon were measured at different delay and width times. As an
example figure 21(a) show OES of the graphite ablation plume at low-resolution in 2000-
2800 Å region monitored at several delay times for a fixed gate width time of 0.1 µs. At early
times (td<2 μs) the predominant emitting species are the C2+ 2p2 1D2 → 2s2p2 1P1 atomic line
at 2296.87 Å and two lines of C3+ at 2524.41 and 2529.98 Å. The three molecular band
systems observed in this spectral region are the C2(E1Σg+-A1Пu; Freymark), C2(D1Σu+- X1Σg+;
Mulliken) and C2(e3Пg- a3Пu; Fox-Herzberg). These bands stay approximately constant as the
delay time is increased up to ~5 μs, and decrease for higher delay times.
Figure 21(b) shows the LIB emission spectrum of graphite plasma plume recorded 1, 1.2,
4 and 11 μs after the CO2 laser irradiation in the spectral region 2850-3600 Å. The main
features in this region are the emission of C+ ionic species and several molecular emission
bands from C2, OH and CN. The C+ emission intensities fall considerably as the delay
changes from 1 to 11 μs. On the other hand, the molecular bands of C2, OH and CN emission
intensities stay approximately constant as the delay time is increased up to ~5 μs, and
decrease for higher delay times. From our results at low-resolution we can appreciated that
the intensity of the C+, C2+ and C3+ ionic emission lines decays rapidly at delay times higher
to 2 μs. On the other hand we can see that at high delay time after plasma ignition, atomic
lines from C, H, O, and molecular bands of C2, CN and OH are still observed.
C+
C2+
td = 1 μs
C+
tw = 0.1 μs
2 2
CH : A Δ - X Πr 0-0
td = 4 μs
CH : B Σ - X Πr 0-0
tw = 0.1 μs
C2+
CN : B - X
2 −
2 + 2 +
CN : B Σ - X Σ Δv=-1
Δv=0
2 2
CH : B Σ - X Πr 0-0
CH : A Δ - X Πr
td = 21 μs
0-0
3600 3700 3800 3900 4000 4100 4200 4300 4400 tw = 0.1 μs
2
0.10
CN : B - X
Laser Power / a. u.
Wavelength / Å
2 −
Δ v=-1
2 + 2 +
CN : B Σ - X Σ
3600 3700 3800 3900 4000 4100 4200 4300 4400 Δv=0
Wavelength / Å
3600 3700 3800 3900 4000 4100 4200 4300 4400

Wavelength / Å
τFWHM= 64 ns t (gate width time)

w
td = 7 μs
0.05
2 2
CH : A Δ - X Πr 0-0
tw = 0.1 μs
CH : B 2Σ− - X 2Πr 0-0
td (gate delay time) 2 + 2 +

CN : B Σ - X Σ
CN : B- X
Δv=-1
Δv=0
CO2 Laser pulse

3600 3700 3800 3900 4000 4100 4200 4300 4400
Wavelength / Å
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Time / μs
Figure 20. A schematic overview of the temporal history of laser-induced breakdown carbon plasma.
Here td is the gate delay time and tw is the gate width time during which the plasma emission is
monitored. Inserts illustrate some spectra observed at different delay and width times. The temporal
shape of the CO2 laser pulse (recorded with the aid of the photon-drag detector) is also shown.
C2+
(b) td=1μs 2+ 2+
CN:BΣ-XΣ
(a)
C+
2 + 2 + Δv=+1
td=1μs CN:BΣ-XΣ
C3+
3 3 2+ 2
C2:eΠg-aΠu OH:AΣ-XΠ Δv=+2
tw=0.1μs
Δv=0
C2+C+
1 + 1 +
1 + 1
C2: DΣu-XΣg 3
C2: eΠg-aΠu
3
C2: EΣg-AΠu
C+
2+
C2+ C+C
C2+
C4+
C++
C
C+
td=1.2μs
td=2μs
td=4μs
td=4μs
td=6μs
td=11μs td=11μs
2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 3500 3600
Wavelength/ Å
Wavelength/ Å
Figure 21. Optical emission spectra of the graphite ablation plume monitored at different delay times
for a fixed gate width time of 0.1 μs in two spectral regions.
In order to get more insight into laser ablation of graphite and to understand the laser-
induced breakdown dynamics, we have scanned in the UV-Visible spectral region with higher
resolution. As an example some results for the spectral region 4640-4750 Å are shown in
figures 22-24. The spectral range was chosen in order to detect both double and triple ionized
carbon species and C2 diatomic molecule. In figure 22, the data acquisition was performed by
averaging the signal over: (a) 20 successive laser shots (td=0 and tw>>30 μs) and (b) 7
successive laser shots (td=4 μs and tw=0.02 μs). The emissions of ionized C2+(1s22s3p 3P02,1,0
→ 1s22s3s 3S1) around 4650 Å, and C3+ around 4658.3 Å are considerably higher in the
spectrum of Figure 22a, while the C2(d-a) Swan Δv=+1 sequence emission is similar. Figures
23a-b and 24 show the typical temporal sequence of laser-induced carbon plasma. At early
times (td≤0.02 μs) emission from C2+ and C3+ is easily detected between 4645-4670 Å (see
inset within figure 23-a). As seen in figure 23-b during the initial stages after laser pulse
(td≤0.04 μs), C2+ emissions dominate the spectrum. As time evolves (0.04 μs≤td≤1.5 μs), C3+
emission dominate the spectrum. As the delay is increased up to 2.5 μs (1.5 μs≤td≤2.5 μs)
again C2+ emission dominates the spectrum. These ionic lines decrease quickly for higher
delay times, being detected up to ∼ 3 μs. Some oxygen and nitrogen ionic lines were also
observed in the spectra at the gate delay from 0.02 μs to 1 μs and its emission intensities
remain approximately constant in this time interval (see Figure 23a). They vanished after the
delay of ∼1.5 μs. It shows that the air is ionized by the CO2 laser pulse and by the collisions
with the laser induced plasma. During the time period up to ∼ 0.5 μs, no apparent C2
emissions were observed. As can be seen from Figure 24, the C2(d-a; Δv=1 band sequence)
emissions were clearly observed from ∼2 μs. The C2 emission intensities increase lightly with
increasing td, reach a maximum at ~5 μs, and then decrease as the time is further increased.
2+ 3+
C C 2
2+ 2 3 0
J' 1 C : 1s 2s3p PJ'
6-5
5-4
3 3
C2: d Πg-a Πu
0
3-2
td = 4 μs
2-1
4-3
2+ 2 3
J" 1 C : 1s 2s3s SJ" tw = 0.02 μs
v'-v"
1-0
7-6
6-5
5-4
3 3
C2: d Πg-a Πu
7-6 4-3
3-2
2+
2-1
C
8-7
v'-v"
1-0
8-7
(b)
(a)
4650 4675 4700 4725 4750 4650 4675 4700 4725 4750
Wavelength / Å Wavelength / Å
Figure 22 (a)-(b). Measured high-resolution pulsed laser ablation of graphite emission spectra observed
in the region 4645-4750 Å region. The data acquisition was performed by averaging the signal over: (a)
20 successive laser shots with td=0 and tw>>30 μs; (b) 7 successive laser shots with td=4 μs and tw=0.02
μs. The assignments of some ionic lines of C2+ and C3+ and molecular bands of C2 are indicated. The
insert in (a) illustrates the rotational structure of one triplet of C2+ line.
2+
C 2+ td=40 ns; tw=20 ns C
3+
2000
C (b) td=500 ns tw=20 ns
7000
200
td= 30 ns
td=0 ; tw=20 ns 6000 td= 20 ns

1500 td=100 ns
3+
C
100 5000
4000 2+
C 2+
1000 0 C
+ 3000
O 4650 4675 4700
Wavelength / Å
2000
500
+ +
O O O
+
1000
0
0
4650 4660 4670 4680 4690 4700 4710 4720 4645 4650 4655 4660 4665 4670
Wavelength / Å Wavelength / Å
Figure 23(a)-(b). Time-resolved high-resolution emission spectra from laser-induced carbon plasma
observed in the region: (a) 4645-4720 Å region monitored at 40 ns delay time; (b) 4645-4670 Å region
monitored at 20, 30, 100, and 500 ns gate delay times for a fixed gate width time of 20 ns. The inset in
(a) displays the spectrum the first 20 ns after incidence of the laser pulse.
Space-and-time resolved OESs laser-induced measurements could be used to estimate

plasma expansion rate. To obtain additional time resolved information about the optical
emission of the plume, wavelength resolved spectra have been recorded at different delay
times at a distance of 9 mm. The temporal evolution of spectral atomic, ionic and molecular
line intensities at a constant distance from the target can be used to construct the time-of-
flight (TOF) profile. TOF studies of the emission provide fundamental information regarding
the time taken for a particular species to evolve after the laser-induced plasma has formed.
Specifically, this technique gives an indication of the velocity of the emitted species. A rough
estimation of the velocity for the different species in the plume can be inferred from the time
resolved spectra by plotting the intensities of selected emission lines versus the delay time,
and then calculating the velocity by dividing the distance from the target by the time where
the emission peaks. This method for determination of plasma velocity should be used with
care due to the superposition of both expansion and forward movements of the plasma plume.
2+
C
3 3
C2:d Πg-a Πu
tw=0.02 μs
5-4
6-5
td=10 μs
3-2
4-3
2-1
2+
C
v'-v"
1-0
8-7
30000
25000
20000
C
3+
td=5 μs
15000
td=3 μs
10000
2+
C
5000
td=2 μs
0
4640 4660 4680 4700 4720 4740
Wavelength / Å
Figure 24. Time-resolved high-resolution emission spectra from laser-induced carbon plasma observed
in the region 4645-4750 Å region monitored at 2, 3, 5, and 10 μs gate delays for a fixed gate width time
of 20 ns.
Figure 25 displays the TOF profile, for ablation experiments induced by CO2 laser pulses,
of several C, C+ and C3+ lines intensities in UV region and C2+ in the visible as a function of
delay time. However, the insert of the figure shows the time dependence of C2+ and C3+ line
intensities in the visible region for ablation induced by CO2 laser pulses in which the tail has
been eliminated by means of the suppression of the N2 in the gas mixture of the active laser
medium. All the data are taken from high-resolution spectra and in the figures the temporal
profiles of both kinds of laser pulses are also plotted. In both cases emissions from C3+ are
stronger than emissions coming from the other species. All the ionic lines follow the time
profile of lasers pulses lasting until four or three microseconds depending on the kind of the
laser pulse. These behaviours may be related to the laser absorption processes on the target
surface. Thus for “non tailed pulses” the line intensities start to growth at 400 ns while for
“tailed pulses” start at 70 ns. Since the energy pumping (rise time) of each kind of pulse is
different, the species reach the maximum intensity at different times: ∼1 μs for non tailed
pulses and 700 ns for tailed ones, indicating that the graphite target needs some energy
threshold to eject the different species. The higher intensity in the 0-400 ns time interval for
the C2+ may be due to the higher sensibility of our ICCD camera in the visible region than in
the UV one. The different behaviour of atomic C can be also observed in figure 25. Atomic C
have a higher rise time and lasting more ( > 15 μs) than ionic species, possibly due to the
continuous recombination of ions with electrons to give excited carbon. From these results
has not been observed excitation dependence on the pulse tail however the energy of the pulse
intensity seems to be the pulse parameter that influence on the graphite ablation process. The
peak velocities estimated for C3+, C2+ and C+ species, from figure 25, are about 7x103 m/s.
CO2 laser pulse

C : 2478.56 Å
+
C : 2509.12 Å CO2 laser pulse
2+
C : 4647.42 Å without tail
3+ C2+:4647.42 Å
C : 2529.98 Å C3+: 4658.3 Å
Intensity / a. u.
Intensity / a.u.
0 1 2 3 4 5
Delay time / μs
0 1 2 3 4 16 18
Delay time / μs
Figure 25. Emission intensity change of C(2478.56 Å), C+(2509.12 Å), C2+(4647.42 Å) and
C3+(2529.98 Å) lines as a function of delay time (fixed gate width time of 20 ns) for a CO2 pulse laser
with a tail of about 3 μs. The insert shows the emission intensity change of C2+(4647.42 Å) and
C3+(4658.3 Å) lines as a function of delay time (fixed gate width time of 20 ns) for a CO2 pulse laser
without tail.
In this section the plasma temperature was determined form the emission line intensities
of several C+ lines observed in the laser-induced plasma of carbon target for a delay time of 1
μs and 0.02 μs gate width. The obtained excitation temperature was 26000 ± 3000 K. The
carbon ionic multiplet line at ∼3920 Å was identified as candidate for electron-density
measurements. Figure 26-a shows, the 3920 Å carbon ionic line with sufficient resolution to
measure the full width at half-maximum at 8 different time delays. All the data points were
fitted with Lorentzian function to determine the Stark line width. By substituting these values
in Eqn. (2.21) and the corresponding value of electron impact parameter W (0.465 Å from
Griem [27] at plasma temperature of 26000 K), we obtain the electron density. Figure 26-b
gives the time evolution of electron density by setting the gate width of the intensifier at 0.02
μs. The initial electron density at 0.02 μs is approximately 3 1016 cm-3. Afterwards, the
density increases over the period of 0.1 μs and reaches a maximum at 0.1 μs (time period of
the peak CO2 laser pulse), and then decrease as the time is further increased. At shorter delay
times (<0.1 μs), the line to continuum ratio is small and the density measurement is sensitive
to errors in setting the true continuum level. For times >0.1 μs, the line to continuum ratio is
within reasonable limits and the values of electron density shown in figure 26-b should be
reliable. Initially the laser-induced plasma expands isothermically within the time of the
duration of the laser pulse. After termination of the peak laser pulse (∼0.1 μs) the plasma
expands adiabatically. During this expansion the thermal energy is converted into kinetic
energy and the plasma cools down rapidly. After 4 μs, the electron density is about 1.5 1016
cm-3. For a long time >4 μs, subsequent decreased C+ emission intensities result in poor
signal-to-noise ratios, and there exits a limitation in the spectral resolution. The decrease of ne
is mainly due to recombination between electrons and ions in the plasma. These processes
correspond to the so-called radiative recombination and three-body recombination processes
in which a third body may be either a heavy particle or an electron.
2 2 0
C :2s 4s S1/2→2s 3p P3/2
+ 2 2
td=4 μs 16
4x10
25000
td=3 μs
2 2 2 2 0
C :2s 4s S1/2→2s 3p P1/2
+ td=1 μs
td=0.5 μs
Electron density / cm-3

20000
td=0.1 μs
tw=20 ns 16
td=0.05 μs 3x10
15000
td=0.02 μs
td=0
10000
16
2x10
5000
0
16
3917 3918 3919 3920 3921 3922 1x10
-3 -2 -1 0
10 10 10 10
Wavelength / Å
Delay time / μs
+
Figure 26 (a). Stark-broadened profiles of the C line at 3920 Å at different delay times for a
fixed gate width time of 0.02 μs. (b) Temporal evolution of electron density at different delay
times from plasma ignition.
In order to further identify properties of the ablation plasma plumes originated from
graphite targets, we have estimated the vibrational temperatures of C2 molecule as function of
delay time. The emission intensities of the C2 d-a Swan Δv=+1 band sequence were used to
estimate these vibrational temperatures Tvib. The estimated vibrational temperatures were
Tvib=8000±500, 8300±600, 7500±600, 4500±900 K at 3, 5, 10 and 15 μs after plasma
ignition, respectively compatible with a cooling stage.
Optical emission accompanying TEA-CO2 nanosecond laser ablation of carbon is very
long lived (∼40 μs) relative to the average radiative lifetimes of the excited levels that give
rise to the observed emission lines. At distances close to the target surface (<9 mm), all of the
emission lines of C, C+, C2+ and C3+ expected in the 2000-10000 Å wavelength range are
observed, illustrating that the excited species giving rise to the optical emission are produced
by non-specific mechanism during the TEA CO2 laser ablation process. However, a direct
excitation-de-excitation mechanism cannot explain the observed emission spectra. EII would
explain the emission intensity variation with the time for C, C+, C2+, C3+ and C2 species. On
the other hand, the formation of the excited molecular species would happen in gas phase by
collisions between atomic or ionic species present in the plume and the residual gas at times
far away from the plasma ignition. The emission process at this plasma stage is divided into
two different process associated, respectively with the shock formation and the plasma
cooling. During the former, the atoms, molecules and ions gushing out from the carbon target
are adiabatically compressed against the surrounding gas. During the latter stage the
temperature of the plasma and consequently the emission intensities of atomic lines and
molecular bands decrease gradually.
The evolution of the TEA-CO2 laser-induced carbon plasma can be divided into several
transient phases. The initial plasma (td<2 μs) is characterized by high electron and ion
densities (1016-1019 cm-3), and temperatures around 2.2 eV. The emission spectrum from this
early stage is characterized by emission lines from C3+, C2+ and C+ ions. Owing to the high
electron density, the emission lines are broadened by Stark effect. The ionic emission lines
(C3+, C2+, C+) decay rapidly being observed up to ∼3 μs. Emission lines from C atoms and
molecular species (C2, CN, CH, OH) in excited electronic states can be found after about 1 μs
time delay. As the plasma expands and cools, the electrons and ions recombine. After the
initial plasma (td>3 μs), the molecular emissions increase slowly up to ∼5 μs and after that
decay slowly up to ∼40 μs.
5. CONCLUSION
This article reviews some fundamentals of LIBS and some experimental studies
developed in our laboratory on the ablation of graphite using a high-power IR CO2 pulsed
laser. In this experimental study we used several laser wavelengths (λ=9.621 and 10.591 µm)
and laser intensity ranging from 0.22 to 6.31 GW cm-2. Ablation was produced typically at
medium-vacuum conditions (∼ 4 Pa). Emissions from the resulting ablation plumes, and from
the collisions with the ablated material and the background gas molecules (N2, O2, H2O, etc),
have been investigated by wavelength-, space-, and time-resolved OES from UV-Vis-NIR.
Wavelength-dispersed spectra of the plume reveal C, C+, C2+, C3+, C4+, N, H, O, N+, O+ and
molecular features of C2, CN, OH, CH, N2, N2+ and NH emissions corresponding to different
electronic band systems. For the assignment of molecular bands a comparison with
conventional emission sources was made. Excitation, vibrational and rotational temperatures,
ionization degree and electron number density for some species were estimated by using
different spectroscopic methods. The characteristics of the spectral emissions from the
different species have been investigated as functions of the ambient pressure, laser irradiance,
the distance from the target and delay time after plasma ignition. Time-gated spectroscopic
studies have allowed estimation of TOF and propagation velocities for various emission
species. Possible production routes for secondary emitters such as C2, CN, OH, CH, N2, N2+
and NH are discussed.
ACKNOWLEDGMENTS
We gratefully acknowledge the support received in part by the DGICYT (Spain) Projects:
MEC: CTQ2007-60177/BQU and MEC: CTQ2008-05393/BQU for this research.
REFERENCES
[1] Kroto, HW; Heath, JR; O’Brien, SC; Curl, RF; Smalley, RE. Nature, 1985, 318,
162-163.
[2] Koinuma, H; Kim, MS; Asakwa, T; Yoshimoto, M. Fuller. Sci. Technol, 1996, 4,
599-612.
[3] Ikegami, T; Nakanishi, F; Uchiyama, M; Ebihara, K. Thin Solid Films, 2004, 457, 7-11.
[4] Ashfold, MNR; Claeyssens, F; Fuge, GM; Henley, SJ. Chem. Soc. Rev., 2004, 33,
23-31.
[5] Yueh, FY; Singh, JP; Zhang, H. In Encyclopedia of Analytical Chemistry; Meyers RA.
Ed. Laser-induced Breakdown Spectroscopy; Elemental Analysis; John Wiley & Sons
Ltd, Chichester, 2000, pp 2066-2087.
[6] Dwivedi, RK; Thareja, RK. Phys. Rev. B, 1995, 51, 7160-7167.
[7] Demyanenco, AV; Letokhov, VS; Puretskii, AA; Ryabov, EA. Quantum Electronics
1998, 28, 33-37.
[8] Vivien, C; Hermann, J; Perrone, A; Luches, A. J. Phys. D: Appl. Phys. 1998, 31,
1263-1272.
[9] Wee, S; Park, S. M. Opt. Comm., 1999, 165, 199-205.
[10] Yamagata, Y; Sharma, A; Narayan, J; Mayo, RM; Newman, JW. J. Appl. Phys., 2000,
88, 6861-6867.
[11] Harilal, SS. Appl. Surf. Science, 2001, 172, 103-109.
[12] Acquaviva, S; Giorgi, ML. J. Phys. B: At. Mol. Opt. Phys., 2006, 35, 795-806.
[13] Saito, K; Sakka, T; Ogata, H. J. Appl. Phys., 2003, 94, 5530-5536.
[14] Zelinger, Z; Novotny, M; Bulir, J; Lancok, J; Kubat, P; Jelinek, M. Contrib. Plasma
Phys., 2003, 43, 426-432.
[15] Saidane, K; Razafinimanana, M; Lange, H; Huczko, A; Baltas, M; Gleizes, A; Meunier,
JL. J. Phys. D: Appl. Phys., 2004, 37, 232-239.
[16] Park, HS; Nam, SH; Park, SM. J. Appl. Phys., 2005, 97, 113103-5.
[17] Fuge, GM; Ashfold, MNR; Henley, SJ., J. Appl. Phys., 2006, 99, 14039-12.
[18] Camacho, JJ, Poyato, JML, Díaz, L; Santos, M. J. Phys. B: At. Mol. Opt. Phys., 2008,
41,105201-13.
[19] Camacho, JJ; Santos, M; Diaz, L; Poyato, JML. Appl. Phys. A., 2009, 94, 373-380.
[20] Camacho, JJ; Diaz, L; Santos, M; Juan, LJ; Poyato, JML. J. Appl. Phys., 2009, 106,
33306-11.
[21] Cremers, DA; Radziemski, LJ. Handbook of Laser-lnduced Breakdown Spectroscopy;
Wiley: Chichester, England, 2006.
[22] Miziolek, AW; Palleschi, V; Schechter, I. (Eds.), Laser-lnduced Breakdown
Spectroscopy; Cambrige. 2006.
[23] Singh, JP; Thakur, SN. (Eds.) Laser-lnduced Breakdown Spectroscopy; Elsevier:
Oxford UK, 2007, Vol. 1, pp 1-427.
[24] Yong-Ill, L., Laser Induced Breakdown Spectrometry; Nova Science Publishers: New
York 2000.
[25] Chan, F. Introduction to plasma physics and controlled fusion; Plenum Press: New
York. 1984.
[26] Griem, HR. Principles of plasma spectroscopy; University Press: Cambridge. 1997.
[27] Griem, HR. Phys. Rev., 1962, 128, 515-523.

[28] MacDonald, AD. Microwave Breakdown in Gases Wiley; New York, 1966.
[29] Raizer, YP. Gas Discharge Physics; Springer: Berlin, Heidelberg. 1991.
[30] Kopiczynski, TL; Bogdan, M; Kalin, AW; Schotwau, HJ; Kneubuhl, FH. Appl. Phys.
B: Photophys. Laser Chem., 1992, 54, 526-530.
[31] Radziemski, LJ; Cremers, DA; Laser-induced plasma and applications; New York:
Dekker. 1989.
[32] Gurevich, A; Pitaevskii, L. Sov. Phys. JETP, 1962, 19, 870-871.
[33] Morgan, CG. Rep. Prog. Phys., 1975, 38, 621-685.
[34] Huddlestone, RH; Leonard, SL. Plasma diagnostic techniques; Academic Press: New
York. 1965.
[35] Hutchinson, IH. Principles of plasma diagnostic; University Press: Cambridge. 2002.
[36] NIST Atomic Spectra Database online at http://physics.nist.gov/PhysRefData/ASD/
index.html
[37] Breene, RG. The Shift and Shape of Spectral Lines; Pergamon: London. 1961.
[38] Bengtson, RD; Tannich, JD; Kepple, P. Phys. Rev. A, 1970, 1, 532-533.
[39] Griem, HR. Spectral line broadening by plasmas; Academic Press: New York.1974.
[40] Herzberg, G. Spectra of diatomic molecules; Van Nostrand: New York. 1950.
[41] Steinfeld, JI. An introduction to modern molecular spectroscopy; MIT Press: London.
1986
[42] Bernath, PF. Spectra of atoms and molecules; Oxford University Press: New York.
1995.
[43] Camacho, JJ; Pardo, A; Martin, E; Poyato, JML. J. Phys. B: At. Mol. Opt. Phys. 2006,
39, 2665-2679.
[44] Kovacs, I. Rotational Structure in the Spectra of Diatomic Molecules. Hilger: London.
1969.
[45] Cabalin, LM; Laserna, JJ; Spectrochim. Acta Part B, 1998, 53, 723-730.
[46] Demtröder, W. Laser Spectroscopy. Vol. 2. Experimental Techniques. Springer. Berlin
2008.
[47] Bogaerts, A; Chen, Z. Spectrochim. Acta Part B, 2005, 60,1280-1307.
[48] Zeldovich, YB; Raizer, YP. Physic of Shock waves and high temperature
hydrodynamics phenomena. Academic, New York 1966.
[49] Drogoff, LB; Margotb, J; Chakera, M; Sabsabi, M; Barthelemy, O; Johnstona, TW;
Lavillea, S; Vidala, F; Kaenela, VY. Spectrochim. Acta Part B, 2001, 56, 987-1002.
[50] Huber, KP; Herzberg, G. Molecular spectra and Molecular structure. IV. Constants of
diatomic molecules; Van Nostrand Reinhold: New York. 1979.
[51] Martin, WC; Zalubas, R. J. Phys. Chem. Ref. Data,1983, 12, 323-380.
[52] Kim, DE; Yoo, KJ; Park, HK; Oh, KJ; Kim, DW. Appl. Spectrosc., 1997, 51, 22-29.
[53] Lu, YF; Tao, ZB; Hong, MH. Jpn. J. Appl. Phys., 1999, 38, 2958-2963.
Chapter 3
INDUCTION TRANSFORMER COUPLED DISCHARGES:

INVESTIGATION AND APPLICATION
I.M. Ulanov* and M.V. Isupov

Kutateladze Institute of Thermophysics SB RAS, Novosibirsk, Russia
ABSTRACT
Researches in the field of low-temperature plasma provide development of devices,
which are widely used by the advanced and high-tech industries. Low-temperature
plasma is applied by such major industries as microelectronics, semiconductor industry,
solar cell production, plasma chemistry, metallurgy, lighting engineering, etc. Among the
known methods of production and use of low-temperature plasma (DC and AC arc
discharges, RF plasmatrons, microwave plasmatrons) the devices based on application of
induction transformer coupled toroidal discharges (TCTD) are the least studied and, thus,
rarely used. Simultaneously, these discharges can be used for development of
electrodeless generators of low-temperature plasma: transformer plasmatrons and new
induction sources of light.
The chapter deals with investigation of properties of TCTD, transformer plasmatrons
and induction light sources on the basis of TCTD. Investigation results on electrophysical
properties of TCTD aimed at development of transformer plasmatrons are presented in
the current paper. Dependences between the strengths of TCTD electric field, discharge
current and gas flow are obtained for different gases within the pressure range of 10÷105
Pa. The thermophysical characteristics of TCTD were determined: device efficiency,
energy balance of a discharge (heat losses to the discharge chamber wall, plasma jet
power). The stable TCTD of the atmospheric pressure in argon and in air was firstly
obtained and studied by the authors of this chapter. The process of plasmachemical
synthesis of NO in air plasma of TCTD was studied. The abnormally high percentage of
NO ~7 % was obtained without product quenching. The transformer plasmatrons of the
10÷200 kW power, operating under the pressures of 10÷105 Pa on argon, air and
argon+hydrogen, argon+oxygen mixtures, were developed on the basis of the studies
performed. The schemes and constructions of these plasmatrons are presented.
* Ulanov@itp.nsc.ru
114 I.M. Ulanov and M.V. Isupov
The electrophysical and spectral characteristics of TCTD were studied in vapors of

mercury and neon. The electrodeless sources of visible and UV radiation with the power
of 100 W ÷ 100 kW were developed on the basis of this research.
INTRODUCTION
The toroidal plasma–confinement devices (TOKAMAKs) are one of the most common
methods for generation of high–temperature plasma [1]. This method is based on gas heating
by a powerful current pulse (several mega amperes), induced by a current pulse in the
transformer primary winding. This method can be also used successfully for generation of
low-temperature plasma both in pulse and continuous modes. In this case we do not need the
system for magnetic confinement of plasma, and a gas discharge is stabilized by the walls of
the gas-discharge chamber.
The use of a closed ferromagnetic core connecting the inductor (the primary coil of
transformer) and toroidal plasma coil has several essential advantages over other types of RF
induction gas discharges: frequency of gas discharge generation is several orders lower, the
phase shift between the current in inductor and applied voltage is lower, and the level of
electromagnetic noises is reduced. Owing to this, the transformer coupled toroidal discharges
can be used successfully for development of efficient electrodeless generators of low-
temperature plasma and gas-discharge light sources.
The first experimental works related to development and construction of electrodeless
gas-discharge devices based on the principle of transformer coupled toroidal discharge were
carried out in the middle of 60s [2].
The gas-discharge ion laser operating on the principle of transformer coupled toroidal
discharge is described in [2]. To get a discharge in argon, the radiation yield in 7 ion lines was
50 mW, and the setup efficiency was about 0.1%. On the ground of these results the author
has made a conclusion about the promising character of gas-discharge lasers, operating by the
principle of transformer coupled toroidal discharge.
The most important of the early works, dealt with investigations of TCTD, are the works
of H.U. Eckert [3, 4]. The possibility of low-temperature plasma generation by means of
TCTD in the low-frequency radio range is studied theoretically and experimentally in [3]. In
particularly, possible development of the induction atmospheric-pressure plasmatron,
operating at current frequencies of 60 or 180 Hz, which does not require application of RF
power supplies, is considered there.
According to the estimates of H. U. Eckert [3], the transformer plasmatron of
atmospheric pressure, operating on argon at current frequency of 60 Hz, requires a core of
60 000 cm2 cross-section; and at this the discharge power will be 1.76 MW. With a rise of
current frequency of up to 180 Hz, the core cross-section decreases to 7000 cm2, and power
consumption reduces to 590 kW.
H. U. Eckert has studied experimentally the transformer coupled toroidal discharge at
current frequencies of 60 and 9600 Hz [3, 4]. At current frequency of 9600 Hz, the power of
transformer coupled toroidal discharge was ~ 12 kW, at argon pressure of ~ 53 kPa. At a
higher pressure the discharge became unstable and died out. At current frequency of 60 Hz,
the power of TCTD was 0.5÷1.9 kW, at argon pressure of 0.6÷3 torr.
Induction Transformer Coupled Discharges: Investigation and Application 115
Measurements of electrical and thermal characteristics of the low-frequency (8 kHz)

transformer coupled toroidal discharge in argon and air at pressures of 0.01÷32 kPa and
powers of 3÷30 kW are presented in [5]. Due to development of current-convective instability
of the toroidal plasma coil, V.M. Goldfarb et al. also failed to achieve the atmospheric
pressure of plasma gas.
For the first time the transformer coupled toroidal discharge of the atmospheric pressure
was obtained by I. M. Ulanov and V. A. Kogan [6]. Application of vortex stabilization
allowed the authors to get firstly a stable stationary discharge of the transformer type under
the atmospheric pressure of argon and air, current frequency of 10 kHz and discharge power
of up to 200 kW. Possible use of the transformer plasmatron for direct plasma-chemical
synthesis of NO from air was also shown [7]. The super-equilibrium yield of NO (~7%) was
reached without application of quenching methods.
The transformer coupled toroidal discharge of the atmospheric pressure in argon and in
the mixture of argon with nitrogen and oxygen was also studied experimentally in [8] at the
discharge power of up to 8 kW.
The problems of mathematical modeling of low-frequency transformer coupled toroidal
discharges in argon are considered in [9, 10]. Also, a simple model that describes the
temporal and spatial characteristics of the voltage and current in conductive toroidal chambers
is given in [11].
Since the transformer coupled toroidal discharge allows generation of super-pure plasma
even of aggressive substances and compounds, it can be used for development of various
devices for etching semiconducting materials, film deposition, surface modification, etc.
There are the patents [12, 13], which describe the gas-discharge devices of the
transformer type for “etching” the semiconducting materials. The experiments on deposition
of titanium nitride films with application of the transformer coupled toroidal discharge are
known [14].
Also the transformer coupled toroidal discharges can be successfully used for creation of
electrodeless light sources with a long service life. The first studies in this field were carried
out by J. M. Anderson [15, 16]. The electrodeless luminescent lamp of the low pressure is
described in [15] and patent [16]. However, the efficiency of this device was low because of
high losses in ferrites (10÷15 W). The construction of high-intensive gas-discharge lamp on
the basis of TCTD is also suggested in patent [17].
The possibility to make the efficient lamps of the transformer type appeared only after
development of high-quality ferrites with low losses. The OSRAM Company has developed
the electrodeless luminescent lamp ENDURA [18]. The parameters of this lamp were studied
experimentally and theoretically in [19].
The authors of this chapter have firstly studied TCTD in the mixture of mercury and
sulfur [20] and in pure neon [21, 22]; for the first time TCTD was studied in detail in mercury
vapors under different conditions of discharge glowing [23].
As it follows from the above brief review of publications, the transformer–coupled
toroidal discharges can be successfully used by various practical applications and for
development of electrodeless gas-discharge devices with a long service life and high
efficiency.
1. TRANSFORMER COUPLED TOROIDAL DISCHARGES:

MAIN PRINCIPLES
1.1. Principle of Operation of Transformer Coupled Toroidal Discharges
The principle scheme of generation of the low-frequency transformer coupled toroidal

discharge is shown in Fig. 1.
Figure 1. Principle of operation of transformer coupled toroidal discharge.1. Gas–discharge chamber.2.

Ferrite core.3. Primary winding.4. 4.Power supply.
The induction discharge of the transformer type is a closed toroidal gas discharge 1,
inductively connected to primary coil 3 via ferromagnetic core 2. In such a discharge plasma
is heated at a release of Joule heat of closed currents.
The closed currents in the toroidal induction discharge of the transformer type are caused
by vortex electric field E, excited by alternating magnetic flux in the magnetic core:
Φ (t ) = Φ meiωt (1)
and alternating magnetic flux, induced by the currents in plasma:
Φ i (t ) = ∫∫ B(t )dS (2)
where ω is circular frequency of current, B is magnetic induction.

In this case circulation of the electric field and current in the discharge chamber can be
described by equations, where the bias current can be neglected:
d
∫ E (t )dl = − dt (Φ(t ) + Φi (t )) (3)
L
rot (B ) = μ0 j
Neglecting active resistance of the primary coil, we will obtain that amplitude Фm does
not depend on the discharge current, whose demagnetizing action is compensated by an
increase in the current strength in the primary coil. To estimate a contribution of component
Ei, excited by alternating magnetic flux Фi, into the discharge current, let’s introduce a current
plasma coil as a closed ring of L=2πRk perimeter. Maximal induction of magnetic field of this
I
current ring occurs in its center and equals Bk = πμ0 . Considering ring area S=L2/(4π),
L
we obtain the value of magnetic flux Фi=μ0·I·L/4.
Respectively, the ratio of electric field components caused by the alternating magnetic
flux in magnetic core Ф, and magnetic flux, induced by current Фi, equals [9]:
⎛ ωΦ i ⎞
⎜ ⎟
Ei ⎝ L ⎠ μ IL πμ I ⎛ I ⎞
= = 0 ≈ 0 ≈ 4 ⋅ 10− 6 ⎜⎜ ⎟ (4)
E ⎛ ωΦ m ⎞ 4 Bm S Bm L ⎝ Bm L ⎟⎠
⎜ ⎟
⎝ L ⎠
The real sizes of plasma ring L varied from 0.2 to 2 m at the discharge current from 1 to
260 А and Bm from 0.1 to 1 T. Correspondingly, we obtain: Ei/E<10-2. Therefore, within the
studied parametrical range of the low-temperature transformer coupled toroidal discharge,
circulation of the electric field differs from zero only because of the magnetic flux changing
with time in the magnetic core:
dΦ (t )
∫ E (t )dl = − dt (5)
L
According to the above estimates, we can conclude that in plasma of the induction
discharge of the transformer type rot(E) = 0, E = - grad φ. With consideration of the Ohm law,
from the continuous equation for current density J, we obtain:
∇σ∇ϕ = 0 (6)
To set the boundary conditions for (6), let’s take conditional surface Σ, separating a
closed current coil into two parts. In this case, the boundary conditions on the Σ – surface,
which reduce the doubly connected domain to the simply connected one, will be
ϕ (Σ −0 ) = 0...ϕ (Σ + 0 ) = iωΦ m
i.e., the potential jump is determined by the rate of magnetic flux alteration by formula (5).
According to (5), the discharge voltage and electric field strength equal:
U (t ) = −iωBm Seiωt = −iωμμ0 H m Seiωt (7)
E=U/L (8)
where Bm is the amplitude value of magnetic induction, Hm is the amplitude value of magnetic
field strength, μ is magnetic conductivity of ferrite, and S is magnetic core cross-section.
Then, in this chapter, U, E, I, etc. mean the mean-root-square values of voltage, field strength,
discharge current, and etc.
It should be noted specially that in the framework of the considered model, the model of
the transformer discharge is similar to the model of electric arc with electrodes, combined
into surface Σ [10]. This gives us the ground to use the term “transformer arc” and describe
the transformer discharge by the same models as the other arc discharges.
Since µ>>1 for ferromagnetic, electric field strength E, required for TCTD glowing, can
be reached in a low-frequency discharge. In fact, the TCTD is the most low-frequent of all
electrodeless discharges [3, 4].
The minimal value of current frequency, sufficient for induction discharge glowing, is
determined by the electric field strength in discharge E. The main criteria for maintenance of
TCTD glowing were considered firstly in the paper of Eckert [3], with regard to a particular
quasistationary discharge in argon. General relationships can be also used for analysis of
criteria of TCTD generation in other plasma-forming media.
1.2. Optimization of the Transformer Coupled Toroidal Discharge
The correct choice of ferromagnetic material allows minimization of losses in the

magnetic core and significant reduction of core sizes for the transformer gas-discharge
device.
According to expression 7, the required voltage of discharge glow U can be achieved by
an increase in the core cross-section S or in current frequency f or in induction Bm. It is
obvious that a significant increase in the core cross-section is unacceptable from the practical
point of view. Thus, to achieve the required voltage, it is desirable to increase maximally the
value of f·Bm, what can be only done with application of special kinds of magnetic materials.
Two groups of materials with minimal volumetric losses at high frequency and induction
of the magnetic field can be distinguished: amorphous soft magnetic alloys and special kinds
of ferrites.
Flat-strip amorphous soft magnetic alloys are characterized by high permeance, low
coercitivity (Hc below 1 A/м), high specific electric resistance (ρ ~ 108 Ohm·cm), low losses
for hysteresis and eddy currents (3÷5 times less than the best crystalline alloys).
Since they are the semiconductors, ferrites have high specific volumetric electric
resistance, exceeding the resistance of steels and alloys by factor of 50 and more. This makes
it possible to use ferrites for high frequencies without a significant rise of eddy-current losses.
Among disadvantages of ferrites there is relatively strong dependence of their magnetic
properties on temperature. In contrast to ferrites, amorphous alloys can operate efficiently up
to ~ 250–300 ºC.
In case of low operation temperatures of the magnetic core, which allow application of
both ferrites and amorphous alloys, the choice of the magnetic material will be determined
from the condition of maximal Bm at the required current frequency and given threshold level
of specific heat losses Pv (W/cm3).
The value of losses in the magnetic core is composed of separate components: eddy-
current losses, hysteresis, and magnetic viscosity (magnetic aftereffect). On the ground of
statistical processing of abundant experimental data, the reference literature on electric
engineering provides the following dependence of specific losses on frequency and induction
of the magnetic field:
α β
PV ⎛ f ⎞ ⎛ Bm ⎞
=⎜ ⎟ ⎜ ⎟ (9)
P* ⎜⎝ f* ⎟⎠ ⎜⎝ Bm* ⎟⎠
where f* = 1 kHz, Bm* = 1 T are the basic values of frequency and induction; P*, α, β are
coefficients, obtained by processing of experimental dependences Pv(f, Bm).
The typical dependence of specific heat losses PV on the amplitude value of magnetic
induction Bm at different current frequencies is shown in Fig. 2 for the strip magnetic cores of
amorphous alloy and ferrites of the Philips 3C96 grade. According to the figure, with a rise of
Bm heat losses in ferrites increase faster than heat losses in amorphous steel because of lower
saturation induction. However, at fixed induction of magnetic field, with a rise of frequency
heat losses in amorphous alloys increase faster than in ferrites (Fig. 2).
The above differences between ferrites and amorphous alloys are visually demonstrated
in Fig. 3. Dependence of f·Bm on frequency f is shown there for the fixed level of specific heat
losses in the magnetic core (0.5 W/cm3), for amorphous alloy and different kinds of ferrites.
Philips 3C96, 25 kHz

Philips 3C96, 200 kHz
Steel 5BD,25 kHz
Steel 5BD,250 kHz
1
3
Specific loss Pv, W/cm
0,1
0,01
0,01 0,1 1
Bm, T
Figure 2. Dependence of specific heat losses in the core PV on magnetic induction Bm.
5BD
3С96
3С94
5x10
4
3С90
3F3
4 3F4
4x10
f·Bm, Hz·T 3F35
4
3x10
4
2x10
4
1x10
0
10 100 1000
f, kHz
Figure 3. Dependence of f·Bm on current frequency f at the fixed level of specific heat losses PV =0.5
W/cm3.
According to Fig. 3, there is the optimal value of current frequency, corresponding to the
maximal value of f·Bm at a constant level of heat losses, and the maximum achievable value of
f·Bm depends essentially on the quality of magnetic material.
In the range of low current frequencies (below 20÷30 kHz) the magnetic cores on the
basis of amorphous alloys have a definite advantage over the best grades of ferrites because
of a weak dependence of specific heat losses on magnetic field induction. At current
frequencies above 100 kHz application of ferrites seems to be the most optimal.
1.3. Matching the Power Source and the Transformer Coupled

Toroidal Discharge
The transformer plasmatron is similar to inductive thermoelectric devices (induction

furnaces). Hence, to match the power source with the transformer plasmatron, we can use the
known calculation schemes, applied for devices of induction heating. In these devices the
oscillating circuit, formed by an inductor and additional capacitor bank, serves as the load.
The equivalent scheme of the power source and transformer plasmatron is shown in Fig.
4a. Here Rg and Xg are active resistance and reactance of the generator, Rtr and Xtr are active
resistance and reactance of the transformer, XC1 and XC2 are capacitances for matching.
Under the steady state it is convenient to present the oscillating circuit, formed by the
transformer and matching vessels, as a parallel equivalent circuit, including equivalent active
resistance Req and reactance Xeq (Fig. 4b). At this, Req and Xeq are determined by the following
formulas [6]:
Rtr ⋅ BC21
Req =
(B12 + X tr ⋅ BC1 ⋅ BC 2 )2 + (Rtr ⋅ BC1 ⋅ BC 2 )2
X eq =
( B12 + X tr ⋅ BC1 ⋅ BC 2 ) ⋅ ( X tr ⋅ BC1 − 1) + (Rtr ⋅ BC1 ⋅ BC 2 )2
(B12 + X tr ⋅ BC1 ⋅ BC 2 )2 + (Rtr ⋅ BC1 ⋅ BC 2 )2
B12 = BC1 − BC 2 (10)
X eq
tgϕ =
Req
Req
Z eq =
1 + (tgϕ )2
where BC1, BC2 are capacitive susceptances; ϕ is phase angle, Zeq is total resistance of the
circuit. The value of Zeq should correspond to the inner resistance of generator, and tgϕ of the
load should fit the ratings of generator to provide the transfer of maximal active power to the
transformer plasmatron.
a) b) X eq
Rg X g X C1
X tr
~ Ig X C2
R tr R eq
Figure 4. Equivalent schemes:a) Transformer plasmatron; b) Parallel equivalent circuit.
Thus, calculations are reduced to determination of Req and Xeq and, hence, XC1 and XC2 for
every specific operating condition (idle running – before discharge ignition; operating
condition – for specific gas and pressure). Besides, it is necessary to determine Rtr, and Xtr
(active resistance and inductance of the transformer plasmatron). The equivalent circuit of the
plasmatron, required for determination of Rtr and Xtr, is the equivalent circuit of the
conventional transformer with active-inductive loading. This circuit is widely spread and
described in electric engineering.
In conclusion it should be noted once more: various electrodeless gas-discharge devices
of the high and low pressure can be developed on the basis of transformer coupled toroidal
discharges for efficient plasma generation with a high density of charged particles of any
chemical composition. Therefore, the transformer coupled toroidal discharges can be used
successfully by many practical applications, which require the use of low-temperature
plasma.
2. TRANSFORMER PLASMATRONS
2.1. Experimental Setup
Initially experiments on generation and investigation of transformer coupled toroidal

discharges in inert and molecular gases were carried out at setup No. 1, shown in Fig. 5a. The
picture of this setup is shown in Fig. 5b.
Figure 5a. Experimental setup No. 1. 1 – primary winding; 2 – sections of magnetic core; 3 – discharge
chamber; 4 – heat exchanger; 5 –main gas feed; 6 – secondary gas feed; 7 – probe No.1; 8 – probe
No.2; 9 – probe No.3; 10 – spectrophotometer; 11 – mass-spectrometer.
Figure 5b. Transformer plasmatron for current frequency of 10 kHz.
The total power of setup No.1 was 180 kW. The magnetic core consisted of 8 separate
sections made of cold-rolled electric sheet steel 3425 of the 80-µm thickness. The outer
diameter of each section was 420 mm, the inner diameter was 160 mm, and the height was 70
mm. Each section had the own system of primary coils. The number of primary coils is 4. All
primary coils were connected either in-parallel or in-series to the machine generator (10 kHz)
depending on the experimental conditions. The discharge chamber was of the closed
“toroidal” shape. The perimeter of discharge chamber along the mean line was L=1.8 m. The
chamber consisted of the metal, water-cooled, mutually insulated sections with the inner
diameter Dint=(80÷100) mm and length (20÷60) mm. The plasma gas was fed to the discharge
chamber through two tangential holes with the diameter of 6 mm and the swirl angle of 20°.
The gas flow was controlled by rotameters.
Experiments on thermal synthesis of ozone from oxygen and natural gas conversion in
the presence of СО2 were carried out in the mixture with argon. For these purposes two
experimental methods were used. By the first method a mixture with the known percentage of
the studied gas in argon was fed into the main feed chamber. By the second method the feed
of argon and required reagents was separated. Oxygen, natural gas and carbon dioxide were
fed through additional vortex chamber 6, located at distance L=60 cm from the point of the
main argon supply. This allowed operation in mixtures of the studied gas and argon within a
small region of the discharge chamber at relatively high concentration of products in
discharge.
Total induced discharge voltage Udisch was measured by an additional insulated coil of
wire, enveloping the magnetic core. The mean mass temperature of gas at discharge chamber
outlet was measured by calorimetric study of heat exchanger 4.
Distribution of the mean mass temperature of gas in the discharge chamber was measured
by calorimetric (enthalpy) probe [24], whose scheme is shown in Fig. 6. The gas temperature
at the outlet as well as the water temperature at the probe inlet and outlet was determined by a
chromel-coppel thermocouple. The gas flow through the probe was determined by
accumulation in a diaphragm evacuated volume. During the experiment the volume was
evacuated constantly, and the pressure drop on the diaphragm was supersonic always. When
measuring the gas flow, buffer volume evacuation was stopped, and the flow rate was
determined by the time of pressure alteration in this volume. The time range was chosen to
obtain a constant supersonic drop on the diaphragm, corresponding to a constant flow rate. A
relative error of mean mass temperature measurements by this method depends on the ratio of
the inlet velocity of the sampling gas to the velocity of the incident flow and makes up from
3% to 17%.
Figure 6. The scheme of calorimetric probe with the system of sampling and analysis
Radial distributions of mean mass temperature of gas in the discharge chamber were
measured for two distances from the point of plasma gas supply: l=60 mm (probe No.1); and
l=800 mm (probe No.2).
A gas sample was analyzed by spectrophotometer CF–26 and mass-spectrometer
Quadrovak. Preliminary, the mass-spectrometer and spectrophotometer were calibrated by the
gas mixtures with given concentrations of substances.
This method of experiments allowed determination of radial distributions of product
concentrations in the discharge chamber together with measurement of temperature
distributions.
Both tubular probes and probes with rectangular cross-section were used in the
experiments. The typical rate of high-temperature gas cooling was estimated by the formulas,
suggested in [25]:
dT
= 0.107 ⋅
(
w ⋅ T f − Tw ) (11)
dt d ⋅ Re0.2
(where w is average gas velocity, Tf, Tw are mean mass temperatures of the flow and probe
wall, d is efficient diameter of the gas-sampling channel); for initial temperature Tw
3000÷6000 K this rate was about 106÷107 К/s for the probe with d = 1.5 mm.
Samples for analysis of product concentration at plasmatron gas outlet were taken by
probe No.3. Probe No.З was located near the heat exchanger (Fig. 5a). The heat exchanger
consisted of the water-cooled tubes with diameters of 8 and 15 cm and lengths of 1 and 1.5 m,
correspondingly. According to the estimate by formula (11), the maximal average rate of gas
cooling in this heat exchanger is dT/dt≈103÷104 К/s.
A disadvantage of the transformer plasmatron (рис. 5b) was the fact that magnetic cores
were made of ordinary transformer steel 3425 and the machine generator with current
frequency of 10 kHz was used as the power source. Therefore, the magnetic cores of the
transformer plasmatron were heavy (~300 kg), and the plasmatron itself was too bulky. Now
it is possible to make more compact transformer plasmatrons because of development of new
high-quality ferrites for power electronics and design of compact and cheap power sources
with current frequency of ~100÷400 kHz [26].
The further works were aimed at reduction of transformer plasmatron sizes with
preservation of high productivity. The works performed allowed development of a compact
transformer plasmatron with the power of 20 kW and current frequency of 100 kHz (Fig. 7a,
7b).
The magnetic cores of plasmatron 1 were assembled of EPCOS ferrite bars of N 87 grade
(the analog of ferrite 3F3, Fig. 3). The total cross-section of magnetic core was 100 cm2, and
the mass was ~ 20 kg. Water-cooled gas-discharge chamber 2 was made of 12 sections,
mutually insulated by silastic gaskets 4. The inner diameter of gas discharge chamber was D
= 45 mm and the gas discharge perimeter along the mean line was L = 1 m. To stabilize a
plasma discharge, the transformer plasmatron was mounted vertically with argon input via
vortex chamber 3. Plasma went out from the opposite side to the chamber of chemical reactor
6. To perform various plasma-chemical reactions, the inlet for chemical reagents 5 was made
in the lower part of the discharge chamber, and branch pipes were provided below for the
input of additional chemical reagents and plasma observations through a window. The
chamber of chemical reactor can be used both for accumulation of reaction products and
allocation of items for plasma processing.
Figure 7a. The scheme of transformer plasmatron with ferrite cores.1 – ferrite cores; 2 – water-cooled
chamber; 3 – vortex tangential input of argon; 4 – silastic gaskets; 5 – reagent input; 6 – outlet to
chemical reactor.
Figure 7b. Transformer plasmatron with ferrite cores of the 20-kW power.
To ignite a discharge, air was evacuated from the plasmatron and chamber to the pressure
of about 10–20 Pa. After powering-on the system, argon was fed into the discharge chamber,
and discharge was ignited. When the atmospheric pressure was reached, a damper was open
to let argon into the atmosphere.
The voltage of discharge glow was measured by an additional coil, enveloping magnetic
cores 1; the discharge current was measured by the current transformer. The thermophysical
characteristics of discharge were determined by calorimetric measurement of water-cooled
chamber 2.
The main achievement of the chapter authors should be marked out specially: for the first
time the stable transformer coupled toroidal discharge of atmospheric pressure was obtained
in argon, air, and argon mixtures with molecular gases (oxygen, hydrogen, carbon dioxide,
and natural gas).
2.2. Electric-Physical and Thermal-Physical Characteristics of Transformer

Plasmatrons
The electric-physical and thermal-physical characteristics of transformer coupled toroidal

discharges were studied in argon, air, and mixtures of argon with natural gas and carbon
dioxide at atmospheric and reduced pressures. Experiments were carried out at setup No.1.
The effect of “glow” conditions on the electric field strength of a discharge in argon was
studied within the range of pressures of 10÷105 Pa; currents of 90÷450 А; and argon flow rate
of 1÷30 g/s. The thermal-physical and electric-physical characteristics of the transformer
coupled toroidal discharge were also studied at atmospheric pressure in the mixture of argon
and hydrogen in the transformer plasmatron with ferrite cores (Fig. 7а).
The effect of argon flow rate on the electric field strength in a discharge at atmospheric
pressure is shown in Fig. 8.
1,5
1,4
Electric field strength, V/cm
1,3
1,2
1,1
1,0
0,9
0,8
0 5 10 15 20 25 30
Ar flow rate, g/s.
Figure 8. Electric field strength vs. argon flow rate.I=200 А, p=1 atm.
1,2
Electric field strength, V/cm

1,1
1,0
0,9
0,8
50 100 150 200 250 300 350 400 450
TCTD current, A
Figure 9. Current-voltage characteristic of transformer coupled toroidal discharge in argon. Argon flow
rate is 4 g/s.
The initial incident region of the curve corresponds to unsteady operation, when gas-
vortex arc stabilization is insignificant because of a low gas flow rate. Therefore, a discharge
is spatially unstable (because of development of current-convective instability), what
increases heat transfer between the arc and discharge chamber wall, and, hence, this leads to a
rise of applied power required for stationary discharge maintenance. At a given current this
increases the electric field strength. The second constant region of the curve corresponds to
stable arc “glow”. A following increase in the electric field strength relates to an increase in a
degree of flow turbulence and increase in heat losses to the plasmatron wall as well as to a
rise of energy, “carried out” by a gas flow. In general, this increases the power required for
discharge maintenance.
The current-voltage characteristic of a discharge in argon was studied for TCTD currents
of 90÷450 А. The typical current-voltage characteristic of the TCTD is shown in Fig. 9 for
argon flow rate G=4 g/s. According to the diagram, at the initial stage (at low currents) the
current-voltage characteristic is descending. Then, after a small region of constant Е(I) there
is an ascending part of the curve. This dependence E(I) is also characteristic for the DC arc
discharges.
Dependence of the electric field strength on the argon pressure is shown in Fig. 10. The
diagram represents the experimental results both with and without vortex gas supply.
According to this diagram, at low pressure (diffusion discharge) the electric field strength
decreases with a rise of pressure. This is caused by the fact that the free path of electrons
decreases with a rise of pressure, and this provides less electron losses to the wall of
discharge chamber; on the other hand, electron concentration increases. This reduces the
electric field strength in a discharge at a given current density. With the following rise of
pressure the discharge contraction occurs, and current-convective instability increases. A
discharge takes a wavy shape, the arc length increase, and heat transfer between the arc and
the wall becomes more intensive. This increases the electric field strength, required for
discharge maintenance. With the following rise of pressure considerable voltage pulsations
occur in a discharge and the instantaneous voltage can exceed the maximal voltage, induced
by the magnetic core.
If the vortex supply of plasma gas is used, current-convective instability of discharge is

expressed weakly. This allows a pressure increase in the discharge chamber of up to the
atmospheric one and higher without significant fluctuations of the discharge voltage.
1,2
1,0
Electric field, V/cm 0,8
0,6
0,4
0,2
0,0
2 3 4 5
10 10 10 10
Argon pressure, Pa
Figure 10. Electric field strength vs. argon pressure. ■ – Without vortex supply of argon (I=100 A). + –
With vortex supply of argon (I=100 A).G=3÷5 l/s
2
8
7
Electric field, V/cm
4 1
0 2 4 6 8 10 12 14 16 18
Air flow rate, g/s.
Figure 11. Average electric field strength E vs. air flow rate. Discharge current I=125 А. 1 – E at initial
arc region in electric-arc plasmatron; 2 – E at developed turbulent region in electric-arc plasmatron.
Dependence of the electric field strength in air flow rate (Fig. 11) has been studied. The
behavior of E=f(G) function in air is similar to the experiments with argon. It is obvious that
the electric field strength in air is 4÷6 times higher than in argon. In the same diagram
empirical dependences of electric field strength on air flow rate are shown at the initial region
(curve No.1) and at the developed turbulent region (curve No. 2); they were derived in [27] at
processing of data on the electric field strength in electric-arc plasmatrons.
Since there are no electrodes we could work with aggressive gas mixtures both at a
reduced and atmospheric pressures. There was no any destruction of discharge chamber walls,
and plasma was not contaminated by admixtures. Dependence of electric field strength in [Аг
+ O2] mixture on oxygen content in this mixture is shown in Fig. 12. According to the
diagram, even at low oxygen concentrations in the mixture the electric field strength increases
significantly. For instance, at oxygen content in argon of ~ 1 % the electric field strength
increases almost twice.
Dependences of electric field strength in [Аг + СО2], [Аr + СН4] and [Ar + H2O]
mixtures on their percentage in argon are shown in Fig. 13. It is clear that for these gases the
electric field strength is relatively high, therefore, a more powerful and efficient setup with a
perfect magnetic core and power source of increased frequency (~100 kHz) is required for the
work with such pure gases.
12
10
0 10 20 30 40 50
O2 concentration, vol. %
Figure 12. Electric field strength in [Ar + O2] mixture vs. oxygen content. Discharge current I=125 А,
argon flow rate G=4 g/s.
12 H 2O
11 CO2
10
9
8 CH4
7
6
5
4
3
2
1
0
0 10 20 30 40 50 60
Concentration, vol. %
Figure 13. Electric field strength in [Ar + H2O, Ar + CO2, Ar + CH4] mixtures vs. substance
concentration in argon. I=125 А, argon flow rate G=4 g/s
Dependence of argon plasma temperature at the atmospheric pressure on argon flow rate
is shown in Fig. 14 for the distance from the axis of discharge chamber r/R=0.3 and discharge
current I=125 А. According to the diagram, with a rise of argon flow rate the high-
temperature zone is localized in the center of discharge chamber, i.e., with a rise of the flow
rate the mean mass temperature of gas decreases within the given radius at the same discharge
current. Temperature distribution in the discharge chamber is shown in Fig. 15 for two
distances from the point of gas supply and different air flow rates. According to the diagrams,
temperature distributions are subjected to two main factors: the flow rate of plasma gas and
the distance from the point of gas supply. A rise of the gas flow localizes the high-
temperature zone in the center of discharge chamber, and the temperature gradient becomes
higher with a rise of the flow rate. With an increase in the distance from the point of gas
supply the temperature profile becomes smoother.
5000
4500
4000
3500
Temperature, K
3000
2500
2000
1500
1000
500
1,0 1,5 2,0 2,5 3,0 3,5 4,0
Ar flow rate, g/s.
Figure 14. Temperature of argon plasma at atmospheric pressure vs. argon flow rate for r/R=0.3, z/R=1
(probe No.1). Discharge current I=125 A
7000
6000
5000
Temperature, K
4000
3000
2000
1000
0
0 10 20 30 40 50
Distance from discharge axis, mm
Figure 15a. Temperature distributions in the discharge chamber for two distances from the point of air
input into the chamber: I=125 A. L =60 mm (probe No.1).• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s
6000
5000
Temperature, K
4000
3000
2000
1000
0 5 10 15 20 25 30 35 40 45 50 55
Figure 15b. Temperature distributions in discharge chamber for two points of air input into the
chamber: I=125 A.L =800 mm (probe No.2).• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s
Discharge power
efficiency, %
Plasma Jet power
20 Heat Losses 40
15 30
Power, kW
10 20
5 10
0 0
0,0 0,1 0,2 0,3 0,4
H2 flow rate, g/s.
Figure 16. Thermal-physical characteristics of transformer plasmatron with ferrite cores. I=65 А, GAr=4
g/s.■ – discharge power;• – power in plasma jet, flowing into plasmachemical reactor;▲ – heat flux to
water-cooled wall of plasmatron;♦ – plasmatron efficiency.
Then, the results of experimental investigations have been used for development of a
compact transformer plasmatron with ferrite cores (рис. 7а, b) for plasmachemical reactions
in a plasma jet [Ar+H2] leaving the plasmatron.
Hydrogen was fed to the lower section of the gas-discharge chamber. Dependence of
plasmatron efficiency, heat losses and power in the leaving plasma jet on hydrogen flow rate
is shown in Fig. 16.
According to the diagram, the maximal plasmatron efficiency is ~ 35%, and it can be
achieved at hydrogen flow of 0.1 g/s. The following decrease in plasmatron efficiency is
explained by a rise of gas flow turbulence and enlargement of heat losses to the plasmatron
walls (Fig. 16).
Dependence of the power of argon-hydrogen plasma jet on hydrogen flow rate was
determined as the difference between discharge power and heat losses. According to the
diagram, the maximal power was obtained at hydrogen flow rate of about 0.1 g/s, with the
following increase in the flow rate the jet power decreased slightly.
Analysis of thermal-physical and electric-physical characteristics of the TCTD allows us
to make a conclusion that characteristics of the transformer coupled toroidal discharge are
similar to characteristics of a positive DC arc column.
It is shown that without vortex stabilization of a discharge current-convective discharge
instability starts developing from the pressures in discharge chamber of 104 Pa, what leads to
significant voltage fluctuations, providing TCTD “extinction”. If there is the vortex gas flow
in the discharge chamber, the described instabilities do not develop, what allows a stable
TCTD at the pressures above the atmospheric one.
2.3. Thermal Production of Ozone in Plasma of Transformer Coupled

Toroidal Discharge
By now the most common plasma chemical method of ozone production is its synthesis
in a barrier discharge. Ozone is produced directly in the zone of a discharge in one stage.
Another plasmachemical method of ozone production is the thermal method. In this case
ozone is produced in two stages. The first stage is oxygen “heating” up to the temperatures of
dissociation. The second stage is the process of fast cooling of dissociated oxygen. In this
process ozone is synthesized at the stage of non-equilibrium cooling of gas, consisting of
oxygen atoms, molecules and radicals.
Let’s consider the main processes occurring at thermal production of ozone.
At the first high-temperature stage ozone concentration under the conditions of
thermodynamic equilibrium is negligibly small, and ozone production by the considered
method is possible only at the stage of non-equilibrium gas cooling, when ozone is formed by
recombination of atomic and molecular oxygen. Let’s estimate the typical rate of cooling of
dissociated oxygen (at the temperatures of 3500÷4500 К), when atomic oxygen turns mainly
into ozone, but not into molecular oxygen:
1. Transition of O–atoms into molecular oxygen and ozone at cooling of dissociated gas
occurs by the following main reactions:
(R1) O+O2+M Æ K1 O3 + M
(R2) O+O3 Æ K2 O2 + O2 (12)
(R3) O+O+M Æ K3 O2 + M
2. A change in the density of atoms, radicals and molecular oxygen in the considered
processes equals:
d [O ]
= − K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] − K 2 ⋅ [O ] ⋅ [O3 ] − K 3 ⋅ [O ]2 ⋅ [M ] (13)
dt
d [O3 ]
= K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] − K 2 ⋅ [O ] ⋅ [O3 ] (14)
dt
d [O2 ]
= − K1 ⋅ [O ] ⋅ [O2 ] ⋅ [M ] + K 2 ⋅ [O ] ⋅ [O3 ] + K 3 ⋅ [O ]2 ⋅ [M ] (15)
dt
Thus,
d ([O ] + [O3 ])
= −2 K 2 ⋅ [O ] ⋅ [O3 ] − K 3 ⋅ [O ]2 ⋅ [M ] (16)
dt
3. A relative change in concentration of oxygen atoms and radicals during the processes
of cooling equals:
1 d ([O ] + [O3 ]) K 2 ⋅ [O ] ⋅ [O3 ] K 3 ⋅ [O ]2 ⋅ [M ]

∫ ([O ] + [O ]) dt = − 2 ∫ ([O ] + [O ]) dt − ∫ ([O ] + [O ]) dt (17)
3 dt 3 3
The expression in the left part is a relative portion of atomic oxygen, converted into
molecular oxygen at cooling [28]. Let’s determine the condition, when it is assumed that
concentration of molecular oxygen at quenching is not changed, i.e., this is the condition,
when almost all atomic oxygen turns to ozone. With this assumption we can consider that the
left part of equation (17) is a small value. Considering the rate of gas mixture cooling
constant (dT/dt = const), we obtain the following condition:
dT ⎡ K 2 ⋅ [O ] ⋅ [O3 ] K 3 ⋅ [O ]2 ⋅ [M ] ⎤
>> ⎢− 2∫ dT − ∫ dT ⎥ (18)
dt ⎢⎣ ([O] + [O3 ]) ([O] + [O3 ]) ⎦⎥
Integrands (18) include equilibrium values of densities, determined by relationships:
[O] + [O3 ] ≈ C [O] = K (T )

[M ] [O ] [O ] (19)
3
where С[O] is concentration of oxygen atoms at the initial temperature, K(T) is the constant of
equilibrium between atomic oxygen and ozone. Integrals converge within the temperature
area, where atomic oxygen turns into ozone. The sought temperature area is determined by
the following condition: K(T) ~ 1, which is satisfied at T ~ 800 K.
Integrals should be taken at the temperatures of 3500÷4500 K, when concentration of
atomic oxygen is high. After substitution of values for T=4500 K, we obtain dT/dt>>108 K/s.
It should be noted that this condition determines ozone formation in assumption of almost
all transition of atomic oxygen into ozone, but not into molecular oxygen. In case of partial
transition of О–atoms into ozone we can expect that formation of not-zero ozone
concentration in the process of dissociated oxygen cooling is possible at dT/dt ~ 107÷108 K/s.
Quenching with the given rates is easy to organize by the recuperative methods because
the process of gas cooling in a heat exchanger is well studied, and this allows us to study
experimentally the effect of the rate of “hot” oxygen cooling on kinetics of the considered
reaction and final output of the product.
Experiments on investigation of thermal synthesis of ozone from the mixture of argon
and oxygen were carried out at setup No. 1. The power of this setup did not allowed operation
with pure oxygen; therefore, the experiments were carried out in the mixture of argon and
oxygen. Two methods of experiment implementation were used. In the first case, the mixture
with given concentration of oxygen was fed into the discharge chamber through the main
section of gas supply. In the second case argon was fed into the main section of gas supply,
and oxygen was fed into the secondary section (see Fig. 6). All experiments were carried out
for the discharge current I = 125A.
To study experimentally the effect of dissociated oxygen quenching on ozone production,
the water-cooled quenching probes of the slot type were made. These probes had the
rectangular cross-sections of the gas sampling channel with the sizes of 0.2х10 and 0.5х10
mm. The geometrical scheme of this probe is shown in Fig. 17.
The probes were located at the distance of 100 mm from the heat exchanger (Probe No. 2
– Fig. 6). The inlet sampling slot of the probe was at distance r/R=0.5. Visually this position
of the probe corresponded to the boundary of the “glowing” conducting channel. Thus, it can
be assumed that the initial temperature of gas was not lower than 4000 К (in these
experiments the initial temperature of a sample was not measured).
Figure 17. Quenching probe of the slot type.
The probes were located at the distance of 100 mm from the heat exchanger (Probe No. 2
– Fig. 6). The inlet sampling slot of the probe was at distance r/R=0.5. Visually this position
of the probe corresponded to the boundary of the “glowing” conducting channel. Thus, it can
be assumed that the initial temperature of gas was not lower than 4000 К (in these
experiments the initial temperature of a sample was not measured).
Experimental results in Fig. 18 are shown for two probes, providing different rates of
high-temperature gas quenching. In the diagram there are the values of volumetric ozone
concentrations, which can be reached at cooling of oxygen-argon plasma in probes,
depending on oxygen concentration in argon.
7
probe 0.2x10 mm, dT/dt~2*10 K/s.
7
0,55 probe 0.5x10 mm, dT/dt~10 K/s)
0,50
0,45
0,40
O3, Vol. %
0,35
0,30
0,25
0,20
0,15
2 4 6 8 10 12 14
O2, vol. %
Figure 18. Dependence of ozone concentration on oxygen concentration in initial mixture [Ar + O2].
According to the diagram, ozone concentration increases with a rise of gas cooling rate
and with an increase in oxygen concentration in the initial mixture. At that it can be seen that
ozone concentration depends non-linearly on oxygen concentration. In other words, initial
concentration of atomic oxygen influences directly the kinetic processes at quenching.
According to investigation of thermal synthesis of ozone in the transformer coupled
toroidal discharge with the use of recuperation quenching devices, at cooling rates of ~107
K/s, from 5 to 10% of oxygen turn into ozone at oxygen content in mixture [Ar+O2] from 15
to 3%, respectively.
2.4. Synthesis of Nitrogen Monoxide in Plasma of the Transformer Coupled

Toroidal Discharge
One of the most examined gas-phase plasmachemical processes is synthesis of nitrogen

monoxide from air. There are many publications dealt with kinetics and thermodynamics of
nitrogen oxide formation in low-temperature plasma. The first detailed investigation of this
plasmachemical process is the research of Ya.B. Zeldovich, P.Ya. Sadovnikov, and D.A.
Frank-Kamenetsky [29], where it is shown that the bimolecular mechanism of nitrogen oxide
formation and decay, used before this research, does not explain the measured times of
reaction. The cited paper suggests the chain mechanism, which disposes the contradiction.
This mechanism includes the following elementary reactions:
1) O2 + M ÆK1 O + O + M…O + O + M ÆK1’ O2 + M

2) O + N2 ÆK2 NO + N……..NO + N ÆK2’ O + N2
3) N + O2 ÆK3 NO + O……..NO + O ÆK3’ N + O2
4) N2 + M ÆK4 N + N + M….N + N + M ÆK4’ N2 + M
5) NO + M ÆK5 N + O + M…N + O + M ÆK5’ NO + M
In assumption of atomic oxygen stationarity (concentration of atomic oxygen is

determined by equilibrium of reaction (1)), Zeldovich [29] has derived the following
expression for the total rate of nitrogen monoxide formation:
⎛ 86000 ⎞
exp⎜ − ⎟
d ( NO )
dt
= 1.5 ⋅ 1013 ⋅ ⎝ RT ⎠
(O2 )
{ }
⋅ [NO ]2 − ( NO )2 mole/cm3/s (20)
where [NO] is equilibrium concentration at estimated temperature.

On this basis the optimal rate of air quenching was determined in [30] within the
temperature range of 3300÷1800 K; at this rate total decay of nitrogen monoxide equals 5%
of the initial equilibrium value:
dT
= 1.5 ⋅ 1013 ⋅ exp⎜ − ⎟ ⋅
{ }
⎛ 86000 ⎞ [NO ] − (NO ) ⋅ (T1 − T2 )
2 2
(21)
dt ⎝ RT ⎠ (O2 ) ⋅ 0.05 ⋅ (NO )1
where T1 – initial temperature, T2 – final temperature, (NO)1 – initial concentration of NO.
According to this expression, to get the maximal output of NO, it is necessary first to
“set” the gas temperature, corresponding to the maximal equilibrium content of the product (T
~ 3400 K), then quenching should be carried out with the rate of ~5·106 К/s, with this rate
reduction to 104 К/s (at the final temperature of ~ 1800 К). It should be noted that the
described process of nitrogen oxide formation by cooling of quasi-equilibrium air within the
temperature range of 1000÷4000 K is proved well by the experimental data [30]. If air is
cooled from the higher temperatures (> 5000 K) there is divergence between experimental
and calculation data: at air cooling from the temperatures above 5000 К, super-equilibrium
concentration of nitrogen monoxide is formed (~6÷7 %) [31].
Together with a practical interest to NO production, abundance of experimental and
theoretical works in this field also seemed very attractive. This allowed us to use this process
as the ‘model’ one for understanding of the features of the transformer coupled toroidal
discharge with regard to plasmachemical technologies.
Experiments on investigation of NO synthesis from air were carried out at setup No. 1.
The plasma-forming gas (air) was fed into the main vortex chamber. Temperature and
concentration distributions over the discharge chamber were measured in these experiments at
different distances from the point of air input into the chamber. The water-cooled tubular
probes-samplers were used for this purpose (Fig. 6). Radial distributions of product
temperature and concentration were studied in two different zones of the discharge chamber
(Fig. 5a):
Concentration of NO after the heat exchanger was also measured in these experiments
(probe No.3). The mean mass temperature of gas at the discharge chamber outlet was
measured via heat exchanger calorimetry.
Distributions of NO concentration over the radius of discharge chamber, obtained by
probes No.1 and No.2 are shown in Fig. 19(a,b) (temperature distributions under the same
conditions are shown in Fig. 15(a,b)). It can be seen that near the point of air input (probe
No.1) distribution of nitrogen monoxide concentration is similar to temperature distribution.
Far from the input point (probe No.2) distribution of nitrogen monoxide concentration does
not depend on the gas temperature. It attracts attention that near the wall of discharge
chamber, where the level of temperature is relatively low (Т ~ 300÷2000 К) NO
concentrations, measured by probes No.1 and No.2, differ significantly. At high temperatures
(Т>5000 К), NO concentrations, measured by probes No.1 and No.2 are close to each other.
8
7
6
5
NO, vol. %
4
3
2
1
0
0 10 20 30 40 50
Figure 19а. Distribution of NO concentration over the chamber of transformer plasmatron, L =60 mm
(probe No.1), I=125 A.Air flow rate.• 3.7 g/s; ■ 6.8 g/s * 12.6 g/s
8
7
6
5
NO, vol. %
4
3
2
1
0
0 10 20 30 40 50
Figure 19b. Distribution of NO concentration over the chamber of transformer plasmatron, L =800 mm
(probe No.2). I=125 A.Air flow rate. • 3.7 g/s; ■ 6.8 g/s * 12.6 g/s
Values of NO concentration vs. the temperature of a gas sample are shown in Fig. 20(a,b)
by circles and squares for probes No. 1 and 2, respectively. The solid line shows the
temperature dependence of equilibrium concentrations of NO. In the same diagram the heavy
line indicates calculated values of NO concentration, obtained in assumption of

thermodynamic equilibrium of gas before quenching.
8
7
6
5
NO, vol. % 4
3
2
1
0
0 1000 2000 3000 4000 5000 6000 7000
Temperature, K
Figure 20a. NO concentration vs. air temperature.L =60 mm (probe No.1), I=125 A. Air flow rate:• 3.7
g/s; ■ 6.8 g/s * 12.6 g/s
8
7
6
5
NO, vol. %
4
3
2
1
0
0 1000 2000 3000 4000 5000 6000
Temperature, K
Figure 20b. NO concentration vs. air temperature.L =800 mm (probe No.2). I=125 A. Air flow rate:•
3.7 g/s; ■ 6.8 g/s * 12.6 g/s The equilibrium curve of NO content in air depending on T.
It can be seen in Fig. 20(a,b) that NO content, measured at low (Т ~ 300÷2000 К) and
high temperatures (Т > 4000 К), is significantly higher than the corresponding equilibrium
NO content in air.
The difference between experimental and calculated concentrations of nitrogen
monoxide, obtained at air cooling from the temperatures of above 5000 К, is the known fact.
The explanations of this phenomenon were discussed actively in [32, 33]. According to one of
these explanations [32], the additional production of NO occurs at quenching of the high-
temperature gas (>4000÷5000 К) due to divergence of oscillating and translation
temperatures at the initial moment of cooling. This phenomenon is not discussed in the
current work, but an attempt to explain a high content of NO in the peripheral gas at
temperatures below 2000 К is emphasized.
Dependence of NO concentrations on the mean mass temperature of gas at the outlet of a
transformer plasmatron (probe No. 3) is shown in Fig. 21(a) by dots. The values of NO
concentrations, obtained by the authors of [30, 34, 35] vs. the mean mass temperature of gas
before the quenching devices are also shown there by the shaded area. It is necessary to note
that in experiments of these authors NO was synthesized in two stages: in the first stage the
air in plasmatron was heated up to some certain temperature, then high-temperature air was
quenched (the cooling rate was ~ 106÷107 К/s). The calculated value of NO, reached at
cooling of thermodynamically equilibrium air with experimentally determined initial
temperature, is shown by the corresponding curve.
7 1
6
3
5
NO, vol. %
1 2
0
2000 3000 4000
Tenperature, K
Figure 21.a. Concentration of nitrogen monoxide at the heat exchanger outlet (probe No. 3) depending
on the mean mass temperature of gas at the heat exchanger inlet.1 – experimental data (probe No. 3).2 –
calculated value of NO concentration at the equilibrium state of air before cooling in heat exchanger;3 –
the area of experimental data of different authors [30, 34, 35], obtained at the use of equilibrium gas
quenching.
7,0 1
6,5
6,0
5,5
NO, vol. %
5,0
4,5 2
4,0
3,5
3,0
2,5
0 2 4 6 8 10 12 14 16
Air flow rate, g/s.
Figure 21.b. Concentration of nitrogen monoxide at the heat exchanger outlet (probe No. 3) depending
on the air flow rate. 1 – experimental data (probe No. 3).2 – calculation by the model.
Dependence of NO concentration on the air flow rate, measured at plasmatron outlet

(probe No. 3), is shown in Fig. 21.b by dots. According to the diagram, at low flow rates (1÷5
g/sс) NO concentration increases almost linearly. With the following increase in the air flow,
NO concentration depends on the flow rate weakly and tends to ~ 7%.
Thus, a high (superequilibrium) content of nitrogen monoxide at the gas outlet from the
plasmatron was determined experimentally at synthesis of nitrogen monoxide in the
transformer coupled toroidal discharge at vortex arc stabilization without application of the
quenching devices, which provide high rates of gas cooling. After measurements of plasma
temperature and NO concentration profiles, it was determined that in the low-temperature
peripheral zone of a vortex flow, which stabilizes the transformer discharge, the NO content
exceeds significantly the thermodynamically equilibrium values.
Let’s estimate the typical times of molecular diffusion under the considered conditions
and show that the diffusion processes cannot provide significant concentration of NO at the
periphery of discharge chamber.
We consider a cylindrical chamber with radial temperature distribution, determined
experimentally. The typical time of diffusion can be estimated as:
τ diff =
[r (T1 ) − r (T2 )]2 (22)
D
As initial temperature Т1 we will take the temperature corresponding to the maximal

thermodynamic-equilibrium concentration of nitrogen monoxide (Т1 ~ 3500 К), and as final
temperature Т2 we will take the temperature, when product stability is provided by a low
decomposition rate of NO (Тк ~ 2000 К). It follows from experimental temperature
distributions (Fig. 15a) that r(T1)-r(T2) ~ 1 cm. Taking into account that in the considered
temperature range (3500÷2000 К) coefficient of diffusion is D ~ 50÷20 cm2/s, we obtain
τdiff≈0.02÷0.05 s (23)
For comparison we should indicate that time τ of nitrogen monoxide decomposition in

the considered temperature range under the atmospheric pressure is [35]:
T, К 3500 3000 2300 2000

τ, s ~10-6 8·10-5 5·10-3 1
Hence, at diffusion of NO molecules from the high-temperature zone (Т ~ 3500 К with

equilibrium NO content of ~ 5%) to the zone with a lower temperature (to 2000 К), in every
considered point gas will be in the state close to thermodynamic equilibrium. Within gas
layers with the temperature below 2000 К, the time of NO decomposition is comparable or
exceeds the typical times of diffusion, and concentration of NO there can correspond to the
equilibrium value for Т ~ 2000 К (NO ~ 0.5 %). Therefore, at the chamber periphery, where
the temperature is below 2000 К, nitrogen monoxide concentration because of diffusion
should be, at its best, about 0.5%.
To explain a high content of NO at Т<2000 К in the transformer coupled toroidal
discharge, the authors of the current work have suggested the model of “self-quenching” [7].
The term “self-quenching” should be understood as the process of fast cooling of turbulent
“hot” formations in peripheral “cold” gas due to turbulent transfer at the boundary between
the forced and free vortices. The scheme of this process is shown in Fig. 22. In the framework
of this model, the presence of nitrogen monoxide in the peripheral low-temperature zone of
the discharge chamber is a result of cooling of high-temperature turbulent formations within
the peripheral gas layers with the rate, sufficient for high concentration of NO.
Figure 22. The scheme of processes in discharge chamber, providing “self-quenching”: 1 –

“transformer arc”; 2 – wall of discharge chamber; Dt – arc thermal size.L – size of turbulent formations;
Q – heat flux to the wall.
Results of calculations by the suggested model [7] are shown in Fig. 21.b by the solid
line. It is obvious that the behavior of curves is qualitatively the same, but their absolute
values differ. Considering a possibility to obtain superequilibrium concentrations of NO at
cooling of the high-temperature air [31], it can be assumed that the suggested model explains
satisfactory the found effect.
Therefore, it is shown at the example of synthesis of nitrogen monoxide from air that
plasmachemical reactions, whose products are stable at the temperatures below 2000 К,
distribution and final yield of the end product can be significantly effected by non-isothermal
character and turbulent characteristics of a vortex flow in the discharge chamber of the low-
frequency induction transformer plasmatron.
2.5. Natural Gas Processing in Plasma of Transformer Coupled Toroidal

Discharge
Recently, due to intensive development of fuel cell technologies and chemical industry,
there is a great interest to hydrogen (or syngas) production from natural gas. The real
industrial technologies require products consumption of tens of thousand of m3/h and, hence,
the setup powers of tens of megawatts. An “appropriate” mean for generation of such flows of
low-temperature plasma can be the transformer plasmatron, which has no principle limitations
for the setup power. Thus, one of the purposes of the current research was the experimental
modeling of real industrial processes at the existing setup.
The plasmachemical technology of syngas (СО + Н2) production from СН4 requires
binding of free carbon both for formation of CO molecule and undesirable product removal
(soot). Therefore, synthesis should be carried out in a presence of atomic oxygen, which can
be taken from the oxygen-bearing compounds. Carbon dioxide can be used as such a
compound.
Experiments were carried out at setup No. 1, shown in Fig. 6. Argon was fed into the
main feed chamber. The mixture of СН4 and СО2 with given stoichiometry was fed into the
additional feed chamber. This experimental scheme allowed us to maintain a stable discharge
in the gas mixture with high contents of СО2 and СН4 at a separate part of discharge. The gas,
reacted in the discharge chamber, was cooled in the heat exchanger (see Fig. 5a). Sampling
was made after gas cooling in the heat exchanger (probe No. 3). Product concentration in a
sample was analyzed by the mass-spectrometer.
Formation of free carbon is possible in reaction CH4 + CO2 Æ 2H2 + 2CO at a lack of
carbon dioxide. To get information about free carbon (soot) in the reaction products, the gas
sampling line included the filters, where carbon particles were deposited.
The absolute flow rates of СН4 and СО2 at the discharge chamber inlet varied from 0.25
to 0.6 m3/h and from 0.15 to 0.47 m3/h, respectively. According to mass-spectrometry results,
under all studied conditions the degree of natural gas conversion in the discharge was about
100%. However, absolute concentrations of СО and Н2 changed depending on natural gas
concentration in the initial mixture.
Experimental results are shown as the diagrams in Fig. 23. The stoichiometric
composition of supplied mixture (CH4/CO2) is plotted on axis x, and the ratio of outlet
volumetric flow of a substance (Н2, СО or С2Н2) to the inlet volumetric flow of СО2 + СН4 is
plotted on axis y. The “pictures” of filter surfaces are shown by circles in the same diagram.
The darker the filter surface, the higher the content of free carbon in a sample. According to
the diagram, with a rise of methane content relative to carbon dioxide, the contents of
hydrogen and soot increase. At this, the content of carbon monoxide decreases. It should be
also noted that there is acetylene in the reaction products.
According to performed experiments on investigation of natural gas conversion into
syngas, it is promising to use the transformer plasmatrons for processing of carbon-bearing
substances.
H2
2,0
1,8
1,6
1,4
1,2
1,0
ηi
0,8
CO
0,6
0,4
0,2 C2H2
0,0
0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8
stochiometric composition CH4/CO2
Figure 23. Dependence of ηi on stoichiometric composition of initial mixture (CH4 and CO2), where
ηi=Gi/(GCH4+GCO2) is the ratio of outlet volumetric flow of a substance (H2, CO or C2H2) to the inlet
volumetric flow rate of CO2+CH4. Circles demonstrate the darkness degree of filters in the gas-
sampling line.
3. TRANSFORMER LIGHT SOURCES

As it is known production of low–temperature plasma by induction (electrodeless) gas
discharges is one of the most promising methods to increase the efficiency and service life of
gas-discharge light sources. The urgency of this problem is obvious because of the wide
application of gas-discharge lamps in different fields of lighting technology (room and street
illumination, biomedical and technical applications, etc).
Several types of induction lamps are produced by industry; they differ both by inductor
and gas-discharge chamber constructions, and by frequency of the discharge current. The
radio-frequency (RF) induction discharge lamps (similar to QL induction lighting systems of
Philips production), and induction lamps, operating by the principle of transformer coupled
toroidal discharge (TCTD, similar to the Endura lamp, produced by Osram) are the most
common now.
It is necessary to note that electrodeless light sources (QL, Endura, Genura, Solara)
produced now are intended only for inner and outer illumination (induction luminescent
lamps). It is obvious that induction gas-discharge lamps can find wide application in other
fields, especially in biomedical technologies. Thus, new efficient and ‘long-living’ ultraviolet
(UV) emitters for water, air and surface disinfection can be developed on the basis of TCTD
in mercury vapor. At the same time, TCTDs in mercury vapor are less studied than
conventional arc discharges, and in particularly:
(a) Dependence of electrical and optical characteristics of TCTD on the main conditions
of discharge (discharge chamber diameter, pressure of mercury vapor, strength of discharge
current) is studied insufficiently, and this does not allow prediction of a change in TCTD
properties with variation of the above parameters, and correspondingly, it is impossible to
choose the optimal parameters of developed electrodeless UV emitters.
(b) Now a few theoretical models of TCTDs have been developed (e.g. [10, 11]);
however, a particular TCTD in mercury vapours, considered here, has been hardly studied.
Besides, it is not clear whether the available theoretical models of the arc discharges in
mercury vapors can be used for calculation of TCTD characteristics in mercury vapours. For
instance, in [19] the authors have calculated the characteristics of an experimental discharge
cell similar to Osram Sylvania’s electrodeless fluorescent lamp in the range of ‘high-loaded’
conditions covering current densities 0.05÷0.6A/cm2 and Hg vapor pressures 1.2÷13 mtorr
with application of two independent self-consistent numerical models of the positive column
(GLOMAC and GLOW).
The experimentally measured concentrations of excited and metastable atoms of mercury
are significantly lower than predictions of the numerical code GLOMAC for the high
discharge currents.
Together with mercury-argon TCTD, the studies of TCTD in neon are of a particular
interest because the new efficient light sources can be developed on this ground.
To create the new electrodeless light sources in the visible and UV spectral regions on the
basis of the TCTD principle, experimental investigations of the TCTDs were carried out in a
mixture of argon and mercury vapour for a wide range of discharge parameters: mercury
vapour pressure (0.1÷40 000 Pa), discharge current (1÷260 A), diameter of a gas-discharge
tube (20÷75 mm), and current frequency (10÷250 kHz). The TCTD was also studied in neon
within the range of neon pressures of 10÷600 Pa, diameters of gas-discharge tubes of 20÷58
mm, discharge currents of 1÷30 A, and current frequencies of 25÷250 kHz.
3.1. Experimental Setups
To study the transformer coupled transformer discharge in mercury vapours, the

experimental setup, shown in Fig. 24, was used. The main elements of this setup are as
follows: 1 – quartz toroidal gas–discharge chamber; 2 – system of ferromagnetic cores with
primary winding 3; 4 – high–frequency power source; 5 – accommodation unit between the
power source and TCTD. In the experiments we have used toroidal (in the form of elongate
torus) gas–discharge tubes, with the following geometrical dimensions: (a) diameter of gas–
discharge tube D = 20 mm, perimeter of gas–discharge tube along the middle line L = 210
mm; (b) D = 35 mm, L = 560 mm; (c) D = 58 mm, L = 880 mm; (d) D = 75 mm, L = 1880
mm. All gas-discharge chambers were filled with buffer gas (argon) under a pressure of 100
Pa and the strictly dosed amount of pure mercury or mercury amalgam with composition of
90%In – 5%Ag – 5%Hg; then, they were soldered. To measure the strength of discharge
current I, current transformer 6 was used. Simultaneously, density of discharge current J was
determined as a ratio between the measured strength of discharge current and the cross-
section of gas discharge chamber S: J = I/S.
To measure the voltage U that drives the TCTD, measurement coil 7, enveloping the gas-
discharge chamber over perimeter L was used. At that, the strength of electric field E was
determined as a ratio between measured voltage U and the discharge perimeter along the
middle line: E = U/L. The wall temperature of gas-discharge chamber T was controlled by
thermocouples 8, which allowed us to determine the mercury vapour pressures p in the
discharge chamber using the known dependence between the pressure of saturated mercury
vapour and the temperature.
To measure the optical and spectral characteristics, the following optic probes 9 were
used:
• Luxmeter, UV–radiometer TKA-PKM (model 06), intended for measuring

illumination (lx) within a visible range and UV–irradiation (mW/m2) within the
range 280 – 400 nm (A + B zones).
• UV-radiometer TKA-PKM (model 12), intended for measurements of UV–
irradiation within three ranges: UVA (315–400 nm), UVB (280–315 nm) and UVC
(200–280 nm).
• Setup for analysis of spectra on the basis of spectrometer AVASPEC–2048FT–2 and
PC, used for analytical investigation in a spectral range 200–800 nm with a very high
photometric sensitivity and optic resolution from 0.04 nm.
While measuring the optical characteristics of TCTD, diaphragming screen 10 with a

window of size h was installed in front of the studied object. The optical detectors were
located at distance l, which exceeded the size of a “window” by a factor of 10 and more, and
so we considered radiation of the outlined tube region as point one. In this case, power,
radiated into the studied range of the spectrum from the outlined part of the gas-discharge
chamber h is calculated as the product of the light flux density (measured by a device) on the
area of a sphere with radius l.
The developed experimental setup was used for determination of dependence of the
electric field strength, UV output, visible radiation and spectral characteristics of a discharge
on the discharge conditions: discharge current strength, pressure of mercury vapour and
diameter of the gas discharge chamber.
Figure 24. Scheme of experimental setup for investigation of TCTD in mercury vapours. 1 – quartz
gas–discharge chamber;2 – ferromagnetic cores; 3 – primary winding; 4 – power source; 5 –
accommodation unit; 6 – current transformer; 7 – measuring coil;8 – thermocouples; 9 – optic
probes;10 – diaphragming screen.
Figure 25. Scheme of experimental setup for investigation of TCTD in neon.1 – quartz gas–discharge
chamber;2 – magnetic core;3 – primary winding;4 – power source;5 – exhaust tubes;6 – gas line; 7 –
vacuum meter;8 – measuring coil; 9 – current transformer; 10 – optic probes;11 – vacuum pump.
To study electric-physical and radiation characteristics of the transformer coupled

toroidal discharges in neon, the experimental setup, shown in Fig. 25, was used. The diameter
of quartz gas-discharge tube 1 in different experiments varied from 20 to 58 mm; the cross-
section of magnetic core 2 – from 5 to 20 cm2; the number of coils in primary winding 3 –
from 1 to 5. To generate a discharge, power sources 4 with current frequency 25 and 250 kHz
were used.
Gas-discharge tube 1 was connected to gas line 6 via exhaust tubes 5, and this allowed
variations in the gas pressure during the experiment. The vacuum meter ILMVAC PIZA 111
7 was used for pressure measurements; it allowed measurements of pressure in the range from
0.1 Pa to 100 kPa with an error of not higher than 10 %. The discharge voltage and current
were measured by additional coil 8 and current transformer 9, respectively, with an error of
not higher than 3%. To register discharge radiation in the visible range of the spectrum, the
digital luxmeter ТКА–01/3 was used. To measure the total radiation flux of a discharge in the
visible and long-wavelength IR spectrum range, the IMO-2N in combination with a “water”
light filter (10) was used. The system was pumped by forevacuum pump 11.
The software–hardware systems for control of spectrometric measurements, developed at
Petrozavodsk State University, were used for spectrometric studies of plasma of an induction
transformer discharge: the software–hardware system “Light” and measuring bench for high-
resolution spectroscopy “Spectrum”.
The block-scheme of hardware system “Light” is shown in Fig. 26. The main units of this
system are: 1 – system of data gathering and control on the basis of modular system CAMAC
and PC; 2 – spectrometer DFS–12, operating by the principle of double monochromator; 3 –
photoelectronic multiplier, connected to the outlet slot of the monochromator; 4 – optic
illumination system (lens + deflecting mirror), forming an image of a light source on the
diaphragmatic inlet slot of the spectrometer; 5 – unit of spatial scanning (system of mirrors,
controlled by a step motor in the horizontal plane); 6 – test subject (a gas-discharge light
source).
Figure 26. Block-scheme of software-hardware system “Light”.1 – System of data gathering and
control on the basis of the modular system CAMAC and PC; 2 – spectrometer DFS–12; 3 –
photomultiplier; 4 – optic illumination system of spectrometer inlet slot; 5 – unit of spatial scanning; 6
– test subject (light source).
This setup allows registration for the spectra of spatially nonuniform emitting objects,
preservation of results on detailed information about experimental conditions, and analysis of
energy characteristics of radiation within the spectral range of 360–800 nm.
The test subject was an induction neon lamp, operating by the principle of induction
transformer discharge with the following parameters (the scheme of this lamp is shown in
Fig. 26): outer radius of torus of ~ 8 cm, inner radius of torus of ~ 4 cm, inner radius of gas-
discharge tube Rk ≈ 17.5 mm, perimeter of gas-discharge tube along the mean line
L=2πRT≈38 cm), and neon pressure inside gas-discharge tube of 1 mm Hg.
While measuring the spectral characteristics of a discharge, the current and voltage of
discharge glow were also measured (with an error of not higher than 3%), together with the
wall temperature of a gas-discharge tube (with an error of not higher than 1 ºС). When
measuring the spectral characteristics the induction lamp was put vertically, the radiance
spectra and profiles were measured in A–B cross-section of a gas-discharge tube. Point O on
the torus mean line was taken as a zero point. The measurement error for the absolute values
of spectral radiance was not higher than 5–10%.
The scheme of a measurement setup for high-resolution spectroscopy “Spectrum”,
intended for examination of spectral line circuits, is shown in Fig. 27. This setup was made by
the standard scheme of spectrometer 1 with Fabry–Perot interferometer 2. Studied gas–
discharge light source 3 is located within the focal plane of lens L1. Second objective L2 is
focused on plane Y of diaphragmatic inlet slot of the spectrograph. The outlet slot of
spectrograph is connected to photoelectronic multiplier 4, whose signal is sent to the ADC
inlet and registered by PC 5.
Since interferometer transmission for monochromatic wave λ, down coming at angle φ to
the mirror normal, is determined by expression
T2
P(λ , ϕ ) =
2d cos(ϕ ) ⎞
(1 − R ) + 4R sin
2 2⎛
⎜π
⎝ λ
⎟
⎠ , (24)
where T is mirror transmission coefficient, R is reflection coefficient, and d is distance

between the mirrors; the system of interference rings, described by expression (24), where
φ=r/F (r is radius, F is focal distance), is formed in plane Y. Distinguishing a light flux in the
central area (φ=0) by a low-radius diaphragm and varying linearly in time parameter d/λ, we
fix periodic signal P'(t), and period of this signal Δt corresponds to interferometer constant
Δλ=λ2/2d. For monochromatic radiation fixed signal P' corresponds to the interferometer
instrument function; if there is line broadening the registered signal is “convolution” of the
spectral line circuit and instrument function.
Parameter d/λ can be changed in the different manners; in the current case the
“deflection” over spectral range Δλ is made by a change in the air pressure in pressure
chamber 6 at fixed distance between mirrors d. For this purpose, the air is evacuated from the
pressure chamber by vacuum pump 7, before the measurements, and then air leak valve 8 was
opened. At the initial stage the pressure in the chamber increases almost linearly, and this
causes a linear increase in air refraction coefficient n. Respectively, the wavelength with
frequency ν changes also linearly: λ=c/nν.
In experiments the interferometer with basis d=10 mm was used, what corresponds to
operation range Δλ of about 0.1 А (for the visible range of spectrum). The coefficient of
mirror reflection for this interferometer was ~74 %. The measurement error for intensity of
distinguished spectral line did not exceed 2%. The discharge current and voltage were also
measured together with spectral characteristics with an error of not higher than 3%.
Figure 27. Scheme of measurement setup “Spectrum”.1 – Spectrograph; 2 – Fabry–Perot

interferometer, (Y– diaphragmatic inlet slot of spectrograph); 3 – light source; 4 – photomultiplier; 5 –
PC with ADC; 6 – pressure chamber; 7 – pump; 8 – leak valve;L1, L2 – lenses with focal distance F.
3.2 Transformer Coupled Toroidal Discharge in Mercury Vapour.
The dependence of electric field strength E on current density J is shown in Fig. 28 for
several levels of mercury vapor pressure p (tube diameter D = 75 mm, current frequency f =
10 kHz). According to Fig. 28, the voltage–current characteristic (VCC) of TCTD is similar
to the VCC of the DC arc discharges: there are a descending branch, area of minimal electric
field intensity and an ascending branch. The position of VCC minimum is determined by the
pressure of mercury vapour: in the range of vapour pressure of 1÷7 kPa, the VCC of TCTD
has no distinct minimum. With the following increase in pressure, the electric characteristics
of discharge change qualitatively: the VCC minimum shifts towards the area of lower
discharge currents and becomes more distinct. Thus, for a pressure of mercury vapor of ~11
kPa, the minimum of E is in the range of 0.65÷1 A/cm2, and for pressure of 35 kPa, it is
within 0.2÷0.3 A/cm2. Dependence E versus p is shown in Fig. 29, for different diameters of
the discharge tube (current density is 1 A/cm2). Dependence E(p) for a DC arc discharge is
shown for comparison (J = 0.35 A/cm2) [36]. According to Fig. 29, in the range of mercury
vapour pressure of 3÷30 Pa, the electric field strength of TCTD has a local minimum. With a
further increase in pressure, the electric field strength increases as E ~ p1/3. For the case of DC
arc discharges in mercury vapour, dependence E(p) takes a qualitatively similar form [36]. A
change in the discharge chamber diameter does not qualitatively effect the form of
dependence E(p), decreasing the field strength with a rise in diameter. Thus, with an increase
in the chamber diameter from 20 to 35 mm, the field strength decreases by a factor of 2.6, and
with an increase of up to 75 mm, it decreases by a factor of 3.2.
p Hg, kPa
35
2,5
22
11
2,0 7
E, V/cm
2
1,5 1
1,0
0,5
0 1 2 3 4 5 6 7
2
J, A/cm
Figure 28. Electric field strength E versus discharge current density J, for different values of mercury
vapour pressure (D = 75 mm, f = 10 kHz).
D f
20 mm, 250 kHz
35 mm, 250 kHz
75 mm, 10 kHz
32 mm [36]
1
E, V/cm
0,1
-1 0 1 2 3 4 5
10 10 10 10 10 10 10
p Hg, Pa
Figure 29. Electric field strength E versus mercury vapour pressure p, for different diameters of
chambers (J = 1 A/cm2).
For the DC arc discharges in the long cylindrical tubes with specific discharge power of
20÷50 W/cm and the middle or high pressure of mercury vapour, interaction between electric
field strength, diameter of the gas–discharge tube and specific mass of evaporated mercury m
is presented by the known dependence, derived by Elenbaas [37]:
ED3 / 2 ≈ 370m7 / 12 (25)
where value E is presented in V/cm, D is in millimeters, and the specific mass of evaporated
mercury m is in mg/cm. The dependence of E · D3/2 on specific mass of evaporated mercury m
(mg/cm) is shown in Fig. 30 for the fixed specific power of TCTD of P = 40 W/cm. Data
shown in Fig. 30 corresponds to the range of mercury vapour pressures of 10÷40 kPa, and the
tube diameter of 35÷75 mm. The similar dependence, obtained by Elenbaas for the mercury
DC arc discharges in the long cylindrical tubes with diameters 12÷38 mm and specific
amounts of mercury 0.3÷50 mg/cm (which corresponds to a pressure of mercury vapour of
17÷300 kPa) for the specific discharge power of 30 W/cm are also shown in Fig. 30 [37].
As it is seen from Fig 30, the experimental data obtained for the TCTD are laid perfectly
on the line, described by expression (25). This proves the fact that, at least, in the range of
mercury vapour pressure of 10÷40 kPa and diameters of gas-discharge tube of 35÷75 mm, the
electrical characteristics of TCTD are similar to those of the DC arc discharges.
The emission yield to non-resonance lines of the visible and UV–spectra as a function of
TCTD specific power is plotted in Figs. 31 and 32 (as a percentage of input power) for
mercury vapour pressure of p ≈ 25 kPa and tube diameter of 75mm (f = 10 kHz).
TCTD DC arc discharge, [37]

10000 D=35 mm, f=250 kHz D=12 mm
D=58 mm, f=25 kHz D=18 mm
D=75 mm, f=10 kHz D=28 mm
D= 38 mm
3/2
E·D , V/cm·mm
3/2 1000
100
0,1 1 10 100
m Hg, mg/cm
Figure 30. Interaction between electric field strength E, diameter of gas–discharge chamber D and
specific amount of evaporated mercury m, for TCTD and DC arc discharges.
404.7 nm 435.8 nm
546.1 nm 577.0/579.1 nm
2,5
Emission yield, %
2,0
1,5
1,0
0,5
0,0
0 100 200 300 400 500 600
Specific power P, W/cm.
Figure 31. Dependence of emission yield to non-resonance lines of the visible spectrum range (400–600
nm) on specific discharge power. D = 75 mm, p ≈25 kPa, f = 10 kHz.
Analysis of data shown in Figs. 31 and 32 demonstrates that dependence of emission

yield to the spectral line on the specific discharge power is determined by its excitation
potential. Thus, the efficiency of the visible triplet 404.7, 435.8 and 546.1 nm with excitation
potential of 7.73 eV reaches its maximal value with a specific discharge power of ~ 50W/cm,
and with a further increase in power, decreases significantly. The feature of this triplet is the
lowest excitation potential (7.73 eV) after the resonance lines (4.88 eV for line 253.7 nm). A
decrease in the efficiency of the lines of the UV and visible spectrum ranges with excitation
potential of 8.85 eV (312.6/313.2, 365.0/366.3, 577.0/579.0 nm) expressed weaker by an
order for the vapour pressure of ~25 kPa. A decrease in emission yield with a rise of
discharge power was not determined for the lines with excitation potential of 9.56 eV (265.2,
302.8/303.6 nm).
265.2 nm 302.8/303.6 nm
312.6/313.2 nm 365.0/366.3 nm
1,5
Emission yield, % 1,0
0,5
0,0
0 100 200 300 400 500 600
Specific power P, W/cm
Figure 32. Dependence of emission yield to non–resonance lines of the UV spectrum range (260–400
nm) on specific discharge power. D = 75 mm, p ≈25 kPa, f =10 kHz.
D=20 mm, I=10 A, f=250 kHz

80 D=35 mm, I=10 A, f=250 kHz
D=75 mm, I=50 A, f=10 kHz
70 D=27 mm, I=4 A [37]
light efficiency, lm/W
60
50
40
30
20
10
0
-1 0 1 2 3 4 5
10 10 10 10 10 10 10
p Hg, Pa
Figure 33. Light efficiency versus mercury vapour pressure for the fixed TCTD current.
A decrease in the light efficiency of TCTD was mentioned in [20] for specific discharge
power of above 80 W/cm and pressure of mercury vapour of ~ 10 kPa; however, there was no
detailed information on energy redistribution in the radiation spectrum. According to
published data [38], the light efficiency of mercury arc discharge should increase
continuously with a rise of power (for the fixed pressure), tending to its limit value (~80
lm/W).
To explain the observed effect, the authors of this paper offer the following hypothesis:
since in gas discharges of the middle and high pressure, the population with excitation
potential Uk is determined as Nk ~ e−Uk/kT, and the temperature of plasma increases with a rise
in discharge power, the population of excited levels should be redistributed in favour of

highly excited states, and correspondingly, the emission yield in the close IR zone of
spectrum (1000–1700 nm) should also increase in the line with excitation potential of 9.5÷10
eV. Unfortunately, the used measurement system did not allow registration of the IR
spectrum of discharge, and this assumption could not be checked experimentally.
The dependence of TCTD light efficiency on the pressure of mercury vapour is shown in
Fig. 33 for the fixed discharge current and different diameters of the gas–discharge tube.
Dependence of DC arc discharge light efficiency is shown in Fig. 33 for comparison (I = 4A,
D = 27 mm, [37]). According to Fig. 33, the dependence of light efficiency of TCTD and DC
arc discharge on mercury vapour pressure are qualitatively similar: the first local maximum of
light efficiency is reached for a pressure of ~10 Pa, and makes up ~20÷30 lm/W; the local
minimum of light efficiency corresponds to the range of mercury vapour pressures of
100÷1000 Pa, and with a further increase in pressure, the light efficiency starts increasing
fast. It can be also seen in Fig. 33 that with an increase in the diameter of the gas-discharge
tube, the light efficiency of TCTD increases; the similar dependence is observed for the DC
arc discharges [38].
The dependence of the emission yield to non-resonance lines of the UV spectrum range
(280–400 nm) on the pressure of mercury vapour is shown in Fig. 34 for the fixed discharge
current. According to the analysis of the results, shown in Figs. 33 and 34, the efficiency of
non–resonance lines of the UV and visible ranges of the TCTD radiation spectrum similarly
depend on the pressure of mercury vapour and the diameter of the gas–discharge tube.
D=35 mm, I=10 A, f=250 kHz

Efficiency in UV range 280-400 nm, %
D=20 mm, I=10 A, f=250 kHz

8
7
6
5
4
3
2
1
0
-1 0 1 2 3 4
10 10 10 10 10 10
p Hg, Pa
Figure 34. Emission yield to non-resonance lines of the UV-range 280–400 nm versus mercury vapour
pressure for the fixed TCTD current.
Emission yield to the line 253.7 nm, %

D=20 mm, I=10 A, f=250 kHz
70 D=35 mm, I=3 A, f=250 kHz
60 D=35 mm, I=7 A, f=250 kHz
D=75 mm, I=50 A, f=10 kHz
50
40
30
20
10
0
-1 0 1 2 3 4
10 10 10 10 10 10
p Hg, Pa
Figure 35. Emission yield to resonance line 253.7 nm versus pressure of mercury vapour for fixed
current of TCTD.
The dependence of the emission yield to resonance line of 253.7 nm on the mercury
vapour pressure is shown in Fig. 35 for the fixed discharge current. As it is clearly seen, the
range of pressures of ~0.5÷1.5 Pa (this corresponds to the temperature of the cold spot of
30÷50 ◦C) is of a particular interest; up to 60% of electric power put into the resonance line
there, and therefore, the mercury discharges of a low pressure are the highly efficient sources
of UV radiation, which can be used in practice.
Since the maximal radiation yield to resonance line 253.7 nm can be obtained in a very
narrow temperature range (30÷50 ◦C), a series of experiments was carried out using a three-
component mercury amalgam of the following composition: 90%In–5%Ag–5%Hg (Fig. 36).
Amalgam Pure Hg
Emission yield to line 253.7 nm, %
70 I=3 A I=3 A
I=10 A I=10 A
60
50
40
30
20
10
0
0 20 40 60 80 100 120
Tube temperature, C
Figure 36. Emission yield to line 253.7 nm versus wall temperature for the fixed current of TCTD
(amalgam and pure mercury filling). Diameter of gas-discharge tube D = 35 mm, f = 150 kHz.
Figure 37. Electrodeless mercury high intensity discharge lamp of 3-kW power.D=60 mm, f=25 kHz.
Figure 38. Electrodeless mercury high intensity discharge lamp of 100-kW power.D=80 mm, f=10 kHz.
Figure 39. Electrodeless mercury UV lamps of 100, 200 and 50 W.F = 150 kHz, D = 35÷40 mm.
The dependence of radiation yield to the resonance line on the temperature of a gas–
discharge chamber wall is shown in Fig. 36 for the TCTD with amalgam and pure mercury
filling. As it can be seen from the figure, using mercury amalgam, the maximum of resonance
radiation yield moves to the area of the higher temperatures (>70 ◦C), and the temperature
range corresponding to the maximal efficiency of resonance line 253.7 nm becomes
significantly wider (70÷120 ◦C).
The unique samples of mercury high-intensity discharge lamps of the 3÷100-kW power
with efficiency of ~60 lm/W were developed on the basis of investigations of the TCTD in
mercury vapors (Fig. 37, 38).
On the basis of experimental studies of the TCTD with amalgam filling, the pilot samples
of bactericidal induction UV lamps with the power of 50, 100, 200 and 500W and current
frequency f = 150 kHz (Fig. 39) were constructed; their emission yield to line 253.7 nm is
~30÷35% of consumed electric power.
90
1,0 120 60
0,8
0,6 150 30 1
0,4 2
0,2 3
0,0 180 0 4
0,2 5
0,4 6
0,6 210 330 7
0,8
1,0 240 300
270
Figure 40. Corner diagram of directivity of the 50Wlamp.1–253.7 nm, 2–312.6/313.2 nm, 3–
365.0/366.3 nm, 4–404.7 nm, 5–435.8 nm, 6–546.1 nm, 7–577.0/9.0 nm.
90
1
1,0 120 60
2
0,8
3
0,6 150 30
4
0,4 5
0,2 6
0,0 180 0
0,2
0,4
0,6 210 330
0,8
1,0 240 300
270
Figure 41. Corner diagram of directivity of the 100W lamp. 1–253.7 nm, 2–312.6/313.2 nm, 3–
365.0/366.3 nm, 4–404.7 nm, 5–435.8 nm, 6–546.1 nm.
The measured corner diagrams of emission directivity of different spectral lines are
shown in Figs. 40 and 41 for developed lamps with the power of 50 and 100 W (in the
reduced form, the lamp position is schematically shown in the figure (vertical one), the axis
of spectral characteristic measurements passes the centre of the lamp). According to Figs. 40
and 41, the corner diagram of radiation directivity of resonance line 253.7 nm differs
maximally from the round one, what is possibly caused both by its intensive absorption in
discharge plasma and partial absorption by the chamber walls.
Before discussing the experimental results, first of all, it is necessary to emphasize the
fact that in contrast to RF induction discharges, the TCTD can be considered as the analogue
of an arc discharge with electrodes, combined in one plane [10]. Therefore, it is interesting to
apply the standard models of arc discharges for the description of TCTD. However, it should
be noted that the most standard models of arc discharges are developed for description of DC
discharges in long cylindrical tubes. As a result, the question arises: whether these models are
applicable for TCTD description. Nevertheless, it can be assumed that for some cases, the
standard models of axisymmetric DC arc discharges will satisfactorily describe the TCTD
characteristics.
Let us consider the main conditions, when the TCTD can be described by the standard
models:
(a.) A small curvature radius of the toroidal gas-discharge chamber. Apparently, in this
case deviations from cylindrical symmetry in a discharge are minimal, and the
discharge itself can be shown as a discharge in a straight cylindrical tube with length
L, where L is the perimeter of the gas-discharge chamber over its middle line.
(b.) Fulfillment of condition f · τ > 1, where f is the current frequency and τ is the typical
time of plasma “damping”. In this case, the parameters of discharge plasma
(conductivity, electron temperature, concentration of excited and metastable atoms)
do not change during discharge current passing through zero, and the TCTD can be
considered and used as the analogue of a DC discharge.
(c.) An absence of the skin-effect on the electrical characteristics of a discharge.
The analysis of experimental data shown in Figs. 33–35 demonstrates that dependences
of the optical characteristics of TCTD on the mercury vapour pressure, discharge current
strength and diameter of the gas-discharge tube correspond qualitatively to similar
dependences for the arc mercury discharges (excluding the abovementioned effect of the
luminous efficiency reduction to a visible triplet with an excitation potential of 7.73 eV with
an increase in discharge power). However, for some cases, there are significant quantitative
differences between the radiation characteristics of TCTD and arc mercury discharges, which
are possibly caused by different “glow” conditions. Thus, even for the pressure of ~ 20 kPa,
the light efficiency of TCTD with D = 75mm is ~ 70 lm/W (Fig. 33), what is relatively close
to the known limit of light efficiency of the arc mercury discharges with a tube diameter of
20÷30 mm ( ~ 80÷85 lm/W), obtained if the pressure of mercury vapour equals several
atmospheres.
The authors of the current paper consider the possibility of using the standard “channel”
model of axisymmetric DC arc discharges [37], based on the assumption of LTE presence in
discharge plasma, for the calculation of the TCTD parameters in mercury vapour. According
to this model, electric field strength E in a mercury discharge is determined by expressions
C1 p 7 / 12 P1 / 2
E= (26)
(P − PT )1 / 3 D1 / 3
Or
C2m7 / 12 P1 / 2
E= (27)
(P − PT )1 / 3 D3 / 2
where C1, C2 are constants, PT is the value of specific heat losses of discharge, P is specific
discharge power, m is the specific amount of evaporated mercury and D is the tube diameter.
Additionally, it is assumed by this model that PT does not depend on the discharge power and
diameter of the discharge tube. In this case, minimal E is reached, if the following condition
is satisfied:
P = 3PT (28)
The analysis of TCTD voltage–current characteristic (Fig. 28) with criterion (28)
provides the following values of specific heat losses: PT = 16 W/cm for p ≈ 10 kPa; 7.9 W/cm
for p ≈ 22 kPa and 7.5 W/cm for p ≈ 35 kPa, what coincides with the results of Elenbaas [37]
(PT = 9÷10 W/cm), obtained similarly for the arc mercury lamps of high pressure in the
following range of discharge parameters: P 15÷60 W/cm, D 12÷38 mm. For the pressures of
mercury vapour below 10 kPa, the VCC of TCTD has no the expressed minimum, what
proves the fact that formulae (26) and (27) can have high errors for calculation of electrical
characteristics of the TCTD. According to formula (26), field strength E for the fixed pressure
of mercury vapour depends on the tube diameter as E ~ D−1/3. The analysis of data in Fig. 29
demonstrates that an increase in the tube diameter from 35 to 75 mm really provides the
reduction of the field strength by a factor of 1.3, whereas for the TCTD with a tube diameter
of 20 mm, the experiment gives an excessive value of the field strength.
However, it should be noted that in this experiment the ratio of perimeter and diameter
L/D for the toroidal chamber was ~10, whereas, for the lamps with the tube diameter of
35÷75 mm, parameter L/D varied from 15 to 25. It can be assumed that for L/D ~ 10, the
condition of axial symmetry is disturbed considerably (visually, the discharge was
significantly “pressed” into the inner wall of the toroidal chamber), therefore, the standard
models of axisymmetric gas discharges cannot be applied for this situation. As it can be seen
from Fig. 30, for the TCTD with the chamber diameter of 35÷75 mm and fixed discharge
power P ≈ 40W/cm, product E · D1/3 can be perfectly described by formula (25) with
constants, obtained by Elenbaas via the analysis of characteristics of the arc axisymmetric
mercury discharges. Therefore, for the pressures of mercury vapor above 10 kPa and ratio
L/D > 15, the TCTD in mercury vapour can be approximately considered as the
axisymmetric, and the “channel” model of axisymmetric discharges can be used for
calculation and analysis of discharge parameters.
3.2. Transformer Coupled Toroidal Discharge in Neon
Dependence of electric field strength of the TCTD on neon pressure is shown in Fig. 42.
Dependence of electric field strength on discharge current is shown in Fig. 43.
D, mm; I, A
2,0 20 ; 1 [39]
1,8 20 ; 5 [39]
1,6
20 ; 10
1,4
E, V/cm
1,2 35 ; 1
1,0 35 ; 8
0,8 51 ; 1 [40]
0,6 51 ; 10 [40]
0,4 58 ; 20
0,2
10 100 100061 ; 10 [39]
p Ne, Pa
Figure 42. TCTD electric field strength vs. neon pressure.[39] – Published data, DC arc discharge.[40]
– Published data, TCTD at current frequency of 450 kHz.
1,1 35 mm; 25 kHz

1,0 35 mm; 250 kHz
0,9 61 mm; [39]
0,8
E, V/cm
0,7
0,6
0,5
0,4
0,3
0,2
1 10
I, А
Figure 43. TCTD electric field strength vs. discharge current. Neon pressure of 1 torr.
According to analysis of data, shown in Figs. 42, 43, dependences E(p, D, I), observed
for the TCTD in neon coincide qualitatively with similar dependences for the positive column
of a gas discharge in neon. However, it was determined that the field strength increases with a
rise of current frequency (Fig. 43).
To analyze the observed increase in the field strength with a rise of current frequency,
let’s estimate the influence of skin-effect on electric-physical characteristics of a discharge.
To simplify calculations, we will represent the studied discharge as a uniform cylindrical
conductor with radius r, with specific electric conductivity σ. In this case efficient depth of
skin layer δ is calculated by formula
2
δ= (29)
2 μ0ωσ
where μ0 is magnetic constant, σ is conductivity of TCTD plasma. Plasma conductivity,

averaged by the discharge cross-section, can be determined from relationship j=σE.
Dependence of skin layer thickness for TCTD in neon on discharge current strength is
shown in Fig. 44. Conductivity is calculated by the voltage-current characteristic, shown in
Fig. 43. According to the figure, the minimal thickness of the skin layer almost twice exceeds
the radius of a gas-discharge tube. In this case, the influence of skin-effect on electric-kinetic
characteristics of a discharge can be described by approximated formula
Rω/R0=1+k4/3 (at k<1) (30)
where Rω is efficient resistance of a “conductor” with radius r to alternating current, R0 is

resistance to direct current, k=r/2δ. It follows from formula (30) that to increase the field
strength by 10% via the skin-effect, the value of k should be ~0.75, what corresponds to the
skin layer thickness of ~1.2 cm. Thus, we can make a conclusion that the observed increase in
the field strength with a rise of current frequency cannot be explained by the skin-effect.
100
25 kHz
250 kHz
10
δ, cm
1
1 10
I, А
Figure 44. Calculated thickness of skin-layer δ vs. strength and frequency of discharge current (p Ne=1
mmHg, D=35 mm).
Dependences of luminous efficiency of TCTD in neon on neon pressure, discharge

current and diameter of the gas-discharge tube are shown in Figs. 45 and 46. Dependence of
specific light flux of TCTD (radiated from 1 cm of the gas-discharge tube) on discharge
current strength in visible and long-wave IR spectral region (400–1200 nm) is shown in Fig.
47. It can be seen in Fig. 45, the maximal luminous efficiency of TCTD is reached within the
pressure range of 100÷200 Pa. According to publications [39], the maximal luminous
efficiency of an arc discharge in neon corresponds to the pressures of 30÷50 and 30÷100 Pa
for tube diameters of 60 and 20 mm, respectively.
D, mm I, A
20 5 А, [39]
35 20 10 А
35 10 А
Luminous efficiency Lm/W

30
58 20 А
25 61 20 А, [39]
20
15
10
0
10 100 1000
p Ne, Pa
Figure 45. TCTD luminous efficiency vs. neon pressure.
Observed divergence can be caused by different experimental conditions: in paper of

Klarfeld [39] electric-physical and radiation characteristics of a discharge were measured in
the sealed gas-discharge tubes, whereas, data in Fig. 45 was obtained at the constant neon
flow through a gas-discharge tube.
D, mm f, kHz
35 25
40 35 250
61 [39]
Luminous efficiency, Lm/W
35
30
25
20
15
1 10
I, А
Figure 46. Luminous efficiency vs. TCTD current strength.
1,6
25 kHz
1,4 250 kHz
specific light flux, W/cm
1,2
1,0
0,8
0,6
0,4
0,2
0,0
0 5 10 15 20 25
I, A
Figure 47. Specific light flux (W/cm) of the TCTD in visible and near IR radiation spectral ranges
(400–1200 nm) vs. discharge current.D=35 mm, p=1 torr.
For the fixed neon pressure (1 torr) and diameter of the gas-discharge chamber D=35 mm
the maximal luminous efficiency of the TCTD is reached at the discharge current of 3÷5 А,
and it makes up ~34÷36 lm/W (Fig. 46), what is close to the maximal value for the arc
discharges in neon (~42 lm/W [39]), and exceeds significantly the efficiency of industrial arc
neon gas-discharge lamps (~10÷15 lm/W) [38]. It should be noted specially that an increase
in current frequency increases the light flux, radiated by the TCTD in visible and long-wave
IR spectral regions (Fig. 47).
According to Figs. 43 and 47, with a rise of current frequency the specific light flux of
TCTD and electric field strength increase similarly, therefore, the luminous efficiency of the
TCTD (Fig. 46), proportional to TCTD efficiency η=F/(IE), stays constant.
1,1 640.2 nm
540.0 nm
1,0
580.4 nm
0,9
Normalized intensity
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0,0
-1,0 -0,8 -0,6 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0
x
Figure 48. Radial distribution of brightness of neon spectral lines. TCTD, p = 1 torr, D = 35 mm, J = 1
А/cm2, L/D = 11.
5
2p 3p x=0
10
8 x=0.85
-3
7
10
nk(x)/gk, cm
5
2p 4d
6
10
5
10
18,0 18,5 19,0 19,5 20,0 20,5 21,0 21,5
Uk, eV
Figure 49. Normalized concentration of excited neon atoms nk(x)/gk vs. excitation potential of level Uk
TCTD, p=1 mm Hg, D=35 mm, J=1 А/cm2.
The spectral characteristics of TCTD were firstly measured in neon: radial distributions
of spectral line brightness and optic spectra of discharge in different areas of a gas discharge.
The shape of spectral line circuits was studied. Experimental investigations were carried out
for the optimal conditions of discharge glow (p=1 torr, D=35 mm, I=5÷10 А). Gas–discharge
tube perimeter L was 38 cm, thus, for the considered case L/D ≈ 11.
Radial distributions of brightness F(x) for different neon lines, measured in A–B tube
cross-section (Fig. 26), are shown in Fig. 48 in the normalized form. The brightness profiles
of radiation transition lines 2p53p–2p53s and 2p54d–2p53p were studied.
According to analysis of data, shown in Fig. 48, radial distributions of brightness are
asymmetrical. The main reason for increased energy release inside the toroidal gas-discharge
tube is a change in the value of longitudinal electric field E(x) with a change in coordinate x.
Thus, it follows from equation (5) that
E(x)=E/(1 - πxD/L) (31)
In particular, for the considered case E(0.8)/E(-0.8)≈1.6, at that Fλ(0.8)/Fλ(-0.8)≈1.7.

Thus, spectral measurements confirm the above assumption that at L/D ~ 10 TCTD is not the
axisymmetric discharge.
Since there is no radial symmetry, we can use the standard Abelian transformation for
determination of spatial distribution of radiation intensity and, hence, distribution of excited
atom concentration. Nevertheless, while analyzing the spectra of neon induction discharge,
measured at different points of tube x, we can calculate concentrations of excited atoms
nk(x)=Nk(x)/l(x), averaged by cross-section l(x) (Fig. 26), where Nk(x)~Fλ(x). Results of
calculations (normalized to statistical weights gk of corresponding excited states k) are shown
in Fig. 49.
According to analysis of data in Fig. 49, the population character of excited states 2p53p
and 2p54d,5s at transition from the central to peripheral regions of the TCTD in neon does not
change qualitatively, only quantitative changes are observed. This proves the fact that the
electron temperature is the same both in the central and peripheral regions of the TCTD.
There was a series of experiments on determination of circuit shapes for neon spectral
lines within the range of 585–700 nm (2p53p–2p53s radiation transitions). These experiments
used the measurement setup of high-resolution spectroscopy “Spectrum”, assembled by the
standard scheme of Fabry-Perot interferometer with a spectrograph (Fig. 27).
According to analysis of TCTD generation conditions in neon, the main mechanisms,
determining the broadening of observed spectral lines, should be the following:
1. Doppler broadening.
2. Line broadening because of self-absorption (re-adsorption) of radiation. According to
the estimates performed, contribution of other mechanisms of broadening can be neglected.
The shape of spectral line circuit with wavelength λ0, widened by the Doppler effect, is
described by expression:
c M ⎛ Mc 2 (λ − λ0 ) 2 ⎞
ϕ (λ , λ0 ) = EXP⎜⎜ − ⎟
⎟
λ0 2πRT ⎝ 2 RT λ20 ⎠, (32)
where λ0 is wavelength, corresponding to the line center, с is light speed, M is molar mass
of neon, R is gas constant, and T gas temperature.
At self-absorption, spectral radiance of the source surface b(λ,λ0) is determined by

expression (in approximation of a uniform plasma layer with thickness y0)
b(λ , λ0 ) = B(1 − exp(−k (λ , λ0 ) y0 ) , k(λ,λ0) ≈(gj/gk)λ04Ajk(8πc)-1Nkφ(λ,λ0) (33)
Thus, due to analysis of the circuit shapes for neon spectral lines (under the considered
conditions of discharge glow), we can determine the following parameters of TCTD plasma:
gas temperature T and population of four lower excited states 2p53s, two out of which are
metastable.
According to analysis performed, for the studied conditions of TCTD glow the average
gas temperature is ~700–800 К, and total atomic concentration on 2p53s levels (4 levels, 2 out
of them are metastable) is ~(4–8)٠1011 cm-3.
The pilot samples of neon induction lamps with the power of 100÷2000 W and efficiency
of 20÷30 lm/W were developed on the basis of investigations performed (fig. 50, 51).
Figure 50. Induction neon lamp of 500-W power.
Figure 51. Induction “demonstration” lamp of 2 000-W power.

CONCLUSIONS
The integrated experimental investigations of transformer coupled toroidal discharges,
aimed at development of new electrodeless generators of low-temperature plasma,
plasmachemical reactors and gas-discharge light sources were carried out. One of the main
achievements of this research is development of the transformer plasmatron, which can
operate at the atmospheric pressures of the plasma-forming gas (argon, air) [6].
The thermal-physical and electric-physical characteristics of the transformer coupled
toroidal discharge were measured in argon, air, and argon mixtures with hydrogen, oxygen,
natural gas and carbon dioxide.
Reactions of plasmachemical synthesis of ozone, nitrogen monoxide and syngas were
studied within the gas-discharge chamber of the transformer plasmatron. Formation of high
(superequilibrium) concentration of nitrogen monoxide (up to 70%) was determined within
the peripheral low-temperature region of the vortex flow, what stabilizes the channel of a
transformer discharge. The physical model explaining this effect was suggested [7].
The process of thermal production of nitrogen from oxygen-argon plasma was studied
experimentally with the use of recuperative quenching methods. It was shown that at cooling
of oxygen-argon plasma with the rates of ~107 К/s, from 5 to 10% of oxygen could transform
into ozone at oxygen content in argon from 15 to 3 %, respectively.
The electric-physical and spectral characteristics of the transformer coupled toroidal
discharge were studied in mercury vapors and neon with the purpose of development of the
new electrodeless light sources [21 – 23].
It is shown that for the pressure of mercury vapour above 10 kPa and ratio L/D > 15, the
electrical characteristics of TCTD can be approximately calculated by the standard “channel”
model of the DC arc discharges [23].
However, the range of the middle pressures of mercury vapor demonstrates the unknown
effect of decreasing emission to the triplet with a potential of 7.73 eV for increasing discharge
power. The qualitative hypothesis, explaining this effect, is suggested.
Based on the obtained experimental results, we have developed and made the pilot
samples of electrodeless lamps: electrodeless mercury high intensity discharge lamps with an
luminous efficiency of 60 lm/W, and power of 3÷100 kW; germicide UV induction lamps of
a low pressure with the power of 50,100, 200 and 500W and luminous efficiency to the
resonant line 253.7 nm at the level of ~30–35% of consumed power, neon induction lamps
with the power of 100÷2000 W and luminous efficiency of 20÷30 lm/W.
REFERENCES
[1] Heinrich, F. B.; Shevel’ko, V. P. Introduction to Physics of Highly Charged Ions.CRC
Press, 2003.
[2] Bell, W.E. Applied Physics Letters. 1965, Vol. 7, 190–191.
[3] Eckert, H. U. AIAA Journal. 1971, Vol. 9, 1452–1456.
[4] Eckert, H. U. IEEE Transactions on Plasma Science. 1974,Vol. PS–2, 308–309.
[5] Goldfarb, V. M.; Donskoy, A.V.; Dresvin, S.V.; Rezvov, V.A. High Temperature.
1979, Vol. 17, 698–702.
[6] Kogan, V. A.; Ulanov, I. M. High Temperature. 1993, Vol. 31, 129–135.
[7] Kolmakov, K.N.; Ulanov, I. M.; Predtechensky, M. R.; Prikhodko, V. G.
Thermophysics and Aeromechanics. 2000, Vol. 7, 419–427.
[8] Shabalin, A. Plasma Sources Sci Technol. 2004, Vol. 13, 588–593.
[9] Kulumbaev, E. B.; Lelevkin, V. M. High Temperature. 1997, Vol. 35, 351–356.
[10] Kulumbaev, E. B.; Lelevkin, V. M. High Temperature. 1999, Vol. 37, 187–193.
[11] Gonzales, J. J.; Shabalin, A. Plasma Sources Sci. Technol. 2003, Vol. 12, 317–323.
[12] Reinberg, A. R. Inductively Coupled Discharge for Plasma Etching and Resist
Stripping. 1984. US patent 4.431.898.
[13] Cox, M. S. Toroidal plasma source for plasma processing. 2002. US patent 6.418.874.
[14] Zhang, B. C.; Cross, R. C. J. Vac. Sci. Technol. A. 1998, Vol. 16, 2016–2020.
[15] Anderson, J. M. Illuminating Engineering. 1969, Vol. 64, 236–244.
[16] Anderson, J. M. Electrodeless Gaseous Electric Discharge Devices Utilizing Ferrite
Cores. 1970. US Patent 3.500.118.
[17] Anderson, J. M. High Intensity Discharge Lamp Geometries. 1979. US patent
4.180.763.
[18] Godyak, V. A. High intensity electrodeless low pressure light source driven by a
transformer core arrangement. 1998. US patent 5.834.905.
[19] Curry, J. J.; Lister, G. G.; J.E. Lawler. J. Phys. D.: Appl. Phys. 2002. Vol. 35, 2945–
2953.
[20] Didenko, A. N.; Ulanov, I. M.; Predtechensky, M. R.; Kolmakov, K. N. Phys.—Dok.,
2000. Vol. 45, 155–157.
[21] Isupov, M. V.; Ulanov, I. M.; Litvintsev, A.Yu. High Temperature. 2004, Vol. 42, 682–
688.
[22] Isupov, M. V.; Ulanov, I. M. High Temperature. 2005. Vol. 43, 169–176.
[23] Ulanov, I. M.; Isupov, M. V.; Litvinsev, A. Yu. J. Phys. D.: Appl. Phys. 2007, Vol. 40,
4561–4567.
[24] Grey, J.; Jacobs, P. F.; Sherman, M. P. Rev. Sci. Instr. 1962. Vol 33, 738–741.
[25] Ambrazyavichus. Heat Transfer at Gas Quenching. Vilnius: Nauka, 1983 (in Russian).
[26] Ulanov, I. M.; Litvinsev, A. Yu.; Mischenko, P. A.; Krotov, S. V. Proc. of the Russian
(International) Conference “Physics of Low-Temperature Plasma–2007”.
Petrozavodsk. 2007, Vol. 1, 240–245. (in Russian).
[27] M. F. Zhukov and A.S. Anshakov. Electric-Arc Generators with Inter-Electrode
Insertions. Novosibirsk: Nauka, 1981. (in Russian).
[28] O.E. Skadchenko. Investigation of Ozone Formation in a Jet of Low-Temperature
Plasma. Thesis. Moscow, 1972. (in Russian).
[29] Zeldovich, Ya. B.; Sadovnikov, P.Ya.; Frank-Kamenetsky, D. A. Oxidation of Nitrogen
at Combustion. Moscow. Izd. AN SSSR, 1947. (in Russian).
[30] Polak, L. S. Kinetics and Thermodynamics of Chemical Reactions in Low-Temperature
Plasma. Moscow: Nauka, 1965. (in Russian).
[31] P.R. Ammann, R.S. Timmins. A. I. Ch. E. Journal. 1966, Vol. 12, 956–963.
[32] Zhivotov, V. K.; Rusanov, V. D.; Fridman, A. A. Diagnostics of Non-Equilibrium
Chemically Active Plasma. Moscow: Energoatomizdat, 1985. (in Russian).
[33] Potapkin, B.V. Chemistry of High Energies. 1983, Vol. 17. 524–530. (in Russian).
[34] Polak, L. S. Application of Plasma in Chemical Reactions. Moscow: Nauka, 1970. (in
Russian).
[35] Parkhomenko, V. D.; Soroka, P. I.; Krasnoutsky, Yu. I. Plasmachemical Technology.

Vol. 4. Low-Temperature Plasma. Novosibirsk: Nauka, 1991. (in Russian).
[36] Klarfeld, B. N. Zh. Tekh. Fiz. 1937, Vol. 7, 1017–1038. (in Russian).
[37] Elenbaas, W. The High Pressure Mercury Vapor Discharge. Amsterdam: North-
Holland, 1951.
[38] Rokhlin, G. N. Discharge Light Sources. Moscow: Energoatomizdat, 1991. (in
Russian).
[39] Klarfeld, B. N.; Taraskov, I. M. Technical Physics. 1934. Vol. 4, 504–514.
[40] Piejak, R.; Godyak, V.; Alexandrovich, B. J. Appl. Phys. 2001, Vol. 89, 3590–3593.
Chapter 4
FEATURES ON THE HIGH FREQUENCY DIELECTRIC

RESPONSE IN FERROELECTRIC MATERIALS
J. D. S. Guerra*
Grupo de Ferroelétricos e Materiais Multifuncionais, Instituto de Física, Universidade
Federal de Uberlândia, 38400-902 Uberlândia-MG, Brazil.
Single crystal and/or polycrystalline ferroelectric materials show a high frequency

dielectric dispersion, which has been attributed as well to a dispersive (relaxation like) as
a resonant mechanism. Physical properties such as relaxation and/or resonant motion
mechanisms can be investigated by analyzing the complex dielectric permittivity (real, ε’
and imaginary component, ε’’) in a broad spectral frequency range (100 MHz–13 GHz).
Especially, for classical (or ‘normal’) and relaxor ferroelectric systems a dielectric
response indistinguishable of dispersion or a resonance mechanism has been found in the
literature. The occurrence of such common dispersion process in so different kinds of
ferroelectric systems has encouraged the development of several mutually excluding
models to explain this physical phenomenon. Nevertheless, the reported results are not
conclusive enough to clearly distinguish each mechanism. In this work, a detailed study
of the dielectric dispersion phenomenon, including the microwave frequencies, carried
out in perovskite structure-type ferroelectric systems, for ‘normal’ and/or relaxor
compositions, is presented. The dielectric response in “virgin” and poled state have been
investigated taking into account the relative direction between the measuring direction
and the orientation of the macroscopic polarization. Results revealed that the dielectric
response in ferroelectric systems may be described as a general mechanism related to an
“over-damped” resonant process rather than a simple relaxation-like dielectric behavior.
1. INTRODUCTION
Since the late 1950s decade [1], the investigation of the high frequency dielectric
properties of dielectric materials has been one of the most challenging tasks in the field of the
*
Corresponding author: E-mail: santos@infis.ufu.br
170 J. D. S. Guerra
Physics Condensed Matter [1]. Microwave measurements of high dielectric permittivity

materials (> 100) will certainly assist in materials selection for high frequencies applications
such as high-speed data transmission and dielectric resonators devices [2]. Specifically
ferroelectric materials, which have relatively high dielectric permittivity, have led to a rising
interest in the gigahertz region and have prompted research for microwave range application
due to its excellent ferroelectric and piezoelectric properties [3]. The dielectric dispersion
(frequency dependence of the dielectric parameters, such as, real, ε’ and imaginary
component, ε’’ of the dielectric permittivity), as shown in the schematic representation of the
Figure 1, can provide information on the dynamics of the mechanism of phase transitions and
shows the frequency region where the ferroelectrics are useful for practical applications [4,
5]. The parameters εs and ε∞ in Figure 1 are the static (high frequency) and optical (high
frequency) dielectric permittivities, respectively.
Figure 1. Representative curves for the high frequency dielectric response, showing the frequency
dependence of the real (ε’) and imaginary (ε’’) component of the dielectric permittivity.
The study of the polarization mechanisms related to the observed behavior requires the
ability to work over both large frequency and temperature ranges, were coaxial cells and
network analyzers are commonly used [6]. Thus, for the study of orientation and relaxation
polarization effects, automated measuring equipments are usually applied and commercially
available today, using the reflectometric method by coaxial technique [7]. Most of the
microwave dielectric studies have been performed in a wide temperature range from room
temperature up to 900 K by using a coaxial line and a discrete varying frequency (resonant
cavity-RC) and/or continuously varying frequency (CVF) method as the principal techniques
[7, 8]. For instance, in the case of incipient ferroelectric materials, like Ca modified SrTiO3
[9, 10], for which a clear interpretation of the dielectric results [11] it is believed is lacking, it
is interesting to investigate the dielectric relaxation mechanisms mainly at low temperatures
(below room temperature) and high frequencies.
However, for high permittivity and high dielectric loss materials, which is the case of
ferroelectrics, the dielectric coefficients become difficult to be measured [12]. For
temperature-dependent measurements, specifically at low temperature region (T < 300 K), the
coaxial line of the high frequency analyzer and the sample holder must be thermally
decoupled because of the thermal contraction/expansion effects of the coaxial line-sample
Features on the High Frequency Dielectric Response in Ferroelectric Materials 171
system. Thus, the dielectric response may result mainly from spurious resonance effects of
the system when CVF methods are used, which may often to mask the true dielectric behavior
of the sample under test. In this way, the best control of the dielectric measurements may be
obtained by adjusting the pressure on the sample. Adjusting the pressure it is possible to
guarantee measurement of the true dielectric response of the material under test when
decreasing temperature.
Within the most commonly used ferroelectric materials, the barium titanate (BaTiO3 -
BT) system and others perovskite-type structure ferroelectric materials have been
distinguished because they presented intrinsic dispersion phenomenon near 1 GHz [3, 13].
This behavior can limit their use specifically in communication systems and electrically
controlled devices, such as, phase shifters [2]. In this context, such dielectric dispersions were
firstly observed in the so-called ‘normal’ ferroelectrics [14], being characterized later in
relaxor ferroelectrics (frequently named as relaxors) [15] and more recently in
antiferroelectrics [16] and incipient ferroelectrics [17]. However, it is well known that
‘normal’ ferroelectrics have micro-sized polar regions (domains) with long-range order [18],
while relaxors and doped-incipient ferroelectrics have nanometric polar regions
(nanodomains) distributed in a non-polar matrix, in which a short-range order prevails [19].
Therefore, the occurrence of such common dispersion process in so different kinds of
ferroelectric systems has encouraged the development of several mutually-excluding models
to explain this physical phenomenon [15, 20, 21-23], in terms of either a grain or ferroelectric
domain resonance [24, 25], as well as the correlation between the ferroelectric polar
structures and their respective dynamical response [26].
Previous investigations on both ceramics and single crystals of BaTiO3 [14] and others
perovskite-type structure ferroelectric materials [27] suggest the presence of a large dielectric
dispersion in the gigahertz region, which was reported to be close to a dipolar character (that
is to say, like a Debye-type relaxation). In spite of not being still clarified, several theoretical
attempts have been proposed to explain the origin of this effect. The relation between the high
frequency dielectric dispersion and the microstructure of ceramics was expressed by Von
Hippel [28], who attributed the piezoelectric resonance of the grain as the principal cause of
the observed microwave anomalies. Kittel [29] suggested that the domain walls motion have
an inertial component. He attributed the decrease of the dielectric permittivity at the
microwave region to the resonance of the domain walls in ceramics. Some others models
based on piezoelectric resonance of individual domains [30] or on correlated hopping of off-
centered ferroelectric active ions between several potential wells [23] have been proposed in
more recent years. In all the cases, the structural disorder on the atomic scale, assuming the
two-minimum potential relief for some lattice ions, was used as the principal cause for the
high frequency dielectric dispersion. As observed, the dielectric dispersion in ferroelectric
materials has been a well investigated subject. However, more recent experimental results
[31] showed evidences of a resonant dielectric behavior, rather than the previous observed
dielectric dispersion. It has been shown, that the resonant behavior may coexist, in the same
studied systems, together with the dispersion behavior, under certain conditions. Therefore, it
seems that the high frequency dielectric response in ferroelectric materials, observed near to 1
GHz, is dictated by a universal mechanism involving a resonant response (damped or over-
damped), which is until now a not fully clarified matter.
The objective of the present work is to show a detailed review concerning the high
frequency dielectric dispersion in ferroelectric systems, by considering the obtained results
172 J. D. S. Guerra
for ‘normal’ and relaxor ferroelectric ceramic compositions with perovskite type structure. A
measuring system suitable for investigation of the high frequency dielectric dispersion, over a
wide temperature range from 90 K to 450 K, is presented. Construction and correct
calibration procedures are also described, including a brief description of the experimental
setup. On the other hand, the present work proposes the fulfillment of the nature of the
microwave dielectric dispersion in ferroelectrics materials viewing a better understanding of
the dispersion and/or resonant behavior verified in ‘normal’ ferroelectric systems at high
frequencies.
2. EXPERIMENTAL DESCRIPTION
Dielectric measurements can be commonly carried out by using high frequency
impedance analyzers. Here, will be described measurements carried out using a Network
Analyzer HP-8719C in the frequency range of 50 MHz to 2 GHz. Technical reasons
associated to the system configuration limited accurate measurements above the mentioned
frequency range. Indeed, at higher frequencies (> 2 GHz), due to natural resonance of the
experimental setup, only measurements at discrete frequencies can be obtained, although the
signal in the analyzer can be generated up to higher frequencies. The dielectric response is
obtained by the reflectometric method where the reflection coefficients (real and imaginary
components: Γ’ and Γ’’, respectively) versus frequency are measured. The setup, currently
known by coaxial probe method, is shown in the Figure 2.
Figure 2. Experimental diagram for dielectric characterization using the reflectometric technique by the
coaxial probe method.
The sample is kept at rest on the end of the probe. However, in the case of very high
dielectric permittivity (ε’ > 100) materials this method presents some disadvantages: a- with
the increase of the frequency the modulus of the reflection coefficient ⏐Γ⏐ becomes close to
1 and its phase φ close to zero. Therefore, such materials are difficult to be measured and the
quality of the results largely depends on the initial calibration of the analyzer; b- air gaps
formed between the sample and the inner/outer conductors cause very large measurement
errors when the sample is badly tooled and c- for high temperatures different thermal
expansions of the sample and inner/outer conductors also can bring errors generated by the air
gaps formation. Thus, for reducing these limitations and successful characterize high
dielectric permittivity materials, as is the case of ferroelectrics, an alternative coaxial line
method has been considered. In order to explore a large frequency spectrum of dielectric
properties of ceramic materials, CVF measurements involving the case of a coaxial line and
the reflectometric technique are appropriate [7]. The coaxial line, including commercial
coaxial semi-rigid connectors, provides a suitable link between the measuring port of the
network analyzer and the temperature-controlled sample holder, as shown in Figure 3.
Figure 3. Experimental diagram for dielectric characterization using the reflectometric technique by the
coaxial line method.
The coaxial line method can be schematically presented by a simple configuration as

shown in Figure 4a. The system is based on a microwave generator signal, a signal detector
and a divider system located into the network analyzer and a 50 Ω coaxial connector coupled
to the coaxial line being terminated by the capacitance of a cylindrical sample. By using the
reflectometric method, the coaxial line (see Figure 3) can be characterized in terms of the
network theory as a terminated two-part network as shown in Figure 4b.
The measured reflection coefficient Γ* (=Γ’ + jΓ’’) is fundamentally affected by the
absorption and phase shift of the signal inside the line, and can be described by the S
parameters (S11, S22, S12 and S21) of the network. Therefore, in order to determine the
frequency dependence of the reflection coefficient of a sample, the coaxial line has to be
previously calibrated. Thus, careful compensation procedures have to be carried out to (i)
eliminate spurious reflections that may result by transmission line discontinuities, and also (ii)
174 J. D. S. Guerra
reduces stray resistances and stray capacitances in the sample holder. Standard terminations
(open, short and 50 Ω) were used over all the operating frequency range at room temperature.
Because of the special construction of the system, the calibration performed at room
temperature was assumed to be valid over the whole temperature range. It is important to
point out that identical dimensions of the sample and standard terminations are necessary to
avoid faults by different measuring planes and different stray fields. Thus, reflection
coefficients data corrected for the line effects can be frequency scanned and automatically
transferred from the network analyzer to the computer. By using the relation between the
1 − Γ*
reflection coefficient and the admittance at the line Y = Yo (where Υo=1/Zo) and
1 + Γ*
taking into account the complex dielectric permittivity (ε=ε’–jε’’), the real and imaginary
component of dielectric permittivity are obtained by using the Eqs. (1) and (2) [12],
respectively,
Figure 4. Experimental diagram of reflectometric method used by the network analyzer.
1 ⎡ − 2Γ ' ' ⎤
ε'= ⎢ 2⎥
(1)
Af ⎣ (1 + Γ') + Γ' ' ⎦
2
1 ⎡ 1 − Γ ' 2 − Γ ' '2 ⎤

ε''= ⎢ ⎥ (2)
Af ⎣ (1 + Γ')2 + Γ' '2 ⎦
where f is the measuring frequency; A is determined by the characteristic impedance Zo of the

π 2r 2
analyzer ( A = 2ε o Z o ); r and t are the radius and thickness of the sample,
t
respectively, and εo is the dielectric permittivity of the free space. To obtain more precise
measurements, the sample dimensions should have (preferably) cylindrical symmetry with a
diameter smaller than the diameter of the central conductor of the coaxial line, in order to
avoid border effects and to ensure the electric field only in axial direction inside the sample.
Heating of the sample can be achieved by a conventional resistive heater, which can be
screwed onto the top of the coaxial line and coupled to a temperature control Flyever
FE50RP, as shown in Figure 3. Finally, this measurement system can be inserted into a
cryogenic system, which allows to perform experiments under either high vacuum level or
nitrogen atmosphere to covers the low temperature ranges. The coaxial line can be inserted
inside a silicate boron tube, in order to isolate the line of the liquid nitrogen. Gassy nitrogen
can be placed inside the silicate boron tube by a gassy nitrogen input to account for a best
good control of the cooling and heating processes.
Air gaps at the interface between the coaxial line and the sample were avoided by an
optimal sample polishing process, in order to guarantee parallel and flat faces. Alumina
powder of 1 μm was used in this polishing process. Gold electrodes were sputtered on the
faces of the discs with 2.0 mm in diameter and 0.5 mm in thickness to insure good electrical
contacts, for all the investigated samples.
3. HIGH FREQUENCY DIELECTRIC MEASUREMENTS

3.1. Microwave Dielectric Characterization in Low Permittivity Materials
(Al2O3)
Firstly, an alumina (Al2O3) ceramic disk (2.0 mm diameter and 0.5 mm thickness) was
used for the dielectric characterization in the range of 50 Hz to 2 GHz at low temperatures.
The Al2O3 ceramic was chosen as reference material for the measurements to successfully
check the correct calibration of the experimental set-up in the low temperature region.
Figure 5 shows the dielectric response of the alumina ceramic obtained at four selected
temperatures (90 K, 150 K, 200 K and 300 K). As can be seen, the obtained data were
characterized by a high stability of the dielectric parameters in the whole frequency interval.
Also, the real component of the dielectric permittivity was found to be around 9.5, in
agreement with previously reported results in the literature (ε’ ≈ 9) [32, 33].
40.0 40
1.0
90 K
30 300 K
150 K
200 K 20
ε'
30.0 300 K 10
Ref. [32]
0
0.5
9 9 9
Ref. [33] 2x10 4x10 6x10
20.0
ε''
ε'
0.0
10.0
LF HF
4 5 6 7 8 9
10 10 10 10 10 10
Frequency (Hz)
176 J. D. S. Guerra
Figure 5. Dielectric response for alumina ceramic including the low (LF) and high frequency (HF)
range (100 Hz–2 GHz), at room temperature.
For comparison, low frequency dielectric data were measured using an HP-4194A
Impedance/Gain-Phase Analyzer (100 Hz–10 MHz), and the results have been added to
microwave properties as shown in the same Figure 5. Since Al2O3 ceramic materials do not
display a dielectric dispersion phenomenon [34] (frequency independent dielectric
permittivity), the smooth continuity in the values of real and imaginary component of
dielectric permittivity for low and high frequency ranges confirms the accuracy of the results
obtained at the microwave frequency range. The ε’ and ε’’ values are again in agreement with
those found in the literature [32, 33].
3.2. Microwave Dielectric Characterization in High Dielectric Permittivity

Materials
In order to demonstrate the range of applicability of the experimental setup and to

illustrate its capabilities when investigating high dielectric permittivity materials, dielectric
measurements were carried out in Sr1-xCaxTiO3 (x=0.1) quantum paraelectric ceramic
samples, hereafter labeled as SCT–90/10, as a function of the frequency (50 MHz–2 GHz)
and temperature (90 K–450 K).
10.0
90 K
150 K
8.0 200 K
250 K
300 K
6.0
ε' ( 10 )
360 K
2
430 K
4.0
2.0
0.0
2.0
SCT-90/10
1.5
ε'' ( 10 )
1.0
2
0.5
0.0
8 9
10 10
Frequency (Hz)
Figure 6. High frequency dependence of the real and imaginary component of dielectric permittivity (ε’
and ε’’, respectively) for the SCT quantum paraelectric system, at several temperatures above the
transition temperature.
In this case, due to additional effects, related to the piezoelectric phenomenon
characteristic in ferroelectric systems, a careful control of the pressure on the sample should
be taken into account. The adjustment and control of the pressure on the sample was made
when decreasing temperature because loss of contacts between the line and the sample holder
originated by the thermal contraction of the different elements. Thus, the dielectric response
of the sample was obtained at each measured temperature for a critical pressure level on the
sample, sufficient to guarantee the electrical contacts. In this case, the frequency dependence
of the dielectric properties (ε’ and ε’’) was carried out at various temperatures. Figure 6
shows such dependences for the temperature and frequency range of 90 K–450 K and 50
MHz–2 GHz, respectively. As can be observed, the frequency dependence of ε’ and ε’’
revealed the characteristic behavior of a dielectric dispersion, i.e., a decrease in the real
component of the dielectric permittivity and a maximum in the imaginary one at the
characteristic frequency (fR). The fR value lies in the order of 700 MHz, in agreement with the
reported values obtained for some perovskite-type structure materials [15, 24].
Similarly to the BaTiO3 type ferroelectric system [35], the dielectric dispersion does exist
for temperatures above the transition temperature, which suggests the existence of polar
regions in the paraelectric state [36], and will be discussed in the next sections. Parallely, low
frequency dielectric measurements were performed at the frequencies of 10 kHz, 100 kHz and
1 MHz, in the same temperature interval. The results, shown in Figure 7, do not present any
dielectric dispersion in this low frequency range, evidencing a good agreement between the
low and high frequency results.
3.3. Microwave Dielectric Characterization in ‘Normal’ and Relaxor

Ferroelectrics
The present section treats about the investigation of the high frequency dielectric
dispersion processes in Pb1-xLaxTiO3 ferroelectrics ceramics for ‘normal’ (x=0.15) and relaxor
(x=0.27) compositions (hereafter labeled as PLT–15 and PLT–27, respectively), in order to
better understand the occurrence of such common dielectric dispersion process in so different
kinds of ferroelectric systems.
16.0
90 K
200 K
12.0 300 K
8.0
ε' ( 10 )
LF HF
2
4.0
0.0
4 5 6 7 8 9
10 10 10 10 10 10
Frequency (Hz)
178 J. D. S. Guerra
Figure 7. Low and high frequency dielectric dispersion for the SCT system for temperatures of 90 K,
200 K and 300 K, respectively.
Figure 8 shows the frequency dependence of the real, ε’ and imaginary, ε’’ component of
the dielectric permittivity for the PLT–15 ceramics, at different temperatures. The Curie
temperature of the samples, TC~393 K, was determined by the dielectric measurements of ε’
at low frequencies (1 kHz). A dielectric response indistinguishable from a dispersive behavior
was observed in the whole analyzed temperature interval. As can be seen, for a fixed
temperature (i.e. at 300 K) and in the frequency range of 70–400 MHz, ε’ decreases slightly
as the frequency increases. Above 400 MHz, ε’ decreases quickly, whereas ε’’ passes through
its maximum value. The frequency corresponding to the maximum of imaginary component
of the dielectric permittivity (fR=700 MHz) is known as the characteristic frequency of the
dispersive process, and it is associated to a polarization mechanism responsible for the
dissipation. The obtained value for the characteristic frequency is in agreement with those
reported for other ferroelectric materials commonly used for microwave applications [24, 37].
1.2 1.5
300K
PLT-15 390K
420K 1.2
ε'' ( 10 )
0.6 440K
ε' ( 10 )
3
445K
3
0.9
0.0 0.6
0.3
8 9
0.0
10 10
Frequency (Hz)
Figure 8. Frequency dependence of the real (ε’) and imaginary (ε’’) component of the dielectric
permittivity for the PLT–15 ceramics, as a function of the temperature.
The origin of this behavior has been attributed to the ferroelectric domain walls
vibrations in the ferroelectric material [38]. In this way, the domain wall motion is well
known to contribute to the polarization of ferroelectric systems. The frequency of the domain
walls vibration (obtained at the Gigahertz region) may be observed by applying an alternate
electric field of very high frequency. For frequencies appreciably lesser than fR, the
ferroelectric domains contribute their full share to the polarization so that the real component
of the dielectric permittivity becomes equal to the static dielectric permittivity (εs) and,
therefore, the losses (associated to the imaginary component of the dielectric permittivity)
vanish. With the increase of the frequency, the domain vibrations increase and consequently
the imaginary component of the dielectric permittivity starts to increase up to its maximum
value. On the other hand, for a frequency higher than fR, the domain wall vibrations are no
longer able to follow the field variations and the real component of dielectric permittivity
approaches its clamping values (ε∞). In this frequency range, therefore, ε’’ passes through its
maximum value and continuously decreases for the highest frequency values.
On the other hand, a closer inspection in the data of Figure 8 seems to support the picture
of a distributed dipolar relaxation mechanism of the dielectric response. Indeed, the obtained
value for the characteristic width (λD) at a half-height for the imaginary dielectric permittivity
peak (known as full-width half maximum, FWHM), specifically at room temperature, was
found to be around 1.91 decades. This value of the FWHM is higher than the FWHM of a
Debye-type peak (λD=1.14 decades) [34]. Therefore, the obtained result reveals that the
observed dielectric behavior cannot be associated to a Debye-type relaxation process. Results
for the dispersive process obtained in the studied PLT–15 samples (in “virgin” state) clearly
show a mechanism with a relaxation times distribution function. The data were fitted taking
into account the most useful distribution functions for the investigation of the relaxation
processes, as reported in the literature [39], and results revealed that the obtained dielectric
response was found to be close to a Davidson-Cole (DC) distribution function [40]. The
obtained values for the characteristic parameters, such as, the mean relaxation time (τ) and the
relaxation time distribution parameter (β), which represents degrees of divergence from the
‘ideal’ Debye model, were 1.45×10-9 s and 0.75, respectively. Compared to single-
exponential Debye behavior corresponding to β=1, values of 0 < β < 1 result in broadened
imaginary dielectric permittivity peaks. Smaller the values of β the greater the deviation with
respect to Debye-type relaxation. Such deviations from Debye behavior are commonly
ascribed to a distribution of relaxation times arising from disorder [40]. The obtained mean
relaxation time is in good agreement with those obtained for similar ferroelectric systems, as
reported in previous works [15].
As can be seen in Figure 8, the dielectric dispersion occurs not only in the ferroelectric
region (T<TC) but also in the paraelectric one (T>TC) close to TC. This behavior, as pointed
out in a recent work [26], suggests the existence of polar regions at temperatures higher than
the transition temperature, which could be one more evidence of the contribution for an
“order-disorder” type paraelectric-ferroelectric phase transition in ABO3 perovskite
structures, as reported in the literature [41, 42]. The existence of an order-disorder component
in the paraelectric-ferroelectric phase transition of PbTiO3 explains the persistence, although
weak, of the dielectric dispersions above TC [26]. It is important to point out that, for higher
temperatures, the data also revealed highest values for the FWHM, showing a distributed
dipolar relaxation mechanism for all the investigated temperature range. At the same time, a
decrease in the distribution parameter (β), as the temperature increases, confirms that the
dielectric spectrum becomes extremely diffuse. Correspondingly, the distribution of
relaxation times becomes extremely wide. As can be observed, similarly to the results
reported by other authors for BaTiO3 [22, 35], PbTiO3 derived materials [8] or some
tetragonal tungsten bronze (TTB) structure systems [43], the obtained dielectric response
curves suggest a dispersive mechanism of the complex dielectric permittivity, without any
evidence of a resonant response.
Figure 9 depicts representative curves of the frequency dependence of ε’ and ε’’ for the
PLT–27 (relaxor composition) measured at different temperatures below and above the
respective temperature of the maximum of the dielectric permittivity, Tm. The data also reveal
strong temperature dependent dielectric dispersions for this relaxor ferroelectric composition,
not only for temperatures below the temperature of the maximum dielectric permittivity, but
also for temperature above Tm.
180 J. D. S. Guerra
4.0 3.0
226K
259K
290K 2.4
2.0 325K
ε'' ( 10 )
ε' ( 10 )
350K
3
1.8
3
0.0
1.2
PLT-27
0.6
0.0
8 9
10 10
Frequency (Hz)
Figure 9. Frequency dependence of the real (ε’) and imaginary (ε’’) component of the dielectric
permittivity for the PLT–27 ceramics, as a function of the temperature.
The high frequency dielectric dispersion can be related to a microscopic polarization

mechanism and, therefore, a possible explanation for the observed behavior could be found
taking into account atomic level considerations. Indeed, the mechanism for this dielectric
behavior is associated with a shift of Ti4+ (the B-site ion in the perovskite structure) inside the
oxygen octahedron. The ion shifts from one potential to the other in a double well potential
model (relaxation motion), as shown in the Figure 10a, being ΔU the high of the potential
barrier [44]. In fact, such mechanism can be represented as real chains (or volumes) of
correlation (named as correlation chains) including a large number of Ti4+ ion cooperative
jumps, which must be considered in the relaxation mechanisms. These chains are isolated
from each other by several types of defects (vacancies, impurities, point defects, etc.) (Figure
10b).
Figure 10. Double potential model (a) and correlation chains scheme (b) for the high frequency
dielectric response.
In order to obtain additional information of the dielectric spectrum, the obtained results
were fitted by using the Eqs. (3) and (4), which consider a damped harmonic oscillator model
[45], where the frequency dependence of the complex dielectric permittivity is governed by
the Eqs. (3) and (4).
BωR2 (ωR2 − ω 2 )
ε ' = A+ (3)
(ω 2
R −ω 2 2
) +γ ω 2 2
BωR2 γω
ε '' = (4)
(ωR2 − ω 2 ) + γ 2ω 2
2
A and B are constant parameters, which are directly related with the dielectric dispersion
parameters. For instance, B=εs–ε∞, defined as dielectric strength (Δε), is the contribution to
the dielectric dispersion of the static dielectric permittivity, εs; A=ε∞ is the contribution to the
dielectric permittivity of the higher frequency electronic processes; ω=2πf is the angular
ωR
frequency and fR = , is the characteristic frequency of the process (defined as,
2π
1
= 2π f R , where τ is the mean relaxation time). The dispersion characteristic parameters,
τ
such as, Δε, fR and γ (the damping coefficient of the dielectric response), can be obtained
directly from the fitting of the obtained dielectric response (frequency dependence of ε’ and
ε’’) with the Eqs. (3) and (4), which undoubtedly might be an essential feature in order to
identify the mechanism involved in the dielectric anomaly.
Figure 11 shows the temperature dependence of the characteristic frequency (fR) and the
dielectric strength (Δε) for both PLT–15 and PLT–27 compositions. The results reveal that Δε
and fR reach simultaneously a maximum and a minimum, respectively, in a temperature that
coincides with its respective maximum real dielectric permittivity, TC and Tm for the PLT–15
and PLT–27 compositions, respectively.
2.8
1.2 PLT-15
2.4
1.0 fR: PLT-15
Δε ( 10 )
2.0
fR ( GHz )
Δε: PLT-15
3
0.8
1.6
0.6 1.2
PLT-27
0.4 0.8
fR: PLT-27
Δε: PLT-27
0.2 0.4
240 270 300 330 360 390 420
Temperature (K)
Figure 11. Temperature dependence of the characteristic frequency (fR) and dielectric strength (Δε) for
the PLT–15 and PLT–27 compositions.
182 J. D. S. Guerra
Referring to the observed results, the essential features can be summarized as follows. a-
the microwave dielectric dispersion process persists at temperatures higher than the
temperature of the maximum dielectric permittivity for both compositions, vanishing in the
temperature interval where 1/ε’ vs. T obeys the Curie law (around TB, the Burns temperature,
characterized by the temperature of nucleation and slowing down of the ferroelectric
domains); b- above Tm, it is noticed that fR is much more temperature dependent for the PLT–
15 than for the PLT–27 composition, vanishing earlier for the former composition; c- the
magnitude of the characteristic frequencies (fR) for both compositions are comparable,
although their polar structures and correlation length differ significantly [22, 46, 47]. Similar
results were found for incipient ferroelectrics and antiferroelectrics [16, 17]. These
experimental results allow us to interpret the microwave dielectric process according to the
scenario described as follows.
First, the fact that the microwave dielectric dispersions became evident even for
temperatures far above Tm clearly demonstrates that the simple presence of polar regions,
independently of their size, volume fraction and correlation length, is the sufficient condition
for the existence of such dielectric dispersion process in perovskite ferroelectric systems.
Nevertheless, the only plausible common mechanism inherent in so different ferroelectric
domain structures is the boundaries of ferroelectric domains and nanodomains. Therefore, it
has been proposed that the field-induced vibration of polar region boundaries of domains
(domain walls) and nanodomains (interphase boundaries between the polar region and the
non-polar matrix) is the common mechanism responsible for the dielectric dispersion process
in the microwave range.
The second intriguing question is why such dispersion occurs around the same frequency
interval (fR) for all the ferroelectric systems, although their apparent distinct nature. In this
context, in analogy with the oscillating membrane theory [38, 48] and in accordance with the
side-way motion of the boundaries of the polar regions discussed above, it is suggested that fR
is governed by the ratio between the effective force constant (κeff) and the effective mass
(Meff.) of the polar regions boundaries, fR ≈ κ eff / M eff . The effective mass is the mass of
domain walls and interphase boundaries for normal ferroelectrics and relaxors, respectively,
while the force constant (κeff) is dictate by elastic properties of the respective polar region
boundaries [19, 38, 49]. Thus, it is believed that this ratio would have almost the same value
for all perovskite ferroelectric systems, justifying the similar values found for fR. For instance,
for normal ferroelectrics the higher force constant would be balanced by the higher effective
mass of domain walls. On the other hand, for relaxors and incipient ferroelectrics, the relative
smaller κeff is balanced by the relative low massive interphase boundaries, thus keeping the
ratio reasonably constant.
Finally, such proposed relation for fR is also able to explain satisfactory the behavior of fR
(slope) from the paraelectric to ferroelectric phase transition. For temperatures higher than TC,
the high thermal energy reduces the dipolar interactions contributing for a low κeff, and
consequently, a relatively low value for fR. With the decrease of the temperature, the thermal
energy decreases in favor of the formation of the polar regions, promoting the increase of
Meff, and consequently fR slightly decreases. For temperatures near TC, a sudden increase in
the interaction energy, and consequently in the κeff, takes place, resulting in an increase of fR.
Indeed, in the case of PLT–15 the paraelectric-ferroelectric phase transition is predominantly
a displacive-type transition. Therefore, it is expected an abrupt crossover between the two

structural phases (paraelectric-ferroelectric), which means an accelerated disappearance of the
polar regions (domains and domain walls). This fact results in a sudden change in both order
parameter and elastic properties, thus reflecting in the force constant. Moreover, the existence
of an order-disorder component in the paraelectric-ferroelectric phase transition explains
satisfactorily the existence of the dielectric dispersion above Tm. On the other hand, it is well
known that relaxor ferroelectrics do not necessarily present a macroscopic structural phase
transition through or bellow the temperature of maximum permittivity (Tm). Therefore, for the
relaxor composition, it is proposed that the low slope of fR above Tm reflects the gradual
condensation and the slowing down of polar nanoregions below TB. Furthermore, the fact the
microwave dielectric dispersion is characterized at temperatures much higher than Tm, for the
PLT–27 composition, confirm the existence of nanodomains at temperatures much higher
than Tm.
20.0
(a) TC = 393 K
15.0
10.0
ε' ( 10 )
3
5.0
0.0
30.0
(b) 300 K
25.0
340 K
20.0 360 K
ε'' ( 10 )
390 K
3
15.0 430 K
10.0
5.0
0.0
8 9
10 10
Frequency (Hz)
Figure 12. Frequency dependence of the complex dielectric permittivity for the PLT–15 samples under
a uniaxial stress, at different temperatures.
3.4. Mechanical and Electrical Driving Field Effects
3.4.1. Stress effect analysis
Because the dielectric response in ferroelectric materials is strongly susceptible to the

influence of electric and/or mechanical fields, the microwave dielectric properties in the
PLT–15 ceramics have been now investigated under the influence of a mechanical uniaxial
184 J. D. S. Guerra
stress, applied parallel to the measurement direction. The dielectric properties were obtained
over the same frequency and temperature range to that obtained for the ‘stress free’ samples
described in the previous section. The results presented in the Figure 12, clearly show that at
room temperature (solid line), ε’ remains essentially flat up to 700 MHz, increases traversing
a maximum, and then decreases to its clamped value (see Figure 12a). As can be seen, this
anomalous behavior obtained for the stressed samples and indistinguishable from a resonant
response, is observed in the whole analyzed temperature range. On the other hand, the
stressed samples also exhibited the highest values of the imaginary component of the
dielectric permittivity, with its ε’’ peak apparently shifted to a higher frequency (Figure 12b),
when compared to that obtained for the ‘stress free’ samples. This dielectric response
obtained for the stressed ceramics corresponds to a true resonance rather than a dispersion
process, with a characteristic frequency around 900 MHz (at room temperature) higher than
the obtained for the ‘stress free’ samples (700 MHz), at the same temperature.
It is interesting to point out, from the Figure 12b, that the high values of the maximum
imaginary dielectric permittivity for the stressed samples, when compared to the maximum
decrease of the real component (εs–ε∞), suggest that is not possible to describe the spectrum
as a classical Debye’s dielectric dispersion. At the characteristic frequency (900 MHz), the
value of ε’’ at room temperature, is about 2 times higher than that predicted by the Debye’s
classical model, [(εs–ε∞)/2], for the stressed samples. This result has an important physical
implication; indeed, it clearly confirms that the high frequency anomalies do not correspond
to a real dispersion process as predicted by the Debye’s model (relaxation-like behavior).
As reported in previous works [21, 22], the assumption that the observed dielectric
anomaly could be, in this case, related to a piezoelectric resonance of the grains or individual
domains suggests that the main influence of the uniaxial stress would be to decrease the
losses related to this process, contrary with the results obtained for the PLT–15 samples.
Therefore, the true mechanism responsible for the obtained anomalies remains still not clear.
The dispersion characteristic parameters, were now also obtained from the fitting of the
obtained dielectric response by using the Eqs. (3) and (4). In order to compare the
temperature evolution of the characteristic parameters results for the stressed and ‘stress free’
samples for the PLT–15 ferroelectric ceramics are presented in the Figure 13. As can be
observed, the temperature dependence of fR and ∆ε is quite similar for the stressed and the
‘stress free’ samples. So that, with the application of the uniaxial mechanic stress, although
the dielectric response evidences resonance characteristics, this behavior remains modulated
by the paraelectric-ferroelectric phase transition. As a result, the thermal evolution indicates
that, in both cases, the characteristic frequency goes through a minimum while the dielectric
strength pass through a maximum around 393 K, which coincides with the paraelectric-
ferroelectric phase transition temperature (TC), as observed in the low frequency dielectric
measurements [26]. This result shows that the maximum dielectric dispersion appears near
the transition temperature, independently of the applied external mechanical driving field.
On the other hand, it is important to point out that the damping coefficient values (γ),
obtained from the fitting of the experimental data by using the Eqs. (3) and (4), at room
temperature, were lower for the stressed samples (0.45⋅109 s-1), than those obtained for the
‘stress free’ samples (8.80⋅109 s-1). This result suggests that the high frequency dielectric
anomalies, dispersion or resonance, can be described as either an over-damped or a damped
resonant response, respectively, where the main contribution to the dielectric response can be
associated to the ferroelastic and/or ferroelectric components of the ferroelectric materials,

which directly modifies the damping of the system.
Referring to the obtained results, the essential aspect to be discussed can be summarized
to the feature that with the application of an external mechanic driving field the dielectric
response of the system clearly pass from a dispersive to a resonant dielectric behavior. In this
way, considering an “over-damped” resonant response, the variations in the strength of the
damping coefficient of the system originate mainly due to the contributions of either
ferroelastic or ferroelectric dipolar components. For lower damping systems, as the case of
the obtained resonant behavior for the stressed samples, the main contribution to the dielectric
response is governed by the ferroelectric dipolar component. On the contrary, for higher
damping coefficient systems, the major contribution of the dispersive behavior (as observed
in the “stress free” samples) is due to the ferroelastic dipolar component, which prevails over
the ferroelectric component.
On the other hand, it is well known that ordinary ferroelectric materials are classic
hybrids ferroics, that is to say, they present a strong coupling between the ferroelectric-
ferroelastic dipolar components [50, 51]. When applying a uniaxial mechanic stress a
reorientation of the electric and elastic dipoles takes place. In the same way, a mechanic strain
in the material can be observed if applying an electric driving field. Then, as in the case of the
stressed samples, applying a uniaxial stress parallel to the measurement direction promotes
the dipolar reorientation in the perpendicular direction of the applied uniaxial stress, that is to
say, an induced polarization may appears in the perpendicular direction to the applied uniaxial
stress direction. Therefore, applying a uniaxial stress parallel to the measurement direction is
equivalent to apply a poling electric field in the perpendicular direction of the measurement
direction. Thus, in order to obtain a better understanding of the current analysis, the observed
anomalies around 1 GHz for the studied PLT–15 ceramics, will be now investigated taking
into account the influence of the relative orientation of the macroscopic polarization on the
dielectric dispersion in the poled samples.
1.2 6.0
fR ; Δε - stressed
fR ; Δε - stress free
0.9 4.5
fR (GHz)
Δε ( 10 )
0.6 3.0
3
0.3 1.5
0.0 0.0
300 330 360 390 420 450
Temperature (K)
Figure 13. Temperature dependence of the characteristic parameters (characteristic frequency, fR and
dielectric strength, Δε) for the stressed and ‘stress free’ PLT–15 samples.
186 J. D. S. Guerra
3.0
'stress free' samples
(a)
1.5
ε' ( 10 )
3
0.0 Unpoled
Poled ( I I )
Poled ( )
2.0 (fR = 1.03 GHz) (b)

(fR = 700 MHz)
ε'' ( 10 )
(fR = 700 MHz)

3
1.0
0.0
8 9
10 10
Frequency (Hz)
Figure 14. Frequency dependence of the complex dielectric permittivity, at room temperature, for the
PLT–15 unpoled ‘stress free’ samples, and the poled samples measured in the parallel (//) and
perpendicularly (⊥) direction to the poling direction; (a)- real and (b)- imaginary component.
3.4.2. Poling field effect analysis
The dielectric response was now performed in poled samples (Ep = 2 kV⋅mm-1) at room
temperature, in both parallel and perpendicular directions to the poling field direction and in
the same frequency interval, following the same experimental procedure previously
described. Thus, the temperature dependence of dielectric permittivity (real and imaginary
component) was obtained parallel (ε’//) and perpendicularly (ε’⊥) to the polarization direction.
The results are shown in Figure 14.
The results reveal that the dielectric response, observed in the sample measured parallel
to the poling direction, presents an ‘apparent’ dielectric dispersion process, which in turn is
similar to that observed for the unpoled sample (see Figure 8). However, it is observed that
the maximum of the imaginary component of the dielectric permittivity (ε’’//), for the sample
measured parallel to the poling direction, is around 4 times higher than (εs//–ε∞//)/2.
Furthermore, a detailed inspection for the FWHM of the imaginary dielectric permittivity
peak for the sample measured parallel to the poling direction revealed that the observed
dielectric dispersion has not a relaxation-like character, since the FWHM of the peak was
found to be around 0.85 decades. This value is lower than the characteristic width of a Debye-
like process (λD=1.14 decades, the low limit value for FWHM). Indeed, it implies that it is not
possible to describe the observed dielectric spectrum as a classic Debye’s dielectric

relaxation, as previously discussed in the section 3.4.1. Therefore, it is possible to affirm that
the observed dielectric anomaly really correspond to a resonant-like dispersion rather than a
simple relaxation-like dielectric behavior.
Figure 14 also displays the dielectric response in the sample measured perpendicularly to
the poling direction. The results show that ε’⊥ slightly increases up to 500 MHz. However,
after that, it abruptly increases and subsequently decreases for higher frequencies to its
clamped values. The maximum of the imaginary component of dielectric permittivity (ε”⊥)
also presents a remarkable high value, now observed around 1.03 GHz. The imaginary
component of the dielectric permittivity for the sample characterized perpendicular to the
poling direction is about 7 times higher than the values predicted by the theoretical Debye’s
model. These results show that the dielectric response obtained for the samples measured
perpendicularly to the poling field is characterized by a resonant mechanism, which may be
associated to a damped resonance process rather than to a simple Debye’s dielectric
relaxation.
The above described experimental results clearly show that in the same sample,
depending on the relative orientation between the measuring direction and the macroscopic
polarization direction, the dielectric dispersion in the GHz region seems to behave as either a
dispersive (relaxation-like) or a resonant mechanism. This change in the dielectric response
can be well described considering a general resonant behavior, which can be associated with
an “over-damped” resonance process. Therefore, it can be affirmed that the microwave
dielectric spectra observed in ferroelectric materials may be described in terms of an “over-
damped” resonance involving either dispersive or resonant behavior, rather than a simple
dispersion process (relaxation-like), which are intimately related to the variation in the
damping strength of the system. At the same time, the damping strength is affected by the
coupling between the ferroelectric and ferroelastic dipolar components. The values of γ were
now obtained for the poled samples, by the fitting of the experimental data, and showed in the
Table I, together with those obtained for the ‘stress free’ and stressed samples. As observed,
the damping coefficient also shows a decrease when the dielectric behavior pass from a
dispersive (relaxation like) behavior to a pure resonant response. As can be seen, the γ value
for the poled samples measured in the parallel direction to the poling direction was around
8.97⋅109 s-1, similar to that obtained for the ‘stress free’ samples and, at the same time, higher
than those obtained for the poled samples measured in the perpendicular direction to the
poling direction.
Table 1. Dispersion characteristic parameters obtained for the PLT–15 samples, at

room temperature.
Sample ε∞ Δε fR (GHz) γ (109 s-1)

Stress free 113 453 0.70 8.80
Stressed 266 691 1.44 0.45
Poled // 145 300 0.70 8.97
Poled ⊥ 149 541 1.04 1.15
188 J. D. S. Guerra
Thus, from the obtained results, it is possible to affirm that the dispersive behavior
observed for the ‘stress free’ samples may be a consequence of an increase of the damping
strength, giving rise an over-damped resonance (reflected by the dispersive behavior). On the
other hand, the resonant behavior observed for the poled samples measured in the
perpendicular direction to the poling direction, as well as in the stressed samples, may be a
consequence of a decrease of the damping strength, giving rise a damped resonance (reflected
by the resonant behavior). It is important to point out that although the dielectric behavior
observed for the samples measured in the parallel direction to the poling direction has a true
resonant-like character, from the previously described analyses, an “apparent” dispersion
behavior (relaxation-like) was observed because of the high component of the damping
coefficient.
Therefore, the gigahertz dielectric anomalies observed in normal ferroelectrics must be
interpreted as a resonance instead of a dispersion mechanism. The question as to whether
these high frequency anomalies behave as a dispersive (relaxation-like) or a resonant process,
described, in general, by an “over-damped” mechanism, may be a consequence of the
coupling between the ferroelectric and the ferroelastic dipolar components, whose
contribution determines the character of the high frequency dielectric dispersion. This aspect
will be discussed in details in the next section.
3.4.3. Discussion
The ferroelectric phenomenon, in most of cases, may be follows by a ferroelastic

behavior, that is to say, a ‘non-intentional’ stress can be present when an electric field E is
applied. As in ferroelectric materials, a ferroelastic crystal contains two or several stable
orientation states when there is no mechanical stress. It is possible to change reversibly from
one state to another by applying a stress σ in defined directions. Thus, there exists a strain-
stress elastic hysteresis with spontaneous strain ss and a coercive stress. In this way, a
transition from a ferroelastic phase to a higher temperature phase, termed paraelastic, occurs
by an increase in symmetry and a change in the crystalline system [52]. The additional ‘non-
intentional’ stress originated when applying an electric field may be sufficient for ferroelastic
switching to occur [53]. The basic assumption of the polarization switching model is that a
single ferroelectric crystallite in a polycrystalline ceramic, which is subjected to an electric
field [54] or to a mechanical stress [55, 56] or both, undergoes a polarization change and a
corresponding strain change. Nevertheless, either partial or complete coupling may exist
between the ferroelectric and ferroelastic properties. When such coupling occurs, ss may be
modified by applying an electric field and Ps may be modified by applying a mechanical
stress. This leads to a set of four hysteresis cycles: polarization–electric field; strain–stress;
strain–electric field; and polarization–stress, as shown in Figure 15 [44]. Crystal growth is
usually accompanied by the formation of ferroelastic and/or ferroelectric domains. The crystal
can be made monodomain by applying a mechanical stress (ferroelasticity), an electric field
(ferroelectricity), or through one of the external actions if coupling occurs between the two
properties. Depending upon the crystal structure and the order parameter, the ferroelectricity
and ferroelasticity in a crystal could be fully or partially coupled.
Figure 15. Schematic representation of the hysteresis loops for ferroelectric-ferroelastic phases [44].
Concerning to the obtained results for the PLT–15 ceramics, it can be noted that two
polarization mechanisms are always present under certain conditions. For all the samples
where an external driving field was applied (a uniaxial stress and/or a poling field applied in
the perpendicular and parallel direction to the measurement direction) a resonant-like
dispersion has been observed. In particular, either uniaxial stress or poling field applied in the
perpendicular direction to the measurement direction promotes redistribution in the
orientation of the dipolar configuration over the perpendicular direction to the measurement
direction. Therefore, the observed resonant behavior, for these conditions, could be associated
to a slightly partial coupling between the ferroelastic and ferroelectric dipolar components,
being the ferroelectric dipolar component the main responsible for the resonance dielectric
response.
On the other hand, for the unpoled and ‘stress free’ samples the observed behavior may
be caused by the coupling between the ferroelectric and the ferroelastic dipolar components.
In these conditions, the dipoles are unable to be reoriented ‘freely’ with the measurements ac
electric fields. Therefore, the dielectric response is mainly governed by the contribution of the
ferroelastic dipolar components, originating the dispersive dielectric behavior.
In order to describe these results, the ferroelastic contributions to the piezoelectric
response should be taken into account [57]. In this case, it will be considered the theoretical
approach in order to investigate the influence of the external poling electric field on the
dielectric response. Thus, the analyses of the data are considered when an oscillating external
electric field is applied parallel or perpendicular to the poling direction. The piezoelectric
effect can be represented by the Eq. (5), where c and e are the converse of the piezoelectric
compliance (or elastic stiffness) and the piezoelectric tensor, respectively, defined as
190 J. D. S. Guerra
[e] = [d ][c E ] [53]. The symbols ν, μ and m correspond to the condensed index of the tensor
notation (ν,μ = 1, 2,…, 6 and m = 1, 2, 3).
σ ν = cνμ
E
sμ + emν Em (5)
In view of the symmetry for a poled ceramic material (∞mm) and considering the
respective symmetry operations, the Eq. (5) can be expanded by considering the probing
electric fields applied parallel and perpendicular to the poling direction (E1 and E3,
respectively) and represented by the following equations:
σ1 = c11s1 + c12 s2 + c13 s3 + e31E3 (6)
σ 2 = c12 s1 + c11s2 + c13 s3 + e31E3 (7)
σ 3 = c13 s1 + c13 s2 + c33 s3 + e33 E3 (8)
σ 5 = c55 s5 + e15 E1 (9)
Some important aspects of the piezoelectric equations (Eqs. (6) – (9)) must be highlighted
when applying a probing electric field in two mutually perpendicular directions. First, when
the probing electric field (E3) is applied parallel to the poling direction (P3), it generates only
compressional/extensional stress (σ1, σ2, or σ3). In contrast, when an electric field (E1) is
applied perpendicular to poling direction, only shear stress (σ5) is generated. The fundamental
difference between both cases is that stress generated by the electric field E3 produces
volumetric changes in the ceramic, while fields applied perpendicularly does not (E1⊥P3)
[58]. In other words, ferroelastic dipoles are modified only by the application of an electric
probing field E3, which is parallel to the macroscopic polarization. It is important to point out
that the elastic stiffness c55, in the Eq. (9), is identical to the coefficient c44, after considering
the respective symmetry operations. In terms of the elasticity theory, the dispersion or
resonant behavior might be associated to the influence of the electric field on the elastic
dipole. In this way, in order to clarify this issue it is necessary to use some basic elasticity
concepts. The behavior of an elastic dipole in the presence of stress can be characterized by
the relation σν = cνμ λμ , being λ the strain tensor, which determines the interaction of
the elastic dipole with the stress field [58]. Formally, σ is defined as the negative stresses
needed to maintain constant the strain per unit concentration of elastic dipoles. Since the λ
tensor represents a strain tensor, it must be symmetric and, therefore, can be characterized by
a strain ellipsoid with three mutually perpendicular principal axes. When expressed in the
coordinate system of the principal axes, the λ tensor becomes diagonal with the three
components λ1, λ2 and λ3 as the principal values. By expanding the relation of the elastic
dipole definition, it is not difficult to notice that results do not depend of the shear
components of stress field. It suggests that the resonant behavior obtained for the
perpendicular direction is governed by the electrical component of the Eq. (5), where only the
shears components of the piezoelectric coefficients play the principal role, without any
contribution of the components along the principal axes (σ1, σ2 and σ3).
Considering that elastic dipoles in displacive ferroelectrics lie parallel to the electric ones,
and that they interact only with the compressional or extensional stress [58], they are excited
only by the application of the poling electric field parallel to the measurements direction. This
suggests that the resonant response obtained for the perpendicular direction is governed only
by ferroelectric components, because only the shear stress components are generated by
piezoelectric contribution, without any ferroelastic contribution. Otherwise, the dielectric
behavior (associated to an over-damped resonance) observed in the parallel direction results
from a coupling between the ferroelastic and ferroelectric contributions. In this case, the
ferroelastic contribution becomes the main responsible for the increase of the damping
coefficient, contributing to the decrease of the characteristic frequency.
It is important to point out that applying a uniaxial stress, parallel to the measurements
direction, the resonant response observed can be explained considering that the ferroelectric
and ferroelastic domains tend to be oriented perpendicularly to the mechanical stress and,
consequently, perpendicular to the measuring electric field direction. As observed, this
condition is analogous to the case of applying a poling electric field perpendicular to the
measurements direction. Thus, either dispersion or resonance mechanism, are always
presented without the presence of an intermediate case. This result confirm that the
ferroelectric and ferroelastic contributions, to the high frequency dielectric anomalies, are
always coupled each other.
4. CONCLUSIONS
In summary, the dielectric microwave properties were investigated in relaxor and normal
ferroelectric ceramics. It was concluded that the vibration of the boundaries of polar regions
is a common mechanism responsible for the microwave dielectric dispersion process in
ferroelectric systems. It was also proposed that the characteristic frequency is controlled by
the ratio between the force constant and the effective mass of such boundaries and the
behavior of fR above the temperature of the maximum of the permittivity reflects the
ferroelectric-type phase transition. On the other hand, the microwave dielectric response of
ferroelectric ceramics was investigated considering the influence of external (electric and
mechanic) driving fields. Two high frequency dielectric anomalies were found in the same
studied material, which were discussed in lights of an over-damped resonant process. It was
confirmed that the obtained anomalies are strongly influenced by the contribution of the
ferroelectric and/or ferroelastic dipolar components, characteristics of ferroelectric materials.
In particular, for all the samples where an external driving field was applied (a uniaxial stress
and/or a poling field applied in the perpendicular and parallel direction to the measurement
direction) a resonant-like dispersion was observed. On the contrary, a dispersive (relaxation-
like) behavior related to an over-damped resonance was obtained for the unpoled (and ‘stress
free’) samples. The results can be well explained by considering the influence of the
192 J. D. S. Guerra
ferroelastic-ferroelectric contributions coupling on the high frequency dielectric response,

which is a common feature for all ferroelectric systems.
ACKNOWLEDGMENT
The author would like to thank to Ferroeletric Ceramics Group (GCFerr), of the Federal
University of São Carlos (UFSCar) for experimental support, especially to Sr. F. J. Picon
(GCFerr) for the technical assistance, and FAPESP (proc. No. 04/09612-0) Brazilian agency
for financial support. The financial support by the ICTP to the Latin-American Network of
Ferroelectric Materials (NET-43) is also gratefully acknowledged.
REFERENCES
[1] Hippel, AV. Dielectric Material and Applications; Technology Press books in Science
and Technology; John Wiley & Sons: New York, 1954, Vol. 1, pp 3-40.
[2] Babbitt, R; Koscica, T; Drach, W; Didomenico, L. Integrated Ferroelectrics, 1995, 8,
65-76.
[3] Zimmermang, F; Voigts, M; Weil, C; Jakoby, R; Wang, P; Menesklou, W; Ivers-Tiffée,
E. J Eur Ceram Soc., 2001, 21, 2019-2023.
[4] Treece, RE; Thompson, JB; Mueller, CH; Rivkin, T; Cromar, MW. IEEE Trans Appl
Supercond, 1997, 7 2363-2366.
[5] Grigas, J. Microwave Dielectric Spectroscopy of Ferroelectrics and Related Materials;
Ferroelectricity and Related Phenomena Series; Gordon and Breach Publisher:
Amsterdam, 1996, Vol. 9, pp 9-108.
[6] Bassora, LA; Eiras, JA. Ferroelectrics, 1999, 223, 285-292.
[7] Haraoubia, B; Meury, JL; LeTraon, A. J Phys E, 1988, 21, 456-460.
[8] Mouhsen, A; Achour, ME; Miane, JL; Ravez, J. Eur Phys J Appl Phys, 2001, 15,
97-104.
[9] Muller, KA; Burkard, H. Phys Rev B, 1979, 19, 3593-3602.
[10] Viana, R; Lunkenheimer, P; Hemberger, J; Böhmer, R; Loidl, A. Phys Rev B, 1994, 50,
601-604.
[11] Venturini, EL; Sâmara, GA; Kleemann, W. Phys Rev B, 2003, 67, 214102.
[12] Bassora, LA. Caracterização dielétrica de materiais ferroelétricos na região de
microondas, Ph. D. Thesis, Federal University of São Carlos, São Paulo, Brazil, 1999,
pp 10-90.
[13] Böhmer, R; Maglione, M; Lunkenheimer, P; Loidl, A. J Appl Phys, 1989, 65, 901-904.
[14] Poplavko, YM; Tsykalov, VG; Molchanov, VI. Sov Phys Solid State, 1969, 10,
2708-2710.
[15] Kamba, S; Bovtun, V; Petzelt, J; Rychetsky, I; Mizaras, R; Brilingas, A; Banys, J;
Grigas, J; Kosec, M. J Phys Condens Matter, 2000, 12, 497-519.
[16] Lente, MH; Guerra, JDS; Eiras, JA; Lanfredi, S. Solid State Commun, 2004, 131, 279-
282.
[17] Lente, MH; Guerra, JDS; Eiras, JA; Mazon, T; Andreeta, MBR; Hernandes, AC. J Eur
Ceram Soc, 2005, 25, 2563-2566.
[18] Cao, WW; Cross, LE. Phys Rev B, 1991, 44, 5-12.
[19] Glazounov, AE; Tagantsev, AK. J Phys Condens Matter, 1998, 10, 8863-8880.
[20] Arlt, G; Böttger, U; Witte, S. Appl Phys Lett, 1993, 63, 602-xxx.
[21] Hippel, AV. Rev Mod Phys, 1950, 22, 221-237.
[22] McNeal, MP; Jang, SJ; Newnham, RE. J Appl Phys, 1998, 83, 3288-3297.
[23] Zhang, L; Zhong, WL; Wang, CL; Zhang, PL; Wang, YG. Solid State Commun, 1998,
107, 769-773.
[24] Maglione, M; Böhmer, R; Loidl, A; Höchli, UT. Phys Rev B, 1989, 40, 11441-11444.
[25] Kazaoui, S; Ravez, J; Miane, JL. J Non-Cryst Sol, 1991, 131, 1202-1205.
[26] Guerra, JDS; Lente, MH; Eiras, JA. Appl Phys Lett, 2006, 88, 102905.
[27] Kersten, O; Rost, A; Schimidt, G. Phys Stat Sol (a), 1983, 75, 495-500.
[28] Hippel, AVZ. Physik, 1952, 133, 158-173.
[29] Kittel, C. Phys Rev., 1951, 83, 458-458.
[30] Turik, AV; Shevchenko, NB. Phys Stat Sol (b), 1979, 95, 585-592.
[31] Tappe, S; Böttger, U; Waser, R. Appl Phys Lett, 2004, 85, 624-626.
[32] Jiang, GQ; Wong, WH; Raskovich, EY; Clark, WG. Rev Sci Instrum, 1993, 64,
1614-1621.
[33] Jiang, GQ; Wong, WH; Raskovich, EY; Clark, WG. Rev Sci Instrum, 1993, 64,
1622-1626.
[34] Jonscher, AK. J Phys D Appl Phys, 1999, 32, 57-70.
[35] Kazaoui, S; Ravez, J; Elissalde, C; Maglione, M. Ferroelectrics, 1992, 135, 85-99.
[36] Petzelt, J; Ostapchuk, T; Gregora, I; Rychetsky, I; Hoffmann-Eifert, S; Pronin, AV;
Yuzyuk, Y; Gorshunov, BP; Kamba, S; Bovtun, V; Pokorny, J; Savinov, M;
Porokhonskyy, V; Rafaja, D; Vanek, P; Almeida, A; Chaves, MR; Volkov, AA;
Dressel, M; Waser, R. Phys Rev B, 2001, 64, 184111.
[37] Böttger, U; Arlt, G. Ferroelectrics, 1992, 127, 95-100.
[38] Arlt, G; Pertsev, NA. J Appl Phys, 1991, 70, 2283-2289.
[39] Garcia, MF; M’Peko, JC; Ruiz, AR; Rodríguez, G; Echevarría, Y; Fernández, F;
Delgado, A. J Chem Educ, 2003, 80, 1062-1073.
[40] Barsoukov, E; Macdonald, JR. Impedance Spectroscopy: Theory, Experiment, and
Applications; John Wiley & Son,s: New Jersey, 2005; Vol. 1, pp 27-42.
[41] Baskaran, N; Ghule, A; Bhongale, C; Murugan, R; Chan, H. J Appl Phys, 2002, 91,
10038-10043.
[42] Zalar, B; Laguta, VV; Blinc, R. Phys Rev Lett., 2003, 90, 037601.
[43] Hornebecq, V; Elissalde, C; Porokhonskyy, V; Bovtun, V; Petzelt, J; Gregora, I;
Maglione, M; Ravez, J. J Phys Chem Sol, 2003, 64, 471-476.
[44] Ravez, J. Chemistry, 2000, 3, 267-283.
[45] Feynman, RP; Leighton, RB; Sands, ML. The Feynman Lectures on Physics;
Commemorative Issue; Addison-Wesley Publishing Co: Redwood, 1989, Vol. 2.
[46] Tagantsev, AK; Glazounov, AE. Phys Rev B, 1998, 57, 18-21.
[47] Bolten, D; Böttger, U; Waser, R. J Eur Ceram., Soc, 2004, 24, 725-732.
[48] Fousek, J; Brezina, B. J Phys Soc Jpn, 1964, 19, 830-838.
[49] Robels, U; Arlt, G. J Appl Phys, 1993, 73, 3454-3460.
[50] Lu, W; Fang, D. N; Li, CQ; Hwang, KC. Acta Mater, 1999, 47, 2913-2926.
194 J. D. S. Guerra
[51] Hwang, SC; Waser, R. Acta Mater, 2000, 48, 3271-3282.

[52] Aizu, K. J Phys Soc Jpn, 1969, 27, 387-396.
[53] Abplanalp, M; Fousek, J; Günter, P. Phys Rev Lett, 2001, 86, 5799-5802.
[54] Hwang, SC; Lynch, CS; McMeeking, RM. Acta Metall Mater, 1995, 43, 2073-2084.
[55] Hwang, SC; Huber, JE; McMeeking, RM; Fleck, NA. J Appl Phys, 1998, 84,
1530-1540.
[56] Wooster, WA; Wooster, N. Nature, 1946, 157, 405-406.
[57] Nye, JF. Physical Properties of Crystals: Their Representation by Tensors and Matrices;
Oxford University Press: New York, 1985; Vol. 1, pp 3-168.
[58] Nowick, AS; Berry, BS. Anelastic Relaxation in Crystalline Solids; Materials Science
Series; Academic Press: New York, 1972; Vol. 1, pp 156-224.
Chapter 5
THE PRINCIPLE THAT GENERATES CONFIGURATION

IN ANIMATE AND INANIMATE SYSTEMS –
A UNIFIED VIEW
Antonio F. Miguel
Geophysics Centre of Evora, Rua Romão Ramalho 59, 7000-671 Evora, Portugal
Department of Physics, University of Evora, PO Box 94, 7002-554 Evora, Portugal
ABSTRACT
The generation of flow configuration (shape, structure, patterns) is a phenomenon
that occurs across the board, in animate and inanimate flow systems. Scientists have
struggled to understand the origins of this phenomenon. What determines the geometry of
natural flow systems? Is geometry a characteristic of natural flow systems? Are they
following the rule of any law? Here we show that the emergence of configuration in
animate flow systems is analogous to the emergence of configuration in inanimate flow
systems, and that features can be put on a unifying theoretical (physics) basis, which is
provided by the constructal law.
All scientific endeavors are based on the existence of universality, which manifests
itself in diverse ways. Here we also explore the idea that complex flow systems with
similar functions have a propensity to exhibit similar behavior. Based on this thought
relationships that characterize animate systems are tested in relation to cities and
countries, and some conclusions are drawn.
1. NATURE'S SHAPES
From the ancient time people have struggled to explain why animals, plants, rivers, etc.,
are shaped the way we find it. Democritus (460-370 B.C.) attributed natural configuration
(shape, structure, patterns) to "chance and necessity." [1]. Aristotle (384-322 B.C.) claims that
shape is more truly than matter. He wrote: “Into what then does it grow? Not into that from
which it arose but into that to which it tends. The shape then is nature” [2]. In other words,
nature is the shape of a fully matured natural object. Therefore, we should understand not
196 Antonio F. Miguel
only matter but especially the shape. He exemplifies “as the teeth, for example, grow by
necessity, the front ones sharp, adapted for dividing, and the grinders flat, and serviceable for
masticating the food”. In 1860 Ralph Waldo Emerson (1803-1882) published the essay
entitled "The Conduct of Life". He wrote [3] "Nature has her own best mode of doing each
thing, and She has somewhere told it plainly, if we will keep our eyes and ears open." In
Kant's (1724-1804) book entitled "Critique of Judgment " he argues that “the forms of nature
are so manifold (…) that there must be laws for these forms (...), if they are to be called laws
(as the concept of nature requires), they must be regarded as necessary virtue of a principle of
the unity of the manifold” [4].
It has been more than 250 years since Pierre de Maupertuis (1698-1759) wrote "(…) laws
of movement and of rest deduced from this principle being precisely the same as those
observed in nature, we can admire the application of it to all phenomena. The movement of
animals, the vegetative growth of plants ... are only its consequences; and the spectacle of the
universe becomes so much the grander, so much more beautiful, the worthier of its Author,
when one knows that a small number of laws, most wisely established, suffice for all
movements. (…)" [5]. This “principle” - the “principle of least action” [6] – was properly
stated later by scientists like William Hamilton. It came from his idea that the very perfection
of the universe demands a certain economy in nature and is opposed to any needless
expenditure of energy. In spite of having a wide applicability in mechanics, electricity,
magnetism and quantum mechanics, this principle only accounts for point-to-point motion but
cannot describe point-to-area and point-to-volume flows [6].
In 1776, Jean Meusnier's study of soap films (very popular among mathematicians of
eighteen century) showed an example in geometric optimization [7]. An ordinary two-
dimensional plane can be twisted infinitely into a helicoid shape (minimal surface) that has
the delicate balance of a soap film. Forty years later, Robert Stirling (1790-1878) patented his
“Heat Economiser” and Sadi Carnot (1796-1832) wrote about the ideal configuration of a
heat engine. Charles Darwin (1809–1882) busied himself with the idea that the configuration
of living systems is determined by evolution and natural selection [8]. Josiah Willard Gibbs
(1839-1903) in the late 19th century refers that a thermodynamic system to be in an
equilibrium state it will configure its components by minimizing the energy. In 1917, D'Arcy
Wentworth Thompson (1860-1948) published the book “On Growth and Form”. Thompson
introduced the idea that they are principles among quite diverse forms of life. From the
observation that the bones of a museum skeleton would lie in a heap on the floor without the
clamps and rods pulling them together, he concluded that the tension holds the skeleton
together as much as weight does. He wrote that "the form of any particle of matter, whether it
be living or dead, and the changes in form which are apparent in its movements and in its
growth, may in all cases be described as due to the action of force" [9]. Forces of tension,
compression and shear occurred in all living structures and influenced both growth and
function. Eighty later, Helbing and Molnar [10] use the concept of “force” to explain the
motion of pedestrians (the so-called “social forces model”).
In 1975, Benoit Mandelbrot introduced the term fractal and wrote “(…) a fractal is a
shape made of parts similar to the whole in some way (…).” [11]. Fractal based description of
natural systems has been widely applied. In spite of its incontestable importance, fractals do
not account for dynamics hence are descriptive rather than predictive.
Natural flow systems are complex and change (evolve) in many ways. Why things in
nature are shaped the way they are? Why do tree-shaped designs occur in river basins, deltas,
The Principle that Generates Configuration in Animate and Inanimate... 197
lightning and lungs? Is it an optimized behavior? Why stony corals and bacterial colonies
present intraspecific variability of patterns? What are the “forces” that shape their growth and
form?
Adrian Bejan is at the origin of the constructal paradigm, which had its start in 1996 [12-
14]. He arrived at this idea from a problem of minimizing the thermal resistance between an
entire heat generating volume and one point [13]. The occurrence of configuration in nature is
a physics phenomenon, and the constructal theory is about the physics principle from which
configuration can be deduced. It is based on the following law (constructal law): “For a finite-
size system to persist in time (to live), it must evolve in such a way that it provides easier
access to the imposed (global) currents that flow through it.” [13]. This law states that there is
a time arrow “associated with the sequence of flow configurations that constitutes the
existence of the system”[14,15] (Figure 1). Besides, the system shape and structure do not
develop by chance but result from the permanent struggle for better performance. Better
performance means minimization and balance of the resistances (i.e., imperfection) faced by
the various internal and external streams under the existing constraints. The morphim
structure is the result of optimal distribution of imperfection.
Figure 1. Time arrow in plant growth [18]
Configuration plays a fundamental role in models used for “perceiving” and

“understanding” nature. Here, it is shown how the constructal theory provides a unifying
picture for the development of flow architectures in natural systems with internal flows. We
also show that the complexity is optimized and is a result of the optimization process.
2. THE CONSTRUCTAL LAW

A flow system (animate and inanimate) is a nonequilibrium thermodynamic system [13-
15]. Classical thermodynamics is not concerned with the configurations of its nonequilibrium
systems. Basically, the first law of thermodynamics is a statement of the conservation of
energy, the second law is a statement about the direction of that conservation (i.e., the
maximization of entropy generation in an isolated system), and the third law is a statement
about reaching absolute zero. The constructal law is a law of thermodynamics distinct of
these three [16]. Likewise the second law it contains an arrow of time but of different things.
The second law proclaims the entropy of a closed system always increases in time up to an
equilibrium state. The constructal law states that the flow architecture morphs in time in the
direction of flowing more easily (i.e., the maximization of flow access). The analogy between
the formalism of equilibrium thermodynamics and that of constructal theory is presented by
[16,17].
Constructal theory begins with the global objective(s) and the global constraint(s) of the
flow system [12-17]. There are two global constraints, one external and the other internal: the
external constraint is the system size, and the internal global constraint is the "amount"
invested in making the flow architecture (e.g., total volume of all the ducts of the flow
structure). How do we identify the configuration that brings flow architecture to the best
performance?
According with constructal law, the architecture of the system must “provides easier
access to the imposed (global) currents that flow through it.” [13]. Consider, for example, the
case of a fluid that has to be drained from a finite-size area or volume. The volume is a
nonhomogeneous porous medium composed of a material of low permeability (high
resistance) and various layers of higher permeabilities. The goal is the optimal arrangement of
these layers such that the global flow resistance is minimal. The global constraints are the
system size (area or volume to be drained) and the total size occupied by the layers. Based on
this, the arrangement is optimized in order to reach maximum performance. In this way, the
designer “constructs” the optimal flow architecture. A detailed description is provided by
[13,14]. The acquisition of shape (architecture) is an evolutionary process not assumed in
advance or postulated. Therefore, the flow architecture (shape) of the system is the result of
the optimum balance between two competing trends – slow (high resistivity) and fast (low
resistivity). If we increase the length of high permeability layer leads to a decrease in the
resistance posed to flow in the area (volume) occupied by low permeability layers, but it also
increases the resistance along the high permeability layer. Therefore, flow optimization is as a
trade-off between competing trends [13-16].
This example is not unique. Behind a broad class of processes in the natural sciences is
the dynamics that combines, for example, Brownian motion (diffusion) with some form of
deterministic drift (convection). Here we shall consider a convection–diffusion equation of
the general form
∂n
+ ∇.(Un ) = D∇ 2 n (1)
∂t
where n denotes the density, U is the velocity and D is the diffusion coefficient. The
spreading of a tracer or a solute, and the transport of heat or fluid are examples that can be
analyzed within the framework of diffusive-convective phenomena. Diffusion is associated
with a high resistivity mechanism, whereas convection is low resistivity. Why the flow
architecture of systems results of the balance between both mechanisms?
By applying scale analysis [13] to Eq. (1), we obtain the time scales for diffusion (tdf) and
convection (tcv)
L2
t df ~ (2)
D
L
t cv ~ (3)
u
Diffusion coefficients are usually much smaller than 1 m2 s-1 (for example, the diffusion
coefficient for oxygen in air is approximately 2 × 10-5 m2 s-1). For very short distances tdf < tcv
and diffusion is the “best” driving mechanism because is “faster”. On the other hand, for
larger distances tdf > tcv and convection perform better. The time of transition from diffusion
to organized flow is t* ~ D/u2. This simple analysis shows us that the two modes of flowing
with imperfection (resistance) should exist in order to enable the best flow architecture.
In summary, optimization means finding the best allocation of resistances (i.e., minimum
imperfection to the global flow architecture), and therefore the configuration of the system is
the one that allows best flow access. To make our point, we illustrate next the application of
the constructal law to a variety of animate and inanimate flow systems.
3. CONSTRUCTAL THEORY OF INANIMATE NATURAL FLOW SYSTEMS

The phenomenon of generation of flow configuration is everywhere in inanimate flow
systems. Agglomerates of aerosol particles often have dendritic shapes instead of spherical
shapes [18]. The reason why this occurs was not clear. Reis et al. [19] relied on the
constructal law of maximization of flow access in order to construct a theory of aerosol
agglomeration. Based on the idea that there are not electrically neutral surfaces in contact
with air [18,19], it is assumed that the forces that make aerosol particles stick onto previously
deposited particles are of the electrical type. There are two possible “modes” of
agglomeration: spherical and conically. The volume of spherical agglomerates is given by
Vsph ~ K 2 t 2 (4)
while the volume of agglomerates of particles with the conical shape is
1/ 2
⎛ q el K14 / 3 ⎞
Vcon ~ ⎜⎜ ⎟⎟ t7 / 3 (5)
⎝ μ el ⎠
where μel is the dipole moment, qel is the charge and K is a quantity function of particle size,
dipole moment, electric charge, Cunningham correction factor, electric permittivity of the air,
surface density of charge and air viscosity [18]. The constructal law is simply about the
physics meaning of the time direction of configuration evolution. Therefore, the architecture
of the aggregate of particles evolves in time in such a way that the global rate of accumulation
of the particles is maximized. In other words, the best configuration is the one that bring the
entire flow system (ambient and particles) to equilibrium in the fastest way possible.
Figure 2. Time evolution of the volume corresponding to conical and spherical agglomerates
The temporal evolution of the accumulation volume is presented in Figure 2. This plot
shows that at the critical time, tct, the volume of conical agglomerates overtakes the volume of
spherical agglomerates. According to Eqs. (4) and (5)
1/ 3
⎛ μ ⎞
t ct ~ ⎜⎜ 1 / 2 el 1 / 3 ⎟⎟ (6)
⎝ q el el K ⎠
This means that the agglomerate must first grow as a sphere (t < tct) and then change to a
conical shape. Existing flow configuration (spherical) is replaced by a configuration (conical,
tree-shaped) that flow more easily. Experimental measurements reported in the literature
confirm the main features of this constructal development [18].
The constructal law also predicted configuration generation during liquid droplet impact
on a wall [20]. For small and slow enough droplets the splat comes to rest viscously, as a
disk. On the other hand, for large and fast enough droplets, it splashes by developing needles
that grow radially until they are arrested by viscous effects. Based on constructal theory,
Bejan and Gobin [20] gave comprehensive explanation of configuration generation and also
present a dimensionless group that governs the selection of geometry. This group is defined
by the ratio of two lengths, the final radius of the disc that dies viscously, divided by the
radius of the still inviscid ring that just wrinkles. The results of the optimization process
match the observed values.
A river basin is the portion of land drained by a river and its tributaries. It encompasses
the entire land surface dissected and drained by many streams and creeks that flow downhill
into one another, and eventually into one river. The final destination is an estuary or an ocean.
A river basin is an example of an area-to-point flow. Based on constructal law, Bejan [13] has
addressed this problem and optimized the channel network that minimizes the overall
resistance (imperfection) to flow. Consider the area allocated to each smallest stream of the
river basin. Rain falls uniformly on the elemental area with a mass flow rate. There is an
optimal elemental shape defined as the ratio between the length and width such that the total
flow rate collected on the elemental area flows with least global flow resistance from the area
through one point on its periphery. The optimized area element becomes a building block
with which larger rain plains can be covered. The tree-shaped flow architecture of the river
basin represents the best allocation of resistances (optimal distribution of imperfection), and
therefore the configuration of the system that allows best flow access from the area to the
outlet [13,18]. The flow architectures of river basins are not the result of change but they
constitute the optimal configuration.
River basins have geometrical features which can be measured, namely the area
(measured on the vertical projection), the elongation ratio (the diameter of a circle with the
same area as the drainage basin, divided by the basin maximum length), the relief (the
difference of elevation between the highest and the lowest points of the drainage area) and the
relief ratio (the basin relief divided by the maximum length of the basin). Based on
experimental data of geometric characteristics, authors proposed several allometric scaling
laws to describe feature of river basins [18]. The well known Horton, Melton and Hack
allometric laws can be anticipated based on constructal law as a result of minimization of the
overall resistance to flow. Reis [21] shows that the: (i) ratio of constructal lengths of
consecutive streams matches Horton's law for the same ratio, while the same occurs with the
number of consecutive streams that match the respective Horton's law; (ii) Hack’s scaling law
is also anticipated from the constructal law and the exponent β that relates the length of
mainstream, Ls, and the basin area, Ab, (Ls~Abβ, β~0.568) presented in a more accurate way
(i.e., β=2τ+1/4τ, where τ is the order of the river basin); and Melton's scaling exponent is
2.45 instead of 2.
Reis and Gama [22] relied also on the constructal law of maximization of flow access in
order to address beachface adjustment as a response to wave swash forcing. They showed that
beachface slope varies with wave height raised to the power 3/4, and sand grain size raised to
the power 4/3.
The largest flow system on earth (atmosphere circulation) was studied from the point of
view of the constructal theory [23,24]. The sun–earth–universe assembly was viewed as a
power plant the power output of which is used to force the atmosphere and hydrosphere to
flow. The constructal optimization was performed in order to deliver the latitude of the
boundary between the Haddley and the Ferrel cells, the boundary between the Ferrel and the
Polar cells, the average temperature of the Earth surface, the convective conductance in the
horizontal direction, as well as other parameters defining the circulation and the Earth surface
temperature. The results of this optimization agree very well with the observed values.
These results and many other examples show that the constructal law explains much of
shape and structure of inanimate flow systems around us. The cracks patterns evolution
during shrinking in soil [13,25], the dendritic crystals formed during rapid solidification [13],
the turbulent flow structure [13], Rayleigh-Bénard convection [13], and the electrokinetic
transfer through porous media [26] may also be mentioned here among other examples.
4. CONSTRUCTAL VIEW OF ANIMATE FLOW SYSTEMS

Animate systems are probably the most complex and diverse system in the universe. The
so-called “life” covers more than 27 orders of magnitude in mass from molecules of the
genetic code to whales and sequoias [27]. To help understand the underlying configuration
(architecture) behind animate systems, three different views (concepts) are employed:
(i) An animate system is made up of interacting subsystems that provide a key

functional characteristic of the overall system. Subsystems such as the circulatory
system, lungs, kidneys, etc. provide a function that is essential to sustain the all
system.
(ii) Colonies of living organisms (e.g., stony corals and bacterial colonies) refers to
several individual organisms of the same species living all time closely together for
mutual benefit.
(iii) Living organisms that are temporally together as an ongoing group to collectively
accomplish a particular purpose (e.g. pedestrians accessing shopping facilities or a
stadium).
Systems at these levels of organization demonstrate the complementary nature of

structure and function. Therefore, the design of animate systems is not only a matter of
molecular biology but also of geometry and physics. To make our point, we review a
collection of studies based on constructal theory that explain in a simple manner the
configuration that occurs in (i), (ii) and (iii).
4.1. Systems View
An animate system is made up of interacting subsystems. Each subsystem provides a key

functional characteristic of the overall system. Are they optimized as a whole? Are the
subsystems also optimized?
The respiratory subsystem was assessed via the constructal principle by Reis et al. [28]
and Reis and Miguel [29], with the structure of the respiratory system being the result of this
appraisal. The primary function of the respiratory system is the supply of oxygen to the blood
and the drainage of carbon dioxide from it, so these in turn delivers oxygen to and remove
carbon dioxide from all parts of the body. There are two alternatives for accomplishing this
purpose: the lung could be a duct system, or a simple single sac (volume) open to the external
air from which the oxygen diffuses to the blood (and carbon dioxide diffuses after being
released from the blood). This second possibility is clearly noncompetitive as compared to a
duct system: the former has a small access time for duct (convective) flow (Eq. 3) of order tcv
~ 1 s ( u ~ 0.5 m/s) whereas the latter has an access time for a diffusive process (Eq. 2) of
order tdf ~ 104 s (L ~ 0.5 m and D ~2 x10-5 m2/s). On the other hand, both solutions have
internal imperfections. A duct system has a large friction resistance to airflow whereas the
single sac has a large diffusive resistance. Tree-shaped flow architectures are the easiest way
to flow between infinity of points (volume, area) and one point or vice-versa [13]. However, a
cavity (alveolar sac) at the end of the tree-shaped network should exist, because for very short
distances diffusion is the “best” driving mechanism (Section 2) and allows easy oxygenation
of the blood from air to the tissues. What are the characteristics of lung configuration to
provide the easiest access for blood to oxygen/carbon dioxide?
According to the constructal principle, lung configuration must provide the easiest access
for blood to oxygen/carbon dioxide under the constraints posed by the space allocated to the
respiratory process (chest cavity). Reis et al. [28] and Reis and Miguel [29] show that the
configuration that performs these functions at the lowest flow resistance is a tree-shaped flow
structure composed of ducts with 23 levels of bifurcation that ends with alveolar sacs from
which oxygen diffuses into the tissues. In summary, lung configuration results of the trade-off
between two competing trends. The configuration that performs best is the one that results of
the “harmony” of the best allocation of resistances to allow the best flow access and not from
the smash of any of the competing trends.
In these studies [28,29], the dimensions of the alveolar sac, the total length of the
airways, the total alveolar surface area and the total resistance to oxygen transport in the
respiratory tree were also obtained. One of the most remarkable findings was that there is a
length defined by the ratio of the square of the first airway diameter to its length which is
constant for all individuals of the same species. Kidneys, vascularised tissues and the nervous
system are also examples of optimized architectures that have been treated from the point of
view of constructal theory [13,16,30].
It’s not only the subsystems that composed the animate systems that are optimized but
also the whole system. In the 1940s, Max Kleiber [31] and Samuel Brody [32], based on
observation of mammals and birds, established for the first time that the interspecies
correlation of metabolic rate scales as 3/4 power of the body mass. Kleiber's and Brody’s
works were generalized by subsequent researchers to intracellular levels, unicellular
organisms, and plants, and an exponent of 3/4 is found over 27 orders of magnitude [33].
A direct explanation for the 3/4 scaling power between metabolic rate and body mass can
be obtained based on constructal theory. The explanation presented by Bejan [34] was
obtained by combining the tree architecture optimized for minimum pumping power and the
convective heat transfer characteristics of two identical fluid trees superimposed in
counterflow.
Constructal theory also anticipates other important empirical allometric laws. A direct
explanation for the 1/4 scaling power between breathing or heart beating time and body mass
was obtained from the minimization of the pumping power required by the thorax for
breathing and the heart for blood circulation. This allometric law derived in [35] is based on
the following realistic assumption: the flow is intermittent (in and out) and “in” interval is of
the same order of magnitude as the “out” interval. Moreover, Bejan [35] also proved that the
heartbeat and the breathing time must be of the same order of magnitude, regardless of body
size.
The allometric laws of the design of the hair coats of animals are also projected by the
constructal theory. By minimizing the combined heat loss by conduction and radiation
through the hair air coat, the proportionality between the hair strand diameter and the animal
body length scale raised to the power of 1/2 , and the hair coats porosity between 0.95 and
0.99 for all animal sizes were obtained [13,36].
Flow configuration (patterns) is generated in space but also in time. The modes of
locomotion (flying, running, swimming) are examples that illustrate the generation of patterns
in time. Bejan [13] and Bejan and Marden [37,38] showed that the constructal law (a single
law) is able to describe the different modes of locomotion. Flying, running and swimming
were attributed to the same principle of configuration generation for greater flow access in
time (constructal law) [37,38]. Flying, for example, involves two losses. One loss is the lifting
of the body weight (vertical loss due to the gravity) and the other is the horizontal loss due to
drag. The total loss per distance travelled is the summation of both losses, i.e.,
Mg 3 / 2 h1 / 2 v −1 + C Dρair v 2 L2b , where M is the body mass , v is the velocity, ρair is the air
density, CD is the drag coefficient, h is the vertical distance and Lb is the characteristic
dimension of the body. The optimal flying speed is obtained by minimizing the total loss,
which is provided by the constructal law. Therefore, the flying speed scales as 1/6 power of
the body mass and flapping frequencies are proportional to the body mass raised to the power
−1/6. These results constitute allometric laws for flying and they are in good agreement with
the speeds of flying organisms ranging from the insects to the big birds. The minimization of
horizontal (friction) and vertical (gravity) losses is also at origin of running and swimming.
More details are provided by [37,38].
Recently, Charles and Bejan [39] collected the heights and weights of the fastest
swimmers (100 m-freestyle) and sprinters (100 m-dash) for world record winners, since the
beginning of the twentieth century. They then plotted the speed data (based on wining times)
versus the size of these athletes. The scaling analyze showed that speed records will continue
to be dominated by heavier and taller athletes. According to [39], this tendency is attributable
to the scaling rules of animal locomotion [37,38], “not to the contemporary increase in the
average body size of humans” [39]. The authors also suggest that if athletes of all sizes are to
compete in these kinds of events, weight classes might be needed, like in certain sports, such
as judo, boxing, weightlifting or wrestling.
The constructal theory was applied to the design of animals but also to predict the
morphology of plants. Bejan and co-authors [40] showed, among other things, that the tree
length is proportional to the wood mass raised to the power 1/3, the tapered shape of the root
and longitudinally uniform diameter and density of internal flow tubes, the near-conical shape
of tree trunks and branches, and the existence of an optimal ratio of leaf volume divided by
total tree volume.
In summary, animate systems are optimized flow architectures. Besides, allometric laws
can be viewed as a manifestation of the underlying dynamics that is optimized, and can be
anticipated by the constructal theory.
4.2. Colonies of Living Organisms
A colony refers to several individual organisms of the same species living closely
together, usually for mutual benefit (e.g., formation of colonies helps escaping predators or
enhance the ability to locate nutrients) despite some detrimental effects such as getting
infected more easily by contagious disease [27]. The formation of dissimilar patterns inside
similar colonies under different environmental conditions is especially intriguing. It seems
that there is an order underlying the apparent variety of shapes. Several authors noticed that
for a given set of growth conditions, the colonies experience similar patterns that can be
reproduced from the experimental point of view [40-42].
The shape of bacterial colonies depends both on the nutrient level and gel (environment)
hardness [41,42]. Bacterial colonies that cope with hostile environmental conditions (i.e., low
level of nutrients or hard agar surface) develop branched forms. On the other hand, colonies
that cope with high levels of nutrients develop a compact shape. Stony corals also consist of
structure of tightly interconnected individuals (called polyps). These corals collected from
exposed growth sites, where higher water currents are found, present more spherical and
compact shape than corals of the same species growing in sheltered sites, which display a
thin-branched patterns [43]. There are 2 questions that need to be answered: how colony
development displays a precise control of the configuration? why colonies develop

characteristic patterns?
Increasing complexity is associated with an increased use of modulatory communication
signals to organize cooperative behavior. These signals do not elicit specific responses in
themselves, but rather operate in a general manner to alter the probability that individuals will
respond to other stimuli [41,44-46]. Ants and other social insects, for example, usually have a
pheromone produced by the endocrine system for marking trails to food [44]. Therefore,
pattern formation requires modulatory communication signals (chemical signals, electrical
impulses or other) between interconnected individuals. According to [41], communication is
on the basis of configuration (pattern) formation in bacterial colonies. This answers our first
question but the following question remains: what is the reason to switch to different
patterns?
The answer is delivered by the constructal law in Miguel [47] and Miguel and Bejan [48].
Bacterial colonies or stony corals may develop branched (tree shaped) or spherical shapes in a
differentiated response to the variability of environmental conditions. Branched and round
massive patterns have different abilities to fill the same space, and thus a different ability to
harvest the nutrients that are available. According to the constructal theory, the survival of
flow systems calls for patterns that promote flow access. The preferred pattern is the one that
allows the living system to deplete the nutrients as fast as possible.
Consider s to be the characteristic external length scale of the living system, and l the
branch (needle) length scale width (l<<s). The volumes of branched and spherical patterns
scale as ~ l2s and ~s3, respectively. Because l2s << s3, spherical patterns are more effective for
filling the space (i.e., the most effective for extracting the available nutrients). In this sense,
spherical means perfection, but in reality branched patterns are also likely to occur. How can
one reconcile such a contradiction?
Consider, for example, stony corals growing at a rate, v, of a few cm/year (for
Pocillopora damicornis the growth speed is 1–6 cm/year [47,48]). When the coral grows in a
sheltered site, where water currents are practically absent, diffusion is the most important
nutrient transport mechanism. The living system starts to grow at birth (t=0). Immediately
after, nutrients close to the system are quickly consumed and depleted. The concentration of
nutrients decreases, which triggers a wave of nutrients defined by the characteristic linear
dimension ~ (Dnt)1/2 (Eq. 2) and the speed of propagation ~ (Dn/t)1/2, where Dn is the diffusion
coefficient for nutrients. The initial speed of propagation is greater than any growth speed of
the living system, but decreases with the inverse of the square root of time. Consequently, the
speed of nutrient propagation drops below the growth speed of the living system when t >
Dn/v2. Before t ≤ Dn/v2, the round (massive) shape is the most effective arrangement for
filling the flow space. After this time scale, the living system begins to grow outside the
nutrient diffusion region (Figure 3). Branches (bio-lanes) develop in order to generate low-
resistance paths for nutrient access, in a phenomenon that resembles the respiratory tree. This
channeling enables the system to continue to experience growth inside the nutrient rich region
from t1critical to t2critical. At times slightly greater than t2critical the coral sticks out of the nutrient
region. New branches grow forward in order to access the nutrient region until a new critical
time is reached (Figure 4). Each branch generates a new group of branches, and the global
feature of this scenario is the dendritic pattern. This configuration represents the most
competitive configuration under these circumstances. In summary, existing flow
configurations are replaced by configurations that flow more easily.
A similar explanation can be applied to bacterial colonies. A Petri dish prepared with
nutrient agar is the place for the bacteria you'll be growing. The nutrient content and the agar
hardness are important parameters because control the colony pattern. Why?
The survival of colony calls for patterns that promote easier flow access (constructal
law). If the agar surface is soft and nutrient abundant, the colony develops a spherical pattern
because, as showed before, is the more effective architecture for filling the space (i.e., the
most effective for extracting the available nutrients). In a low-nutrient environment, nutrients
close to the system are quickly consumed and depleted. In a hard agar surface, both the
resistance of colony and nutrient to movement is high. In both cases, the contact surface of
colony should be larger and the bacteria should “follow paths” of less resistance, to continue
to experience growth inside the nutrient region. For both a tree shaped pattern represents the
most competitive configuration [13], the optimal shape for survival in nature.
In summary, corals and bacteria colonies evolve in time as the result of the continuous
search for easier flowing configurations. Existing flow configurations are replaced by
configurations that flow more easily. Therefore, patterns do not develop by chance when the
flow configurations are able to morph in time.
Figure 3. Time evolution of characteristic dimensions of living system and nutrients: the speed of
nutrient propagation falls behind the growth speed of the living system at t > Dn/v2
Figure 4. Time evolution of characteristic dimension of a living system and critical times: coral
branches at critical times t1critical, t2critical and t3critical.
4.3. Living Organisms that are Temporally Together as an Ongoing Group
The formation of dissimilar patterns inside a similar group is not an exclusive of coral
and bacterial colonies. Walking is the most basic form of transportation, undertaken by
almost every citizen. Systematic observations of pedestrian revealed that in standard
situations (for instance, running to catch a departing bus; panic situation excluded) pedestrian
motion reveals regularities. Figure 5 shows the empirical walking speed – pedestrian density
relation according to data from several authors [49]. These data show that pedestrians are able
to walk fast and slow. If the interpersonal distance between the pedestrians is large enough (or
the density of pedestrians is small enough) the motion is not disturbed and pedestrians walk
with a constant walking speed (domain I). For a density higher than ~1 pedestrian/m2, the
walking speed decreases with the pedestrian density (domain II). Recently, Miguel [49]
suggested that each pedestrian at domain II may be modeled as a “particle” subjected to some
“effective forces” arising out of its interaction with the obstacles and the other pedestrians. He
consider two sub-domains in this domain II: (i) a sub-domain which the pedestrians’
interpersonal distances are large enough and the deviation from the desired speed is due to a
necessary deceleration to adjust their own speed to the speed of neighboring pedestrians, and
(ii) a sub-domain which the interpersonal distances between pedestrians are small enough that
“repulsive forces” will affect the walking speed. Humans (but also birds, fishes or other) in a
group cannot anticipate the behavior of the other elements. The decrease of walking speed in
domain II helps to maintain the so-called “territorial effect” and avoid or reduce physical
contact with other pedestrians. In this study [49], the optimal flow capacity of a system (i.e.,
maximum pedestrian flow) and the optimal level of service provided to the pedestrian (i.e.,
minimum travel time) were also analyzed based on the constructal theory.
Figure 5. Walking speed versus pedestrian density (empirical data)
Observations performed by different authors [13,50] show that in crowded spaces the
movement of individuals self-organizes naturally into lanes with a specific direction. In
addition, when a stationary crowd stands in the way and needs to be passed through,
pedestrians organize themselves into river-like streams. They act together, more or less
automatically, to accomplish a task. How and why they develop these patterns?
Such organized collective behavior is based on a coordinated and orchestrated use of
communication between pedestrians. Pedestrians rely on their senses to communicate. Similar
to pheromones widely used by social insects such as ants, pedestrian must rely on their vision
for spatial orientation. Vision guarantees a precise control of the pedestrian position in space
and consequently, thereafter, a precise pattern control. But why pedestrians develop specific
patterns when confronted with specific situations?
Consider pedestrian that proceed from one point to every point of a finite-size territory
(e.g., a square, a shopping mall, etc.). According to the constructal law, the emerging
architecture will be the one that promotes the easiest flow of pedestrians. Eq. (1) reveals that
are two flow mechanisms to accomplish this purpose: diffusion (slow flow or high resistivity)
and streams (fast flow or low resistivity). These two competing mechanisms provide greater
flow access than one mechanism alone. What is then the best mechanism that drives
pedestrian movement to evolve in space and in time?
To estimate diffusion and streams time scales (Eqs. 2 and 3), we need the pedestrian
diffusion coefficient (D) and the pedestrian walking speed (u). Several experimental studies
(see for example [50]) provide valuable data to obtain these quantities. According to field
surveys the pedestrian walking speed is about 1,34 m/s. Worth noting is that pedestrians in a
shopping mall or busy city street exhibit a speed that is related to the free area available
around each individual. From an analogy with kinetic theory in Miguel and Bejan [48], the
pedestrian diffusion coefficient is related to the walking (random) speed and the mean free
interpersonal distance via the Einstein-Smoluchowski equation. On the basis of the empirical
data, the relation between the pedestrian diffusion coefficient, D, and the mean interpersonal
distance, λ, can be established as
D = 1.97λ − 0.61 0.31≤λ≤3.16 m and L>0.31 m (7)

Here L is the distance to access. Why does pedestrian crowd motion exhibit dissimilar
patterns (diffusion and streams)? When the time of transition from diffusion to organized
flow (streams) is t* ~ 1.1λ-0.34, both diffusion and streams promote the easiest flow of
pedestrians. However, diffusion is more effective than streams if tdf < t* (streams perform
better if tcv < t*). So, diffusion is more effective when λ > 0.68 L + 0.31 or L < 1.47 λ - 0.45.
This means that the diffusion is only more competitive for large mean free interpersonal
distances between pedestrians and for accessing small distances. On the other hand, streams
are more competitive for small mean free interpersonal distances between pedestrians and for
accessing large distances. Therefore, pedestrians follow the configuration that flows more
easily. These results are in agreement with empirical observations that pedestrians can diffuse
among themselves only at very small pedestrian densities [48,50].
Another striking phenomenon in pedestrian dynamics is the occurrence of spontaneous
lanes of pedestrians with uniform walking speed in very crowded open spaces. When facing a
stationary crowd, pedestrians spontaneously self-organize in river-like streams (''rivers of
people'' [13]) in order to penetrate the crowd [50]. Why do pedestrians exhibit such self-
organization?
The constructal approach begins with recognition of a purpose. In line with this approach,
pedestrians move in configurations that provide greater access. According to Miguel and
Bejan [48], the argument of minimum resistance led to the conclusion that streams/lanes
provide the best flow access as long as the mean free interpersonal distance between the
pedestrians is less than ~ 0.8 m (or density larger than ~ 1.6 pedestrians per square meter).
Therefore, configuration evolves in the direction of greater flow access. This result explains
why pedestrians self-organize into lanes when their density is high enough, as observed by
Helbing et al. [51].
The occurrence of rivers of people through stationary crowds can be predicted from the
constructal law in the same way as the structure of river basins [13,14,16]. The stationary
crowd is the river basin, and the space vacated by the crossing pedestrians is the eroded river
bed. Each pedestrian opens a small space toward the stationary crowd, thereby creating the
conditions for a successor to follow. The next pedestrians follow those who are already in
motion, giving rise to organized lines that form through the crowd. The channels due to the
coalescence of such paths are tree-like structures similar to river branches. The physics
meaning of the evolution of the configuration is the greater flow access (constructal law).
In crowds that panic, the above formulation is not applicable because the lanes and the
river-like streams of people are destroyed: individuals do not know the right way to escape.
They strive to go forward, thereby reducing interpersonal distances, which might induce
interpersonal contact (collision) or even loss of balance.
5. CITIES/COUNTRIES AND THE CONSTRUCTAL APPROACH

Cities and countries are complex systems composed of animate and inanimate sub-
systems. To the Romans, the city was the center of culture and development [52]. It was so
important that they built cities in the lands they conquered in order to spread their civilization.
Since Romans, the city configuration and size was a preoccupation. A Roman city started
with a grid layout, with major buildings in the center and the whole area surrounded by walls.
During the Roman Empire, the designers tried to limit cities’ sizes because they thought that
the quality of life of huge cites decreases [52]. In our days, the dynamic of city growth stills a
major issue [53].
Cities are characterized by specific shape and dynamics. They increase in size and
number (evolve). Cities and countries can be treated in thermodynamic terms, for these are
also open structures regulating the flow of energy in and out. They have also remarkable
general resemblances with living organisms. They acquire and consume resources, produce
and discard wastes, all the while employing energy for a variety of tasks: transportation,
communication, maintenance, and reproduction of the complexity and organization. Modern
cities and countries are as much a product of evolutionary events as any organism. They are
also voracious users of energy. Why do these “bodies” have the configuration that they have?
5.1. Zipf’s Law and the Constructal Law
At the end of the 19th century, Vilfredo Pareto suggested that income distribution in a
society is described by a power law (scaling law) [54]. In the meantime, other scaling laws
were founded in biology, economy and other areas of knowledge [18,31-33,55].
In 1949, George Zipf [56] found that city sizes obey to an astonishingly simple
distribution law. He attributed this feature to the “least effort principle” of human behavior
similarly to the idea of Pierre de Maupertuis (Section 1). Zipf’s law has been gaining
continuous interests for its accurate description of city size distributions in many different
countries and at different times within a country.
Zipf distribution of city ranks versus city sizes can be derived from the constructal law in
the same fashion as patterns of animate and inanimate flow structures. Bejan [12] shows that
this distribution can be obtained from the optimal allocation of flow paths to areas. This and
others Zipf distributions have their origin on tree-shaped flow systems with patterns optimally
allocated in space, in a similar way to scaling laws of river basins. These distributions are
counterparts of the scaling laws of biology. In both cases, they reveal easier flowing
configurations, and in this sense, scaling laws are a synonym of optimal structures.
In summary, scaling laws are useful tools in studying complex systems because they
reflect underlying generic features that are characteristics of the systems.
5.2. Bringing New Insights to Science: Cities/Countries and Scaling Laws
The conception and performance of living organisms have long been an inspiration and
subject of study in different domains [27]. From Leonardo da Vinci investigations into human
flight, based on the flight of birds, to modern bionics humans created systems that
resemblance working of living bodies. The constructal theory shows that the improvement of
function (easier access to currents) creates configurations that characterize both animate and
inanimate systems (Sections 3 and 4). Therefore, “imitate” nature's design maybe relevant
because it may contribute to solve problems in an efficient and sustainable way.
Organisms have evolved over thousands of years towards optimal configurations and the
“existence of a scaling relationship is a hint that there is an underlying constructal
evolutionary process” [39]. Could cities and countries have reached a similar level of
performance?
All scientific efforts are based on the existence of universality, which manifests itself in
diverse ways. Four decades ago, for example, studies on phase transitions and critical
phenomena conducted to the development of two important concepts: scaling and
universality. These developments were based both on the scaling argument of B. Widom [57]
and the full theory of the renormalization group of K. G. Wilson [58]. For systems in the
same class, exponents and scaling functions are the same (universal) in the vicinity of the
critical point. Scaling and universality support much our current conceptualization in different
subjects. Here we explore the thought that complex systems far from equilibrium with similar
functions have a propensity to exhibit similar scaling correlations. The purpose is to examine
and interpret the metabolism of cities and countries in the light of scaling laws for biology.
These laws relate a characteristics or functions of organisms to their body mass. Therefore,
we should obtain the body mass of cities and countries.
Cities and countries are composed by groups of subsystems that perform specific
functions towards commons and specific goals. Similar to organs of living organisms, each
subsystem (entity) has a “metabolic” relationship with the rest of the body (city, country),
while each subsystem has its own role in the overall metabolism. The whole-system (city,
country) “metabolism” results from the summation of the average metabolic rate of all
subsystems having a total effective mass. The mass of animate and inanimate subsystems
must be accounted. The animate subsystem is assumed to be formed by inhabitants and
livestock. Conversely, dwellings, power plants, road and rail networks, ports and airports, all
compose the inanimate subsystem. The effective mass of a city/country results of the
summation of the whole mass of these subsystems.
Figure 6. Metabolic power versus the mass (cities [59] and countries [60])
The 3/4-power scaling relationship [31-33] and the metabolic power (power
consumption) are depicted in Figure 6. Data for cities and countries were obtained from
Isalgue et al. [59] and Miguel [60], respectively. This plot shows that the metabolic power of
cities of USA and France are above the allometric law of animate systems. The same
tendency is followed by the countries (USA and France). On the other hand, other countries
are below the slope line of the allometric law. Unfortunately, there are no data available for
cities in these countries, which make it impossible to verify this tendency. Figure 6 also
shows that the measured values of the power consumption for countries are not very different
from those obtained with the scaling law.
Figure 7. CO2 emission versus the country’s mass in 2002 [60]
Figure 8. Gross domestic product (GDP) versus the country’s mass in 2002 [60]
For 12 countries (Australia, Brazil, Denmark, France, Germany, Italy, Japan,

Netherlands, South Africa, Turkey, UK and USA) carbon dioxide emissions based on data of
2002 and the 3/4 power scaling line are shown in Figure 7. This plot shows the existence of
countries that fall on the 3/4 power line for living organisms (Netherlands, France and Italy)
but that there are also countries above (USA, Japan, Germany and UK) and below (Brazil,
Turkey, Denmark, Australia and South Africa) this line. Except for these three countries
(Netherlands, France and Italy), all countries release more and less CO2 than expected for an
optimal behavior. The “metabolism” of countries like the USA is very fast and releases about
3.5 times more than the CO2 of an optimal system with similar mass. On the other hand,
countries like Brazil have a “slow metabolism”, and it releases about 2.3 times less CO2 of an
optimal system. To check the possible connection between the speed of “metabolism” and the
economic growth, the gross domestic product (GDP) data is considered for each country
(Figure 8). The 3/4 power scaling line was adjusted in order to match the Netherlands (i.e.,
the country that follows a 3/4 power scaling for CO2 emission). According to Figure 8, we
come to the conclusion that countries have a similar behavior to that displayed in Figure 7.
Except for France (perhaps due to the fact that the majority of the electrical power utilized in
France comes from nuclear power plants and not from fossil fuel), the countries with a “slow
metabolism” present GDP values below the 3/4 power scaling line. Therefore, it seems that
there is a strong relationship between CO2 emissions and GDP which agrees with the findings
of [61].
In summary, for CO2 emission there are countries that obey to different scaling laws. The
scaling exponent for the countries with a “fast metabolism” (USA, Japan, Germany and UK)
is 1.04 (r2=0.996) and for the countries with a “slow metabolism” (Brazil, Turkey, Denmark,
Australia and South Africa) is 0.90 (r2=0.888). Notice that all countries are composed by
animate and inanimate sub-systems with similar (constructal) configurations. Therefore, the
existence of different scaling relationships indicates these groups of countries might have
experienced different evolutionary processes. Notice that all the countries with a “fast
metabolism” are member of the Group of Eight (G8) highly industrialized countries. None of
the countries with a “slow metabolism” is member of this G8 group.
6. CONSTRUCTAL THEORY – A UNIFYING PERSPECTIVE

Nature displays an enormous variety of shapes (patterns, configurations). Why is shape
so important? Is there a principle from which shapes can be deduced?
As noted by Feynman [62] “[There is a] rhythm and pattern between the phenomena of
nature which is not apparent to the eye, but only to the eye of analysis; and it is these rhythms
and patterns which we call Physical Law.”. Constructal law is the view that the generation of
flow configuration is a physics phenomenon. The outcome of constructal law is that system
shape and internal flow architecture do not develop by chance, but result from the permanent
struggle for better performance. The shape of the system evolves just as the envelope of
underlying dynamic processes. The constructal law is the statement that for a flow system to
persist in time it must evolve in such a way that it provides easier flow access. This is a very
attractive configuration principle for both animate and inanimate systems. Therefore, the
constructal theory provides a common theoretical basis for the existence of design in nature.
The constructal theory described in the previous section is robust and can be also applied
to engineered systems [13,14]. This theory is now a fast growing field with contributions
from many sources, and with leads in many directions (see for example,
[12,13,14,16,18,27,63,64]).
REFERENCES
[1] Bejan, A. (1999). How nature takes shape: extensions of constructal theory to ducts,
rivers, turbulence, cracks, dendritic crystals and spatial economics. Int. J. Therm. Sci.,
38, 653-663
[2] Aristotle, Physics. Adelaide: eBooks, Adelaide; 2007
[3] Emerson, R. W. (2006). The Conduct of Life. Lanham: University Press of America.
[4] Kant, I. (2008). Critique of Judgement. Radford: Wilder Publications.
[5] Glass, B., Temkin, O. & Straus Jr., W. (editors) (1959). Forerunners of Darwin:
1745-1859. Maryland: Johns Hopkins University Press.
[6] Reis, A. H. (2006) Constructal theory: from engineering to physics, and how flow
systems develop shape and structure. Applied Mechanics Reviews, 59, 269-282.
[7] Colding, T. H. & Minicozzi, W. P. (2006) Shapes of embedded minimal surfaces.
Proceedings of the National Academy of Sciences, 103, 11106-11111.
[8] Darwin, C. (1995). The Origin of Species. New York: Gramercy.
[9] Thompson, D. W. (1992). On Growth and Form. Cambridge: Cambridge University
Press.
[10] Helbing, D. & Molnár, P. (1995) Social force model for pedestrian dynamics.
Phys. Rev., E, 51, 4282 – 4286.
[11] Mandelbrot, B. B. (1982). The Fractal Geometry of Nature. New York: W.H. Freeman
and Co.
[12] Bejan, A. (2006). Advanced Engineering Thermodynamics, 2nd ed.. Hoboken: Wiley;
1997; Advanced Engineering Thermodynamics, 3rd ed.. Hoboken: Wiley.
[13] Bejan, A. (2000). Shape and Structure, from engineering to nature. Cambridge:
Cambridge University Press.
[14] Bejan, A. & Lorente, S. (2008). Design with Constructal Theory. New Jersey: Wiley.
[15] Bejan, A. & Lorente, S. (2006) Constructal theory of generation of configuration in
nature and engineering. J. Applied Physics, 100, 2006.
[16] Bejan, A. & Lorente, S. (2004) The constructal law and the thermodynamics of flow
systems with configuration. Int. J. Heat and Mass Transfer, 47, 3203–3214.
[17] Bejan, A. & Lorente, S., (2004). Equilibrium and nonequilibrium flow system
architectures. Heat Technol., 22, 85–92.
[18] Bejan, A., Lorente, S., Miguel, A. F. & Reis, A. H. (2006). Along with Constructal
Theory. Workshop series nº 1 (Hernandez & Cosinschi, editors). Lausanne: University
of Lausanne Press.
[19] Reis, A. H., Miguel, A. F. & Bejan, A. (2006). Constructal theory of particle
agglomeration and design of air-cleaning devices. J. Phys. D, 39, 2311–2318.
[20] Bejan, A. & Gobin, D. (2006) Constructal theory of droplet impact geometry. Int.
J. Heat and Mass Transfer, 49, 2412–2419.
[21] Reis, A. H. (2006). Constructal view of scaling laws of river basins. Geomorphology,
78, 201–206.
[22] Reis, A. H. & Gama, C. (2009). Sand size versus beachface slope — an explanation
based on the constructal law. Geomorphology, in press
[23] Bejan, A. & Reis, A. H. (2005). Thermodynamic optimization of global circulation and
climate. Int. J. Energy Res., 29, 303-316.
[24] Reis, A. H. & Bejan, A. (2006). Constructal theory of global circulation and climate.
Int. J. Heat Mass Transfer, 49, 1857–1875.
[25] Bejan, A., Ikegami, Y. & Ledezma, G. A. (1998). Constructal theory of natural crack
pattern formation for fastest cooling. Int. J. Heat Mass Transfer, 41, 1945–1954.
[26] Lorente, S. (2007). Constructal view of electrokinetic transfer through porous media.
J. Phys. D, 40, 2941–2947
[27] Miguel, A. F. (2009). Pattern Formation and Self-Organization in Living Systems: a
unified view for coral colonies and crowd dynamics. In: "Constructal Human
Dynamics, Security and Sustainability" A. Bejan, S. Lorente, A. F. Miguel, A. H. Reis
(editors). Series Human and Societal Dynamics – Vol. 50. Amsterdam: IOS Press.
[28] Reis, A. H, Miguel, A. F. & Aydin, M. (2004). Constructal theory of flow architecture
of the lungs. Medical Physics, 31, 1135-1140.
[29] Reis, A. H. & Miguel, A. F. (2006). Constructal theory and flow architectures in living
systems. J. Thermal Science, 10, 57-64.
[30] Bejan, A., Dincer, I., Lorente, S., Miguel, A. F. & Reis, A. H. (2004). Porous and
Complex Flow Structures in Modern Technologies. New York: Springer.
[31] Kleiber, M. (1947). Body size and metabolic rate. Physiological Reviews, 27, 511-541.
[32] Brody, S. (1945). Bioenergetics and Growth. New York: Reinhold.
[33] West, G. B., Woodruff, W. H. & Brown, J. H. (2002). Allometric scaling of metabolism
from molecules and mitochondria to cells and mammals. Proc. Natl. Acad. Sci. U.S.A.,
99, 2473.
[34] Bejan, A. (2001). The tree of convective heat streams: its thermal insulation function
and the predicted 3/4-power relation between body heat loss and body size. Int. J. Heat
and Mass Transfer, 44, 699-704.
[35] Bejan, A. (1997). Theory of organization in Nature: pulsating physiological processes.
Int. J. Heat Mass Transfer, 40, 2097-2104.
[36] Bejan, A. (1990). Optimum hair strand diameter for minimum free-convection heat
transfer from a surface covered with hair. Int. J. Heat Mass Transfer, 33, 206-209.
[37] Bejan, A. & Marden, J. H. (2006). Unifying constructal theory for scale effects in
running, swimming and flying. J. Exp. Biol., 209, 238-248.
[38] Bejan, A. & Marden, J. H. (2009). The constructal unification of biological and
geophysical design. Physics of Life Reviews, 6, 85-102.
[39] Charles, J. D. & Bejan, A. (2009). The evolution of speed, size and shape in modern
athletics. J. Exp. Biol., 212, 2419-2425.
[40] Bejan, A., Lorente, S. & Lee, J. (2008). Unifying constructal theory of tree roots,
canopies and forests. J. Theoretical Biology, 254, 529–540.
[41] Ben-Jacob, E., Cohen, I., Shochet, O., Aronson, I., Levine, H. & Tsimering, L. (1995).
Complex bacterial patterns. Nature, 373, 566-567.
[42] Matsushita, M., Wakita, J., Itoh, H., Ràfols, I., Matsuyama, T., Sakaguchi, H. &
Mimura, M. (1998). Interface growth and pattern formation in bacterial colonies.
Physica A, 249, 517-524.
[43] Kaandorp J. A. & Sloot P. M. A. (2001). Morphological models of radiate accretive
growth and the influence of hydrodynamics. J. Theor. Biol., 209, 257–274.
[44] Tumlinson, J. H.,. Silverstein, R. J., Moser, J. C., Brownlee, R. G. & Ruth, J. M.
(1971). Identification of the trail pheromone of a leaf-cutting ant, Atta texana. Nature,
234, 348 – 349.
[45] Partan, S. R. & Marler, P. (1999). Communication goes multimodal. Science, 283,
1272–1273.
[46] Anderson, C. & McShea, D. W. (2001). Individual versus social complexity, with
particular reference to ant colonies. Biological Review, 76, 211–237.
[47] Miguel, A. F. (2006). Constructal pattern formation in stony corals, bacterial colonies
and plant roots under different hydrodynamics conditions. J. Theoretical Biology, 242,
954-961.
[48] Miguel, A. F. & Bejan, A. (2009). The principle that generates dissimilar patterns inside
aggregates of organisms. Physica A, 388, 727-731.
[49] Miguel, A. F. (2009). Constructal theory of pedestrian dynamics. Physics Letters A,
373, 1734-1738
[50] Miguel, A. F. (2007). Constructal Patterns Formation in Nature, Pedestrian Motion and
Epidemics Propagation. In: Constructal Theory of Social Dynamics (Bejan and Merkx,
editors), New York: Springer, chapt. 5, 85-114.
[51] Helbing, D., Schweitzer, F., Keltsch, J. & Molnar, P. (1997). Active walker model for
the formation of human and animal trail systems. Phys. Rev. E, 56, 2527-2539.
[52] Macaulay, D. (1983). City: a Story of Roman Planning and Construction. Boston:
Houghton Mifflin Company.
[53] Bluestone, B., Stevenson, M. H. & Williams, R. (2008). The Urban Experience:
Economics, Society, and Public Policy. New York: Oxford University Press.
[54] Pareto, V. (1897). Le Cours d’Economie Politique. London: Macmillan.
[55] West, G. B., Brown, J. H. & Enquist, B. J. (1997). A general model for the origin of
allometric scaling laws in biology. Science, 276, 122–126.
[56] Zipf, G. K. (1949). Human Behavior and the Principle of Least Effort. Cambridge:
Addison-Wesley.
[57] Widom, B. (1965) Surface tension and molecular correlations near the critical point. J.
Chem. Phys., 43, 3892-3897
[58] Wilson, K. G. (1971) Renormalization Group and Critical Phenomena. I.
Renormalization Group and the Kadanoff Scaling Picture. Phys. Rev. B, 4, 3171-3183
[59] Isalgue, A., Coch, H. & Serra, R. (2007). Scaling laws and the modern city. Physica A,
382, 643–649.
[60] Miguel, A. F. (2009). Quantitative study of the CO2 emission to the atmosphere from
biological scaling laws. Int. J. Global Warming, 1, 129-143.
[61] Shafik, N. (1994) Economic development and environmental quality: an econometric
analysis, Oxford Economic Papers, 46, 757-773.
[62] Feynman, R. F. (1967). The Character of the Physical Law. Cambridge: MIT press.
[63] Bejan, A., Badescu, V. & de Vos, A. (2000) Constructal theory of economics. Applied
Energy, 67, 37-60.
[64] Bejan, A., Badescu, V. & de Vos, A. (2000) Constructal theory of economic structure
generation in space and time. Energy Conversion and Management, 41, 1429-1451.
Chapter 6
COMPUTATIONAL STUDIES ON DRAG REDUCTION

EFFECT BY SURFACE GROOVES
Haosheng Chen* and Yongjian Li

State Key Laboratory of Tribology, Department of precision instrument and mechnology
Tsinghua University, Beijing, 100084, China
ABSTRACT
The drag reduction effect caused by the periodic surface grooves were studied using
computational fluid dynamic method in two different flow conditions: laminar flow in a
slide-disk interface, and turbulent flow on the grooved surface immerged in water.
In the first part, the drag results by Reynolds equation that is commonly used in
lubrication calculation is compared with the CFD result based on Navier-Stokes equation.
It was validated that the Reynolds equation is not suitable when the groove depth is
higher than 10% of the interface distance. Then, the drag forces on the surface with
transverse and longitudinal grooves are calculated using CFD method. It was found that
the ‘side wall’ effect causes the drag reduction, which means the drag reduction would
appear when the loss of the drag on the groove’s bottom can not be compensated by the
pressure drag or the viscous drag on the side walls of the grooves.
In the second part, the drag force on the transverse rectangular grooves in turbulent
flow is analyzed based on the RANS equations coupled with the RNG k-ε turbulent
model. It was found that the pressure drag force makes up a large proportion of the total
drag, and the turbulent vortex structure in the grooves affects the drag characteristic of
the surface. The ‘side wall’ effect also functions in turbulent flows, and the drag
reduction is determined by the synthesized effect of the reduction of viscous drag and the
increment of pressure drag.
*
Corresponding author: chenhs@tsinghua.edu.cn
218 Haosheng Chen and Yongjian Li
SECTION I
Drag Reduction on Groove Surface in Laminar Flow
Contents
Usually, Reynolds equation is used in hydrodynamic lubrication calculation, and a

discrete probability distribution function is used in the equation to represent the effect of
transverse and longitude grooves. Here, the drag force on the square groove surface by the
Reynolds equation is validated by the Navier-Stokes equation, and it is found that the results
acquired by the two methods are the same when the groove height is below 1% of the film
thickness. When the height exceeds 1% of the film thickness, the forces acquired by Reynolds
equations are smaller than those acquired by CFD, and the difference between them exceeds
10% when the groove height is higher than 10% of the film thickness. Side wall effect is
considered to be the main reason for the difference, and Reynolds equation is believed not
suitable for calculating the drag force under such conditions.
Therefore, forces on side wall of the transverse and longitudinal grooves are numerically
analyzed based on Navier-Stokes equation using finite volume method. The viscous force on
the groove’s bottom plane is far less than the force on corresponding plane of smooth surface,
but the loss of the drag force on groove’s bottom surface is compensated by either the
pressure drag on side walls of transverse groove, or the viscous force on side walls of
longitudinal groove. According to the numerical results for the squared grooves as calculated
here, the pressure drag on side wall of transverse groove is higher than the viscous force on
side wall of longitudinal groove, and drag reduction happens on the surface with longitudinal
grooves.
1. CFD EVALUATION ON APPLICABILITY OF REYNOLDS EQUATION

The drag reduction on groove surface has been studied for many years [1]. The pioneer
work done by Dowson [2] and Shelly [3] has revealed the effect of artificially made grooves
in a journal bearing lubrication. It has been proved that proper surface roughness increases the
load capacity and reduces the friction coefficient [2-5], and specially designed surface
roughness is also investigated in improving the head-disk interface lubrication recently [6,7].
On the other hand, MEMS technology makes it possible to design micro structures, such as
micro squared grooves, on the working surface to develop the lubricating results [8,9].
However, there are some problems when using the traditional Reynolds equation to
calculate the hydrodynamic lubrication under the effect of squared grooves. First, the average
Reynolds equation [10, 11] is commonly used and the surface roughness effect is represented
by the stochastic model, where Gaussian distribution function is used for random roughness.
But the Gaussian distribution function is not suitable for the artificially designed periodical
grooves. Second, as Elrod [1] has pointed out, Reynolds equation can only apply to the
“Reynolds roughness” where the roughness height is an order of magnitude smaller than the
film thickness. But in most of the applications where micro grooves are made on the surfaces
[6,9], the groove depth is in an order of the magnitude of the lubricant film thickness, so the
Computational Studies on Drag Reduction Effect by Surface Grooves 219
work on the applicability and precision of the Reynolds equation under the effect of surface
grooves still needs to be continued.
In the following studies, the hydrodynamic lubrication on a surface with 50 squared
transverse grooves is calculated by Reynolds equation and CFD method, respectively. When
using Reynolds equation, Gaussian distribution function is replaced by a discrete probability
distribution function to represent the expectancy of the film thickness. This modified
Reynolds-type equation (MRE) is derived to play the same role in the analysis of groove
surfaces as the ordinary Reynolds equation does. Then, the differences between the results
from the less approximating CFD approach and modified Reynolds equations are compared.
Based on the comparison, the applicability of Reynolds equation in calculating the effect of
squared groove surface is discussed.
1.1. Geometry Models
Figure 1 shows the geometry model for CFD, it is a typical slider-disk interface used in
lubrication study. The upper surface is smooth and moves at a speed of U. The pitch angle of
the upper surface is 1°. The lower surface is rough and stationary, and there are 50 squared
transverse grooves with the wavelength λ on the lower surface. The geometry of the lubricant
film is split up into two parts. One part which denoted the nominal film thickness h ( x)
measures the large scale variation in film geometry. The other part is the part due to groove
depth as measured from the nominal level, and it varies with the amplitude of the roughness
height σ. The relationship between the two parts is σ =β (h0+h1) / 2.
As a common method, a CFD software FLUENT is used as a less approximating
approach to achieve more precise results. FLUENT uses a control-volume-based technique to
convert the governing equations, such as Navier-Stokes equations, to algebraic equations that
can be solved numerically. The continuity equation and momentum equation are adopted in
the calculation, which are applicable for continuous medium. The values of the parameters
used in the lubrication are listed in Table 1. The boundary conditions for CFD are defined as
follows: the upper surface is a moving wall, while the lower surface is a stationary wall; the
left side is the pressure inlet and the right side is the pressure outlet. Under the low Reynolds
number, the lubricant is considered to perform a stable laminar flow according to Chin’s
viewpoint [12], and a laminar model is adopted in CFD.
U z
x
h1
h(x)
h(x)
h0
σ
σ
B=50λ
Figure1. Geometry model of a slider-disk with 50 squared transverse grooves

Table 1. Values and parameters used in the lubrication calculation
Parameters Value
U 10 m/s
h1 2.4364e-004 m
h0 1.5636e-004 m
B 5e-3 m
η 1 Pa.s
β 0.001~0.2
There are 55000 quadrangle grids total for calculation. To show the pressure and fluid
flow near the wall more clearly, the grid number increases from the center to the wall and the
grow factor is 1.1. The grid size is about 2×2 μm2, which is small enough in the laminar flow
compared with the size of the squared transverse waves. The residual definitions are used for
judging convergence. This criterion requires that the residuals decrease to 10-6 for the
continuity equation and the velocity equation. During the calculation, the residuals reach the
convergence criteria after 25000 iteration steps, which indicate that a quite precise CFD result
is achieved.
1.2. Modified Reynolds Equations for Groove Surface
A stochastic model for rough surface with transverse roughness is shown as Eq.(1),
where ψi(H) is the expectancy of the film thickness that has been demonstrated detailedly by
Christensen [13].
∂ ⎡ ∂p ⎤ ∂
⎢ Ψ 1 (H )⎥ = 6μU Ψ 3 (H ) (1)
∂x ⎣ ∂x ⎦ ∂x
For random transverse roughness, the film thickness function ψi (H) was given by
Christensen [13]. For surface with square grooves, the Gaussian distribution function used for
random roughness is replaced by the discrete probability distribution function to calculate the
expectancy of the film thickness. In the Eq.(2), E is the expectancy, xi is the variance and pi is
the probability respect to xi.
n
E = ∑ xi pi (2)
i =1
so, the expressions of the functions are:
ψ 1 = ( 2 β3 ) h 3 ; ψ 3 = ( β 2 β3 ) h ; (3)
2 2 3 3
⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞
Where, β2 = ⎜ ⎟ +⎜ ⎟ ; β3 = ⎜ ⎟ +⎜ ⎟
⎝ 1+ β ⎠ ⎝ 1− β ⎠ ⎝ 1+ β ⎠ ⎝ 1− β ⎠
Then, the Reynolds equation is modified as following:
∂ ⎡ ∂p 3 ⎤ β 2 ∂h
⎢ h ⎥= 6 μU (4)
∂x ⎣ ∂x ⎦ 2 ∂x
According to the geometry model shown in Figure 1, the pressure is acquired through
integrating equation (4) on both sides with respect to x twice. With the boundary conditions
p|h=h0=0 and p|h=h1=0, the expression of the hydrodynamic pressure is derived as Eq. (5),
where K=( h1-h0)/ h0.
β2 B ⎡ 1 h0 h1 1 1 ⎤
p = 6μU ⎢− + + ⎥ (5)
2 Kh0 ⎣ h h0 + h1 h 2
h0 + h1 ⎦
1.3. Numerical Results
In the calculation, the groove depth is not higher than 20% of the film thickness, i.e. β ≤
0.2, which indicates that the lubrication is under the full film lubricating condition. The load
capacity, friction force and the friction coefficient are calculated by CFD and MRE methods,
respectively. Such forces are compared with those on smooth surface using normalized
percentage (F-Fs)/Fs%, where F is the force for the rough surface while Fs is the force for the
smooth surface.
(a) MRE (b) CFD

15 40
0.190
0.190
12 0.185
30
1 2 3 0.185 1 2 3 0.180
9
Friction coefficient
0.175
(F-Fs)Fs%
(F-Fs)Fs%
0.180 20
6
0.170
Load capacity Load capacity
0.175 0.165
3 Friction force 10 Friction force
Friction coefficient Friction coefficient 0.160
0 0.170
0 0.155
-3 0.165 0.150
1E-3 0.01 0.1 1 1E-3 0.01 0.1 1
β β
Figure 2. Lubrication results acquired by CFD and MRE compared with those on smooth surface
As the groove depth increases, both of the load capacity and the friction force increase
compared with forces on smooth surface, but the increment of the friction force is lower than
that of the load capacity, which results in the decrement of the friction coefficient. The
increasing process of the lubricant results can be divided into 3 characteristic stages. In stage
1, β<0.01, the effect of the surface groove on lubricating results is not obvious, the load
capacity and the friction force on the groove surface is almost the same as those on smooth
surface, it indicates that the terms considering the roughness effect can be omitted in the
Reynolds equations. In stage 2, β<0.1, the load capacity and the friction force on the rough
surface begin to increase, and the friction coefficient decreases correspondingly. However,
the increments of the forces are below 5% and 15% acquired by MRE and CFD, respectively.
So the groove effect should be considered in some conditions where the precision is required.
In stage 3, β ≥ 0.1, the film thickness is greatly changed by the surface roughness, and the
load capacity increases so rapidly that the groove effect must be included in the lubrication
calculation.
According to the results shown in Figure 2, it is found that there is a marked difference
between the forces acquired by MRE and CFD methods, especially when β ≥ 0.01. Figure 3
shows the differences more clearly using the normalized percentage, where the subscript
represents the method used in the calculation. The differences on the load capacity and the
friction force both increase as the groove depth increases. Moreover, the increment speed of
load capacity is higher than that of the friction force, which results in the lower friction
coefficient acquired by the CFD method. As Almqvist [14] has validated, CFD is a more
precise method than the Reynolds equation, so it is believed that a conservative lubrication
effect is acquired by the Reynolds equation. For example, the friction coefficient and the load
capacity are 10% lower and 20% higher than those acquired by the MRE method,
respectively. Based on the comparison, it is considered that the Reynolds equation is good
enough for calculating the hydrodynamic lubrication including the roughness effect when the
roughness height is below 1% of the film thickness, while it is not suitable when the squared
groove depth is in an order of the film thickness.
20
Friction force
Load capacity
(FCFD-FMRE)/FMRE%
10
-10
1E-3 0.01 0.1 1
β
Figure 3. Comparison on lubrication results acquired by CFD and MRE
One of the important reasons that cause the different lubrication results acquired by MRE
and CFD is considered to be the side wall effect caused by the squared transverse grooves.
Figure 4 shows the shape of the 25th groove when β = 0.1. There are two side walls in each
groove. Plane 4 is the windward of the flow direction while plane 2 is the leeward of the flow
direction. Pressures on the windward plane will increase while pressures on the leeward plane
will decrease, when the fluid passes the transverse groove. The contour lines of the pressures
on the two planes are marked in Figure 4(a), and the value of each line is also given in the
figure. It is found that the highest pressure is about 20% higher than the lowest pressure. This
result conflicts with one of the basic assumption for Reynolds equation that the pressure does
not change so much across the fluid film. As Sun [15] has point out, the pressures generated
in the fluid along x and z directions can be represented by Eq. (6) and (7) , respectively.
(a) β =0.1 3.91Mpa 6.07Mpa A 600 (b)
Pressure along A-A cross section(kPa)

4.22Mpa 5.76Mpa
z β=0.1
4.53Mpa 5.45Mpa
550
4.83Mpa 5.14Mpa
U
500
groove
β=0.01
Plane2
Plane4
450 groove
β=0.001
Groove groove
x
400
Plane3 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3
A Height (m)
Figure 4. Pressure distributions on the side wall of the 25th squared transverse wave
∂u ∂u
px − =ε2 (6)
∂z ∂x
∂v ∂v
pz = ε 2 +ε4 (7)
∂z ∂x
When the groove depth is small, the terms on the right-hand side of equations are
neglected, and then the traditional Reynolds-type equation is established. However, if the
groove depth is in an order of the film thickness, such terms should be retained to reveal the
roughness effects. In the average Reynolds equation [10,11], the effect of Eq.(6) is
compensated by the flow factors, but the effect of Eq.(7) is neglected. In the calculation
results acquired by CFD, such terms are kept, and Figure 4(b) describes the pressure variation
across the film thickness at the cross-section A-A when roughness height is 0.001, 0.01 and
0.1, respectively. It is proved by the results that the pressure along z direction almost keeps
constant when the groove depth is small, while it changes a lot when the groove depth
exceeds 1% of the fluid film thickness. Compared with the results from MRE, it is believed
that the pressure variation across the film increases the hydrodynamic pressure, which
directly causes the increment of the load capacity.
Compared with the load capacity, the reason for the increment of the friction force is
more complex. The total friction force consists of two parts. One is the viscous drag force on
the surface parallel to the flow direction caused by the fluid viscosity, the other is the pressure
drag caused by the side walls of the grooves shown as Figure 4(a). On the windward side
wall, the pressure increases from the bottom to the top of the side wall. But on the leeward
plane, the pressure variation process is reversed. The difference of pressures on the two side
walls of the transverse groove causes a significant pressure drag.
As the groove depth increases, the velocity of the fluid in the groove decreases and the
viscous friction force on the bottom wall is then decreases. On the other hand, the pressure on
the windward side wall increases as the groove depth increases, then the pressure drag is
inevitably increased. So whether the friction force on the rough surface is higher or lower
than that on the smooth surface is determined by whether the pressure drag generated on side
walls can compensate the viscous drag loss on the bottom walls or not. The result is
illustrated by Figure 5, and the speed of the pressure drags increment on the side walls is
higher than that of the viscous drag loss on the bottom walls, which results in the increment
of the total friction force on the rough surface. Furthermore, the increment of roughness
height not only increases the pressure drag, but also decreases the viscous drag, so the total
increment speed of the friction force is less than that of the load capacity, which is the reason
for the decrement of the friction coefficient. The great difference between the results acquired
by CFD and MRE indicates that the Reynolds-type equations are not suitable for the friction
force calculation under the side wall effect.
300
250
200
Drag force (N)
150 pressure drag

viscous drag
100 total drag
drag on smooth surface
50
0
1E-3 0.01 0.1 0.5
β
Figure 5. Friction forces on transverse groove surface using CFD
Thus, some conclusions are drawn according to the comparison:
(1) Compared with the smooth surface, surface grooves begin to show the effect when
groove depth is larger than 1% of the film thickness, and the effect can not be
neglected when the height exceed 10% of the film thickness. Squared transverse
grooves on the surface cause higher load capacity, higher friction force, and lower
friction coefficient, and the groove effect becomes stronger as the groove depth
increases.
(2) Compared with the CFD results, lubrication results acquired by the modified
Reynolds equation is more conservative when the groove depth exceeds 1% of the
film thickness, and the Reynolds equation is considered not applicable when the
groove depth is higher than 10% of the film thickness because the load capacity
acquired by the Reynolds equation is 20% lower than that acquired by more precise
CFD method.
(3) Side wall effect is believed to be one of the important reasons causing the difference
between CFD and MRE results. The side wall effect causes the pressure variation
across the fluid film and increases the hydrodynamic pressure, which directly
increases the load capacity compared with that acquired by MRE. On the other hand,
the side wall effect both increases the pressure drags on the side walls and decreases
the viscous drags on the bottom walls. Whether the pressure drags can compensate
the loss of viscous drags determines the appearance of the drag reduction.
2. DRAG REDUCTION OF SURFACE GROOVES

Recently, the effect of surface groove patterns on drag reduction has been studied
considering the parameter γ , which represents the different surface groove directions. Their
calculations show that the γ affects the lubrication results through changing the flow factors
[6], it is predicted in the numerical results [5,6] that the surface with longitudinal pattern has
the highest load capacity compared with the transverse and isotropic surface patterns, and it
correspondingly decreases the friction coefficient.
According to the previous studies, when the surface roughness height is in the same order
of the lubrication film thickness, there is great difference between the results obtained from
the Reynolds equation and the results obtain from the less approximating CFD-approach
[14,16]. This continuous studies is to determine which kind of surface patterns is best for the
drag reduction under the condition that Reynolds equation is not applicable. Forces on
different planes of the groove surface are numerically calculated based on Navier-Stokes
equation using finite volume method, and analysis on the drag force reduction effect of
different surface roughness is also given. Same to the previous studies, all the calculations are
under the condition of laminar flow.
2.1. Models and Method
The geometry models of the surface with grooves are shown in Figure 6 schematically.
The dimension unit is millimeter. The upper surface is smooth and it moves at a velocity U.
The lower surface is stationary and grooves are on it. There are two typical kinds of patterns
on stationary surface, longitudinal grooves and transverse grooves. The cross-section of the
groove is rectangular and the surface pattern parameter γ is adopted here to represent the
length-to-width ratio of groove [6]. γ < 1 for transverse groove, and γ >1 for longitudinal
groove. The orientation of the longitudinal groove is parallel to the flow direction, while the
orientation of transverse groove is perpendicular to the flow direction. The isotropic pattern
can be seen as the superimposition result of the longitudinal pattern and the transverse
pattern.
The groove surface is divided into 5 rectangular planes as marked by c~g in Figure 6.
Plane 1,3,5 are parallel to x-z plane, they are called parallel planes. Plane 2 and plane 4 are
perpendicular to the x-z plane, they are called perpendicular planes. There are 5 walls in the
cross-section of the groove surface as shown in Figure 6(d). Wall 1 and 5 are the top walls,
while wall 2 and 4 are the side walls; wall 3 is the bottom wall. The smooth surface is also
modeled to be compared with the surface pattern, it is equalspaced divided into 3 walls named
wall 1, 3 and 5 respectively, which is shown in Figure 6(c). The groove is a steep-walled
surface and the depth of the groove is in the same order of the fluid thickness. Therefore, the
Reynolds equation is not applicable under such conditions. The effect of perpendicular planes
on the fluid flow and pressure needs to be included in the calculation. The numerical method
is same to that in previous studies in section 1, and the finite volume method (FVM) is
adopted to calculate the force on the different planes.
2.2. Numerical Analysis
Pressure drag on perpendicular planes
Between the two transverse perpendicular planes, plane 4 is the windward of the flow
direction while plane 2 is the leeward of the flow direction. The contour lines of the pressure
on the two planes are numbered from 1~5 in Figure 5(a). The value of each line is also given
in the figure, P4 represents the pressure on plane 4 and P2 represents the pressure on plane 2.
The difference of pressures on the two transverse perpendicular planes causes a significant
pressure drag. It is one of the important reasons that increase the drag force. A rectangular x-y
coordinate is established on plane 2 and plane 3 shown in Figure 7(a). More detailed pressure
along plane 4 from the bottom to the top of the interface is shown in Figure 7(b). The
pressures in Figure 7(b) are normalized, they are the ratio of the pressures to the maximum
pressure. The maximum pressure is marked in Figure 7(b). The center line represents the
groove depth. The existence of perpendicular plane causes a severe variation of the pressure
in the groove, and the pressure achieves a stable value rapidly when the height exceeds the
groove depth. The pressure variation on perpendicular planes affects the pressure drag, so the
pressures on side wall of grooves with different directions needs to be considered when
analyzing the effect of surface patterns to drag force reduction.
y
U = 10
2
m/sec
1 z 1
5 0.5
21
5
1 1 4 4
3 1 3 5
1
2 3 3
x
3
(a) Transverse pattern (b) Longitudinal pattern (c) Smooth pattern
y Wall 1 Wall 5
Wall 4
Wall 2
x Wall 3
(d) Walls and Mesh on cross-section A-A of groove model
Figure 6. Geometry models of surface patterns

1.0 (b) 1: P2=-1.90E3; P4=1.39E3 2: P2=-1.57E3; P4=1.07E3

3: P2=-1.24E3; P4=0.74E3 4: P2=-0.91E3; P4=0.41E3
5: P2=-0.58E3; P4=0.08E3
0.8 y (a)
U
5
5
Plane 1 4 43 Plane 5
0.6 1 3 2
1
P / P max
2
Groove
Plane 4
Plane 2
0.4
Plane 3
x
0.2
Groove deepth
PEO, Pmax=1.47E3
0.0
0.5 mm
-0.2
0 0.5 1.0 1.5 2.0
y / [mm]
Figure 7. Pressure on the perpendicular planes of transverse groove
Viscous force on perpendicular planes
Not same to the effect of transverse perpendicular planes, there is no pressure drag on the
longitudinal perpendicular plane. However, obvious viscous force is generated on the planes.
In the same fluid, the viscous force can be represented by the strain rate of the fluid. The
strain rates of the fluid on perpendicular planes of three different grooves are compared in
Figure 8. All the shear rates are on plane 2 and along direction x. For smooth surface, there is
no perpendicular plane and the shear rate is equal to zero. But for the later analyses on
viscous drag and pressure drag on different surfaces, it is provided here and compared with
the shear rate on other two surfaces. It is found from Figure 8 that there is not only pressure
drag but also viscous drag on the perpendicular planes of transverse grooves. But, the strain
rate on longitudinal perpendicular plane is obviously higher than the strain rate on other
corresponding perpendicular plane in the groove. It is one of the important part of the drag
force on the surface with longitudinal grooves.
2000
Groove
depth
1500
Shear rate / [1/s]
1000
Longitudinal
500 Smooth
Transverse
0 0.5 1.0 1.5 2.0

y / [mm]
Figure 8. Strain rate on the perpendicular planes

0.005 Viscous force on transverse 0.016

Longitudinal
Viscous force on longitudinal
Pressure force on transverse Transverse
0.004 Total force
0.012 Force on plane 1,3,5
Drag force / [Pa] Force on plane 2 ,4
0.003
Drag / [Pa]
0.008
0.002
0.004
0.001
0.000 0.000
plane1 plane 2 plane 3 plane 4 plane 5 200 400 600 800 1000 1200
Surface Shear rate / [1/s]
(a) Total force comparison when velocity is 2 m/s (b) Force comparison under different shear rates
Figure 9. Forces on planes of transverse and longitudinal grooves
Surface pattern effect
Total drag force on the surface consists of pressure drag and viscous force. As discussed
above, the forces on the perpendicular planes are obtained using finite volume method based
on Navier-Stokes equation. The pressure drag and viscous force on five planes are shown in
Figure 9(a).
The sum of viscous forces on plane 1 and plane 5 of different grooves are almost the
same. But the viscous force on perpendicular planes of longitudinal groove is obviously
greater than the force on perpendicular planes of transverse groove. On the other side, there is
no pressure drag generated on perpendicular planes of longitudinal groove, while there is
significant pressure drag on the perpendicular planes of transverse groove. Moreover, the sum
of pressure drag on transverse perpendicular planes is greater than the sum of viscous force
on longitudinal perpendicular planes. It indicates that the transverse perpendicular planes will
cause more drag force than the longitudinal perpendicular planes does.
As shown in Figure 9(b), the forces on perpendicular planes and on parallel planes are
both increasing as the shear rate increases. But the increment of the force difference on
perpendicular planes is always greater than the increment of the force difference on parallel
planes. It leads to the result that the total drag force on transverse grooves is more and more
greater than that on longitudinal grooves as the shear rate increases.
For smooth surface, the drag force is caused by the viscous force. Like the treatment on
rough surface, the smooth surface is also divided into three planes, the shape of each plane is
same to plane 1, plane 3 and plane 5 of transverse grooves, respectively. The force on each
plane is shown in Figure 10(a), it is compared with the forces on corresponding planes of
transverse groove and longitudinal groove. The viscous forces on plane 1 and plane 5 of
grooves are greater than the viscous force on corresponding smooth surface. But on plane 3,
the viscous force on smooth surface is obviously greater. It is because the flow velocity is
rapidly slowed down in the grooves, and the shear rate is also slowed down. The strain rate of
fluid at the center point of plane 3 from bottom to the top of the film thickness is shown as
Figure 10 (b). The strain rate on smooth surface is far greater than the strain rates on rough
surface. It leads to the higher viscous force on plane 3 of smooth surface compared with the
forces on rough surface. So, the sum of viscous forces on three parallel planes of grooves is
less than the force on smooth surface.
0.014 1600
Viscous force on transverse groove
Viscous force on longitudinal groove
Viscous force on smooth surface
Pressure force on transverse groove 1200
Shear rate / [1/s]

Drag force / [Pa]
800
0.007
Longitudinal
Roughness
400
height
Smooth
Transverse
0
0.000
plane1 plane2 plane3 plane4 plane5 all 0 0.5 1.0 1.5 2.0
Planes y / [mm]
(a) Comparison of drag force on different surfaces (b) Strain rate along y direction
Figure 10. Effect of surface pattern on drag reduction compared with smooth surface
However, the decrement of viscous force on parallel planes of grooves is compensated by

the pressure drags and viscous forces on perpendicular planes. Including the forces on the
perpendicular planes, the sum of forces on transverse groove is greater than the force on
smooth surface, and the sum of forces on longitudinal groove is a little bit less than force on
smooth surface. Thus, drag reduction effect appears only on the longitudinal groove surface.
3. CONCLUSIONS
In this section, the side wall effect on groove surface is numerically studied. Because of
it, the Reynolds equation is not suitable and the force on transverse groove surface is different
from that on longitudinal groove surface. According to the numerical results, some
conclusions can be drawn as following:
(1) The effect of groove surface needs to be considered when the groove depth is higher
than 1% of the lubricant film thickness.
(2) Reynolds equation is considered not applicable when the groove depth is higher than
10% of the film thickness, and the side wall effect causing the high pressure gradient
on the wall is believed to be one of the important reasons causing the difference
between CFD and MRE results.
(3) Pressure drag and viscous force on the perpendicular planes play important roles in
drag reduction effect. The viscous force on the bottom plane of the groove is far less
than the force on the corresponding plane of smooth surface. However, drag force on
groove surface is compensated by the pressure drag on transverse perpendicular
plane, or by the viscous force on longitudinal perpendicular plane. Since the pressure
drag appears to be always higher than the viscous force on the same perpendicular
plane under different average shear rates, the longitudinal groove surface has the
lowest drag force in laminar flow.
SECTION II
Drag Characteristic on Grooved Surface in Turbulent Flow
Content
In this section, the drag force on the transverse rectangular grooves in turbulent flow is
analyzed by CFD methods based on the RANS equations coupled with the RNG k-ε turbulent
model. The pressure drag force induced by the side wall effect could make up 80% of the
total drag force when the Reynolds number is higher than 50, 000. So, the side wall effect of
the transverse rectangular grooves is more significant in the turbulent flow than that in the
laminar flow.
The flow structure of the turbulence is more complicated and it is affected by the
Reynolds number and the shape of the grooves. As the increase of the width-to-depth ratio of
the grooves, the vortex structure in the grooves changes from single to double. This variation
could change the pressure difference between the windward side and the leeward side of the
grooves and the magnitude and direction of the viscous friction force acting on the bottom of
the grooves. The pressure distribution and flow downstream of the grooves also vary as the
increase of the width-to-depth ratio, the flow separation reduce the area acting by viscous
drag, and also arise the pressure difference. This induces the reduction of the viscous drag and
the increment of the pressure drag. As the former one is lower than the latter one, the total
drag force of surface with transverse rectangular grooves is always larger than that of the
smooth surface.
1. SIMULATION OF DRAG CHARACTERISTIC OF TRANSVERSE

GROOVES IN TURBULENT FLOW
Drag reduction by grooves (also called “riblets”) in turbulent flow is initialed by Walsh
and his colleagues at NASA-Langley laboratory[17]. They had proved that the geometrical
modified grooves aligned with the flow stream can reduce the drag force. Nowadays,
abundant researches are still continuously carried out to deepen the understanding of the
mechanism and optimize the cross-section shape and scale of these longitudinal grooves[18-
20]. However, the researches on the other kind of grooves – the transverse grooves, are much
less than the former one. In some of the pioneer study[21] on the drag characteristic of the
transverse pattern, only the height parameter is adopted to represent the effect of the patterns,
but the simulation result are not adequate. In most of the industrial problems, the pattern’s
height or depth is in the same order of the thickness of the boundary layer, so the dimension
of the patterns in the direction aligned with the flow must be taken into consideration. So far,
the drag characteristic of the surface with transverse grooves is still uncertain[22].
In the following study, the CFD methods based on the RANS equations coupled with the
RNG k-ε model to analysis the drag characteristic of the surface with transverse grooves are
demonstrated and the drag force composition on the grooved surface are solved. The effect of
the pressure drag force is discussed.
1.1.Geometry Models
Figure 1 shows the two-dimensional geometry model of a flat surface with periodic
transverse rectangular grooves. Three geometric parameters are defined to demonstrate this
kind of grooves, and they are the groove width (w), the groove depth (k), and the groove
wavelength (λ).
Figure 1. Geometry model of surface with periodic transverse rectangular grooves
1.2. CFD Simulation Method for Grooved Surface in Turbulent Flow
The Reynolds average Navier-Stocks (RANS) equation [23-25] is one of the most
popular physical models for the turbulent flow. It split the motion of the flow into two parts:
the average motion and the random fluctuation. The former one is calculated based on the
traditional Navier-Stocks equation and the latter one is noted as the Reynolds stress term. The
RANS equation is written as follows:
∂ ( ρ u i ) ∂ ( ρu i u j ) ∂p ∂ ⎛ ∂u i
⎜μ
⎞
⎟ − ρ u i′u ′j
+ =− + (1)
∂t ∂x j ∂xi ∂x j ⎜ ∂x ⎟
⎝ j ⎠
− ρ ui'u 'j
where, is known as the Reynolds stress term or turbulent stress term.
To solve the RANS equation, turbulent models for Reynolds stress term should be
chosen. k-ε model is one of the turbulent models which is widely used nowadays. Here k is
k = u ′u ′ 2
the turbulent kinetic energy, which can be defined as i i , and ε is the turbulent
dissipation rate of the fluctuation kinetic energy per unit mass, which represents the transfer
rate of the energy of the isotropy small scale vortices from mechanical to thermal:
ε = μ (∂u i′ / ∂x j ) ⋅ (∂u i′ / ∂x j ) / ρ . k and ε could be solved by the follow two differential
equations:
∂ ( ρk ) ∂ ( ρkui ) ∂ ⎡ ∂k ⎤
+ = ⎢α k (μ + μt ) ⎥ + Gk + ρε (2)
∂t ∂xi ∂x j ⎢⎣ ∂x j ⎥⎦
∂ ( ρε ) ∂ ( ρεu i ) ∂ ⎡ ∂ε ⎤ C1∗ε ε ε2
+ = ⎢α ε (μ + μ t ) ⎥+ G k − C 2ε ρ (3)
∂t ∂xi ∂x j ⎢⎣ ∂x j ⎥⎦ k k
Cμ ρk 2 ⎛ ∂ui ∂u j ⎞ ∂ui
where, Gk = ⎜ + ⎟ , the generation term of the turbulent kinetic induced
ε ⎜⎝ ∂x j ∂xi ⎟⎠ ∂x j
by the average velocity gradient;
1 ⎛⎜ ∂u i ∂u j ⎞
⎟ , η = (2 E ⋅ E )1/ 2 k / ε ,
Eij = +
2 ⎜⎝ ∂x j ∂xi ⎟
⎠
ij ij
C1∗ε = C1ε − η (1 − η / η 0 ) /(1 + βη 3 ) ,
Cμ = 0.0845 , C1ε = 1.42 , C2ε = 1.68 , α k = α ε = 1.39 , η 0 = 4.377 , β = 0.012 .
The model constants above are for the RNG k-ε model which is short for renormalization
group k-ε model. This model[26] expand the unsteady Navier-Stokes equations according to
the Gauss statistic expansion, and derive the high Re number k-ε model using the method that
was used to filter the wave-number section of the fluctuation spectrum. RNG k-ε model has
the same form of the k and ε as the standard k-ε model, but its five model parameters are
obtained theoretically but not experimentally. The values above are the newest set of
parameters for RNG k-ε model being given in reference [27].
As the coefficient C1ε of the generation term of the ε equation is calculated basing on the
time-averaged strain rate Ei,j in the RNG k-ε model, it is a function of the coordinates in
space. At present, the RNG k-ε model is believed to be one the best two-equation models with
good combination property[28-31].
After the solving of the equations for k and ε, the turbulent viscosity stress (μt) can be
calculated according to Eq.(4), and finally the turbulent stress can be figured out according to
the Boussinesq assumption (Eq.(5)).
μ t = C μ ρk 2 / ε
(4)
⎛ ∂u ∂u j ⎞ 2⎛ ⎞
− ρ u i′u ′j = μ t ⎜ i + ⎟ − ⎜ ρk + μ t ∂u i ⎟δ ij
⎜ ∂x ⎟ 3 ⎜ ∂xi ⎟⎠
⎝ j ∂xi ⎠ ⎝ (5)
where, δij is the Kronecher delta. when i = j, δij = 1; and when i ≠ j, δij = 0.
In the solving process using RANS methods, if the first layer mesh next to the wall
locates outside the iso-stressed zone, then the wall function should be adopted at the first
layer of the mesh as the boundary condition; and else if the first layer mesh can reach the
viscous sub-layer, then the RANS model should be modified to solve the near wall problems.
The common method to modify the model is to introduce the damp coefficient fμ, and the
vortex viscosity coefficient can be written as:
ν t = c μ f μ u 'l
(6)
where, u is the characteristic velocity of the turbulent fluctuation, u =

'
'
k ; l is the
characteristic length, l = k / ε . In the calculations of this article with RANS methods, the
3/ 2
first layer mesh are all set to be within the viscous sub-layer, so Eq.(6) is conducted to modify
the low Reynolds number effect in the near wall zones.
Figure 3.2 shows the mesh model of the surface with periodic square cross-section
grooves. The grooves are located on the lower surface of the two-dimensional channel. d is
the groove depth, w is the groove width, and λ is the groove wavelength. The upper surface of
the channel is assumed to be flat and smooth. H is the height of the channel. The upper and
lower surfaces of the channel are set to be non-slip. Periodic boundary condition is applied
and three geometric periodicities are modeled in each case. The Reynolds number of the flow
is calculated according to the half-height of the channel and the main flow velocity. The mesh
around the grooves was refined as the variations of the flow around those areas are significant
and more accurate details are necessary.
Finite volume method are used to solve the RANS equations of incompressible fluid with
the commercial CFD software FLUENT. The turbulent models are based on the RNG k-ε
model and pressure gradient effect is considered. QUICK format is employed in the
discretization of momentum equations, PRESTO format are used to discrete the pressure
terms and SimpleC algorithm is applied in the coupling of the pressure and velocity. Second
order upwind format is conducted in the solving process of k and ε. Reference [32]~ [34] have
introduced the details of the numerical algorithms mentioned above.
Figure 2. Mesh mode Figure 3. Examine results of mesh independency

Examination of the mesh independence and validation of the numerical results
The fineness of the mesh is important for the numerical simulation. Mesh independence
mush have been examined to validate the credibility of the numerical results. As the thickness
of the boundary layer of the turbulent flow is quick small that if the mesh is not fine enough,
the accurate parameters of the flow could not be obtained.
As shown in Figure 3, four sets of simulation results have been obtained base on different
meshes. The fineness of these four set of mesh increase gradually. The mesh number of the
first set is 2688, and it is the coarsest one. The mesh of the latter one is one times finer than
the former one both in the flow direction and the normal direction.
Figure 3 shows the numerical simulation results based on these four sets of mesh. The
pressure gradient and drag coefficient vary as the increase of the mesh number. According to
the comparison of the simulation results, it can be found that for this problem when the mesh
density is as fine as or finer than the second set, the calculation results are almost mesh
independent.
The numerical results could also be validated by comparing with the experimental results.
Djenidi, et al[35] (1999) used the LDV to test and observe the flow around the transverse
grooves (w/d 1, d=5mm) in the water tunnel whose cross section is 250×250mm2. They
measured the velocity profile of a serious of sections located in the middle of the grooves.
The lower starting point in normal direction of each measurement is the top of the grooves.
The average velocity profile obtained in the test is based on the thickness of the boundary
layer. For convenience, numerical results and test data are compared base on the maximum
velocity location, and they are in good agreement (as shown in Figure 4).
Figure 4. Comparison of the average velocity profile between experiments and simulations
To sum up, it has been discussed above that the Reynolds averaged Navier-Stokes
equations, coupled with the RNG k-ε turbulence model with the near wall modifying
treatment, is adopted to investigate the turbulent flow over a plat surface with transverse
grooves.
General drag characteristic of the transverse grooves in turbulent flow
The objective of this research is the make further understanding of the drag characteristic
of the grooved surface in turbulent flow. As mentioned in Section 1, for transverse grooves,
the drag acting on the surface could be classified into two groups: the pressure drag which is
acting on the grooves’ surface perpendicular to the flow, and the viscous drag, which is acting
on the grooves’ surface parallel to the flow. Here, we can define the dimensionless coefficient
of the drag for comparison among grooved surfaces in different scales.
2 Dd 2( Ddp + Ddv )
Cd = =
ρU ∞ Ω t
2
ρU ∞2 Ω t (7)
2 Ddp 2 ∫ τ cos(t 1 , x1 )dΩ t

Ωt
C dp = =
ρU Ω t
2
∞ ρU ∞2 Ω t (8)
2 Ddv 2 ∫ p cos(n 1 , x1 )dΩ t

Ωt
C dv = =
ρU ∞ Ω t
2
ρU ∞2 Ω t (9)
where, Cd, Cdp and Cdv are total drag coefficient, the pressure drag coefficient, and the viscous
drag coefficient respectively; Dd, Ddp, and Ddv are the total drag force, the pressure drag and
the viscous drag respectively; Ωt is the wetted perimeter; τ is the viscous shear stress, p is the
pressure, t1 is the unit vector in the tangential direction, and n1 is unit vector in the exterior
normal direction.
For periodic square grooves (w/d = 1, λ/w = 2), the groove width could be defined as the
characteristic length of the groove.
As shown in Figure 5, the pressure rise induced by the grooves increases as the rise of the
characteristic length of the grooves (Re = 4.5×104). This pressure fluctuation causes
remarkable pressure difference between the windward side and leeward side of the grooves
and forms a significant pressure drag. Figure 6(a) and (b) show the variation of the drag
forces (the total drag, the pressure drag and the viscous drag) and the proportion of the
pressure drag and viscous drag to the total drag.
According to the curves and charts, two conclusions can be drawn for the drag
characteristic of the surface with periodic transverse square grooves:
(1) The drag coefficients are almost independent with the scale of the grooves;
(2) The pressure drag is about 4 times as large as the viscous drag.
As mentioned above the pressure drag is caused by the pressure difference acting on the
grooves’ surface perpendicular to the flow, and it is a kind of side wall effect which has been
discussed in Section 1. The difference between the results obtained in laminar flow and
turbulent flow is that the side wall effect is more significant in the latter case. So it should be
concluded that the real profile of the groove is important for the consideration of the effect of
grooves on the drag characteristic of the grooved surface. In some traditional researches on
drag reduction of rough surface in turbulent flow, where only the friction drag (the same as
the viscous drag mentioned in this article) is taken into consideration, were not adequate. The
pressure drag must be considered, and Reynolds equations should not be used to analyze the
drag characteristic of the transverse grooves in turbulent flow.
Figure 5. Pressure fluctuation cause by square grooves with different characteristic length
Figure 6. Drag coefficients of square grooves with different characteristic length
2. TRANSVERSE GROOVES ON DRAG CHARACTERISTIC OF THE

SURFACE IN TURBULENT FLOW
2.1. Effect of Reynolds Number
The relation between the drag characteristic of the grooved surface and Reynolds number
has been investigated as follows. Figure 7(a) shows variation of the drag coefficient of the
periodic transverse square grooves. As the increase of the Reynolds number, the drag
coefficients drop down at the same time. When Reynolds number increases from 10, 000 to
50, 000, drop velocity of the viscous drag coefficient is much larger than that of the pressure
drag coefficient. When Reynolds number is larger than 100, 000, the variation of the three
drag coefficients slow down significantly. Figure 7(b) demonstrates the variation of the
proportion of the pressure and viscous drag coefficient to the total drag coefficient as the
increase of Reynolds number. When Reynolds number increases from 10, 000 to 50, 000, the
quotient of the viscous drag decrease quickly, and when Reynolds number rises continuously,
the variation of the quotient of the viscous drag coefficient are slight.
Figure 7. Drag coefficients of square grooves with different Reynolds number
Reynolds number is a parameter represents the state and structure of the flow. The results
above indicate that the drag characteristic of the grooved surface is affected by the structure
of the flow. For square grooves, in the range that Reynolds number varies from 10, 000 to 50,
000, the structure of the flow may change more rapidly.
2.2. Effect of Width-to-Depth Ratio
Computational analysis in this part was performed to investigate the drag characteristics
of the surface grooves with different geometric parameters. For the rectangular grooves,
effect of the width-to-depth ratio of the grooves was focused on. The following simulation
were carried out in condition that Reynolds number of the flow is 10, 000.
Figure 8 shows the velocity stream line diagram and static pressure distribution on the
cross-section of grooves in different width-to-depth ratio (from 1/4 to 10/1).
When w/d is small (less than 2), single vortex almost takes up the whole groove. The
rotating center of the vortex and the center of the groove almost coincide. This is useful to
prevent the separation of the flow. But when w/d increases gradually, a negative pressure
region will generate just in the downstream zone of the windward side of the groove and
induce the local flow separation. Generally speaking, the separation induced by the grooves
will reduce the friction drag but increase the pressure drag. So w/d could be regarded as a
critical parameter to determine the flow state and drag characteristic.
Figure 8. Streamlined diagram and pressure distribution rectangular grooved surfaces with different w/d
ratios
When w/d rises to 7, the single vortex separates into two. These two vortices locate in the
two ends of the groove respectively. When w/d is larger than 7, the main stream will almost
reach the bottom of grooves.
Besides the negative pressure region around the downstream zone of the windward side,
the center of the vortex’s pressure is quick low too. The pressure in these two region decrease
as the grooves’ width increases. The former one changes more rapidly.
Drag characteristic of the rectangular grooves
Figure 9 shows the variations of the drag coefficients as the change of the width-to-depth
ratio of the grooves. The variation of the total drag coefficient, the pressure drag coefficient
and the viscous drag coefficient are drawn respectively. According to the curves, it can be
concluded that the grooves with small width-to-depth ratio are more favorable in drag
reduction design.
While w/d is between 1/1.5 and 2/1, the drag coefficient rises quickly in a linear way.
When w/d=3, the rise of the drag coefficient slows down. While w/d is between 6 and 7, the
drag coefficient would reach it maximum value. It should be noted that when w/d equals to 7,
two totally separate vortices are just going to generate.
Figure 9. Drag coefficients of square grooves with different width-to-depth ratio
When w/d<1/1.5, the viscous drag is larger than the pressure drag, and when w/d = 1/4,
viscous drag makes up more than 80% of the total drag. When w/d >1/1.5, pressure drag rises
quickly, while the viscous drag reduce slowly.
When w/d is large than 2, the viscous drag becomes negative which means the viscous
drag is in the opposite direction of the flow. This phenomenon can be explained as follows: as
the increase of the grooves’ width, the separation aggravate and the area where the viscous
drag is in the direction of the flow reduces, while the friction drag in the grooves whose
direction is opposite to the flow direction increases, and the integral effect is that the total
viscous drag becomes negative.
When w/d is larger than 1/1.5, the pressure drag becomes the major part of the total drag
force, which make the effect of viscous drag become less and less important. When w/d is
between 6 and 7, the pressure drag coefficient reaches its maximum value. At the same time,
the viscous drag coefficient arrived at it minimum point. When w/d is larger than 7, the main
flow begins to reach the bottom of the grooves directly, and the viscous drag starts to rise.
And the pressure drag force begins to reduce slowly after the generation of the two vortices as
the continuous increase of the grooves’ width.
To sum up, the drag characteristic of the rectangular grooved surface is concerned with
flow influenced by the grooves. The vortices in the grooves and their variation process play
an important part. The processes have direct effects on the main stream flow, the distribution
of the turbulent intensity and the flow condition next to the surface. The combined effect will
determine the drag force on the wall. Decreasing the width of the grooves is an effective way
to reduce the drag force.
3. CONCLUSION
In this section, the drag characteristic of the surface with transverse grooves in turbulent
flow is numerically studied. The pressure drag plays an important role in the total drag force.
The pressure drag and viscous friction drag are all relative to the turbulent flow structure.
According to the numerical results, following conclusions can be drawn:
(1) The side wall effect of the transverse grooves are more significant in turbulent flow
than in the laminar flow, the pressure drag inducing by the side wall effect could
make up about 80% of the total drag force when the Reynolds number of the flow is
higher than 50, 000.
(2) The turbulent flow structure in the grooves is considered to be the reason for the
variation of the pressure and viscous drag. And it is affected by the shape of the
grooves, such as the width-to-depth ratio. The generation and slitting of the vortexes
can change the direction of the net viscous drag, and also change the pressure
difference between the windward and leeward side of the grooves.
(3) For rectangular grooves, as the reduction of the viscous drag is lower than the
increment of the pressure drag, there is an increase in the overall surface with
transverse rectangular grooves comparing with the smooth one.
NOMENCLATURE
p = Fluid pressure
μ = Viscosity of fluid
h(x) = nominal film thickness
h1 = maximum film thickness
h0 = minimum film thickness
B = length of the disk
U = moving speed of the slider
σ = surface roughness height
d = groove depth
λ = wavelength of the squared transverse waves
β = ratio of roughness height to the film thickness; also model constant for RNG
k-ε model
ε = ratio of average film thickness to disk length; also turbulent dissipation rate
k = turbulent kinetic energy
E = expectancy function
ψi(H) = expectancy of the film thickness function
h ( x) = nominal film thickness
τij = stress on direction i and j
γ& = shear rate.
ρ = density of fluid.
x,y,z = Coordinates directions
u,v,w = Velocity along direction x, y and z, respectively; w: also groove width
u = velocity tensor
τ = stress tensor
ACKNOWLEDGEMENT
This work is supported by the State Key Laboratory of Tribology with the funding (No.
SKLT08B03), NSFC Project (No. 50975158 and No. 50505020) and the Doctoral Program
of Higher Education (No. 20070003103).
REFERENCES
[1] Elrod, HG. A General theory for laminar lubrication with Reynolds roughness, ASME
Journal of Lubrication Technology, 1979, 101, pp: 8-14.
[2] Dowson, D; Whomes, TL. The effect of surface roughness upon the lubrication of rigid
cylindrical rollers- I. Theoretical, Wear, 1971, 18, pp: 129-150.
[3] Shelly, P; Ettles, C. Effect of transverse and longitudinal surface waviness on the
operation of journal bearings, Jnl. Mech. Engrg. Sc., 1972, 14, pp: 168-172.
[4] Chang, L. Deterministic modeling and numerical simulation of lubrication between
rough surfaces- a review of recent developments. Wear, 1995, 184, pp: 155-160.
[5] Chiang, HL; Hsu, CH. Lubrication performance of finite journal bearings considering
effects of couple stresses and surface roughness, Tribology International, 2004, 37, pp:
297–307.
[6] Wang, LL; Cheng, IW. “An average Reynolds equation for non-Newtonian fluid with
application to the lubrication of the magnetic head-disk interface,” Tribology
Transaction, 1997, 40(1), pp: 111-119.
[7] Elsharkawy, AA. On the hydrodynamic liquid lubrication analysis of slider/disk
interface. International journal of mechanical sciences, 2001, 43, pp: 177-192.
[8] Yang, M, Talke, F. E. Effects of gas composition, humidity and temperature on the
tribology of the head/disk interface- Part I. Experiments. Tribology Transaction, 1996,
39(6), pp: 615-620.
[9] Boxenhorn, B; Greiff, P. Monolithic silicon accelerometer. Sensor and actuators, 1990,
A21-A23, pp: 273-277.
[10] Christensen, H. Stochastic models for hydrodynamic lubrication of rough surfaces.
Proc. Inst. Mech Engrs., 1969, 184, pp: 1013-1022.
[11] Patir, N; Cheng, HS; An average flow model for determining effects of three-
dimensional roughness on partial hydrodynamic lubrication. Journal of lubrication
technology, 1978, 100(1), pp: 12-17.
[12] Chin, R. Numerical studies of slow viscous flow between a moving plane wall and a
stationary wavy wall. Case western reserve university report, FTAS/TR, 1972,(6), PP:
72-79.
[13] Christensen, H; Tonder, K. The hydrodynamic lubrication of rough bearing surfaces of
finite width. Journal of lubrication technology, 1971, (7), pp: 324-330.
[14] Almqvist, T; Larsson, R. Some remarks on the validity of Reynolds equation in the
modeling of lubricant film flows of the surface roughness scale. Journal of tribology,
2004, 126(10), pp: 703.
[15] Sun, DC; Chen, KK. First effects of Stokes roughness on hydrodynamic lubrication.
Journal of lubrication technology, 1977, (1), pp: 2-9.
[16] Elsharkawy, AA. On the Hydrodynamic Liquid Lubrication Analysis of Slider/Disk
Interface. International journal of mechanical sciences, 2001, 43, pp: 177-192.
[17] Walsh, MJ; Lindemann, AM. Optimization and Application of Riblets for Turbulent
Drag Reduction. AIAA-84-0347.
[18] Walsh, MJ. Riblets as a viscous drag reduction technique. AIAA Journal, 1983, 21(4),
pp: 485-486.
[19] El-Samni, OA; Chun, HH; Yoon, HS. Drag reduction of turbulent flow over thin
rectangular riblets. International Journal of Engineering Science, 2007, 45, pp:
436-454.
[20] Liu, ZH; Dong, WC; Xia, F. The effects of the tip shape of V-groove on drag reduction
and flow field characteristics by numerical analysis. Journal of Hydrodynamics, 2006,
21(2), pp: 223-231.
[21] Nikuradse, K. Stromungsgesetze in Rauhen Rohren. [English Trans]. NACA Tech.
Memo No. (1292).
[22] Wang, SL; Shi, XJ; Zhao, SH; and et al. The technical research and application
development of groove surface in turbulent drag reduction. Journal of Southwest
Petroleum University, 2008, 30(1), pp: 146-150.
[23] Versteeg, HK; Malalasekera, W. An Introduction to Computational Fluid Dynamics:
the Finite Volume Method. Wiley, New York, 1995.
[24] Breuer, M; Jovicic, N; Mazaev, K. Comparison of DES, RANS and LES for the
separated flow around a flat plate at high incidence. International Journal for
Numerical Methods in Fluids, 2003, 41(4), pp: 357-388.
[25] Walter, DK; Leylek, JH. A new model for boundary layer transition using a single-
point RANS approach. Journal of Turbomachinery, 2004, 126(1), pp: 193-202.
[26] Yakhot, V; Orzag, SA. Renormalization group analysis of turbulence: basic theoy.
J Scient Comput, 1986, (1), pp: 3-11.
[27] Speziale, CG; Thangam, S. Analysis of an RNG based turbulence model for separated
flows. Int. J. Engng. Sci., 1992, (10), pp: 1379-1388.
[28] Chen, Q. Comparison of different k-e models for indoor air flow computations. Numer
Heat Transfer, Part A, 1995, 28, pp: 353-369.
[29] Rodi, W. Experience with two-layer models combining the k-e model with a one-
equation model near the wall. AIAA, 1991-0216.
[30] Yakhot, V; Orzag, SA; Thangam, S; Gastki, TB; Speziale CG. Development of
turbulence models for shear flows by a double expansion technique. Phys Fluids A,
1992, 4(2), pp: 1510-1520.
[31] Kim, SW; Chen, CP. A multiple-time-scale turbulence model based on variable
partitioning of the turbulent kinetic energy spectrum. Numerical Heat Transfer, Part B,
1989, 16, pp: 193-211.
[32] Hayase, T; Humphrey, JAC; Greif, R. A consistently formulated QUICK scheme for
fast and stable convergence using finite-volume iterative calculation procedures.
Journal of Computational Physics, 1992, 98, pp: 108-118.
[33] Van Doormal, JP: Raithby, GG. Enhancement of the SIMPLE method for predicting
incompressible fluid flows. Numerical Heat Transfer, 1984, 7, pp: 147-163.
[34] Patanker, SV; Spalding, DB. A calculation procedure for heat, mass and momentum
transfer in three-dimensional parabolic flows. Int J Heat Mass Transfer, 1990, 15, pp:
1787-1806.
[35] Djenidi, L; Elavarasan, R; Antonia, RA. The turbulent boundary layer over transverse
square cavities. J. Fluid Mech, 1999, 395, pp: 271-294.
Chapter 7
CURRENT LIMITING IN OXIDE CERAMIC

STRUCTURES
Alexander Bondarchuk*
Technological University of Mixteca, Huajuapan de Leon, Oaxaca CP. 69000, Mexico
ABSTRACT
The review of physical phenomena which lead to current limiting behavior (current
is increased weaker than voltage, saturated and even decrease) in oxide ceramic
structures are presented. Particular attention is given to the mechanisms of current
limiting in materials whose electrical conductivity is controlled by potential barriers at
the grain boundaries. In particular, the current saturation effect in the nano-grained
ceramic films is examined. The consideration of physical models describing the current
limiting in specific material is followed by the short posing of main experimental data.
For some models, the computational modeling has been performed.
INTRODUCTION
The electrical properties of oxide ceramic structures are directly linked to electrically
active interfaces and contacts between the adjacent grains. There are two types of contacts
between grains in polycrystalline semiconductors [1]: (I) contacts modeled by the double
Schottky barrier (Figure 1a) and (II) the intergranular regions in form of a neck (Figure 1b, c).
Usually, the grain contacts of the first type in oxide ceramic structures can appear at the
sintering in the oxidizing atmosphere. The contacts of second type (the neck) between the
adjacent grains arise at the sintering of ceramics in the vacuum or in the inert atmosphere [1].
The neck, depending on its thickness, defect concentration in the intergranular region and
a magnitude of the surface charge, can be (Figure 1b) or can not be (Figure 1c) a barrier for
the flow of electronic current between grains. If the potential barrier in the neck exists and its
*
Corresponding author: alexbond@mixteco.utm.mx
246 Alexander Bondarchuk
height is larger than kT , then such intergranular region is called the closed neck [1-3]. This
type of neck (Figure 1b) exists at the overlapping of the depletion layers in the intergranular
region that occur at the small thickness of the neck with a large magnitude of the surface
charge. If the neck thickness is sufficiently large and the overlapping of depletion layers does
not take place (Figure 1c), this is the open neck [1-3]. In this case, the barrier for the flow of
the electronic current does not exist.
Figure 1. Schematic layout of a contact between grains in polycrystalline n-type semiconductors: (a)
modeled by the double Schottky barrier; (b) open neck; (c) closed neck.
Potential barriers make the grain boundaries (GBs) highly resistive compared to the grain
interiors and therefore the current flow through the material bulk is controlled by GBs. In
polycrystalline n-type semiconductors, the potential barrier is formed by capturing of
electrons by localized states in the GB, resulting in a negative surface charge [4-6]. There are
several ways which such grain-boundary states can arise. The lattice mismatch alone is
sufficient to cause a planar array of localized states at the grain boundaries. Also, the
boundary regions may contain a large number of point defects and impurities. According to
[1, 7], the interface states can be generated by the adsorbed oxygen. Any of these conditions
can cause a space-charge layer to form on the grain boundary. This negative interface charge
is compensated by ionized, positively charged donors in a space charged region formed on
each side of the grain boundary. The formed grain-boundary barrier determines the current
flow between adjacent grains. Therefore the conductance of the material is controlled by
grain-boundary barriers and its nonlinear electrical property (deviation from Ohm’s law) is a
cumulative effect caused by individual GBs.
Deviation from Ohm’s law in oxide ceramic structures can be displayed in the form of
the superlinear current-voltage characteristic I ~ U ,
β
β > 1 (varistor effect) or the sublinear
Current Limiting in Oxide Ceramic Structures 247
I (U ) dependence with β < 1 (current limiting effect). The varistor effect is related to the
barrier height lowering on the grains boundary with the voltage increase. A detailed
description of such processes and the theory of the varistor behavior have been presented in
[4-6]. In the present time, the superlinear current-voltage characteristic has been observed in
ceramics on the basis of various n-type semiconductors ZnO, SnO2, TiO2, WO3 [8-20]. Such
ceramics with their polycrystalline nature and a large number of intergranular barriers (i.e., a
high energy absorption capability) exhibit the superlinear current-voltage characteristics and
are used as active components in overvoltage protection devices, in the varistores. The
varistores behave as variable resistors: they exhibit the high resistance at the low voltage
(insulators) and the low resistance at the high voltage (conductors). When the voltage exceeds
a certain value, for instance, during a voltage transient or surge, the varistor becomes highly
conducting and draws the current through it, usually to the ground. Due to their high
reliability and the high energy absorption capability, the varistors play an important role in
the radio engineering and power electronics. To make such electronic devices the deep
understanding of the nature of the varistor effect is required.
Ceramics with the ability to limit the current might have a great potential in new
overvoltage protection devices as well. However, in contrast to the varistor effect, the nature
of the current limiting behavior can be different in different materials. In many cases, the
mechanism of current limiting is not studied completely yet and the experimental facts of its
observation are not described sufficiently in the literature or this information is fragmentary.
In this chapter the short review of physical phenomena which lead to the current limiting
effect in the oxide ceramic structures are presented. This work was performed with the aim to
collect known experimental facts of current limiting in the oxide ceramic structures and to
consider the physical models describing such current behavior. The author hopes that this can
be useful for the systematical development of such materials and improvement of their
engineering performance.
07.03.2009 Alexander Bondarchuk
I. CURRENT LIMITING BEHAVIOR AS A RESULT OF THE CAPTURE OF

ELECTRONS AT GRAIN-BOUNDARY STATES
Current limiting behavior (current is increased weaker than voltage) caused by capturing
of electrons by localized states at the GBs can be observed in polycrystalline semiconductors
whose conductance is controlled by grain-boundary barriers [4, 21-24]. Sublinear current-
voltage dependence ( β < 1 ) appears at intermediate voltages between Ohmic ( β = 1 ) and
superlinear ( β > 1 ) regions (Figure 2) and can be explained in terms of the barrier model in
following way.
Let us assume that the potential barrier at a single grain boundary in polycrystalline
semiconductor can be modeled by the double Schottky barrier (two back-to-back connected
Schottky barriers) with zero thickness of intergranular layer (Figure 3). It is assumed too that
at the surface of the adjacent grains the localized electronic states exist due to the oxygen
adsorption, the presence of defects and dopants. Such grain-boundary states can catch
electrons moving from the interior of the grains to the boundary. Under equilibrium some of
these states, i.e., roughly those below the Fermi level, are occupied by electrons, giving a
negative surface charge (full circles in Figure 3a). However, some grain-boundary states
(above the Fermi level) are empty (open circles in Figure 3a). The negative charge localized
at the grain boundary is shielded by a region of positive space charge situated on the two
sides of the grain boundary. Assume as well that this positive space charge is due to the fully
ionized shallow donors. Then the condition of electroneutrality for the grain boundary can be
presented in the form:
m = N D ( L1 + L2 ) (1)
where m is the surface density of electrons captured at the grain-boundary states, N D is the
donor density in the grain, L1 and L2 are the thicknesses of the space-charge regions in
grains.
Figure 2. Schematic of current-voltage characteristic (I— Ohmic, II— sublinear, and III– superlinear
region) for single grain boundary in polycrystalline semiconductor.
Assuming that the discrete nature of the charges can be ignored, the magnitude of the
potential barrier can be calculated by solving the Poisson equation for the potential, Φ (x ) ,
from a knowledge of the grain-boundary charge density, ρ ( x ) [20]:
d2 ρ ( x)
Φ( x) = (2)
dx 2
εε 0
where ε is the relative permittivity and ε 0 is the permittivity of free space.

Under equilibrium at zero applied bias voltage (Figure 3a), the magnitude of the potential
barrier, ϕ 0 , is [20, 23] :
q 2 L20 N D q 2 m0 2
ϕ0 = = (3)
2εε 0 8εε 0 N D
Here q is the elementary charge, L0 is the thicknesses of the space-charge region in

separate grain at zero applied bias voltage ( L0 = L1 = L2 ), m0 is the surface density of
electrons captured at the grain-boundary states at U = 0 .
Figure 3. Energy level diagram for a grain boundary (a) at zero applied bias voltage and (b) under
applied voltage U .
From the solution of the Poisson equation, the following expressions can be written [23]:
q 2 L12 N D
ϕ 0 − qU1 = (4)
2εε 0
q 2 L22 N D
ϕ 0 + qU 2 = (5)
2εε 0
U = U1 + U 2 (6)
where U is a voltage applied across the grain boundary (Figure 3b).

Using the equations (1) and (3)-(6), the barrier height of forward-biased Schottky barrier
ϕ = ϕ 0 − qU 1 at a given voltage U (Figure 3b) can be obtained in the form
2
b m2 ⎛ ⎞
ϕ (U ) = ⎜1 − qU ⎟ , (7)
4 ⎜⎝ b m 2 ⎟⎠
where b = q
2
(2εε 0 N D ) .
Then the current density across the grain boundary (Figure 3b) in diode approximation
can be presented as
j = j1 − j2 ,
where j1 = 1 4 qυN D exp[− ϕ (U ) kT ] , j2 = 1 4 qυN D exp[− (ϕ (U ) + qU ) kT ] .

Here υ is the average thermal velocity of electrons, k is the Boltzmann constant and T is the
absolute temperature.
Or in compact form:
q υ ND ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞
j= exp⎜ − ⎟ ⎜1 − exp⎜ − ⎟⎟ , (8)
4 ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠
The surface density of trapped electrons is determined by the capture of electrons and the
emission processes at grain-boundary states [23]:
dm ⎛ dm ⎞ ⎛ dm ⎞
=⎜ ⎟ −⎜ ⎟ ,
dt ⎝ dt ⎠1 ⎝ dt ⎠ 2
⎛ dm ⎞ ⎛ j j ⎞
⎜ ⎟ = σ ( M − m ) ⎜⎜ 1 + 2 ⎟⎟,
⎝ dt ⎠ 1 ⎝ q q ⎠
⎛ dm ⎞ ⎛ ES ⎞
⎜ ⎟ = σ υ m N C exp⎜ − ⎟.
⎝ dt ⎠ 2 ⎝ kT ⎠
Here (dm dt )1 is the capture rate of electrons which is proportional to the quantity of
electrons crossing the grain boundary, the capture cross-section σ and the density of empty
grain-boundary states (M − m ) (where M is the surface density of the localized electronic
states); (dm dt )1 is the emission rate of electrons from the grain-boundary states into the
conduction band, E S is the ionization energy of the grain-boundary state.
Thus,
dm υ ND ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞
= σ ( M − m) exp⎜ − ⎟ ⎜1 + exp⎜ − ⎟⎟
dt 4 ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠ (9)
⎛ E ⎞
− σ υ m N C exp⎜ − S ⎟.
⎝ kT ⎠
The equation (9) describes the filling of grain-board states with voltage increase. The
general picture of such processes can be given in the following way.
Under equilibrium, some of the grain-board states remain empty at zero applied bias
voltage (Figure 3a). Application of dc voltage causes the barrier decrease. Therefore the
capture rate of electrons (dm dt )1 is increased and the surface density of trapped electrons
m is increased. The increase of negative charge localized at the grain boundary leads to
some increase of the barrier height and the rate of capture process (dm dt )1 is decreased.
The capture process is “autohampering” [24].
At low voltages, the change of the occupation of the grain-boundary states is negligible
due to a weak change of the barrier height and Ohm’s law is performed. Further voltage
increase leads to the stronger barrier decrease and the capture rate of electrons is increased.
The capture of additional electrons at the grain boundary results in some rise of the barrier
height that somewhat compensates the decrease of the barrier height after an application of
voltage. Therefore the current density j is increased weaker than voltage, i.e. the sublinear
j (U ) dependence takes place.
The voltage increase leads to the filling of the grain-boundary states and the density of
empty grain-boundary states, ( M − m ) , is decreased. Therefore, the intensity of a capture
process is decreased at high voltage and the occupation of the grain-boundary states is weakly
changed. In these conditions, as a result of the sharp lowering of the barrier height under
applied high voltage, the sublinear region at j (U ) curve is changed to the superlinear one.
Calculated dependences of j and ϕ versus U are shown in Figure 4. The calculus was
performed at the assumption that the electronic equilibrium ( dm dt = 0 ) at the grain
boundary are reached at each discrete value of voltage. The next parameters were used [23]:
18 −3
the effective density of states in the conduction band Nc = 6.0 × 10 cm , the average
thermal velocity of electrons υ = 1.0 × 107 cm s −1 , the donor density in the grain
N D = 1.0 × 1017 cm −3 , the ionization energy of grain-boundary states E S = 0.6 eV , the
relative dielectric permittivity of the grain is assumed to be ε = 8.5 , the absolute
temperature T = 290 K , and the value of the barrier height ϕ 0 = 0.5 eV at U = 0 .
For convenience, the results of calculations are presented in the normalized form. The
current density is normalized on the value j0 = qυN D 4 and multiplied by coefficient 10 .
10
The magnitude of the potential barrier and voltage is normalized on the value of the barrier
height at U = 0 , ϕ 0 / q .
Figure 4. Normalized dependences of the current density (curve 1) and the barrier height (curve 2) on
the voltage.
Calculated current-voltage dependences (Figure 4) are contained sublinear region

(current limiting) in compliance with the qualitative consideration given above and the
experiment data [21, 22].
II. CURRENT LIMITING EFFECT CAUSED BY ADSORPTION

PROCESSES IN ELECTRIC FIELD
Recently [25–29] it was reported about preparation of In2O3-SrO ceramics with the
capability to limit the current by voltage increase. This type of current-voltage dependence in
In2O3-SrO ceramics sample is shown in Figure 5 in linear (curve 1) and logarithmic (curve 2)
scale. It contains linear region ab at low voltages (Ohm’s law), superlinear bc at higher
voltages which is followed by the current limiting region: sublinear part cd and region de
where current is even decreased (Figure 5, curve 1). Next measurements give nearly the same
curve with current limiting region but it is slightly shifted at low voltages (Ohm’s law) to
higher currents due to the degradation phenomena related to water adsorption [29].
The adsorption effects were studied at room temperature just after the cooling the
ceramics sample from high temperature in the air [29]. The increase and subsequent decrease
of the ceramics resistance on time at fixed voltage from Ohm’s region was found. During
about 1–2 h just after the cooling to room temperature oxygen adsorption takes place and
resistance is increased. Later on water adsorption causes the decrease of the resistance.
Adsorption processes at the grain surface in the bulk of this material are likely due to its high
porosity revealed by scanning electron microscopy [29].
It was found out that current saturation region is observed if measurements were carried
out in air and it is not observed if measurements were performed in argon [29]. In argon
I (U ) dependence was nearly linear. In addition it was shown that even if current saturation
region is disappeared as a result of long-term degradation, current limiting effect in In2O3-SrO
ceramics can be reproduced in air again after the heat treatment of a sample in Ar at 1070 K.
In that way, it was ascertained sensibility of ceramics to external environment. Current
limiting region is observed only if there is enough oxygen in atmosphere and it is not
observed if there is a low oxygen content in atmosphere. Therefore, this current limiting
effect rather can be related to the adsorption of oxygen.
Figure 5. Current versus voltage for In2O3-SrO sample in air in linear (curve 1) and logarithmic
(curve 2) scale.
However, the current limiting effect in In2O3-SrO ceramics can not be explained by the
adsorption of oxygen which running independently of the applied voltage. Next experimental
facts show that. The dependence of current on time in In2O3-SrO ceramics at fixed voltage is
changed qualitatively with increase of voltage [25]: current is slightly increased with time at
low voltages and it is decreased with time at voltages in the current limiting region ce in
Figure 5. It means that processes responsible for discussed effect are controlled by voltage. In
addition, the study of current-voltage dependences was always started in a few hour’s time
after the cooling of a sample to room temperature, when water adsorption was dominated and
the sample resistance was decreasing [29]. It confirms the fact of slight shift of I (U )
dependence in the Ohm’s law region at each measurement to higher current region. In view of
these facts, the assumption about the relation of observed current limiting behavior to the
oxygen adsorption (which can take place independently of the applied voltage) should be
excluded. However, the oxygen adsorption can be induced and controlled by applied voltage
[30, 31].
An application of voltage can cause a capture of electrons at empty grain-boundary states
and, therefore, a sublinear I (U ) dependence can appear between linear and superlinear
regions (see section I). However, in In2O3-SrO ceramics, on the contrary, the sublinear region
ce in Figure 5 is observed after the superlinear region. This fact cannot be explained in the
frames of the model discussed in section I.
For the explanation of the observed current limiting effect in In2O3-SrO ceramics it is
useful to modify the existing barrier model. Let us assume that the density of grain-boundary
states is increased with voltage due to the adsorption of oxygen:
(i) the grain-boundary states are caused by the adsorbed oxygen; the capture of electrons
at the grain-boundary states means the transition of adsorbed oxygen from neutral to
charged form; the density of empty and filled grain-boundary states are equal to the
density of adsorbed oxygen in neutral and charged forms, respectively; the
correlation between the charged and neutral forms of adsorbed oxygen is controlled,
according to Volkenshtein [7], by the position of the Fermi level at the surface of a
sample;
(ii) the adsorption equilibrium at the grain surface is kept up at the expense of adsorption
and desorption oxygen only in the neutral form; oxygen in the charged form does not
take part in an exchange with the atmosphere.
On the basis of these assumptions the general picture of processes can be given in the
following way [29]. At low voltages electronic equilibrium at the grain boundary is broken
and the capture of electrons is dominated over their emission from the grain-boundary states.
It leads to some increase of the barrier height and the rate of capture process is decreased. The
capture process is “autohampering” [24]. Therefore, current limiting region can be observed
after the Ohm’s law region (see section I).
As a result of the electron capture the transformation of adsorbed oxygen to a charged
form takes place, amount of adsorbed oxygen in the neutral form is decreasing and
adsorption-desorption equilibrium at the grain boundaries is broken. This causes the
additional oxygen adsorption. However, at low voltages it is negligible due to a weak
deviation of the density of adsorbed oxygen in a neutral form from its equilibrium value.
Respectively, at low voltages the additional oxygen adsorption cannot have strong influence
on the number of empty grain-boundary states.
With increase of voltage the barrier height is decreased and current is increased with
voltage superlinearly. However, some capture processes are continued and, respectively, the
density of adsorbed oxygen in a neutral form is decreased. It means that deviation from the
adsorption equilibrium is being increased with voltage. This leads to a strong increase of
additional oxygen adsorption. Therefore, the density of empty grain-boundary states can be
gradually increased with voltage. As a result, the capture of electrons becomes stronger and
superlinear region of I (U ) curve is replaced by the sublinear one or even the decrease of
current could take place. The growth of the barrier height leads to the decrease of the capture
rate and, therefore, it leads to the decrease of oxygen adsorption rate. Both processes are
“auto-hampering” ones. In this way the equilibrium between capture and emission of
electrons and between adsorption and desorption of oxygen is settled.
The necessary condition for additional oxygen adsorption and realization of assumed
mechanism of the barrier height growth is the presence of empty adsorption centers at the
grain surface. However, during a long stay of a sample in air these centers can be filled with
adsorbed water, physically adsorbed oxygen, etc. As a result, additional oxygen adsorption
caused by voltage application is negligible and the density of grain-boundary states is almost
constant. Accordingly, I (U ) dependence of In2O3-SrO ceramic sample kept in air for a long
time does not contain the current limiting region after superlinear one but the heat treatment
of a sample in Ar at 1070 K again creates empty adsorption centres at the grain surface and
current limiting region after superlinear one is appeared again. This can explain observed
degradation behavior.
Assume that the potential barrier at a single grain boundary in In2O3-SrO ceramics can be
modelled by the double Schottky barrier considered in the section I. In this case, the current
density j across the grain boundary in diode approximation will be determined by the
following system of equations (more detail see in section I):
q υ ND ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞
j= exp⎜ − ⎟ ⎜1 − exp⎜ − ⎟ ⎟, (1)
4 ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠
2
b m2 ⎛ ⎞
ϕ (U ) = ⎜1 − qU ⎟ , (2)
4 ⎜⎝ b m ⎟⎠
2
dm υ ND ⎛ ϕ (U ) ⎞ ⎛ ⎛ qU ⎞ ⎞ ⎛ E ⎞
= σ ( M − m) exp⎜ − ⎟ ⎜1 + exp⎜ − ⎟ ⎟ − σ υ m N C exp⎜ − S ⎟, (3)
dt 4 ⎝ kT ⎠ ⎝ ⎝ kT ⎠ ⎠ ⎝ kT ⎠
where ϕ is the magnitude of the potential barrier, m is the surface density of electrons
captured at the grain-boundary states, U is a voltage applied across the grain boundary, q is
the elementary charge, M is the density of the localized electronic states, N D is the donor
ε is the relative permittivity and ε 0 is the permittivity of free space, υ
density in the grain,
is the average thermal velocity of electrons, σ is the capture cross-section, k is the
Boltzmann constant and T is the absolute temperature; b = q (2εε 0 N D ) .
2
According to the model suggested in [29], the application of voltage causes the additional
trapping of electrons at the grain-boundary states and this leads to the increase of the density
−
of adsorbed oxygen in charged form ( N ). Therefore, the surface density of adsorbed
oxygen in neutral form N (t ) is decreased:
N (t ) = M − m(t ). (4)
The density of empty grain-boundary states M − m(t ) is equal to the surface density of
adsorbed oxygen in neutral form N (t ) .
The decrease of N (t ) due to the capture of electrons (transition of adsorbed oxygen
from neutral to charged form) breaks the adsorption equilibrium and causes the additional
adsorption of oxygen. This process can be described by the equation:
dN dm
= α (N SS − N ) − . (5)
dt dt
Here first term gives the rate of additional adsorption caused by the application of voltage, α
is the adsorption rate coefficient. First term is proportional to the difference between the
steady-state and variable surface densities of adsorbed oxygen in neutral form. The second
term gives the rate of N (t ) decrease due to the transition of adsorbed oxygen from neutral to
charged form.
In a general case the barrier height and the current density are described by equations
(1–5). The steady-state surface density of adsorbed oxygen in neutral form N SS can be found
from equation (3) at the conditions dm dt = 0 and U = 0 :
⎛ ϕ (0) − E S ⎞
N SS U =0 = M − m0 = 2m0 N C N D−1 exp⎜ ⎟, (6)
⎝ kT ⎠
where m0 = (4ϕ (0) b )

12
is the equilibrium ( U = 0 ) surface density of trapped electrons.
The solution of equations (1–5) was obtained by the numerical methods. The next
parameters were used: the barrier height ϕ 0 = 0.3 eV at U = 0 (taken from the
temperature dependence of electrical conductivity [27]); the electron capture cross-section
σ = 1.0 × 10 −15 cm 2 , the effective density of states in the conduction band
Nc = 6.0 × 1018 cm −3 , the average thermal velocity of electrons υ = 1.0 × 107 cm s −1 , the
17 −3
donor density in the grain N D = 2.0 × 10 cm , the ionization energy of grain-boundary
states E S = 0.45 eV , the relative dielectric permittivity of the grain is assumed to be
ε = 8.5 , the absolute temperature T = 290 K , the adsorption rate coefficient
α = 0.004 s −1 (this value was estimated from experimental time dependence of
current[29]). In calculations of j (U ) dependence, voltage was increased by the steps
ΔU = 0.004 V with the duration of each step Δt = 3.0 s . In the experiment with ceramic
sample (Figure 5) the step voltage is higher1 but ceramic sample contains a great number of
grain boundaries connected in series.
For convenience, the results of calculations (Figure 6 and Figure 7) are presented in the
normalized form. The current density is normalized on the value j0 = qυN D 4 and
6
multiplied by coefficient 10 . The magnitude of the potential barrier and the voltage are
normalized on the value of the barrier height at U = 0 , ϕ 0 / q , the density of empty and
1
Voltage was increased consecutively by the steps ΔU = 0.2 V with the duration of each step Δt = 3.0 s [29].
filled grain-boundary states at U ≠ 0 are normalized on the value m0 , the density of filled
grain-boundary states at U = 0 .
Calculations of j (U ) curves are performed for two cases. In the first case voltage is
raised fairly slowly so the adsorption equilibrium ( dN dt = 0 ) and electronic equilibrium
( dm dt = 0 ) at the grain boundary are reached at each discrete value of voltage. To model
this situation equations (1–3) at a condition dm dt = 0 were solved jointly. Equation (5)
gives simply N = N SS . Obtained results are shown in Figure 6.
In the second case voltage is raised relatively quickly (what frequently takes place in an
experiment) so the adsorption and electronic equilibrium at the grain boundary rather are not
reached. To model this situation equations (1–5) were solved jointly. Obtained results are
shown in Figures 7.
If voltage is raised fairly slowly (first case), j (U ) dependence is sublinear (Figure 6a,
curve 1). In this case the density of empty grain-boundary states is constant (Figure 6b, curve
2) because the surface density of adsorbed oxygen in neutral form N (t ) is kept unchanged
(there is enough time for the adsorption of additional amount of oxygen). This ensures fairly
strong trapping of electrons (Figure 6b, curve 1). As a result, the decrease of the barrier height
is slightly slowed down (Figure 6a, curve 2) and current is increased with voltage weaker than
linearly (Figure 6a, curve 1). Similar behavior of j (U ) dependence was found in the
experiment if current was measured nearly in the steady-state conditions [25].
If voltage is raised relatively quickly (second case), j (U ) curve is more complicated
(Figure 7a, curve 1). At low voltages additional adsorption is negligible due to a weak change
of the occupation of the grain-boundary states (Figure 7b, curve 1). However, there are
a lot of empty grain-boundary states (Figure 7b, curve 2). The trapping of electrons leads to
some not too sharp decrease of the barrier height (Figure 7a, curve 2).
Therefore, at low voltages some slight sublinear region at j (U ) curve takes place
(Figure 7a, curve 1). Further voltage increase leads to the filling of the grain-boundary states
(Figure 7b, curve 1) and, therefore, the intensity of a capture process is decreased. In these
conditions, the sublinear region at j (U ) curve is changed to even slightly superlinear one
(Figure 7a, curve 1). However, after the filling of a majority of empty states some deficiency
of oxygen in neutral form appears and additional adsorption becomes a substantial factor.
Increase of oxygen adsorption causes the strong increase of the density of empty grain-
boundary states (Figure 7b, curve 2).
The curve 2 in Figure 7b has minimum and the density of empty grain-boundary states
raises at higher voltages. As a consequence, trapping is increased strongly (Figure 7b, curve
1) and the barrier height is passed through a minimum and then it is increased (Figure 7a,
curve 2). Accordingly, superlinear region at j (U ) curve is changed to sublinear one and the
current density is even decreased with voltage (Figure 7a, curve 1).
Figure 6. (a) Normalized dependences of the current density (curve 1) and the barrier height (curve 2),
(b) the density of filled (curve 1) and empty (curve 2) grain-boundary states on the voltage.
Calculations are performed for slow voltage increase (adsorption equilibrium takes place).
Figure 7. (a) Normalized dependences of the current density (curve 1) and the barrier height (curve 2),
(b) the density of filled (curve 1) and empty (curve 2) grain-boundary states on the voltage.
Calculations are performed for fairly rapid voltage increase (adsorption equilibrium does not take
place).
In that way, considered model of grain-boundary controlled conduction can be applied to

explain observed nonlinear current-voltage characteristics of In2O3-SrO ceramics having
superlinear region which is followed by the sublinear region at higher voltages. The key point
of the model and its radical difference from the known models is a relation between electronic
and adsorption processes: capture of electrons at the application of voltage causes an
additional oxygen adsorption and subsequent capture of electrons, respective increase of the
barrier height and appearance of the current limiting region at current-voltage curve.
Calculations give reasonable approximation to experimentally observed current-voltage
dependences.
Similar processes leading to the current limiting effect may be possible in other oxide
ceramics with barrier-controlled conduction. How it has been reported in [32], in SnO2-based
ceramics sintered at low temperature the current limiting effect appears if there is enough
oxygen in atmosphere and it is not observed if there is a low oxygen content in atmosphere
(Figure 8, curves 1-4). The low-field conductivity in argon is slightly lower than in air (Figure
8). It can be assumed that such behaviour is related to the adsorption of water molecules
which are in air. In contrast, in a flux of dry argon the ceramic surface is substantially cleaned
by water. The low-field conductivity in SnO2–Co3O4–Nb2O5–Cr2O3 ceramics is highly
sensitive to the variation in the relative humidity of air [33] and this behavior is typical for
SnO2-based materials [34, 35].
Figure 8. Current versus voltage in the SnO2–Co3O4–Nb2O5–Cr2O3 sample recorded in argon (curves 1
and 2) and in air (curves 3 and 4) at increase (1, 3) and decrease (2, 4) of voltage.
In addition, the current limiting effect in SnO2–Co3O4–Nb2O5–Cr2O3 ceramics can be

transformed in oxidizing atmosphere into the varistor effect [32]. The first application of
voltage to the fresh sample (which was not subjected to the electric field influence) in air
always leads to the current limiting behaviour. Current is saturated and even decreased with
the rise in voltage (Figure 8, curve 3). However, the decrease in the voltage gives a varistor-
like superlinear curve (Figure 8, curve 4). The rise in voltage in the following measurements
again gives a current limiting effect but the I (U ) curve is shifted to lower currents and the
mentioned effect becomes less expressed. Further applications of voltage in air (5–10
measurements at a fixed voltage polarity) give a practically reproducible superlinear curve

shifted slightly lower than I − U curve obtained at first measurement. Thus, the current
limiting effect in tin dioxide ceramics is transformed into the varistor effect [32].
According to the barrier model considered above, such behavior of current versus voltage
in tin dioxide based ceramics can be explained in following way. In SnO2–Co3O4–Nb2O5–
Cr2O3 ceramics the application of the voltage in air can cause additional oxygen adsorption.
As a result of it the barrier height grows and the current limiting effect takes place. At the
voltage decrease additional oxygen adsorption is negligible, only electronic processes take
place and therefore the varistor behaviour is observed.
III. CURRENT LIMITING IN CERAMICS BASED ON BaTO3 INDUCED BY

THE CHANGE OF THE CRYSTALLINE STRUCTURE
Very stable and highly reproducible current-limiting characteristics have been observed
for polycrystalline ceramics prepared from sintering mixtures of coarse-grained, donor-doped
BaTiO3 (tetragonal) as the major phase and ultrafine, undoped cubic perovskite such as
BaSnO3, BaZrO3, SrTiO3 or BaTiO3(cubic) as the minor phase [36, 37].
The current-voltage curves of such materials have a strong initial maximum and then the
current decreases to a nearly steady value as the voltage is increased. At low voltages the
current increases linearly (Figure 9a) and the power dissipation through Joule heating raises
the body temperature. Close to the Curie temperature ( TC ) the resistance of the material
increases sharply (Figure 9b) and the current drops to a near limiting value.
The Curie temperature is defined as the point at which an alteration of the crystalline
morphology occurs. The crystal structure of BaTiO3 ceramics changes with heating from its
initial tetragonal, non-symmetrical configuration below the Curie temperature (~120 ◦C), to a
cubic perovskite symmetrical configuration above the Curie temperature. This phase
transitions is accompanied by a marked peak in the relative permittivity ( ε ) vs. temperature
curve [38-46]. With increasing temperature the permittivity is increased below the Curie point
and is decreased above TC . Above the Curie temperature the relative permittivity in BaTiO3
ceramics follows Curie-Weiss law and can be described accurately by a simple relationship
[44-46]:
C
ε= (1)
T − TC
where C is the Curie constant and T is the temperature.

The anomalous increase of the resistivity in BaTiO3 ceramics is known as the PTCR
effect and is used for Positive Temperature Coefficient (PTC) thermistors. The generally
accepted model that describes this behavior was proposed by Heywang [40, 41] and later on
was extended by Jonker [42]. According to the Heywang model, the PTCR effect in BaTiO3
ceramics originates from the potential barrier that is formed at the grain boundaries increasing
ρ S in the grain-boundary region above the Curie

the electrical resistance. The resistivity
temperature depends exponentially on the magnitude of grain-boundary barrier ϕ :
⎛ϕ ⎞
ρ S ~ exp⎜ ⎟ , (2)
⎝ kT ⎠
where ϕ ~ 1 ε (T ) , ε (T ) is the relative permittivity of the grain-boundary region that

depends on the temperature; k is the Boltzman constant.
Figure 9. (a) Static current-voltage and (b) resistance-temperature characteristics of PTC Thermistor
produced by Murata Manufacturing co., Itd.
Above the Curie temperature, the relative permittivity ε decreases with heating (see (1))
and the magnitude of grain-boundary barrier ϕ increases. According to (2), this leads to the
sharp increase of resistivity at the temperature T > TC . Thus, the resistance increase of
ceramics is due to the change in the dielectric constant in the boundary region between
BaTiO3 grains. For the first time this idea was expressed by Heywang [40, 41] and later on
verified by other researchers [44, 46-48].
Unfortunately, the examined model can not explain the electrical behavior of BaTiO3
materials below the Curie temperature, in particular the low resistivity of ceramics and the
influence of polarization on the resistivity which takes place at the room temperature.
The explanation for electrical properties of BaTiO3 ceramics below the TC was proposed
by Jonker [42]. His model is based on the ferroelectric behavior of BaTiO3 below the Curie
temperature. According to Jonker, the spontaneous polarization in domains near the grain
boundary regions neutralizes partly the charge localized at the grain boundaries. This affects a
decrease in the potential barrier height allowing for many electrons to transport across the
grain boundary.
In spite of the convincing explication of the PTCR-effect base in BaTiO3 ceramics, the
examined models can not interpret in detail all aspects of this phenomenon. The PTCR
phenomenon is a complex interaction of mechanisms involving grain boundary effects,
ferroelectric crystal effects, and defect mechanisms. Therefore the creation of a unified
theory, which takes into consideration all the interdependent parameters and provides good
correlation with experiment, is indeed a difficult task. The investigations in this direction are
still continued. In particular, the non-linear current-voltage characteristics of such materials
have been observed and modelled by Mallick and Emtage [47], Al-Allak et. al [48], Zhang
and Cao [49]. The detailed consideration of these models is beyond the scope of this section
and can be found in literature [47-49].
IV. CURRENT SATURATION IN NANO-GRAINED OXIDE CERAMICS

As reported by Neto et al. in [51], current limiting effect (current is increased weaker
than voltage and even saturated) have been observed in nano-structured ZnO ceramic films
prepared by sol-gel processing. Schematic of current-voltage characteristic for such materials
is presented in Figure 10. The current vs. voltage curve contains three regions: linear at low
voltages, superlinear at intermediate voltages ( β > 1 ), and sublinear at higher voltages
( β < 1 ).
Figure 10. Schematic of current-voltage characteristics (I — Ohmic, II — sublinear and III superlinear
region) for ZnO ceramic films obtained in [51].
As follows from the experimental data reported in [51], the saturation current (the region
III in Figure 1) exhibits Arrhenius behavior with an activation energy on the order of a few
tenths of an electron volt; the activation energy of conductivity in a strong field ( E2 ) is
somewhat lower than that in a weak field ( E1 ); a specific feature of obtained ZnO ceramic
films is a small grain size ( l g ~ 4 nm ) . However, unfortunately, an explanation of these
facts and of the nature of current limiting behavior was not presented in [51].
The current limiting effect observed in [51] was explained by the simple grain-boundary
model in [52]. This model takes into account the fundamental feature of semiconducting
oxide ceramics — the presence of grain-boundary potential barriers and the small grain size
of ZnO ceramics obtained in [51]. The proposed mechanism for current limiting behavior in
such nano-structured oxide ceramics consists in the following [52].
Let us assume that the voltage U applied to two adjacent grains separated by grain-
boundary barrier (in detail see in section I). Then, the height of the forward-biased barrier ϕ
is
2
⎛ qU ⎞
ϕ (U ) = ϕ 0 ⎜⎜1 − ⎟⎟ , (1)
⎝ 4ϕ 0 ⎠
where ϕ 0 is the equilibrium magnitude of the grain-boundary barrier at U = 0 ,

ϕ 0 = q 2 N D L20 ( 2εε ) −1 . Here q is the elementary charge, N D is the donor density in the
grain, L0 is the equilibrium (at U = 0 ) thickness of the space-charge region on one side of
the boundary, ε is the relative permittivity and ε 0 is the permittivity of free space.
In a diode approximation, the current through the grain boundary is equal to the
−1
difference between the forward ( I 0 exp[ −ϕ ( kT ) ] ) and reverse
−1
( I 0 exp{−[ϕ + qU ]( kT ) } ) currents. For qU >> kT , the current-voltage dependence
can be represented by
⎡ ϕ ⎤
I = I 0 exp ⎢ − ⎥, (2)
⎣ kT ⎦
where ϕ is given by (1), k is the Boltzmann constant and T is the absolute temperature.
With increasing voltage, the thickness of the space-charge region in the lower potential
grain decreases slightly, while that in the higher potential grain increases. At the saturation-
onset voltage VS , the thickness of the depletion region in the higher potential grain attains the
grain size l g . Upon a further increase in voltage, the positive space charge in the higher
potential grain cans no longer increase. This implies by virtue of the constant negative charge
localized at the grain boundary and the electroneutrality condition that the height of the
forward-biased barrier, ϕ , which determines the current, remains constant as a function of

voltage, and the current saturates at I S .
In the current-saturation regime, the thickness of the space-charge region in the higher
potential grain is constant, equal to l g and the bottom of the conduction band is lowers in this
1 dϕ
grain. The electric field in this grain, given by , is higher.
q dx x =0
Thus, the current saturation results from the constant height of the current-controlling
barrier at a sufficiently small grain size, l g > L0 .
~
The saturation-onset voltage VS can be found by solving jointly the equation (1) and the
expression ( )
ϕ + qU = q 2 N D 2εε 0 l g2 which was obtained at the assumption that the
space-charge region is spread with voltage over the whole grain:
ϕ ⎛ lg ⎞
US = 4 ⎜⎜ − 1⎟⎟ . (3)
q ⎝ L0 ⎠
Taking (1) and (3) into account, we obtain from (2)
⎛ E ⎞
I S = I 0 exp⎜ − ⎟, (4)
⎝ kT ⎠
where the activation energy E at saturation is given by
2
⎛ l ⎞
E = ϕ 0 ⎜⎜ 2 − g ⎟⎟ . (5)
⎝ L0 ⎠
Figure 11 shows the logarithm of the saturation current as a function of the square of the
mean grain size (experimental data reported by Neto et.al. [51]). The saturation current rises
with mean grain size because of the increase in carrier-depletion voltage. As apparent from
2
Figure 11, log I S is proportional to l g , in full accord with relations (4) and (5).
The saturation current shows thermally activated behavior (Figure 12), in accordance
with (4), because the probability of carriers overcoming the barrier of fixed height rises with
temperature. The activation energy derived from the data in Figure 12 is E2 = 0.16 eV .
How it was reported in [51], the activation energy of conductivity in strong fields, E2 , is
somewhat lower than that in weak fields, E1 . In the model [52], E1 = ϕ 0 + ς , where ς is
the energy separation between the Fermi level and the bottom of the condition band in the
quasi-neutral part of the grain, and E2 = E + ς . Therefore, we obtain from (5) E2 < E1
since l g > L0 , in agreement with experiment.
~
Figure 11. Logarithm of the saturation current ( U = 380 V ) as a function of the square of the mean
ZnO grain size [51].
Figure 12. Arrhenius plot of the saturation current ( U = 380 V ) for ZnO film [51].
According to [52], one would expect from (3) and (5) that the model is applicable in the
range 1 < l g L0 < 2 . Although the upper limit on l g L0 is probably a characteristic of the
model, the proposed mechanism can be responsible for the sublinear current-voltage behavior
of fine-grained ( l g > L0 ) semiconductor ceramics. In particular, the examined model [52]
~
would possibly explain the current limiting observed in nano-grained SnO2 ceramic films
[53]. Unfortunately, the experimental data presented in [53] are not enough to verify this
supposition.
V. SUBLINEAR CURRENT-VOLTAGE DEPENDENCE IN Pb

(ZrXTi1-X)O3 THIN FILMS
The electrical properties of lead-zirconate-titanate thin film were the subject of inquiry in
many works [54-65]. This interest can be explained by a ferroelectric behavior of such
materials which are promising candidates for a variety of electronic applications [65]. In
addition, a significant feature of these materials is a sublinear current-voltage dependence at
intermediate- and high voltage.
Figure 13 presents a schematic drawing of typical I − U curves obtained at room
temperature for Pb(ZrxTi1-x)O3 (PZT) thin films. In that current-voltage curve, one can
distinguish a region of a steep current increase followed by a sublinear I (U ) dependence at
intermediate- and high voltage. This structure of the I − U curve is observed for films
prepared in the different ways: sol-gel, metalorganic chemical vapor deposition, and
sputtering [65]. The current response in Pb(ZrxTi1-x)O3 thin films includes a time dependent
current which is strongly dependent on the measuring technique, prehistory of the sample,
and some other factors [64, 65].
Figure 13. Schematic of current-voltage characteristics (I — Ohmic and II — sublinear region) for
Pb(ZrxTi1-x)O3 thin films obtained in [63].
The different samples can exhibit different types of conduction behavior depending on their
thickness, grain structure and structure of interfaces [65]. In addition, it is known that the
high-field conduction is strongly controlled by properties of the electrodes [56, 59]. The
shape of I − V curves as well as the effects of temperature cycling, electrical pretreatment,
and light pretreatment strongly suggest that electrode-controlled injection occurs in PZT thin
films, which is in turn influenced by the electric field of the depletion and entrapped space
charge [65]. Such features of a leakage current complicate the comprehension of the
conduction mechanism in this system.
Different physical models describing the leakage current behavior in PZT thin films were
proposed in [55-61]. However, as it was shown in [62, 65], none of these models is able to
provide an adequate explanation for the totality of the known experimental data.
The space-charge influenced-injection model describing the main features of the
observed current-voltage characteristics for lead-zirconate-titanate thin film was developed by
Stolichnov and Tagantsev. Their model is based on the following experimentally statements
[65]:
(i) The leakage current is determined by injection of carriers through the electrode- PZT
contacts.
(ii) The carriers injected are holes. According to the energy band structure of the Pt-PZT-
Pt films, the conditions for the hole injection are much more favorable than for the
electron injection because of the high work function of Pt [62].
(iii) Being in contact with Pt electrodes, the film contains a positive space charge due to
the positive difference in work function δΦ between Pt and PZT [62], which
corresponds to upward band bending (for the side parts with respect to the middle of
the films), i.e., the built-in charge in the film is of the same sign as the carriers
injected.
(iv) The built-in charge is assumed to be related to deep trapping centers and oxygen
vacancies, immobile during the measurement time, and homogeneous. According to
Simmons [66] for a positive δΦ , this charge can be formed due to a depletion of the
electrons from the states lying in an energy stripe δΦ below the Fermi level of the
18 19 −3
bulk material. For a reasonable impurity concentration of 10 − 10 cm ,
following Ref.63 and Ref.64, the thickness of the charged nearby electrode layer for
PZT-Pt contact can be estimated as a few tenths of a micron. Model calculations
taking into account oxygen vacancy migration give similar values [67]. From these
considerations, it is expected that the built-in space charge is indeed homogeneous in
300 nm films which were researched in [65].
Based on these suppositions Stolichnov and Tagantsev [65] consider the band profile of
the Pt-PZT-Pt system with a uniform positive space charge of depletion within the film
(Figure 14). The interfacial potential barrier for holes injection Φ h is mainly determined by
the difference of work functions of Pt and Pb(ZrxTi1-x)O3 [62]. However, the upward band
bending caused by the space charge results in formation of the additional potential barrier
ΔΦ [65]. If no external electric field is applied, the maximum of the potential barrier is
located in the middle of the film (Figure 14a).
Once an electric field is applied, the band profile of the system changes as shown in
Figure 14b. As the electric field increases, the potential barrier decreases and its maximum
moves towards the cathode [65]. This evolution of the band profile is accompanied by a
strong nonlinear increase in current.
At some critical field FC , the maximum of the potential barrier reaches the interface
(Figure 14c) and the further current increase is only possible due to a barrier lowering
according to the Schottky mechanism [65, 68]. Thus, at higher fields, the current will be
controlled by the potential barrier Φ h . This barrier is mainly determined by the difference of
work functions of materials, therefore the change of its height is negligible with increasing
applied electric field. As a result of constant height of the current-controlling barrier at
electric field F > FC , the sublinear I (U ) dependence appears.
According to [65], the current-temperature characteristic of the system is described by the

activation energy of about 0.9 eV in the temperature range 70-200◦C. Therefore, the change
of the slope of the current-temperature curve described in literature can be related to the
measuring procedure because of an incomplete saturation of current-time response to a
measuring voltage step [65]. The time dependence of the measured current response in
experiments where measuring times are greater than 100 s and fields higher than
30 − 40 kV cm is attributed to a time drift of parameters of the conductive system of the
films caused by the influence of the injected charge entrapment during measurement [65].
Figure 14. Schematic energy band profile of the PZT film [65]: (a) without external electric field, (b) at
electric field F < FC applied, and (c) at electric field F > FC applied.
One can conclude that the model proposed by Stolichnov and Tagantsev reproduces fairly
well the main qualitative features of current-voltage curves of Pt-PZT-Pt thin films and gives
us a reasonable description of the current limiting behavior obtained in this system.
ACKNOWLEDGMENTS
The author would like to thank Dr. A. Glot for his work as co-author in the papers which
were used in this chapter, Dr. V. Rybak and Dr. Y. Temko for reading some sections and
useful comments.
REFERENCES
[1] Myasnikov, I. A., Sukharev, V. Ya., Kupriyanov, L. Yu., & Zavjalov, S. A. ( 1991).
Semiconductor sensors in physico-chemical studies; Nauka: Moscow, pp 87-88.
[2] Beekmans, N. M. J. (1978). Chem. Soc. Faraday Trans., 74, 31-45.
[3] McAleer, J. F., Moseley, P. T., Norris, J .O. & Williams, D. E. J. (1987). Chem. Soc.
Far. Trans., 83, 1323 - 1331.
[4] Pike, G. E. & Seager, C. H. J. (1979). Appl. Phys., 50, 3414-3422.
[5] Mahan, G. D., Levinson, L. M. & Philipp, H. R. J. (1979). Appl. Phys., 50, 2799-2812.
[6] Vanadamme, L. K. J. & Brugman, J. C. J. (1980). Appl. Phys., 51, 4240-4244.
[7] Volkenshtein, F. F. (1987). Electron processes at semiconductor surface at
chemisorption; Nauka: Moscow, pp 118-123.
[8] Matsuoka, M. & Jap, J. (1971). Appl. Phys., 10, 736-746.
[9] Glot, A. B., Chakk, A. M., Chernyi, B. K. & Yakunin, A. Y. A. (1974). Inorganic
Mater., 10, 1866-1868.
[10] Einzinger, R. (1978). Appl. Surf. Sci., 1, 329-340.
[11] Clarke, D. R. J. (1978). Appl. Phys., 49, 2407-2411.
[12] Yan, J. F. & Rodes, W. W. (1982). Appl. Phys. Lett., 40, 536-537.
[13] Glot A. B., Hogarth C. A., Bulpett R. & Reynolds A. J. J. (1985). Mat. Sci. Lett., 4,
963-966.
[14] Makarov, V. O. & Trontelj, M. J. (1994). Mat. Sci., Lett., 13, 937-939.
[15] Bueno, P. R., De Cassia-Santos, M. R., Leite, E. R., Longo, E., Bisquert, J.,
Garcia-Belmonte, G. & Fabregat-Santiago, F. J. (2000). Appl. Phys., 88, 6545-6548.
[16] Glot, A. B. (2007). In: Ceramic Materials Research Trends; Ed.: Lin, P. B., Nova
Science Publishers, Inc., New York, p. 227-273.
[17] Vijayanandhini, K. & Kutty, T. R. N. (2006). Appl. Phys. Lett., 88, 123513-3.
[18] Kutty, T. R. N. & Philip, S. (1995). Materials Science and Engineering B, 33, 58-66.
[19] Glot, A. B., Sandoval-Garcia, A. P., Gaponov, A. V., Bulpett, R., Jones, B. J. &
Jimenez-Santana, G. (2008). Adv. in Tech. of Mat. and Mat. Proc. J., 10, 21-32.
[20] David, R. C. J. (1999). Am. Ceram. Soc., 82, 485–502.
[21] Mueller, R. K. J. (1961). Appl. Phys., 32, 635-639.
[22] Taylor, W. E., Odell, N. H., & Fan, H. Y. (1952). Phys. Rev. B, 88, 867-871.
[23] Gaidyalis, V. I., Markevich, N. N. & Montrimas, E. A. (1968). Fizicheskie protsessy v
elektrofotograficheskikh sloyakh ZnO (Physical processes in ZnO electrophotographic
layers); Vilnius: Mintis, 227-230.
[24] Goldman, E. I. & Zhdan, A. G. (1976). Sov. Phys. Semicond, 10, 1839-1845.
[25] Glot, A. B., Bondarchuk, A.N. & Mazurik, S.V. (2000). Photoelectronics, 9, 110-113.
[26] Glot, A; Behr, G. & Werner, J. (2001). In Proceedings of the 7th Conference of the
European Ceramic Society, Brugge, Belgium, Key Eng. Mater., 206–213, Trans. Tech.
Publications, Switzerland (2002). 1441-1444.
[27] Glot, A., Bondarchuk, A., Mazurik, S., Behr, G. & Werner, J. (2002). In Proceedings of
the 7th Conference of the European Ceramic Society, Brugge, Belgium, 2001; Key Eng.
Mater., 206–213, Trans. Tech. Publications, Switzerland, 1437- 1440.
[28] Bondarchuk, A. N. & Glot, A. B. (2004). Visnyk of Dniepropetrovsk Univ. Ser. Phys.
Radiophys, 12, 165-167.
[29] Bondarchuk, A., Glot, A., Behr, G. & Werner J. (2007). Eur. Phys. J: Appl. Phys., 39,
211-217.
[30] Constantinescu, M., Segal, E. & Vass, M. (1976). Rev. Roumaine Chim., 21, 503-508.
[31] Kunishima, Y. N., Miyayama, M., & Yanagida, H. (1996). Jpn J. Appl. Phys., 35,
3478-3482.
[32] Bondarchuk, A. N., & Glot, A. B. J. (2008). Phys. D: Appl. Phys., 41,175306 (3pp) doi:
10.1088/0022-3727/41/17/175306 (to be published).
[33] Gaponov, A. V; Glot, A. B., Ivon A. I. & Chack A. M., (2007). Jimenez-Santana, G.
Mater. Sci., Eng. B, 145, 76-84.
[34] Korotcenkov, G., Blinov, I., Brinzari, V. & Stetter, J. R. (2007). Sensors Actuators B,
122, 519-526.
[35] Nenov. T. & Yordanov, S. (1992). Sensors Actuators B, 8, 117-122.
[36] Kutty, T. R. N. & Ravi, V. (1991). Journal of Materials Science: Materials in
Electronics, 2, 79-88.
[37] Zhang, D., Zhou, D., Jiang, S., Wang, X. & Gong, S. (2002). Sensors and Actuators A,
101, 123-131.
[38] Harwood, M. G., Popper, P. & Rushman, D. F. (1947). Nature, 160, 58-59.
[39] Megaw, H. D. (1957). Ferroelectricity in crystals; Methuen & Co Ltd: London,
pp 15-20.
[40] Heywang, W. (1961). Solid-State Electronics, 3, 51-58.
[41] Heywang, W. J. (1964). Am. Ceram. Soc., 47, 484-490.
[42] Jonker, G. H. (1964). Solid-State Electronics, 7, 895-903.
[43] Kataoka, N., Hayashi, K., Yamamoto, T., Suguwara, Y., Ikuhara, Y. & Sakuma, T. J.
(1998). Am. Ceram. Soc., 81, 1961-1963.
[44] Lines, M. E. & Glass, A. M. (1977). Principles and applications of ferroelectics and
related materials; Clarendon press: Oxford, pp 101-105.
[45] Heywang, W. (1984). Amorphe und polykristalline Halbleiter; Springer-Verlag: Berlin,
Heidelberg, New York, Tokyo, pp 101-115.
[46] Vittayakorn, N. J. (2006). Appl. Sci. Res., 2, 1319-1322.
[47] Mallick, G. T., Emtage, Jr., & Emtage, P. R. J. (1968). Appl. Phys., 39, 3088- 3094.
[48] Al-Allak, H. M., Illingsworth, J., Brinkman, A. W. & Woods, J. (1989). J. Phys. D, 22,
1393-397.
[49] Zhang, F. & Cao. Z. J. (1996). Appl. Phys., 79, 2487-2490.
[50] Sinclair, D. C. & West, A. R. I. J. (1989). Appl. Phys., 66, 3850-3856.
[51] Neto, J. A. D., Pulcinelli, S. H. & Santilli, C.V. (1996). In Proceedings of the
International Conference on Electronic Ceramics and Applications. University of
Aveiro: Aveiro, Portugal, 409-412.
[52] Glot, A. B. & Bondarchuk, A. N. (1999). Inorg. Mater., 35, 532-534.
[53] Leite, E. R., Lee, E. J., Ribeiro, C. & Longo, E. J. (2006). Am. Ceram. Soc., 89,
2016-2020.
[54] Chen, J., Chen, H. & Lee, J. Y. (1996). Appl. Phys. Lett., 69, 4011-4015.
[55] Moazzami, R., Hu, C. & Shepherd, W. (1990). Proceeding International Reliability
Physics Symposium, 231-236.
[56] Sudhama, C., Campbell, A. C., Maniar, P. D., Jones, R. E., Moazzami, R., Mogab, C. J.
& Lee, J. C. J. (1994). Appl. Phys., 75, 1014-1019.
[57] Waser, R. (1997). Integr. Ferroelectr. 15, 39-43.
[58] Mihara T., Watanabe, H. & Jpn. J. (1995). Appl. Phys., 34, 5664-5673.
[59] Chen, X., Kingon, A. I., Al-Shareef, H. & Bellur, K. R. (1994). Ferroelectrics, 151,
133-138.
[60] Hu, H. & Krupanidhi, S. B. J. (1994). Mater. Res., 9, 1484-1498.
[61] Blom, P. W., Wolf, R. M., Cillessen, J. F. M. & Krijin, M. P. C. M. (1994). Phys. Rev.
Lett., 73, 2107-2110.
[62] Wouters, D. J., Willems, G. J., & Maes, H. E. (1995). Microelectron. Eng., 29,
249-256.
[63] Takasu, H. (1997). Integr. Ferroelectr, 14, 1-10.
[64] Tagantsev, A., Kholkin, A., Colla, E., Brooks, K. & Setter, N. (1995). Integr.
Ferroelectr, 10, 189-204.
[65] Stolichnov, I. & Tagantsev, A. J. (1998). Appl. Phys., 84, 3216-3225.
[66] Simmons, J. G. J. (1971). Phys. Chem. Solids, 32, 2581-2591.
[67] Brennan, C. J. (1995). Integr. Ferroelectr, 7, 93-109.
[68] Sze, S. M. (1969). Physics of Semiconductor Devices; Wiley: New York, pp 301-312.
Chapter 8
CHARACTERISATION OF SILICIDE THIN FILMS FOR

SEMICONDUCTOR AND NANOTECHNOLOGY
ELECTRONICS
Madhu Bhaskaran1*, Sharath Sriram1 and David R. G. Mitchell2,3

1
Microelectronics and Materials Technology Centre, School of Electrical and Computer
Engineering, RMIT University, GPO Box 2476, Melbourne, Victoria 3001, Australia.
2
Institute of Materials Engineering, Australian Nuclear Science and Technology
Organisation (ANSTO), PMB 1, Menai, New South Wales 2234, Australia.
3
Present address: Electron Microscopy Unit, University of Sydney, Sydney, New South
Wales 2006, Australia.
ABSTRACT
Nanotechnology devices require low resistance contacts, which can be fabricated by
the incorporation of silicide thin films This chapter discusses in detail the study of silicide
thin films using a suite of materials characterisation tools. The silicides of interest in this
study were titanium silicide (TiSi2) and nickel silicide (NiSi), given their low resistivity
and low barrier heights to both n-type and p-type silicon. The silicide thin films were
formed by vacuum annealing metal thin films on silicon substrates. Silicide thin films
formed from metal films deposited by DC magnetron sputtering and electron beam
evaporation were compared. The composition, crystallographic orientation, and
morphology of these thin films were studied using spectroscopy (AES, SIMS, RBS, in
situ Raman spectroscopy), diffraction (Bragg-Brentano and glancing angle XRD,
RHEED), and microscopy techniques (TEM, SEM, and AFM).
*
Corresponding author: madhu.bhaskaran@gmail.com, madhu.bhaskaran@rmit.edu.au
274 Madhu Bhaskaran, Sharath Sriram and David R. G. Mitchell
1. INTRODUCTION
Advancements in nanotechnology have created the need for efficient means of
communication of electrical signals to nanostructures [1]. Electrical contacts made to such
nanodevices need to pose minimum possible contact resistance. In order to study and evaluate
the resistance of such contacts or the resistance posed by the interface(s) in such contacts,
accurate test structures and evaluation techniques need to be used. These will pave the way
for the identification of new materials and/or contact architectures to develop nanoscale low
resistance contacts. These low resistance contacts incorporate silicide thin films. Titanium
silicide and nickel silicide thin films are very popular due to numerous advantages. These thin
films are also suitable for applications in the MEMS industry as mask materials for silicon
bulk-micromachining [2].
Titanium silicide (TiSi2) is widely used as a local interconnect material in complementary
metal-oxide semiconductor (CMOS) technology [3,4]. One of the advantages of titanium
silicide is that it can exist in a low resistivity phase, suitable for local interconnects in CMOS
technology. It is also suitable for CMOS contacts, with low barrier heights to both n-type and
p-type silicon [5]. Titanium silicide is also stable at high temperatures in the range of 800-
1000 ºC [4,6]. While substantial characterisation and application results for TiSi2 have been
previously reported (examples of such work include [3,4,7-10]), the TiSi2 thin films discussed
in this chapter were comprehensively analysed by a variety of spectroscopy, diffraction, and
microscopy techniques. Many of these results add further to those previously published (such
as those in [3,7-10]).
Nickel silicide (NiSi) is a material which finds application in both CMOS and
microsystems devices. NiSi thin films are used in ohmic contacts in the CMOS industry due
to their low resistivity [3,6] and comparable performance for both nMOS and pMOS devices
[11,12]. These advantages have made NiSi the CMOS industry standard from 2005. This
chapter discusses the synthesis of nickel silicide thin films and details materials
characterisation results obtained from spectroscopy, microscopy, and diffraction techniques.
2. CHARACTERISATION OF TITANIUM SILICIDE THIN FILMS

This section reports on the comprehensive analysis of C54 titanium silicide thin films by
a variety of spectroscopy, diffraction, and microscopy techniques. The formation of titanium
silicide thin films using sputtered and evaporated titanium has been compared. The desired
phase of titanium disilicide (TiSi2) is C54, as it exhibits low resistivity (14-17 μΩcm) and is
more stable. The undesirable, high resistivity (60-90 μΩcm) phase of the same stoichiometry
titanium silicide is termed C49. C49 TiSi2 is formed initially, which at higher temperatures
(above 450-500 ºC) starts to convert to C54. The conversion of C49 to C54 is usually
complete at temperatures of 650 ºC to 750 ºC (depending on the initial thickness of titanium
[3,13]). In this research, the focus has been on synthesising and studying C54 titanium
disilicide (TiSi2).
Characterisation of Silicide Thin Films for Semiconductor and Nanotechnology... 275
2.1. Silicide Thin Films from Sputtered and Evaporated Titanium
The formation of titanium silicide (TiSi2) by annealing titanium thin films (deposited
either by sputtering or evaporation) on silicon in vacuum is discussed in this section. The
advantages of vacuum annealing (over rapid thermal processing, generally reported [3])
include (i) complete reaction of the metal with silicon, and therefore, no necessity to remove
unreacted metal by etching; (ii) more uniform heating due to direct contact with the substrate
heater; and (iii) lesser impurities in silicide thin films formed in vacuum. Thin films of
titanium (100-110 nm) on silicon which were either sputter-deposited or electron beam
evaporated were placed on a substrate heater in a vacuum chamber.
Table 1. DC magnetron sputtering conditions for titanium.

Target Titanium (99.99%)
Target diameter 100 mm
DC power 100 W
Target to substrate distance 70 mm
Process gas Argon (99.999%)
Base pressure 1.0 x 10-5 Torr
Sputtering pressure 1.0 x 10-2 Torr
Sputtering duration 7 minutes
Titanium thin films were deposited on n-type (100) silicon of resistivity 1-10 Ωcm. The
silicon samples were cleaned, prior to metal deposition, by etching in buffered hydrofluoric
acid. Thin films of Ti were deposited either by DC magnetron sputtering (using conditions in
Table 1) or by electron beam evaporation, from a 99.99 % pure source in vacuum (2.0 x 10-7
Torr). The resulting metal thin films were about 100 nm thick, with a 6 % or 15 % variation
in thickness across a three-inch silicon wafer (for evaporation or sputtering, respectively).
Thin films of Ti (100-110 nm) on silicon, which were either sputter-deposited or electron
beam evaporated, were placed on a substrate heater in a vacuum chamber. The anneal process
was carried out at a temperature of 800 ºC for 60 minutes in vacuum (1.0 x 10−5 Torr). The
temperature was ramped up at 15 °C/min and ramped down at 10 °C/min.
An AES depth profile of a titanium silicide thin film formed by vacuum annealing
sputtered titanium is shown in Figure 1. There is complete reaction of titanium and silicon to
form a uniform TiSi2 thin film, as is apparent from the 1:2 titanium-silicon ratio. Though
there is slight oxygen contamination on the surface, the oxygen signal drops rapidly
indicating an oxygen-free silicide-to-silicon interface.
The average surface roughness of titanium silicide thin films formed from annealing
sputtered titanium was 26 nm, as obtained from AFM scans (an example of which is shown in
Figure 2).
Evaporated titanium thin films vacuum annealed at 800 ºC for 60 minutes also formed
TiSi2, with the composition of the silicide (Figure 3) being 33-34% titanium and 66-67%
silicon. When compared to the AES depth profile shown in Figure 2 (TiSi2 formed from
sputtered titanium), these films do have a considerable amount of oxygen contamination on
the surface. As with the titanium silicide formed from sputtered titanium, the oxygen signal
drops to a very low concentration giving an oxygen-free silicide-to-silicon interface.
Figure 1. AES depth profile of titanium silicide thin film (formed from sputtered titanium).
Figure 2. AFM surface scan of titanium silicide (formed from sputtered titanium) thin film over a 5 μm
x 5 μm area.
Figure 3. AES depth profile of titanium silicide thin film (formed from evaporated titanium). (Reprinted
with permission from Ref. [14].)
The average surface roughness of these thin films was also studied by AFM, and was
found to be lower (Ra = 20 nm). These results are discussed in sub-section 2.2.
Both sputtered and evaporated titanium formed TiSi2 under the same conditions, but
oxygen contamination was higher on the surface of the silicide film obtained by annealing
evaporated titanium. The oxygen contamination on the surface of TiSi2 thin films formed
from evaporated titanium thin films is due to the snow-ploughing of oxygen from native
oxide on the silicon substrate (discussed in detail in sub-section 2.2); this could be removed
by etching in concentrated hydrochloric acid (HCl) as it was a form of titanium oxide. The
physical bombardment by material being deposited while sputtering will remove the thin
native oxide on silicon; and is probably the reason for lower oxygen contamination on the
TiSi2 film surface. This oxide formed on the surface of TiSi2 thin films from sputtered
titanium was a Ti-Si-O compound, which could not be removed by HCl.
Both types of films had oxygen-free silicide-to-silicon interfaces and reasonably rough
surfaces, though the average surface roughness of films formed from evaporated titanium was
lower than those from sputtered titanium thin films.
Due to ease of removal of surface oxide and uniformity in thickness of titanium deposited
by electron beam evaporation, these films were chosen over TiSi2 thin films formed from
sputtered titanium. Further detailed analysis was carried out on TiSi2 thin films formed by
vacuum annealing evaporated titanium. The results of these analyses are discussed in the rest
of this section.
2.2. Composition and Surface Morphology Analysis
Titanium silicide (TiSi2) thin films were studied by spectroscopy techniques – Auger
electron spectroscopy (AES) and secondary ion mass spectrometry (SIMS) – to study the
composition, uniformity, and oxygen contamination through the films. Atomic force
microscopy (AFM) was used to study the surface morphology, in order to observe the degree
of regularity of grains (with respect to structure and grain size) and to estimate the average
surface roughness of the films. Experimental conditions are detailed in Ref. [14].
Both depth profiles (AES and SIMS, discussed later) indicated the presence of a thin
layer of titanium oxide on the silicide surface. This can be removed by etching in
concentrated hydrochloric acid (HCl). Hydrofluoric acid (HF) cannot be used to remove this
oxide as HF attacks titanium silicide.
Silicon wafers were placed face-down on samples during the anneal process, to minimise
oxygen interaction with the samples. The upper silicon wafers (prior to use during the anneal
process) were dipped in buffered hydrofluoric acid (BHF) to remove native oxide. These
silicon wafers showed evidence of titanium diffusion after the anneal process. This titanium
diffusion was verified be performing AES depth profiles at different spots. The silicon wafers
which covered patterned titanium, showed titanium diffusion in regions with identical
dimensions as the patterned titanium [see Figure 4(a)]. This proved to be a novel method of
selectively diffusing titanium into silicon. In addition, for a continuous titanium thin film,
when a weight exerting a contact pressure of about 620 N/m2 is placed on the top face-down
silicon, the upper and lower silicon samples fuse together, forming an unbreakable bond
[Figure 4(b)]. Silicon wafers coated with titanium nitride (TiN) to prevent titanium diffusion
were experimented with, and did reduce the extent to which titanium diffused into the upper
wafer. However, this did not reduce the amount of oxide on the surface of the TiSi2 formed.
(a) (b)
Figure 4. Schematic images depicting (a) titanium diffusion into the upper BHF-dipped silicon wafer
placed over a patterned titanium sample and (b) the unbreakable bond formed when the upper BHF-
dipped silicon wafer is placed over a continuous titanium thin film. (Not to scale.)
TiSi2 formation is not affected by native oxide present on silicon (which occurs at the
interface of titanium and silicon). Previous studies clearly show that the diffusing species (Si,
from substrate) starts the Ti-Si reaction through the native oxide layer, and the oxygen atoms
are slowly driven out towards the surface, resulting in an oxygen-free interface. The silicon
diffusion is accompanied by the disintegration of the oxide layer. This indicates that a layer of
oxide on the silicide surface will occur in most cases, as has been observed [10,15,16].
The AES depth profile indicates very uniform composition for titanium and silicon, and
shows an ideal 1:2 ratio of titanium to silicon (Figure 3). SIMS depth profiling was carried
out to study the uniformity in composition of titanium and silicon through the thin film. The
result (Figure 5) indicates a very homogenous thin film, and no observable oxygen
contamination at the interface.
Figure 5. SIMS depth profile of titanium silicide thin film. (Reprinted with permission from Ref. [14].)
Figure 6. AFM surface scan over a 5 μm × 5 μm area of titanium silicide thin film. (Reprinted with
permission from Ref. [14].)
AFM surface scans of the TiSi2 formed from evaporated titanium show a regular grain
structure with an average grain size between 100 and 125 nm (Figure 6), and the thin film
exhibits an average surface roughness of 20 nm. The surface also shows the presence of a
regular array of well-defined crystallites.
2.3. Cross-Section Transmission Electron Microscopy Analysis
Cross-section transmission electron microscopy (XTEM) studies of the thin films was
carried out to extract a variety of results, and to verify certain results obtained by other
analyses. The specimen preparation details are given in [17]. XTEM analysis of these samples
indicates a uniform thin film with crystallites on the surface at regular intervals (Figure 7).
These crystallites are 100-150 nm sized on large titanium silicide grains (flat plate-like grains
approximately 1.5 μm in diameter). The thickness of the titanium silicide thin film was
determined to be 244±10 nm.
Jump ratio maps (Figure 8) have shown that the crystallites have the same composition as
the thin film. The ratio maps for titanium, silicon, oxygen, and carbon were obtained for the
region corresponding to the bright field image shown in Figure 8(a). These maps show the
presence of titanium and silicon only in the thin film and the crystallites [see Figure 8(b) and
8(c)]. The titanium map [in Figure 8(b)] shows a high titanium signal at the outer surface of
the silicide, which is also rich in oxygen [Figure 8(d)]. This corresponds to the oxide pile-up
on the titanium silicide surface due to the ‘snowplough’ effect (as seen in AES and SIMS
depth profiles). The bright field image contrast [Figure 8(a)] and the silicon map [Figure 8(c)]
show that some re-deposition of silicon during ion milling has occurred. This has oxidized
and results in a high silicon and oxygen signal at the outer edge of the epoxy. Figure 8(e)
indicates the presence of carbon in the hydrocarbon-based epoxy used for TEM specimen
preparation. Energy dispersive X-ray analysis also confirmed that the crystallites had the
same composition as the titanium silicide thin film. These results show that the crystallites are
of the desired silicide composition, and are unaffected by the oxide pile-up.
TiSi2 grain ~1.5 μm in size
Figure 7. Surface crystallites on titanium silicide thin films formed by vacuum annealing observed in an
energy-filtered bright field XTEM image. The grain boundaries are apparent due to diffraction contrast.
(Reprinted with permission from Ref. [17].)
Figure 8. (a) Reference bright field XTEM image for jump ratio maps. (b-e) Jump ratio maps for
titanium, silicon, oxygen, and carbon, respectively. (Reprinted with permission from Ref. [17].)
Figure 9. High resolution cross-sectional transmission electron microscopy image showing the interface
between titanium silicide (left) and silicon (right). The dark (~7 nm wide) band in the TiSi2 at the
interface is a diffraction artefact. The image highlights the abrupt interface and the absence of
amorphous oxide. Inset: SAED patterns for two phases [-112] TiSi2 and [110] silicon. (Reprinted with
High-resolution XTEM images highlight the orthorhombic arrangement in the titanium

silicide thin film (Figure 9) and the presence of a well-defined interface between the
thermally formed silicide and the silicon substrate. No amorphous oxide layer was present at
the interface, supporting the AES and SIMS results.
2.4. Orientation Analysis
The titanium silicide thin films formed were also studied by diffraction techniques. XRD
analysis was carried out on thin film samples, and only the desired C54 peaks were detected
[as shown in Figure 10(a)] [18]. The peaks due to the silicide thin films (approximately 250
nm thick) were overshadowed by the peak due to the silicon substrate, under the Bragg-
Brentano XRD conditions. To overcome this, the films were analysed using glancing angle
XRD (GA-XRD, at an incidence angle of 5º), and more pronounced peaks at identical 2θ
positions were obtained [Figure 10(b)].
The results shown in Figure 10 have been shifted to 2θ positions corresponding to the
wavelength of copper kα1 radiation (λ = 0.154056 nm). This was done to enable comparison
with the standard diffraction files from the International Centre for Diffraction Data. The
background variations in Figure 10(b) are an artefact of the diffractometer (observed in all
glancing angle XRD analysis on that system), and are possibly due to reflections from the
sample holder.
(a)
(b)
Figure 10. Diffractograms obtained for titanium silicide from (a) Bragg-Brentano X-ray diffraction and
(b) glancing angle X-ray diffraction. (Reprinted with permission from Ref. [14].)
Figure 11. Reflection high-energy electron diffraction pattern obtained for titanium silicide with an
accelerating voltage of 100 kV. (Reprinted with permission from Ref. [14].)
Surface orientation analysis was performed using RHEED. Prior to analysis, the samples
were dipped in concentrated hydrochloric acid, to remove surface oxide. The resulting pattern
(as shown in Figure 11) indicated a polycrystalline surface orientation, as expected from
XRD. The rings obtained correspond to the XRD results, and to expected values of lattice
spacing for C54 titanium silicide. RHEED analyses the top few monolayers of the thin film,
while GA-XRD analyses a few hundred nanometres of the film (based on the films X-ray
absorption properties). A comparison of these results indicates that the orientations presented
in the upper monolayers of the film (from RHEED) correspond to the averaged orientation
results (obtained from GA-XRD) for the bulk of the thin film.
SAED patterns of the silicide obtained during the XTEM analysis were calibrated using
the silicon SAED patterns (zone axis [011]), both of which were obtained under identical
conditions. The silicide SAED corresponded to the C54 phase of titanium silicide as shown in
Figure 12 [19], and the obtained lattice spacings corresponded to those listed in the standard
used for XRD data [18].
Figure 12. Indexed selected-area electron diffraction pattern of a region of the titanium silicide thin film
with the zone axis of [-112]. (Reprinted with permission from Ref. [14].)
Figure 13. Energy filtered hollow cone dark field image confirming that the crystallites have the same
orientation as the underlying grain. Inset: SAED patterns along TiSi2 [011] for the thin film and a
crystallite. (Reprinted with permission from Ref. [17].)
Dark field imaging with hollow cone illumination was carried out to study the
crystallographic orientation of titanium silicide crystallites at a grain boundary in the titanium
silicide thin film. This indicates that the crystallites have the same orientation as the
underlying titanium silicide grain. This is shown in Figure 13. This was also confirmed by
SAED patterns (inset images in Figure 13).
3. CHARACTERISATION OF NICKEL SILICIDE THIN FILMS

This section presents results from extensive materials characterisation of nickel silicide
thin films by a variety of spectroscopy, diffraction, and microscopy techniques. Nickel
silicide thin films formed from sputtered and evaporated nickel have been compared. The
section also discusses Raman spectroscopy results which compares nickel silicide thin film
formation on (100) and (110) silicon. The focus has been on synthesising low resistivity
monosilicide (NiSi) thin films.
3.1. Silicide Thin Films from Sputtered and Evaporated Nickel
Nickel silicide growth involves the consumption of silicon, when it is formed by

thermally reacting nickel thin films deposited on silicon substrates. Thin films (50 nm) of
nickel on (100) n-type silicon substrates, with resistivity between 1 and 10 Ωcm, were
subjected to different vacuum annealing conditions in order to investigate the formation of
nickel silicide of the desired monosilicide composition (NiSi, 1:1 nickel and silicon).
The native oxide on silicon substrates was removed with buffered hydrofluoric acid, prior
to nickel (Ni) deposition. Ni thin films were either deposited by DC magnetron sputtering or
by electron beam evaporation. Sputtering was carried out under the conditions listed in Table
2 and evaporation of 50 nm of Ni was performed from 99.99% pure nickel sources, after
pumping down to a base pressure of 2 x 10-7 Torr.
Table 2. DC magnetron sputtering conditions for nickel.
Target Nickel (99.99%)

DC power 80 W
Process gas Argon (99.999%)
Sputtering duration 1 minute
Nickel thin films coated on (100) n-type silicon samples were subjected to a one-step or
two-step contact anneal process. The following temperatures were among those used to
anneal both evaporated and sputtered nickel films (50 nm) on silicon:
(i) 200 ºC for 1 hour, followed by 350 ºC for 3 hours

(ii) 250 ºC for 1 hour, followed by 350 ºC for 3 hours
(iii) 250 ºC for 1 hour, followed by 400 ºC for 3 hours
(iv) 350 ºC for 1 hour
(v) 350 ºC for 30 minutes
These temperatures were chosen because the formation of nickel silicide (NiSi) from
silicon takes place through a series of stoichiometric transformations [20]. Nickel thin films
on silicon react to form Ni2Si at about 250 ºC, NiSi at 350 ºC, and NiSi2 above 650-700 ºC.
The samples were placed on a substrate heater in a vacuum chamber. The anneal process was
started under vacuum of 1.0 x 10-5 Torr. The composition of thin films formed as a result of
reacting sputtered nickel and evaporated nickel on silicon was compared using AES depth
profiles.
Depth profiles show that the sputtered nickel films annealed under the above conditions
did not react completely with silicon, giving predominantly Ni2Si films (Figure 14). The AES
depth profile of sputtered nickel films vacuum annealed at 600 ºC for 90 minutes (Figure 15)
shows more complete reaction with silicon, giving NiSi with a composition of 48 % nickel to
52 % silicon. AES depth profiles of other sputtered films annealed under conditions (i) and
(iii) were found to be similar to Figure 14.
Evaporated nickel films reacted using any of the conditions (i)-(iii) formed NiSi films,
with the composition of the silicide (Figure 16) being 49-50% nickel and 51-50% silicon.
Evaporated nickel films were also annealed in vacuum at 350 ºC for different time periods of
30 minutes and 1 hour. Depth profiles of these films (formed using 50 nm evaporated nickel)
showed them to be NiSi (with nickel and silicon ratios of almost 50:50). Figure 16 depicts the
depth profile obtained for a NiSi film which was formed after 30 minutes of vacuum
annealing at 350 ºC.
Figure 14. AES depth profile of sputtered nickel thin film vacuum annealed at 250 ºC for 1 hour,
followed by 350 ºC for 3 hours.
Figure 15. AES depth profile for sputtered nickel thin film vacuum annealed at 600 ºC for 90 minutes.
Figure 16. AES depth profile for evaporated nickel thin film vacuum annealed at 350 ºC for 30 minutes.
Compared to evaporated nickel films, it was found that more thermal effort was required
to form NiSi using sputtered nickel films. Only sputtered nickel films which were annealed at
a high temperature of 600 ºC appeared to have completely reacted. This could be attributed to
variations in the diffusivity of metals in silicon, based on the metal thin film deposition
technique [21].
Further materials characterisation using a suite of tools were carried out for NiSi thin
films formed by vacuum annealing evaporated nickel thin films at 350 ºC for 30 minutes.
3.2. Composition and Surface Morphology Analysis
The composition of nickel silicide thin films has been analysed using AES depth profiles,
SIMS, and RBS. AFM has been used to study the surface morphology of the nickel silicide
thin films. Details on experimental conditions are given in [22].
The atomic percentage versus nickel silicide depth was obtained using a combination of
sputter etch and AES. The AES depth profile (Figure 16) shows a uniform NiSi film, with a
composition of 49-50 % nickel and 51-50 % silicon. No oxygen was observed, either on the
surface or at the silicide-silicon interface.
SIMS depth profiles were used to study the composition, uniformity, and potential
oxygen contamination of the nickel silicide thin films. The profile (Figure 17) indicates a
homogenous thin film with no trace of oxygen contamination on its surface. The presence of
an abrupt interface is indicated by the sharp drop in the nickel signal on approaching the
silicide-silicon interface (also verified using XTEM).
Figure 17. SIMS depth profile of a nickel silicide thin film. (Reprinted with permission from Ref. [22].)
RBS results indicate a nickel silicide layer of thickness 114 nm, with the simulated
spectrum in close agreement with obtained data (Figure 18). The simulated spectrum was
obtained by considering a multi-layered structure of nickel silicide thin films with nickel and
silicon composition varying between 48-52 % for both elements. The two significant layers
used in the simulation were a 40 nm layer of 51.0 % nickel and 49.0 % silicon and a 74 nm
layer of 48.5 % nickel and 51.5 % silicon. The 114 nm thick layer with a stoichiometry close
to NiSi agrees well with profilometry results.
Figure 18. Spectrum from Rutherford back-scattering for a nickel silicide thin film. (Reprinted with
Figure 19. AFM 1 μm × 1 μm scan in tapping mode of a nickel silicide thin film. (Reprinted with
AFM has been used to study the surface morphology of the nickel silicide thin films. The
average grain size and average surface roughness of these films was found to be 35 nm and
0.67 nm, respectively (Figure 19). The films were found to be very smooth with the
roughness values in agreement with those reported in literature [23]. The film surface was
studied using a FEG-SEM, but the very smooth film surface (which is as smooth as silicon)
prevented good resolution and contrast in the images of the film surface.
3.3. Transmission Electron Microscopy Analysis
This section discusses the transmission electron microscopy (TEM)-based investigation

of nickel silicide thin films: energy filtered imaging to study the thin film cross-sections,
jump ratio maps to confirm the composition uniformity of the thin films, high resolution
imaging (HRTEM) to characterise the silicide-silicon interface, and selected-area electron
diffraction (SAED) to determine the orientation of the thin films.
Cross-sectional transmission electron microscopy analysis was carried out along the
length of the films, to study the uniformity in thickness and grain structure of the nickel
silicide thin films. The film thickness was estimated to be about 110 nm (agreeing with the
profilometry results) and remained reasonably uniform with an equiaxed grain structure
evident along the length of the film. It was significant that a repetitive and uniform crystalline
structure, with well defined equiaxed grains about 50-60 nm in size, had resulted from the
reaction of the nickel film deposited on silicon.
The elastic (zero loss) image in Figure 20(a) shows that the grains had an equiaxed grain
structure, with grain diameters of about 60nm. Moiré fringes were present at oblique grain
boundaries in the film [as indicated in Figure 20(b)]. The interface between the NiSi and Si
was not flat but showed some evidence of weak faceting beneath individual NiSi grains. This
extended in the plane perpendicular to the image in Figure 20(a, b), giving rise to an oblique
interface. This resulted in the interface appearing slightly blurred. This effect was more
pronounced in the thicker regions of the TEM foil, as there was a greater projected thickness
and consequently a greater pile-up of features in projection. In the very thinnest regions of the
foil (those used for HRTEM – discussed later), the interface appeared much sharper. Plan
view images highlighted the existence of small (~60 nm) crystalline grains [Figure 20(c)],
comparable to the equiaxed grains seen in cross-section.
The compositional uniformity was studied using jump ratio maps. The intensity in these
images is a function of the areal concentration of the element of interest. The ratio maps for
nickel and silicon were obtained for the region corresponding to the bright field image shown
in Figure 20(a). These maps (Figure 21) show no significant variation in the Ni and Si
compositions. Some modulation due to diffraction contrast effects was present, but these were
readily identified as such by their sensitivity to specimen tilt. No unreacted Ni was found. The
bright line on the outer surface is an EFTEM artefact induced by specimen drift during the
long acquisition and ratio map processing, since no equivalent features were present in the
bright field image [Figure 20(a)].
High resolution transmission electron microscopy (HRTEM) imaging indicates the
presence of a very abrupt nickel silicide to silicon interface (Figure 22). The silicon lattice
appears to abruptly transform into the nickel silicide lattice in the span of one atomic layer.
On studying this interface at different points along the nickel silicide film, the faceting
introduced by the silicide grains appears to have introduced a curvature to the silicide-silicon
interface.
Figure 20. Results from TEM analysis of NiSi thin films: (a) XTEM highlighting equiaxed grains in the
NiSi thin film in which Moiré (interference) fringes due to orientation differences between grains can
be observed; (b) Notable features in the as-obtained image (a) are indicated; (c) Plan view, elastic
hollow cone dark field image of the film, highlighting individual grains with dimensions of 60-200 nm;
and (d) Plan view TEM image showing polygonal NiSi grains. (Reprinted with permission from Ref.
[24].)
Figure 21. Jump ratio maps corresponding to content of (a) nickel and (b) silicon. (Reprinted with
Figure 22.HRTEM of the nickel silicide to silicon interface depicting the abrupt silicide-silicon
interface. (Reprinted with permission from Ref. [24].)
3.4. Orientation Analysis
The orientation of the thin films was studied using a combination of Kikuchi patterns (or
EBSPs) obtained by electron back scatter diffraction (EBSD) and glancing angle X-ray
diffraction (GA-XRD). The Kikuchi patterns, also called electron backscatter patterns
(EBSPs), have been used to verify the existence of preferential orientations across the thin
film surface. (TiSi2 thin films had significant surface roughness, making them unsuitable for
this analysis.)
Table 3. Nickel silicide unit cell parameters for EBSD.
Composition Silicon 50%, Nickel 50%

Space group 62, Pnma
Laue group 3, mmm
Unit cell lengths 5.18 Å, 3.34 Å, 5.62 Å
Unit cell angles 90º, 90º, 90º
Unit cell volume 97.23 Å3
The nickel silicide crystal phase used for the indexing the EBSPs is shown in Table 3.
This crystal phase was obtained from [25] and also corresponds to the International
Committee for Diffraction Data (ICDD) nickel silicide powder diffraction file 85-0901 [26].
The multi-part figure (Figure 23) shows a sample set of the simulations generated and results
obtained. A three-dimensional simulation of the EBSPs is shown in Figure 23(a) overlaid on a
2 × 2 × 2 nickel silicide unit cell array. An example of an experimentally obtained EBSP is
shown in Figure 23(b). The ultra-fine grain size of the nickel silicide thin films prevents better
contrast in these patterns and also limits the number of locations on the film surface which
produce usable patterns. Due to this, indexing of selected patterns was performed manually to
compare the results with the simulated pattern. Based on simulations, recorded patterns were
indexed, as shown by the combination of Figs. 23(b, c). An additional example of a recorded
EBSP is shown in Figure 23(d).
EBSD analysis was used to investigate the uniformity in crystallographic orientation
across the thin film. The orientations resulting from the indexing of EBSPs compared very
well with the results obtained from GA-XRD, and these indicate the existence of preferential
orientations. There is no available literature showing that EBSD has been used to analyse
nickel silicide thin films, and the results obtained indicate that further work is required in
refining the EBSP acquisition parameters to account for the ultra-fine grain size of the nickel
silicide thin films before detailed orientation maps can be generated.
Figure 23. (a) Simulated three-dimensional EBSD pattern corresponding to the crystal phase defined in
Table 1. Inset is a 2 × 2 × 2 array of the nickel silicide unit cell. (b, c) Example of a recorded EBSP and
the simulation used for indexing this EBSP. (d) Example of an EBSP showing a different orientation to
the EBSP in (b). The low contrast in the EBSPs is due to the ultra-fine grain structure of the nickel
silicide thin films. (Reprinted with permission from Ref. [22].)
Figure 24. Glancing angle X-ray diffraction results obtained for a nickel silicide thin film sample
indicate preferential orientation. (Reprinted with permission from Ref. [24].)
GA-XRD (Figure 24) of the nickel silicide thin films indicates the films to be
polycrystalline with preferential orientation. The peaks obtained in the diffractograms were
compared with the International Committee for Diffraction Data powder diffraction file
(ICDD PDF) 85-0901 (Cu kα) for indexing [26]. These results also indicate that the films are
of the NiSi (monosilicide) phase, and not undesirable Ni2Si or NiSi2 phases.
3.5. In Situ XRD and Raman Spectroscopy Analysis
The in situ techniques employed for this study include micro-Raman spectroscopy and X-
ray diffraction (XRD); in both cases the variations for temperatures up to 350 ºC has been
studied. The evolution of NiSi from Ni2Si is discussed and the influence of substrate
orientation is investigated. These techniques have also been used to study variations in nickel
silicide formation between n-type and p-type silicon. Experimental details can be found in
[27].
3.5.1. Raman analysis of reference NiSi sample
A Raman spectrum obtained for the reference nickel silicide thin film (formed by vacuum
annealing at 350 ºC) is shown in Figure 25. The NiSi thin film used as a reference was
extensively characterised to verify it was of the required low resistivity monosilicide phase,
and so, the corresponding Raman spectrum was used to identify transformation of Ni to NiSi
in the nickel films under in situ study.
NiSi belongs to the MnP-type orthorhombic structure (space group Pnma, D2h16);
therefore, the film is Raman active. In accordance with group theory (see, for example [28]
and [29]), 12 Raman active optical phonons exist for NiSi. Each of these 12 optical phonons
can be observed in the Raman spectra of NiSi single crystal collected in a specific
measurement geometry. Raman spectra of NiSi films, NiSi powder, and polycrystalline NiSi
will contain only some of these modes. Depending on the degree of texture, some Raman
active modes may be more pronounced than others in thin film spectra. Donthu et al. [29]
identifies eight phonon peaks for nickel silicide powder at 197, 214, 255, 288, 314, 332, 360,
and 397 cm-1. Peaks at 197, 216, 256, 289, 314, 332, and 362 cm-1 were observed in the
Raman spectra for the nickel silicide thin film shown in Figure 25. According to factor group
analysis, the peaks 197, 216, 332, and 362 cm-1 can be assigned to the Ag symmetry, while
peaks at 256 and 314 cm-1 may belong to either B2g or B3g symmetry [29]. The five other
phonon peaks, expected from group theory and, in particular, the peak at 397 cm-1 [29], were
not observed in this work.
Figure 25. Micro-Raman spectrum of nickel silicide film formed at 350 ºC by vacuum annealing
(reference sample). (Reprinted with permission from Ref. [27].)
3.5.2. In situ measurement results for nickel on n-type (100) silicon
Raman spectra obtained while heating nickel thin films on n-type (100) silicon are shown
in Figure 26(a). The absence of peaks for the as-deposited Ni film (see spectrum at 24 ºC, for
example) is typical for metallic films that have no optic lattice vibrations. Ni2Si peaks are
expected at 100 and 140 cm-1, but due to the presence of strong background in this region of
the spectra and relatively large noise, the appearance and disappearance of the Ni2Si phase
cannot be conclusively determined. However, these two peaks are typically accompanied by a
small peak at ~217 cm-1 and a very weak peak at ~ 190 cm-1 [30,31] related to the existence
of the NiSi phase. Therefore, in this work, the presence of the Ni2Si phase has been
determined based on the appearance of a small single peak at ~215 cm-1, while confirmation
on the formation of NiSi phase was made from the appearance of a double peak at ~196 and
215 cm-1. Based on this it seems that the Ni2Si and NiSi phases start to form at 250 ºC and
290 ºC, respectively.
The diffractograms in Figure 26(b) show the XRD peaks obtained while the sample was
heated to 350 ºC. Nickel and silicide peaks are concentrated in the 2θ range of 40-50º; the
peaks at 44.54º and 45.88º are those of nickel (from the sample) and platinum (from the
heating stage), respectively [32,33]. Due to thermal expansion, peaks shift left; the shifts are
more prominent in nickel than in platinum. This could be due to platinum and nickel being
located at different heights with respect to the detector and the different thermal expansion
coefficients of the two metals. When the sample reaches a temperature of 250 ºC, there is a
drop in intensity of the nickel peak and a small peak appears at 47.09º which corresponds to
Ni2Si [34]. At 325 ºC, the nickel silicide (NiSi) peak at 45.21º becomes prominent. The
diffractogram labelled ‘350 ºC (30 minute dwell)’ in Figure 26(b) corresponds to the XRD
data collected after the sample was held at 350 ºC for 30 minutes and this indicates a nickel
silicide thin film with a preferential (211) orientation.
(a)
(b)
Figure 26. (a) Micro-Raman spectra and (b) X-ray diffractograms registered at different temperatures
for nickel thin film deposited on n-type (100) silicon substrate. (Reprinted with permission from Ref.
[27].)
(a)
(b)
Figure 27. (a) Micro-Raman spectra and (b) X-ray diffractograms registered at different temperatures
for nickel thin film deposited on p-type (100) silicon substrate. (Reprinted with permission from
Ref. [27].)
3.5.3. In situ measurement results for nickel on p-type (100) silicon
Raman measurements were carried out for these samples under conditions similar to
those for n-type (100) silicon substrates. The spectra obtained at various temperatures are
shown in Figure 27(a). The weak single peak at ~215 cm-1 (Ni2Si) again appears at 250 ºC
and double peaks at 215.6 and 196.5 cm-1 (NiSi) start to appear at 290 ºC. The diffractograms
in Figure 27(b) show the XRD peaks obtained while the sample was heated to 350 ºC. The
peaks at 44.54º and 45.88º are those of nickel (from the sample) and platinum (from the
heating stage), respectively [32,33]. The drop in intensity of the nickel peak occurs at 325 ºC.
After dwelling at 350 ºC for 30 minutes, the nickel silicide (NiSi) peak at 45.21º
corresponding to (211) orientation becomes prominent. There is a small peak at 46.75º which
corresponds to nickel silicide (121), and appears only during the initial 350 ºC measurements.
The peak for Ni2Si at 47.09º is not prominent during this measurement cycle.
3.5.4. In situ measurement results for nickel on p-type (110) silicon
Raman spectra obtained while heating nickel thin films on p-type (110) silicon are shown
in Figure 28. The formation of Ni2Si and the transformation from Ni2Si to NiSi seem to occur
at slightly higher temperatures on these samples when compared to the nickel deposited on
(100) silicon substrates. The spectrum at 350 ºC indicates a polycrystalline film with strong
NiSi peaks at 215.6 and 196.5 cm-1 and weak NiSi peaks at 255, 289, 310, 332 and 362 cm-1.
These results were conclusive, overcoming necessity of performing in situ XRD
measurements.
Figure 28. Micro-Raman spectra registered at different temperatures for nickel thin film deposited on p-
type (110) silicon substrate. (Reprinted with permission from Ref. [27].)
4. CONCLUSION
This chapter presents results from a comprehensive characterization of titanium silicide
(TiSi2) and nickel silicide (NiSi) thin films formed from thin films of metal deposited on
silicon. TiSi2 thin films formed from evaporated and sputtered titanium were compared, and
thin films formed from evaporated titanium were subject to further analyses. The titanium
silicide films formed by annealing electron beam evaporated titanium films on silicon have a
thin layer of oxide on the surface, but an oxygen-free TiSi2–Si interface. The stoichiometry,
uniformity in composition with depth and the surface morphology of these thin films were
studied using AES and SIMS depth profiles and AFM surface scans. All these results point to
a chemically uniform film of the desired composition and the presence of crystallites on the
surface at regular intervals. These crystallites were shown to have the same composition and
orientation as the underlying thin film. XTEM analysis of the thin films was used to verify the
presence of a layer of oxide on the surface, and high resolution XTEM images have shown a
well-defined silicide–silicon interface. SAED employed during the XTEM analysis confirmed
the presence of C54 titanium silicide. The XRD results showed the existence of a
polycrystalline film, with all orientations present corresponding to the desired C54 phase of
TiSi2. Glancing angle XRD and RHEED were also used to analyse the thin films, in order to
increase the intensity of reflections from the thin film (and decrease substrate effects).
NiSi films formed by vacuum annealing of sputtered and evaporated nickel films have
been compared using AES depth profiles. Nickel silicide of NiSi composition was formed
using evaporated nickel and several annealing temperatures and time periods. Higher thermal
energy was required to form NiSi using sputtered nickel films, which did not react uniformly
with silicon at lower temperature conditions.
Nickel silicide thin films formed by vacuum annealing of nickel thin films on silicon
substrates have been studied by spectroscopy, microscopy, and diffraction techniques.
Spectroscopy results from AES, SIMS, and RBS indicate (i) the presence of the desired 1:1
nickel and silicon stoichiometry, (ii) uniform composition with depth, and (iii) lack of oxygen
contamination either on the film surface or at the silicide-silicon interface. AFM scan results
and TEM analyses have shown that the films have an extremely smooth surface (as smooth as
silicon with an average surface roughness of 0.67 nm) comprising of ultra-fine grains
approximately 30-50 nm in diameter. HRTEM showed that the nickel silicide to silicon
interface was atomically abrupt. GA-XRD confirmed that the phase formed was NiSi; with
the thin films preferentially oriented with (211) NiSi planes parallel to the (100) planes of the
silicon substrate. EBSD analysis has shown that the films are preferentially oriented,
confirming results from XRD analysis.
The evolution of nickel silicide (NiSi) from the nickel film was observed using both
Raman and XRD in situ studies. Raman studies of nickel silicide formation on both n-type
and p-type (100) silicon substrates suggest that the transformation temperatures for both are
similar (~ 290 ºC); though XRD studies indicate the transformation temperature to be ~ 325
ºC for p-type (100) silicon. This discrepancy could be due to the fact that the laser used for
Raman measurements analyses a much smaller area (few μm2) compared to XRD (few mm2).
Raman spectra also indicate that the transformation from Ni2Si to NiSi occurs approximately
at 300 ºC for (110) silicon as opposed to 290 ºC for (100) silicon.
REFERENCES
[1] Roukes, M. L. (2001). Sci. Amer., 285, 48-57.
[2] Bhaskaran, M., Sriram, S. & Sim, L. W. (2008). J. Micromech. Microeng, 18, 095002.
[3] Zhang, S. & Ostling, M. (2003). Crit Rev Solid State Mater Sci., 28, 1-129.
[4] Murarka, S. P. (1995). Intermetallics, 3, 173-186.
[5] Plummer, J. D., Deal, M. D. & Griffin, P. B. (2000). Silicon VLSI Technology:
Fundamentals, Practice and Modeling; Prentice-Hall: NJ, p 700.
[6] Chen, J., Colinge, J. P., Flandre, D., Gillon, R., Raskin, J. P. & Vanhoenacker, D.
( 1997). J. Electrochem. Soc., 144, 2437-2441.
[7] Da Silva, A. N. R., Furlan, R. & Santiago-Aviles, J. J. (1995). In Silicide Thin Films –
Fabrication, Properties, and Applications; Tung, R. T., Maex, K., Pellegrini, P. W.,
Allen, L. H., Eds.; Materials Research Society Symposium Proceedings; Materials
Research Society: Pittsburgh, PA, Vol. 402, pp 119-124.
[8] Maury, D., Regolini, J. L. & Gayet, P. (1995). In Silicide Thin Films – Fabrication,
Properties, and Applications; Tung, R. T., Maex, K., Pellegrini, P. W., Allen, L. H.,
Eds.; Materials Research Society Symposium Proceedings; Materials Research Society:
Pittsburgh, PA, Vol. 402, pp 283-294.
[9] Kim, K. H., Lee, J. J., Seo, D. J., Choi, C. K., Hong, S. R., Koh, J. D., Kim, S. C., Lee,
J. Y. & Nicolet, M. A. (1992). J. Appl. Phys., 71, 3812-3815.
[10] Bender, H., Chen, W. D., Portillo, J., van den Hove, L. & Van der Vorst, W. (1989).
Appl. Surf. Sci., 38, 37-47.
[11] Morimoto, T., Ohguro, T., Momose, H. S., Iinuma, T., Kunishima, I., Suguro, K.,
Katakabe, I., Nakajima, H., Tsuchiaki, M., Ono, M., Katsumata, Y. & Iwai, H. ( 1995).
IEEE Trans. Electron Devices, 42, 915-922.
[12] Iwai, H., Ohguro, T. & Ohmi, S. I. (2002). Microelectron. Eng., 60, 157-169.
[13] Ezoe, K., Kuriyama, H., Yamamoto, T., Ohara, S. & Matsumoto, S. (1998). Appl. Surf.
Sci., 130-132, 13-17.
[14] Bhaskaran, M., Sriram, S., Short, K. T., Mitchell, D. R. G., Holland, A. S. &
Reeves, G. K. (2007). J. Phys. D: Appl. Phys., 40, 5213-5219.
[15] Dexin, C. X., Harrison, H. B. & Reeves, G. K. (1988). J. Appl. Phys., 63, 2171-2173.
[16] Wee, A. T. S., Huan, A. C. H., Osipowicz, T., Lee, K. K., Thian, W. H., Tan, K. L., &
Hogan, R. (1996). Thin Solid Films, 283, 130-134.
[17] Bhaskaran, M., Sriram, S., Mitchell, D. R. G., & Holland, A. S. (2008). Semicond. Sci.
Technol., 23, 035021.
[18] Powder Diffraction Pattern Files, International Centre for Diffraction Data (ICDD,
formerly the Joint Committee for Powder Diffraction Studies), Newtown Square, PA
19073, Card 35-0785.
[19] Andrews, K. W., Dyson, D. J. & Keown, S. R. (1971). Interpretation of Electron
Diffraction Patterns; Ed. 2; Plenum: NY, p 152.
[20] Clevenger, L. A. & Mann, R. W. (1995). In Properties of Metal Silicides; Maex, K.,
van Rossum, M., Eds., INSPEC: London, p 64.
[21] Thomä, A. (1990). J. Phys.: Condens. Matter, 2, 3167-3175.
[22] Bhaskaran, M., Sriram, S., Holland, A. S. & Evans, P. J. (2008). Micron, 40, 99-103.
[23] Tsuchiya, Y., Tobioka, A., Nakatsuka, O., Ikeda, H., Sakai, A., Zaima, S. & Yasuda, Y.
(2002). Jpn. J. Appl. Phys., 41, 2450-2454.
[24] Bhaskaran, M., Sriram, S., Mitchell, D. R. G., Short, K. T., Holland, A. S. & Mitchell,
A. (2008). Micron, 40, 11-14.
[25] Villars, P. & Calvert, L. D. (1991). Pearson's Handbook of Crystallographic Data for
Intermetallic Phases; Ed. 2; ASM: Ohio, p 2887.
19073, Card 85-0901.
[27] Bhaskaran, M., Sriram, S., Perova, T. S., Ermakov, V., Thorogood, G. J., Short, K. T.,
& Holland, A. S. (2008). Micron, 40, 89-93.
[28] Hayes, W. & Loudon, R. (1978). Scattering of light by Crystals; Wiley: NY.
[29] Donthu, S. K., Chi, D. Z., Tripathy, S., Wong, A. S. W. & Chua, S. J. (2004).
Appl. Phys. A: Mater. Sci. Process., 79, 637-642.
[30] Lee, P. S., Mangelinck, D., Pey, K. L., Shen, Z. X., Ding, J. & Osipowicz, T. (2000).
See, A. Electrochem. Solid State Lett., 3, 153.
[31] Nemanich, R. J., Tsai, C. C., Stafford, B. L., Abelson, J. R. & Sigmon, T. W. (1984). In
Thin Films and Interfaces II; Baglin, J. E. E., Campbell, D. R., Chu, W. K., Eds.;
Materials Research Society Symposium Proceedings; Elsevier Science Publishing: New
York, Vol. 25, p 9.
19073, Card 87-0712.
19073, Card 04-0802.
19073, Card 80-2283.
Chapter 9
MAGNETICALLY MODIFIED BIOLOGICAL

MATERIALS AS PERSPECTIVE ADSORBENTS FOR
LARGE-SCALE MAGNETIC SEPARATION PROCESSES
Ewa Mosiniewicz-Szablewska1, Mirka Safarikova2,

and Ivo Safarik2,3
1
Institute of Physics, Polish Academy of Sciences, Al.
Lotników 32/46, 02-668 Warsaw, Poland,
2
Department of Biomagnetic Techniques, Institute of Systems Biology and Ecology,
Academy of Sciences, Na Sadkach 7, 370 05 Ceske Budejovice, Czech Republic
3
Department of Medical Biology, Faculty of Science, University of South Bohemia,
Branisovska 31, 370 05 Ceske Budejovice, Czech Republic
ABSTRACT
Novel magnetically modified biological materials, containing magnetic iron oxides
nanoparticles as labels, have been successfully developed and applied as magnetic
affinity adsorbents for the magnetic separation of various biologically active compounds
and xenobiotics.
The main attention was focused on cheap and easy to get magnetic adsorbents which
could be applied for large-scale processes. Among them magnetically modified plant-
based materials (sawdust) and microbial cells (yeast and algae) were taken into
consideration. An inexpensive, extremely simple procedure was proposed for the
preparation of such magnetic adsorbents using standard water-based ferrofluids
containing maghemite nanoparticles with the diameter of about 12 nm. Such ferrofluids
can be prepared in a simple way (almost in any lab) and such nanoparticles can be used to
prepare biocomposite materials enabling their simple magnetic separation with standard
permanent magnets. Both of these properties are important for possible large-scale
applications.
The structural, adsorption and magnetic properties of the developed materials were
studied in detail by means of scanning electron microscopy, transmission electron
microscopy, spectrophotometric measurements, ESR spectroscopy and conventional
magnetic methods (DC magnetization and AC susceptibility measurements). The
302 Ewa Mosiniewicz-Szablewska, Mirka Safarikova and Ivo Safarik
prepared materials efficiently adsorbed selected biologically active compounds and

xenobiotics (mainly different enzymes, water-soluble organic dyes and heavy metal ions).
Their magnetic behavior was dominated by the superparamagnetic relaxation of isolated
single domain maghemite nanoparticles, although a little amount of agglomerates was
also present. However, these agglomerates were sufficiently small to show at static
conditions the superparamagnetic behavior at room temperature which allows to use the
developed materials as magnetic adsorbents in the magnetic separation techniques.
Moreover, the prepared materials exhibit the peculiar features enabling their rapid and
efficient removal not only from solutions, but also from suspensions. Such materials
could be efficiently used to isolate rare biologically active compounds from difficult-to-
handle materials including raw extracts, blood and other body fluids, cultivation media,
environmental samples, etc.
Inexpensive raw materials, extremely simple preparation method, affinity to various
biologically active compounds and both organic and inorganic xenobiotics, and
distinctive magnetic properties make the developed materials greatly suitable as magnetic
adsorbents for large-scale magnetic separation processes.
1. INTRODUCTION
Magnetically modified biocompatible materials, containing magnetic nanoparticles as
labels, have attracted much attention because of their great potential as magnetic affinity
adsorbents for various biologically active compounds. They have been successfully applied
for the magnetic separation of various proteins (enzymes, antibodies, antigens, receptors,
lectins), nucleic acids (DNA, RNA, oligonucleotides), low-molecular weight biologically
active compounds (drugs) and xenobiotics (carcinogens, water soluble dyes, heavy metal
ions, radionuclides) [1,2].
Magnetic separation techniques have many interesting applications in different areas of
biosciences ranging from cells separation [3] to removal of xenobiotics from aqueous wastes
[4]. They are alternative methods to gravitational, centrifugal or filtration separation
techniques and enable a simple magnetic manipulation with the adsorbents using an external
magnetic field.
Magnetic affinity adsorbents can be efficiently used for work in difficult-to-handle
materials including raw extracts, blood and other body fluids, cultivation media,
environmental samples, etc. However, for large-scale applications (e.g., in biotechnology or
environmental technology), the finding of relatively cheap and readily available magnetic
adsorbents is necessary. That is the reason why an extremely simple procedure was here
proposed for the preparation of such magnetic adsorbents using standard water-based
ferrofluids (which can be prepared in any lab) and inexpensive raw materials (sawdust and
microbial cells).
The sawdust and microbial cells have been chosen for magnetic modification because of
their well known affinity for various xenobiotics (mainly water-soluble organic dyes and
heavy metal ions [5-8]) which allows to use these materials for the large-scale removal of
xenobiotics from polluted water sources.
A large amount of environmental contaminants (among them dyes) is produced every
year in different branches of industry. A substantial part of them pollutes many water sources.
Even a very small amount of dye in the water (10 – 50 mg/dm3) affects the water
transparency and aesthetic values [9, 10]. There are more than 100.000 commercially
Magnetically Modified Biological Materials as Perspective Adsorbents... 303
available dyes and over 7x105 t of them are annually used in textile, paper, leather, plastics,
food and cosmetics industries [11]. It is estimated that 20-50 % of these dyes are lost into
wastewaters, causing environmental contaminations.
Dyes may be toxic and mutagenic. They contaminate not only the environment but also
traverse through the entire food chain, leading to the biomagnification. To decrease the dyes
concentration, different procedures such as coagulation, flocculation, chemical degradation,
oxidation, photodegradation, aerobic and anaerobic biodegradation, etc., are used [10].
However, such methods are often very expensive and cannot be used on a large scale.
Moreover, many synthetic dyes are difficult to remove by the conventional wastewater
systems, due to their complex chemical structure [10]. Therefore, the adsorption on
appropriate adsorbents seems to be an efficient procedure for their removal.
There are many adsorbents available, but the main attention is focused on cheap and easy
to get materials which could be applied for large-scale processes. Among them, living or dead
microorganisms (yeast, bacteria, fungi, algae) are intensively studied (see [8] and references
therein). Each microorganism is able to bind or degrade several types of dyes and on the other
side each dye can have an affinity to various microorganisms. In addition, the microbial cells
efficiently interact with magnetic nanoparticles present both in low-pH and high-pH ionic
magnetic fluids, leading to the formation of magnetically labeled cells, which could be easily
separated from the system using an appropriate magnetic separator [8].
Another cheap and easy available adsorbents which could be applied for large-scale
processes are plant-based materials (e.g. different types of sawdust). It has been known for a
long time that these materials have an affinity for different biologically active compounds
such as enzymes (e.g. trypsin [12], urokinase [13], elastase [14], cellulases [15]) or
polyphenols [16]. Moreover, many organic compounds such as acid and basic dyes and oils
[5, 6] and heavy metal ions [7] have been efficiently adsorbed on these materials. Therefore
the sawdust seems to be also promising adsorbent for the removal of dyes.
Currently, extensive studies are performed in many laboratories to find the optimal
magnetic adsorbents with best magnetic and adsorption properties enabling to separate
contaminating xenobiotics from large volumes of polluted water. These materials should be
superparamagnetic to that they would exhibit magnetic properties when placed within a
magnetic field (thus enabling their simple magnetic separation from the treated systems), but
retained no residual magnetism when removed from the magnetic field. They should form
stable colloidal suspensions and they should not aggregate in magnetic fields. They should
also have an affinity to adequate xenobiotics.
Inspired by these experiments, we directed our attention to the possibility of application
of magnetically modified sawdust and microbial cells as new inexpensive and readily
available magnetic adsorbents, which could be used for the separation and purification of
biologically active compounds and xenobiotics. Therefore new ferrofluid-modified materials
– spruce sawdust [17, 18], baker’s yeast (Saccharomyces cerevisiae) cells [19], brewer’s
yeast (Saccharomyces cerevisiae subsp. uvarum) cells [20, 21], fodder yeast (Kluyveromyces
fragilis) cells [22] and unicellular algae (Chlorella vulgaris) cells [23] – containing
maghemite nanoparticles as magnetic labels – were prepared and tested as possible adsorbents
for binding of different substances. The structural and magnetic properties of the developed
materials were studied in detail by means of scanning electron microscopy, transmission
electron microscopy, ESR spectroscopy and conventional magnetic methods (DC
magnetization and AC susceptibility measurements). These studies are of considerable
interest for development of new inexpensive magnetic affinity adsorbents, which could
exhibit the peculiar features enabling their rapid and efficient removal not only from
solutions, but also from suspensions. Such materials could be efficiently used to isolate rare
biologically active compounds from difficult-to-handle materials including raw extracts,
blood and other body fluids, cultivation media, environmental samples, etc. They are also
greatly suitable as magnetic adsorbents for large-scale magnetic separation processes.
2. MAGNETIC MODIFICATION OF BIOLOGICAL MATERIALS

An absolute majority of biological materials is diamagnetic (i.e. is not attracted to a
magnetic field). Therefore, when a magnetic separation technique is applied, these materials
have to be first magnetically modified, usually by forming complexes with magnetic particles.
Different procedures are available to convert diamagnetic biological materials into their
magnetic derivatives [8]. However, for large-scale applications (e.g. in biotechnology or
environmental technology), relatively cheap and readily available magnetic adsorbents are
necessary. That is the reason why the new, extremely simple procedures for magnetic
modification of biological materials were developed and presented here.
According to these procedures the biological materials are magnetically modified by a
contact with water-based ionic magnetic fluids stabilized with perchloric acid or
tetramethylammonium hydroxide. The ferrofluids are prepared using the standard Massart
procedure [24, 25] and contain magnetic iron oxide nanoparticles (usually in the form of
maghemite) with the diameter of about 12 nm. Figure 1 shows exemplary TEM image of the
obtained perchloric acid stabilized magnetic fluid and the corresponding log-normal
distribution of particle diameters.
60
<D> = 12.6 nm
Number of particles
50
σ = 0.31
40
30
20
10
0
0 5 10 15 20 25 30
D [nm]
Figure 1. TEM image of perchloric acid stabilized magnetic fluid obtained according to Massart
procedure (the bar corresponds to 200 nm) and the corresponding log-normal distribution of particle
diameters. The vertical bars are the experimental data, whereas the solid line represents the fit to the
log-normal distribution function.
The above-mentioned ionic ferrofluids can be prepared in a simple way (almost in any
lab) and such nanoparticles can be used to prepare biocomposite materials enabling their
simple magnetic separation with standard permanent magnets. Both of these properties are
important for large-scale magnetic separation techniques.
Having in mind possible large-scale applications, the inexpensive raw materials should be
also used. Therefore spruce sawdust (waste material) and microbial cells (baker’s, brewer’s
and fodder yeast and unicellular algae, all of them produced in large quantities and available
at low price) were taken into consideration.
In order to produce magnetically modified spruce sawdust the raw material was sieved to
obtain particles with the diameter less than 0.5 mm. Then 500 mg of sawdust was suspended
in 7 ml of methanol in a test tube and 1 ml of perchloric acid stabilized ferrofluid was added.
The suspension was mixed on a rotary mixer (Dynal) for 1 hour. During this time almost
complete adsorption of ferrofluid nanoparticles on the sawdust occurred. Then the magnetic
sawdust was repeatedly washed with water and the suspension was stored at 4oC [18].
Alternatively, the modified sawdust was washed with methanol and air-dried.
Figure 2. Raw (SEM image) and magnetically modified (optical micrograph) spruce sawdust particles.
The scale bars correspond to 50 μm.
Figure 2 shows microscope images of spruce sawdust particle before and after magnetic
modification. It is seen that maghemite nanoparticles are attached to the sawdust particle
surface as the individual magnetic particles or agglomerates of these particles. The prepared
magnetic sawdust can be easily manipulated by means of magnetic field and therefore it can
be used as the magnetic adsorbent in magnetic separation techniques.
A different procedure was used when working with dried fodder yeast (Kluyveromyces
fragilis) and unicellular algae (Chlorella vulgaris) cells. The microbial cells were first washed
six to eight times with an excess of 0.1 M acetic acid in order to remove substantial portion of
soluble macromolecules which otherwise caused spontaneous precipitation of magnetic fluid.
Then 1 ml of perchloric acid stabilized ferrofluid was added to 3 ml of the suspension of
washed cells in acetic acid (1 + 3, v/v) and the suspension was mixed at room temperature for
1 h on a Dynal MX1 sample mixer (Invitrogen, USA). The adsorption of maghemite
nanoparticles onto the microbial cells was fast; majority of nanoparticles was adsorbed within
several minutes. The residual ferrofluid was removed by washing with 0.1 M acetic acid and
then by repeated washing with water, until the supernatant was clear. The magnetized cells
were captured using an appropriate magnetic separator. The resultant magnetic adsorbents
were stored in water suspensions at 4 oC [19, 20].
Figure 3. Cross-sections of fodder yeast cells (before and after magnetic modification) observed by
TEM. The scale bars correspond to 200 nm.
Figure 4. TEM picture of original dried Chlorella vulgaris cell (left) and magnetically modified cell
(right). The bar lines correspond to 1 μm.
Figures 3 and 4 show TEM images of original dried and magnetically modified microbial
cells. The drying process caused damage to the cell walls, often followed by the release of the
intracellular components. Magnetic modification influenced the whole cells not the cell
fragments. Analysis of TEM micrographs showed the presence of both isolated magnetic
nanoparticles and their agglomerates on the cell surface. The nanoparticles were roughly
spherical in shape and externally attached to the microbial cells walls. The outer cell surface
preferentially accumulated magnetic nanoparticles even in the case of ruptured cells; only
negligible binding of magnetic nanoparticles on the inner cell wall surface was observed.
Obtained this way magnetically modified microbial cells could be easily separated using
commercially available magnetic separators or strong permanent magnets and therefore they
can be used as cheap magnetic adsorbents. These biocomposite materials were stable even
after one year storage of the suspension at 4 ºC.
It was found recently that the procedure used for sawdust modification (methanol / acid
ferrofluid) can be also successfully used for dried microbial cells modification (unpublished
results). To prepare magnetically responsive baker’s yeast cells with highly active
intracellular enzymes, tetramethylammonium hydroxide stabilized ferrofluid in glycin-NaOH
buffer was used; such cells have been used as non-toxic and efficient magnetically responsive
whole cell biocatalysts for hydrogen peroxide decomposition and invert sugar formation [26].
As mentioned above the new developed procedures of obtaining of magnetically
modified biological materials are cheap and extremely simple and may be performed almost
in any lab. Therefore they allow to produce also other new magnetic adsorbents which can be
used for large-scale magnetic separation.
3. MAGNETIC CHARACTERIZATION OF MAGNETICALLY MODIFIED

BIOLOGICAL MATERIALS
Magnetically modified biological materials obtained according to above-mentioned
preparation procedure were characterized by means of EPR spectroscopy and conventional
magnetic methods in order to test if they can be used as magnetic affinity adsorbents in
magnetic separation procedures. As mentioned above, such magnetic adsorbents should be
magnetic field, but retained no residual magnetism when removed from the field. They should
also form stable colloidal suspensions and they should not aggregate in magnetic fields.
3.1. DC Magnetization Measurements
DC magnetization was measured by means of an extraction magnetometer MagLab 2000

System (Oxford Instruments Ltd.) in the applied magnetic field ± 3 kOe and wide temperature
range 4 – 300 K. Figure 5 shows temperature dependencies of magnetization measured in
zero field cooled – field cooled (ZFC-FC) regime at the applied magnetic field of 50 Oe for
magnetically modified spruce sawdust and two kind of magnetically modified microbial cells
(Kluyveromyces fragilis and Chlorella vulgaris). The ZFC curves were obtained by first
cooling the samples in zero magnetic field from 300 to 4 K. Then the magnetic field H = 50
Oe was applied and the magnetization was measured with increasing temperature. The FC
curves were obtained in a similar manner except that the samples were cooled in the same
measuring field H = 50 Oe.
It is seen that the ZFC-FC curves split at T < TB (TB = 190K, 210 K and 250 K for fodder
yeast cells, algae cells and spruce sawdust, respectively), indicating the existence of the
irreversible processes. The observed behavior is reminiscent of a blocking process of small
single domain particles, which turn to a superparamagnetic state with increasing temperature.
The ZFC curves for all investigated materials show a maximum associated to the
transition between the superparamagnetic and blocked state. Moreover these maxima are very
broad and a clear Curie-Weiss law behavior is not observed above the blocking temperature
TB. This indicates the existence of dipole-dipole interactions between the particles, which
may lead to the formation of small agglomerates and thus a wide distribution in particle size
ranging from ultrafine isolated particles up to particle aggregates. The magnetization in the
FC curves for all materials does not fulfill the Curie-Weiss law, which also reveals the
presence of non-negligible dipole-dipole interactions between the particles. Such behavior has
been observed in several iron oxide particle systems [27 - 29].
At about 250 K and 280 K a kink is seen in both ZFC and FC curves for microbial
materials and spruce sawdust, respectively, which is associated with the melting point of the
solution.
Above the blocking temperature, the field dependent magnetization curves of all prepared
materials show the superparamagnetic behavior indicated by the absence of hysteresis (see
Figure 6). At lower temperatures the magnetization curves display the hysteresis, which
confirms that the maghemite nanoparticles are in the ferrimagnetic state.
0,08 fodder yeast cells

FC
0,07
FC algae cells
0,06
FC
M [emu/g]
0,05
spruce sawdust
0,04
0,03 ZFC
0,02
ZFC ZFC
0,01
0 50 100 150 200 250 300

T[K]
Figure 5. Temperature dependencies of magnetization for magnetically modified fodder yeast and algae
cells and spruce sawdust recorded at the applied magnetic field of 50 Oe.
0,4 fodder yeast cells
0,3 T = 4.2 K
algae cells
0,2
0,1
M [emu/g]
spruce sawdust
0,0
-0,1
-0,2
-0,3
-0,4
-3 -2 -1 0 1 2 3
H [kOe]
0,4
fodder yeast cells
0,3 T = 300 K
0,2 algae cells
0,1
M [emu/g]
spruce sawdust
0,0
-0,1
-0,2
-0,3
-0,4
-0,3 -0,2 -0,1 0,0 0,1 0,2 0,3
H [kOe]
Figure 6. Field dependent hysteresis loops for magnetically modified fodder yeast and algae cells and
spruce sawdust recorded at 4.2 and 300 K.
At T = 4.2 K, the remanence-to-saturation ratio for all prepared materials, R = MR/MS =

0.37, is smaller than the expected R = 0.5 value for non-interacting, randomly oriented
particles with uniaxial symmetry [30]. It is an additional confirmation for the existence of
interparticle magnetic dipole-dipole interactions.
3.2. AC Susceptibility Measurements
In order to study the effects of interparticle interactions on the dynamics of the blocking
process, temperature dependence of the in-phase (real) component χ' (T) and the out-of-phase
(imaginary) component χ '' (T) of the AC magnetic susceptibility were measured for different
frequencies ƒ of the excitation field for all magnetically modified materials. Measurements
were performed by means of an extraction magnetometer MagLab 2000 System (Oxford
Instruments Ltd.) using an excitation field of 10 Oe and driving frequencies in the range 36 –
2237 Hz.
The experimental data (see Figure 7) for all prepared materials exhibit the expected
behavior of a superparamagnetic system, i.e., the occurrence of a maximum in both
components χ' (T) and χ'' (T) at different temperatures Tm' and Tm'' which shift towards
higher temperatures with increasing frequency ƒ [31]. At about 260 K and 280 K a kink is
seen in both χ' (T) and χ '' (T) curves for microbial materials and spruce sawdust,
respectively, which is associated with the melting point of the solution. At low temperatures
all χ'' (T) curves show oscillations, which are due to the imperfection of the experimental
system.
The real component of the AC susceptibility of all materials shows a value not equal to
zero for T approaching zero (see Figure 7). It may be due to the presence of small
agglomerates of particles coupled by magnetic dipolar interactions. These interactions modify
the magnetic behavior of the system, introducing a collective component which has the
influence on the low temperature magnetic relaxation. It has been shown [31] that the
magnetic relaxation of an interacting nanosized magnetic particle system at low temperatures
is extended towards longer time scales as compared to the relaxation of a non-interacting
particle system. Another indication of the influence of magnetic dipole-dipole interactions on
the dynamics of the samples comes from the increasing with increasing frequency of the
height of the peak in χ'' (T) for all investigated materials (see Figure 7), whereas it is almost
constant with frequency for a non-interacting system [31].
The empirical parameter Φ [32], which represents the relative shift of the temperature Tm
per interval of frequency,
ΔTm
Φ= , (3.1)
Tm Δ log10 ( f )
calculated from the frequency dependence of the maximum in the imaginary χ'' (T) part of
the AC susceptibility is equal to 0.10, 0.09 and 0.08 for spruce sawdust, yeast cells and algae
cells, respectively. These values are close to the 0.1 – 0.13 values found for
superparamagnetic systems [32]. The slightly lower values originate from interparticle
magnetic dipole-dipole interactions.
20 20
36 Hz 36 Hz
216 Hz 216 Hz
χ" x 10 [emu/g*Oe]
16
[emu/g*Oe]
1117 Hz 1117 Hz
15 2237 Hz
2237 Hz
12
10
8
-5
-4
χ' x 10
5
4
fodder yeast cells fodder yeast cells
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
T [K] T [K]
12 36 Hz 12 36 Hz
216 Hz 216 Hz
[emu/g*Oe]
χ" x 10 [emu/g*Oe]
1117 Hz 1117 Hz
2237 Hz 2237 Hz
8 8
-4
-5
χ' x 10
4 4
algae cells algae cells

0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
T [K] T [K]
4
36 Hz 36 Hz
4 216 Hz 216 Hz
[emu/g Oe]
937 Hz
[emu/g Oe]
3 937 Hz
2237 Hz 2237 Hz
3
2
2
-4
-5
χ' x 10
χ" x 10
1 1
spruce sawdust spruce sawdust
0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
T [K] T [K]
Figure 7. Temperature dependencies of the real (χ') and imaginary (χ'') components of the magnetic
susceptibility for magnetically modified fodder yeast and algae cells and spruce sawdust recorded at
different excitation frequencies. Arrows indicate increasing frequencies. The data were taken with an
external magnetic field H = 10 Oe.
3.3. Electron Spin Resonance (ESR) Measurements
In order to give a deeper insight into the nature of the magnetic structures described
above, ESR measurements were performed. ESR spectra were recorded by means of a
standard X-band spectrometer (Bruker EMX – 10/12) operating at 9.46 GHz with 100 kHz
field modulation. Resonance absorption was measured as a first derivative of the absorbed
microwave power versus magnetic field.
Figure 8 shows the ESR spectra of all magnetically modified materials plus the spectra of
the concentrated ferrofluid (1:1) used for magnetic modification and diluted ferrofluid
(1:100). The huge dilution (from 1:1 to1:100) of the ferrofluid sample causes a relative shift
in the resonance field of about 50 Oe. However, the shift observed in the resonance field
between the concentrated ferrofluid sample (1:1) and the ferrofluid-modified materials is
about 196, 319 and 362 Oe for spruce sawdust, fodder yeast cells and algae cells,
respectively. This remarkable resonance field shift is the signature of a strong interaction
among the maghemite nanoparticles when attached to the biological materials. The ESR data
confirm that maghemite nanoparticles do attach to the microbial cell wall and spruce sawdust
surface after incubation of these materials with the biocompatible magnetic fluid sample as
revealed by the microscope micrographs (see Figures 2, 3 and 4).
The exemplary ESR spectra of magnetically modified biological materials recorded at
various temperatures are presented in Figure 9. The room temperature spectra for all materials
show well-defined single broad signals with an effective g value of about 2.1 and the peak-to-
peak line width ΔHpp = 1040, 990 and 880 Oe for spruce sawdust, fodder yeast cells and algae
cells, respectively. The line widths of these signals considerably exceed the
magnetocrystalline-anisotropy-determined minimum value, ΔHpp = 400 Oe, for non-
interacting single domain maghemite particles [33]. It suggests the existence of non-
negligible dipole-dipole interactions between nanoparticles.
15 spruce sawdust
ESR signal intensity [arb. units]
10 fodder yeast cells
algae cells
5
-5
-10 diluted ferrofluid 1:100
-15
ferrofluid 1:1
-20
1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0
H [kOe]
Figure 8. Room temperature ESR spectra of maghemite nanoparticles in labeled fodder yeast and algae
cells and spruce sawdust and maghemite nanoparticles suspended as a stable magnetic fluid samples at
different concentrations.

T = 300 K
1 T = 4.2 K
T = 100 K
-1
fodder yeast cells
0 1 2 3 4 5 6
H [kOe]
2
T= 300 K
T= 4.2 K
1 T= 100 K
-1
-2 algae cells
0 1 2 3 4 5 6
H [kOe]
T = 100 K
T = 300 K
2 T = 4.2 K
2+
Cu
-2
free radicals spruce sawdust
0 1 2 3 4 5 6
H [kOe]
Figure 9. ESR spectra of magnetically modified fodder yeast and algae cells and spruce sawdust
recorded at various temperatures.
On the spruce sawdust resonance line two more ESR signals are superimposed (see
Figure 9), which are clearly seen only at low temperatures due to their paramagnetic
behavior:
• A sharp signal, at g = 2.00, which seems to be due to free radicals, usually found in
biological materials [34].
• A narrow (ΔBpp = 19 mT) resonance at g = 2.27, which seems to be attributed to the
existence of paramagnetic Cu2+ complexes, sometimes found in wood [34].
Upon decreasing the temperature the main resonance signals for all materials shift to
lower fields and gradually broaden, closely following the predictions for the ESR of
superparamagnetic nanoparticles systems [35]. Therefore, it may be expected that the
broadening and shift of these signals is associated with a blocking of the magnetization in
maghemite nanoparticles.
In general, for a superparamagnetic system of particles having a statistical distribution of
shapes and sizes, the simple power relation between the resonance field shift δHres and the
peak-to-peak line width ΔHpp can be expressed as [35]:
δHres ~ (ΔHpp)n (3.2)
where n = 2 (3) is predicted for partially (randomly) oriented particles.
In order to test this power relation we have plotted the data of figure 9 on a double
logarithmic scale (see Figure 10). It is seen that a straight line with a slope close to 3 (3.08 for
spruce sawdust and 3.32 for microbial cells) can well approximate the data for all prepared
materials. Thus, the broad ESR signal is shown to be due to the superparamagnetic behavior
of small isolated randomly oriented maghemite particles.
200
200
180 spruce sawdust
algae cells
160 100
80
δHres [Oe]
δHres [Oe]
60
fodder yeast cells
140
40
n
δHres ~ (ΔHpp)
120 n = 3.08 n
δH ∼ (Δ H ) 20
res pp
n = 3.32
1000 1050 1100 1150 1200 1250 900 1000 1100 1200 1300
ΔHpp [Oe] ΔHpp [Oe]
Figure 10. Relation between the resonance field shift (δHres) and peak-to-peak line width (ΔHpp) of the
ESR signal for magnetically modified spruce sawdust and fodder yeast and algae cells. Points
correspond to the experimental data, the solid line is the best fit according to Equation 3.2.
The presented ESR results for all prepared materials are in good agreement with the
magnetization, susceptibility and TEM measurements and confirm that the magnetic behavior
of these materials is dominated by the superparamagnetic relaxation of isolated single domain
maghemite particles, although a certain amount of agglomerates of these particles coupled by
magnetic dipole-dipole interactions is also present. However, these agglomerates are
sufficiently small to show at static conditions the superparamagnetic behavior at room
temperature. Therefore, the obtained results are very promising from the point of view of
using the prepared materials as the magnetic affinity adsorbent in the magnetic separation
techniques.
4. ADSORPTION PROPERTIES OF MAGNETICALLY MODIFIED

BIOLOGICAL MATERIALS
Magnetically modified spruce sawdust and microbial cells were tested as possible
adsorbents for binding of different substances [17-23]. They efficiently adsorbed different
organic compounds, such as water-soluble organic dyes, and mercury ions. Eight different
dyes belonging to four dye classes were tested, namely, crystal violet and aniline blue
(triphenylmethane group), amido black 10B, congo red, Saturn blue LBRR and Bismarck
brown Y (azodyes group), acridine orange (acridine group) and safranin O (safranin group).
Commercially available dyes were used during the experiments. They were dissolved in
distilled water without buffering the solution. Preliminary experiments also showed that
adsorption properties of spruce sawdust and microbial cells were not significantly influenced
by magnetic modification.
The adsorption of the tested dyes reached equilibrium in approximately 30-120 min. In
order to achieve equilibrium during the adsorption process, an incubation time of 3 h was
used for all adsorption experiments.
Measurements were performed in test tubes containing 50 μl of sedimented magnetically
modified biological material, appropriate amount of tested dye solution and water to a total
volume of 10 ml. The suspension was mixed for 3 h at room temperature. Then the magnetic
adsorbent was separated from the suspension using a magnetic separator (Dynal MPC-1 or
MPC-6, Invitrogen, USA) and the clear supernatant was used for the spectrophotometric
measurement. The concentration of free (unbound) dye in the supernatant (Ceq) was
determined from the calibration curve. The amount of dye bound to the unit volume of the
adsorbent (qeq) was calculated using the following formula [20]:
Dtot − 10C eq
q eq = (4.1)
50
where Dtot is the total amount of dye used in the experiment.
Examples of equilibrium adsorption isotherms for the unbuffered water solutions of

selected tested dyes using magnetically modified fodder yeast and algae cells as adsorbents
are presented in Figure 11. These isotherms represent distribution of dyes between the
aqueous and solid phases as the dye concentration increases. They follow the typical
Langmuir adsorption pattern given by [22, 23]:
Qmax bC eq
q eq = (4.2)
1 + bC eq
where qeq (expressed in mg/g or mg/ml) is the amount of the adsorbed dye per unit mass or
sedimented volume of magnetically modified biomass and Ceq (expressed in mg/l) is the
unadsorbed dye concentration in solution at equilibrium. Qmax is the maximum amount of the
dye per unit mass or sedimented volume of biomass to form a complete monolayer on the
surface bound at high dye concentration and b is a constant related to the affinity of the
binding sites (expressed in l/mg).
The Langmuir model allows simple calculation of maximum adsorption capacity Qmax,
which is a very important parameter describing the adsorption process. The results obtained
for the magnetically modified fodder, baker’s and brewer’s yeast and algae cells and spruce
sawdust are presented in Table 1.
Table 1. Comparison of maximum adsorption capacities of magnetically modified

microbial cells and spruce sawdust for the tested dyes
Maximum adsorption capacity of magnetically modified biomaterials [mg/g]
Fodder Brewer’s Baker’s Algae cells Spruce sawdust

Dye yeast cells yeast cells yeast cells
[22] [20] [19] [23] [18]
Acridine orange 62.2 82.8 24.1

Amido black 10B 29.9 11.6
Aniline blue 228.0 430.2 257.9
Bismarck brown 75.7 201.9 52.1
Congo red 49.7 93.1 156.7
Crystal violet 42.9 41.7 85.9 42.9 52.4
Safranin O 138.2 46.6 90.3 115.7 25.0
Saturn blue LBRR 33.0 24.2
The results show that dye sorption by ferrofluid-modified biomass follows a chemical,
equilibrated and saturable mechanism. Thus, the adsorption increases when the initial dye
concentration rises, as long as binding sites are not saturated. The fact that the Langmuir
isotherm fits the experimental data very well may be due to the homogeneous distribution of
active sites on the magnetically modified sawdust particles and microbial cells [36].
The prepared magnetically modified microbial cells were tested as possible adsorbents
for binding of heavy metal ions. Heavy metal pollution represents an important environmental
problem due to the toxic effects of metals, and their accumulation throughout the food chain
leads to serious ecological and health problems. Magnetically modified brewer’s yeast were
used to study Hg2+ biosorption-desorption process using synthetic solutions in batch system.
The biosorption process was fast; 80% of biosorption occured within 60 min and equilibrium
was achieved at around 90 min. The maximum Hg2+ biosorption capacity was 114.6 mg/g at
35 ºC and this value decreased to 76.2 mg/g at room temperature. The adsorption was well
fitted to the Langmuir isotherm. Results suggest that chemisorption processes could be the
rate-limiting step in the biosorption process. The yeast biomass could be easily and efficiently
regenerated by 0.1M HNO3. Biosorption of other heavy metal ions from artificial wastewater
was also studied; the biosorption capacities were 29.9 mg/g for Cu2+, 14.1 mg/g for Ni2+ and
11.8 mg/g for Zn2+ at room temperature [21].
fodder yeast cells Bismarck brown

160 algae cells
Congo red
Safranin O
120 Safranin O
qeq [mg/g]
80
Bismarck brown
Congo red
40
0
0 50 100 150 200 250
3
Ceq [mg/dm ]
Figure 11. Equilibrium adsorption isotherms (calculated from experimental data using Langmuir
equation (4.2)) for unbuffered water solutions of tested organic dyes using magnetically modified
fodder yeast [22] and algae cells [23]. Ceq – equilibrium liquid-phase concentration of the unbound dyes
(mg/dm3); qeq – equilibrium solid-phase concentration of the adsorbed dyes (mg/g).
The prepared magnetically modified spruce sawdust was additionally tested as a possible
adsorbent for binding of different enzymes [17]. It was successfully used for the batch
separation of proteolytic enzymes (trypsin, bacterial protease produced by a strain of Bacillus
sp.) and hen egg white lysozyme. The degree of lysozyme purity increased from 65% (a
technical preparation) to 96% (after the magnetic sawdust treatment) [17].
5. CONCLUSION
The main aim of this work was to find cheep and easy to get magnetic adsorbents which
could be used for large-scale magnetic separation procedures. Such materials should be
magnetic field, but retained no residual magnetism when removed from the magnetic field.
They should form stable colloidal suspensions and they should not aggregate in magnetic
fields. They should also have an affinity to corresponding biologically active compounds and
xenobiotics.
We proposed an inexpensive, extremely simple procedure for the preparation of such
magnetic adsorbents using standard water-based ferrofluids containing magnetic iron oxides
(mainly maghemite) nanoparticles with the diameter of about 12 nm. Such ferrofluids can be
prepared in a simple way and such nanoparticles can be used to prepare biocomposite
materials enabling their simple magnetic separation with standard permanent magnets. Both
of these properties are important for large-scale applications.
Inexpensive and easily available raw materials – spruce sawdust, fodder yeast
(Kluyveromyces fragilis), baker’s yeast (Saccharomyces cerevisiae), brewer’s yeast
(Saccharomyces cerevisiae subsp. uvarum) and unicellular algae (Chlorella vulgaris) cells –
were used for magnetic modification. It allows to use these materials for large-scale
applications.
The prepared magnetic adsorbents efficiently adsorbed several water-soluble organic
dyes, belonging to different groups, and also heavy metal ions. It means that they can thus be
new promising magnetic affinity adsorbents which may be used to the large-scale removal of
environmental contaminants from polluted water sources.
The magnetic measurements show that the magnetic behavior of the prepared materials is
mainly dominated by the superparamagnetic relaxation of isolated single domain maghemite
nanoparticles, although a little amount of agglomerates is also present. However, these
agglomerates are sufficiently small to show at static conditions the superparamagnetic
behavior at room temperature. The magnetic properties of the prepared magnetic adsorbents
enable their rapid and efficient removal not only from solutions, but also from suspensions, so
they could be used in the separation process performed directly in unprocessed samples such
as waste water, biological fluids, fermentation media, etc.
Summarizing, inexpensive raw materials, extremely simple preparation method, affinity
to various biologically active compounds and both organic and inorganic xenobiotics, and
distinctive magnetic properties make the developed materials greatly suitable as new
magnetic adsorbents for large-scale magnetic separation processes.
REFERENCES
[1] Safarik, I; Safarikova, M. In Scientific and Clinical Applications of Magnetic Carriers;
Hafeli, U; Schutt, W; Teller, J; Zborowski, M. Ed; Plenum: New York and London,
1997, pp 323-340 .
[2] Safarik, I; Safarikova, M. In Encyklopedia of Separation Science; Wilson, I. D; Adlard,
T. R; Poole, CF; Cool, M; Ed; AcademicPress Ltd.: London, 2000, pp 2163-2170.
[3] Safarik, I; Safarikova, MJ. Chromatogr. B., 1999, 722, 33-53.
[4] Ebner, AD; Ritter, JA; Ploehn, HJ; Koche, RL; Navratil, JD. Separ. Sci. Technol., 1999,
34, 1277-1300.
[5] Shukla, A; Zhang, YH; Dubey, P; Margrave, J. L; Shukla, SS. J. Hazard. Mater., 2002,
95, 137-152.
[6] Garg, VK; Kumar, R; Gupta, R. Dyes Pigment., 2004, 62, 1-10.
[7] Yu, B; Zhang, Y; Shukla, A; Shukla, SS; Doris, KLJ. Hazard Mater, 2001, 84, 83-94.
[8] Safarik, I; Safarikova, M. China Particuology, 2007, 5, 19-25.
[9] Banat, I. M; Nigam, P; Singh, D; Marchant, R. Bioresour. Technol., 1996, 58, 217-227.
[10] Robinson, T; McMulan, G; Marchant, R; Nigam, P. Bioresour. Technol, 2001, 77,
247-255.
[11] Zollinger, H. Colour Chemistry: Synthesis, Properties and Applications of Organic
Dyes and Pigments; VCH Publishers: New York, 1987.
[12] Safarik, IJ. Chromatogr. A, 1984, 294, 504-506.

[13] Kobayashi, O; Matsui, K; Minamiura, N; Yamamoto, TJ. Chromatogr. A, 1981, 210,
180-185.
[14] Fujimoto, KI; Ogawa, M; Saito, N; Kosaki, G; Minamiura, N; Yamamoto, T. Biochim.
Biophys. Acta, 1980, 612, 262-267.
[15] Palonen, H; Tjerneld, F; Zacchi, G; Tenkanen, M. J. Biotechnol. 2004, 107, 65-72.
[16] Sakanaka, SJ. Agr. Food Chem., 2003, 51, 3140-3143.
[17] Safarik, I; Safarikova, M; Weyda, F; Mosiniewicz-Szablewska, E; Ślawska-
Waniewska, AJ. Magn. Magn. Mater., 2005, 293, 371-376.
[18] Safarik, I; Lunackova, P; Mosiniewicz-Szablewska, E; Weyda, F; Safarikova, M.,
Holzforschung, 2007, 61, 247-253.
[19] Safarik, I; Ptackova, L; Safarikova, M. Eur. Cells Mater., 2002, 3 Suppl. 2, 52-55.
[20] Safarikova, M; Ptackova, L; Kibrikova, I; Safarik, I. Chemosphere, 2005, 59, 831-835.
[21] Yavuz, H; Denizli, A; Gungunes, H; Safarikova, M; Safarik, I. Sep. Purif. Technol.,
2006, 52, 253-260.
[22] Safarik, I; Rego, LFT; Borovska, M; Mosiniewicz-Szablewska, E; Weyda, F;
Safarikova, M. Enz. Microb. Technol. 2007, 40, 1551-1556.
[23] Safarikova, M; Rainha Pona, BM; Mosiniewicz-Szablewska, E; Weyda, F; Safarik, I.
Fresenius Environ. Bull., 2008, 17, 486-492.
[24] Massart, R. IEEE Trans. Magn., 1981, 17, 1247-1248.
[25] Berger, P; Adelman, N. B; Beckman, K. J; Campbell, D. J; Ellis, A. B; Lisensky, G. C.
J. Chem. Educ., 1999, 76, 943-948.
[26] Safarikova, M; Maderova, Z; Safarik, I. Food Res., Int., in press.
[27] Del Barco, E; Asenjo, J; Zhang, XX; Pieczyński, R; Julia, A; Tejada, J; Ziolo, RF;
Fiorani, D; Testa, AM. Chem. Mater., 2001, 13, 1487-1490.
[28] Mamiya, H; Nakatami, I; Furubayashi, T. Phys. Rev. Lett., 1998, 80, 177-180.
[29] Dormann, JL; D’Orazio, F; Lucari, F; Tronc, E; Pren, P; Jolivet, J; Fiorani, D;
Cherkaoui, R; Nogues, M. Phys. Rev. B, 1996, 53, 14291-14297.
[30] [Denham, CR; Blackemore RP; Frankel, RB. IEEE Trans. Magn., 1980, 16, 1006-1007.
[31] Jonsson, T; Norblad, P; Svedlindh, P. Phys. Rev. B, 1998, 57, 497-504.
[32] Dormann, JL; Fiorani, D; Tronc, E. Adv. Chem. Phys., 1997, XCVIII, 326-330.
[33] Dunlop, DJ; Ozdemir, O. Rock Magnetism: Fundamentals and Frontiers; Cambridge
University Press: New York, 1997.
[34] Swartz, HM; Bolton, JR; Borg, DC. Biological Applications of Electron Spin
Resonance; Wiley-Interscience, a Division of John Willey & Sons, Inc: New York,
1972.
[35] Nagata, K; Ishihara, AJ. Magn. Magn. Mater., 1992, 104-107, 1571-1573.
[36] Ozacar M; Sengil, IA. Bioresour. Technol. 2005, 96, 791-795.
Chapter 10
OPTIMISATION OF DEPOSITION CONDITIONS FOR

FUNCTIONAL OXIDE THIN FILMS
Sharath Sriram* and Madhu Bhaskaran

ABSTRACT
This chapter describes a systematic approach to determining the optimal conditions
required to deposit thin films of complex functional oxides using RF magnetron
sputtering. The motivation of this study was to attain films of designed stoichiometry and
of preferentially oriented perovskite crystal structure, as the composition and structure of
complex oxides determines their functionality (e.g. ferroelectricity, piezoelectricity, etc.)
which is exploited by a variety of applications. The process has been demonstrated with
strontium-doped lead zirconate titanate (PSZT) thin films using a combination of
materials characterisation techniques – X-ray diffraction, X-ray photoelectron
spectroscopy, atomic force microscopy, and crystal structure calculations. The major
variables in the deposition process using RF magnetron sputtering were identified as the
deposition substrate temperature, post-deposition thermal processing, oxygen partial
pressure during deposition, and the choice of metallisation on silicon substrates. Starting
with these variables the influence of each was systematically determined. The results
highlighted the advantages of slow cooling to promote perovskite growth, the influence
of oxygen on the composition and crystal structure of the thin films, and the presence of
modified unit cell structure for the PSZT thin films on gold- and platinum-coated silicon
substrates. The combination of these results led to the limiting the variables in deposition
to finite values and arriving at two definite sets of deposition conditions for deposition
based on the dependence of microstructure on thermal conditions and the choice of
substrate. The validity of the conditions chosen is then demonstrated by deposition of
PSZT thin films on thermally-grown silicon dioxide and attaining nanocolumnar
preferentially oriented thin films.
*
Corresponding author: sharath.sriram@gmail.com, sharath.sriram@rmit.edu.au
320 Sharath Sriram and Madhu Bhaskaran
1. INTRODUCTION
In the pursuit of enhanced functionality and miniaturisation, high performance thin film
coatings are increasingly incorporated into device designs. Oxides form a very large category
of materials used for processes reliant on functionality. Oxide compounds can exhibit a wide
variety of properties such as semiconduction, piezoelectricity, ferroelectricity, gas sensitivity,
etc. Oxides can exhibit one or more functional properties, with properties often enhanced in
thin films, leading to the widespread use of the terminology ‘multifunctional oxide thin
films’.
The ability to exhibit multifunctional behaviour by an oxide is determined by its chemical
composition and crystal structure. With thin film deposition often done from pure source
material of desired composition, the resulting structure is the primary concern (though the
stoichiometry of the thin films will need to be verified). Most multifunctional oxides often
have complex crystal structure symmetries, such as tetragonal or rhombohedral phases [1].
This structure is often collectively termed ‘perovskite’ [2,3]. Such compounds generally have
the ABO3 chemical structure. An example of such a multifunctional oxide with perovskite
structure is lead titanate (PbTiO3) which exhibits piezoelectric, ferroelectric, and electro-optic
properties. In this case lead (Pb) is the A-site compound and titanium (Ti) is the B-site
compound.
The properties exhibit by such multifunctional oxide compounds can be further enhanced
by the addition of dopants. These dopants can replace either a small percentage of A- or B-site
atoms or both. This results in ‘complex multifunctional oxides’. One of the most popular
compounds in this category, with a B-site substituent is lead zirconate titanate [PZT: Pb(Zr1-
xTix)O3], in which the ferroelectric and piezoelectric properties are enhanced, resulting also in
a more stable compound. Additional substitution of A-site lead atoms of PZT results in
compounds with either new properties or further enhancement in existing ones: (i) lanthanum-
doped PZT (PLZT) is known for enhanced electro-optic behaviour [4]; (ii) stannum-doped
PZT and niobium-doped PZT (PNZT) are known to exhibit shape memory behaviour [5] (not
exhibited by PZT); and (iii) strontium-doped PZT (PSZT) has been reported to have
improved piezoelectric, ferroelectric, dielectric tenability, and pyroelectric (infra-red radiation
sensors by thermal effects) behaviour [3,6-8].
The possible multidisciplinary applications of oxide thin films are apparent, and the
multifunctionality displayed is dependent on the chemical and crystal structure. Accurate
control of deposition process and understanding of variables which control the resulting films
is vital to harnessing the capabilities multifunctional oxide thin films present. This chapter
uses a selected complex oxide and demonstrates the process of optimisation of deposition
conditions. The compound chosen is strontium-doped PZT (PSZT) which has both A- and B-
site dopants. The composition chosen of (Pb0.92Sr0.08)(Zr0.65Ti0.35)O3 is reported to exhibit
enhanced properties [3,6-8] and is expected to have the complex rhombohedral crystal
structure at room temperature [9]. The deposition process of choice is RF magnetron
sputtering, which enables large (wafer-) scale deposition and is suitable for commercial
production.
Optimisation of Deposition Conditions for Functional Oxide Thin Films 321
2. EXPERIMENTAL DETAILS
This chapter focuses on determination of optimised deposition conditions for thin films
by RF magnetron sputtering. PSZT thin films were deposited under the conditions shown in
Table 1, which highlights the variables in this process.
Table 1. Variables in PSZT thin film deposition conditions.
Target (Pb0.92Sr0.08)(Zr0.65Ti0.35)O3
RF power 100 W
Sputtering pressure variable
Process gas 10% oxygen in argon
Substrate temperature variable
Temperature ramp-up rate 10 ºC/min
Temperature ramp-down rate variable
Sputtering parameters principally determining the deposition rate of the thin films were
kept constant. These parameters (deposition power and target-substrate distance) were chosen
so as to achieve a compromise between a slow deposition rate to enable grain growth and
crystallisation, while achieving a reasonable deposition rate to achieve micron-thick films.
Initial tests showed that the choice of forward RF power of 100 W (maintaining the reflected
power at ~0 W) and a target-substrate distance of 70 mm, resulted in a deposition rate of 300-
400 nm/hour. This variation is over 3-inch substrates due to the absence of substrate rotation
option in the sputtering system. This deposition rate was considered satisfactory to achieve
micron-thick films, and as a result these parameters were kept constant during further study.
Sputtering parameters such as post-deposition cooling rate, sputtering pressure, and substrate
temperature which are expected to have a significant influence on the composition, crystal
structure, and microstructure of the thin films were varied during this study.
3. INFLUENCE OF POST-DEPOSITION COOLING RATE

While the substrate temperature during deposition is a critical variable in determining the
crystallography of these complex perovskite-structured PSZT thin films (the influence of
which is discussed in Ref. [10]), post-deposition thermal processing was expected to
influence the degree of grain growth and crystallisation. Extensive X-ray diffraction analysis
of samples deposited at high temperatures and those deposited at lower temperatures followed
by annealing was carried out. This work found that slower post-deposition cooling results in
thin films exhibiting more pronounced perovskite structure. The optimal post-deposition
cooling rate was 5 ºC/min. An example of the results obtained is presented in Fig. 1, where
the switch in the intensity of the pyrochlore and perovskite peaks at around 2θ of 30º is
apparent.
Figure 1. X-ray diffraction results for PSZT thin films deposited at 600 ºC: (a) as deposited and cooled
at 15 ºC/min, (b) as deposited and cooled at 5 ºC/min, (c) cooled at 15 ºC/min and post-deposition
annealed at 1000 ºC for 30 min in argon. (Reprinted with permission from Ref. [10].)
The process of slower cooling was also found to be better than post-deposition annealing,
as it:
(i) prevented the occurrence of microcracks in the thin films and

(ii) resulted in stronger preferential orientation, rather than a collection of perovskite
orientations.
This study [10] supported results previously presented for other ferroelectric material by
Wasa et al. [11], and helped limit the main deposition variables to two.
4. INFLUENCE OF OXYGEN PARTIAL PRESSURE

The level of piezoelectric response exhibited by PZT compounds depends on the thin
films having a composition closely corresponding to the ideal ABO3 structure and on having a
preferential perovskite orientation. Lead loss and lead excess have been reported to have a
marked influence on the materials properties of PZT type compounds [12,13], with the
composition variations affecting perovskite orientation.
Deposition of PSZT thin films was carried out in an atmosphere of 10% oxygen in argon.
Argon is an inert gas and exerts no influence on thin film composition. Changes in sputtering
pressure are directly related to changes in the partial pressure of oxygen present in the
sputtering environment. This section reports on the influence of the oxygen partial pressure
on the chemical composition, crystal structure, and surface morphology of the deposited
PSZT thin films. Deposition was carried out on platinum-coated silicon substrates at 650 ºC –
the substrate temperature was kept constant to study the influence of partial pressure. The
results obtained from films deposited at oxygen partial pressures of 1, 2, 3, 4, and 5 mTorr
(relating to sputtering pressures from 10-50 mTorr) are discussed.
4.1. Composition Analysis
X-ray photoelectron spectroscopy (XPS) analysis was carried out to study the
composition variations in the PSZT thin films deposited at varying oxygen partial pressures.
Peaks for lead, zirconium, titanium, and oxygen expected at around 137 eV, 181 eV, 454 eV,
and 531 eV, respectively, were observed. Examples of the spectra recorded are shown in Fig.
2. Peaks for the dopant (strontium) were not pronounced due to its low concentration, and
possible detected peaks at ~133 eV for strontium could not be resolved satisfactorily due to
their vicinity to the peaks of lead.
Figure 2. XPS spectra for PSZT thin films deposited at sputtering pressures of 10, 20, and 30 mTorr.
The spectra for 10 mTorr and 30 mTorr overlap, corresponding to the similar stoichiometries estimated.
(Also see Table 2. Reprinted with permission from Ref. [14].)
XPS analysis highlights the marked influence of variations in oxygen partial pressure
during sputtering, observed by variations in oxygen (O) concentration in the thin films, and in
some cases by the undesirable decrease in lead (Pb) concentration in the thin films (Table 2).
Films deposited at sputtering pressures of 10 and 30 mTorr (oxygen partial pressures of 1 and
3 mTorr, respectively) have resulting compositions which are acceptably close to the PSZT
8/65/35 stoichiometry of the sputtering target. The composition of films deposited at

sputtering pressures of 20 mTorr and 50 mTorr indicates insufficient and excess lead,
respectively. The zirconium (Zr) to titanium (Ti) ratio of films deposited at 10, 30, and 40
mTorr was reasonably constant (the ratio varying between 1.90 and 2.15), while the lead and
oxygen content varied relative to the other. As can be observed from Table 2, the films
deposited at 10 mTorr appears to have the composition closest to the desired ABO3, albeit
mildly lead deficient.
Table 2. Variations in composition of PSZT thin films with sputtering pressure.
Sputtering O2 partial Elemental composition (%)

pressure pressure
(mTorr) (mTorr) Pb Zr Ti O
10 1 18.18 14.72 7.73 59.36
20 2 12.80 17.94 9.19 60.08
30 3 22.21 14.28 6.81 56.70
40 4 25.83 14.23 6.63 53.24
50 5 26.41 13.14 5.40 55.05
The trend observed in the composition (with the exception of films deposited at 20
mTorr) agrees with earlier studies on diffusive transport regime assisted deposition of heavier
atoms [15,16]. With increasing sputtering pressure, lighter atoms experience more collisions
(with decreased mean free path), as a result of which the deposited films become richer in the
heavier species which constitute the film. A combination of the previous studies on transport
regimes during sputtering and the results presented here provide a technique for optimising
lead content (by altering the sputtering pressure) in lead-based or PZT-type films, where lead-
loss during sputtering often occurs, resulting in poor electrical performance.
4.2. Crystallographic Orientation Analysis
Glancing angle X-ray diffraction (GA-XRD) has been used to study the crystallographic
orientation of the deposited PSZT thin films. PSZT film peaks in diffractograms obtained
using Bragg-Brentano XRD were over-shadowed by substrate peaks; this was overcome by
using GA-XRD. All deposited thin films were poly-crystalline with preferential
orientation(s). Figures 3(a) and 3(b) show expected perovskite peaks at 2θ of 29.6º and 34.3º
and these correspond to (104) and (006) orientations for a rhombohedral PSZT unit cell [17].
Figure 3(c) shows another expected peak for PSZT thin films corresponding to the (108)
orientation. The relative shifts of the perovskite peaks with respect to variation in oxygen
partial pressure are evident in the diffractograms shown in Fig. 3. Films deposited at 10 and
30 mTorr, and which show similar composition, show similar orientation and least variations
from expected 2θ positions. High levels of lead (as in films deposited at 50 mTorr) results in
an increase in the exhibited d-spacings which has caused the peaks to shift left. High levels of
oxygen (as in films deposited at 20 mTorr) results in a decrease in d-spacing from expected
values which has caused the peaks to shift right. This result corresponds well with the radius
of the atoms, with lead having a larger atomic radius than oxygen.
(a) (b) (c)
Figure 3. Comparison of glancing angle X-ray diffraction results for PSZT thin films deposited under
different sputtering pressures. Parts (a-c) of the image show significant 2θ positions for PSZT.
(Reprinted with permission from Ref. [14].)
4.3. Surface Morphology Analysis
The surface morphology of PSZT thin films deposited at different oxygen partial
pressures has been analysed using atomic force microscopy (AFM). The average surface
roughness (Ra) values and grain sizes measured are listed in Table 3. An AFM scan of a PSZT
thin film deposited at 10 mTorr is shown in Fig. 4(a); these films had the least surface
roughness and had uniformly sized grains. Figure 4(b) shows an AFM scan of a PSZT thin
film deposited at 30 mTorr illustrating the presence of crystalline agglomerates. Films
deposited at 20 and 50 mTorr were similar to this, also with crystalline agglomerates. Films
deposited at 50 mTorr contain the maximum amount of lead, and are extremely rough. The
observable effect that composition had on orientation is not apparent in the case of the AFM
results, with no obvious trend being present in this case.
(a) (b)
Figure 4. AFM contact mode 5 μm × 5 μm surface scans of PSZT thin films deposited at sputtering
pressures of (a) 10 mTorr and (b) 30 mTorr. (Reprinted with permission from Ref. [14].)
Table 3. Average surface roughness (Ra) and grain size of PSZT films deposited under
different sputtering pressures.
Sputtering pressure O2 partial pressure Ra Grain size

(mTorr) (mTorr) (nm) (nm)
10 1 10.93 170
20 2 44.82 170
30 3 37.04 140
40 4 19.80 210
50 5 43.93 140
4.4. Section Summary
This section summarises the results obtained from the analyses of PSZT thin films
deposited under different sputtering pressures. The composition, orientation, and surface
roughness of thin films sputtered at oxygen partial pressures of 1, 2, 3, 4, and 5 mTorr have
been characterised and compared. XPS and GA-XRD results complement each other, with
observable effects due to either excess lead or oxygen. Increasing the sputtering pressure
increased lead concentration due to diffusion-based transport of species (and this can be used
to prevent lead loss in deposited films). High levels of lead resulted in an increase in the
exhibited d-spacings and high levels of oxygen resulted in a decrease in d-spacing, from
expected values. Surface morphology studies do not indicate an identifiable trend, but
highlights the presence of agglomerates in some cases. The PSZT thin films deposited at a
sputtering pressure of 10 mTorr (oxygen partial pressure 1 mTorr) have given the best results
– desired composition (close to the desired ABO3) and orientation, with low surface roughness
and uniformly sized grains.
5. IDENTIFICATION OF MODIFIED PSZT UNIT CELL

Having identified suitable deposition conditions in terms of post-deposition cooling rate
and sputtering pressure, the choice of temperature of deposition of the PSZT thin films was
addressed. Substrate temperatures of 500 to 700 ºC are normally used for the deposition of
thin films of PZT compounds, to attain desired perovskite structure [10,11]. In order to
incorporate these piezoelectric thin films into standard devices, deposition on silicon
substrates is desirable. Deposition of PZT has been conventionally carried out on platinum-
coated silicon substrates [10,11]. Considering the inert properties of gold and that its lattice
spacings are close to those of PSZT [18], deposition on gold-coated silicon substrates was
also investigated. As the gold-silicon eutectic point is 363 ºC, deposition on gold coated
substrates was limited to 300 ºC. Considering these preliminary choices, deposition at 300 ºC
on gold and at 600-700 ºC on platinum was investigated.
This section reports on the influence of multi-layered bottom electrodes (with gold or
platinum) on the orientation of deposited PSZT thin films of the composition
(Pb0.92Sr0.08)(Zr0.65Ti0.35)O3. For this composition, the films are expected to have a
rhombohedral unit cell at room temperature (corresponding to space group R3c) [9]; and so,
the corresponding powder diffraction file was used as a reference [19]. Bragg-Brentano X-ray
diffraction (XRD) was used to experimentally determine the d-spacings in the deposited
PSZT thin films, in order to calculate the unit cell parameters. These results have been used to
index X-ray diffractograms for PSZT thin films deposited on different multi-layer bottom
electrode configurations, and to identify those that promote perovskite orientation.
The XRD analysis carried out showed that a majority of the PSZT peaks were shifted
from the expected 2θ positions [19]; X-ray photoelectron spectroscopy was used to confirm
that composition of the films was very close to that of the reference used [14,19]. As the
diffractometer was regularly calibrated using a quartz standard, errors introduced by the
system were ruled out. The as-deposited metal films (gold and platinum) at room temperature
were expected to be nanocrystalline; but the temperatures at which PSZT deposition was
carried out resulted in grain growth with preferential orientation, manifested as strong (111)
metal peaks. These (111) metal peaks were at the expected 2θ positions [18,20], again
verifying that the PSZT peaks were shifted from the expected. From these results, unit cell
parameters for PSZT deposited on gold and platinum were calculated, which indicated that
two modified unit cells existed based on deposition temperature and substrate.
5.1. Modified PSZT Unit Cell on Gold at 300 ºC
Using d-spacings obtained from XRD (for the 2θ range of 20º-60º) the unit cell
parameters for the PSZT thin films deposited on gold were estimated. The results obtained are
summarised in Table 4.
In a similar manner to the above, the unit cell parameters for the PSZT thin films
deposited on platinum were estimated. The results obtained are summarised in Table 5. These
results show the equivalent d-spacings increase or decrease based on the dominating unit cell
parameters for the chosen orientation. The resulting changes in the rhombohedral unit cell
parameters a and c were –4.92 % and +9.49 %, respectively; this change is from the expected
a of 5.732 Å and c of 14.317 Å to 5.450±0.010 Å and 15.675±0.015 Å, respectively. This
corresponds to a decrease in the unit cell volume by about 1 %.
These results show that the majority of the equivalent (i.e. calculated) d-spacings show an
increase of about 5.26-5.31 %. The resulting change in both the rhombohedral unit cell
parameters a and c is +5.29 %; from the expected a of 5.732 Å and c of 14.317 Å to
6.035±0.010 Å and 15.075±0.015 Å, respectively. This corresponds to an increase in the unit
cell volume by about 16 %.
Table 4. Comparison between expected, experimental, and

equivalent lattice spacings for PSZT on gold.
Experimental Difference in
Expected d-spacing values Equivalent equivalent
Orientation d-spacings based on from d-spacings calculated d-spacings
ICDD PDF XRD from experimentally from expected
04-002-5985 Å Å derived unit cell Å values %
(012) 4.0793 - 4.295 +5.288
(004) - 3.767 3.769 -
(104) 2.9033 3.058 3.057 +5.294
(110) 2.8661 - 3.018 +6.661
(113) 2.4570 2.623 2.587 +5.291
(006) 2.3862 2.503 2.513 +5.314
(202) 2.3451 - 2.469 +5.283
(024) 2.0397 2.152 2.147 +5.261
(211) 1.8604 - 1.959 +5.300
(116) 1.8338 1.932 1.931 +5.300
(122) 1.8150 1.809 1.911 +5.289
(018) 1.6836 1.774 1.773 +5.310
(214) 1.6618 - 1.750 +5.308
(300) 1.6547 - 1.742 +5.276
(125) 1.5694 - 1.652 +5.263
(034) - 1.586 1.581 -
(028) - 1.531 1.528 -
Table 5. Comparison between expected, experimental, and equivalent lattice spacings

for PSZT on platinum.
Experimental Equivalent Difference in

Expected d-spacing values d-spacings equivalent
Orientation d-spacings based on from calculated from d-spacings
ICDD PDF XRD experimentally from expected
04-002-5985 Å Å derived unit cell Å values %
(012) 4.0793 4.055 4.043 –0.890
(104) 2.9033 3.014 3.015 +3.847
(110) 2.8661 2.784 2.725 –4.923
(113) 2.4570 - 2.416 –1.669
(006) 2.3862 2.614 2.613 +9.505
(202) 2.3451 2.106 2.260 –3.346
(024) 2.0397 2.027 2.022 –0.868
(211) 1.8604 - 1.772 –4.752
(116) 1.8338 1.956 1.886 +2.847
(122) 1.8150 - 1.739 –4.187
(108) 1.6836 1.849 1.810 +7.508
(214) 1.6618 - 1.624 –2.275
(300) 1.6547 1.577 1.573 –4.937
(125) 1.5694 - 1.550 –1.236
(028) - 1.510 1.508 -
5.2. Modified PSZT Unit Cell on Platinum at 650 ºC
In order to overcome this undesirable reaction, a 200 nm silicon dioxide layer was
introduced to isolate the metals from silicon. This increased the preferential (111) orientation,
relative to the (200) peak, of the underlying gold layer [18] and also the relative intensity of
the perovskite (104) PSZT orientation as shown in Fig. 5(b). Cross-sectional transmission
electron microscopy studies have confirmed that this added silicon dioxide layer prevents the
reaction between gold and silicon. Deposition duration did not influence the orientation of the
PSZT thin films, with the ratio of relative intensities of the preferential orientations remaining
unaffected. However, the extent of preferential orientation in the underlying gold layer
increased, due to continued grain growth.
5.3. Orientation of PSZT Thin Films on Gold at 300 ºC
PSZT thin films were deposited on silicon samples coated with 150 nm of gold with a 15
nm titanium adhesion layer. Figure 5(a) shows the resulting diffractogram for such films; here
two perovskite PSZT peaks can be observed. Cross-sectional transmission electron
microscopy (JEOL 2010F TEM) of these samples indicated that interdiffusion had occurred
between the bottom electrode layers and silicon [21]. This could be attributed to the fact that
the PSZT deposition on these samples was carried out at 300 ºC, which is close to the eutectic
point of gold and silicon – the eutectic point could have been reached as the vapour pressure
at the sputtering pressure would have been lower.
(a)
Figure 5. (Continued).
(b)
Figure 5. X-ray diffractograms obtained for PSZT thin films deposited on silicon samples coated with
150 nm Au using a 15 nm Ti adhesion layer without and with an intermediate 200 nm SiO2 layer are
shown in (a) and (b), respectively. (Reprinted with permission from Ref. [17].)
5.4. Orientation of PSZT Thin Films on Platinum at 650 ºC
PSZT thin films were deposited on a number of platinum-based bottom electrode

configurations on silicon substrates, as listed below:
(i) 200 nm Pt / 20 nm Ti / Si
(ii) 200 nm Pt / 20 nm Ti / 200 nm SiO2 / Si
(iii) 300 nm Pt / 70 nm Ti / 200 nm SiO2 / Si
(iv) 200 nm Pt / 20 nm TiO2 / Si
PSZT thin films on platinum require heating to temperatures of 600 ºC and above to
enhance the perovskite structure; hence, all depositions of PSZT on platinum were carried out
at 650 ºC [10,11]. Films deposited on electrode configuration (i) were found to have high
surface roughness due to the formation of platinum-titanium silicide at these temperatures
(verified by Auger electron spectroscopy), but perovskite PSZT peaks were observed [Fig.
6(a)]. A silicon dioxide layer was introduced to inhibit this silicide formation [configurations
(ii) and (iii)], along with experimenting with thicker platinum and titanium layers
[configuration (iii)], in order to preserve the smooth platinum surface. PSZT thin films
deposited on electrode configurations (ii) and (iii) were not usable, as the underlying layers
delaminated due to stress imbalance between the various layers. This is attributed to the
differences in coefficients of thermal expansion and ratio of thicknesses of the different layers
[22]. Titanium dioxide layers (known to be stable and inert) were experimented with, and
PSZT thin films deposited on electrode configuration (iv) were very smooth, as the reaction
between platinum and titanium was prevented, and preferential perovskite PSZT orientation
was obtained [Figure 6(b)].
(a)
(b)
Figure 6. X-ray diffractograms obtained for PSZT thin films deposited on (a) 200 nm Pt on 20 nm Ti on
Si and (b) 200 nm Pt on 20 nm TiO2 on Si. (Reprinted with permission from Ref. [17].)
Deposition of PSZT thin films on platinum was also carried out at 600 ºC and 700 ºC,
and there were no significant differences in the preferential orientations with temperature.
Deposition durations of two and four hours also resulted in almost identical diffractograms,
with just the substrate peaks suppressed for samples from longer depositions.
5.5. Section Summary
This section discusses the orientation dependence of PSZT thin films on bottom electrode
architectures. XRD results highlighted variations in obtained 2θ peak positions from expected
positions, indicating the existence of modified unit cells for PSZT thin films on gold and
platinum. The modified rhombohedral unit cell parameters and the equivalent d-spacings are
discussed. The variations in unit cell parameters from expected values could be attributed to
guiding effects from the crystalline bottom electrode layers; and variations appear consistent
with the larger lattice spacings for gold in comparison to platinum. The structure obtained on
gold-coated silicon at 300 ºC is unique to such gold-coated substrates, apparently due to the
guiding effect exhibited by (111) gold on (104) PSZT. No preferential orientation was
observed on other materials (such as LiNbO3, Pt, etc.) at 300 ºC. Considering the second
variable of deposition temperature, additional work was carried out with PSZT deposition on
silicon substrates with thermal silicon dioxide.
6. DEPOSITION ON THERMAL SILICON DIOXIDE

PSZT thin films were deposited either at room temperature followed by an annealing
process or at a substrate temperature of 700 ºC. PSZT thin films deposited at room
temperature were subject to post-deposition furnace annealing at 700 ºC for 3 hours in the
presence of high purity argon. In the case of samples deposited at temperatures of 700 ºC, the
samples were heated to deposition substrate temperature at a ramp rate of 10 ºC/min and
cooled, subsequent to deposition, at 5 ºC/min; previous work has shown that these conditions
improve the degree of perovskite orientation in the thin films, in line with previous results
(Section 3 and Ref. [10]).
6.1. X-Ray Diffraction Analysis
The X-ray diffractogram of PSZT thin films deposited at room temperature and subjected
to post-deposition annealing is shown in Fig. 7(a), and consists of peaks at expected 2θ
positions [17]. These peaks are characterised by weak reflections and broad peak widths;
indicative of weak preferential orientation, limited grain growth, and a nanocrystalline
structure.
Figure 7(b) is representative of diffractograms obtained for PSZT thin films deposited at
substrate temperature of 700 ºC. Expected perovskite peaks at 2θ of 29.6º and 49.3º, for (104)
and (108) orientations, for a rhombohedral PSZT unit cell [17] were observed. Smaller peaks
obtained at 34.3º and 58.5º correspond to (006) and (300) orientations, respectively. The
temperature at which these thin films were deposited was chosen to encourage thermally-
driven grain growth and the diffractogram [Fig. 7(b)] confirms that this resulted in uniform
crystal growth, manifested as strong and sharp peaks in the diffractogram. These results, with
strong c-axis preference, also promise strong columnar growth in the thin films (discussed in
sub-section 6.3).
These results indicate that though post deposition annealing encouraged grain growth,
only thin film deposition at high temperatures (in situ substrate heating) results in strong
preferential orientation. The PSZT thin film samples deposited at 700 ºC, with promising
XRD results, were subject to further analyses.
(a)
(b)
Figure 7. X-ray diffractograms obtained for PSZT thin films: (a) deposited at room temperature with
subsequent furnace annealing at 700 ºC for 3 hours and (b) deposited at a substrate temperature of 700
ºC for 3 hours.
6.2. Atomic Force Microscopy Analysis
AFM surface scan results for a PSZT thin film deposited at 700 ºC are shown in Fig. 8.
The topography images in Figs. 8(a) and 8(b) depicts tightly packed nanocrystalline grains
with an average grain size of 80-100 nm. The average surface roughness (Ra) of these films,
measured to be about 9-11 nm, indicates that the pronounced grain structure is regular. The
accentuated faceting of the grain faces is only conveyed by the peak roughness (Rp) values
measured to be ~90 nm. The polygonal structure of the grains is apparent in the deflection
image shown in Fig. 8(c).
(a)
(b) (c)
Figure 8. Atomic force microscopy scan results for PSZT thin film deposited at 700 ºC: (a) three-
dimensional representation of film surface showing facetted tightly packed grains; (b) topography
image obtained over an area of 1 μm × 1 μm; and (c) the deflection image corresponding to the
topography in (b).
6.3. Electron Microscopy Analysis
Cross-sectional scanning electron microscopy (SEM) showed that the films had
nanocolumnar grains extending through the thickness of the PSZT thin films (Fig. 9). The
structure and width of these grains matched those expected from XRD and AFM analyses.
There are regions where the PSZT grains ends abruptly [especially in Fig. 9(b)]; this was due
to the brittle nature of the ceramic PSZT thin films which prevented better sectioning.
(a)
(b)
Figure 9. Cross-sectional scanning electron micrographs in (a) and (b) showing the regular
nanocolumnar grains spanning the thickness of the film. (Observed at specimen tilt of 9º.)
Plan view TEM analysis results for PSZT thin films deposited on silicon dioxide are
presented in Fig. 10. The nanocrystals observed in the images correspond to grain sizes of 80-
100 nm for the columnar preferentially oriented grains in the structure. A majority of the
grains appear to be triangular in shape. AFM scans confirmed that the grains in the PSZT thin
film are densely packed, and this can be observed in the top-left of Fig. 10(a). In the other
region, the ion milling process for the TEM sample preparation has resulted in numerous
regions devoid of material. This is a by-product of the very rough crystalline film surface, on
which milling from the back results in the majority of the material being removed, with only
the capping regions of the grains left behind.
The thin region of the specimen gives us valuable information regarding the
nanostructure of the thin films. Figure 10(b) shows that the grains vary in size from 80 to 100
nm, with well defined crystalline and polygonal structure. Strong Bragg diffraction from
many grains in Fig. 10 indicates that they share the same orientation. These nanocrystals
extend all the way through the thickness of the specimen forming the columnar structure
observed in the cross-sectional analysis. Selected area electron diffraction of the plan view
specimen showed that the nanocrystals exhibited the expected perovskite structure, but were
randomly distributed – there was no preferential orientation along the substrate surface (XRD
indicated preferential orientation perpendicular to the substrate).
(a)
(b)
Figure 10. Plan view transmission electron microscopy results obtained for PSZT thin film deposited at
700 ºC at two different magnifications are shown in (a) and (b).
Figure 11. Schematic representation of the PSZT thin films deposited on thermal SiO2, showing the
nanocolumnar polygonal grain structure (Not to scale.)
This section presents results from the first instance of deposition of preferentially
oriented, nanocrystalline, and nanocolumnar PSZT thin films directly on thermal silicon
dioxide. A schematic depicting the overall structure of the deposited films is shown in Fig.
11. No intermediate seed or activation layers were used between PSZT and SiO2. A substrate
temperature of 700 ºC was found to be suitable for obtaining the desired perovskite structure.
This relates to the summary of the previous section and helps arrive at the conclusion that at
higher temperatures, thermal effort dominates the resulting microstructure in such complex
oxide thin films.
7. CONCLUSION
This chapter demonstrates a systematic approach in determining optimal deposition
conditions for complex oxide thin films. A material with the ABO3 perovskite structure, with
both A- and B-site dopants, was chosen for this study. This was strontium-doped lead
zirconate titanate. The influence of oxygen partial pressure, bottom metal electrode choice,
and substrate temperature on the composition and crystal structure of the resulting thin films
was analysed. The potential variables identified in Table 1 have now been replaced with
optimised parameters (Table 6).
During the course of identification of optimal deposition conditions, the following key
results have been obtained:
A. Conditions to achieve desired PSZT thin film composition have been identified.
B. In studying the influence of oxygen partial pressure on thin film composition,

conditions for deposition of controlled lead-rich films have been identified. This will
serve as a technique to design film composition in order to overcome lead loss due to
subsequent processing.
C. Suitable bottom electrode architectures for device designs and piezoelectric response
measurements have been identified.
D. Two different modified unit cell structures of PSZT, based on the guiding effect of
gold (at 300 ºC) or as a consequence of temperature (at 650 ºC), have been identified.
This result demonstrates an approach for tailoring the crystal structure of the
material.
E. The first instance of deposition of perovskite piezoelectric thin films directly on
thermal silicon dioxide (without using intermediary seed layers) is demonstrated [23-
25].
Table 6. Optimised PSZT thin film deposition conditions.
Target (Pb0.92Sr0.08)(Zr0.65Ti0.35)O3
RF power 100 W
Process gas 10% oxygen in argon
Substrate temperature 300 ºCa or 600-700 ºCb
Temperature ramp-up rate 10 ºC/min
Temperature ramp-down rate 5 ºC/min
a
Suitable for deposition on gold
b
Suitable for deposition on platinum
REFERENCES
[1] Voigt, W. Lehrbuch der Kristallphysik; B. G. Teubner: Leipzig and Berlin, 1910.
[2] Randall, CA; Kim, N; Kucera, JP; Cao, W; Shrout, TR. J. Am. Ceram. Soc., 1998, 81,
677-688.
[3] Yu, Y; Tu, J; Singh, RN. J. Am. Ceram. Soc., 2001, 84, 333-340.
[4] Tunaboylu, B; Harvey, P; Esener, SC. Integr. Ferroelectr. 1998, 19, 11-32.
[5] Yu, Y; Singh, RN. J. Appl. Phys., 2000, 88, 7249-7257.
[6] Zheng, H; Reaney, IM; Lee, WE; Jones, N; Thomas, H. J. Am. Ceram. Soc., 2002, 85,
207-212.
[7] Bedoya, C; Muller, Ch; Baudour, JL; Madigou, V; Anne, M; Roubin, M. Mater. Sci.,
Eng. B, 200, 75, 43-52.
[8] Araujo, EB; Eiras, JA. J. Phys. D: Appl. Phys., 2003, 36, 2010-2013.
[9] Jaffe, B; Cook, WR; Jaffe, H. Piezoelectric Ceramics; Academic: New York, 1971, p
136.
[10] Sriram, S; Bhaskaran, M; Holland, AS. Semicond. Sci. Technol., 2006, 21, 1236-1243.
[11] Wasa, K; Kitabatake, M; Adachi, H. Thin Film Materials Technology: Sputtering of
Compound Materials; Springer-Verlag: Heidelberg, 2004.
[12] Miclea, C; Tanasoiu, C; Miclea, CF; Amarande, L; Gheorghiu, A; Spanulescu, I;
Plavitu, C; Miclea, CT.; Cioangher, MC; Trupina, L; Iuga, A. J. Eur. Ceram. Soc.,
2007, 27, 4055-4059.
[13] Garg, A; Agrawal, DC. Mater. Sci. Eng. B, 1999, 56, 46-50.
[14] Sriram, S; Bhaskaran, M; Du Plessis, J; Short, KT; Sivan, VP; Holland, AS. Micron
2009, 40, 104-108.
[15] Rossnegal, SM; Yang, I; Cuomo, JJ. Thin Solid Films, 1991, 199, 59-69.
[16] Castaldi, L; Gibbs, MRJ; Davies, HA. J. Appl. Phys., 2003, 93, 9165-9169.
[17] Bhaskaran, M; Sriram, S; Short, KT; Mitchell, DR. G; Holland, AS. Thin Solid Films
2008, 516, 8101-8105.
19073, Card 04-0784.
19073, Card 04-002-5985.
19073, Card 04-0802.
[21] Sriram, S; Bhaskaran, M; Mitchell, DRG; Short, KT; Holland, AS; Mitchell, A.
Microscopy Microanal, 2009, 15, 30-35.
[22] Lu, DX; Pun, EYB; Wong, EMW; Chung, PS; Lee, ZY. IEEE T. Ultrason. Ferr. Freq.
Contr., 1997, 44, 675-680.
[23] Ramesh, R. International Patent WO/1994/013471. Available online at
http://www.wipo.int/pctdb/en/wo.jsp?IA=US1993010387
[24] Zhao, J; Lu, L; Thompson, CV; Lu, Y; Song, WD. Proc. SPIE, 2002, 4426, 221-224.
[25] Sriram, S; Bhaskaran, M; Mitchell, A; Mitchell, DRG; Kostovski, G. Nanoscale Res.
Lett., 2009, 4, 29-33.
Chapter 11
OPTICAL WAVEGUIDES PRODUCED BY ION

IMPLANTATION IN OXIDE GLASSES
Feng Chen∗, Xue-Lin Wang, Lei Wang and Ke-Ming Wang

School of Physics, Shandong University, Jinan, 250100, People’s Republic of China
ABSTRACT
Optical waveguides can restrict light propagation in very small size of order of
several microns, reaching high optical intensities even at low pumps; consequently, some
properties of the bulk materials may be considerably improved in waveguide structures,
such as non-linear responses, laser actions and optical signal amplifications. As one
mature technique for material property modifications, ion implantation has been applied
to construct optical waveguides in many optical materials, including insulating crystals
and glasses, semiconductors and organic materials. Oxide glasses receive much attention
for various photonic and telecommunication applications for its competitive costs and
excellent optical features. In this chapter, we reviewed the research results obtained for
optical waveguides in oxide glasses produced by ion implantation techniques, by giving a
brief introduction of basic fabrication principles and methods and a summary of
interesting results obtained in this topic. The prospects of possible practical applications
of ion-implanted oxide glass waveguides in photonics are also demonstrated.
Keywords: Optical waveguides; ion implantation; glasses.
I. INTRODUCTION
An optical waveguide structure is defined as a region with higher refractive index
surrounded by lower-index mediums, which allows the light confinement within its
boundaries by total internal reflection. The restricted inside dimensions of the light are
∗
Corresponding author. Tel.: +86-531-88364655; fax: +86-531-88565167; E-mail: drfchen@sdu.edu.cn.
342 Feng Chen, Xue-Lin Wang, Lei Wang et al.
typically with order of several microns, which is somehow comparable with the light
wavelength [1,2]. Planar or slab waveguides restrict light propagation in one dimension (1D
waveguide), whilst channel or ridge waveguides confine the light in two dimensions (2D
waveguide). The small-size cross sections of waveguide structures offer high light intensities
produced by even very low powers; consequently, the nonlinearities or laser actions in
waveguides may be considerably improved with respect to those in bulk materials [3-6]. In
addition, the components based on waveguides usually have very small sizes, which make the
related optical chips very compact. As a result, they may be connected together with optical
fibers quite simply because the dimensions of waveguides are directly compatible with fibers
used in telecommunications, in which waveguides are often used as amplifiers, signal
switches, modulators, etc [7]. Another advantage of waveguide technology is the low costs of
the guiding devices, which makes components with diverse functions easily introduced into
the market for practical applications.
In principle, many optical materials can be used as the substrates for optical waveguide
formation, such as insulating crystals, glasses, semiconductors, and organic materials.
Glasses, with their many attractive features, e.g. large wafers, low costs, diverse
compositions, and various optical properties with excellent performances, have been
extensively used for production of passive as well as active waveguide devices [8-10].
Among this large glass family, the group that has been mostly investigated for guiding
structures so far is oxide glass, which receives most extensive applications in integrated
photonics and modern telecommunications. By certain chemical methods, such as ion
exchange [11,12], optical waveguides in both 1D (planar) and 2D (channel) configurations
can be constructed. Ion implantation is a well-known technique for modifications of chemical,
physical and optical properties of material surface [13]. In 1968, this technique was applied to
form waveguide structures in fused silica, which was the first successful attempt for ion
implantation on the waveguide fabrication [14]. Since the ion implantation is an un-
equilibrium process, which does not depended on the chemical properties of the substrates, in
principle, this technique can be applicable to most optical materials in solid states, which is
one of the most advantageous characteristics [13,15]. Up to now, waveguides have been so
far fabricated in more than 100 optical materials (including insulating crystals, glasses,
semiconductors and organic polymers) by implantation of various ions at the energies of
several hundred kilo-electron-volt (keV) up to several mega-electron-volt (MeV) or even
higher [15]. In addition, the ion technique offers precise determination of the refractive index
in selected regions with considerable modulations. One can easily control the RI profiles of
the substrates by adjusting the implantation parameters (species, energies, fluences of the
incident ions) and geometries, and further optimize waveguide properties by post-
implantation treatments (e.g. conventional or rapid thermal annealing, pulsed laser annealing,
etc.) [13,15]. Ion implantation has been used to produce waveguides in several glass families
with diverse compositions.
In this chapter, we will give a review of the ion implanted optical waveguides in glasses.
In details, in Section II, a brief introduction of fundamental models of ion implanted glass
waveguides will be given; Section III summarizes the fabrication methods for both 1D and
2D waveguides in glasses; research results of the related topics that have been obtained so far
for several glass families are overviewed in Section IV; finally, possible applications such
waveguides are raised in Section V.
Optical Waveguides Produced by Ion Implantation in Oxide Glasses 343
II. FUNDAMENTAL MODELS

There is no a unique explanation on ion-implanted waveguide formation for all materials,
because of the diversity of configuration and properties. However, there is one simple model,
i.e. so-called “barrier”-type model, which was firstly raised by Townsend in 1976 as a
prediction, which seems to be a reasonable demonstration [13,15,16]. When implantation of
light ions at energies of several MeV are performed, a majority of the damage occurs at the
end of the ion track inside the substrates through nuclear collisions, which results in a
decrease of physical density by means of volume expansion and hence a reduced refractive
index of an optical barrier. Such a barrier confines the light in a narrow layer with relatively
high refractive index between itself and the sample surface, forming an optical waveguide.
This model is applicable to some glass wafers. Figure 1 (a) shows the typical refractive index
profile of the barrier-confined waveguide, which is produced by 2.8 MeV He+ ions into a
phosphate glass at dose of 1×1016 ions/cm2.
Figure 1. Refractive index profile of (a) barrier-confined waveguide in phosphate glass and (b) “well +
barrier” restricted waveguide in silicate glass.
In some cases, the ion implantation may induce positive changes of refractive index in
the surface regions of the substrate, forming enhanced well that confines the light propagation
inside the waveguide in a non-leaky manner (referring to “enhanced well” model). The
physical mechanism may be due to the compressing effect in the implanted regions, where the
physical density may increase because of the volume reduction. In most cases, the barrier may
also happen at the end of the ion track, which bring out a “well + barrier” model. Figure 1 (b)
shows the refractive index profile of a proton implanted silicate glass waveguide. Note that
the different refractive index alternations may happen for glasses wafers belonging to one
same glass family, since the glasses are amorphous solids with no fixed compositions.
III. FABRICATION METHODS

Ion implantation is a mature technology for semiconductor production, and has been
widely used in the optical communication devices. In the research area of the ion-implanted
waveguides, accelerators are more often used because they offer high energies of specific
implanted ions at acceptable doses. The implanted ions, normally with positive charges, are
extracted out from the sources, experiencing acceleration and mass/energy selection, and
bombarded into the target materials by beam scanning technique, by which uniform
irradiation is ensured over the sample surface. This method has been adopted for waveguide
fabrication in a large number of optical materials. An improved method, a so-called focused
ion beam (FIB) implantation, could provide an ion beam with diameters ranging from several
microns to 100 nm or even much less, which provides directly selective implantation to
regions with small dimensions, forming directly-writing waveguides [17,18].
The planar waveguides can be produced in glass wafers by means of simple ion
implantation processing. The waveguide may be confined by a buried damage layer (optical
barrier), or an enhanced refractive index well, or both of them together. One may select
suitable implantation parameters to control the refractive index alternations of the glasses,
which in principle offer possibility of designable guide devices.
Figure 2. Schematic plots of ridged waveguide formation: (a) planar waveguide layer construction by
single ion implantation; and (b) selected milling of planar waveguide by ion beam etching method.
Figure 3. Selected ion implantation for channel waveguide formation with a stripe mask.
Figure 4. Focused ion beam writing for channel waveguide formation.
For 2D waveguide formation in glass by means of ion implantation, several solutions are
available. If the waveguide is confined by optical barrier, one may form ridge waveguide
structures by milling selected parts of the planar waveguide substrate, see Figure 2 (a) and
(b). If the waveguide is confined by enhanced refractive index well, single implantation may
be efficient with a stripe mask, as shown in Figure 3. Alternatively, one may create high-
refractive-index stripes by using focused ion beam writing, which avoids the lithographic
masking (Figure 4). All these techniques are applied to fabricate waveguides in oxide glasses.
Since the implantation creates color centers (optical absorption sites) by electronic energy
deposition, and destroys somehow the equilibrium of the origin structures by nuclear
collisions, which both inevitably increase the waveguide losses, therefore annealing treatment
is commonly necessary for all the ion-implanted waveguides in glasses as well crystals. Such
anneals are often useful for considerable improvement of waveguide qualities [13,15].
IV. MATERIALS
In this section, the development of ion-implanted waveguides in oxide glasses will be
summarized. The progress of non-oxide glass waveguides produced by ion implantation is
also mentioned briefly.
1) Fused Silica
Fused silica is a high purity synthetic amorphous silicon dioxide. The excellent
properties, such as near zero thermal expansion, good chemical inertness, low dielectric
constant and losses, make fused silica extensively used in the optical fields. In addition, since
its refractive index is very close to that of the commercial fibers, fused silica is also very
attractive for telecommunication applications. Optical waveguides in this material could be
fabricated by means of directly femtosecond fiber laser with high repetition rate [19]. The
earliest waveguide in fused silica was reported by Schineller et al., who used proton
implantation [14], and later work was performed on the investigation of He+, N+, O+, Si+, P+,
Ti+, or Ge+ ions [20-28]. The refractive index of the fused silica increases due to the
compaction under the influence of the nuclear and electronic damage, forming index-
enhanced wells. This non-tunneling index potential often means fine confinement of the light
propagation within the waveguides and, as a result, low propagation losses. Johnson et al.
investigated the compaction effects of Si+ ion implanted fused silica waveguides, where a
minimum loss coefficient of 0.15 dB/cm was measured for the samples with post-
implantation annealing at 500ºC [22]. In addition, Leech et al. studied the formation and the
properties of the MeV Si, P or Ge ion implanted channel waveguides in fused silica [24]. The
propagation losses of the annealed channels were only of 0.1-0.2 dB/cm at the
telecommunication wavelength of 1.3 and 1.5 μm. More recently, Wang et al. reported on
planar waveguides in fused silica by 3.0 MeV O+ ion implantation at the dose of 1×1015
ions/cm2, and attenuation coefficient of 0.14 dB/cm was obtained for the thermal treated
samples [23].
In addition, Drouard et al. performed Ti ion implantation into bulk fused silica, and both
planar and channel waveguides were fabricated. The mode fields of the waveguides may
match those of the fibers by adjusting the doses of Ti ions [20]. It is also possible to form
buried waveguides in fused silica because of the index increase induced by the compaction
effects of the implantation. Von Bibra et al. reported on buried waveguides with graded
depths fabricated by a focused 3.0 MeV proton ion beam writing in fused silica through a
tapered film varying in thickness of 5-40 μm [26]. The technique also provides possibility of
the formation of photonic lattices in fused silica. Moreover, Chen et al. studied second order
nonlinearity in a 5 MeV Ge2+-ion implanted waveguide after thermal poling treatment [27].
The results show that, with the periodical UV erasure of the second order nonlinearity, the
quasi-phase-matched SHG from 1064 to 532 nm in the silica waveguide arrays could be
realized. Recently Chao et al. reported the channel waveguide formation in similar silica hosts
by Ge2+ ion implantation combined with lithographic masking [28]. All of the above results
show that ion implantation is a practical method for effective waveguide formation in fused
silica comparable with other techniques.
2) Silicate Glasses
Silicate glasses are excellent candidates for waveguide amplifiers and lasers because of
their high solubility for rare-earth ions, such as Nd3+, Er3+, and Yb3+, as well as the chemical
robustness. By the exchange of K+/Na+ or Ag+/Na+ ions, planar and channel waveguides
could be fabricated in silicate wafers [29,30]. Ion implantation is also effective for the
modulation of refractive index of silicate glass. However, the silicate is amorphous,
particularly with no fixed composition; different silicate may have different ratio for the same
composition.
Earlier work on the ion-implanted waveguides in silicate glasses was done by
Kakarantzas et al. in 1992 [31]. From their results, standard optical barrier-type waveguides
were formed by He+ ion implantation at the doses of ~1016 ions/cm2. A considerable large
index decrease happened in the surface region, which degraded the waveguide quality [32].
Because silicate is a mixture, refractive index behaviors may be different from other silicate
glasses. We reported on optical waveguide formation in one silicate glass, which exhibited
index increase in the surface region, resulting in non-tunneling waveguide structures [33-35].
The results show that, for light ion-implanted silicate waveguides, a refractive-index
decreased barrier exists at the end of ion range, whilst in heavier-ion implanted case, no such
barrier is built up.
Waveguides with large index decrease exhibit high propagation losses. Even for annealed
samples, the loss could be ~15dB/cm, which is too high to be acceptable for practical
applications. However, low loss may be obtained for silicate waveguides confined by index-
increased wells due to the non-leaky properties. We reported on a single-mode Nd-doped
silicate waveguide, which was formed by 3.0 MeV Si+-ion implantation at low dose of ~1014
ions/cm2, with propagation loss of only ~0.6 dB/cm after annealing treatment at 218ºC for 60
min [36]. For another silicate waveguide fabricated by 6 MeV carbon ion implantation, the
losses of the first two modes were 0.42 and 0.91 dB/cm after moderate annealing treatments,
respectively [35]. Low propagation attenuation suggests the potential usage in related fields.
In addition, more recently, Wang et al. reported the first ion implanted channel waveguides in
a Nd doped silicate glass by 6 MeV O3+ at dose of 1×1015 ions/cm2 [37]. Figure 5 shows the
3D near-field intensity distribution of the quasi-TE modes of channel waveguides.
Figure 5. 3D near field intensity distribution of the quasi-TE modes of the channel waveguide by O ion
implanted into Nd-doped silicate glass: from the left to right, TE00, TE10, and TE20.
3) Phosphate Glasses
Er3+/Yb3+ co-doped phosphate glasses are extensively used in the field of

telecommunications to form erbium-doped waveguide amplifiers (EDWA). The most
common method for waveguide formation in these glasses is ion exchange. Alternatively,
implantation of light as well as heavy ions could be used to fabricate waveguides in
phosphate glasses [38-40]. The earlier work from Kakarantzas et al. showed that similar
refractive index behaviors were observed for the He+ ion implanted waveguides in their
phosphate and silicate glass samples [31]. Recently we studied the waveguide properties in an
Er3+/Yb3+ co-doped phosphate glass formed by means of He or Si ion implantation [33]. It
was found that the refractive indices in the surface and the barrier regions of the waveguide
were reduced by 0.012 and 0.021 after the He ion implantation at the dose of 5×1015 ions/cm2,
while in the Si implanted case (at the dose of 9×1014 ions/cm2), these index decrease in the
two regions were 0.018 and 0.029, respectively. Such waveguides usually have high losses
(~5dB/cm for the He- and ~20 dB/cm for the Si-implanted one) due to the relatively poor
confinement of the light inside the guide layers. Tan et al. reported ridge waveguide
formation in the He-implanted phosphate glass, which shows a designable configuration of
waveguide devices [38]. Figure 6 shows the comparison of the experimental near-field mode
intensity distribution with the calculated modal profile. Sum et al. investigated the channel
waveguides by focused proton beam writing technique [39]. In their work, 2.0 MeV proton
beams at the doses of 0.4-1.0×1015 ions/cm2 with spot size of 1.0 μm were used to write
channel waveguides in the phosphate wafers. The atomic force microscopy (AFM)
investigations of the waveguides revealed that positive rather than negative changes of the
refractive index occurred near the end of the ion range, which means the different
mechanisms between the formation of low and high dose light ion implanted waveguides
[40]. The propagation losses of the waveguides could be reduced from 3.2 to 0.8 dB/cm after
the annealing at 220 ºC.
Figure 6. Comparison of (a) the calculated modal profile and (b) the experimental near-field mode
intensity distribution in a He-implanted Er/Yb co-doped phosphate glass waveguide.
4) Tellurite Glasses
Tellurite glass is proposed as a host for broadband erbium-doped fiber amplifiers because
of their excellent optical and chemical properties. Berneschi et al. reported the channel
waveguide formation in an Er3+-doped tellurite glass by 1.5 MeV Ni+ ion implantation at
doses of 0.5~4×1016 ions/cm2, with a 75 μm-thick silicon stripe mask [41]. The Ni+ irradiation
created positive RI changes; hence the guiding structures did not require barrier confinement
of the guided light.
5) Germanate Glasses
Germanate glass is another promising candidate for integrated photonic applications.

Kakarantzas et al. studied the planar waveguide structures in lead germanate glass, in which
low-temperature (at 77K) He ion implantation at energy of 2.5 MeV and doses of 4-5×1016
ions/cm2 was performed to fabricate the waveguides [32]. The surface refractive index
increased 0.34%, forming an enhanced well, and the barrier was with 0.12% index lower than
the substrate, constructing a “well plus barrier” refractive index profile. Because of the non-
leaky mechanisms, the waveguide has a low propagation loss of 0.15-0.3 dB/cm after
annealing at 200-300 °C in air. In addition, the laser action at wavelength of ~1.9μm was
realized in a Tm3+-doped lead germanate glass waveguide formed by 2.9 MeV He ion
implantation at dose of 4×1016 ions/cm2 [42].
6) Other Glasses
Other works were performed in the topics of ion implantation into some other oxide
glasses, such as soda-lime glasses [43], filter glasses [44], etc. Barrier-confined waveguides
were obtained for those glasses after the implantations.
V. APPLICATIONS
There are several basic applications for ion-implanted oxide glass waveguides. These
waveguides can be either used as passive devices or active components, according to the
diverse properties of the glass wafers.
For all glass waveguides, it may be used as passive devices, which can be designed to
guide light propagation at diverse wavelengths. Drouard et al. used Ti ion implantation to
fabricate planar lightwave circuit in bulk silica glass [19]. The results show that the refractive
index distributions can be tailored by the doses of the Ti ions. This was helpful to fit the
guided mode of the standard single-mode fibers, or to allow a sharp radius of curvature of
bent waveguides. The waveguide devices exhibit low propagation attenuations of 0.1-0.8
dB/cm at wavelength of 1.5 μm, which can be used for practical applications in
telecommunication windows. Moreover, since the refractive index profiles of such waveguide
devices can be adjustable, more flexibility is expected for components with more functions.
In fact, other oxide glass can also be used as passive guide-wave devices, although there has
not been too much work presented in this topic so far.
The rear-earth ion doped glass waveguides can also serve as active devices, such as
waveguide laser modules and waveguide amplifiers. Shepherd et al. reported the first planar
waveguide laser in the 2μm region, based on a He+-implanted Tm3+:lead germanate glass
[40]. The energy and the dose of the He+ ions are 2.9 MeV and 4×1016 ions/cm2. Large index
increase of ~0.2 happened at the most path of the incident ions, which constructed an
enhanced well to confine the light. After the annealing treatment at 200ºC, the fluorescence of
the waveguides was quite similar to that of the original bulk materials. With suitable pump,
the threshold was expected to be as low as 10 mW. Although there were very limited
literatures reported on the laser actions in ion-implanted glass waveguides, in principle many
rear-earth ion doped glasses could be the waveguide hosts for this work.
Erbium doped waveguide amplifiers (EDWA) receive extensively applications at
telecommunication window at ~1.5 μm. By using focused proton writing, EDWA can be
produced in Er3+-doped glasses, for example, Er3+/Yb3+ co-doped phosphate glass. Liu et al.
reported buried channel waveguides in Er3+/Yb3+ co-doped phosphate glass, which serve as
EDWA [38]. The proton doses were of 1014-1015 ions/cm2, and the energy was 2 MeV. The
maximum net gain of the waveguide amplifiers at 1.534 µm wavelength was measured to be
~1.72 dB/cm with 100 mW pump power at 975 nm.
Another promising application for ion-implanted waveguides in glasses is for frequency
doubling (second harmonic generation, SHG). Chen and Chao et al. reported first order SHG
from 1064 nm to 532 nm in planar and channel waveguides produced by Ge ion implantation
into fused silica, based on a quasi-phase-matched (QPM) configuration [26]. For channel
waveguide, the second-harmonic peak power of ~1.8 μW was obtained with fundamental
peak pumping power of ~780 mW, which meant SH conversion efficiency was 6.1× 10-4
%/W·cm2.
CONCLUSION
We reviewed the progress of ion-implanted waveguides in oxide glasses, by giving basic
fabrication methods and experimental data obtained for several kinds of glasses. Applications
as both passive and active devices for these waveguides were also mentioned briefly. As it
was summarized, ion implantation is very promising for fabricating a number of attractive
optical waveguide devices in hosts of various oxide glasses. Future work may focus on the
designable waveguide structures and realization of devices with more functions.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support of Program for New Centrary
Excellent Talent in University, China (under grant No. NCET-08-0331).
REFERENCES
[1] Lifante, G. (2003.). Integrated Photonics: Fundamentals; John Wiley and Sons Ltd:
Atrium, England,
[2] Hunsperger, R. G. (2002.). Integrated Optics: Theory and Technology, Springer-
Verlag, Berlin,
[3] Kip, D. (1996). Appl. Phys. B, vol 67, 131-150.
[4] Stegeman G. I. & Seaton, C. T. (1985). J. Appl. Phys., vol 58, R57-R78.
[5] Mackenzie, J. I. (2007). IEEE J. Sel. Top. Quantum Electron., vol 13, 626-637.
[6] Murphy, E. J. (1999). Integrated Optical Circuits and Components: Design and
Applications, Marcel Dekker Inc. New York.
[7] Polman, A. (1997) J. Appl. Phys., vol 82, 1-39.
[8] Yan, Y. C., Faber, A. J., De Waal, H., Kik, P. G. & A Polman, A. (1997). Appl. Phys.
Lett., vol 71, 2922-2924.
[9] Takahashi, H., Suzuki, S., Kato, K. & Nishi, I. (1990). Electron. Lett., vol 26, 87-
88.
[10] Ramaswamy, R. V. & Srivastava R. (1988). J. Lightwave Technol., Vol 6, 984-1000.
[11] Silva, B., Melo, R. P., Falc˜o-Filho, E. L. & De Araújo, C. B. (1997). Appl. Opt., vol.
36, 5949-5950.
[12] Florea, C. & Winick, K. A. (1999). J. Lightwave Technol., vol. 17, 1593-1601.
[13] Townsend, P. D., Chandler, P. J. & Zhang, L. (1994). Optical Effects of Ion
Implantation, Cambridge University, Cambridge.
[14] Schineller, E. R., Flam, R. P. & Wilmot, D. W. (1968). J. Opt. Soc. Am. vol. 58,
1171-1176.
[15] Chen, F., Wang, X. L. & Wang, K. M. (2007). Opt. Mater., Vol. 29, 1523-1542.
[16] Townsend, P. D. (1976). Ion Implantation in Optical Materials, Institute of Physics
Conference Series Vol. 28 Chapter 3, 104.
[17] Fu, Y. & Ngoi, B. K. A. (2005). Opt. Eng., vol. 44, 128002.
[18] Jarupoonphol, W., Murakami, K., Sakata, K., Takai, M., Hosono, A. & Okuda, S.
(2003). J. Vac. Sci. Technol. B, vol. 21, 1598-1601.
[19] Shah, L., Arai, A., Eaton, S. & Herman, P. (2005). Opt. Express, 13, 1999-2006.
[20] Drouard, E., Escoubas, L., Flory, F., Tisserand, S. & Roux, L. (2004). J. Lightwave
Technol., vol. 22, 2310-2315.
[21] Webb, A. P., Houghton, A. J. & Townsend, P. D. (1976). Rad. Effects, vol. 30, 177.
[22] Johnson, C. M., Ridgway, M. C. & Leech, P. W. (1996). Appl. Phys. Lett., vol. 67, 984-
986.
[23] Wang, X. L., Wang, K. M., Fu, G., Li, S. L., Shen, D. Y., Ma, H. J. & Nie, R. (2004).
Opt. Express, vol. 12, 4675-4680.
[24] Leech, P. W., Ridgway, M. & Faith, M. (1996). IEE Proc.-Optoelectron., vol. 143,
281-286.
[25] Rao, E. V. K. & Moutonnet, D. (1975). J. Appl. Phys., vol. 46, 955-957.
[26] Von Bibra, M. L., Roberts, A. & Dods, S. D. (2000). Nucl. Instr. Meth. B vol. 168,
47-52.
[27] Chen, H. Y., Lin, C. L., Yang, Y. H., Chao, S., Niu, H. & Shih, C. T. (2005). Appl.
Phys. Lett., vol. 86, 081107.
[28] Chao, S., Chen, H. Y., Yang, Y. H., Wang, Z. W., Shih, C. T. & Niu, H. (2005).
Opt. Express, vol. 13, 7190-7196.
[29] Peters, P. M., Funk, D. S., Peskin, A. P., Veasey, D. L., Sanford, N. A., Houde-Walter,
S. N. & Hayden, J. S. (1999). Appl. Opt., vol. 38, 6879-6886.
[30] Ondráček, F., Salavcová, L., Míka, M., Lahodný, F., Slavík, R., Špirková J. & Čtyroký,
J. (2007). Opt. Mater., vol. 30, 457-461.
[31] Kakarantzas, G., Zhang, L., & Townsend, P. D. (1992). E-MRS Symp. Proc., vol. 29,
97.
[32] Kakarantzas, G., Townsend, P. D. & Wang, J. (1993). Electron. Lett., vol. 29, 489-490.
[33] Chen, F., Wang, X. L., Li, X. S., Hu, L. L., Lu, Q. M., Wang, K. M., Shi, B. R., &
Shen, D. Y. (2002). Appl. Surf. Sci., vol. 193, 92-101.
[34] Li, S. L., Wang, K. M., Chen, F., Wang, X. L., Fu, G., Lu, Q. M., Hu, L. L., Shen, D.
Y., Ma, H. J., Nie R. (2005). Surf .Coat. Technol., vol. 200, 598-601.
[35] Li, S. L., Wang, K. M., Chen, F., Wang, X. L., Fu, G., Lu, Q. M., Hu, L. L., Shen,
D. Y. & Ma, H. J. (2005). J. Phys. D: Appl. Phys., vol. 38, 2899-2903.
[36] Chen, F., Wang, K. M., Wang, X. L., Li, X. S., Lu, Q. M., Shen, D. Y., & Nie, R.
(2002). J. Appl. Phys., vol. 92, 2959-2961.
[37] Wang, L., Chen, F., Wang, X. L., Wang, K. M., Jiao, Y., Wang, L. L., Li, X. S., Lu, Q.
M., Ma, H. J. & Nie, R. (2007). J. Appl. Phys., vol. 101, 053112.
[38] Tan, Y., Chen, F., Hu, L. L., Xing, P. F., Chen, Y. X., Wang, X. L. & Wang, K. M.
(2007). J. Phys. D, vol.40, 6545-6548.
[39] Sum, T. C., Bettiol, A. A., Liu, K., Ren, M. Q., Pun, E. Y. B., Rao, S. V., Van Kan, J.
A., & Watt, F. (2005). Nucl. Instr. Meth. B, vol. 231, 394-399.
[40] Liu, K., Pun, E. Y. B., Sum, T. C., Bettiol, A. A., Van Kan, J. A. & Watt, F. (2004).
Appl. Phys. Lett., vol. 84, 684-686.
[41] Berneschi, S., Conti, G. N., Bányász, I., Watterich, A., Khanh, N. Q., Fried, M., Pászti,
F., Brenci, M., Pelli, S. & Righini, G. C. (2007). Appl. Phys. Lett., vol. 90, 121136.
[42] Shepherd, D. P., Brinck, D., Wang, J., Tropper, A. C., Hanna, D. C., Kakarantzas, G. &
Townsend, P. D. (1994). Opt. Lett., vol. 19, 954-956.
[43] Skelland, N. D. & Townsend, P. D. (1994). J. Non-Cryst. Solids, vol. 188, 243-253.
[44] Okur, I. & Townsend, P. D. (1997). Nucl. Instr. Meth. B, vol. 124, 76-80.
Chapter 12
THIN FILM PIEZOELECTRIC RESPONSE

COEFFICIENT ESTIMATION TECHNIQUES
Sharath Sriram*, Madhu Bhaskaran and Arnan Mitchell

ABSTRACT
The response of piezoelectric materials is quantified using charge and voltage
coefficients. One such coefficient is d33, which numerically describes the resulting effect
for an applied cause along the same direction. There is a lack of established techniques
for quantitative estimation of d33 for piezoelectric thin films. This initiated an
investigation into the development of such techniques, as a consequence of which two
new techniques for piezoelectric coefficient estimation, under the inverse piezoelectric
effect, have been developed. One technique capitalises on the measurement accuracy of
the nanoindenter in following thin film displacement, while the other uses a standard
atomic force microscope in contact imaging mode to estimate d33. Both techniques were
developed by rigorously testing them against standard materials and avoiding commonly
reported sources of error. Full details on the development, scope, and limitations of both
techniques are presented in this chapter.
1. INTRODUCTION
The piezoelectric effect is exhibited by the category of materials in which charge is
generated when the materials are under stress. The inverse or converse effect also exists;
whereby, the materials undergo dimensional deformation when exposed to an electric field.
The piezoelectric response is quantified using charge and voltage coefficients. This enables
comparison between the piezoelectric behaviour of different piezoelectric materials. One such
*
Corresponding author: sharath.sriram@gmail.com, sharath.sriram@rmit.edu.au
354 Sharath Sriram, Madhu Bhaskaran and Arnan Mitchell
coefficient is d33, which numerically describes the resulting effect for an applied cause along
the same direction. The subscript ‘3’ nominally corresponds to ‘z’ in the standard axis system.
Under the direct piezoelectric effect, when a force is applied along direction ‘3’ (equivalent to
stress experienced by this direction), a quantity of charge results along this direction. In this
case, d33 has the units of pC/N. Under the inverse piezoelectric effect, when a voltage (or an
electric field) is applied along direction ‘3’, the dimension of the material along this direction
changes. In this case, d33 has the units of pm/V.
A schematic representation of the direct and inverse effect, with respective d33
computations, is shown in Figure 1. Though d33 is described under these two effects, and with
two different units, the numerical value in both cases has been shown to be the same based on
Maxwell’s thermodynamic principles [1]. This law breaks down in the case of thin films, due
to the substrate clamping effect on such two-dimensional structures [2], where the length and
width dimensions are significantly greater than the thickness.
(a) (b)
Figure 1. Two-dimensional schematic of (a) the direct and (b) the inverse piezoelectric effects. The
corresponding expressions used to calculate d33 are shown.
When applying existing d33 measurement techniques to thin films, numerous challenges
were faced, considering the small variations (in charge or displacement) expected from the
thin films. The most commonly reported approach, with an ever increasing number of
publications each year, is piezoresponse force microscopy (PFM) [3]. PFM uses atomic force
microscopes (AFM) with additional options: to apply an electric voltage through the tip,
which is generally metal-coated to make it conductive; to access signals from the four-
quadrant laser detector; electronics to process these signals; and software configuration to
map these results as images. PFM excels in relative mapping to study piezoelectric response
variations, but is only semi-quantitative in d33 measurements. Approaches for quantitative
measurement of d33 have been demonstrated using nanoindenters by Koval et al. [4], under
the direct piezoelectric effect.
Using strontium-doped lead zirconate titanate (PSZT) and reference thin film materials –
amorphous silicon dioxide (non piezoelectric) and aluminium nitride (standard piezoelectric)
– two new and relatively simple techniques for piezoelectric response estimation have been
developed. The techniques aimed to enable piezoelectric response estimation using standard
equipment, such as the nanoindenter and atomic force microscope, without the need for
extensive modifications or additional options.
Thin Film Piezoelectric Response Coefficient Estimation Techniques 355
The capabilities of the techniques described in this chapter have been demonstrated using
PSZT thin films deposited on gold-coated silicon substrates. PSZT is a doped variant of the
popular piezoelectric lead zirconate titanate (PZT). Deposition details for the PSZT thin films
can be found in [5-7].
2. SAMPLE PREPARATION FOR d33 MEASUREMENTS

The need to use standard equipment necessitates a couple of additional steps in
preparation of samples for d33 measurements. To estimate the piezoelectric response of the
deposited thin films in terms of d33, samples were fabricated to form a sandwich electrode
structure. The following sample preparation description applies for any film to be studied, but
has been described with reference to PSZT.
During PSZT deposition, a portion of the bottom metal electrode (gold or platinum) was
covered with a piece of silicon to prevent sputtering of PSZT on the selected region of the
sample; this was done to create electrical access to the bottom electrode. This form of shadow
masking was performed in order to study the film in its as-deposited form, without exposing
it to wet chemical etchants. The top metal electrode (600 nm aluminium) was deposited at
room temperature by electron beam evaporation. In this case also, evaporation was carried out
through a shadow mask with small openings to define the top electrode. On all samples, two
or more electrodes were defined to study the uniformity in response of the thin films.
Individual samples were firmly mounted onto glass slides using non-conductive double-sided
tape. The glass slides used had gold islands defined by photolithographic patterning, with the
gold having been deposited by electron beam evaporation (at room temperature) to a
thickness of 150 nm, using a 15 nm titanium adhesion layer. The top and bottom electrodes
were wire-bonded to individual gold islands on the glass slide using gold ribbon (76 μm wide
and 12.5 μm thick). Fine gauge wire was bonded using silver epoxy to these gold islands, to
serve as connectors to external devices, such as the function generator. All connections were
tested to ensure that no electrical shorts were present. A schematic representation of such a
sample is shown in Figure 2.
Figure 2. A schematic representation of samples prepared for piezoelectric response measurements.

(Not to scale.)
3. ESTIMATION OF d33 USING A NANOINDENTER

Published work on the measurement of the piezoelectric response of thin films using a
nanoindenter [8-11] are generally based on the direct piezoelectric effect, and use the
nanoindenter to apply stress on the thin films. While these techniques are expected to give the
same numerical value for d33, they require an in-depth understanding of indenter mechanics,
dynamics, and principles. Whereas, the technique presented below uses the indenter only as a
means of accurately measuring displacement. Related work by Rar et al. [11] has been carried
out for piezoelectric substrates using the inverse piezoelectric effect, and highlights that the
indentation approach leads to better quantification of d33, compared to the popular
piezoresponse force microscopy (PFM) technique.
3.1. Experimental Details
An UMIS 2000 nanoindenter (CSIRO, Australia) was used to carry out the
measurements, given its precision for displacement measurements. A blunt metallic tip was
used to prevent deformations to the thin film being tested, and contact of the tip to the film
was made with a force of 1 mN. The indenter tip was brought into contact with the surface of
the top electrode or the piezoelectric thin film. The piezoelectric thin film studied was
strontium-doped lead zirconate titanate (PSZT) deposited for a duration of 4 hours at 300 ºC
on gold-coated silicon substrates (PSZT on Au/Ti/Si with a PSZT thickness of ~1900 nm). A
schematic of this arrangement is shown in Figure 3. The top electrode was connected to
ground of the function generator. This was done to ensure that no electrostatic interaction
between the tip and the sample affected the measured displacements. The indenter software
(IBIS, version 1, 2004 from Fischer-Cripps Laboratories Pty. Ltd.) was used to follow the
displacement of the surface of the films over time. Most measurements were made by placing
the tip on the surface of the aluminium top electrode, to ensure the response to the entire
electric field is measured. Low frequency (ranging from 25-500 mHz) electrical signals with
peak-to-peak voltages of 10 V and 32 V were used, due to the limitations in the sampling rate
of the software. Mapping options in the software were used to precisely position the
nanoindenter tip and perform line scans and area scans to study piezoelectric response
variations.
3.2. Results and Discussion
Initial piezoelectric response measurements were done on the surface of the aluminium
top electrode, in order to study the response of the piezoelectric thin film to the entire applied
electric field. Applied peak-to-peak voltages of 10 V and 32 V at one particular point,
resulted in peak-to-peak film thickness changes of 4.34 nm and 13.97 nm respectively. These
relate to d33 values of 434 pm/V and 437 pm/V respectively, indicating a linear and consistent
result. These are typical of results obtained for d33 measurements done at different points.
Mapping of the thin film piezoelectric response over an area of 80 μm by 80 μm on the
aluminium top electrode showed a region of high piezoelectric response, with the varying
response around that region (Figure 4). This is due to the variations in the alignment of
domains in the region being studied. A minimum d33 value of 458 pm/V and a maximum d33
value of 608 pm/V were obtained, with an average value of 545 pm/V.
Figure 3. Schematic of the arrangement used to measure the piezoelectric response of the PSZT thin
films. (Not to scale.)
Figure 4. Results from mapping the piezoelectric response of a PSZT thin film over an 80 μm × 80 μm
area (on the aluminium electrode surface) for an applied peak-to-peak voltage of 32 V. (Reprinted with
The piezoelectric response was mapped along a line of length 1250 μm, at intervals of
250 μm, and the resulting data is shown in Figure 5. The response is high and reasonably
consistent on the top of the electrode, reaches a maximum closer to the edge of the electrode,
and drops to a lower value when measured directly on the thin film. These results follow the
expected variations in electric field strength. The field is expected to be at its maximum
between the top and bottom electrodes, with its strength decreasing with distance from the
electrode area.
Figure 5. Results from a linear map of the piezoelectric response for an applied peak-to-peak voltage of
32 V, to study variations in response with distance from the top aluminium electrode. (Reprinted with
In order to validate the measurement technique, a sample with the same overall
arrangement, but with amorphous silicon dioxide instead of PSZT was tested. The resulting
response under an applied electric potential for silicon dioxide was a null response, similar to
those obtained when no electric potential was applied to the piezoelectric PSZT thin film.
This supports the conclusion that the film thickness variations measured were a direct
consequence of the piezoelectric behaviour. The range of d33 values was also verified using
the technique presented in the next section. Discussion on the physical processes resulting in
the measured piezoelectric response coefficient and comparison of these values to those in
published literature is presented in Ref. [12].
4. ESTIMATION OF d33 USING AN ATOMIC FORCE MICROSCOPE

This section describes a technique for the quantification of d33 using atomic force
microscopes (AFM), without the need for modifications; to support and simplify other
reported techniques, such as piezoresponse force microscopy (PFM) [13] and piezo-nano-
indentation (PNI) [4]. The AFM is used under the standard contact mode imaging settings, as
is done for most surface topography scans. Samples are prepared (as described in Section 2)
to study the piezoelectric response of the thin films under the inverse piezoelectric effect. The
technique uses the scan result at a given point to extract the peak-to-peak displacement of the
thin film for a specific applied peak-to-peak voltage. This results in a quantified value for d33
in pm/V.
4.1. Experimental Details
PSZT thin films deposited on Au/Ti/SiO2/Si substrates at 300 ºC for 4 hours were used
during the development of this technique. These films were subsequently annealed in the
sputtering chamber, during the cooling down process, at 250 ºC for 1 hour. A Digital
Instruments Dimension 3100 atomic force microscope with a Nanoscope IIIa controller was
used for these experiments. The instrument was calibrated for ‘z’ displacements before use,
using a 100 nm step height standard. The system was used in the contact mode imaging
arrangement, with non-conductive silicon nitride tips.
Figure 6 is a schematic representation of the testing arrangement used (similar to Figure
3). The AFM tip is placed at a chosen point on the surface of the top electrode. An electric
field is applied between the top and bottom electrodes with the top electrode connected to the
ground of the function generator to prevent charging up of the non-conductive tip surface
(this could cause attractive or repulsive forces, and so, could introduce errors). The electric
field in Figure 6 is represented using two components: Vp represents the peak value or
amplitude of the signal and is half the peak-to-peak voltage value, and f(t) represents the
signal type used and corresponds to the sine function. Frequencies ranging from 500 mHz to
10 Hz were experimented with, limited by the sampling rate and tip mechanics of the system.
All scan results incorporated in this article were carried out with resolution settings of
512 pixels/line and 512 lines in an image. The scan area value is set to zero and the line scan
frequency was set to 1 Hz. The electric field was applied using an Agilent Technologies
33220A function generator without a 50 Ω termination. In order to accurately evaluate the
electric signal applied across the film, an oscilloscope was connected in parallel during all
measurements.
4.2. Results and Discussion
This section describes the results obtained using the proposed piezoelectric response
measurement technique, and discusses the manner in which the electro-mechanical coefficient
d33 can be obtained from the scan results.
Figure 7 shows the scan results obtained with a scan rate of 1 Hz and for varying applied
signal frequencies, from 1 to 4 Hz. Considering that the AFM tip is not scanning over any
region (it is at one chosen point on the top electrode), the displacements observed in Figure 7
correspond to variations in thickness over time at a point chosen on the sample. This is
different from normal AFM scan results in which displacements are mapped over a chosen
area. Figure 7 [especially Figure 7(b)] clearly depicts that every scan line has sinusoidal
variations, with the number of cycles directly corresponding to the ratio of the driving
frequency (1-4 Hz) and the scan frequency (1 Hz); each sinusoid corresponds to a set of bands
(one dark and one light) in the scan result. The bands in the scan result shift for each line,
with an overall diagonal appearance; this is a result of delays introduced by the AFM
feedback system after each scan line.
Figure 6. Schematic of the arrangement used for piezoelectric response measurements using the AFM.
(Not to scale. Reprinted with permission from Ref. [14].)
This result demonstrates that the thin film displacements occur at the applied electrical
signal frequency, as expected. Samples with amorphous silicon dioxide in place of PSZT
were tested under the same arrangement and exhibited no displacement variations (null
response – only background noise in scan results). This clearly indicates that the
displacements being measured are due to the piezoelectric behaviour of the PSZT thin films.
Figure 8 was used to verify the technique further. During the initial portion of the scan,
no electric field was applied across the thin film. The result is just background noise in the
scan result. On applying an electric field with a frequency of 1 Hz, the film starts to respond
instantly, reflected by approximately one sinusoidal cycle in the middle portion of the scan.
The number of sinusoids increases to two in the last section of the scan, when the frequency is
doubled.
Figure 9 highlights the ability of the proposed technique in identifying displacement
variations with applied voltage at a constant driving frequency; larger displacements increase
contrast in the image while the number of bands in the image remain constant. During the
course of the scan, the amplitude of the driving voltage was changed. From an initial peak-to-
peak amplitude of 12 V, it was increased to 16 V, with the final scan region being recorded at
20 V. The corresponding peak-to-peak displacements are 1.3 nm, 2.7 nm, and 4.1 nm.
(a)
(b)
Figure 7. Scan result when the frequency of the applied sinusoidal electric field was varied from 1 to 4
Hz, in steps of 1 Hz, with a fixed peak-to-peak amplitude of 20 V. The dark and bright bands seen in (a)
are due to the applied sinusoidal voltage’s positive and negative cycles, respectively. The three-
dimensional representation of (a) shown in (b) highlights the sinusoidal nature of the displacement
variations. (Reprinted with permission from Ref. [14].)
From these scan results, the piezoelectric voltage coefficient d33 can be estimated in
pm/V, as the ratio of the peak-to-peak displacement (in pm) to the applied peak-to-peak
voltage (in V). For example, the displacement variations in Figure 9 of 1.3 nm for 12 V, 2.7
nm for 16 V, and 4.1 nm for 20 V, correspond to d33 values of approximately 109, 169, and
205 pm/V, respectively (these values are lesser than those for films discussed in Section 3.2
due to lesser degree of preferential orientation, as described in Refs. [5,12,14]).
Figure 8. Scan result where three sectors of the image correspond to time periods of no applied signal
and to signal frequencies of 1 Hz and 2 Hz for an applied sinusoidal electric field of fixed peak-to-peak
amplitude of 20 V. (Reprinted with permission from Ref. [14].)
Figure 9. Scan result obtained under an applied 4 Hz sinusoidal voltage. The peak-to-peak amplitude of
the voltage was increased at regular intervals. The applied voltage and resulting displacement are
shown as peak-to-peak values. (Reprinted with permission from Ref. [14].)
5. CONCLUSION
This chapter presents two new and simplified techniques for the estimation of the
piezoelectric charge coefficient d33. The following key results have been obtained:
A. A technique has been developed to estimate thin film d33 using a nanoindenter under
the inverse piezoelectric effect.
B. The capabilities of using the nanoindenter to map piezoelectric response data over an
area and along a line, to study influence of electrodes or domain/grain distribution,
are presented.
C. An atomic force microscope in the contact imaging mode has been used to estimate
thin film d33. By following the displacement at one point, a representation of thin
film variation as a function of frequency and amplitude of the applied electric field
can be studied.
The two techniques described in this chapter enable quantification of piezoelectric

response by enabling accurate estimation of the piezoelectric response coefficient d33 using
simplified techniques and easily accessible instrumentation.
REFERENCES
[1] Cady, W. G. (1964). Piezoelectricity; Dover Publications: New York, Vol. 1.
[2] Lefki, K. & Dormans, G. J. M. (1994). J. Appl. Phys., 76, 1764-1767.
[3] Gruverman, A. & Kalinin, S. V. (2006). J. Mater. Sci., 41, 107-116.
[4] Koval, V., Reece, M. J. & Bushby, A. J. (2005). J. Appl. Phys., 97, 074301.
[5] Bhaskaran, M., Sriram, S., Short, K. T., Mitchell, D. R. G. & Holland, A. S. (2008).
Thin Solid Films, 516, 8101-8105.
[6] Sriram, S., Bhaskaran, M., Du Plessis, J., Short, K. T., Sivan, V. P. & Holland, A. S.
(2009). Micron, 40, 104-108.
[7] Sriram, S., Bhaskaran, M., Mitchell, A., Mitchell, D. R. G. & Kostovski, G. (2009).
Nanoscale Res. Lett., 4, 29-33.
[8] Karapetian, E., Kachanov, M. & Kalinin, S. V. (2005). Philos. Mag., 85, 1017-1051.
[9] Sridhar, S., Giannakopoulos, A. E. & Suresh, S. (2000). J. Appl. Phys., 87, 8451-8456.
[10] Algueróa, M., Calzada, M. L., Bushby, A. J. & Reece, M. J. (2004). Appl. Phys. Lett.,
85, 2023-2025.
[11] Rar, A., Pharr, G. M., Oliver, W. C., Karapetian, E. & Kalinin, S. V. (2006). J. Mater.
Res., 21, 552-556.
[12] Sriram, S., Bhaskaran, M., Holland, A. S., Short, K. T. & Latella, B. A. (2007). J. Appl.
Phys., 101, 104910.
[13] Kalinin, S. V. & Bonnell, D. A. (2004). In Nanoscale Phenomena in Ferroelectric Thin
Films; Hong, S.; Ed.; Kluwer Academic: Boston, MA, pp 183-217.
[14] Sriram, S., Bhaskaran, M., Short, K. T., Matthews, G. I. & Holland, A. S. (2009).
Micron, 40, 109-113.
Chapter 13
PLASMA TECHNOLOGY: AN ALTERNATIVE TO

CONVENTIONAL CHEMICAL PROCESSES FOR
HYDROGEN PRODUCTION
María Dolores Calzada*

Grupo de Espectroscopia de Plasmas, Edificio Einstein (C-2), planta baja, Campus de
Rabanales, Universidad de Córdoba, 14071 Córdoba (Spain)
ABSTRACT
There is currently a strong debate on the need to look for new alternative energy
sources to replace fossil fuels (particularly oil), due to atmospheric pollution derived
from their use, probable reserve exhaustion and productive countries’ excessive
dependence on future political events. Therefore, many countries are currently giving an
incentive to research aimed at the development of new energy sources, among which
hydrogen —and its use by means of fuel-cells— is to be included. Within the so-called
hydrogen economy, hydrogen production —from different raw materials such as
hydrocarbons, alcohols and some others— becomes an important aspect, since hydrogen
is not a natural product but is obtained by means of water vapour. However, such process
leads to the production of high amounts of CO2. The use of plasma technology allows
generating hydrogen by decomposing organic compounds (alcohols and methane), so
plasma acts as a reactive environment; this way, biogas —which can be easily found in
city landfills— can also be used for hydrogen obtention. Thus, the use of plasma with this
purpose arises as an alternative process not only to reforming of hydrocarbons and other
substances with water vapour but also to other purely chemical processes, avoiding or
minimizing CO2 emissions (involving greenhouse effects). A subproduct —obtained with
H2 and which also deserves special attention— is solid carbon, which has important
added value, since it may work as a base element for chemical industry and other
technological fields. Given the foregoing, research in the use of plasma technology for
hydrogen production is one of the research lines within Applied Physics with higher
applicability potential within the energy sector.
*
Corresponding author: Email: md.calzada@uco.es
366 María Dolores Calzada
INTRODUCTION
The energy sector itself constitutes an important part of the economic activity in any
given country, is an indispensable service for its citizens’ everyday life and contributes an
undeniable value to the remainder sectors in such economy. Studies show that 80 % of the
total energy generated and consumed by the human being comes from fossil fuels, whose
combustion involves the formation of CO2, which is considered a non-environmentally-
friendly and greenhouse-effect gas.
The current rate of CO2 emissions from different human activities is around 25 % of the
natural rate of atmospheric CO2-recycling by photosynthetic processes. Although physical
and chemical exchanges with oceans should be included into the global carbon-CO2 cycle, the
previous figure shows the existence of a strong artificial perturbation in a natural cycle whose
atmospheric component plays a relevant role in greenhouse effect and, therefore, in global
warming. Figures of molar CO2 concentration in air have increased with a progressively more
marked tendency since the beginning of the 20th century, currently exceeding 380 molar ppm.
On the other hand, studies completed in Norwegian fiords have shown up that atmospheric
CO2 levels affect oceans, given that the absorption of carbon anhydride increases water’s
acidity levels, thus producing an excessive increase of plankton and altering equilibrium in
oceans. All the previous has given rise to the establishment of limits for CO2 emissions
collected in different protocols such as that of Kyoto. Furthermore, additional problems
related to the use of fossil fuels for energy generation should also be taken into account, such
as forecasts on the exhaustion of world oil reserves within several decades and instability in
domestic policy and international relations in the main oil-producer countries.
Consequently, European Union member countries and other countries such as United
States, Japan, China and Brazil have impelled R&D policies aimed at promoting the
development of the so-called alternative energies such as solar and wind energies, the use of
biomass and hydrogen as fuels, thus contributing to guarantee energy supply, reduce their
dependence from oil-producer countries and contribute to diminish —drastically in some
cases— CO2 emissions into the atmosphere. Particularly, hydrogen is considered as an
alternative fuel to fossil ones, especially within the automotive sector, being extremely clean,
since the chemical energy accumulated in H-H bonds is freed when combined with oxygen,
thus producing only water as a sub-product of the reaction which takes place in the so-called
fuel cells or in internal combustion engines; this energy is subsequently transformed into
electric current to feed vehicle engines. Thus, hydrogen beings to look like an efficient, non-
pollutant and abundant energy source at a reasonable price in medium to long term. However,
hydrogen is not a natural resource and must be obtained from other raw materials (water,
biomass, fossil fuels, etc.), resulting in some cases in a less-clean-than-expected production
process.
HYDROGEN ECONOMY
The so-called hydrogen economy tackled different issues such as production, storage
(safety) and use of hydrogen as a fuel. The Issues regarding its storage (safety) and use in the
automotive industry are those in which research has reached the most advanced stages so far.
Plasma Technology: An Alternative to Conventional Chemical Processes... 367
Thus, hydrogen can be stored at different pressure values both as a liquid and a gas and also
in the shape of hydrides. Regarding its use, energy can be obtained from internal combustion
engines or by means of the so-called fuel cells.
However, one of the most important issues still demanding further in-depth research is
the production of enough quantities of hydrogen to supply the current demand. Therefore,
great part of the efforts in this hydrogen economy should be aimed at finding the most
economic and simple method for hydrogen production. For hydrogen generation at industrial
scale, water electrolysis and hydrocarbons reforming with water steam are regularly used.
Water decomposition into hydrogen and oxygen is produced by the process of
electrolysis. This decomposition takes place within an electrolytic cell containing two
electrodes separated by a gas-proof diaphragm. Oxygen is produced in the anode and
hydrogen in the cathode according to the following reactions:
1
2OH → H 2 O + O 2 + e − (anode)
2
2H 2 O + 2e → H 2 + 2OH − (cathode)
−
(1)
1
H 2 O → H + O 2 (total reaction)
2
This process consumes a high quantity of electric energy, given that it is necessary to
apply 1 amp for 24 hours to decompose a mole of water (18 grams). Thus, the production of
1 amp per second shall be 0.22 cm-3 of H2 and 0.11 cm-3 of O2, which entails a high cost in
electric energy. For this reason, electrolysis is usually carried out by means of photovoltaic
solar energy or wind energy.
Hydrogen production from hydrocarbons reforming with water steam takes place in two
stages. The first of them according to the following reaction:
CmHn + mH2O → mCO + (m+n/2)H2 (2)
This reaction —independently from the used hydrocarbon— is endothermic and its
enthalpy (ΔH) depends on the heat of formation of the hydrocarbon in question. Thus, for
instance, such value is 206 and 260 kJ/mole in the cases of methane and ethane, respectively.
The second stage comprises the following reaction:
mCO + mH2O → mCO2 + mH2 (3)
In this case, the reaction is exothermic (ΔH = -m × 41.1 kJ/mole). Thus, it is observed
that hydrogen production from hydrocarbons reforming with water steam entails massive
production of CO2 in two steps, the first of them being CO production, also considered as a
greenhouse effect gas. This means that the level of emissions is not reduced, but may even
increase. For this reason, efforts have been recently aimed at finding methods for hydrogen
production which enable reducing or even eliminating CO2 emissions.
PLASMA: REACTIVE MEDIUM TO OBTAIN HYDROGEN

A new alternative option to that of hydrocarbons reforming with water steam is
transformation of hydrocarbons into hydrogen and carbon through controlled pyrolytic
processes according to the following reaction:
CnH2m → nC + mH2 (3)
This is also an endothermic reaction. Thus, for instance, absorbed energy in the case of
methane is 75 kJ/mole, 84 kJ/mole for ethane, 100 kJ/mole for propane and 125 kJ/mole for
butane. The main inconvenient of these reactions is that they require very high temperatures,
over 500 K, while it has been ascertained that, for instance, at temperatures over 1000 K
practically all methane molecules enclosed in a recipient become split.
A medium which can be considered as one of the most appropriate ones for Reaction (3)
to take place is plasma. This statement is based on the fact that plasma is a partially-ionized
gas in which electrons, ions and atoms coexist with enough energy to break up, by means of
collisions, the bonds of substances introduced into it. In plasmas generated at reduced
pressure (below atmospheric pressure), electron energy may reach 20000 K vs. 300 to 400 K
found in heavy particles (atoms, ions and radicals), while in plasmas at atmospheric pressure,
electron energy may be ≥ 7000 K and that of heavy particles ≥ 1200 K. On the other hand,
this temperature difference between plasma particles is characteristic of Non-Local
Thermodynamic Equilibrium (LTE) plasmas, which allows varying population density and
temperature of its species by modifying the operation conditions of the discharge (applied
power, gas kind and flow, etc.), influencing the reactions which take place in the discharge, in
order to make them follow the desired direction. Therefore, Non-Local Thermodynamic
Equilibrium plasmas may be considered as one of the most appropriate methods to produce
certain reactions which cannot take place or are not effective enough through conventional
methods. Thus, if a hydrocarbon flowing is conducted into plasma generated with an inert
gas, the decomposition of such hydrocarbon gives rise to the production of H2 and C with no
CO2 emissions. Such decomposition may be constituted by several stages depending on the
operation conditions under which plasma is generated and the kind of hydrocarbon; not only
electrons can take part in such reactions but also radicals themselves formed after a first stage
of decomposition. However, not only hydrocarbons can be considered as a raw material from
which hydrogen is to be obtained by using plasma. Other substances such as alcohols
(methanol, ethanol and bioethanol), pure methane, natural gas and even biogas may be used
with this purpose.
The use of plasma as a reactive medium to decompose different raw materials with the
aim of obtaining hydrogen involves a series of advantages, not only from an environmental
viewpoint but also from an operation viewpoint. Among its advantages it is to stand out the
wide interval of frequencies, from some MHz to GHz, at which plasmogen gases can be
generated. These gases can be both pure and mixed, and working pressure ranges between
some Torr to some atmospheres. Furthermore, there are several energy applicators. All the
previous allows studying plasmas generated under different operation conditions to find those
which fit best the kind of raw material from which hydrogen is extracted. However, the main
weakness of plasmas is the fact that, in most cases, it is necessary to use gas or easily-
vaporizable raw materials.
The following figures are an example of the decomposition of methanol by microwave
plasma generated in Ar at atmospheric pressure into which a flowing of the afore-mentioned
alcohol has been introduced. The capacity of this plasma to break up methanol molecules was
observed in collected spectra corresponding to the characteristic radiation emission of the
plasma used in these experiments [1]. Figure 1 shows the emission spectrum of plasma
generated in pure Ar, showing the lines of atomic Ar and some molecular bands
corresponding to impurities of different species (at level of traces) in the gas used as
plasmogen gas. Figure 2 shows not only Ar lines but also the molecular bands of radicals
from the decomposition of the methanol molecules introduced into the plasma. Both the
increase of the line Hα of hydrogen and the appearance of the band C2 involve the formation
of molecular hydrogen and solid carbon at plasma end. This shows that, apart from hydrogen,
other sub-products —which contribute value added to the process— may be obtained. Among
these sub-products it is to stand out carbon, with applications within different scientific and
technologic fields such as Mechanics, Electronics, Medicine, etc.
4
1.5x10
4
1.0x10
I (u.a)
3
5.0x10
NH Hα
OH
0.0
300 400 500 600 700
λ (nm)
Figure 1. Spectrum of pure-Ar plasma generated at atmospheric pressure [1]
4
5x10
4
4x10 CN
4
3x10
Hα
I (u.a)
4
2x10 OH and CH
CN and C2 C2
4
1x10
NH C2 C2
0
300 400 500 600 700
λ (nm)
Figure 2. Spectrum of Ar plasma generated at atmospheric pressure when a methanol flowing is

introduced into the plasma [1]
Studies carried out in this line have shown that, at plasma end, the percentage of species
recombination which gives rise to the primitive substance introduced into the plasma is very
small, always obtaining H2 and C, as well as other sub-products depending on the raw
material used [1-12]. Figures 3, 4 and 5 are an example of mass spectra at the discharge exit
after methanol was decomposed by plasma at atmospheric pressure [1]. The first twenty
spectra were recorded while the plasma was switched-on and the next twenty ones with the
plasma switched-off. In Figure 3, one observes molecular hydrogen production from
methanol decomposition by plasma. Spectra corresponding to Figure 4 show the formation of
acetylene (26 amu), which is obtained from the association of CH radicals. It results in the
possibility of obtaining other sub-products from alcohol decomposition.
I (a.u.)
Molecular Hydrogen
Atomic Hydrogen
Plasma off
Plasma on
Mass
Figure 3. Mass spectra: production of molecular hydrogen [1]
Water
t
I (a.u.)
Methanol
Acetylene Oxygen
Nitrogen
Plasma off
Plasma on
Mass
Figure 4. Mass spectra: production of acetylene [1]
With respect to CO2 (44 a.m.u.), Figure 5 shows the same amount when the plasma is
switched-on and switched-off, which implies non-formation of CO2 by plasma. This
substance comes mainly from argon, alcohol impurities and the air which is in contact with
the MS fiber. This is the reason why the use of the discharge at atmospheric pressure not only
simplifies its implementation on an industrial scale but also allows it to be considered a non-
contaminating hydrogen molecular source.
Argon
I (a.u.)
CO2
Plasma off
Plasma on
Mass (amu)
Figure 5. Mass spectra: CO2 due to air in contact with the MS fiber [1]
This result also shows another advantage in the use of plasma as a medium to decompose
substances containing hydrogen: In case of using natural gas or biogas, whose content is not
100 % methane, the molecules of the remainder component substances also undergo
decomposition, so less pollutant sub-products are obtained. This behaviour of plasma has
already been observed in the purification of Kr and Xe gases from air distillation and
destruction of SF6 [13-14], a gas generated during the manufacture of microelectric
components and circuits.
FINAL COMMENT
The use of hydrogen as a fuel in the automotive sector is currently a priority research
objective for different automobile brands —General Motors, Honda, Mazda, Nissan, Renault,
Volkswagen, Hyundai and Toyota among them. It is in the issue of hydrogen production
where plasma may play a decisive role as a reactive medium to produce the decomposition of
different substances containing hydrogen, without the emission of greenhouse effect gases as
sub-products of the reactions taking place in the plasma.
Finally, it should also be pointed out that the research field of hydrogen production by
plasma is a multidisciplinary and still wide-open field which needs synergies from different
scientific and technologic fields, among which Physics plays an essential role, since it is the
field which may contribute the definitive understanding of the processes taking place in
plasma and their optimization. This way, with the participation of different sectors, hydrogen
shall really become the fuel of a not very far future.
REFERENCES
[1] Jiménez, M; Yubero, C; Calzada, MD. J. Phys D: Appl. Phys. 2008, vol. 41, 155201
(6pp).
[2] Deluga, GA; Salge, JR; Schmidt, LD. Verykios, X.E. Science, 2004, vol. 303, 993
[3] Bromberg, L; Cohn, DR; Rabinovich, A; Alexeev, N; Samokhin, A; Ramprasad, R;
Tamhankar, S. Int. J. Hydrogen Energ., 2000, vol. 25, 1157-1161.
[4] Bromberg, L; Cohn, DR; Rabinovich, A; Alexeev, N. Int. J. Hydrogen Energ. 1999,
vol. 24, 1131-1137.
[5] Bromberg, L; Cohn, DR; Rabinovich, A; Heywwod, J. Int. J. Hydrogen Energ. 2001,
vol. 26, 1115-1121.
[6] Lutz, AE; Bradshaw, RW; Bromberg, L.; Rabinovich, A. Int. J. Hydrogen Energ.,
2004, vol. 29, 1401-1424.
[7] Deminsky, M; Jivotov, V; Potapkin, B; Rusanov, V. Pure Appl. Chem., 2002, vol. 74,
413-418.
[8] Sarmiento, B; Brey, JJ; Vera, IG; González-Elipe, AR; Cotrino, J; Rico, VJ. J. Power
Sources, 2007, vol. 169, 140-143.
[9] Li, H; Zou, J; Zhang, Y; Liu, C. Chem. Lett., 2004, vol. 33, 744-745.
[10] Aubry, O; Met, C; Khacef, A; Cormier, JM. Chem. Eng. J., 2005, vol. 106, 241-247.
[11] Tanabe, S; Matsuguma, H; Okitsu, K; Matsumoto, H. Chem. Lett., 2000, vol. 10,
1116-1117.
[12] Yanguas-Gil, A; Hueso, JL.; Cotrino, J; Caballero, A; Gonzalez-Elipe, AR. Appl. Phys.
Lett. 2004, vol. 85, 4004-4006.
[13] Nantel-Valiquette, M; Kabouzi,Y; Castaños-Martinez, E; Makasheva, K; Moisan, M;
Rostaing, JC. Pure Appl. Chem., 2006, vol. 78, 1173-1185.
[14] Kabouzi, Y; Moisan, M; Rostaing, JC; Trassy, C; Guerin, D; Keroack, D; Zakrzewski,
Z. J. Appl. Phys., 2003, vol. 93, 9483-9496.
[15] Rostaing, JC; Bryselbout, F; Moisan, M; Parent, JC. C. R. Acad. Sci. Paris, 2000, vol.1
(Série IV), 99-105.
Chapter 14
TARGET-PLASMA-FILM INTERACTIONS IN HIGH

POWER PULSED MAGNETRON SPUTTERING
(HPPMS)
K. Sarakinos*
Materials Chemistry, RWTH Aachen University, 52074, Aachen, Germany
ABSTRACT
Growth of films by plasma-assisted physical vapor deposition (PVD) techniques
provides means for tailoring their properties and improving their functionality in
technological applications. State of the art plasma-assisted PVD techniques, like direct
current magnetron sputtering (dcMS), suffer from low degree of ionization and thus, low
ion-to-neutral ratios in the flux of the deposited material. The implementation of external
sources for the enhancement of the ionization is in many cases technically complicated
and increases the end-product cost. High power pulsed magnetron sputtering (HPPMS) is
a novel sputtering technique that elegantly enables the conversion of conventional
sputtering source into an ion source. By applying the power in unipolar pulses of low
frequency (<2 kHz) and low duty cycle (<10%) high values of peak target current density
up to ~ 5 Acm-2 are achieved. This leads to the generation of ultra dense plasmas with
electron densities of ~1019 m-3. These values are up to four orders of magnitude higher
than those obtained by dcMS and result in a high degree of ionization of the sputtered
material (up to ~90%). Moreover, due to the low duty cycles, the average target current
densities are maintained at values of ~0.5 Acm-2 which are comparable to those in dcMS
processes. Therefore, HPPMS allows for highly ionized deposition fluxes without
excessive thermal load at the target.
In the current short communication the target-plasma interactions and their effect on
the deposition rate in non-reactive and reactive HPPMS are discussed. Furthermore, the
interactions of the plasma species in HPPMS with the growing film and their implications
on the microstructure and the phase formation of metal nitride and oxide films are
demonstrated.
*
Corresponding author: sarakinos@mch.rwth-aachen.de
374 K. Sarakinos
1. INTRODUCTION
Thin films are used in diverse technological applications, such as surface protection and
decoration, data storage, optical and microelectronic devices. The increasing demand for new
functional films has been a strong incentive for intensive research towards understanding
fundamentals and technical aspects of their growth and developing new deposition
techniques. Among the various methods employed for film growth, Physical Vapor
Deposition (PVD) techniques, such as magnetron sputtering, are widely used [1]. Magnetron
sputtering is a plasma-based technique, in which inert gas (commonly Ar) atoms are ionized
and accelerated by a negative potential towards a solid material source (target) causing the
ejection (sputtering) of atoms which condensate on a surface (substrate) and form a film [1,
2]. The film growth is a complex non-equilibrium process in which both thermodynamics and
kinetics allow for tailoring of the film properties [2, 3]. The latter can be achieved by (i) the
growth temperature [2, 4] and (ii) bombardment by energetic species [5-9], which can be
employed either in conjunction with the growth temperature or as an alternative when heating
of the substrate is not possible. In general, there are two kinds of bombarding species, i.e.
neutrals and ions. Ions are favored over the neutrals, since their energy, flux and trajectory
can be easily controlled by using external electric and/or magnetic fields [5] providing thus
well defined conditions of energetic bombardment. In these terms conventional magnetron
sputtering techniques, such as dc magnetron sputtering (dcMS), are not always the best
solution. The dcMS plasmas exhibit a relatively low electron (plasma) density ne (ne=1014-
1016 m-3) [10] while the penning ionization is the main ionization mechanism [11]. These
features have as a consequence an ionization degree of a few percent for both the sputtered
material and the Ar gas. This in turn results ion low ion-to-neutral ratios available at the
substrate, while Ar+ ions are the dominant bombarding species [10, 11]. The low ion-to-
neutral ratios often require relatively high energies of the individual bombarding species in
order for the effect of the energetic bombarding on the growing film to be significant [8]. The
high bombarding energies may result in subplantation of Ar+ ions into the film [12, 13]
leading to generation of lattice defects [12, 13], high residual stresses [14-18], deterioration of
the quality of the film/substrate interface [16, 19], and poor film adhesion. Thus, the increase
of the ion-to-neutral ratio and primarily the increase of the ionization degree of the film
forming species has been an objective of many research works during the past decades.
Several solutions have been demonstrated, such as the use of inductively coupled plasmas,
external electron and ion sources in conjunction with a magnetron plasma [20], as well as
other ionized PVD techniques such as cathodic arc [20]. These approaches have successfully
served the protective coatings, microelectronic and optical industries by enabling the
deposition of high quality films on complex shaped substrates. However, in the case that the
use of magnetron sputtering based processes without the complexity of external sources is
required the challenge is to turn a conventional magnetron source into an ion source. The
latter can be achieved by increasing the plasma (electron) density and thus the frequency of
the electron impact ionization collisions [11, 21]. It can be shown that the plasma density and
thus the ionization degree are proportional to the power density applied to the target [22]
which implies that this approach is by definition limited by the maximum thermal load that
the cathode can accommodate.
Target-Plasma-Film Interactions in High Power Pulsed Magnetron... 375
In the mid-90’s Mozgrin et al. [23] Bugaev et al. [24] and Fetisov et al. [25]
demonstrated that the operation of a conventional sputtering source in a pulsed mode, with a
pulse duration ranging from 1 µs to 1 s and a frequency less than 2 kHz, allowed for pulse
target power up to three orders of magnitude higher than the average power in dcMS, while
maintaining the average target power at the dcMS level [23-25].These high values of pulse
power resulted, in turn, in ultra-dense plasmas with electron densities in the order of 1018 m-3
[23-25], which is much higher than the values of 1014-1016 m-3 commonly obtained for dcMS
[10, 11]. A few years later in 1999, Kouznetsov et al. [26] demonstrated the operation of a Cu
target in this high power pulsed mode. They showed that high plasma densities obtained using
this pulsed technique resulted in a total ion flux two orders of magnitude higher than that of a
dcMS plasma, and a Cu ionization of ~70 %. The new deposition technique was called High
Power Pulsed Magnetron Sputtering (HPPMS) and in the years after Kouzsetov’s report it has
extensively been used as an elegant approach to increase the ionization degree in
conventional magnetron sputtering devices. Later, several research groups [20] adopted the
alternative name High Power Impulse Magnetron Sputtering (HiPIMS) for this technique.
The high degree of ionization in HPPMS has been shown to allow for the growth of ultra-
dense and ultra-smooth films [27], homogeneous deposition on substrates of complex
geometry [26-28], engineering of the film/substrate interface [29], control over the phase
composition [30, 31] and more recently exotic applications such as gasless sputtering [32].
Another issue which related to the HPPMS process and has attracted considerable attention is
the deposition rate, since it has been reported that depending on the target material, the pulse
configuration and whether the deposition is performed in a metallic or a reactive mode, rates
ranging from 15% to 120% of the values achieved by dcMS are obtained, when the same
average power is used [20, 33-40].
The achievements of HPPMS in the last decade show the great potential of this
technique. The establishment of HPPMS as a mainstream magnetron sputtering technology
goes through a deeper understanding of the underlying physics which would enable
development of products (films and processes) unique for this technique. The key to achieve
this is to consider, understand and describe the HPPMS process in light of the target-plasma-
film interactions. As shown schematically in Figure 1, the interactions of the plasma with the
target determine the plasma density and ionization, the ejection (sputtering) rate of the target
material and the transport of ionized and neutral species from the target to the substrate.
These in turn have implications on the deposition rate and determine the properties of the
growing film. Along these lines in section 2 the issue of the deposition in both non-reactive
and reactive HPPMS is discussed. In section 3 the concept of the target-plasma-film
interactions is used in order to understand the effect of HPPMS on the microstructure and the
phase formation of metal nitride and oxide films.
2. THE DEPOSITION RATE IN HPPMS

The deposition rate in HPPMS is an area that has been attracting considerable attention. It
has been reported that depending on the target material, the pulse configuration and whether
the deposition is performed in a metallic or a reactive mode, rates ranging from 15% to 120%
of the values achieved by dcMS are obtained, when the same average power is used [20, 33-
376 K. Sarakinos
40]. Since the deposition rate is a quantity of significant technological importance, several
research groups have attempted to identify mechanisms that determine its magnitude in
HPPMS. In the following paragraphs, the issue of deposition rate in non-reactive and reactive
HPPMS is discussed.
Figure 1. The target-plasma-film interactions during HPPMS are of key importance for the further
understanding and the development of this technique.
2.1. Non-Reactive HPPMS
The use of HPPMS for non reactive deposition has been shown to give lower deposition
rates than those obtained by dcMS at the same average target power [20, 34]. In order to
understand the mechanisms behind this loss of deposition rate, researchers have focused their
attention on two stages of the film deposition process, (i) the interactions of the plasma
species with the target and their effect on the target erosion (sputtering) rate, and (ii) the
transport of the ionized sputtered species from the target to the substrate [41-44]. In the
following the effect of the target sputtering rate on the deposition rate is discussed,
In a first approximation, the deposition rate in sputtering processes is proportional to the
target erosion rate ( Φ ), i.e. the number of species ejected from the target per time and area
unit. The erosion rate is, in turn, equal to
Φ = ji × Y (1)
where ji is the number (current) of ions impinging onto the target and Y designates the
sputtering yield, namely the number of particles sputtered per impinging ion. In conventional
sputtering processes (e.g. dcMS) the ion target current consists mostly of Ar+ species [10], i.e.
dcMS +
ji ≈ j Ar (2)
which implies that the erosion rate is equal to

+ +
Φ dcMS ≈ j Ar × Y Ar (3)
The total ion current is proportional to the average current applied to the target ( I Tav ).
The sputtering yield, for a given ion-target combination, exhibits a power-law dependence on
the energy ( E ) of the impinging ions [45], i.e.
Y ∝ Em (4)
with m < 1 . The ion energy is, in turn, equal to
E = q × eVT (5)
where VT is the target voltage, q stands for the charge state of the impinging ion and e
designates the elementary charge. Deposition at a constant average target power ( PTav )
implies that
PTav = I Tav × VT = const. (6)
It is known that the HPPMS discharges are driven by cathode potentials in the range of
400 to 2000 V [46, 47], exceeding the target potentials commonly employed in dcMS (~300
eV) [46, 47]. This fact in combination with Eq. (6) means that the use of a constant average
target power leads to lower I Tav values in HPPMS. The lower I Tav values result in a decrease
in the flux of ions to the target and thus, in a decrease in target erosion rate (Eq. (1)) [46].
This decrease cannot be compensated by the increase of the target voltage, due to the non-
linear dependence of the sputtering yield on the ion energy [46, 48], as demonstrated in Eq.
(4). This argument implies that, when the target voltage is increased and the average power is
maintained constant, a loss of deposition rate should be expected, irrespective of the
deposition process used. For most target materials the values of the relative HPPMS-to-dcMS
deposition rates reported in the literature [20] are lower than those which correspond to the
sputtering process described by Eq. (3), i.e. sputtering of the target by Ar+ ions. Thus, a
different mechanism has to be invoked, in order to explain these findings. A number of early
studies on the deposition rate in HPPMS [20, 22] suggested that the high degree of ionization
obtained with this technique leads to an increased probability for ionized sputtered species
(M+) to be re-directed towards the target and self-sputter resulting in an ion current
HPPMS + +
ji = j Ar + j M (7)
The presence of M+ ions in the ion flux towards the target results in an erosion rate equal
to [49],
+ + + + +
Φ HPPMS = j Ar × Y Ar ( E ) + j M × Y M ( E ) − j M (8)
378 K. Sarakinos
M+
with Y (E ) being the energy dependent sputtering yield of M+ ions (also referred to as
M+
self-sputtering yield). The term “ − j ” on the right-hand side of Eq. (8) accounts for the
self-sputtering and results in a decease of the total number of sputtered species as compared
Ar +
to dcMS. The sputtering yield ( Y ) is for most metals higher or comparable to the self-
+
) [22]. This further contributes to the decrease of Φ
M HPPMS
sputtering yield ( Y with respect
to Φ
dcMS
.
It is evident that the understanding of the mechanisms that determine the ionization and
the re-direction of ionized sputtered species is of key importance in order to better clarify the
deposition rate changes observed in HPPMS. In general, an increase of the fraction of ionized
species can be obtained via an increase of the efficiency of the ionization process. In HPPMS
the electron impact is the dominant ionization mechanism [21]. Thus, higher electron
densities and energies, as well as longer interaction times of the sputtered species with the
energetic electrons would lead to an increase of the ionization [21]. This can be, for instance,
achieved by increasing the target voltage, the average and peak target power (or current)
during the HPPMS operation [44, 46, 47, 50], as well as by increasing the pulse-on time [34,
51]. In all these cases a decrease of the deposition rate has been demonstrated for a number of
target materials [20, 46, 49, 52]. Moreover, the high values of peak target current are decisive
for the plasma composition in the target’s vicinity, since they result in high ionization degrees
of the sputtered species and depletion of Ar gas in front of the target, i.e. gas rarefaction [53].
Investigations of the plasma composition in a number of Ar-metal HPPMS discharges have
demonstrated that the high ionization, the gas rarefaction, and the loss of the deposition rate
are observed simultaneously [34]. The gas rarefaction signifies that less Ar is available for
ionization during the pulse on-time, which contributes to the abundance of M+ species in the
target’s vicinity and to the high re-direction probability [34, 46, 47]. In order to identify
whether the rarefaction is a necessary condition for the loss of the deposition rate to occur,
Alami et al. [46] studied the plasma composition and the deposition rate of Cr films grown by
HPPMS and dcMS. In this study a constant average target current was used [46], which
allows for an easy understanding of the underlying physics, since in a first approximation [46]
the total ion target current remains constant at all deposition conditions. Thus, according to
Eqs (1)-(8), changes in the target erosion rate can be assigned to changes in the composition
of the ion target current and the ion energy (i.e. target voltage) [46]. In Figure 2 the deposition
rate of Cr is plotted as a function of the peak target current (ITp) for two pulse on/off time
configurations. It is seen that the increase of the pulse off-time from 950 to 2450 µs led to a
higher peak target current and a decrease in the deposition rate. However, when the pulse
on/off time configuration was held constant, the increase of the peak target current up to a
value of ~28 A resulted in HPPMS rates equal to the dcMS ones (dotted lines in Figure 2).
Above this value, and irrespective of the pulse on/off time configuration used, the HPPMS
deposition rates deviated from the dcMS rates. Time resolved optical emission spectroscopy
measurements [46] showed that at these conditions (ITp > 28 A) the ionization of the sputtered
material increased significantly, while rarefaction was observed. Gas rarefaction was also
observed below the characteristic value 28 A, where the ionization degree significantly lower
than at peak target current values above 28 A [46]. This implies that simultaneous high
ionization and Ar rarefaction are prerequisites for triggering the self-sputtering and loss of the
deposition rate [46]. Based on Eqs. (1)-(8) and using the constant average current approach
HPPMS
Sarakinos et al. [49] derived the following expression for the HPPMS ( R d ) to the
dcMS
dcMS rate ratio( R d ):
+ +
R HPPMS Ar + Y Ar ( E ) M+ (Y M ( E ) − 1)
∝ f ⋅ +
+ f ⋅ +
(9)
R dcMS Y Ar ( E 0 ) Y Ar ( E 0 )
Figure 2. Dependence of the deposition rate of HPPMS grown Cr films on the peak target current. The
films have been deposited using pulse off-times (950 and 1450 µs) at a constant pulse on-time of 50 µs.
The deposition rate deviates from the dcMS one (dotted lines) at the well defined peak target current
value of 28 A irrespective of the pulse on/off time configuration used. Above peak target current value
significant increase of the ionization of the sputtered material, as well as rarefaction of the Ar gas takes
place (data taken from [46]).
Ar + M+
In Eq. (9) f and f are the relative fractions of Ar+ and target ions (M+) in the
Ar + +
target current. Y (E ) and Y M (E ) are the sputtering and the self-sputtering yields as
Ar +
functions of the energy of the sputtering ions, while Y ( E 0 ) is the sputtering yield at the
reference dcMS conditions [49]. In order to quantify the dependence of the composition of
the ion target current on the target material at the various deposition conditions, Eq. (9) was
fitted to experimentally determined deposition rates for C, Cr, and Cu [49]. The
M+
corresponding f values are plotted as a function of the target voltage (i.e. ion energy) in
Figure 3. The analysis showed that in the case of C the current almost exclusively consisted
of Ar+ ions at all deposition conditions [49] On the other hand, the results for Cr and Cu
indicated a transition from and Ar+- to a M+-dominated current, when the target voltage or
equivalently the peak target current was increased [49]. The absence of C+ ions in the ion flux
to the target calculated for C could be attributed to the lower sputtering yield and the lower
ionization probability of this, in comparison to Cr and Cu [49]. The lower sputtering yield
results in lower fluxes of sputtered species, which in turn lead to a less pronounced gas
rarefaction [53] while the lower ionization probability contributes further to the relative
reduction of the local C+ population [49].
380 K. Sarakinos
Figure 3. Relative fraction of ionized sputtered species (fM+) for a number of target materials as a
function of the ion energy (i.e discharge voltage) as estimated by the formalism developed by Sarakinos
et al. [49]. The fM+ values decrease when the sputtering yield and the ionization probability of the
sputtered material decrease (data taken from [49]).
2.2. Reactive HPPMS
The addition of a reactive gas into the sputtering atmosphere leads to target coverage, i.e.
chemisorption and/or implantation [54,55] of reactive gas species at the target surface and the
sub-surface layers, respectively. Deposition from a fully covered target (referred to as the
compound sputtering mode) allows for growth of stoichiometric compound films, i.e.
compound films with sufficient incorporation of the reactive gas atoms [55]. At these
conditions deposition rates lower than those obtained from an elemental (e.g. metallic) target
are commonly achieved [55]. Growth of stoichiometric compound films with relatively high
rates can be facilitated in the intermediate target coverage regime (referred to as transition
zone) between the metallic and the compound mode [55]. In reactive dcMS the transition
sputtering zone is frequently unstable, and a hysteresis of the process characteristics is often
observed [55]. This is particularly pronounced during reactive dcMS of metal oxides [56]. As
a consequence, stoichiometric films can only be obtained in the compound sputtering mode
[55], unless a feedback system for controlling the target coverage is employed [57]. Earlier
studies on the reactive HPPMS of various metal oxides (TiO2, ZrO2, Ta2O5) revealed
deposition rates ranging between 25 % and 120% of dcMS rates [36, 38]. In these studies the
obtained deposition rates were interpreted in light of the self-sputtering phenomenon [36, 38].
More recently, systematic investigations of the reactive HPPMS of Al2O3 [58] and ZrO2 [59]
showed that, in contrary to dcMS, the HPPMS process exhibits a hysteresis free and stable
transition zone, as demonstrated for the reactive deposition of ZrO2 in Figure 4. The
stabilization of the transition zone allows for deposition of stoichiometric films at a lower
target coverage, when compared to the compound mode in dcMS [58, 59], and has been
shown to result in rates similar to [58] or up to two times higher [59] than the dcMS rates
(Figure 5).
Figure 4. Target voltage versus O2 flow during deposition of ZrO2 using (a) dcMS and HPPMS with a
pulse on-time of 50 µs and pulse of times (b) 450 µs and (c) 1450 µs. HPPMS allows for a stable
transition zone as opposed to dcMS (data taken from [59]).
Figure 5. Normalized (with respect to the average target power) deposition rate of ZrO2 grown using (a)
dcMS and HPPMS with a pulse on-time of 50 µs and pulse of times (b) 450 µs and (c) 1450 µs. The
stabilization of the transition mode in HPPMS (see Figure 4) allows for deposition of transparent ZrO2
films with rates up to 2 times higher than those in dcMS (data taken from [59]).
In general, the stability of the transition zone in reactive sputtering processes is

determined by the competition between the formation and the removal (sputtering) of the
compound from the target surface [55]. According to the formalism developed by Berg et al.
[55] the steady-state target coverage, ϑ t , is equal to,
ji Ycϑt
− a 2 F (1 − ϑt ) = 0 (10)
q
where ji is the ion target current density, Yc the compound sputtering yield, q the
elementary charge, a the sticking coefficient of the reactive gas and F the flux of the
reactive gas molecules towards the target [55]. Stabilization of the transition mode can be
382 K. Sarakinos
JYcϑt
achieved, if for a certain nominal partial pressure/flow of the reactive gas the term “ ”
q
(removal of the compound) increases and/or the term “ a 2 F (1 − ϑt ) ” (formation of the
compound) decreases. The pulsed character of the HPPMS discharge affects both terms in Eq.
(10). The high peak currents (up to three orders of magnitude higher than the average values)
result in a significantly higher instantaneous erosion rate during the pulse on-time, which may
effectively clean the target surface and displace the oxidation onset at higher reactive gas
flows [58]. In addition, the absence of plasma during the pulse off-time results in a limited
activation of the reactive species [58, 60]. Under these conditions relatively high levels of
reactive gas exposure are necessary for the formation of the compound on the target surface,
i.e. this mechanism leads to lower effective values of the sticking coefficient a in Eq. (10).
Furthermore, as a result of the high target voltage in HPPMS [46] higher compound
sputtering yields Yc and as a consequence higher values of the compound erosion rate can be
achieved [59]. The high peak target current in HPPMS results in rarefaction of the neutral
species in the target’s vicinity, as discussed in Section 2.1, which affects not only the Ar but
also the reactive gas [37]. The rarefaction implies that the reactive species flux F lower than
the value which corresponds to the nominal partial pressure of the reactive gas should be
expected [37, 59].
3. HPPMS FOR TAILORING THE FILM MICROSTRUCTURE

AND PHASE FORMATION
3.1. The Effect of HPPMS on the Microstructure of Metal Nitride Films
Polycrystalline films grown by PVD techniques exhibit a variety of microstructures with

respect to the size, the morphology, and the relative orientation of the crystallites [8, 61].
These features have, for instance, implications on the mechanical strength [62] and the
electrical conductivity [63-65] of the film. The microstructure is determined primarily by
surface and bulk diffusion processes [8, 61], and may be controlled by variation of the growth
temperature (Ts). In the case of metallic films, the effect of the temperature can be quantified
Ts
using the so-called homologous temperature (Tm is the melting temperature of the
Tm
deposited material) and the obtained microstructural features can be classified according to
Ts
the structural zone models (SZM) [4, 6, 7, 9]. At < 0.2 (Zone I), the size of the grains is
Tm
determined by the nucleation density, the film consists of uninterrupted fibrous columns and
exhibits porous and rough morphologies and a random texture [8, 61]. In zone T
Ts
( 0.2 < < 0.4 ) the surface diffusion has considerable influence on the growth. This fact
Tm
results in V-shaped grains at the interface due to competitive growth. At larger film
thicknesses faster growing grains overwhelm slower ones giving rise to a columnar dense
morphology, smoother surfaces and strong texture [8, 61]. At higher temperatures
Ts
( > 0.4 ; Zone II) bulk diffusion becomes significant giving rise to a dense columnar
Tm
microstructure which is retained from the interface to the film surface [8, 61]. In Zone III the
structure is characterized by equi-axed three-dimensional grains. This growth mode is a result
of periodic interruption of the crystal growth and under the presence of contaminants in the
film it can be observed at any substrate temperature [8, 61].
Figure 6. Cross sectional TEM micrographs of CrN films grown at room temperature by (a) dcMS and
(b) HPPMS at a peak target current of 44 A. The dcMS grown films exhibit a columnar underdense
morphology with inter-columnar porosity. The use of HPPMS results in dense films and suppression of
the columnar growth (data taken from [68]).
The simultaneous use of low flux ion bombardment (typical for conventional magnetron
sputtering processes) allows for the transitions between the structural zones to occur at lower
temperatures [6, 7, 9] and for the morphological features predicted by the SZM’s to remain
valid [6, 7, 9]. In contrast, growth of thin films during HPPMS is characterized by the high
ion flux during the pulse on-time which can be up to three orders of magnitude higher than in
dcMS [20]. Observations of the morphology of films grown by this technique have shown
that the SZM in its classical form is no longer valid [29, 66-68]. An example is given in
Figure 6 where cross sectional TEM images of CrN films grown by HPPMS and dcMS under
otherwise the same conditions are shown [68]. In contrast to the underdense columnar
morphology and high surface roughness obtained by dcMS (Figure 6 (a)), the use of HPPMS
resulted in film densification, surface smoothening and suppression of the columnar structure
, as shown in Figure 6 (b). In addition, some columns grew on the top of existing columns,
i.e. renucleation is observed. As mentioned above, in films grown under low flux ion
irradiation the renucleation occurs when the individual grain growth is inhibited by
contaminants [61]. The interruption of the columnar growth can be enhanced when the
384 K. Sarakinos
plasma ionization and thus the substrate ion current are increased, e.g. by increasing the peak
target current [68]. This can facilitate the transition from a dense polycrystalline to a
featurelesss/nanocrytalline structure [68], as shown in Figure 7. It is therefore, evident that the
low-energy high-flux ion irradiation during HPPMS can be used, in order to overcome the
characteristically underdense and rough microstructures and obtain morphologies unique for
low temperature deposition [68].
Figure 7. Cross sectional SEM micrographs of CrN films grown at room temperature by (a) dcMS and
HPPMS at a peak target current values of (b) 44, (b) 74 and (d) 180 A. The increase of the peak target
current results in a transition from a columnar to a featureless microstructure (data taken from [68]).
3.2. Phase Composition Tailoring of Metal Oxide Films Deposited By

HPPMS
The influence of the energetic bombardment on the phase formation of HPPMS deposited
films can be unraveled using TiO2. In general, TiO2 films can grow in an amorphous and in
two tetragonal crystalline structures; the rutile and the anatase phases [69]. Deposition and/or
annealing at high temperatures (700–900 °C) have been reported to lead to the formation of
the rutile phase [69]. The anatase phase is generally obtained at lower temperatures [69],
while deposition at room temperature often leads to the formation of amorphous films [69]. In
addition, the energetic bombardment by positively charged ions (facilitated by applying a
negative bias voltage on the substrate) can promote the crystallization of TiO2 films at room
temperature [70]. When HPPMS is employed, the rutile phase can be achieved even at room
temperature and the increase of the flux of energetic species towards the growing films favors
the formation of the rutile at the expense of the anatase phase [31, 71-73]. This can be, for
instance, achieved by increasing the peak target current [72] or decreasing the working
pressure during deposition (Figure 8) [71, 73]. In general, when O2 is used as a reactive gas,
the oxidized target surface is the source for the generation of negatively charged oxygen ions
[74]. These ions are accelerated in the cathode sheath by the negative target potential towards
the substrate [75] and it has been shown to have implications on the structure formation and
the phase composition of reactively sputtered metal oxide films [76]. In HPPMS target
voltages higher than those in dcMS are commonly observed, as shown in the previous
sections. This implies that metal oxide films deposited by HPPMS do not only experience the
high-flux low-energy bombardment by ionized sputtered species but they are also irradiated
by O- species with energies higher than those during dcMS. The trajectory of the O- ions,
when targets with no pronounced erosion track are used, is largely directional and
perpendicular to the target surface [77]. Therefore, substrates placed perpendicular to the
target surface will encounter a significantly suppressed O- flux, as compared to parallel
oriented substrates, as well as high positive ion-to-neutral ratios due to an abnormal transport
of ions observed in HPPMS discharges [43]. The effect of the substrate orientation on the
phase composition is shown in Figure 9. Deposition on substrates facing the target leads to
the growth of rutile-rich films (Figure 9) [73]. The deposition on the inclined substrates
(Figure 9 (b)) also results in rutile films with a morphology that consists of crystallites
embedded in an amorphous matrix [73]. These results outline an alternative pathway for the
synthesis of crystalline TiO2 films; in the absence of high energy bombardment (by means of
O- ions), i.e. crystalline (rutile-rich) TiO2 films can be deposited when a sufficient flux of
low-energy (positive) ion irradiation is available [73].
Figure 8. X-ray diffraction (XRD) patters of TiO2 films grown at room temperature by HPPMS at
working pressures of (a) 0.5, (b) 1.0, (c) 2.0, (d) 2.5 and (e) 3.0 Pa. The capitals “A” and “R” designate
the position the XRD peaks in unstrained bulk anatase- and rutile-TiO2 phase, respectively. The
increase of the working pressure results in a transformation from rutile rich films to film that consist of
mixtures from anatase- and rutile phase (data taken from [73]).
386 K. Sarakinos
Figure 9. X-ray diffraction patters of TiO2 films grown at room temperature by and at a working
pressure of 0.5 Pa by HPPMS for substrate planed at an angle of (a) O (parallel) and (b) 90°
(perpendicular) with respect to the target. The capitals “A” and “R” designate the position the XRD
peaks in unstrained bulk anatase- and rutile-TiO2 phase, respectively. In both cases films with a
primarily rutile-TiO2 crystal structure are obtained. However, the XRD pattern for the film grown on
perpendicularly oriented substrate implies a microstructure that consists of TiO2 crystals embedded in
an amorphous matrix (data taken from [73]).
4. SUMMARY
High power pulsed magnetron sputtering is a novel ionized physical vapor deposition
technique with great potential for improving already existing and developing new functional
films. One issue related to HPPMS is the lower film deposition rates as compared to
conventional magnetron sputtering techniques (e.g. dcMS) for the same average target power
when the discharge operates in a non-reactive mode. This is an intrinsic characteristic of the
HPPMS process which can be largely attributed to the re-direction of ionized sputtered
species to the target resulting in self-sputtering. Prerequisites for the self-sputtering to occur
are the high density of ionized sputtered species and the depletion of Ar species (gas
rarefaction) in the target’s vicinity. This implies that target materials with a low sputtering
yield and low ionization probability would exhibit a less pronounced loss of the deposition
rate. In addition, for a given material the loss of the rate can be suppressed by choosing
discharge conditions that correspond to a relatively low degree of ionization, i.e. a
compromise between deposition rate and ionization of the sputtered material can be always
found depending on the requirements of the specific application. In the case of reactive
deposition, operation in HPPMS mode can provide a stable transition zone. This, in turn,
facilitates deposition of stoichiometric compound films at target coverage lower and higher
rates as compared to dcMS. As far as it concerns film properties, the high fluxes of ionized
material in HPPMS enable the room-temperature growth of metal oxide and nitride films with
unique microstructure and phase composition
5. REFERENCES
[1] Mattox, D. M. (1998). Handbook of Physical Vapor Deposition (PVD) Processing,
Noyes Publications, Westwood.
[2] Ohring, M. (1992). The Material Science of Thin Films, Academic Press, San Diego.
[3] Michely, T. & Krug, J. (2004). Islands Mounts and Atoms, Springer, Berlin.
[4] Mochan, B. A. & Demchishin, A.V. (1969). Fiz. Met. Metalloved., 28, 33.
[5] Hopwood, J. (2000). Ionized Physical Vapor Deposition, Academic Press, San Diego.
[6] Thornton, J. A. & Vac, J. 1986). Sci. Technol., A 4, 3059.
[7] Messier, R., Giri, A. P., Joy, A. R. & Vac, J. (1984). Sci. Technol., A 2.
[8] Petrov, I., Barna, P. B., Hultman, L., Greene, J. E. & Vac, J. (2003). Sci. Technol.,
A 21, S117.
[9] Grovenor, C. R. M., Hentzell, H. T. G. & Smith, D. A. (1984). Acta Metall., 32, 773.
[10] Christou, C., Barber, Z. H. & Vac, J. (2000). Sci. Technol., A 18, 2897.
[11] Chapman, B. (1980). Glow Discharge Processes, John Wiley & Sons.
[12] Lifshitz, Y., Kasi, S. R. & Rabalais, J. W. (1989). Phys. Rev. Lett., 62, 1290.
[13] Robertson, J. (2002). Mat. Sci. Eng., R37, 129.
[14] Janssen, G. C. A. M. & Kamminga, J. D. (2004). Appl. Phys. Lett. 85, 3086.
[15] Pauleau, Y. (2001). Vacuum, 61, 175.
[16] Koch, R. & Phys, J. (1994). C.: Condens. Matter. 6, 9519.
[17] Windischmann, H. (1992). Crit. Rev. Solid. Stat. Mater. Sci., 17, 547.
[18] Davis, C. A. (1993). Thin Solid Films, 226, 30.
[19] Hovsepian, P. (1988). Arch. Metall., 33.
[20] Helmersson, U., Lattemann, M., Bohlmark, J., Ehiasarian, A. P. & Gudmundsson, J. T.
(2006). Thin Solid Films, 513, 1.
[21] Gudmundsson, J. T. (2008). J. Phys. : Conf. Ser., 100, 082013.
[22] Christie, D. J. (2005). J. Vac. Sci. Technol., A 23, 330.
[23] Mozgrin, D. V. Fetisov, I. K. Khodachenko, G. V. (1995). Plasma Phys. Rep., 21, 401.
[24] Bugaev, S. P., Koval, N. N., Sochugov, N. S., & Zakharov, A. N. (1996). XVIIth
International on Discharges and Electrical Insulation in Vacuum, p. 1074.
[25] Fetisov, I. K., Filippov, A. A., Khodachenko, G. V., Mozgrin, D. V. & Pisarev, A.A.
(1999). Vacuum, 53, 133.
[26] Kouznetsov, V., Macak, K., Schneider, J. M., Helmersson, U., Petrov, I., (1999). Surf.
Coat. Technol, 122, 290.
[27] Alami, J., Persson, P. O. Å., Music, D., Gudmundsson, J. T., Bohlmark, J. &
Helmersson, U. (2005). Vac, J. Sci. Technol., A 23, 278.
[28] Chistyakov, R., Abraham, B. & Sproul, W. D. (2006). Society of Vacuum Coaters 49th
Annual Technical Conference Procedings, Washington, DC, p. 88.
[29] Ehiasarian, A. P., Münz, W. D., Hultman, L., Helmersson, U., Petrov, I. (2003). Surf.
Coat. Technol., 163-164, 267.
[30] Alami, J., Eklund, P., Andersson, J. M., Lattemann, M., Wallin, E., Bohlmark, J.,
Persson P., Helmersson, U., (2007). Thin Solid Films, 515, 3434.
[31] Konstantinidis, S. Dauchot, J. P. & Hecq, M. (2006). Thin Solid Films, 515, 1182.
[32] Andersson, J. Anders, A. (2008). Appl. Phys. Lett., 92, 221503.
388 K. Sarakinos
[33] Sproul, W. D., Christie, D. J. & Carter, D. C., (2004). Society of Vacuum Coaters 47th
Annual Technical Conference Proceedings, Dallas, TX, p. 96.
[34] Konstantinidis, S., Dauchot, J. P., Ganciu, M., Ricard, A. & Hecq, M. (2006). Appl. J.
Phys., 99 013307.
[35] Bobzin, K., Bagcivan, N., Immich, P., Bolz, S., Alami, J., Cremer, R. & Mat. J. (2008).
Proc. Technol.
[36] Glocker, D. A., Romach, M. M., Christie, D. J. & Sproul, W. D. (2004). Society of
Vacuum Coaters 47th Annual Technical Conference Proceedings, Dallas, TX, p. 183.
[37] Sarakinos, K., Alami, J., Wuttig, M., Phys. J., (2007). D: Appl. Phys., 40, 2108.
[38] Davis, J. A., Sproul, W. D., Christie, D. J., Geisler, M., (2004). Society of Vacuum
Coaters 47th Annual Technical Conference Proceedings, Dallas, TX, p. 215.
[39] Alami, J., Eklund, P., Emmerlich, J., Wilhelmsson, O., Jansson, U., Högberg, H.,
Hultman, L., Helmersson, U. (2006). Thin Solid Films, 515, 1731.
[40] DeKoven, B. M., Ward, P. R., Weis, R. E., Christie, D. J., Scholl, R. A., Sproul, W. D.,
Tomasel, F., Anders, A., (2003). Society of Vacuum Coaters 46th Annual Technical
Conference Proceedings, San Francisco, CA, p. 158.
[41] Konstantinidis, S., Dauchot, J. P., Ganciu, M. & Hecq, M., (2006). Appl. Phys. Lett.,
88, 021501.
[42] Sevvana, S., Chennadi, S., Lakkaraju, R. K., Li, J. & Rohde, S., (2005). Society of
Vacuum Coaters 48th Annual Technical Conference Proceedings, Denver, CO, p. 720.
[43] Lundin, D., Larsson, P., Wallin, E., Lattemann, M., Brenning, N., Helmersson, U.,
(2008). Plasma Sources Sci. Technol., 17, 035021.
[44] Vlcek, J., Kudlacek, P., Burcalova, K., Musil, J., Vac. J., (2007). Sci. Technol.,
A 25, 42.
[45] Berisch, R., (1982). Sputtering by Particle Bombardment I, Springer (Berlin).
[46] Alami, J., Sarakinos, K., Mark, G., Wuttig, M., (2006). Appl. Phys. Lett., 89, 154104.
[47] Ehiasarian, A. P. New, R., Münz, W. D., Hultman, L., Helmersson, U. & Kouznetsov,
V., (2002). Vacuum, 65, 147.
[48] Emmerlich, J., Mraz, S., Snyders, R., Jiang, K. & Schneider, J. M., (2008). Vacuum, 82,
867.
[49] Sarakinos, K., Alami, J., Dukwen, J., Wördenweber, J., Wuttig, M. & Phys. J., (2008).
D: Appl. Phys., 41, 215301.
[50] Bohlmark, J., Alami, J., Christou, C., Ehiasarian, A. P., Helmersson, U. & Vac, J.
(2005). Sci. Technol., A 23, 18.
[51] Anders, A., Andersson, J. & Ehiasarian, A. P. J. (2007). Appl. Phys. 102, 113303.
[52] Horwat, D., Anders, A. & Phys. J., (2008). D: Appl. Phys. 41, 135210.
[53] Rossnagel, S. M., Kaufman, H. R. & Vac, J., (1987). Sci. Technol., A 5, 2276.
[54] Depla, D., De Gryse, R. & Vac. J. (2002). Sci. Technol., A 20, 521.
[55] Berg, S. & Nyberg, T. (2005). Thin Solid Films, 476, 215.
[56] Severin, D., Kappertz, O., Kubart, T., Nyberg, T., Berg, S., Pflug, A., Siemers, M. &
Wuttig, M. (2006). Appl. Phys. Lett., 88, 161504.
[57] Sproul, W. D., Graham, M. E., Wong, M. S., Lopez, S., Li, D. & Scholl, R.A., (1995).
J. Vac. Sci. Technol., A 1188.
[58] Wallin, E. & Helmersson, U. (2008). Thin Solid Films, 516, 6398.
[59] Sarakinos, K., Alami, J., Klever, C. & Wuttig, M. (2008). Surf. Coat. Technol., 202,
5033.
[60] Depla, D. & De Gryse, R. (2004). Surf. Coat. Technol., 183, 184.
[61] Barna, P. B. & Adamik, M. (1998). Thin Solid Films, 317, 27.
[62] Koehler, J. S. (1970). Phys. Rev., B2, 547
[63] Tellier, C. R. & Tosser, A. J. (1982.). Size Effects in Thin Films, Elsevier, Amsterdam.
[64] Munoz, R. C., Finger, R., Arenas, C., Kremer, G. & Moraga, L. (2002). Phys. Rev. B
66, 205401.
[65] Van Attekum, P. M., Woerlee, P. H., Verkade, G. C. & Hoeben, A. A. (1984). Phys.
Rev., B 29, 645.
[66] Chistyakov, R., Abraham, B., Sproul, W. D., Moore, J. & Lin, J. (2007). Society of
Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY, p.
139.
[67] Paulitsch, J. Mayrhofer, P.H. Mitterer, C. Münz, W.D. Schenkel, M., (2007). Society of
Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY,
p. 150.
[68] Alami, J., Sarakinos, K., Uslu, F., Wuttig, M. &. Phys. J. (2009), D: Appl. Phys., 42,
015304.
[69] Wiggins, M. D., Nelson, M. C., Aita, C. R. & Vac. J., (1996). Sci. Technol., A14, 772.
[70] Zhou, W., Zhong, X. X., Wu, X. C., Yuan, L. Q., Shu, Q. W., Li, W., Xia, Y. X. &
Phys, J., (2007). D.: Appl. Phys., 40, 219.
[71] Stranak, V., Quaas, M., Wulff, H., Hubicka, Z., Wrehde, S., Tichy, M., Hippler, R. &
Phys. J., (2008). D: Appl. Phys. 41, 055202.
[72] Bandorf, R., Vergöhl, M., Giesel, P., Wallendorf, T. & Mark, G. (2007). Society of
Vacuum Coaters 50th Annual Technical Conference Proceedings, Louisville, KY,
p. 160.
[73] Alami, J., Sarakinos, K., Uslu, F., Klever, C., Dukwen, J., Wuttig, M. & Phys. J.,
(2009). D: Appl. Phys., 42, 115204.
[74] Tominaga, K. K., Ito, D. & Miyamoto, M. (2006). Vacuum, 80, 654.
[75] Mraz, S. & Schneider, J. M. (2006). Appl. J., Phys., 100, 023503.
[76] Ngaruiya, J. M., Kappertz, O., Mohamed, S. H. & Wuttig, M., (2004). Appl. Phys. Lett.,
85, 748.
[77] Severin, D., Kappertz, O., Nyberg, T., Berg, S. & Wuttig, M. (2007). Thin Solid Films
515, 3554.
INDEX
AFM, xii, 30, 34, 39, 42, 273, 275, 276, 277, 279,
A 287, 288, 289, 298, 325, 333, 334, 335, 348,
354, 358, 359, 360
Abelian, 163
Ag, 11, 15, 46, 52, 56, 144, 154, 294, 347
absolute zero, 198
agar, 204, 206
absorption, 8, 9, 10, 14, 18, 19, 20, 37, 56, 58, 68,
age, 5
69, 71, 72, 78, 79, 81, 84, 93, 97, 106, 157,
agent, 24
163, 164, 173, 247, 311, 345, 366
agents, 11
absorption coefficient, 20, 79
aggregates, 216, 307
absorption spectra, 8, 56
aid, 82, 103
acceleration, 72, 344
air, x, 57, 60, 64, 69, 85, 86, 87, 88, 89, 90, 92,
acceptor, 12, 48, 50, 53
93, 94, 96, 97, 98, 99, 101, 104, 113, 115, 126,
accommodation, 144, 145
128, 130, 131, 135, 136, 138, 139, 140, 141,
accuracy, xiii, 21, 77, 176, 353
143, 148, 165, 173, 199, 202, 203, 214, 243,
acetic acid, 305
252, 253, 254, 259, 260, 305, 349, 366, 370,
acetone, 24
371
acetylene, 87, 93, 142, 370
air quality, 58
achievement, 126
air-dried, 305
acid, 275, 277, 283, 285, 303, 304, 305, 306
airports, 211
acidic, 23
airways, 203
acidity, 366
alcohol, 369, 370
acoustic, 84
alcohols, xiii, 365, 368
acoustical, 52
algae, xii, 301, 303, 305, 307, 308, 309, 310, 311,
activation, 11, 50, 56, 83, 263, 264, 268, 337, 382
312, 313, 314, 315, 316, 317
activation energy, 11, 263, 264, 268
algorithm, 233
active site, 315
alkaline, 9
actuators, 242
alloys, 118, 119, 120
ADC, 147, 148
alternative, xiii, 173, 302, 365, 366, 368, 374,
adhesion, 3, 329, 330, 355, 374
375, 385
adjustment, 177, 201
alternative energy, xiii, 365
adsorption, xii, 3, 53, 163, 247, 252, 253, 254,
alternatives, 202
255, 256, 257, 258, 259, 260, 301, 303, 305,
aluminium, 61, 354, 355, 356, 357, 358
314, 315, 316
alveoli, 6
adsorption isotherms, 314, 316
amalgam, 144, 154, 156
aerobic, 303
ambient pressure, ix, 63, 64, 109
aerosol, 199
amorphous, 21, 23, 36, 37, 53, 58, 118, 119, 120,
281, 344, 346, 347, 354, 358, 360, 384, 386
392 Index
amplitude, 117, 118, 119, 219, 359, 360, 361, back, 39, 83, 84, 247, 288, 291, 335
362, 363 background noise, 360
anaerobic, 303 bacteria, 5, 206, 303
analog, 124 bacterial, 197, 202, 204, 205, 206, 207, 215, 216,
anatase, 384, 385, 386 316
angular momentum, 76, 80 band gap, 5, 11, 14, 21, 44, 45, 48, 49, 50
aniline, 314 barium, 171
animals, 195, 196, 203, 204 barrier, xi, 2, 132, 180, 245, 246, 247, 248, 250,
anisotropy, 52, 311 251, 252, 254, 255, 256, 257, 258, 259, 260,
annealing, xi, 15, 59, 273, 275, 277, 280, 284, 261, 262, 263, 264, 267, 273, 274, 343, 344,
286, 287, 293, 294, 298, 321, 322, 332, 333, 345, 347, 348, 349
342, 345, 346, 347, 348, 349, 350, 384 barriers, xi, 55, 245, 246, 247, 263
anode, 81, 367 batteries, 9
anomalous, 184, 260 beams, 10, 37, 71, 84, 348
anti-bacterial, 5, 9 beating, 203
ants, 205, 208 behavior, xi, xii, 2, 20, 35, 51, 62, 128, 141, 169,
aqueous solutions, 5 170, 171, 172, 177, 178, 179, 180, 182, 184,
arc plasma, 128 185, 187, 188, 189, 190, 191, 195, 197, 205,
argon, x, 24, 27, 34, 113, 114, 115, 118, 123, 207, 208, 210, 213, 245, 247, 253, 255, 257,
124, 125, 126, 127, 128, 129, 130, 132, 134, 259, 260, 262, 263, 264, 265, 266, 268, 302,
144, 165, 252, 259, 321, 322, 323, 332, 338, 307, 309, 313, 314, 317
370 behaviours, 106
argument, 209, 211, 377 bending, 80, 267
Aristotle, 195, 214 benign, 57
arrow of time, 198 bias, 4, 51, 116, 248, 249, 251, 384
arsenic, 50 bifurcation, 202
arsenide, 81 binding, 2, 38, 48, 49, 94, 142, 303, 306, 314,
ASD, 111 315, 316
assignment, ix, 63, 76, 85, 87, 88, 89, 92, 93, 109 binding energies, 38
assumptions, 93, 254 binding energy, 38, 48, 49, 94
athletes, 204 biocatalysts, 306
atmosphere, 6, 24, 34, 86, 94, 97, 126, 175, 201, biocompatible, 302, 311
216, 245, 253, 254, 259, 323, 366, 380 biocompatible materials, 302
atmospheric pressure, x, 69, 81, 113, 114, 115, biodegradation, 303
126, 129, 130, 140, 165, 368, 369, 370 bioethanol, 368
atomic force microscope, xiii, 353, 354, 358, 359, biogas, xiv, 365, 368, 371
363 biologically active compounds, xii, 301, 302,
atomic force microscopy (AFM), 30, 39, 41, 319, 303, 316, 317
325, 333, 348 biomass, 315, 316, 366
atoms, 2, 37, 46, 66, 68, 69, 71, 73, 74, 76, 77, biosciences, 302
79, 80, 83, 94, 95, 108, 109, 111, 132, 133, biosensors, 9
144, 157, 162, 163, 278, 320, 324, 368, 374, biosorption, 315
380 biotechnology, 302, 304
attachment, 69, 70, 71 bipolar, 53
attacks, 277 birds, 203, 204, 207, 210
Auger electron spectroscopy, 277, 330 bleaching, 17, 23
automotive sector, 366, 371 blood, xii, 3, 4, 6, 7, 8, 9, 54, 56, 202, 203, 302,
availability, 10, 53 304
averaging, 52, 104, 105 blood flow, 54
body fluid, xii, 302, 304
body mass, 203, 211
B body size, 203, 204, 215
body temperature, 260
Bacillus, 316
Index 393
body weight, 203 ceramic, xi, 59, 172, 173, 175, 176, 188, 190,
Bohr, 67 245, 246, 247, 255, 256, 259, 262, 263, 265,
Boltzmann constant, 67, 250, 255, 261, 263 334
Boltzmann distribution, 93 ceramics, 171, 177, 178, 180, 183, 184, 185, 189,
bonding, 2, 12, 45, 57 191, 245, 247, 252, 253, 254, 255, 259, 260,
bonds, 2, 11, 12, 45, 46, 366, 368 261, 262, 263, 265
boundary conditions, 117, 219, 221 CFD, xi, 217, 218, 219, 220, 221, 222, 223, 224,
Boussinesq, 232 225, 229, 230, 231, 233
breakdown, ix, 39, 63, 64, 65, 68, 69, 70, 72, 77, CH4, 129, 142, 143
78, 83, 94, 98, 102, 103, 104 channels, 209, 346
breathing, 203 charge coupled device, 81
bremsstrahlung, 69, 72, 84, 95 charge density, 10, 248
broadband, 348 charged particle, 65, 66, 72, 121
Brownian motion, 198 chemical bonds, 2
buffer, 123, 144, 306 chemical degradation, 303
buildings, 18, 210 chemical energy, 366
bulk materials, xiii, 3, 341, 342, 350 chemical industry, xiv, 142, 365
burning, 87 chemical properties, 46, 342, 348
butane, 87, 368 chemical reactions, 5, 66, 124
chemical reactor, 124, 125
chemical vapor deposition, 17, 23, 62, 266
C chemical vapour, 49, 50, 51
chemical vapour deposition, 49, 50, 51
calculus, 251
chemisorption, 269, 316, 380
calibration, 83, 172, 173, 174, 175, 314
chest, 202
calorimetry, 136
circulation, 116, 117, 201, 203, 214, 215
cancer cells, 5
classes, 204, 314
candidates, 266, 347
classical, x, 68, 71, 169, 184, 383
capacitance, 173
cleaning, 5, 58, 214
capillary, 56
clinical trial, 54
carbon, xiv, 4, 5, 6, 7, 8, 63, 64, 79, 82, 85, 86,
clinical trials, 54
87, 88, 89, 92, 94, 95, 96, 97, 98, 99, 101, 102,
clusters, 10, 11, 12, 13, 14, 26, 57, 94
103, 104, 105, 106, 107, 108, 109, 123, 126,
CMOS, 274
142, 165, 202, 212, 280, 281, 347, 365, 366,
Co, vii, 113, 116, 118, 120, 132, 135, 142, 149,
368, 369
159, 166, 193, 214, 270
carbon atoms, 95
CO2, vii, ix, xiii, 54, 63, 64, 80, 82, 83, 84, 85,
carbon dioxide, 4, 5, 6, 7, 8, 79, 96, 97, 123, 126,
87, 88, 89, 92, 94, 95, 96, 97, 100, 102, 103,
142, 165, 202, 212
104, 106, 107, 108, 109, 129, 142, 143, 212,
carbon monoxide, 5, 142
213, 216, 365, 366, 367, 368, 370, 371
carcinogens, 302
coagulation, 303
Carnot, 196
coatings, 2, 5, 9, 28, 53, 320, 374
carrier, 4, 5, 37, 44, 45, 47, 50, 52, 264
coil, 114, 115, 116, 117, 123, 126, 144, 145, 146
catalysis, ix, 1, 2, 3
collisions, 64, 69, 70, 74, 79, 94, 95, 104, 108,
catalyst, 21, 29, 30, 34, 59
109, 324, 343, 345, 368, 374
cathode, 81, 83, 267, 367, 374, 377, 385
combined effect, 46, 70, 240
cation, 3, 44, 47
combustion, 93, 366, 367
cavitation, 84
commons, 211
cavities, 243
communication, xiv, 171, 205, 208, 210, 274,
cell, x, xiii, 6, 17, 47, 82, 113, 142, 144, 291,
344, 373
292, 306, 311, 319, 324, 326, 327, 328, 331,
communication systems, 171
332, 338, 367
community, 44
cell surface, 306
compaction, 346
compensation, 49, 173
394 Index
competition, 381 control, 2, 3, 13, 17, 34, 45, 46, 58, 71, 83, 146,
competitor, 49 147, 171, 175, 177, 205, 206, 208, 219, 320,
complement, 326 342, 344, 375
complex systems, 209, 210, 211 convection, 198, 199, 201, 215
complexity, 53, 89, 197, 205, 210, 216, 374 convective, 115, 127, 128, 132, 198, 201, 202,
compliance, 189, 252 203, 215
complications, 54 convergence, 220, 243
components, 65, 72, 81, 93, 117, 119, 172, 185, convergence criteria, 220
187, 188, 189, 190, 191, 196, 247, 306, 309, conversion, xiv, 2, 123, 142, 274, 350, 373
310, 342, 349, 359, 371 cooling, xiii, 18, 83, 102, 108, 124, 132, 133,
composition, xii, 3, 38, 47, 50, 53, 60, 64, 66, 75, 134, 135, 136, 138, 139, 140, 141, 142, 165,
97, 121, 142, 143, 144, 154, 179, 182, 183, 175, 215, 252, 253, 307, 319, 321, 322, 326,
230, 241, 273, 275, 277, 278, 280, 284, 285, 359
287, 288, 289, 298, 319, 320, 321, 322, 323, copper, 45, 65, 81, 282
324, 325, 326, 327, 337, 338, 347, 375, 378, coral, 205, 207, 215
379, 385, 386 corona, 66
compounds, xii, xiv, 39, 46, 115, 142, 301, 302, correlation, 8, 171, 180, 182, 203, 254, 262
303, 304, 314, 316, 317, 320, 322, 326, 365 correlations, 211, 216
comprehension, 266 corrosion, 2, 4, 8
Computational Fluid Dynamics, 242 cosmetics, 303
computational modeling, xi, 245 costs, xiii, 18, 341, 342
concentrates, 44 Coulomb, 18, 66, 71, 72
concentration, 37, 44, 45, 47, 50, 51, 52, 53, 70, coupling, 76, 185, 187, 188, 189, 191, 192, 233
123, 124, 127, 129, 132, 133, 134, 135, 136, covalent, 2, 12, 14, 45, 46
137, 138, 139, 140, 141, 142, 157, 162, 163, covalent bond, 2, 12, 45, 46
164, 165, 190, 205, 245, 267, 275, 289, 303, covalent bonding, 2
314, 315, 316, 323, 326, 366 covering, 144
conception, 210 CPD, 21, 22, 23
conceptualization, 211 crack, 30, 41, 215
condensation, 183 cracking, 83
condensed matter, 45 CRC, 55, 165
conductance, 201, 246, 247 credibility, 234
conduction, 5, 10, 12, 21, 44, 46, 49, 52, 53, 61, critical points, 7
62, 203, 250, 251, 256, 259, 264, 266 cross-sectional, 281, 336
conductive, 17, 61, 65, 115, 268, 354, 355, 359 crystal growth, 332, 383
conductivity, xi, 9, 11, 12, 22, 23, 44, 45, 46, 48, crystal structure, xii, 23, 45, 188, 260, 319, 320,
50, 51, 52, 53, 58, 118, 157, 159, 160, 245, 321, 323, 337, 338, 386
256, 259, 263, 264, 382 crystal structures, 23, 45
conductor, 17, 159, 160, 175 crystalline, 21, 23, 33, 34, 36, 59, 60, 118, 188,
configuration, xi, 12, 17, 44, 46, 67, 80, 82, 172, 260, 289, 324, 325, 332, 335, 336, 384
173, 189, 195, 196, 197, 198, 199, 200, 201, crystalline solids, 21
202, 203, 205, 206, 209, 210, 213, 214, 260, crystallisation, 321
330, 343, 348, 350, 354, 375, 378, 379 crystallites, 279, 280, 284, 298, 382, 385
confinement, 96, 114, 341, 346, 348, 349 crystallization, 384
conservation, 197 crystals, xiii, 50, 171, 201, 214, 270, 341, 342,
constraints, 197, 198, 202 345, 386
construction, 9, 114, 115, 174, 344 cubic system, 47
consumption, 114, 142, 212, 284 cultivation, xii, 302, 304
contaminants, 383 culture, 209
contamination, 275, 277, 278, 287, 298 current limit, xi, 245, 247, 252, 253, 254, 255,
continuity, 69, 176, 219, 220 259, 260, 262, 263, 265, 268
cycles, xiv, 23, 188, 360, 361, 373
cycling, 19, 23, 266
Index 395
cyclotron, 72 diffusion region, 205

diffusivity, 287
diodes, 49, 62
D dipole, 190, 199, 307, 309, 311, 314
dipole moment, 199
damping, 157, 181, 184, 185, 187, 188, 191
discharges, x, 113, 114, 115, 118, 121, 122, 126,
data gathering, 146, 147
127, 143, 144, 146, 149, 150, 151, 152, 153,
database, 90
154, 157, 158, 162, 165, 377, 378, 385
decay, 66, 73, 100, 109, 135, 136
discretization, 233
decomposition, 9, 140, 306, 367, 368, 369, 370,
discs, 175
371
diseases, 6
defects, 3, 5, 43, 45, 48, 49, 57, 60, 61, 62, 180,
disinfection, 143
246, 247, 374
disorder, 171, 179, 183
deficiency, 52, 58, 257
dispersion, x, 2, 10, 169, 170, 171, 176, 177, 178,
definition, 191, 374
179, 180, 181, 182, 183, 184, 185, 186, 187,
deformation, 49, 353
188, 189, 190, 191
degenerate, 44, 80, 93
displacement, xiii, 353, 354, 356, 359, 360, 361,
degradation, 252, 253, 255, 303
362, 363
dendrimers, 11
dissociation, 13, 132
density, ix, xiv, 3, 9, 10, 11, 12, 14, 16, 18, 20,
dissolved oxygen, 7, 8
24, 30, 35, 43, 45, 46, 52, 63, 65, 66, 67, 68,
distillation, 371
69, 70, 71, 72, 73, 74, 75, 77, 78, 79, 82, 87,
distilled water, 21, 314
91, 94, 96, 101, 102, 107, 108, 109, 117, 121,
distress, 54
127, 132, 144, 145, 149, 198, 199, 203, 204,
distribution, 10, 24, 25, 34, 46, 52, 66, 71, 73, 93,
207, 208, 209, 234, 241, 248, 249, 250, 251,
130, 136, 140, 141, 162, 163, 179, 197, 201,
252, 254, 255, 256, 257, 258, 263, 343, 344,
210, 218, 219, 220, 230, 237, 238, 240, 304,
368, 373, 374, 375, 381, 382, 386
307, 313, 314, 315, 347, 348, 363
deposition, xii, xiv, 11, 17, 21, 23, 24, 28, 30, 34,
distribution function, 179, 218, 219, 220, 304
49, 50, 51, 53, 57, 60, 62, 64, 115, 266, 275,
divergence, 78, 82, 84, 136, 138, 161, 179
280, 285, 287, 319, 320, 321, 322, 324, 326,
diversity, 343
327, 329, 332, 337, 338, 345, 355, 373, 374,
DNA, 7, 302
375, 376, 377, 378, 379, 380, 381, 384, 386
domain structure, 182
deposition rate, xiv, 321, 373, 375, 376, 377, 378,
domain walls, 171, 178, 182, 183
379, 380, 381, 386
domestic policy, 366
derivatives, 304
donor, 49, 50, 53, 248, 251, 255, 256, 260, 263
desorption, 3, 254, 315
donors, 48, 49, 246, 248
destruction, 129, 371
dopant, 44, 50, 53, 323
detection, 9, 10, 15, 75, 77, 79, 81, 83, 90, 100
dopants, 49, 50, 247, 320, 337
deviation, 179, 207, 246, 254
doped, xii, 5, 6, 44, 45, 47, 50, 51, 53, 57, 60, 61,
diaphragm, 82, 84, 123, 148, 367
171, 260, 319, 320, 337, 347, 348, 349, 350,
dielectric constant, 49, 261, 346
354, 355, 356
dielectric materials, 169
doping, 44, 47, 48, 49, 50, 51, 52, 61, 62
dielectric permittivity, x, 169, 170, 171, 172, 174,
Doppler, 72, 73, 94, 163
175, 176, 177, 178, 179, 180, 181, 182, 183,
drainage, 201, 202
184, 186, 187, 251, 256
drugs, 302
dielectric strength, 181, 184, 185
drying, 306
differential equations, 231
durability, 17, 19, 23
diffraction, xii, 10, 27, 36, 50, 78, 82, 86, 273,
duration, 73, 77, 78, 82, 84, 107, 256, 275, 285,
274, 280, 281, 282, 283, 284, 289, 291, 293,
329, 356, 375
298, 319, 321, 322, 324, 325, 327, 336, 385,
dyes, xii, 13, 302, 303, 314, 315, 316, 317
386
dynamical properties, 64
diffusion, 4, 6, 23, 50, 70, 71, 127, 140, 198, 199,
202, 205, 208, 209, 277, 278, 326, 382
diffusion process, 140, 382
396 Index
249, 250, 251, 253, 254, 255, 256, 257, 259,

E 262, 267, 368, 378
elongation, 201
ears, 196
emission, ix, 10, 12, 23, 42, 51, 57, 58, 63, 64,
earth, 66, 201, 347, 349
65, 69, 71, 72, 73, 74, 75, 76, 78, 79, 81, 82,
EBSD, 291, 292, 298
83, 85, 86, 87, 88, 89, 90, 92, 93, 94, 96, 97,
ecological, 315
99, 100, 101, 102, 103, 104, 105, 106, 107,
econometric analysis, 216
108, 109, 150, 151, 152, 153, 154, 156, 157,
economic activity, 366
165, 212, 213, 216, 250, 254, 369, 371, 378
economic growth, 213
emission source, ix, 63, 109
economics, 214, 216
emitters, 35, 43, 73, 109, 143
education, 241
encapsulation, 11
elasticity, 190
endocrine, 205
election, 344
endocrine system, 205
electric arc, 87, 118
endothermic, 367, 368
electric charge, 65, 199
energy characteristics, 147
electric conductivity, 159
energy density, 65, 79
electric current, 366
energy supply, 366
electric energy, 367
energy transfer, 13, 16, 80
electric field, x, 5, 9, 10, 68, 69, 73, 74, 77, 78,
engines, 366, 367
86, 113, 116, 117, 118, 126, 127, 128, 129,
enlargement, 131
144, 145, 149, 150, 151, 157, 159, 162, 163,
entrapment, 268
175, 178, 185, 188, 189, 190, 191, 259, 264,
entropy, 198
266, 267, 268, 353, 356, 358, 359, 360, 361,
environment, xiv, 71, 204, 206, 253, 303, 323,
362, 363
365
electric potential, 16, 358
environmental conditions, 204, 205
electric power, 154, 156
environmental contaminants, 302, 317
electrical conductivity, xi, 48, 245, 256, 382
environmental contamination, 303
electrical power, 213
environmental protection, 5
electrical properties, 65, 245, 262, 266
environmental technology, 302, 304
electrical resistance, 261
enzymes, xii, 302, 303, 306, 316
electricity, 196
epitaxy, 49, 51, 61
electrodes, 13, 17, 18, 50, 118, 129, 157, 175,
epoxy, 280, 355
266, 267, 326, 355, 358, 359, 363, 367
EPR, 307
electroluminescence, 51
equilibrium, ii, 45, 66, 67, 73, 75, 76, 77, 95, 115,
electrolysis, 367
132, 133, 136, 137, 138, 139, 140, 196, 198,
electrolyte, 17, 21, 23
199, 211, 248, 251, 254, 255, 256, 257, 258,
electromagnetic, 10, 13, 53, 65, 68, 114
263, 314, 315, 316, 342, 345, 366, 374
electromagnetic fields, 65
equilibrium state, 139, 196, 198
electromagnetic wave, 68
erbium, 347, 348
electron beam, xii, 16, 17, 34, 35, 37, 38, 41, 45,
erosion, 376, 377, 378, 382, 385
46, 47, 61, 273, 275, 277, 285, 298, 355
ESR, xii, 301, 303, 310, 311, 312, 313, 314
electron charge, 20, 68
ESR spectra, 310, 311, 312
electron density, 65, 66, 67, 68, 69, 70, 94, 102,
ESR spectroscopy, xii, 301, 303
107, 108, 109
estimating, 76
electron diffraction, 34, 283, 289, 336
etching, 115, 275, 277, 344
electron gas, 10
ethane, 367, 368
electron microscopy, xii, 24, 27, 31, 32, 39, 40,
ethanol, 368
252, 280, 281, 289, 301, 303, 329, 334, 336
evacuation, 123
electronic materials, 60
evaporation, xii, 11, 17, 23, 34, 46, 47, 57, 59,
electronic structure, 12, 48, 57
60, 61, 78, 79, 273, 275, 277, 285, 355
electrons, 5, 10, 12, 16, 21, 35, 37, 39, 43, 45, 46,
47, 48, 52, 65, 66, 67, 68, 69, 70, 71, 79, 81,
94, 95, 98, 107, 108, 109, 127, 246, 247, 248,
Index 397
evolution, 10, 64, 66, 68, 97, 102, 105, 107, 108, flow rate, 11, 46, 48, 54, 123, 126, 127, 128, 129,
109, 184, 196, 199, 200, 201, 206, 207, 209, 130, 131, 132, 137, 138, 139, 140, 142, 143,
215, 267, 293, 298 200
evolutionary process, 198, 211, 213 fluctuations, 77, 78, 128, 132
excitation, 16, 64, 65, 69, 75, 79, 80, 89, 94, 100, FLUENT, 219, 233
107, 108, 151, 152, 157, 162, 309, 310 fluid, xi, 198, 203, 217, 220, 222, 223, 224, 225,
expansions, 173 227, 228, 233, 240, 241, 243, 304, 305, 311
experimental condition, 11, 52, 89, 122, 147, 161, fluorescence, ix, 1, 3, 8, 9, 10, 12, 15, 16, 57, 65,
287 79, 350
exposure, 382 fluorescent lamps, 65
external environment, 253 focused ion beam, 344, 345
extinction, 37, 38, 132 focusing, 65, 77, 81, 100
extraction, 16, 19, 37, 307, 309 foils, 23, 60
eye, 213 food, 196, 205, 303, 315
eyes, 196 forests, 215
fossil fuel, xiii, 213, 365, 366
fossil fuels, xiii, 365, 366
F fractals, 196
free energy, 72
fabricate, 50, 345, 348, 349
free radicals, 313
fabrication, xiii, 59, 341, 342, 344, 350
freedom, 69
failure, 4, 54
friction, 202, 204, 218, 221, 222, 223, 224, 225,
faults, 174
230, 236, 237, 239, 240
feedback, 360, 380
fuel, xiii, 3, 4, 16, 142, 213, 365, 366, 371
fermentation, 317
fuel cell, 16, 142, 366, 367
Fermi level, 21, 22, 248, 254, 264, 267
fulfillment, 172
ferrite, 118, 124, 125, 126, 131
funding, 241
ferroelectrics, 170, 171, 172, 173, 177, 182, 183,
fungi, 303
188, 191
furnaces, 120
ferrofluids, xii, 301, 302, 304, 316
fusion, 110
ferromagnetic, 114, 116, 118, 144, 145
FWHM, 73, 74, 77, 82, 179, 186
Feynman, 193, 213, 216
FIB, 344
fiber, 54, 346, 348, 370, 371 G
fiber membranes, 54
fibers, 54, 342, 346, 349 G8, 213
field-emission, 58 GaAs, 81
filament, 46 gallium, 81
film formation, 284 gas diffusion, 4
film thickness, 18, 20, 53, 218, 219, 220, 221, gas exchange, 6, 54
222, 223, 224, 225, 228, 229, 240, 241, 289, gas phase, 100, 101, 108
356, 358, 383 gas sensors, 16, 59
filters, 142, 143 gases, x, 7, 76, 113, 122, 126, 129, 368, 371
filtration, 302 gauge, 82, 355
financial support, 192, 350 Gaussian, 73, 78, 218, 219, 220
finite volume, 218, 225, 226, 228 GDP, 212, 213
finite volume method, 218, 225, 226, 228 gel, 12, 17, 51, 60, 61, 62, 204, 262, 266
flame, 87, 93 General Relativity, ii
flexibility, 349 generation, xi, xiv, 3, 6, 14, 44, 56, 68, 69, 70,
flight, x, 63, 105, 210 114, 115, 116, 118, 121, 122, 142, 163, 195,
flocculation, 303 198, 199, 200, 203, 213, 214, 216, 232, 240,
flow field, 242 350, 366, 367, 373, 374, 385
generators, x, 113, 114, 165
genetic code, 201
398 Index
germanium, 81 heart disease, 6

glass, xiii, 5, 11, 16, 17, 18, 24, 34, 37, 51, 341, heartbeat, 203
342, 343, 344, 345, 346, 347, 348, 349, 350, heat, x, 80, 113, 116, 118, 119, 120, 122, 123,
355 124, 127, 131, 132, 134, 136, 139, 141, 142,
glasses, xiii, 341, 342, 344, 345, 346, 347, 348, 158, 196, 197, 198, 203, 215, 243, 253, 255,
349, 350 367
global warming, 216, 366 heat capacity, 81
goals, 3, 211 heat loss, x, 113, 118, 119, 120, 127, 131, 132,
gold, xiii, 10, 21, 50, 65, 81, 319, 326, 327, 328, 158, 203, 215
329, 331, 338, 355, 356 heat transfer, 127, 203, 215
GPO, 273, 319, 353 heating, 18, 21, 29, 31, 35, 114, 120, 132, 175,
grades, 120 260, 261, 275, 294, 297, 330, 332, 374
grain, xi, 171, 201, 245, 246, 247, 248, 249, 250, heavy metal, xii, 302, 303, 315, 317
251, 252, 253, 254, 255, 256, 257, 258, 259, heavy particle, 108, 368
260, 261, 262, 263, 264, 265, 266, 277, 279, height, 30, 42, 66, 122, 179, 201, 218, 219, 222,
280, 284, 289, 291, 292, 321, 325, 326, 327, 223, 224, 225, 226, 230, 233, 240, 241, 246,
329, 332, 333, 335, 337, 363, 383 247, 250, 251, 252, 254, 256, 257, 258, 259,
grain boundaries, xi, 245, 246, 254, 256, 260, 260, 262, 263, 264, 267, 309, 359
262, 280, 289 height growth, 254
grains, 184, 245, 246, 247, 248, 262, 263, 277, heterogeneous, 2, 6
280, 289, 290, 298, 325, 326, 333, 334, 335, heterogeneous catalysis, 2
336, 382 high pressure, 96, 97, 150, 152, 158, 229
graph, 94 high resolution, 289, 298
graphite, ix, 63, 64, 82, 83, 86, 89, 93, 94, 99, high temperature, 49, 66, 84, 94, 111, 137, 138,
100, 101, 102, 103, 104, 105, 106, 108, 109 173, 252, 274, 287, 321, 332, 368, 384
gratings, 83 high-speed, 170
gravity, 203 high-tech, x, 113
greenhouse, xiv, 365, 366, 367, 371 hip, 81, 210, 212
grids, 220 Hm, 118
gross domestic product, 213 homogenous, 278, 287
groups, 118, 211, 213, 235, 317, 375, 376 host, 348
growth, xiii, 2, 3, 10, 11, 14, 17, 19, 27, 28, 29, hostile environment, 204
30, 34, 39, 45, 46, 50, 57, 59, 61, 62, 68, 69, household, 9
70, 71, 106, 188, 196, 197, 204, 205, 206, 210, HRTEM, 289, 291, 298
213, 215, 254, 284, 319, 321, 327, 329, 332, human, 3, 6, 7, 8, 9, 54, 56, 210, 216, 366
374, 375, 380, 382, 383, 385, 386 human behavior, 210
growth mechanism, 34 humans, 204, 210
growth temperature, 374, 382 humidity, 6, 60, 241, 259
hybrid, 16
hybrids, 185
H hydrides, 367
hydro, xiii, 55, 365, 367, 368
H2, xiv, 131, 143, 365, 367, 368, 370
hydrocarbon, 93, 280, 367, 368
HA, 339
hydrocarbons, xiii, 55, 365, 367, 368
haemoglobin, 3, 4, 6, 7, 8, 56
hydrochloric acid, 277, 283
halogen, 83
hydrodynamic, 218, 219, 221, 222, 223, 224,
hardness, 204, 206
241, 242
harm, 203
hydrodynamics, 111, 215, 216
harmony, 203
hydrofluoric acid, 275, 277, 285
harvest, 205
hydrogen, x, xiii, 3, 4, 6, 8, 15, 67, 74, 79, 86, 89,
health, 4, 315
113, 126, 131, 132, 142, 165, 306, 365, 366,
health care, 4
367, 368, 369, 370, 371
health problems, 315
hydrogen peroxide, 306
heart, 4, 6, 203
Index 399
hydrophilicity, 55 insertion, 19, 83

hydrosphere, 201 insight, 65, 89, 104, 310
hydroxide, 304, 306 inspection, 179, 186
hypothesis, 152, 165 inspiration, 210
hysteresis, 118, 119, 188, 189, 307, 308, 380 instabilities, 66, 132
hysteresis loop, 189, 308 instability, 115, 127, 128, 132, 366
instruments, 81
insulation, 215
I insulators, 44, 247
integration, 71, 72, 83, 95
IBIS, 356
interaction, 5, 16, 64, 73, 74, 77, 95, 150, 182,
ICU, 56
190, 207, 262, 277, 311, 356, 378
id, 62, 329
interactions, xiv, 13, 64, 71, 96, 182, 307, 309,
identification, 49, 274, 337
311, 314, 373, 375, 376
illumination, 143, 145, 146, 147, 284
intercalation, 16, 17, 18, 19, 20, 21, 22, 23, 37
images, 42, 81, 278, 281, 284, 289, 298, 305,
interface, xi, 3, 6, 50, 175, 217, 218, 219, 226,
306, 333, 335, 354, 383
241, 246, 267, 274, 275, 278, 281, 287, 289,
imaging, xiii, 9, 284, 289, 353, 359, 363
291, 298, 374, 375, 382
implants, 4
interference, 148, 290
implementation, xiv, 134, 370, 373
internal combustion, 366, 367
impurities, 3, 180, 246, 275, 369, 370
international relations, 366
IMS, 277
interpersonal contact, 209
in situ, xii, 54, 273, 293, 297, 298, 332
interphase, 182
in transition, 59
interstitial, 48
in vitro, 4, 7
interstitials, 48
in vivo, 4
interval, 83, 104, 106, 175, 177, 178, 182, 186,
incentive, xiii, 365, 374
203, 309, 368
incidence, 102, 105, 242, 282
intravascular, 54
income, 210
intrinsic, 45, 48, 51, 52, 72, 171, 386
income distribution, 210
ion beam, 344, 345, 346
incompressible, 233, 243
ion bombardment, 383
incubation, 311, 314
ion implantation, xiii, 341, 342, 344, 345, 346,
incubation time, 314
347, 348, 349, 350
independence, 234
ionic, 2, 12, 46, 64, 65, 67, 68, 72, 74, 75, 79, 86,
indexing, 291, 292, 293
87, 92, 95, 97, 100, 102, 103, 104, 105, 106,
indication, 29, 106, 309
107, 108, 109, 303, 304
indices, 348
ionization, x, xiv, 46, 63, 64, 67, 68, 69, 70, 71,
indium, 16, 44, 45, 46, 47, 48, 60, 61
72, 76, 77, 79, 94, 95, 109, 250, 251, 256, 373,
indium tin oxide (ITO), 16
374, 375, 377, 378, 379, 380, 384, 386
induction, x, 113, 114, 116, 117, 118, 119, 120,
ionization energy, 69, 70, 250, 251, 256
141, 143, 146, 147, 156, 157, 163, 164, 165
ionization potentials, 46, 71, 94
inductor, 114, 120, 143
ionosphere, 66
industrial, ix, 2, 4, 142, 162, 230, 367, 370
ions, xii, 16, 21, 35, 47, 65, 66, 67, 68, 69, 70, 71,
industrialized countries, 213
73, 77, 94, 95, 107, 108, 109, 171, 302, 303,
industry, x, xiv, 2, 44, 65, 113, 142, 143, 274,
314, 315, 317, 342, 343, 344, 346, 347, 348,
302, 365, 366
349, 350, 368, 374, 376, 377, 378, 379, 384
inelastic, 69, 95
IR spectra, 160, 162
inert, 122, 245, 323, 326, 330, 368, 374
iron, xii, 4, 301, 304, 307, 316
inertness, 346
irradiation, 34, 35, 56, 82, 103, 145, 344, 348,
infinite, 16
383, 385
infrared, ix, 56, 63, 64, 80, 81
island, 24, 25
injection, 16, 21, 266, 267
isothermal, 141
inorganic, xii, 16, 302, 317
isotherms, 314, 316
insects, 204, 205, 208
400 Index
isotropic, 225 light scattering, 10

isotropy, 231 light-emitting diodes, 62
iteration, 220 lighting systems, 143
ITO, 16, 17, 21, 24, 44, 53 limitation, 108
limitations, xiii, 3, 7, 28, 142, 173, 353, 356
linear, xiii, 64, 73, 74, 80, 90, 91, 148, 205, 239,
J 252, 253, 262, 341, 356, 358, 377
linear dependence, 377
Joule heating, 260
liquid nitrogen, 175
judo, 204
lithium, 37, 50, 58
liver, 4
K loading, 121
localization, 12, 45, 46
K+, 347 location, 30, 234
Kant, 214 locomotion, 203, 204
kidneys, 202 losses, x, 71, 79, 82, 113, 115, 118, 119, 120,
kinetic energy, 10, 68, 69, 79, 108, 231, 241, 243 127, 131, 132, 158, 178, 184, 203, 345, 346,
kinetics, 2, 3, 4, 10, 11, 14, 17, 23, 134, 135, 374 347, 348
low power, 68, 342
low temperatures, 10, 51, 53, 74, 170, 175, 309,
L 313
low-temperature, x, 61, 113, 114, 117, 121, 135,
L1, 147, 148 140, 141, 142, 165, 349
L2, 117, 147, 148 lubrication, xi, 217, 218, 219, 220, 221, 222, 224,
LA, 56, 57, 85, 88, 92, 96, 98, 192 225, 241, 242
lamina, xi, 217, 219, 220, 225, 229, 230, 235, lumen, 54
240, 241 luminescence, 11, 23, 35, 42, 43, 60
laminar, xi, 217, 218, 219, 220, 225, 229, 230, lung, ix, 1, 3, 4, 6, 7, 54, 56, 202
235, 240, 241 lung disease, 4, 6
land, 200 lung function, 6
landfills, xiv, 365 lungs, 4, 6, 54, 197, 202, 215
Langmuir, 315, 316 lysozyme, 316
lanthanum, 320
laser, ix, xiii, 11, 49, 50, 56, 57, 62, 63, 64, 65,
68, 69, 70, 71, 72, 77, 78, 79, 80, 81, 82, 83, M
84, 85, 86, 87, 88, 89, 90, 91, 92, 94, 95, 96,
97, 100, 102, 103, 104, 105, 106, 107, 108, macromolecules, 305
109, 114, 298, 341, 342, 346, 349, 354 maghemite, xii, 301, 302, 303, 304, 305, 307,
laser ablation, ix, 17, 63, 64, 82, 100, 102, 104, 311, 313, 314, 316, 317
105, 108 magnetic, vi, xii, 2, 46, 51, 52, 61, 72, 114, 116,
Laser Induced Breakdown Spectroscopy (LIBS), 117, 118, 119, 120, 122, 123, 124, 126, 127,
63, 64, 65, 71, 74, 77, 78, 79, 81, 82, 84, 102, 129, 146, 160, 241, 301, 302, 303, 304, 305,
109 306, 307, 308, 309, 310, 311, 314, 316, 317,
laser radiation, 64 374
lasers, 35, 65, 79, 80, 106, 114, 347 magnetic field, 46, 51, 52, 72, 117, 118, 119, 120,
lattice, 5, 16, 24, 34, 45, 47, 49, 50, 171, 246, 302, 303, 304, 305, 307, 308, 310, 311, 316,
283, 289, 294, 326, 328, 332, 374 374
leakage, 266, 267 magnetic fluids, 303, 304
LED, 49, 51 magnetic materials, 2, 118
lens, 78, 82, 100, 146, 147 magnetic particles, 304, 305
lenses, 53, 148 magnetic properties, xii, 2, 61, 118, 301, 302,
lifetime, 73 303, 307, 316, 317
ligands, 11 magnetic relaxation, 309
light beam, 81 magnetic structure, 310
Index 401
magnetism, 2, 196, 303, 307, 316 methane, xiv, 142, 365, 367, 368, 371
magnetization, xii, 301, 303, 307, 308, 313, 314 methanol, 305, 306, 368, 369, 370
magnetoresistance, 55 microbial, xii, 301, 302, 303, 305, 306, 307, 309,
magnetron sputtering, xii, xiv, 24, 27, 34, 46, 50, 311, 313, 314, 315
58, 60, 273, 275, 285, 319, 320, 321, 373, 374, microbial cells, xii, 301, 302, 303, 305, 306, 307,
375, 383, 386 313, 314, 315
magnets, xii, 2, 301, 305, 306, 317 microelectronics, x, 2, 113
mainstream, 201, 375 microorganism, 303
maintenance, 4, 80, 118, 127, 210 microorganisms, 303
mammals, 203, 215 microscope, xiii, 305, 311, 353, 354, 359, 363
management, 4 microscopy, xii, 24, 27, 30, 32, 39, 40, 41, 42, 60,
manifold, 196 252, 273, 274, 277, 280, 281, 284, 289, 298,
manipulation, 302 301, 303, 319, 325, 329, 334, 336, 348, 354,
man-made, 65 356, 358
manners, 148 microstructure, xiii, xiv, 50, 171, 319, 321, 337,
mapping, 354, 357 373, 375, 382, 384, 386
market, 342 microstructures, 30, 382, 384
mask, 171, 274, 345, 348, 355 microwave, x, 68, 113, 169, 170, 171, 172, 173,
masking, 345, 346, 355 176, 178, 182, 183, 187, 191, 311, 369
mass spectrometry, 50, 277 migration, 267
material surface, 15, 342 military, 80
mathematicians, 196 mitochondria, 215
matrices, 194 mobility, 44, 45, 47, 48, 50, 51, 52, 53, 61
matrix, 24, 83, 171, 182, 385, 386 modeling, xi, 115, 142, 241, 242, 245
MBE, 49 models, x, xi, 54, 118, 144, 157, 158, 169, 171,
measurement, xiii, 8, 12, 21, 30, 45, 46, 52, 53, 197, 215, 225, 226, 231, 232, 233, 242, 243,
78, 107, 124, 126, 144, 147, 148, 153, 163, 245, 247, 259, 262, 266, 342, 382
171, 173, 175, 184, 185, 189, 191, 234, 253, modulation, 11, 12, 18, 19, 44, 289, 311, 347
260, 267, 268, 293, 294, 296, 297, 314, 353, modules, 349
354, 356, 358, 359 modulus, 172
measures, 219 mole, 136, 367, 368
mechanical stress, 188, 191 molecular beam, 49, 61
media, xii, 118, 201, 215, 302, 304, 317 molecular beam epitaxy, 49, 61
melt, 94 molecular biology, 202
melting, 307, 309, 382 molecular oxygen, 132, 133
melting temperature, 382 molecular weight, 302
membranes, 4, 54 molecular-beam, 51
memory, 14, 16, 35, 320 molecules, 5, 10, 13, 66, 68, 69, 71, 73, 80, 94,
MEMS, 218, 274 108, 109, 111, 132, 140, 201, 215, 259, 368,
mercury, x, 114, 115, 143, 144, 145, 149, 150, 369, 371, 381
151, 152, 153, 154, 155, 156, 157, 158, 165, molybdenum, 81
314 momentum, 52, 76, 80, 219, 233, 243
mercury, 149, 167 monochromator, 81, 146
metabolic, 54, 203, 211, 212, 215 monolayer, 315
metabolic rate, 203, 211, 215 monolayers, 283
metabolism, 211, 213, 215 morbidity, 4
metal ions, 316 morphological, 17, 35, 43, 383
metal organic chemical vapor deposition, 62 morphology, xii, 3, 10, 24, 25, 29, 30, 31, 39, 41,
metal oxide, ix, 1, 2, 3, 4, 11, 17, 19, 21, 44, 46, 60, 204, 260, 273, 277, 287, 289, 298, 323,
59, 380, 385, 386 325, 326, 382, 383, 385
metal oxides, ix, 1, 2, 4, 19, 21, 44, 46, 380 mortality, 4
metallurgy, x, 113 motion, x, 73, 169, 171, 178, 180, 182, 196, 198,
metals, 2, 10, 49, 56, 287, 295, 315, 329, 378 207, 209, 231
402 Index
motivation, xii, 319 NIST, 72, 75, 87, 89, 90, 111
movement, 21, 196, 206, 208 nitride, xiv, 50, 115, 277, 354, 359, 373, 375, 386
MPI, 68, 69, 71, 95 nitrogen, 50, 104, 115, 135, 136, 138, 139, 140,
MRS, 62, 351 141, 165, 175
multidisciplinary, 320, 371 noble metals, 10
multiplication, 81 noise, 45, 81, 83, 108, 294, 360
multiplier, 146, 147 nonequilibrium, 197, 214
mutagenic, 303 nonequilibrium systems, 197
nonlinearities, 342
non-Newtonian fluid, 241
N nonstoichiometric, 39
non-thermal, 95
NA, 193, 194
non-uniform, 18
Na+, 16, 347
non-uniformity, 18
NaCl, 81, 82
normal, 39, 80, 83, 147, 182, 188, 191, 234, 235,
nanocrystalline, 59, 327, 332, 333, 337
304, 359
nanocrystals, 335, 336
normal distribution, 304
nanodevices, 274
n-type, xi, 12, 49, 50, 51, 246, 247, 273, 274,
nanomaterials, 27
275, 284, 285, 293, 294, 295, 296, 298
nanometer, 13
nuclear, 46, 213, 343, 345, 346
nanometers, 43
nuclear charge, 46
nanoparticles, xii, 3, 5, 10, 15, 34, 42, 43, 56, 57,
nuclear power, 213
59, 301, 302, 303, 304, 305, 306, 307, 311,
nuclear power plant, 213
313, 316, 317
nucleation, 3, 10, 14, 34, 182, 382
nanorods, 23, 24, 30, 34, 43, 58
nucleic acid, 302
nanoscale materials, 36
numerical analysis, 242
nanostructured materials, 53
nutrient, 204, 205, 206
nanostructures, 3, 23, 58, 274
nutrients, 204, 205, 206
nanotechnology, 3, 274
nanotube, 23, 35
nanotubes, 23, 43 O
nanowires, 15, 21, 23, 24, 25, 27, 28, 29, 30, 34,
35, 43, 58, 59 observations, 9, 84, 124, 207, 209
NASA, 230 observed behavior, 170, 180, 189, 307
National Academy of Sciences, 214 oceans, 366
natural, xi, xiii, 66, 69, 73, 123, 126, 142, 165, Ohmic, 247, 248, 262, 266
172, 195, 196, 197, 198, 215, 365, 366, 368, oil, xiii, 114, 117, 289, 365, 366
371 oils, 303
natural gas, 123, 126, 142, 165, 368, 371 oligonucleotides, 302
natural science, 198 one dimension, 23, 342
natural sciences, 198 ophthalmic, 53
natural selection, 196 optical, ix, xiii, 1, 3, 5, 7, 8, 9, 11, 13, 14, 16, 17,
Navier-Stokes equation, xi, 217, 218, 219, 225, 18, 19, 20, 35, 43, 45, 47, 48, 49, 50, 56, 57,
228, 232, 234 58, 60, 61, 69, 76, 78, 79, 81, 82, 83, 105, 108,
Nd, 11, 347 143, 145, 157, 170, 293, 305, 341, 342, 343,
negativity, 46 344, 345, 346, 347, 348, 350, 374, 378
neon, x, 114, 115, 144, 146, 147, 159, 160, 161, optical absorption coefficient, 20
162, 163, 164, 165 optical coatings, 5, 9
nervous system, 203 optical density, 17
nickel (Ni),, xi, 67, 77, 79, 94, 95, 273, 274, 284, optical detectors, 145
285, 286, 287, 288, 289, 290, 291, 292, 293, optical fiber, 342
294, 295, 296, 297, 298, 348, 351 optical properties, 11, 20, 47, 48, 56, 57, 58, 61,
niobium, 320 342
NIR, ix, 63, 86, 109 optical pulses, 79
Index 403
optical transmission, 50 particles, 5, 6, 10, 14, 25, 26, 56, 57, 64, 65, 66,
optics, 9, 10, 82 71, 72, 73, 74, 100, 121, 142, 199, 304, 305,
optimization, 35, 43, 45, 196, 197, 198, 199, 200, 307, 309, 311, 313, 314, 315, 368, 376
201, 214, 371 partition, 75
optimization method, 45 passive, 342, 349, 350
optoelectronic, 44, 48, 49 patents, 115
optoelectronic devices, 44, 48, 49 patients, 7, 54
orbit, 69 patterning, 355
ores, 120, 124, 125, 126, 131, 144 Pb, 266, 267, 320, 323, 324
organ, 4, 6, 23, 208 pedestrian, 207, 208, 209, 214, 216
organic, xii, xiii, 5, 13, 16, 51, 62, 302, 303, 314, pedestrians, 196, 202, 207, 208, 209
316, 317, 341, 342, 365 peptide, 11
organic compounds, xiv, 303, 314, 365 periodic, xi, 16, 46, 57, 148, 217, 231, 233, 235,
organic polymers, 342 237, 383
organism, 210 periodic table, 46
organometallic, 23 periodicity, 21
orientation, x, xii, 25, 36, 169, 170, 185, 187, permeability, 198
188, 189, 208, 225, 273, 283, 284, 289, 290, permit, 66
291, 292, 293, 295, 297, 298, 322, 324, 325, permittivity, x, 169, 170, 171, 173, 174, 175, 176,
326, 327, 329, 330, 331, 332, 336, 361, 382, 177, 178, 179, 180, 181, 182, 183, 184, 186,
385 187, 191, 199, 248, 251, 255, 256, 260, 261,
orientation state, 188 263
orthorhombic, 281, 293 perovskite, x, xii, 169, 171, 172, 177, 179, 180,
oscillation, 68 182, 260, 319, 320, 321, 322, 324, 326, 327,
oscillations, 10, 80, 309 329, 330, 332, 336, 337, 338
oscillator, 71, 181 peroxide, 14, 57, 306
oxidation, 2, 4, 5, 7, 10, 11, 14, 23, 39, 50, 59, perturbation, 366
62, 303, 382 Petri dish, 206
oxide, ix, xi, xiii, xiv, 1, 2, 3, 9, 10, 11, 12, 13, petroleum, 242
14, 15, 16, 17, 18, 19, 21, 22, 23, 24, 27, 30, PF, 111
34, 35, 36, 37, 38, 40, 43, 44, 45, 46, 47, 48, pH, 7, 8, 303
53, 56, 57, 58, 59, 60, 61, 135, 136, 245, 246, phase transitions, 170, 211, 260
247, 259, 263, 274, 277, 278, 280, 281, 283, pheromone, 205, 215
285, 298, 304, 307, 320, 337, 341, 342, 345, phonon, 49, 52, 294
346, 349, 350, 373, 375, 385 phonons, 293
oxide clusters, 11 phosphate, 343, 347, 348, 350
oxide nanoparticles, 43, 57, 59, 304 phosphate glasses, 347
oxides, ix, xii, 1, 2, 3, 4, 10, 11, 12, 16, 19, 21, phosphorous, 50
44, 46, 52, 53, 301, 316, 319, 320, 380 photocatalysis, 4, 5, 6, 8
oxygen plasma, 51, 58, 62 photocatalysts, 3, 4, 5, 56
oxygenation, 4, 6, 7, 8, 202 photocorrosion, 7
ozone, 123, 132, 133, 134, 135, 165, 166 photodegradation, 303
photoelectron spectroscopy, xii, 319, 323, 327
photoexcitation, 12
P photographs, 42
photoionization, 72, 79
pacing, 24, 283
photoluminescence, 13, 42, 56, 57, 59, 60
paints, 5
photon, 14, 68, 69, 71, 72, 73, 77, 78, 79, 82, 83,
parabolic, 81, 243
84, 86, 103
paramagnetic, 313
photonic, xiii, 56, 341, 346, 349
parameter, 67, 74, 107, 148, 158, 179, 183, 188,
photonic devices, 56
225, 230, 237, 309, 315
photonics, xiii, 9, 81, 341, 342
Pareto, 210, 216
photons, 66, 68, 69, 95
particle collisions, 95
404 Index
photooxidation, 55 population density, 71, 368

photosynthetic, 366 pores, 25
photovoltaic, 367 porosity, 203, 252, 383
physical chemistry, 59 porous, 23, 198, 201, 215, 382
physical properties, 10, 35, 42, 49, 65, 77 porous media, 201, 215
physical sciences, ix portability, 53
physicists, 66 ports, 211
physics, ii, ix, xi, 1, 12, 44, 65, 110, 111, 195, powder, 59, 175, 291, 293, 327
197, 199, 202, 209, 213, 214, 375, 378 powders, 23
physiological, 215 power plant, 201, 211
piezoelectric, xiii, 2, 170, 171, 177, 184, 189, power plants, 211, 213
190, 191, 320, 322, 326, 338, 353, 354, 355, power-law, 377
356, 357, 358, 359, 360, 361, 362, 363 powers, 11, 115, 142
piezoelectric properties, 170, 320 precipitation, 305
piezoelectricity, xii, 319, 320 predators, 204
PL spectrum, 43 prediction, 143, 343
planar, 58, 246, 342, 344, 345, 346, 347, 349, 350 preference, 332
Planck constant, 78 press, 214, 216, 270, 318
plane waves, 78 probability, 68, 69, 71, 72, 73, 75, 95, 205, 218,
plankton, 366 219, 220, 264, 377, 378, 379, 380, 386
plants, 195, 196, 203, 204, 211, 213 probability distribution, 218, 219, 220
plasma physics, 110 probe, 8, 21, 23, 47, 122, 123, 124, 130, 131,
plasmatrons, x, 113, 124, 142 134, 136, 137, 138, 139, 140, 142, 172
plasmons, 9, 10 production, x, xiii, 3, 7, 23, 44, 109, 113, 132,
plastic, 53 134, 136, 138, 142, 143, 165, 320, 342, 344,
plastics, 303 365, 366, 367, 368, 370, 371
platelets, 39, 40 productivity, 124
platinum, xiii, 21, 294, 297, 319, 323, 326, 327, program, 76
328, 330, 331, 338, 355 propagation, x, xiii, 63, 64, 83, 84, 100, 101, 109,
play, 13, 79, 80, 191, 219, 229, 240, 247, 371 205, 206, 341, 342, 344, 346, 347, 348, 349
PLD, 11, 50 propane, 87, 368
ploughing, 277 proportionality, 203
PMI, 95 proposition, 29, 30
point defects, 48, 61, 180, 246 protection, 5, 247, 374
point-to-point, 196 protective coating, 374
poisoning, 39 proteins, 302
Poisson equation, 248, 249 proteolytic enzyme, 316
polarity, 260 protocols, 8, 366
polarization, x, 2, 169, 170, 178, 180, 185, 186, protons, 21
187, 188, 189, 190, 262 pseudo, 48
pollutant, 366, 371 p-type, xi, 12, 44, 45, 49, 50, 51, 52, 61, 62, 273,
pollutants, 5 274, 293, 296, 297, 298
pollution, xiii, 4, 6, 315, 365 pulse, 11, 64, 65, 68, 77, 78, 81, 82, 83, 86, 95,
polycrystalline, x, 169, 188, 245, 246, 247, 248, 102, 103, 104, 105, 106, 107, 114, 375, 378,
260, 283, 293, 297, 298, 384 379, 381, 382, 383
poly-crystalline, 324 pulsed laser, ix, 11, 17, 50, 57, 60, 63, 64, 65, 79,
polymer, 57 82, 100, 102, 105, 109, 342
polymers, 342 pulsed laser deposition, 11, 50, 57, 60, 64
polyphenols, 303 pulses, xiv, 79, 106, 373
polyps, 204 pumping, 106, 203, 285, 350
poor, 53, 78, 108, 324, 348, 374 pumps, xiii, 341
population, 4, 6, 66, 67, 71, 76, 80, 152, 163, purification, 303, 371
164, 368, 379 PVA, 24
Index 405
receptors, 302
Q recognition, 209
recombination, 5, 10, 68, 69, 70, 71, 72, 77, 101,
QDs, 13
107, 108, 132, 370
quality of life, 210
recombination processes, 108
quanta, 80
reconcile, 205
quantitative estimation, xiii, 353
rectification, 51
quantum, 5, 6, 7, 10, 13, 57, 67, 89, 93, 176, 177,
recycling, 366
196
red blood cells, 4
quantum dots, 13, 57
red shift, 37
quantum mechanics, 196
redistribution, 152, 189
quartz, 6, 51, 82, 144, 145, 146, 327
refining, 292
quasi-equilibrium, 136
reflection, 53, 148, 172, 173, 341
reflectivity, 9, 81, 82
R refractive index, 5, 37, 53, 341, 342, 343, 344,
345, 346, 347, 348, 349
R&D, 366 refractive indices, 348
radial distribution, 71, 124, 162, 163 regression, 90
radiation, x, 13, 45, 53, 64, 65, 71, 72, 73, 81, 84, regular, 39, 89, 279, 280, 298, 333, 335, 362
114, 145, 146, 147, 148, 152, 153, 154, 156, relationship, 71, 160, 210, 211, 212, 213, 219,
157, 161, 163, 203, 282, 320, 369 260
radio, 51, 114, 143, 247 relationships, xi, 118, 133, 195, 213
radionuclides, 302 relaxation, x, xii, 64, 86, 89, 97, 169, 170, 171,
radius, 67, 71, 78, 82, 130, 136, 145, 147, 148, 179, 180, 181, 184, 186, 187, 188, 191, 302,
157, 159, 160, 174, 200, 324, 349 309, 314, 317
rail, 211 relaxation process, 89, 179
rain, 200 relaxation processes, 179
Raman, xii, 9, 11, 13, 56, 57, 58, 65, 273, 284, relaxation time, 179, 181
293, 294, 295, 296, 297, 298 relaxation times, 179
Raman scattering, 56 reliability, 23, 247, 270
Raman spectra, 293, 294, 295, 296, 297, 298 renal disease, 4
Raman spectroscopy, xii, 11, 57, 65, 273, 284, renormalization, 211, 232
293 reparation, 335, 355
random, 208, 218, 220, 231, 382 reproduction, 210
range, x, 2, 10, 11, 18, 23, 24, 34, 37, 45, 49, 51, reserves, 366
65, 66, 71, 74, 80, 81, 82, 86, 93, 95, 97, 102, residuals, 220
104, 108, 113, 114, 117, 120, 123, 126, 136, resistance, xi, 2, 45, 46, 53, 117, 118, 120, 121,
140, 144, 145, 146, 147, 148, 149, 150, 151, 160, 197, 198, 199, 200, 201, 202, 203, 205,
152, 153, 154, 156, 158, 160, 163, 165, 169, 206, 209, 247, 252, 253, 260, 261, 273, 274
170, 171, 172, 174, 175, 176, 177, 178, 179, resistive, 175, 246
182, 184, 237, 265, 268, 274, 294, 307, 309, resistivity, xi, 23, 49, 50, 51, 52, 53, 62, 198, 208,
327, 347, 348, 358, 377 260, 261, 262, 273, 274, 275, 284, 293
RAS, 113 resolution, 72, 83, 85, 86, 87, 88, 89, 92, 94, 96,
ratings, 121 98, 102, 103, 104, 105, 106, 107, 145, 146,
raw materials, xii, xiii, 302, 305, 317, 365, 366, 147, 163, 281, 289, 298, 359
368 resonator, 80
Rayleigh, 201 resources, 210
reaction mechanism, 16 respiratory, 4, 54, 202, 203, 205
reactivity, 45, 46 respiratory failure, 4, 54
reading, 268 response time, 23
reagent, 125 Reynolds, xi, 61, 217, 218, 219, 220, 221, 222,
reagents, 123, 124 223, 224, 225, 226, 229, 230, 231, 233, 234,
reality, 205 236, 237, 240, 241, 242, 269
406 Index
Reynolds number, 219, 230, 233, 236, 237, 240 search, 11, 206
rhombohedral, 320, 324, 326, 327, 331, 332 seed, 337, 338
rhythm, 213 selected area electron diffraction, 34
rhythms, 213 selecting, 75
rings, 21, 78, 148, 283 self-organization, 209
river basins, 196, 201, 209, 210, 214 SEM, xii, 24, 25, 26, 28, 29, 30, 34, 39, 40, 41,
rivers, 195, 209, 214 273, 289, 305, 334, 384
RNA, 302 SEM micrographs, 384
robustness, 347 semiconductor, x, 5, 14, 45, 49, 81, 113, 247,
rods, 16, 18, 25, 196 248, 265, 269, 274, 344
Roman Empire, 210 semiconductors, xiii, 2, 44, 46, 49, 53, 61, 62,
room temperature, xii, 48, 51, 53, 55, 66, 170, 118, 245, 246, 247, 341, 342
174, 176, 179, 184, 186, 187, 252, 253, 262, sensing, 23, 53, 60
266, 302, 305, 311, 314, 315, 317, 320, 326, sensitivity, 53, 65, 81, 100, 145, 289, 320
327, 332, 333, 355, 383, 384, 385, 386 sensors, 2, 8, 9, 15, 16, 57, 59, 269, 320
room-temperature, 386 separation, xii, 70, 230, 237, 239, 264, 301, 302,
roughness, 11, 218, 219, 220, 222, 223, 224, 225, 303, 304, 305, 306, 307, 314, 316, 317
240, 241, 242, 275, 276, 277, 279, 289, 291, series, 75, 78, 86, 122, 154, 163, 214, 256, 285,
298, 325, 326, 330, 333, 383 368
rutile, 47, 384, 385, 386 shape, xi, 10, 72, 73, 76, 78, 82, 84, 102, 103,
123, 127, 163, 195, 196, 197, 198, 199, 200,
201, 204, 205, 206, 210, 213, 214, 215, 222,
S 228, 230, 240, 242, 266, 306, 320, 335, 367
sharing, 24
Saccharomyces cerevisiae, 303, 317
shear, 190, 191, 196, 227, 228, 229, 235, 241,
safety, 366
243
sample, 18, 19, 21, 23, 35, 36, 37, 38, 39, 41, 42,
shear rates, 227, 229
43, 51, 52, 64, 65, 77, 78, 96, 124, 134, 137,
shock, 84, 108
142, 170, 172, 173, 174, 175, 177, 186, 187,
shock waves, 84
252, 253, 254, 256, 259, 266, 278, 282, 291,
short-range, 171
293, 294, 297, 305, 311, 335, 343, 344, 355,
shoulder, 38
356, 358, 359
sign, 52, 74, 267
sampling, 64, 65, 123, 124, 134, 142, 143, 356,
signals, 13, 81, 205, 274, 311, 313, 354, 356
359
signal-to-noise ratio, 81, 108
sand, 201
silica, 342, 346, 349, 350
sapphire, 50, 61, 62
silica glass, 349
satellite, 93
silicate, 175, 343, 344, 347, 348
saturation, xi, 7, 16, 18, 19, 20, 119, 245, 252,
silicate glass, 347
263, 264, 265, 268, 309
silicon, xi, xiii, 12, 24, 58, 81, 242, 273, 274,
sawdust, xii, 301, 302, 303, 305, 306, 307, 308,
275, 277, 278, 280, 281, 282, 283, 284, 285,
309, 310, 311, 312, 313, 314, 315, 316, 317
287, 288, 289, 290, 291, 293, 294, 295, 296,
scaling, ix, 2, 201, 203, 204, 210, 211, 212, 213,
297, 298, 319, 323, 326, 329, 330, 332, 335,
214, 215, 216
337, 338, 346, 348, 354, 355, 356, 358, 359,
scaling law, 201, 210, 211, 212, 213, 214, 216
360
scaling relations, 210, 212, 213
silicon dioxide, xiii, 319, 329, 330, 332, 335, 337,
scaling relationships, 213
338, 346, 354, 358, 360
Scanning electron, 40
silver, 3, 9, 10, 11, 12, 13, 14, 15, 16, 45, 46, 47,
scanning electron microscopy, xii, 24, 27, 39,
56, 57, 61, 355
252, 301, 303, 334
simulation, 76, 230, 234, 237, 241, 288, 291, 292
scanning tunneling microscopy, 60
simulations, 76, 234, 291
scatter, 291
sine, 359
scattering, 10, 44, 52, 53, 56, 288
single crystals, 171
Schottky, 245, 246, 247, 250, 255, 267
sintering, 245, 260
Schottky barrier, 245, 246, 247, 250, 255
Index 407
SiO2, 330, 337, 359 steady state, 120

sites, 3, 18, 19, 20, 43, 47, 204, 315, 345 steel, 119, 122, 124
skeleton, 196 stiffness, 189, 190
skin, 70, 157, 159, 160 stochastic, 218, 220, 242
SME, 241 stochastic model, 218, 220
software, 146, 147, 219, 233, 354, 356 stoichiometry, xii, 18, 19, 142, 274, 288, 298,
soil, 201 319, 320, 324
solar, x, 5, 9, 16, 113, 366, 367 storage, 2, 10, 13, 17, 35, 43, 306, 366, 374
solar cell, x, 5, 16, 113 strain, 28, 29, 34, 41, 53, 185, 188, 190, 227, 228,
solar cells, 5, 16 232, 316
solar energy, 367 streams, 197, 200, 201, 208, 209, 215
sol-gel, 17, 51, 60, 62, 262, 266 strength, 2, 19, 20, 71, 72, 73, 117, 118, 126, 127,
solid phase, 315 128, 129, 143, 144, 145, 149, 150, 151, 157,
solid state, 342 158, 159, 160, 161, 162, 181, 184, 185, 187,
solidification, 201 188, 358, 382
solid-state, 17 stress, 183, 184, 185, 186, 187, 188, 189, 190,
solubility, 347 191, 231, 232, 235, 241, 330, 353, 356
soot, 142 stretching, 11, 80
sorption, 315 strikes, 81
spatial, 10, 115, 146, 147, 163, 208, 214 strong interaction, 311
species, ix, xiv, 7, 44, 49, 63, 64, 65, 66, 68, 69, strontium, xii, 319, 320, 323, 337, 354, 356
70, 71, 74, 75, 86, 94, 95, 98, 100, 101, 102, structural modifications, 42
103, 104, 105, 106, 108, 109, 202, 203, 204, structure formation, 385
278, 324, 326, 342, 368, 369, 370, 373, 374, substances, xiv, 115, 124, 142, 303, 314, 365,
375, 376, 377, 378, 379, 380, 382, 384, 386 368, 371
specific heat, 118, 119, 120, 158 substitutes, 4, 50
spectral component, 81 substitution, 133, 320
spectrophotometric, xii, 301, 314 substrates, xi, xiii, 21, 24, 34, 50, 51, 53, 58, 59,
spectroscopic methods, 79, 109 61, 273, 284, 285, 296, 297, 298, 319, 321,
spectroscopy, ix, xii, 11, 57, 63, 64, 65, 71, 79, 323, 326, 330, 332, 342, 343, 355, 356, 359,
81, 110, 111, 146, 147, 163, 273, 274, 277, 374, 375, 385
284, 293, 298, 301, 303, 307, 319, 327, 330, suffering, 4
378 sugar, 306
spectrum, 15, 37, 38, 39, 43, 45, 51, 61, 64, 76, sulfur, 115
85, 86, 89, 93, 96, 102, 103, 104, 105, 109, superconducting, 2
145, 146, 148, 151, 152, 153, 173, 179, 181, superimposition, 225
184, 187, 232, 243, 288, 293, 294, 297, 369 supernatant, 305, 314
speed, 17, 19, 65, 78, 163, 170, 204, 205, 206, superposition, 33, 106
207, 208, 209, 213, 215, 219, 222, 224, 240 supply, 6, 14, 116, 123, 124, 127, 128, 130, 134,
speed of light, 78 202, 366, 367
spin, 2, 51, 62 suppression, 24, 27, 34, 39, 106, 383
spintronic devices, 2 surface area, 5, 7, 23, 54, 203
sputtering, xiii, xiv, 11, 12, 17, 19, 24, 28, 34, 39, surface component, 98
45, 46, 49, 59, 61, 266, 275, 277, 319, 321, surface diffusion, 382
323, 324, 325, 326, 329, 355, 359, 373, 374, surface energy, 3
375, 376, 377, 378, 379, 380, 381, 386 surface layer, 380
stability, 18, 19, 23, 44, 49, 51, 140, 175, 381 surface modification, 115
stabilization, 115, 127, 132, 140, 380, 381 surface properties, 2
stabilize, 124 surface region, 344, 347
stages, 104, 132, 139, 221, 366, 367, 368, 376 surface roughness, 11, 218, 222, 225, 240, 241,
standard model, 157, 158 242, 275, 276, 277, 279, 289, 291, 298, 325,
Stark effect, 71, 73, 74, 109 326, 330, 333, 383
statistical processing, 119 surgical, 80
408 Index
survival, 205, 206 threshold, 51, 70, 72, 78, 83, 106, 118, 350
susceptibility, xii, 301, 303, 309, 310, 314 threshold level, 118
suspensions, xii, 302, 303, 304, 305, 307, 316, thrombosis, 54
317 tin, 16, 44, 260
swimmers, 204 tin oxide, 16
switching, 17, 18, 19, 188 titania, 4
Switzerland, 269 titanium, ix, xi, 1, 3, 4, 5, 55, 58, 115, 273, 274,
symbols, 190 275, 276, 277, 278, 279, 280, 281, 282, 283,
symmetry, 80, 157, 158, 163, 175, 188, 190, 294, 284, 298, 320, 323, 324, 329, 330, 355
309 titanium dioxide, 4, 5, 55, 58
synchronization, 83 titanium oxide, 3, 5
synthesis, x, 11, 23, 59, 65, 113, 115, 123, 132, torus, 144, 147
134, 135, 136, 140, 141, 142, 165, 274, 385 total energy, 366
systems, ix, x, xi, 1, 3, 10, 18, 44, 45, 46, 47, 66, total internal reflection, 341
79, 89, 103, 109, 143, 146, 169, 171, 177, 178, toxic, 5, 303, 306, 315
179, 182, 185, 191, 195, 196, 197, 198, 199, toxic effect, 315
201, 202, 203, 204, 205, 209, 210, 211, 212, trade-off, 198, 203
213, 214, 215, 216, 303, 307, 309, 313 trajectory, 374, 385
transfer, 5, 13, 16, 18, 19, 48, 54, 58, 80, 121,
127, 141, 201, 203, 215, 231, 243
T transformation, 163, 254, 293, 297, 298, 368, 385
transformations, 285
targets, 102, 108, 385
transition, 2, 17, 19, 59, 67, 71, 72, 73, 75, 76,
technical assistance, 192
89, 90, 93, 133, 163, 177, 179, 182, 184, 188,
technology, ix, xiii, 1, 4, 16, 56, 142, 143, 218,
191, 199, 209, 242, 254, 255, 256, 307, 379,
242, 274, 302, 304, 342, 344, 365, 375
380, 381, 384, 386
teeth, 196
transition metal, 2, 17, 19, 59
telecommunication, xiii, 341, 346, 349, 350
transition temperature, 177, 179, 184
telecommunications, 342, 347
transitions, 10, 18, 19, 20, 43, 71, 72, 75, 80, 89,
TEM, xii, 30, 32, 34, 273, 280, 281, 289, 290,
163, 170, 211, 260, 383
298, 304, 306, 314, 329, 335, 383
translation, 138
temperature dependence, 79, 137, 181, 184, 186,
transmission, xii, 31, 46, 50, 84, 147, 148, 170,
256, 309
173, 280, 281, 289, 301, 303, 329, 336
temperature gradient, 130
transmission electron microscopy, xii, 31, 280,
temporal, 68, 71, 82, 102, 103, 104, 105, 106,
281, 289, 301, 303, 329, 336
115, 200
transparency, 45, 302
tension, 196, 216
transparent, ix, 1, 3, 16, 17, 21, 44, 48, 49, 53, 61,
test data, 234
81, 381
textile, 303
transplant, 54
thermal energy, 35, 108, 182, 298
transplantation, 54
thermal evaporation, 11, 17, 23, 57, 59, 60
transport, 4, 44, 45, 46, 52, 53, 198, 203, 205,
thermal expansion, 34, 41, 84, 173, 294, 330, 346
262, 324, 326, 375, 376, 385
thermal load, xiv, 373, 374
transportation, 207, 210
thermal resistance, 197
travel time, 207
thermodynamic, 45, 66, 67, 73, 132, 138, 140,
tribological, 23
196, 197, 210, 354
tribology, 241, 242
thermodynamic equilibrium, 45, 66, 67, 73, 132,
triggers, 83, 205
138, 140
trypsin, 303, 316
thermodynamics, 135, 197, 214, 374
tubular, 124, 136
thermonuclear, 65
tumours, 5
thorax, 203
tungsten, 3, 4, 16, 18, 19, 20, 21, 22, 23, 24, 27,
three-dimensional, 15, 59, 242, 243, 291, 292,
29, 30, 34, 35, 36, 37, 38, 39, 40, 43, 58, 59,
334, 361, 383
60, 179
three-dimensional representation, 334, 361
Index 409
tunneling, 2, 60, 346, 347 vibrational modes, 80

turbulence, 127, 131, 214, 230, 234, 242, 243 viscosity, 8, 119, 199, 223, 232, 233
turbulent, xi, 128, 141, 201, 217, 230, 231, 232, visible, ix, x, 6, 10, 11, 13, 14, 37, 45, 47, 56, 63,
233, 234, 235, 236, 240, 241, 242, 243 69, 81, 93, 106, 114, 144, 145, 146, 148, 150,
turbulent flows, xi, 217 151, 153, 157, 160, 161, 162
two-dimensional, 81, 196, 231, 233, 354 vision, 208
VLS, 34
volumetric changes, 190
U vortex, xi, 115, 116, 123, 124, 125, 127, 128,
132, 136, 140, 141, 165, 217, 230, 233, 237,
ultra-fine, 291, 292, 298
239
ultraviolet (UV), ix, x, 7, 8, 13, 35, 63, 69, 86,
vortices, 141, 231, 239, 240
104, 106, 109, 114, 143, 144, 145, 150, 151,
152, 153, 154, 155, 156, 165, 346
uncertainty, 73, 75, 95 W
unification, 215
uniform, 15, 24, 25, 39, 40, 159, 164, 204, 209, walking, 207, 208, 209
267, 275, 278, 280, 287, 289, 298, 332, 344 wall temperature, 144, 147, 154
universality, xi, 195, 211 waste water, 317
universe, 196, 201 wastes, 302
urokinase, 303 wastewater, 303, 316
UV irradiation, 35 wastewaters, 303
UV light, 7, 8 water, xi, xii, xiii, 3, 4, 6, 7, 8, 21, 24, 55, 56, 82,
UV radiation, x, 114, 154 123, 124, 125, 126, 131, 134, 136, 143, 146,
UV spectrum, 152, 153 204, 205, 217, 234, 252, 253, 254, 259, 301,
302, 303, 304, 305, 314, 316, 317, 365, 366,
367, 368
V water vapour, xiii, 365
water-soluble, xii, 302, 314, 317
vacancies, 3, 39, 43, 44, 45, 47, 48, 180, 267
wave number, 11
vacuum, ix, xi, 3, 4, 15, 21, 34, 49, 53, 60, 63, 64,
waveguide, xiii, 341, 342, 343, 344, 345, 346,
82, 86, 91, 94, 96, 109, 146, 148, 175, 245,
347, 348, 349, 350
273, 275, 277, 280, 284, 285, 286, 287, 293,
waveguide technology, 342
294, 298
wavelengths, ix, 10, 63, 69, 95, 109, 349
valence, 5, 12, 18, 19, 21, 38, 44, 45, 46, 48, 57
weakness, 369
validation, 18, 56, 234
weapons, 80
validity, xiii, 67, 242, 319
welding, 80
van der Waals, 2, 74
wells, 171, 346, 347
vanadium, 4
wetting, 55
vapor, xiv, 17, 23, 34, 62, 94, 143, 144, 149, 158,
white blood cells, 7
165, 266, 373, 386
wide band gap, 5, 21, 44, 45, 49
vapor-liquid-solid, 34
wind, 66, 366, 367
variability, 197, 205
windows, ix, 1, 3, 5, 16, 18, 57, 82, 349
variables, xii, 77, 319, 320, 321, 322, 337
wires, 15, 16, 18, 24
variance, 220
wood, 204, 313
variation, 20, 21, 23, 90, 108, 143, 187, 219, 223,
workers, 52
224, 226, 230, 235, 236, 239, 240, 259, 275,
writing, 60, 344, 345, 346, 348, 350
289, 321, 324, 363, 382
vector, 235
velocity, 46, 54, 67, 71, 73, 79, 106, 123, 124, X
198, 203, 220, 223, 225, 228, 232, 233, 234,
237, 241, 250, 251, 255, 256 xenobiotics, xii, 301, 302, 303, 316, 317
vessels, 120 x-ray absorption, 283
vibration, 11, 80, 178, 182, 191 x-ray analysis, 280
410 Index
x-ray diffraction, 50 yield, 5, 10, 44, 114, 115, 141, 150, 151, 152,
X-ray diffraction (XRD), xii, 27, 36, 273, 281, 153, 154, 156, 376, 377, 378, 379, 380, 381,
282, 283, 291, 292, 293, 294, 297, 298, 319, 386
321, 322, 324, 325, 326, 327, 328, 331, 332,
334, 336, 385, 386
X-ray photoelectron spectroscopy (XPS), 21, 38, Z
39, 323, 326
zinc, 1, 3, 4, 44, 48, 50, 51, 61, 62, 81
zinc oxide, 61
Y zirconium, 323, 324
Zn, 47, 48, 50
yeast, xii, 301, 303, 305, 306, 307, 308, 309, 310, ZnO, 44, 45, 48, 49, 50, 51, 53, 60, 61, 62, 247,
311, 312, 313, 314, 315, 316, 317 262, 263, 265, 269

Applied Physics in The 21

Încărcat de

Informații document

Titlu original

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

Applied Physics in The 21

Încărcat de

Drepturi de autor:

Formate disponibile

HORIZONS IN WORLD PHYSICS SERIES

APPLIED PHYSICS IN THE 21ST

APPLIED PHYSICS IN THE 21ST

Nova Science Publishers, Inc.

NOTICE TO THE READER

LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA

Available upon request.

ISBN: 978-1-61728-139-6 (E-Book)

Published by Nova Science Publishers, Inc. 

Chapter 9 Magnetically Modified Biological Materials as Perspective

general mechanism related to an “over-damped” resonant process rather than a simple

process using RF magnetron sputtering were identified as the deposition substrate

EMERGING CONCEPTS AND CHALLENGES IN NANO

A. Subrahmanyam1,2, T. P. J. Ramesh3, A. Karuppasamy4,

SUMMARY OF THE ARTICLE

A. INTRODUCTION TO METAL OXIDES

adsorption or desorption energies—defined as the energy required to remove a metal atom to

1. TITANIUM OXIDE THIN FILM FOR LUNG ASSISTED DEVICES

1.3. Introduction to Photo-Catalysis

Photocatalysis is an established phenomenon shown by several metal oxide thin films:

TiO2 (Eg = 3.2 eV)

Figure 1. Photocatalytic mechanism operative in Titanium oxide

1.4. Basics of Water Splitting by Photo-Catalysis

The concept of water splitting by photocatalysis [1.7] is being adopted in the

TiO 2 + hν → h + + e − (h + = hole vacated by e- )

1.5. Oxygenation of Blood by Photo-Catalysis

[1.34 X Hb gm/dl X SaO2 (expressed as fraction)] + (0.003 X Po2 mmHg) (1)

Figure 2. Basic concept of a photolytic lung assisted device (in vitro)

Figure 3. Photolytic oxygenation of whole human blood

1.6. Challenges in the Process of Oxygenation by Photocatalysis

Figure 4. Optical absorption of photocatalytic oxygenated whole human blood

2. NANO SILVER – SILVER OXIDE THIN FILMS

2.2. Surface Plasmons and Fluorescence in Silver – Silver Oxide Nano

2.3. P- type Conductivity in Mixed Phase Silver Oxide Thin Films

2.4. Fluorescence from Photo-Activated Nano-Silver Oxide

630 640 650 660 670 680

600 620 640 660 680

600 620 640 660 680 700

3. NANO TUNGSTEN OXIDE THIN FILM FOR ELECTRO-CHROMICS

Electrochromism is defined as the ability of a material to change its optical states in

WO3 + xe − + xM + ↔ M xWO3 ( M = Li, Na, H ) 3.1

Figure 3.1 Schematic of an Electro-Chromic Device

Electrochromic device consists, in general, of two (transparent) electrodes, an electrolyte,

3.3. The Optical Absorption Process in Tungsten Oxide Thin Films

The optical absorption process, in electro-chromic materials in general, and tungsten

3.3. Kelvin Probe Studies

qVCPD= ΦR– ΦF 3.6

500 colored WO3 -2

3.4. A Brief Outline on Nano-Electro-Chromics Materials and Processes

3.5. Structural Properties of WO3 Nanowires

3.6. Morphology of WO3 Nanowire Thin Films

3.7. Growth of Layered Tungsten Oxide Thin Films

3.8.1. Structural properties of electron bombarded WO3 films

3.8.2. Optical properties of electron bombarded WO3 films

3.8.3. Compositional studies of electron bombarded WO3 films

These changes in the spectra are interpreted in earlier reports as a consequence of

3.8.4. Surface morphology of electron bombarded WO3 films

Figure 3.28. SEM image of the electron bombarded sample S3

The surface morphology of electron bombarded samples S2 and S3 is alike and so a

3.8.5. Luminescence in electron bombarded samples

With the optimization of nanostructured electron emitters (nanorods, nanowires and

4. ZINC OXIDE FOR TRANSPARENT ELECTRONICS

4.3. Measurements on p- type TCOs

Published by Nova Science Publishers, Inc.

e[ε ≥IP(A)] + A → A+ + 2e; e[ε ≥IP(A+)] + A+ → A2+ + 2e. (2.11)