Advances in Inorganic Chemistry and Radiochemistry 1 (1959)

Advances in
I NORGANIC CHEMISTRY
AND
RADIOCHEMISTRY
Volume 7
This Page Intentionally Left Blank
Advunces in
INORGANIC CHEMISTRY
AND
RADIOCHEMISTRY
EDITORS
H. J. EMELEUS
A. G. SHARPE
University Chemical laborafory
Cambridge, England
VOLUME 7
7959
ACADEMIC PRESS INC.
NEW YORK
COPYRIGHT(O
1959
BY
ACADEMIC
PRESS
INC.
ALL RIGHTS RESERVED

NO PART OF THIS BOOK MAY BE REPRODUCED IN ANY FORM
BY PHOTOSTAT, MICROFILM, OR ANY OTHER MEANS,
WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS
ACADBMIC PRESS INC.

111 FIFTHAVENUE
NEW YORK,NEW YORK10003
United Kingdom Edition
Published by
ACADEMIC PRESS INC. (LONDON) LTD.
BERKELEY
SQUARE
HOUSE,LONDONW. 1
Library of Congress Catalog Card Number: 59-7692
First Printing, 1959

Second Printing, 1961
Third Printing, 1965
PRINTED IN THE UNITED STATES OF AMERICA
LIST OF CONTRIBUTORS
D. H. F. ATKINS, United K i n g d m Atomic Energy Research Establishment, Harwell, Didcot, England
E. 0. FISCHER,
Institut fur anorganische Chemie, Universitat Munchen,
Munich, Germany
H. P. FRITZ,
Anorganisch-chemisches Laboratorium, Technische Hochschule, Munich, Germany
R. J. GILLWPIE,University College, London, England *
GARMANHARBOTTLE,
Department of Chemistry, Brookhaven National
Laboratory, Upton, New Ymk
WILLIAMN. LIPSCOMB,
School of Chemistry, University of Minnesota,
Minneapolis, Minnesota
N. L. PADDOCK,
Albright and Wilson (Mfg.) Ltd., Oldbury, Birmingham,
England
E. A. ROBINSON,
University Callege, London, England *
W. RUDORFF,
University of Tubingen, Tuebingen, Gemnany
H. T. SEARLE,
Albright and Wilson (Mfg.) Ltd., Oldbury, Birmingham,
England
A. A. SMALES,
United Kingdom Atomic Energy Research Establishment,
Harwell, Didcot, England
NORMAN
SUTIN,Department of Chemistry, Brookhaven National Laboratory, Upton, New York
H. TAUBE,
Department of Chemistry, University of Chicago, Chicago,
Illinois
T. C. WADDINGTON,
The University Chemical Laboratories, Cambridge,
England
* Present address : Hamilton College, McMaster University, Hamilton, Ontario,

Canada.
PREFACE
The world-wide interest in inorganic chemistry and radiochemistry
has created a keen demand for articles by experts which review the fields
of current interest to research workers and which serve, also, to inform
the nonspecialist of recent progress. It is with this in mind that the
editors venture to commend this series of volumes to the reader. It is
hoped that, in the succeeding volumes, the topics presented will, in a
relatively short period of time, cover many of the major fields. Thereafter, there will, of necessity, be other articles which bring these topics
up to date. The editors are most grateful to those who have contributed
to this volume and to others who have promised contributions for later
voIumes. They are particularly glad that the response has been international, since they feel that the subject has an unusually wide appeal and
each country has much to learn from the approach of workers in others.
Perusal of the material presented in this volume will show that the
definition of inorganic chemistry adopted here is a broad one. Modern
inorganic chemistry depends for its very existence on the application of
physical and physicochemical principles to chemical phenomena, and it
is intended that the integration of structural, kinetic, and thermodynamic
data with descriptive chemistry shall be one of the main features of this
series.
H. J. EMELI~US
A. G. SHARPE
January, 1959
Vii
CONTENTS
LIST OF CONTRIBUTORS
. . . . . . . . . . . . . . . .
FOREWORD
. . . . . . . . . . . . . . . . . . .
vii
Mechanisms of Redox Reactions of Simple Chemistry
H TAUBE
I. Introduction . . . . . . . . . . . . . . . . .
I1. Electron Transfer in the Gas Phase . . . . . . . . . .
I11. Interaction of Reactant Ions with Environment . . . . . .
IV. General Consideration of the Electron Transfer Process in Solution
V. The Outer-Sphere Activated Complex . . . . . . . . .
VI . The Bridged Activated Complex . . . . . . . . . . .
VII . Reactions Proceeding by Mechanisms of Uncertain Classification * .
VIII. Systems in Which a Net Two-Electron Change Is Involved . . .
IX . Reactions in Nonaqueous Solvents . . . . . . . . . .
X . Conclusion
. . . . . . . . . . . . . . . . .
References
. . . . . . . . . . . . . . . . .
1
2
4
8
12
19
32
41
44
46
50
Compounds of Aromatic Ring Systems and Metals
E . 0. F~SCHER
and H . P . FRITZ
I . Introduction . . . . . . . . . . . . . .
I1. Complex Formation by Five-Membered Ring Systems .
I11. Complex Formation by Six-Membered Ring Systems . .
IV. The Existence of Complexes of Seven-Membered Aromatic
. . . . . . . . . . . . . .
References
. 56
57
. . . 97
Systems 107
. . . 109
Recent Studies of the Boron Hydrides
WILLIAM
N . LIPSCOMB
I. Introduction . . . . . . . . . . . . . . . . .
I1. The Boron Hydride Structures . . . . . . . . . . .
I11. The Valence Theory . . . . . . . . . . . . . .
IV . Nuclear Resonance (NMR) Studies of the Boranes. Substituted Boranes. and Related Compounds . . . . . . . . . . .
V. The Equations of Balance . . . . . . . . . . . . .
VI . A Topological Theory of Boron Hydrides . . . . . . . .
VII . Possible Boron Hydride Ions . . . . . . . . . . . .
VIII . The Boron Halides . . . . . . . . . . .
. . .
IX . Reactions among Boron Hydrides . . . . . . . . . . .
Notes Added in Proof . . . . . . . . . . . . . .
References
. . . . . . . . . . . . . . . . .
ix
118
118
121
131
138
138
144
146
145
151
154
CONTENTS
Lattice Energies and Their Significance in Inorganic Chemistry
T. C . WADDINGTON
I . General Introduction and Account of the Born-Haber Cycle . .
I1. The Calculation of Lattice Energies . . . . . . . . . .
I11. Individual Values of the Lattice Energies of Alkali Metal and Alkaline Earth Salts . . . . . . . . . . . . . . .
IY. Uses of Calculations of Lattice Energies . . . . . . . . .
References
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
158
162
188
203
218
Graphite Intercalation Compounds
W . RUDORFF
I . Crystal Structure and Mode of Reaction of Graphite .
I1. Graphite Compounds with Homopolar Bonding . .
I11. Graphite Compounds with Polar Bonding . . . .
IV . Residue Compounds . . . . . . . . . . .
V . Comparative Survey
.
.
.
.
.
.
.
.
.
.
References
.
.
.
.
. . . . . . . .
.
.
.
.
.
.
224
226
. 236
. 262
. 263
. 264
The Srilard-Chalmers Reaction in Solids

GARMAN
HARBOTTLE
and NORMAN
SUTIN
I . Introduction . . . . . . . .
11. Chemical Effects of Nuclear Recoil .
I11. Post-Recoil Annealing Effects . . .
IV . Conclusion
. . . . . . . .
V.Appendix . . . . . . . . .
List of Symbols . . . . .
.
References
. . . . . . . .
. 271
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
268
281
309
.
. 309
. . 311
. . 312
.
.
Activation Analysis
D . H . F. ATKINSand A . A . SMALES
.
. . . . . . . . .
I . Introduction
.
.
.
.
I1. Irradiation
. . . . .
I11. Radiation Sources . . . . . . . .
1V. Experimental . . . . . . . . . .
V . Conflicting Nuclear Processes . . . . .
VI . Applications of Neutron Activation Analysis
VII . Activation Analysis with Charged Particles .
VIII . Conclusion . . . . . . . . . .
. . . . . . . . . .
References
315
316
.
.
.
.
.
.
.
319
.
.
.
.
.
.
.
321
.
.
.
.
.
.
. 331
. . . . . . . 334
. . . . . . . 339
. . . . . . . 341
. . . . . . . 342
xi
CONTENTS
The Phosphonitrilic Halides and Their Derivatives
N . L . PADDOCK
and H. T. SEARLE
I. Introduction . . . . . . . . . . . . .
I1. The Preparation of the Phosphonitrilic Halides . . .
I11. Properties . . . . . . . . . . . . . .
I V. Structure . . . . . . . . . . . . . .
V . Discussion
. . . . . . . . . . . . .
VI . Conclusion . . . . . . . . . . . . .
List of Symbols . . . . . . . . . . . .
References
. . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
348
348
353
365
375
379
380
381
The Sulfuric Acid Solvent System
R . J . GILLESPIE
and E. A . ROBINSON
I. Introduction . . . . . . . . . . . . . . . . . 386
I1. Solutions in Sulfuric Acid . . . . . . . . . . . . . 394
I11. The Behavior of Some Inorganic Compounds in Sulfuric Acid . . 405
References
. . . . . . . . . . . . . . . . . 421
AUTHOR INDEX
. .
. 425
SUBJECTINDEX
. . . . . . . . . . . . . . . . 442
MECHANISMS OF REDOX REACTIONS OF SIMPLE CHEMISTRY

H. Taube
Department of Chemistry, University of Chicago, Chicago, Illinois
I. Introduction
. . . . . . . . . . . . . . . . .
11. Electron Transfer in the Gae Phase . . . . . . . . . .
111. Interaction of Reactant Ions with Environment . . . . . .
V. The Outer-Sphere Activated Complex . . . . . . . . .
VI. The Bridged Activated Complex . . . . . . . . . . .
VII. Reactions Proceeding by Mechanisms of Uncertain Classification .
VIII. System in Which a Net Two-Electron Change Is Involved . .
IX. Reactions in Nonaqueous Solvents . . . . . . . . . .
X. Conclusion . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . .
. .
.
. .
.
.
.
.
. .
. .
. .
.
.
1
2
4
8
12
19
32
41
44
46
50
1. Introduction
The principal concern in this chapter will be the mechanism of reaction

of metal ions in which a change of oxidation state takes place on reaction.
These reactions, of which
Fe++
+ Ce(1V)
Fe+++ Ce+++
(1)
may be taken as typical, are part of the material of instruction in general

and analytical chemistry and as such are familiar to anyone with training
in chemistry. Nevertheless, fundamental questions as to the means by
which the change in oxidation state takes place were until recently not
formulated, let alone answered.
The general subject includes as a proper part problems of specific chemistry such as are encounted in the reaction, for example, of Fe++ with
Mn04-: What are the steps by which the system proceeds to the final
products, and what are the properties of the intermediate oxidation states
of Mn (or of Fe) which must be involved for such a complex over-all reaction? Important for inorganic chemistry as such questions are, for the
most part they have been set aside, and attention is directed rather to the
description of the individual steps. Given a process of simple order, we
shall consider questions such as these: What is the closest distance of approach of oxidant and reductant? What is the arrangement of other groups
besides the reactant metal ions in the activated complex? What motions
of these groups are necessary to consummate the reaction? How are the
1
H. TAUBE
rates dependent on the nature of the metal ions and on the other groups
which may be present in the activated complex? These questions will be
given a more specific formulation when we consider the various proposals
which have been made describing the electron transfer process.
11. Electron Transfer in the Gas Phase
It is instructive to consider some features of the data on charge-transfer processes in the gas phase between simple molecules before considering
the systems as they are complicated by the interactions of the central
atoms with molecules of the solvent or other groups. The earliest experimental evidence for the occurrence of electron transfer between species in
the gas phase was obtained from mass spectroscopic measurements (120).
Data on the cross sections for such reactions are derived principally from
two sources: measurement of the drift velocity of ions in various gases and
measurements of the attenuation of ion beams brought about by inelastic
collisions with molecules. The cross section for charge transfer is dependent
on the kinetic energy of ions, so that the data on the drift velocity, involving speeds more nearly those of thermal energies, will be considered first.
The subject of drift velocities, particularly as it pertains to inert gas
systems, was discussed by Hornbeck (58) (experimental results) and
Holstein ( 5 7 ) (theoretical) as part of the program for a symposium on
Electron Transfer Processes in general, held a t Notre Dame in 1952. The
significant observation is that the drift velocity of an ion such as He+.is
much less than is expected if account is taken only of the usual processes
for energy transfer, including polarization of He by the positive ion. Similar effects are noted for the other inert gas ions and have been recorded also
for Nz+ in N2 (90). The effective collision cross section is increased by
symmetry effects which include electron transfer as a component. Table I
TABLE I
COMPARISON
OF CROSSSECTION
FOR ELECTRON
TRANSFER
Ti
WITH GASKINETIC
CROSSSECTION
2.
T,
He
Ke
A
(cm*)
54
65
134
T.
X 1016 (cmz)
15
21
42
presents the data taken from Hornbecks paper comparing the cross section for charge transfer with the normal cross section for ion-molecule
interaction including atom polarization. Ticonsiderably exceeds T,, and
MECHANISMS OF REDOX REACTIONS
is large enough so that electron transfer over nuclear separations as large

as 5-8 Bohr radii must be invoked. For N2+ in N2 the effect on drift velocity is smaller than is observed for the inert gases, and the mobility is only
7.5% less than the normal gas kinetic mobility. The difference between an
inert gas on the one hand and a more complex molecule on the other can
be attributed to the circumstance that the equilibrium internuclear separations for N2+ and N2 are different, as are the separations of the vibrational levels. Since the electron transfer process is adiabatic, the energies
at the two sites must match closely to make transfer possible.
During the Notre Dame Symposium, Muschlitz and Simons (93) summarized work on the cross sections for inelastic scattering of protons by
various gases. The cross sections for 100-volt protons on He, Ne, Kr, and
A are in the ratios 0, 1.5, 48, and 28 (52); for a complex molecule such as
C2Ha,a value of 112 is observed; there is a decrease in the cross section
for inelastic scattering as the kinetic energies of the ions increase (118).
The interesting qualitative features are: the maximum in the cross section
for the inert gas as atomic number increases and the large cross section
for C2Hs (and other complex molecules). Presumably, the large cross section for complex molecules undergoing electron transfer to H+ comes
about because, by using vibrational states, these molecules can provide
many ways of bringing about the energy matching.
Since shielding by other electrons at large distances for a test electron
is complete, and since the cross sections for electron transfer when simple
atoms are involved are large compared to atomic dimensions, hydrogenlike wave functions can be used to describe the electron distribution a t
the large radii in question. Thus calculations of the probability of electron
transfer, at least in simple cases, meet with a fair degree of success. Such
calculations have been made by Holstein (57) for the symmetrical inert
gas systems, using Hartree-Fock wave functions for the outermost shells.
Gurnee and Magee (48), dealing with the same systems, have used the
Slater wave functions (119) and have chosen in each case a value for the
parameter Q (which in the hydrogen-like case would be (21) where 1is
the ionizing potential) such that the one-electron wave function is approximately satisfied for the range of distances in question. Satisfactory agreement of calculated and experimental cross sections for Ne Ne+ and
He 3- He+ is obtained. The treatment of Gurnee and Magee also includes
double charge transfer reactions; the theory and observations made for
such reactions (141) are germane to the issue of 1 versus 2e changes in
redox reactions. Experimentally, the cross section for Ne++ Ne =.Ne
Ne++ is about 1/4 of that for the l e process in the same system, while for
A, the ratio is about 1/2. Theory (48) suggests a ratio of about 4/2 for both
sets of reactants. For the discussion of charge transfer in diatomic mole-
H. TAUBE
cules, Gurnee and Magee used Heitler-London functions for the electronic
part of the wave functions and harmonic oscillator and rigid rotator functions for the internal motions. Of the internal motions, only the vibrational
part affects the probability of electron transfer. Gurnee and Magee have
tabulated values of the vibrational overlap integrals for H2-H2+, HDHD+, D2-D2+, and Nz-Nz+-these range from 0.195 for D2-D2+ to 0.940
for N2-Nz+.The case of widest application is that for which the energies
of the initial and final states differ. At small relative velocities, the probability of transfer decreases sharply as the energy difference increases; for
an energy difference as small as 0.05 ev, electron transfer is possible only
when relative velocities are of the order of loe cm sec-l or larger (48).
111. Interaction of Reactant ions with Environment
For the systems with which we are concerned, other molecules are always in close proximity to the species undergoing charge transfer. Since
even in a solvent of low dielectric constant the energy of interaction with
the medium of a charge residing on a sphere of atomic dimensions amounts
to several tens of kilocalories per mole, and since a redistribution of charge
always occurs in the reactions of present interest, due consideration must
be given to these interactions. These interactions are often discussed in a
way that glosses over structural features, as is done for example in the
application of the Born equation. But from a chemists point of view it is
necessary to inquire into the structure of the solvent surrounding the ions,
especially in a solvent containing polar molecules, and in particular to
distinguish groups in the first coordination sphere from those further out
which may also be influenced by the electric field of the central ions. This
distinction is naturaI, not only because the forces binding the first layer
are greater than for those further out, but also because the residence time
for a solvent molecule adjacent to a cation of high charge may be considerably greater than for solvent having only solvent as neighbor. The forces
that give individuality to the different ions, whether we consider the influence of the ligands on the central ion or vice versa, are largely expended
in the first coordination sphere, so that differences in the interaction which
two ions such as A1 (OH2)a+ + + and Fe (OH216+ + + ,having the first coordination spheres completed, have with the surrounding medium can be understood largely as differences that arise from the slightly different radii
of the two central ions. The distinction between groups in the first coordination sphere and other groups may become unsharp in certain cases [of
which Cr(HzO)e++ may be an example] in which not all the groups in the
first coordination sphere are equivalent. However, even in most such cases,
the less firmly bound solvent molecules experience a much greater electric
field than do those in the second sphere of coordination, so that the distinction can still be maintained.
Several aspects of solvation phenomena will be considered : solvation
of cations, interaction of cations with other groups, and phenomena of
electrolytic dissociation. The essential general features will be covered if
we consider on the one hand a solvent of high dielectric constant, such as
water, and on the other, remark on the differences in the state of an electrolyte .produced by dissolving it in solvents of low dielectric constant.
Special emphasis will be given to the subject of hydration of ions, because
most of the work on redox reactions has been done with water as solvent.
Over the last 15 years great progress has been made in understanding
hydration of cations. The existence of well-defined hydrates of cations in
crystals leads naturally t o the supposition that cations in solutions will
also be hydrated. Certainly the energies of hydration of cations with water
are great enough (ranging from approximately 60 kcal mole-' (73)
for Cs+ to > lo3 kcal for a tripositive cation such as Al+++) so that it is
not unreasonable to expect cations and water in the first sphere of coordination to constitute definite chemical species. I n spite of the force which
such general considerations may have, hydrated cations have not been
part of the careful chemist's vocabulary of molecules, and for good reasons. With only evidence of the kind cited for the existence of hydrated
ions in solution, questions can be raised as to their existence as definite
species. In the solid, species of particular formula may be stabilized by
the forces producing the ordered arrangement. In the liquid, several different configurations may have approximately the same energy; even when
the energy differences for different configurations are appreciable, the
residence time of a particular set of water molecules in the first coordination sphere may be so short that the concept of the hydrated cation as a
molecule loses significance. In a program for the study of hydration of
ions, the experimenter seeks to establish the formulae of the hydrated ions,
their labilities and the energy differences between states of different coordination numbers.
Most of the classical physicochemical methods for the study of hydration of ions fail to distinguish between water in the first coordination
sphere and water more remote from the central ion which aIso comes under
the influence of its charge. Some of the methods more recently applied
have provided a clearer picture. The oxygen isotope exchange method,
where applicable, not only can define the composition of the first coordination sphere but also can be used to measure the lability of the aquo
ion. Thus it has served to establish Cr(OH2)s+++ (64); (NHs)&o
(OHa) + + + (114) ; and (NH3)&o (OH,) 2++ + (106) as webdefined
species in solution and also to fix the half-time for exchange of
H. TAUBE
these species (tn of the order of 20 to 40 hr at 25'). I n a further development (10) of this method, which makes accessible shorter sampling
times, it has been shown that Al+++aq. can indeed be represented as
Al(OH2)6+++,and that the half-time for the exchange of bound water
exceeds 0.005 sec. Experiments on the effect which cations exert on the
relative fugacities of H2016and H2018 have served to indicate that even
cations which form labile hydrates (for example, H+aq.) nevertheless
form definite hydrates (40).
Perhaps the greatest progress has been made in the subject of hydration of cations by the studies of the electronic spectra of transition metal
ions. The theoretical developments on the influence of ligand fields on the
energy levels of ions, which culminated in the work of Penney and Schlapp
(102)and of Van Vleck (133) on magnetic properties of ions, have been
applied by Hartmann and co-workers (51,65), as well as others (91),to
the problem a t hand. The theoretical ideas can be applied in two ways. In
one application the spectrum of an ion in solution is compared with that
in a solid in which the composition and constitution of the hydrated ion
are known. However unsatisfactory the simple application of crystal field
theory may be in predicting quantitatively the energy levels of the ion in
a hydrate, it can in many cases nevertheless be trusted to indicate the magnitude of the changes in spectrum which can be expected t o accompany a
change in coordination number. I n another application, less empirical but
also less dependable, an analysis of the spectrum of the ion, applying the
principles of crystal field theory, can be used to determine the symmetry
of the ligand field. Difficulties can be encountered in the second application arising from the incomplete resolution of the electronic bands and
from unresolved questions of intensities. The work on the electronic spectra has, hovever, served to establish the formulae of the hydrates of most
colored transition metal ions in water.
A relatively new technique (139) for measuring the substitution labilities of hydrated paramagnetic cations is observation of the line broadening of the nuclear magnetic resonance absorption of ligand nuclei caused
by the central ion. The effect of the line broadening is particularly marked
for ions which permit rapid exchange of water in the first coordination
sphere and is much weaker for an ion such as Cr(H20),+++. This technique has been applied by Werte (139) to C1- substitution and by Connick (30) to H 2 0 substitution in hydrated cations. There has been a revival of interest (22) in the application of X-ray diffraction to a study of
aqueous solutions, and further significant progress can be expected in the
near future using this technique.
Specific hydration of anions is not dealt with here, not because the
energy of hydration is not large but because there is greater question of
the existence of definite species in the sense implied in the discussion of the
cations. The hydration of anions is different in this sense, that simple ion
dipole (hydrogen bond) interactions come into question, and it is likely
that the residence time of a particular water molecule adjacent to an
anion will not be much longer than for water adjacent to water.
Conventional physicochemical methods, the most powerful of which
depend on the application of the mass law, can be used to establish the
composition of a complex ion with respect to a nonsolvent ligand. Even
when the composition with respect to nonsolvent ligands is established,
many questions remain which are particularly hard to answer for substitution labile systems. Thus when an ion of formula FeC14- is proposed,
one can ask, how many molecules of water in addition to the C1- comprise
the first coordination sphere? If the formula is FeC14(H2O)2-, what is the
ratio of the cis to trans form a t equilibrium? Furthermore, for oppositely
charged ions of high charge, the distribution between two forms must be
considered, in only one of which is there a direct bond of the ligand to
central ions. Thus the studies on hydration of cations need to be extended
also to complex ions which are coordinatively unsaturated with respect
to nonsolvent ligands. Studies with complex ions which are not labile to
substitution help greatly in assessing the importance of outer-sphere1
forms. For the system (NH3) &oOH2+ + + SO4=, inner-sphere and
outer-sphere forms are of about equal stability (131). The outer-sphere
affinity of Cr(H,O)B+++ for C1- is very small. This latter conclusion,
based on the careful work of Gates and King (@), casts doubt on the
strong outer-sphere association of halides and tripositive ions indicated
by the work of Linhard (78) and of Evans and Nancollas (39).
The incomplete dissociation of ion pairs which are coordinatively independent, observed for ions of opposite and high charge, even in water,
is a common phenomenon in solvents (71)of low dielectric constant (D) .
I n liquid NH3 with D = 27 at -60", strong electrolytes such as NaN03
or NH4Cl have dissociation constants of the order of
When dioxane
is the solvent (D = 2.18), a salt such as Bu4NC104,though comprised of
large ions of low charge, has a dissociation constant of the order of 10-l8.
For such solutions the model of an ion atmosphere surrounding each ion
collapses to one in which the electrolyte exists as ion pairs, but with the
further complication that a t reasonable concentrations of electrolyte association of the ion pairs takes place (41).
'The term requires definition. It refers to regions beyond the first coordination
sphere of B cation considered aa center, but not to anions considered as centers.
Thus, for the system Al+++aq. C1- aq., the distinction is between AlCl++ aq. and
Al(OH&+++ C1-, and the question of the hydration of C1- is not raised (for example, no distinction is made between structures such as AI+++(OH& C1- and
AI+++(OHB)~
(OH,), C1-1.
H. TAUBE
Let us consider first eome general features of electron transfer between

ions in solution without specific, detailed discussions of individual systems.
The conditions which must be met to make electron transfer between two
ions possible are matching the energy of the electron at the two sites and
sufficient orbital overlap between the two sites to provide for a reasonable
probability of transfer (76, 104). Various factors are irivolved in meeting
t8heseconditions, and these are considered qualitatively in the present section discussing mainly reactions in media of high dielectric constant but
also referring briefly to media of low dielectric constant.
The mechanism of the electron transfer itself has been given considerable discussion, and attention has been directed to it as a barrier penetration phenomenon (84,104,137). This emphasis is justified only to the
extent that other reasonable possibilities exist. Let us examine at the outset one such alternative process which has also come up for consideration,
namely one in which the electron spends rmfficient time in the solvent to
be solvated. For a reaction with Fe++ as reducing agent, the mechanism
corresponding to this process would be formulated as follows:
1 1
Fe++ aq. 2 Fc+++aq.

I;-,
+ e- aq.
+ e- aq. - products.
/,
Oxidizing agent
--+
(3)
A mechanism of this kind, consisting of discrete steps, can be rejected on

the basis of kinetic evidence for all systems which have thus far been
studied in detail (although it would be dangerous to reject it as a possibility for very powerful reducing agents in water or liquid ammonia). For if
reaction (2) forward is slow compared to (3), the rate of reaction will be
independent of the concentration of oxidizing agent, while if (2) is maintained as an equilibrium with (3) rate determining, the rate of reaction
will be inverse in the concentration of Fe+++. Such rate laws have not
heen observed for the systems of simple chemistry which we are discussing.
IIowxer: the feature that the rates can be inverse in the concentration of
the product formed from the reducing agent has frequently been observed
for systems of complex chemistry. In every such case, this kinetic feature
can reasonably he attributed to the formation of an intermediate oxidation
state.
If the model is modified by postulating that the electron is strongly
affected by the presence of the oxidizing agent, the conditions for barrier
penetration are also met, and it is difficult to see what experimental criteria would distinguish the two cases, whether the electron surmounts a
barrier the shape and magnitude of which is affected by the reactant ions,
or whether it penetrates such a barrier. An answer would be given if the
energy levels of the electron in the activated complex were known and if
the shape and height of the barrier could be calculated. Knowledge of the
distribution of electrons in the complex molecules in question is not sufficiently quantitative to settle the issue at present.
The suggestion (34) analogous to that contained in Eqs. (2) and (3),
but having the oxidizing agent remove electrons from the solvent, can be
rejected in many systems for reasons similar to those advanced for
the electron in the solvent. In any case, such processes would seem to
be reasonable possibilities only for the strongest oxidizing agentsfor HzO = HO H+ e- lies at approximately -2.8 volts (72)-and
there is no evidence that an oxidizing agent such as Co+ + +aq. with Eo
at -1.8 volts makes use of a solvent-hole mechanism.
The limited conclusions reached in the preceding paragraph should
not be taken to imply that the description of these processes as barrier
penetration phenomena is always apt. It is useful to consider the extreme
mechanisms: in one the geometries of the oxidized and reduced forms, at
least as regards the first sphere of coordination, are the same, and substitutions in the first coordination sphere are not readily accepted; in the
other there are important changes in the first coordination sphere accompanying the electron transfer, and further, at least one of the species readily accepts substitution, and the coordination spheres interpenetrate in
the activated complex. It will be noted that more than one condition has
been specified for each case. The conditions are not necessarily combined
as they appear here, and other cases can arise. The particular combinations have been chosen to provide the greatest contrast in behavior which
can be expected.
The systems of the first class afford the closest approach to a simple
barrier penetration process, and perhaps they more readily respond to a
theoretical analysis. It can reasonably be supposed that for these systems
orbital overlap for the two ions is small, so that the frequency of the electronic transition is small, and there is no substantial binding between the
two exchanging centers. A model of this kind presumably corresponds to the
weak overlap cases as defined and discussed by Marcus (82).In attempting to calculate the rates of these reactions, besides the problem of
the shape and height of the barrier for the electron transfer, electrostatic
interaction of the reactants must be dealt with and the energy necessary
to distort the solvent and ionic atmosphere about each ion to make the
energy of the electron equal at the two sites. Different workers have emphasized different ones of these factors, and serious differences of opinion
are recorded.
30
H. TAWE
The electrostatic interactions are allowed for in the various treatments

of the problem (82?,84,137),but the quantitative validity of the treatments
covering close distances of approach has been correctly questioned (104).
In discussing the frequency of electronic transition, Libby (7'6) uses the
radial dependence of hydrogen-like wave functions, with due regard for
the charges of the ions, and discusses electron exchange between transition
metal ions over distances of several Angstrom units; Zener (142), in discussing the overlap of d electrons of nearest neighbor transition metal
cations (nearest neighbor Mn cations are separated by O=) in a solid
such as LaMn03, states, ". . . Mn+3 ions are sufficiently far apart as to
have no appreciable overlapping." Marcus (82) considers the ions, including the first coordination shell, as conducting spheres, takes the distance
of approach for electron transfer to be the sum of the radii of the spheres
and concludes that factors other than the frequency of transition of the
electron determine the rate of Teaction. Marcus et al. (84) leave the probability of the electron transfer, depending as it does on distance, as a running parameter, balancing the increasing electrostatic repulsion on close
approach against the greater probability for barrier penetration under
these circumstances. The assumption of the complex ion as a conducting
sphere, necessary because knowledge of the electron distribution in such
ions is so little understood, does, however, gloss over the extraordinary
differences in rate observed for different ligands, and no theory can be
considered satisfactory that does not make allowance for the chemical
individualities of the systems.
For reactions in which there are no changes in tlie first coordination
sphere, the effects arising from the requirement for nonequilibrium polarization of the surroundings are relegated to regions beyond the first sphere
of coordination, but they are nevertheless important. Marcus (82) has
given a mature discussion of the contribution made to the free energy of
activation of these processes by the requirement for solvent reorganization. Electrolyte effects also exist, for the transfer of an electron from one
site to another requires the eventual redistribution of ions; as in the case
of dipole interactions, the changes must occur prior to the electron transfer
as part of the requirement that no energy change accompany electron
transfer. There may in fact be an interplay between the solvent dipole and
ion atmosphere contributions, and it is conceivable that the transfer of a
single ion from one site to another can make up a large part of the total
energy requirement of the activation of the environment. A specific mechanism (53, 112) of this type which has been suggested is a proton shift
from the coordination shell of the reducing agent t o that of the oxidizing
agent on electron transfer (or described as H-atom transfer).
To make the discussion concrete, let us take as the activated complex
11
for the reactions of the second type a species of geometry: LsM+"+lX

M+nW5,where L , X , and W are ligand groups which may all be alike,
and the M's are metal ions. In arriving a t a configuration having X as a
bridging group, either M+"+l or M+" (or both) has undergone substitution in the first coordination sphere. We will also assume that the electronic
transfer or redistribution of the electron itself is extremely rapid. This is
an independent assumption which does not follow from the assumed geometry, but it seems to be required to explain why certain systems adopt this
kind of activated complex in undergoing electron transfer reactions. Important differences as compared to the previous class are realized for these
systems. The electron is effectively distributed over the two sites, and the
binding resulting therefrom lowers the energy needed for the formation
of the activated complex. Whereas in the previous case the electron waits
for an appropriate fluctuation without being able to influence the changes
required in the site to which it will transfer, in the case under present discussion it affects the energy required to produce the fluctuation there. Presumably the reaction is consummated by some fluctuation which causes
the separation of Mn+l and Mn, and may require other changes in the GOordination spheres as well. If, for example, the fluctuation in question is
the transfer of X from an equilibrium position close to the oxidizing cation
to one close to the reducing cation, the redistribution of the electron affects
the energy that is necessary to stretch the Mn+l-X bond and to compress
the Mn-X bond.
A feature of the bridged activated complex is that the bridging group
may move from the oxidizing agent to the reducing agent, that is, in the
direction opposite to that of the electron. When the bridging group is negatively charged, the process is electrically equivalent to the transfer of an
atom from the oxidizing agent to the reducing agent, and thus the demand
for reorientation of solvent is less than in the previous mechanism. This
feature also operates to encourage traffic by a path which in certain other
respects appears to be highly unfavorable to reaction.
For reactions of this class, even less than for those discussed earlier,
is it possible to account quantitatively for the rates. In fact, i t can be
considered an ambitious goal even to predict whether in a particular instance substitution to form the binuclear complex is rate determining or
whether, as was assumed in the foregoing discussion, equilibrium to form
the binuclear complex is established rapidly, and the slow step is disruption of the binuclear complex. For a quantitative discussion, the energy
of interadion in such a binuclear complex must be assessed. This energy
will depend greatly on the electronic structures of the metal ions, on the
properties of the bridging group, and on the properties of the other ligands.
Similarly, the energies required to produce the changes in coordination
12
H. TAUBE
sphere necessary to complete the reaction will be dependent on the same

facturs. At the present time, far from being able to treat these factors
quantitatively, we still lack a full description of the activated complexes for systems of the widely different electronic structures which must
be considered.
In solvents of low dielectric constant the differences between the two
types of activated complex become. less pronounced. Thus, considering for
example the system involving electron exchange between CIOz and C102-,
when the solvent is water the mechanism would fairly clearly be of the
first type; however, in a solvent of low dielectric constant, free C102does not exist, but it will always be in association with the counter cation.
When the dipole moment of the solvent is low, the only mechanism for
bringing about the matching of energy of the electron at two sites is the
motion of the cation, and we thus have the parsimony of atomic readjustments characteristic of the second class. The electron will follow the motion of the cation from C102- to C102, and the activated complex can be
considered to be of the bridged type [C102NaC102] in the case under present discussion.
The two types of activated complex that have been discussed present
t,he gross geometrical features of two types which have been experimentally delineated; however, the types that are observed do not necessarily
present the other features which were referred to in the cases chosen for
the qualitative theoretical discussion. In the next two sections the experimental evidence on the nature of outer sphere and bridged activated
complexes is discussed (the terms have no connotations other than the
geometry of the activated complexes implied by them), and the general
observations relating to the kinetic behavior of the corresponding systems
will be outlined.
V. The Outer-Sphere Activated Complex
The discussion of experimental methods which opens this section applies equally well to the next two sections. The development of new methods of measurements of rates has been an important part of the progress
in this field. The majority of the kinetic work done has been for systems
in which there is no net chemical change. The reasons for this choice of
problem are understandable; they arise in part from the desire to have
the chemistry simple and partly from the novelty and excitement of using
isotopic tracers to measure rates of reactions, which until recently could
not be measured by any other means. In fact, when the significance of
isotopic methods in this field is appraised, it becomes obvious that of principal importance is the fact that the new methods attracted many capablc
13
workers who otherwise might not have developed an interest in this area
of chemistry.
The principal method for measuring the rate of exchange in a system
in which there is no net chemical exchange is to observe the redistributiw
of nuclei between the two forms. In such applications separations of the
oxidation states after various intervals of time are made. These can involve, singly or in combination, precipitation, complex ion formation,
solvent extraction, ion exchange, and separation by diffusion. The method
of quenching and making separations by precipitation, in combination
with a flow technique for mixing, has been applied particularly successfully by Wahl and his students. An indirect method which does not require
separation of the species undergoing electron exchange has been used in
the special case of the Cr++aq.-Cr(OHz)a+++ system (103): exchange
of electrons between these two forms provides a path for rapid water exchange of the ion Cr(OHe)6+++. A powerful new technique involves
paramagnetic resonance measurements ; the lifetime of the paramagnetic
state which is being observed is reduced by electron exchange with the reaction partner, resulting in line broadening (136).Another new approach
depends on the broadening of the magnetic resonance absorption of a nucleus, caused by the presence of the exchange partner. (80)
A very ingenious method which in essence satisfies the conditions of no
net change, and which can be applied to ions that exist in d-1 forms, was
introduced by Dwyer and Gyarfas (36).This method exploits the change
in the rotatory power of a system in which a net change of the following
kind is taking place:
d-MCa++
+ Z-MCa+++ +d-MC;+++ + l-MCs++.
(4)
It is regrettable that with all the activity that has developed in measuring rates of virtual changes there has been so little acceleration of the
work for systems involving net chemical changes. Many of the important
questions of mechanism which are posed can be answered as well by the
study of orthodox reactions as they can by that of more exotic ones and, in
most instances, by the expenditure of much less effort. The techniques
which have been used are conventional, but for many of the important
reactions further development of methods for the measurements of rates
of rapid reactions is called for. Many of the systems of interest here involve intensely colored ions, so that the flow spectrophotometric method
can often be applied.
A m n g the reagents for which electron transfer can take place without
net rearrangement of the coordination sphere are the followingz: MnOrThe abbreviation phen represents 1,lO-plienanthroline, and dip represents
2,2-bipyridine.
14
H. TAUBE
- Mn04=, I r C l p - IrCls=, Fe(CN)64- - Fe(CN),F, M O ( C N ) ~ 'Mo(CN)s=, F e ( ~ h e n ) ~ +-+Fe(phen)g+++, Fe(dip),++ - Fe(dip):i

+ + + , 0s (dip) 3+ + - 0 s (dip) 3+ + + , Co (dip) 3+ + - Co (dip) 3 + + +. The
list is not intended to be exhaustive but only to provide an introduction to
reagents of this class. Oxyions, CN- and halide complexes are represented,
various central metal ions, cations and anions, and species of Coordination
number 4,6, and 8.
Both members of each couple are substitution inert, and for all leredox reactions which are possible for mixtures within this group, whether
involving virtual or net chemical changes, electron transfer takes place
much more rapidly than does substitution in the first coordination sphere.
Thus, even in the absence of direct proof that no labilization of the coordination sphere occurs in the activated complex, it is aimost certain that no
interpenetration of the coordination spheres occurs. The outer sphere activated complexes are not necessarily restricted to reagents of this kind,
but it is only for them that it is possible to assert with some confidence
that this type of activated complex operates. It is also characteristic of
the molecules in this group that the change in oxidation state takes place
with little change in the dimensions of the molecule; thus these systems
approach fairly closely the conditions which were chosen for the first class
which was given theoretical discussion in the previous section.
The group considered here has been restricted to molecules which can
reasonably be regarded as coordinatively saturated. There are numerous
molecules, some of which are particularly important, for which certain
elements only of the structure of the first coordination sphere are known
to be preserved on electron transfer. These include3: C102- - C102,U02+
- U02++,etc., Fe(cpn)* - F e ( ~ p n ) ~ +etc.,
, FeII(porphyrin) - Fe"'
{porphyrin) ,etc. Here the ligands specified in the formulas do not provide
for coordinative saturation of the molecule. In UOz+ + , for example, the
oxygens are colinear with the U atom, and groups held around the girdle
of this molecule are readily replaced; furthermore, it is not known whether
changes in the girdle coordination accompany electron transfer. On the
basis of structures of the reactants alone, no conclusions as to the geometries of the activated complexes can be reached for this class of reagents.
The system among those of present interest which has been most thoroughly studied is exchange between Mn04- and Mn04=. A number of
have done experiments with this reaction, but the
workers (21,61,77,115)
most successful and complete study is that of Sheppard and Wahl (115).
The rate of the reaction has been proved to be first order in each of the
reactants, as was tacitly assumed in the earlier discussion of reactions of
this class. The specific rate a t 0" is reported as 710 M-I sec-l, E as
a The
abbreviation cpn represents the cyclopentadiene radical, GH;.
15
10.5 kcal mole-1 and A S as -9 e.u., for a medium which contains NaOH
at 0.16 M.There is fairly direct evidence that in the Mn04= - Mn04reaction there is no interpenetration of coordination spheres. Symons (124)
has shown that when Mn04- is reduced to Mn04= by labelled water containing OH-, the isotopic composition of Mn04- is unaltered. This reaction cannot occur without accompanying electron transfer between MnO4and Mn04=. If, or example, the electron transfer were to take place by
an activated complex such as [03MnOMn03]-, Mn04- would be brought
into rapid exchange with the solvent. Some exchange does occur if Mn04=
is allowed to accumulate, but this is attributable to exchange of Mn04=
rather than to oxygen exchange in the act of electron transfer.
TABLE I1
SALTE F F E ~INB THE Mn0,- - Mn04- EXCHANGE
REACTION
(TEMPERATURE, 0
'
)
Medium
(electrolyte, molarity)
k (M-1 sec-1)
NaOH, 0.16
NaOH, CsOH, each 0.08
CsOH, 0.16
NaOH, 0.16, Co(NH&Cla, 0.001
NaOH, 0.16, NaaFe(CN)a, 0.001
730
1730
2470
1860
1180
The data obtained by Sheppard and Wahl for the Mn04- - Mn04=
reaction feature some interesting salt effects. Table I1 contains a summary
of the salient data reported by them.
Two effects are to be looked for in considering the influence of salts in
reaction rate: a general ion-atmosphere effect and, when ions of high
charge are involved, ion-pairing effects. Both appear to be illustrated by
the data of Table 11. The equilibrium ion-atmosphere distribution is unfavorable to the electron transfer so that a readjustment of these atmospheres must occur prior to the transfer. This readjustment may actually
involve an anion moving in a direction opposing that of electron transfer,
or a cation in the same direction. The difference in rate brought about by
changing the electrolyte from 0.16 M NaOH to 0.16 M CsOH seems very
large for an ordinary ion-atmosphere effect. The direction of change is
such as to suggest that the mobility of the cation is an important factor
in the electron transfer. In the present system, this may well be the case
in spite of the high mobility of the OH-; for an activated complex of
negative charge, cations will predominate in the ion atmosphere, and a
special sensitivity to cation influences will prevail. Even though the equilibrium properties are determined by a general ion atmosphere distribu-
16
H. TAWE
tion, for the rate process it may be economical of time to provide for
energy matching at the two sites by specific motion of a few ions rather
than by small readjustments for many. Such a means for electrostatic readjustment is all the more likely when there is strong ion pairing. Thus when
Co(NH8) 6 + + + is added to the Mn04--Mn04- solution, a substantial
fraction may be present as the outer sphere complex C O ( N H ~ ) ~ + + +
Mn04=. The electron then can follow the motion of thc tripositive
cation from Mn04= to Mn04-. Thus it is not a foregone conclusion for
these systems that electron transfer will be more rapid for the so-called
free ions than it is, for example, when one of the ions is present as part of
a complex. For example, the pair HC102 - CIOz may undergo electron
exchange as rapidly as CIOz- - CIOz; this relationship is actually indicated by the incomplete data which were obtained for this system ( 3 1 ) .
Whatever advantage, in respect to energy of activation, the system may
lose in changing from C102- to HC102, may be made up in the greater
economy of motion needed in the latter case, therefore leading to a more
favorable entropy of activation.
The acceleration in reaction rate produced by low concentrations of
Fe(CN)e" may be a result of catalysis by the Fe(CN)64- - F e ( C N ) F
couple, and the specific rate indicated for the reaction of Mn04= with
Fe(CN)6" is of the order of loa M-' sec-l (115).
Wahl and Deck (134) have succeeded in getting an estimate of the
specific rate for electron exchange between Fe (CN) 64- - Fe (CN)6= ; they
report for the rate coefficient a t 4", 1 x lo3 M-I sec-l.
For the majority of the systems which have been investigated, only
lower limits on the rates have been established. George and Irvine (44)
report for the reactions: Fe(dip) 3+ + - Fe(phen) 3+ + + , Fe (dip) + Ru (dip) 3+ + + , Fe (CN)64- with Fe (phen) 3+ + +, Ru (dip) 3+ + + or IrC16=,
M O ( C N ) ~~ -IrC16=, a lower limit for the specific rates of lo5 M-l
sec-l at 18'. Since for these systems involving net changes no question of
separation-induced exchange can be raised, the lower limit for the specific
rate reported can be regarded as established. Other reactions to which the
same remarks apply, and for which the minimum specific rate has been set
even higher ( k > 2 X lo6 M-l sec-l at O"), are: O ~ ( d i p ) ~ +with
+
Fe(phen)3+++,Fe(dip)3++, and F e ( ~ h e n ) ~ with
+ + R ~ ( d i p ) ~ + +($7).
+
Eichler and Wahl repeated the experiment of Dwyer and Gyarfas ($6)
with the Os(dip)s++ - O ~ ( d i p ) ~ + +system,
+
but the results do not agree
quantitatively. Whereas Dwyer and Gyarfas report that with solutions at
5x
M at 5", more than a minute is required for complete reaction,
Eichler and Wahl find that the reaction is complete in 15 sec, even when
the concentrations are reduced to lo-* M . From the work of the latter
authors, the specific rates of this reaction can be set as > 105 M-I
sec-'. The electronic paramagnetic resonance (EPR) method has been
17
used to set an upper limit of 4 x lo8 M-l sec-' on the rate of electron
exchange for the W(CN)8*- - W(CN)8= (138).
A number of other reactions have been studied using orthodox isotopic
tracer techniques for which complete exchange in the time of separation
was observed. The lower limits for the specific rates that can be calculated
are considerably smaller than those to which we have referred, and since
in many cases no proof could be adduced that the separation method did
not cause the exchange, these results are not reported. The review by
Amphlett ( 5 ) gives references to many of the literature reports on these
and other reactions in the entire field.
For the systems which will be discussed now, it is not as certain as for
those already discussed that the activated complexes are of the outersphere type. These systems differ in that one reaction partner is substitution labile (this is not so certain for Co(phen)3++ but is certain4 for the
Co (NH3)g++ and Co (en)3+ + complexes),so that the coordination sphere
of the reducing cation can readily be entered. However, because the data
indicate that enough amine appears in the activated complex to complete
the Coordination sphere of the reducing agent, and because no suitable
bridging group is present on the oxidizing agent, it seems almost certain
that the electron moves through the coordination spheres of both reaction
partners. The results on rates and energetics of the activated complexes
are summarized in Table 111.
TABLE 111
RATESOF ELECTRON
EXCHANGE
FOR SOME COBALT
COMPLEXES
IJ
Co(phen)3++- Co(phen)$+++
Co(en)a++ - Co(en)s+++a
CO(NH~)G,++- Co(NH&+++
Co(en),++ - Co(NH&+++
0.98
0.98
0.98
No specific effect of C1-
1-M level.
rtt
Temperk
ature
("C)
(M-l sec-l)
3.5
-0
25
45
25
6 X lo-'
<7 x 10-6
2
10-4
E
(kcal) Reference
-14
38
75
75
75
Not enough systematic work has been done to evaluate the various
factors that affect the rates for the class of reactions which has been considered, but a few general observations may be in order. Contributing t o
the high rates which are observed for most of them is the circumstance
that because of the electronic structures, very little change in geometry
takes place in the first coordination sphere on electron transfer, and such
'The abbreviation en represents ethylenediamine.
18
H. TAUBE
rearrangements as are necessary are relegated to outer spheres where the

intensities of the interactions are much reduced. The factors that contribute to the remarkable result that for many of these ions an electron
can be accepted without much increase in internuclear distance are fundamental to understanding the results on electron transfer. For most members of the class the electron population is such that the electrons can
avoid the ligands (the exceptions are in the Co+ + complexes), but this
is not a sufficient condition. Thus, while in the couple V(OHa)a++ V ( O H Z ) ~ + + the
+ electrons can also avoid the six ligands around each
ion, a considerable expansion in the first coordination sphere results when
V + + + absorbs an electron. Strong x interactions of the dc electrons with
ligand orbitals must be invoked, but to explain the observations it must
further be supposed that the de electrons of the metal are bonding rather
than antibonding (100, 111).One is led to conclude that bonding of this
type for dr electrons is particularly effective when the ligands have multiple bonds or conjugated multiple bonds, but it may not be limited to
such cases. Thus in terms of the geometrical features which have been
considered, I r C l p - IrCla= probably does not differ from Fe(CN)$- F e ( C N ) p , for example, but when a parallel case for the first row transition series is considered, for example VClz v e r m VC13, a large change
in dimension again results on change in oxidation number.
Besides the influence that the ligand has in determining the geometrical
and electronic structure of the complex, there may also be the direct (but
related) influence of affecting the distribution of the d electrons in the
complex and thus the probability of barrier penetration. Unfortunately,
there are no observations which expose separately the geometrical and
conductivity factors, but the second factor will readily be discernible
in some of the systems discussed in the next section.
The slower rates for the Co systems, summarized in Table 111 compared to those for systems described earlier, can be attributed to the
circumstance that large readjustments of the bonds must be made to provide sites equally favorable to occupancy by the electron. This circumstance is itself a result of the electronic structure. The seven d electrons
of Co++ can not all be accommodated in the dc orbitals and so prevent
the close approach of ligands to the metal; when there are only six d
electrons, as in Co+ + +, these can be accommodated in the dr set of orbitals and thus do not interfere as markedly with the close approach of
the ligands to the central ion. The difference in dimensions of Co(II1) and
Co(I1) complexes is probably much less marked for the Co(phen) complexes than it is for those of the Co-N-
type. Although Co(phen)3++
has a magnetic moment of 5.2 Bohr magnetons (%I),

the complex prob-
19
ably has a configuration close to that which would make the electronic
state with one unpaired electron the ground state. Thus the readjustment
required in the Co (phen) system preceding electron transfer is probably
a great deal less than it is for the Co-N-
systems. The role of the con-
jugated system of double bonds in phenanthroline in bringing the d electrons to the surface of the complex ion may also be important in causing
a greater rate of electron transfer for the Co(phen) as compared to
Co-N-
case.
In summary of the work discussed in this section it can be said that

the composition of the activated complex is known at least with respect to
the exchanging species, and it is further known, at least for some systems,
that reaction takes place without interpenetration of the coordination
spheres. Important work which remains to be done is, an extension of the
study of the salt effects, and the influence of ligand properties. Important
fundamental questions which remain to be answered are: What is the optimum distance of approach for electron transfer? Precisely how is the
energy matching brought about? How does the nature of the ligands and
the solvent influence the probability of barrier penetration? Rough ideas
as to the optimum distance of approach can perhaps be obtained from
studies of the influence of ionic strength. If for a given salt concentration,
the dimension of the ionic atmosphere is large compared to the separation
of the ions undergoing reaction, a normal salt effect can be expected, but
for the reverse case anomalies can be expected. If the distance of approach
required for electron transfer is large, these anomalies will appear in dilute
solution. It should be stressed that if the electron transfers over a large
distance salt effects will be observed even for a reaction involving a neutral
species, for example, C102 - ClOz-, and it is precisely for such a system
that there is the greatest hope of learning something about the geometry
of the activated complex from study of the salt effects.
VI. The Bridged Activated Complex
The term bridged will be used to imply a configuration in which an

atom or group of atoms is part of the coordination sphere of both cation
partners in the redox reaction. The characteristic feature is that the coordination sphere of a t least one of the partners has been entered, so that
new bonds are established in making the activated complex. It is by no
means necessary that this geometry correspond to the strong overlap
(8.2) case, although, as will be adduced from the evidence for the cases
20
H. TAUBE
which have been studied, strong electronic interaction between the cations
must be invoked.
It is precisely the feature that bonds are broken and new bonds are
formed in generating the activated complex that provides experimental
accesa to some features of mechanism for these systems. In properly
chosen systems, it can be expected that the new bonds formed in the process of activation will be retained when the metal ions separate to form
products. In contrast to the systems discussed in the previous section, in
which only the outer-sphere environment is changed on electron transfer,
the inner coordination sphere is changed, and the substitution lability
of the new combinations will in general be lower for the present cases.
Even though generally lower labilities for the new rearrangements formed
on electron transfer may actually obtain for the majority of systems, the
substitutions are still so rapid compared to electron transfer that only the
equilibrium configurations for the products are detected, leaving no hint
as to the path by which they are arrived at.
Choosing as representative a system in which the bridged activated
complex leads to net transfer of a group from oxidizing agent to reducing
agent, it is seen that the conditions which must be met to make possible a
conclusion about mechanism by examination of products are these: the
oxidizing agent and the oxidized product must undergo substitution slowly
compared ta electron transfer, but the reducing agent must be labile with
respect to substitution (to form the bridged activated complex, a t least one
partner must be labile to substitution). A variety of oxidizing agents qualify for such a test of mechanism, but of the aquo-ion reducing agents, only
Cr++ satisfies the imposed conditions.
A typical system for which the bridged activated complex can readily
be demonstrated is the following. When Cr++ is added to a solution containing (NH3)&oC1++ (this oxidizing agent is chosen because C1- is
only very slowly replaced by HzO or other groups), a rapid redox reaction
takes place forming CrCl++, Co++, and NH4+. The significant observation is that CrCl++ rather than Cr(HzO)e+++ is formed. If electron
transfer took place with the reaction partners widely separated, Cr
(OHZ)B+ + + would be the product, for once formed, Cr(OHz)g+ + + will
not change to CrCl++ rapidly enough for the formation of the latter by
this route to come into question. Thus we can conclude that the Cr-C1 bond
must have been formed prior to the electron transfer. An additional result
which confirms the conclusion that C1 transfer is direct is the following:
When the experiment is repeated with free radioactive C1- present, substantially none of this radioactivity appears in the product CrCl+ +. Thus,
it is shown that C1- transfer is not by way of loss and re-entry but rather
that it is direct (130).
21
Complex ions of the series (NH~)&o"~Lcan be formed in great variety, and these are useful for the present purposes in providing a survey of
groups which will act in the same capacity as C1-. Efficient transfer from
(NH3)5C0111Lto Cr++ is observed also for F-, Br-, I-, sod- ( I W ,N3-,
CNS-, carboxylic acids, PZOT4- (l95),PO4=, OH- (92). Particularly
significant is the fact that OH- transfers, because a path involving the
hydroxy complex is commonly observed in redox reactions of aquo cations
with each other. In the reaction of (NH3)&oOH++ with Cr++, transfer
of oxygen is substantially quantitative, but for the equally important
reaction of the aquo complex, the situation is much less clear. Some transfer, a t least, is indicated, but the results are quite erratic. The experiments
in which the aquo path predominates are difficult to perform, and various
factors operate to reduce the apparent transfer (92). The results reported
in reference 195 for the phosphate complex are wrong because the complex
salt used was not really the inner sphere complex. The inner sphere form
has since been prepared and shows normal transfer to Cr++. The only
oxidizing agent of this class for which net transfer has not been observed
is (NH3)5C~N03+
+. The reaction takes place rapidly, much more rapidly
than with (NHs) &oOHz+ + +, so that attack by Cr+ + a t NOS- is indicated; however, CrNOs++ is not observed. The failure to observe net
transfer may simply be the consequence of a high substitution lability for
CrN03++.
The reaction
(NH&CoOHr+++
+ Cr++
(5)
merits special discussion since it is a representative of the important class

of reactions in which both members are aquo ions. The system has the
useful and distinctive feature that a variety of testa of mechanism can be
brought to bear, and definite conclusions can be derived from the results.
In common with other systems of this class (see below), the rate law has
two terms (92) [the residue C O ( N H ~(or
) ~Cr(NH3)a) will frequently be
represented by Ro (or Rr)] :
The kinetic data are summarized in Table IV. The specific rate kz which is
the coefficient for the term (RoOH++) (Cr++) is obviously equal to kz'/
K where K is the dissociation constant of RoOH2+ + +. The value of K
has beendetermined (16) as 1.2 x
at 25" and p = 1.00; the associated
value of AH is 10 kcal. The value of k, is a revision of that reported
earlier (99), but the new value of k i agrees well with the earlier one.
The tracer work on oxygen atom transfer has already been referred
22
H. TAUBE
to, Of equal significance are the experiments on the isotope fractionation

effects for atoms bound to the oxidizing agent. There is a strong discrimination (1.035)in favor of the reaction of 016Hcompared to 018H in the
bridging position. This result indicates that the bridging does not merely
function to bond the metal ion centers but that stretching of the ColI1OH- bond is, in fact, part of the activation process. A significant discrimiTABLE IV
KINETICDATA(143)
Jt1
.\I-'
ser-'
FOR THE
REACTION
RoOH2+++ Cr++
k?'
k?
E,
E2
sec-1
M-1 sec-1
kcal
hcal
(p =
1.00;AT 20C)
A&$
e.u.
A&$
e.u.
~~
In HzO
In D20
0.50
.I3
1.58
0.40
1.5 X loG
-
3.3
4.8
-48
- 15
-
nation (1.0035for XI4compared to Pi)is also reported for the nitrogen in

the bound ammonias so that stretching of Co1"-NH3 bonds in the activation process in indicated. However, complete interpretation of this result
is not possible, because it is not known whether the observed effect is
distributed over the fiye nitrogen positions or m-hether it is localized in a
single bond.
The data on the effect of changing the solvcnt froiii H20 to D20 are
incomplete, because the change in K with solvent in not known. It is unlikely, however, that the change in K will account completeIy for the decrease in kz' by a factor of 3.8 which is observed when DzO is used in
place of HzO [it should be noted that exchange of the aquo protons is
rapid, so that in DzO the reactant species is (NH31sCoODz+++],and a
residual Hz0-D20 effect is undoubtedly left. The origin of this effect may
lie in this, that stretching of the 0-H bond takes place in the activation
process. This conclusion is also strongly indicated by the marked change
in kl which occurs when the isotopic change in the solvent is made. However, the conclusion cannot be insisted on for the kl reaction, because the
case for oxygen transfer in the reaction is not as strong as for the k2' path.
In the context of these results it should be mentioned that even for a system which does not involve a hydrogen-containing bridging group (and
which certainly does not react by H-atom transfer), namely the reaction
of (NH3)&rCl++ with C r + + , a decrease in rate by approximately 30%
is observed (98) in changing the solvent from H20 to D20. Thus there is
a noticeable solvent isotope effect even short of stretching H-0 bonds in
the bridging group or transferring H from the coordination sphere of reducing agent to oxidizing agent.
23
For a large number of reactions of Cr(II1) complexes with Cr++, a

bridged activated complex is obviously also involved. Among these is a
reaction of almost classical interest: the catalysis by Cr+ + of the dissolution of anhydrous CrC13 (I).The product of the reaction has been shown
to be CrC1+ + [rather than Cr (OH2)e+ + + as would be expected for ordinary dilute solutions if complete equilibrium were rapidly established] ,
and the C1- retained has been proved not to have passed through the solution (129). The reaction can be formulated as
CrCla
+ Cr*++ = Cr++ + 2C1- + Cr*Cl++.
(7)
It is difficult to see an alternative formulation that accommodates the

observations which have been cited. The rapid dissociation of CrC12+ under the influence of Cr++ (I29)-this also proceeds rapidly only to the
CrCl+ + stage-can be understood in a similar manner:
CrC12+
+ Cr*++-+ Cr++ + C1- + Cr*Cl++.
(8)
+ Cr*+++ Cr++ + Cr*Cl++
(9)
The next stage :

CrCl++
does not lead to a net change, but through the use of isotopic tracers it
has been shown (198) to be a rapid change also, and the similar reactions
with F-,NCS-, and N3- as bridging groups have also been studied (11).
A group of reactions which is analogous to those just mentioned, but
in which the operation of the mechanism in question leads to net changes,
is exemplified by
(NH&CrCl++
+ 5H+ = 5NH4+ + CrCl++.
(10)
The systems take this course only when Cr++ is present; in the ordinary
aquotization process the products (NH3)&rOH2+ + + X - are formed.
The rates of reaction are strictly first order in the Cr(II1) complex and
in the catalyst Cr++, and are independent of acidity over a wide range
of concentration. The mechanisms are obviously similar to those discussed
in the previous paragraph. Kinetic data on the reactions of the Cr(II1)
complexes are summarized in Table V.
The data shown in Table V make possible comparison of the efficiency
of various groups as electron mediators.6 As shown by the quantitative
data for Cr (111) complexes, and indicated also by qualitative observations with Co(II1) complexes, the iodo complexes yield to reduction much
more rapidly than the fluoro. This result indicates that the electron transfer does ngt occur by direct overlap (76) of the d orbitals of the metal
ions but rather through the agency of the bridging groups (148).The order
Use of this term was suggested t o us by Prof. John R. Platt, Department of

Physics, University of Chicago.
24
H. TAUBE
TABLE V
THERATESOF REACTION
OF Cr++ WITH V.uuons Cr(II1) COMPLEXES
k
Oxidant
CrF++
CrCl+
CrBr++
CrNCS++
CrN3++
(NH&CrF++
(NH&CrCl++
(NH&CrBr++
(NHa)5CrI+
frane-CrC12+(69)
+
(M-1 Sec-1)
Temper
ature
(C)
2.6 x lo-*
8.3 f 2, 9.1 & 1
>60
1.8 x 10-4
>l.2
2.7 x 10-4
5.1 X 10-2
3.2 X 10-1
5.5 f 1.5
-23 x 102
27
0
0
27
0
25
25
25
25
A m
kcal
AS$
e.u.
13.7
-
-20
13.4
11.1
8.5
-30
-23
-33
-
Reference
11
128,ll
11
11
11
98
98
98
98
189
of rates for the Cr(II1) halide complexes is the same as is observed for
attack of a series of organic halides by a free radical reagent such as Na.
Although the observations cited thus far are accommodated by the simple
statement that we are dealing with halogen atom removal by Cr++, this
is not an apt description for the entire field of phenomena. The circumstance that d electrons or orbitals are not used in the same way for binding as are the orbitals in question for a carbon atom center introduces
features which are, as we shall see, absent in the carbon case.
The rate comparison for RrNCS++, and RrN3++, which holds qualitatively also for the corresponding Co complexes, is especially interesting.
Part of the reason for the greater rate for N3- may be that attack at the
remote end in this case leads to a stable species, but with NCS-, reaction
at the remote site necessarily forms a system of higher energy. Ball and
King (11) have pointed out that, if the reducing agent maintains octahedral coordination, attack at the atom bearing Co or Cr is impossible for
steric reasons.
An important comparison is that of the rate of reaction of Cr++ with
(NH3)&o+ + + and with (NEfS)&oOH2+ + +. For the completely ammoniated species, the redox reaction is very slow, slower by at least a
factor of 100 than for the aquo ion. The difference in rate can be ascribed
to the availability of an electron pair when H20 is coordinated to a central
ion; all electron pairs are occupied for a coordinated NH3. The mechanism by which the hexammino ion is reduced is not known; since the
bridged activated complex has been made difficult of access, electron
transfer may in fact take place through the coordination spheres of the
MECHANISMS OF BEDOX BEACTIONS
25
two reactant ions. To arrive at a bridged activated complex in the casc

of coordinated NHs would require the dissociation of protons. Even with
OH- as the attacking base, this requires about 14 kcal (7) in the way of
activation energy; when only water is available, the activation energy
would be considerably greater.
Experiments with oxidizing agent8 of the pentamminecobalt (111)
class and with complex organic molecules occupying the sixth coordination position have led to some new and interesting observations. In all of
the systems referred to, only a single (NHs)&o is attached to each ligand;
in every system, transfer of the organic ligand to chromium is observed on
electron transfer, so that direct attack of Cr++ on the organic ligand can
with confidence be accepted as a feature of the reaction mechanism. The
data on these systems are summarized in Table VI.
In the reaction of the acetato and butyrato complexes, attack can be
only at the carbonyl group, and following the argument of Ball and King,
it is likely that it takes place on the oxygen which does not have the
Co(II1) residue. In view of the similarity of the rate of reaction of the
acid succinate complex to that of the acetato or butyrato, it is reasonable
to conclude that for the acid succinate the attack is also at the carbonyl
adjacent to the Co(II1) center; this conclusion is supported by the fact
that the methylsuccinate complex also reacts at about the same rate. The
acceleration noted in proceeding to the succinate ion can be attributed to
the influence of negative charge and/or of chelation of Cr++ in the activated complex.
Fumarate and succinate appear to afford a clear cut intercomparison ;
the more rapid rate for fumarate, in spite of the circumstance that there
is no possibility of benefit from chelation in the activated complex, suggestn that, for this ion, attack is not restricted to the carbonyl adjacent
to the Co(II1) but takes place at the remote end. The conjugated bond
system provides a mechanism for electron transport, and there is made
available to the system the benefit of avoiding close approach of the positive charges of the cations. The comparisons including the phthalate complexes support the conclusion reached. For the meta and ortho complexes,
the conducting systems of bonds are lacking, in the first case because of
the relative positions on the benzene ring, and in the second because the
close approach of the carboxyls prevents them from assuming a coplanar
configuration. In the p-phthalate complex, the conjugated bond system
can be achieved, and the rate of reaction is correspondingly more rapid.
An important new effect of acid does not fit the form adopted for reporting the data in Table VI. Examples of the usual behavior, that the
reaction with a bridging anion is more rapid than with the acid in the
bridging position, are shown there. With fumarate as the bridging group,
26
H. T A W E
TABLE VI
CARBOXYLIC
ACIDSAS ELECTRON
MEDIATORS,
AL
Ligand
Acetate
Butyrate
Crotonate
Succinic acid
Succinate ion
Me-succinate ester
Oxalic acid
Maleic acid
Fumaric acid
Fumarate ion
o-Phthalic acid
o-Phthalate ion
m-Phthalic acid
p-Phthalic acid
Temperature
("C)
25
25
25
25
25
25
5
5
5
5
25
25
25
25
E
(kcd)
k
(M-*sec-l)
= 1.00
G
S
e.u.
Reference
126
186
126
114
114a
114a
126
0.15
0.08
0.18
0.19
-1.0
0.22
>20
126
>20
0.5
-2
0.057
10
0.10
7.5
33
114a
114a
5.1
47
114~
2.6
56
114a
114a
114a
-40
these features are also present, but there is an added term in the rate law
of the form k(RoFH++) (H+)(Cr++). The acceleration by acid is so
marked that in 1 M HC1O1, 80% of the reaction proceeds by the acid accelerated path. This effect, which is marked also for the p-phthalate complexes, presumably arises from this: In a complex constituted as follows,
(?r:H3)&0"' 0
I
c-c=c-C
0 Cr"
//
(11)
'OH
conjugation between the metal ion centers is incomplete. Placing a proton

on the carbonyl oxygen adjacent to the Co(II1) causes a redistribution of
the carbonyl electrons and improves the conjugation through the molecule.
As is required by this interpretation, there is no hint of acceleration by
acid when bifunctional ligands which are unconjugated (succinate, 0- and
m-phthalates) are the bridging groups.
The very rapid rates of reaction observed for the oxalate and maleate
complexes are not accommodated by any of the factors which have been
discussed thus far. The experiments with the succinate complexes show
the chelation effect to be small. Chelation implies attack by Cr++ at the
carbonyl adjacent to the Co(II1), and if this is the case the benefit from
having a conjugated structure is not required; in fact, if there is chelation
with maleate, the coplanar arrangement of the carboxyls required for effec-
27
tive conjugation is impossible. In explanation of the difference between

maleate and fumarate, it is suggested that two different mechanisms of
electron transfer by the bridging groups must be discussed. In one of these
mechanisms, exemplified by fumarate, as electron density is removed a t
one end of the conjugated 7r system, it is replaced at the other; in the second, there is net transfer of an electron from the reducing agent to the
bridging group. Such a mechanism does not require the same configuration of atoms as does the one discussed for the fumarate ion and will come
into question for systems which have unoccupied low-lying orbitals. The
fact that, of all the bridging groups discussed, oxalate (88) and maleate
are the most easily reduced, fits in with the suggestion made about the
mechanism of electron transport for these groups as electron mediator.
The suggestion also explains the difference between maleate and o-phthalate, which have similar geometry about the carboxyls, but differ in reducibility. It should be mentioned that with neither fumarate nor maleate
is there a change in configuration of the bridging group on electron transfer. Isomerisation of the maleate is not required by the mechanism proposed because the Cr+ +, in attacking maleate, may impose restrictions on
its geometry.
Thus far the only ligand effects which have been discussed are those
concerned with their role as electron mediators in bridging positions. This
takes account of only one of the eleven coordination positions which need
to be considered for a bridged activated complex between two reactants of
normal coordination number six. It is to be expected that changing groups
in nonbridging positions will also influence rates, and such effects can
easily and unambiguously be demonstrated. Taking (NHJ &oOHz+ + +
as oxidizing agent ana Cr++ as reducing agent, it is observed (125)that
there is a marked acceleration of the reaction by SO4=, and particularly
by pyrophosphate, and both of these ions are incorporated into the product
Cr(II1) complex. That a group other than these ligands is involved as a
bridging group is demonstrated by using (NH,) &oCl+ + as oxidizing
agent; when pyrophosphate is present, both C1- and pyrophosphate are
incorporated in the product Cr (111) complex. Groups differ enormously in
their capacity to accelerate the rate of electron transfer by simple ligand
intervention. Thus, in the system now under discussion, pyrophosphate
is very effective, SO,= less so, and an effect of C1- at 0.1 M level is not
discernible (92).It seems likely that there will be a parallelism between
the ability of the ligand to stabilize Cr(II1) over Cr(I1) and its ability
to accelerate the rate of oxidation by simple attachment to Cr++ in the
activated complex (the relationship would presumably hold in the reverse
way also). The ligand effect of C1-, though slight, must exist to explain
cataIysis by Cr++ of the CrCI---C1- exchange (128) ; or, looking at it
28
H. TAUBE
in the reverse direction, to explain why CrC12+ reacts so much more rapidly
with Cr++ than does CrCl++ (see Table V).
Cis and trans positions can be distinguished in the bridged activated
complexes under present discussion, and in considering the influence of
nonbridging ligands, it is of interest to differentiate the effects at the
two positions. Orgel (99) has suggested that for Co(II1) and Cr(II1)
complexes the incoming electron is accepted in the d,2 orbital, the energy
of this orbital being lowered to the necessary extent by moving groups
trans to each other from the metal ion center. The stretching of the
Co(II1)-OH- bond, when OH is the bridging group, has been demonstrated (92) ; Orgel's suggestion that the group trans to the bridging group
also moves out helps explain some recent observations which have been
made. It is found (197) that trans-Cl2en2Cof++ is reduced more rapidly
by Cr++ (and other reducing agents) than is the cis form. The result is
surprising, a t least in the context of a philosophy that omits consideration
of the individuality of different central ions. Thus, attention (1-44) has
been directed to the advantages of double bridges for electron transfer,
and it does seem reasonable, if account is taken only of forces external
to the ions, to use both negative ions in reducing the energy of interaction between ions of the same charge. On Orgel's interpretation the
trans effect operates in this way: stretching of a HsN-Co(III) bond trans
to the bridging group is necessary to lower the dS2orbital to the same extent as is necessary for a trans-C1-. In a similar way, the relative rates
for (NH3),CrC1++, (H20)5CrCI++, and truns-C1(H~O)4CrCl++(see
Table V) can be understood. But there is a disturbing feature in the
comparison of (NH3)&rCl+ + with (H20)&rCl+ + , for the difference in
rate appears to be in the entropies rather than in the energies of activation (Table V ) .
Some general observations on the energies and entropies of activation
of redox reactions which proceed by bridged activated complexes are in
order. These quantities, even for the few systems for which they have been
determined, cover the range 4 to 14 kcal and -20 to -45 e.u. respectively.
The ranges overlap with those for the outer-sphere activated complexes
and, except possibly in extreme cases, it is not safe to use the magnitude
of these quantities as diagnostic of mechanism. The comparison of A S
for the process
(NH&CrBr++
and
+ Cr++ +;
A S = -33 at p = 1.0 (98)
(12)
aS = -16 at, p = 0 (S)

(NR3MbBr++ Hg+++;
(13)
is instructive. In both systems the activated complexes presumably have
the 8811163 general geometry and are of the same charge type, so that the
entropy effects caused by concentration of charge in the dielectric should
MECHANISMS OF REWX REACTIONS
29
be much the same. The disparity in the values of A S is very great, however, and would probably be even greater if the entropies were compared
at the same ionic strength (4). The differences in A S are in part attributable to this, that much more in the way of simultaneous bond readjustment may be required in the electron transfer than in the substitution
reaction. There is evidence that, in the latter case, a true intermediate of
coordination number 5 is formed from the ammine complex (106),and
thus the principal bond dislocations may be only the motion of Br- from
Co(II1) to Hg++. In the electron transfer case under discussion, simultaneous with the movement of Br- from Co(II1) to Cr(II), motion of a
group trans to the bridging Br- away from the Co, and again trans to the
Br-, toward the Cr may be necessary. In addition, a change in the bond
angle from something which at equilibrium may be less than 180, to
180, may be required, for in the colinear arrangement, there will be the
most efficient overlap of the d,, orbitals of the metal ion with a PO orbital
in the Br- (142).
It is important to consider whether the bridged complexes which have
been discussed are merely activated complexes or whether binuclear species
of similar geometry must be invoked as intermediates. In the cases encountered thus far, the concentration of such intermediates appears to be
so small that direct detection is difficult, if not impossible, yet there are
powerful arguments which support such a formulation of the mechanism.
When an activation energy as small as 4 kcal is in question for the
bridged activated complex, we face the difficulty that the activation energy
for substitution on the Cr++ probably exceeds this value. A mechanism
in which the reaction occurs in a single step,
(NH&.CoOHr+++ Cr+++,
(14)
would require activation energy at least as great as the activation energy

for substitution in the more labile partner (here Cr++), and considerable
additional contributions from other sources can be expected as well. The
situation would not be materially improved by adopting the formulation
(we assume Cr+ +aq. is a hexa-aquo ion) :
Cr(OH2),++P Cr(OH2)6++ 4- HzO
(15)
Cr(OH&,++4-(NH8)6CoOH2+++ products
(16)
AH# over-all would necessarily exceed AH for process (15), and 4

kcal appears to be a conservative lower limit for AH, even taking into
consideration the probable tetragonal symmetry of Cr (OH,) g+ +. The
high cost in energy for substitution can be avoided if we formulate the
substitution process giving rise to the binuclear intermediate species as an
30
H. TAUBE
equilibrium. In this situation, the reaction can be represented as:

I:,
RoX++
+ Cr(OH?),++C [RoXCr(H,0)6]++ + H,O
07)
1-1
[RoXCrj++-+ products.
1. Y
Then the rate of reaction will be given by

k,k,
(L
+ k2) (Cr+*) (RoX++),
If k v 1 > > k2, AH$ over-all = AH1 + AH2$. The value of AH1 can be
considerably less than AH$ for substitution, because energy is paid back
in the new Cr++-X bond formed, and there may also be a gain from energy of interaction of the d electrons of the metal ion centers. But we must
now consider why k-l can be larger than k2, even though the corresponding activation energy is probably larger. The compensating factor is necessarily in the entropies of activation. The entropies of activation for substitution are usually small in simple cases, and in a situation such as the
present, involving separation of positive charges, may well be positive;
for a reaction such as (18), as we have already seen, the entropies of activation have large negative values.
This analysis suggests, particularly for systems with low values of
hHS total, fairly strong interactions of d electrons of the metal ion tenters. This interaction leads to a lowering of AH1 (and in certain cases may
be so low as t o make it possible t o detect the binuclear intermediates) , but
it benefits the rcaction path in process (Is),also. In the case of strong
overlap, as is postulated, the energy required to compress the Cr-ligand
bonds and to stretch the Co-ligand bonds is reduced, if the electron in
question is to some extent distributed over both centers. In the absence of
effects such as these, it is difficult to understand how a reaction between
a dipositive and a tripositive center can take place by such intimate approach as obtains in a bridged activated complex with an activation energy
as low as 4, or even as low as 10, kcal. The most instructive way to compare various groups as electron niediators would be to refer to the coefficients for the rates expressed in terms of the concentration of the binuclear
intermediates, but so little is known of the relative stabilities of such species that an analysis of this kind must be put aside for the present.
The electronic interpretations of the observations for the particular systems which have been discussed will form an important part, and perhaps
the basis, of generalizations which include other cases, but they cannot
be adopted without careful scrutiny and possible modification for systems
of different electron structure. We have been dealing in every instance
with R situation in which the oxidizing agent has dc orbitals only occupied
31
and cannot readily accept additional electrons, and the reducing agent, in
every case Cr++, has a dy electron outside a stable half-filled subshell of
the dc: electrons, In fact, even the feature characteristic of the systems so
far discussed, that group transfer from oxidizing agent to reducing agent
accompanies the electron transfer, does not necessarily apply to other
systems in which a bridge activated complex nevertheless provides the
path for reaction. For example, a binuclear complex is undoubtedly formed
in the reaction of IrC16= with Cr++, Cr and I r sharing the bridging group,
yet the net reaction, after the short time needed to permit dissociation of
the binuclear product complex has elapsed, is to form IrC16= and
Cr(OH2)6+++.In this system, the IrCle group, having a vacancy in the
de set of orbitals, can absorb an electron without a major readjustment of
distances; the Cr-C1 distance does decrease when the electron is lost to the
I r ; the product complex, however, more readily undergoes substitution at
the Cr(II1) than at the Ir(II1) center when it is aquotized to form the
separate (129) ions. In the inorganic systems, the case can readily be
imagined that electron transfer is accompanied by the movement of an
, electronegative group from reducing agent to oxidizing agent-it
is for this
reason if for no other that it is inappropriate to discuss these reactions as
atom transfers in analogy to the case of free radical attack at carbon.
A system which would undoubtedly show such reverse transfer is that
of a CrfIV) complex reacting with Cr(II1). Cr(1V) complexes are almost
certainly substitution labile (97, 126), so that when Cr*X+ + and Cr (IV)
react to form Cr*(III)XCr(IV), the ion in the lower oxidation state
brings the bridging group into the activated complex; on electron transfer
Cr (111) becomes Cr (IV) ; when the product binuclear complex dissociates, it separates into Cr(1V) and CrX++. Attempts (97) to demonstrate this reaction have thus far failed, not necessarily because the chemistry outlined is faulty, but probably because Cr(1V) , being unstable, is
a rather intractable species.
To summarize this section: the evidence for a bridged binuclear complex as an activated complex and an intermediate has been outlined, the
efficiency of various groups as bridging group compared, the particular role
of the bridging groups discussed, as well as the role of groups in the activated complexes in nonbridging positions, and the rationalization of these
observations in terms of the electronic structures of the central atoms and
ligands attempted. Notably lacking is the extension of the conclusions to
systems other than Co(II1) and Cr(II1) as oxidihg agents and Cr++
as reducing agent. This does not imply that the particular mechanism applies only to these reactants, but the examples have purposely been limited
to those in which the conclusions about mechanism are unambiguous. In
the next section, some of these results are applied to an analysis of observa-
32
H. TAUBE
tion for the more labile systems, in which direct conclusions about mechanism are more di5cult to derive.
VII. Reactions Proceeding by Mechanisms of Uncertain Classification
The characteristics which have made possible the conclusions about

mechanism for the systems considered thus far are (a) both oxidizing
agent and reducing agent undergo substitution much more slowly than
electron transfer or (b) the oxidizing agent and the oxidized product complex undergo substitution much less rapidly than the electron transfer
takes place, but the reducing agent is substitution labile. The majority of
the systems do not meet these sets of conditions, and for them the conch:ions about mechanism are more limited in scope. The reactions are, nevertheless, as interesting and important, and it is a challenge to devise new
experiments in the hope of differentiating between the possible mechanisms. The models for the activated complexes which have been proposed
to explain the electron transfer in a typical case, such as F e + + - F e + + + ,
tire the two which have been discussed thus far, together with a modification of the outer-sphere activated complex in which a specific role is assigned to protons in carrying the electrons from the reducing agent to the
oxidizing agent. Because the greatest effort has been expended on the
Fe+ + - Fe+ + + reaction, with respect to both experimentation and theorizing, this system will be considered first, and such conclusions about
mechanism as seem justified drawn. The problems of mechanism for other
systems are similar, but usually some new element is featured, so that
most of the systems for which rates have been measured, and which have
come to the attention of the author, are referred to.
The early lore on the rate of Fe+ + - Fe+ + + electron exchange is confused by paradoxes (19,68,79,152,136).
Partly because of this, the problem challenged a large number of skillful experimenters] and a general
growth of activity in the entire field ensued. The first definitive experiments were done by Dodson (32); his results are consistent with those of
Linnenbom and Wahl (79)and with those of Betts et al. (19),but not
with those of van Alten and Rice (I,%?), or those of Kierstead (68).
I n Table VII is a summary of the data which have been obtained on
the Fe++ - Fe+f+ reaction. The observation that the values of E are so
nearly the same for a variety of paths has inclined some authors (112)
to the view that a common process, assumed to be H-atom transfer between the hydration spheres, is taking place for all the systems. This view
has been strengthened by the observation made by Hudis and Dodson
(62) that the specific rate coefficients for the terms
(Fe++)(Fe+++)and (Fe++)(F'e++
(H+)
A)
33
MECHANISMS OF BEDOX REACTIONS
TABLE VII
SUMMARY OF KINETICDATAON Fe++ - Fe+++EXCHANGE
(TEMPERATURE
25", p = 0.5)
ko
(M-1 Sec-1)
Fe+++
FeOH++
FeNCS++
FeF++
FeC1++
FeF2+
FeCIz+
FeF8
0.87
1010.
41.5
9.7
9.7
2.5
15
-0.5
MS
Reference
9.9
7.4
7.5
9.1
-25
117
ii7
74
8.8
9.5
- 20
-28
-21
-24
-22
63
117
63
I17
63
each decrease by a factor of approximately 2 in changing to DzO as the solvent. Neither argument is convincing. The HzO - D20experiments with
(NHs)&oOH2+ + + as oxidizing agents show that isotope effects even
larger than those reported by Hudis and Dodson are observed when transfer of hydrogen atoms does not take place-although substantial weakening of an 0-H bond must be inferred. Comparisons of the energies and
estropies of activation can be made with systems which are known to
react by bridged activated complexes and can equally well be taken to
indicate that the Fe++ - Fe(II1) reactions are similar in mechanism.
Since the Fe(II1) complexes in question are substitution labile, the
composition of the activated complex does not establish its geometry, nor
can the form of the rate law (FeX++) (Fe++) ,for example, even be taken
to mean that Fe(II1) brings the group X- into the activated complex.
Thus, for the rate term (FeX++) (Fe++), all of the following formulations will satisfy equally the kinetic requirements:
(a) Fe+++aq.
(b)
(4
(4
XFe++aq.
(no X--Fe bonds)
(XFe++)aq.
Fe++aq.
Fe+++aq.
FeX+ aq.
[FeXFe]4+
Observations made by Lewis et aE. (75), Anderson and Bonner ( 6 ) , and

others (98),for systems of the charge type +2, +3, show that when substitution in the first sphere of coordination of the oxidizing agent is excluded, C1- does not exert its usual catalytic effect. Thus it seems likely
that, in the Fe++ - Fe+++ systems, activated complexes (a) and (c) can
34
H. TAUBE
be excluded. It is admitted that this conclusion is based on reasoning by

analogy, but there appears to be no basis for concluding that such reasoning will fail here. The argument is strengthened when it is noted that,
when C1- is placed in the first coordination sphere of Co (111) and Cr(II1) ,
the usual large acceleration of rate by C1- is observed. The fact that the
reaction of Cr+ +, in catalyzing the dissociation of CrC1+ + [for this purpose an activated complex such as either (b) or (e) would serve], is very
slow compared to electron transfer accompanied by atom transfer [ activated complex (d) 1, taken with the previous arguments, suggests that in
the present system the activated complex (d) also serves as the reaction
path.
The comparison (12.9) of the relative rates at which a series of Fe(II1)
complexes react with Cr++, with the relative rates a t which the same
complexes react with Fe+ + , suggests that similar mechanisms operate in
the two series. The argument that the Cr++ - Fe(II1) reactions, at least
with catalytic groups other than OH- in the activated complex, proceed
by a bridged activated complex, is this. When Cr++ reacts with Fe+++ in
the presence of X - , catalysis by X - is observed and CrX++ is formed
quantitatively by this path (129).When Cr (OH,) 6 + + + is oxidizing agent,
vatalysis by C1- has not been observed (6, 92), and CrCl++ is not
formed. The formation of CrCl++ in the reaction of Fe(II1) with Cr++,
in the presence of C1-, proves that a Cr-CI bond is joined in the activated
eomplex. The comparison of the sensitivity to C1- catalysis of Fe(III),
which is substitution labile, to that of Cr(II1) or Co(III), which are not,
shows that substitution in the first coordination sphere is a condition for
marked catalysis by C1-. Hence, the conclusion that C1- makes a bond
not only to Cr+ + but also to Fe+ + + in the activated complex is strongly
indicated if not, in fact, proven. This line of argument, therefore, also
supports the bridged activated complex for the reaction of Fe++ with
Fe+++.
At best, the kind of argument used here can indicate that a major part
of the reaction takes place by a particular reaction path. Based as it is
on gross comparisons of rate, it does not make possible an assessment of
the contribution to the total reaction of rival, but minor, paths. And yet,
only such a complete description can be regarded as satisfactory. To take
n particular example, we are perhaps prepared to accept the conclusion
that with X = F a part of the reaction takes place by the bridged activated
*Plane and Taube report marked catalysis by C1- of the Cr++- Cr+++reaction,
but the data of Anderson and Bonner allow at most a slight effect. It is quite likely
that the conclusion of the former authors is wrong. They observed erratic and
unexplained catalysis in several of their experiments and were probably misled by
such accidental catalysis appearing in the experiment on the effect of C1-.
35
complex. But in this case it is not unlikely that an activated complex such
as
b+44+
also accommodates a good fraction of the reaction. It obviously would be

of some interest to discover how the reaction divides between the two
routes for the series of the halides.
Very little that is definite can be said about paths involving more than
one nonsolvent ligand in the activated complex. When a single anionic
ligand is present, it is reasonable, in view of the rather general behavior
that anions are better electron mediators than neutral molecules, that this
ligand be used in the bridging position. The most reasonable disposition
of a second anionic ligand is much more difficult to choose; further work
on effects for nonbridging ligands using substitution-inert oxidizing agents
is required to illuminate this subject. As before, it is likely that only configurations in which the ligands are directly attached to either Fe++ or
Fe+++ would appear to come into consideration. But even granting this,
the second X - may be in association with the oxidant, with the reductant,
or with both simultaneously as in a system in which a double bridge is involved. The trans effect noted in the reactions of Co(II1) complexes with
Cr++ does not necessarly hold for oxidizing agents of differing electron
structure. Thus when Fe+ + + is being reduced the dislocations required
to cause this center to accept the incoming electrons are probably more
symmetrical than those sufficient for Cr(II1) or Co(II1) as the oxidizing
agent.
A number of other electron exchange reactions of the +2, $3 charge
type have been studied, those of Cr, Co, V, and Eu. Anderson and Bonner
(6) observed for the Cr++ - Cr+++ exchange the rate law
Even in strongly acidic solution the acid inverse path is the dominant one,
M-l
and a good value of k, was not obtained. They report 18 x
sec- as an upper limit for kl at 25C and p = 1.0 and k2 under the same
sec-l. From the data of Plane and Taube
conditions as 1.0 X
(103) on the catalysis by Cr++ of the Cr(OH2)6+++- H20 exchange,
and using Anderson and Bonners conclusion that the acid inverse path is
dominant, kb is calculated as 1.6 x
sec-l a t 25OC and p = 6.0, in
satisfactory agreement with the more direct measurement of Anderson
and Bonner. Using the known value (107)of K for Cr(O.&)s+++ at
p = 0.068, and assuming the same variation of K with p as for Fe+++aq.
(89), kz,the coefficient for the term (Cr++) ( O H + + ) is calculated as
36
H. TAUBE
1.4 M-l sec-l. Finally, using the reported temperature coefficient for
kz' and 9.4 kcal as the value of aHDfor Cr (OHz)8+ ++ (lor),EZis found
to be 12.6f 2, and ASis calculated as -16
7 eu. There is no proof that
the reaction of Cr++ with Cr(OH2)8+++ involves a bridged activated
complex, but in view of the known behavior of other Cr(II1) complexes
with Cr++, there is every likelihood that this is indeed the case. Such
proof could be obtained by comparing the specific rate for the Cr++ Cr (OHe)6+ + + exchange measured using radioactive Cr with the value
obtained by the method of Plane and Taube; in calculating the rate of
electron transfer from data on the rate of water exchange catalysed by
Cr++, an assumption is made as to the number of HzO molecules which
exchange per electron transfer. The value reported was calculated assuniing 6; if a bridged activated complex is involved, then only 5 will be exchanged, and the value of k calculated should be only 5/6 of that reported.
Careful measurements would be required to expose the 2076 difference in
rate, and the present data are by no means accurate enough to settle the
question.
The specific rate for the exchange between Co++aq. and Co+++aq.
has been measured as 0.77 M-l sec-I a t 0" and 1 M HCIOa (2%). No data
have been published on the variation of rate with acidity. The comparison
of the rate of this reaction with that of Co(NH3)6+++ - Co(NH3)gf+
is interesting. Part of the difference is probably attributable to the circumstance that the bridged activated complex is readily accessible for the
aquo ion but not for the ammonated one; an additional factor likely is
this, that with NH3 as ligand, a greater distortion of the coordination
sphere is required to make the energy of the electron to be transferred
and C0Y-I (Y repmatch at the two sites. The exchange between
resents the ethylenediaminetetraacetate) has been shown ( 3 ) to be very
slow. The specific rate a t 85" is 2.1 X
M - 1 sec-l, and E is 22
kcal. The nature of the activated complex in this system is not knownelectron transfer may be through the intact coordination shells or may require some dislocation on the substitution-labile Co (11) complex. Adamson and Vorres ( 3 ) have discussed the relation of crystal field stabilization
of electronic states to rates of electron transfer in Co(I1) - Co(II1)
systems.
The exchange of Mn+ + with Mn+ + + appears to be measureable. The
experiments by Adamson (2)indicate that k at p = 3.2 and 25OC is about
100 M-l sec-l. The exchange reaction V++aq. - V + + + aq. would provide an interesting comparison with the systems already described in this
section because, for the pair, only electrons in the dr levels come into question. In the only work published on this reaction (70), complete exchange
in 1 min at a concentration level of 0.1 M is reported, and no conclusions
37
MECHANISMS OF REDOX BEACTIONS
as to the rate of electron transfer can be drawn. The issue of relation of

rate and mechanism to electronic structure is also raised in comparing the
kinetic behavior of Eu++aq. - Eu+++aq. (86) with the systems already
discussed. The rate of exchange is very slow, and no contribution by an
aquo or even hydroxo path could be detected (the reduction of water by
Eu+ + limits the time during which exchange can be measured). Exchange
in the presence of C1- was observed, and this reaction path is described by :
Rate (Msec-1) = 6.5 X 10~le-20~800/RT(E~++)(E~+++)(C1-),
at p = 2.0.
This rate law can be cast into the same form as those for the Fe+++ Fe++ exchange, but the data on the stability of EuCl++ are lacking, and
a meaningful and dependable comparison cannot be made. It seems likely,
however, that even when expressed in the form (Eu++) (EuCl++), the
corresponding activation energy will be as large as 15 kcal.
The rate of electron exchange between Cu(1) and Cu(I1) in 12 F HC1
has been measured (80) by a nuclear resonance technique. The specific
rate is reported as 5 x 107 M-l sec-1. In the context of this result, the
observation by Gordon and Wahl (45) that the bimolecular reaction between Ag(1) and Ag(I1) leading to exchange is less than 10 M-l sec-'
is all the more surprising. The environment is different, 5.9 M HC1O4 in
place of 12 M HC1, and perhaps the rate comparison reflects the difference,
which C1- in place of C104- or HzO, exerts on the rate of electron transfer for these ions. The reaction which carries the exchange in the case of the
Ag(1) - Ag(I1) reaction is 2 Ag(II)+. The specific rate at 0
' is 1020 40
F-l sec-l and E is 12.5 1.2 kcal (45). For the exchange of oxidation
state between AuC14- and Au(I1) (the latter as a chloride complex of
unknown formula), a specific rate a t 0" in excess of 107 M-l sec-l has
been estimated (110)and, for the disproportionation of Au(II), a specific
rate in excess of 10* M-l sec-l. The comparison of the latter value with
the corresponding one measured for the disproportionation of Ag (11)again
may be taken to illustrate the great sensitivity of the electron exchange
reactions of ions of this group to chloride ions. The disproportionation reaction of Ag(I1) does not take place between ions of charge +2, but extensive loss of protops takes place in forming the activated complex ( 4 5 ) .
Considerable work on electron exchange reactions of cations of oxidation state +3, +4 has been done. In only one of the systems of thia class
studied are the ions involved (reactants and products) actually of charge
+3 and +4. Keenan (66) reports the rate of electron exchange between
Pu+++ and Pu4+ to be given by:
Rate (Msec-1) = 3 X ~O~(PU+++)(PU~+)~-'*'C"/~~
+ 2.2 X 106(Pu+++)(PuOH+++)e-2*800/RT,at
p =
2.0.
(21)
38
H. TAUBE
We encounter again an extremely low activation energy, as in the Cr+ + (NH3)&oOH++ reaction, and perhaps again the formation of a binuclear
complex in the system is indicated.
The kinetic interpretation of reactions in which C e ( N ) is a reactant
is complicated because Ce (IV) is apparently considerably hydrolyzed
even in acidic solution (116) and furthermore may condense to polynuclear species (54).The pioneer work of Gryder and Dodson (47) was
checked by the later work of Duke and Parchen (35) in most of the essential features, except in this: Gryder and Dodson reported a path for
the reaction in which the rate is independent of the concentration of
Ce(1V) , but no evidence for this remarkable feature is found in the later
work. The dependence of rate on acidity is complex, and Duke and
Parchen interpret the data as indicating activated complexes of composition ( C e + + + )(Ce(OH)2++),(Ce+++)(Ce(OH)s+), (Ce+++) (CeOCe
OH6+).The reaction is subject to strong catalysis by F- (60), soh= and
H2P04- (8), but the effect of C1- is very slight (8,60).
The strong influence of acid is noted in other systems, in which extensive changes in degree of hydrolysis for one or both metal ion centers accompany reaction. Thus Furman and Garner (48) report the rate law for
the V (111)-V (IV) exchange as
V(IV) is presumably present as VO++, thus loss of H+ from the V(II1)

center must take place to convert it to V(IV). Inverse acid dependence
is reported for the reaction of Go+++ with Cef3; the conclusion that the
activated complex for this reaction also contains C104- is somewhat questionable (1231.
The redox reactions of the ions of U, Np, P u , and Am provide a fertile
field for exploring the influence of ( H + ) on the rates of reactions which
involve a change in the state of hydrolysis of the metal ion center. A singular feature of the chemistry is that, whereas species of oxidation state
+3 and f 4 can exist as ions of the same respective charge, those of + 5
and + 6 in water exist as ions of the type Tu02+ and TuOz++. With the
exception of Am4+, ions corresponding to each of the oxidation states + 3
to +6 exist for each element, and a large number of reactions between
members of this class, as well as with other reagents, are possible.
A complete review of the reactions of these interesting ions is not attempted, and the reader is directed to papers by Newton (95) and Hindmsn
(55) for a more complete discussion. A few of the reactions are referred to
in order to develop some of the features of the chemistry. Among the reac-
39
tions in which the extent of hydrolysis is increased are:
+ Fe+++ + 2H20 = NpOz+ + Fe++ + 4H+

+ 2Fe++++ 2H20 = UOZ+++ 2Fe++ + 4H+
U4+ + 2Ce(IV) + 2H20 = UOZ++ Ce+++ + 4H+.
Np4+
(23)
U4+
(24)
(25)
The rate laws are, respectively,

k(Np4+)(Fe+++)(H+)-a (81)
+
(U4+)[Ce(IVI)[k(H+)-z + k(H+)-a]
(26)
(U4+)(Fe+++)[k(HH+)-1 k(H+)-*)] (18)

(9)
(27)
(28)
In the reactions of U4+,the primary products of the bimolecular steps are

presumably formed by l e - transfer. In the three reactions the loss of proton is needed, not only to provide a proper bridging group (if indeed the
mechanism is of this type), but also as a pure ligand effect. Direct loss of
an electron from an ion such as U*+ or Np4+ may require an oxidizing
agent of exceptional electron affinity. Prior proton loss, so that the state of
hydrolysis of the reducing agent approaches that of the product, does give
an opportunity for energy matching with oxidizing agents of ordinary potential. It is not at all clear, however, why the differences in the compositions of the activated complexes appear; indeed, it would be expected that
Ce (IV) in reacting with U (IV) could get by with a weaker inverse dependence on (H+) than is the case for Fe+++, contrary to the case actually
observed.
In another group of reactions, the extent of hydrolysis of some metal
ion centers is diminished on electron transfer. It will suffice to cite the rate
laws for the four disproportionation reactions:
2 T ~ 0 2 + 4H+ = T%Oz++
+ Tu + 2Hz0.
(29)
They are
+ k(H+)]
k(UOz+)(H+) (63,67),
(NpOz+)[k(H+)
k(PuOt+)a(H+) (1091,
k ( A m ~ o ~ + ) ( H +(111).
)~
(661, ( 1 2 1 ~ ) ~
The necessity for the increase in rate with acidity can qualitatively be
understood by following the arguments given earlier. It is difficult to understand, however, why such apparently similar substances show such
variety in the rate laws for similar reactions. Particularly striking is the
extreme behavior of AmOz+.
An important result which is significant for all the reactions which have
been considered was obtained by Newton (96) in a study of the reaction
2PuO2++
+ U4++ 2Hz0 = UOz++ + 2PuOz+ + 4H+.
(30)
40
H. TAUBE
The form of the rate law suggests that U4+and Pu02++ react to form an
intermediate containing both U and Pu; all three reactions, formation of
intermediate, intermediate to reactants, and intermediate to products, are
rate determining. It is d s c u l t to formulate an intermediate which behaves
in this manner without invoking a species with definite linkages between
U and Pu, and thus the observation indicates a bridged activated complex
for this reaction as well.
The exchange reaction Np02+ - Np02++ has been studied thoroughly
by Coheo et al. (873. The interesting feature of this system is that part
of the coordination sphere remains intact on electron transfer. However,
since substitution in girdle positions can take place readily, i t is by no
means certain that the activated complex is of the outer-sphere type, and
in fact the authors incline to the view that a bridged activated complex is
involved in some paths. The reaction is first order in each of the redox
species and is independent of (H+) over a considerable range; k a t O'C,
p = 1, is 29 M-I sec-', E = 8.3 kcal M-I and ASS = -24 e.u. C1- increases the rate of reaction (88). The observations can be interpreted as
follows: Exchange for Np02+ - Np02C1+ is more rapid than for NpOz+
- Np02+ + or Np02+ - Np02C12. An important result (29) is the demonstration that the rate of reaction is independent of the ethylene glycol or
sucrose content over a range wide enough so that the dielectric constant
is reduced to 68; this result is in striking variance with the requirements
of the quantitative theory proposed by Marcus et al. (84). A reduction in
rate by 40% in changing H2Q to D2O was noted (122).As has already
heen indicated, this kind of observation a t the present level of understanding these systems is not a particularly searching diagnostic tool.
In the reactions considered thus far, ions having charges of the same
sign have been involved. For these, particularly, there can be serious question ivhether the intimate contact obtaining in a bridged activated complex is allowed. However, it has been shown that, even in such cases, a
reaction path of this kind provides a means for electron transfer. When
the reactants are oppositely charged, and at least one is substitution labile,
it seems likely that in the usual case there will be intimate contact between reductant and oxidant. It is highly unlikely that in the reaction of
Fef + + with hydroquinone ( 1 4 1 5 ) reaction takes place by electron transfer through the intact coordination shell of Fe+ + + (as suggested by Marcus, 851, when substitution on Fe+ + + can so easily occur. In the reaction
of Cr+ + with quinone, chromium-oxygen bonds are established (127) ;
by analogy, therefore, a similar mechanism can be expected in the Fe+++
reaction. Not many reactions of this kind, of simple chemistry, have been
studied. An interesting reaction recently studied (IS) is that of Fe++
with CO(C~O~)~=.
The reaction is quite rapid: k at 20.3"C, p = 0, is 11 x
MECHANISMS OF REDOX RBACTIONS
41
lo3 M - l sec-l, E = 12.6 kcal/mole-l and A S N 0. Because the reaction

can be studied at low concentration, it proves useful in testing theories of
the influence of inert ions on the rates of ionic reactions.
No claim can be made that a complete review has been made of the
reactions which qualify for discussion in this section. An attempt was
made, however, to present enough of the observations so that the different
aspects of behavior which seem significant are outlined. Much interesting
work readily suggests itself; it is much more difficult, however, to devise
experiments for these labile systems which will lead t o definite answers to
the important question: how are the various groups known to be in the
activated complex arranged with respect to each other?
VIII. Systems in Which a Net Two Electron Change Is Involved
The systems which are treated here are those in which a net 2e- change
takes place for both partners on reaction. In some instances, the reaction
may proceed by a series of two le- steps; in others, it is almost certain
that the reaction proceeds directly by a 2e- change.
A feature which distinguishes the 2e- processes as a class from the leclass is this, that with few exceptions, serious dislocations in the coordination sphere accompany reaction. Thus in the changes Cl(VI1) + Cl(V)
+ Cl(II1) + C1(I),there is a progressive reduction in coordination number as the oxidation state decreases by 2e- steps. The reason for this is
quite obviously that the incoming electrons, occupying as they do s and p
orbitals, interfere with the ligand electrons. The ions for which coordination number may be preserved on a 2e- change are those which have d
orbitals available for occupancy, for example, Mn(VI1) 4Mn(V) ,
Cr (dip) 3 + + + + Cr (dip)
But even in these cases, considerable changes
in geometry may accompany the change in oxidation state.
In certain cases, clear-cut evidence that the redox process is accompanied by atom transfer has been obtained. Thus when SOS" brings about
the changes c103-+ CIOz-, CIOz- + C10-, BrOo- + BrO-, essentially
complete transfer of an atom of oxygen for each 2e- stage of oxidation
occurs (49). The reaction can be formulated (using C103- as an example
of an oxidizing agent), as involving as an intermediate
(OnSOC10*)-.
The reaction is completed by decomposition into SO3 and ClOz- (or, if a

base such as H20 is required, into SO4- 2H+ C102-). Since C1(V)
undergoes substitution less readily than S(1V) ,the bridging oxygen is derived from c103-rather than SO3=, and net transfer is therefore observed.
This kind of mechanism apparently operates also when MnOz is the oxi-
42
H. TAUBE
dizing agent; when SO4= is formed, one atom of oxygen is derived from
the oxidizing agent (49)
The kind of mechanism in which atom transfer occurs can perhaps be
reasonably expected for these systems in which a state of changed electron
population but unchanged coordination would presumably be a t a very
high energy. A mechanism in which changes in the coordination sphere of
sach reactant accompanies electron transfer over a distance, even if favorable energetically, is excluded on probability grounds. The accident
that independent changes, involving large dislocations a t the two sites,
mould take place to match the energies while the sites are close enough
for clectron transfer to occur is likely rare indeed. By adopting a more
intimate association, the changes a t one site influence those at the other.
This arrangement is particularly happy when, as in the case of c103- SO3=, the oxidizing agent needs to lose one oxygen atom, and the reducing
agent to gain one, in completing the net change.
It seems likely that in a system such as Tl(1) - T l ( I I I ) , similar ideas
may be applicable. The difference between these reagents and Cl(V) S(1V) is one of degree rather than kind. The arrangement of water molecules about T1 is undoubtedly strongly disturbed by the pair of s electrons
that constitute the electronic difference between the two states of oxidation. However, there is no basis for a definite pronouncement about the
mechanism in this or other of the metal ion systems which will be referred
to. The treatment of these systems will feature mainly a review of the esperimental observations.
In their study of the T1(I) - Tl(II1) exchange, Prestwood and Wahl
(108) used acidic nitrate media, and Harbottle and Dodson (50) used
perchlorate media ; except for differences attributable to difference in the
media, the two sets of experiments agree. Evidence is obtained for the separate paths: (Tlf) (T1+++), (T1+) (TIOH++) , and (T1+) (TINOS++)
(108).For the path (Tl+) (TIOH++) , k a t p = 6.0 and a t 25C is reported
(108) as 2.6 x
M-' sec-'; E is 14.7 kcal mole-l and A S is -32.
C1- exerts a marked effect (50), acting a t low concentration to diminish the rate and at higher concentration to increase it. Complex formation
must be invoked to explain the decrease in rate and is supported by other
evidence. The interesting problem is to explain why the specific rate for
the term (TICI++) ( T l f ) is so much less than for (T1+) ( T l + + + ) and
why, with an increase in the degree of complex formation, the rates increase. The similar system with Br- added has been thoroughly worked
aut (26). It shows the interesting behavior that the exchange rate first decreases, then rises to a maximum a t about
144 Br-, falls to a minimum
between lo-* and 10-1' AT Br- and then rises again. These data have been
43
quantitatively accounted for by the rate law:
k~(Tl+)(Tl+++) k2(T1Brz+) ks(TIBra)
+ k,(TlBrz-)(TlBr'-).
The second and third terms presumably correspond to the establishment of

the redox equilibrium with TI+ and Br2, the first and fourth to direct exchange. The coefficients at 25", p = 0.5 M H+ are 1.2 X lov4 M-'
sec-l, 6.2 x 10-7 sec-', 2.7 x lo-' sec-l, and 7.4 x 10-1 M-l sec-l.
With CN-, a diminution in rate at low CN- is observed, and the increase sets in after approximately enough CN- has been added to form
T1(CN)4- (101). Brubaker and Michael (66) require, to explain their
observations on the effect of SO4=, complexes which in addition to TI+
and Tl+++ contain respectively, zero, one, and three sulfate ions.
It is impossible at the present level of knowledge of this subject to find
a unique explanation of the interesting ligand effects which have been
observed. Carpenter and Dodson (26')propose for the activated complex
of composition (TlBr2-) (TlBr4-) the structure
[.
]-
Br Br Br
T1 T1
Rr Br Br
and this is certainly reasonable. It is not clear, however, why a double

bridge should be required in these systems, nor why the activated complex (Tlf) (TI+++) (Br-)2 is not also an effective pathway to products.
Further, it is difficult to see why (TlCl++) (Tl+) is actually less effective
or why SO4= is more effective than Br- or C1- in
than (TI+) (Tl+++),
promoting the electron transfer.
The exchange between Sn(I1) and Sn(IV) in strong HC1 has been
studied (64). The solutions show strong interaction absorption, and it is
of interest to inquire into the relation between the species causing the absorption and the activated complexes. That they are not precisely the
same is true by definition, but it is possible that a small dislocation of
binuclear species causing the interaction absorption suffices to form the
activated complex. The observations on the Sn(I1) - Sn(IV) exchange
are summarized by the equation
Rate (M sec-l) = (Sn(I1)) (Sn(1V)) 7.5 X 10se-10s500/RT; Ai3$
= -50e.u.
(31)
Unfortunately, the formulae of the species Sn(I1) and Sn(IV) are not
known, nor is the composition of the activated complex defined with respect to C1- (or HzO) content.
Some attention (60,94) has been paid to the exchange of Sb(II1) and
Sb(V) in strong HC1. A considerable advance in understanding these reactions was made by Neumann and Brown (94). Realizing that SbC1,-
44
H. TAUBE
reaches dissociation equilibrium only slowly, they studied the exchange

between Sb(II1) and SbCls- in HCl solution. There is a diminution in rate
at high C1-, suggesting the conclusion that the optimum composition for
the activated complex is Sb+ + + * 3C1- SbC16-. A reasonable structure
for an activated complex of this formula is two condensed octahedra sharing one face.
Thus, evidence for multiple halide bridges in bringing about exchanges
comes up also in the Sb(II1) - Sb(V) case, and in fact finds support also
from work in nonaqueous media.
For the cases considered thus far, it suffices to construct a relatively
simple bridged activated complex, and no evidence for complicated chemistry involving intermediate oxidation states is obtained. Not all systems
of the present class are this simple. Thus Rona (113) finds for the exchange
U4+ - U02++ the rate law
and this and other similar observations add a new dimension to the field:
Why in some cases, but not in others, can the transfer of oxidation states
be brought about simply by transfer of one or more groups?
IX. Reactions in Nonaqueous Solvents
There are certain experiments done in hydrogen-labile solvents which

could economically have been discussed in relation to the corresponding
experiments in water. However, work in solvents as little different from
HzO as CH30H promises such new and striking results, that it seemed
preferable to emphasize the importance of the work by collecting the
relevant data in one section.
Experiments on the Fe+ + - Fe+ + + exchange in various alcohols have
been done by Horne and Dodson (59).It is remarkable that the rate of
exchange decreases to very low values at zero content of water, so low
that within the precision of the experiments no exchange is detected. As
HzO is added, the specific rate first increases linearly, then a t a diminishing rate. At comparable concentrations of water, say 1 M , the rate in isopropanol is much less (perhaps by a factor of
than it is in ethanol.
If a bridged activated complex is indeed involved, the results indicate that
an alcohol is much less effective than is water, a result that seems qualitatively reasonable on the basis of such a formulation of mechanism; for
if substantial separation of the groups in the bridge must be achieved in
the activated complex, this is more difficult for alcohol than for H2O. The
large difference between ethanol and water does not seem explicable on
45
the basis of a simple barrier penetration mechanism. But left unexplained

still is the tremendous difference in rate between isopropanol and ethanol
as solvents. Is there perhaps a change in coordination number for Fe (111)
between ethanol and isopropanol as solvents? The large bulk of isopropano1 would favor a lower coordination number. To complete the hypothesis,
it would be necessary to assume that the coordination number of Fe++
remains at 6, From this brief essay into the changes produced by the substitution of alcohols for H20, it is evident that extension of this work is
very worthwhile. An added attractive feature is that the CHsOH remains
liquid down to -98". At this temperature other ions may behave like Cr++
does in water at room temperature and definite evidence about mechanism
may be obtainable simply from product analysis.
Experiments in NHs would also appear to have special significance in
view of the fact that the bridged activated complex, at least in acid solution, is not readily accessible for an ammoniated ion. A beginning in this
field has been made by Grossman and Garner (46) who studied the Co
(11)-Co(III) exchange. The exchange rates were observed to be much the
same as in water; the authors attach significance to the circumstance that
the rates are comparable to those for the Co (NH;O8+ + + - NHs exchange
( I @ ) , and the comparison does in fact suggest that more than electron
penetration of the coordination spheres is involved.
The experiments of Ward and Weissman (135) have already been referred to. They represent the only work on redox reactions of highly polar
substances in solvents of low dielectric constant and are to be regarded
as the beginning of a field of work which can be expected to develop considerably. An interesting feature of the results is not only the magnitude
of the specific rates they were able to measure but also the dependence
of rate on the nature of the cation. Table VIII contains a summary of
their data.
TABLE VIII
THE EXCJUNQE
OF ALKALI
NAPETEALENIDES
WITH NAPETHALENJZ
Cation
Solvent
K*
Na+
Li+
Tetrahydrofurane
Tetrahydrofurane
Tetrahydrofurane
l,%Dimethoxyethane
1,Z-Dimethoxyethane
5.7
IS+
Na+
~~~
(M-1 Sec-1)
*-1107x 10'
4.6 f 3 X 108
7.6 3 x 107
*-10'
Interesting differences in the kinetics of the Sn(I1) - Sn(1V) and

Sb(II1) - Sb(IV) reactions are produced by changing from water to
media in which the species exist as discrete molecules. For the conditions
46
K. TAUBE
under which the reactions are studied in water, the system can make up
from the supply of C1-, or deliver to the surrounding medium, whatever
is needed to meet the demands of the activated complex. But for the conditions under which the following experiments were run, the systems must
make do with solvent molecules or with reactant molecules. When CClp
is solvent, little help can be expected from it for any influences in the
first coordination sphere of each reactant which might be required.
The exchange reaction of SnClz with SnCll in CzH50H (86) takes
place by the rate law
Rate ( M sec-1) = 1.4 X
1018 e--23,700/RT (SnCI2)(SnCL);
(32)
ASS is calculated as -0.4 e.u. The activation energy for the reaction is
much higher than that in concentrated HCl; presumably, some use is made
in the latter case of the large fund of additional C1- available. When
CH30H is the solvent (87) the rate is given by
5.5
1010
e-20*m/RT
(SnCls)(SnCL).
(33)
The rate law for the exchange reaction of SbCls with SbCl5 in CCll is:
Rate ( M sec-1)
= 106 e - l Q f " / R T
(SbClb)
+ 4 X 106ee-15s000/RT(SbC&)(SbCls)'.
(34)
The first term presumably corresponds to operation of the equilibrium to

form CI2 and SbC13; the second demands a remarkable activated complex
which Barker and Kahn (12) have succeeded in formulating in a plausible
way. For the analogous exchange of P C 4 with Pels (17) only the decomposition term is observed. For this path,
Rate (Msec-1) = 1.2 x 10' e-15933/RT (PCb).
(35)
These kinds of reactions, extended to include systems in which there

is a net change, and emphasizing solvents such as CCla, deserve much
more attention. There appears to be a better opportunity to define the
activated complex than is the case when water is a solvent, in part because
the specific influence of this solvent is probably much reduced, and in part
because the substitution lability of the molecules has been much reduced.
Thus, in the SbC& - SbC13 case, it may be possible to study C1 exchange
between the two forms, and a comparison of these data with those on Sb
exchange may illuminate the mechanism further.
X. Conclusion
Much of the work reported has taken place in the last few years. During this time interest has continued in questions of specific chemistry, but
research has also penetrated to the more general questions with which we
have been concerned. The resulting investigations have brought about
47
advances not only in the direction of the general goals, but also have
brought out new facts of chemistry. Further developments can be expected
as research workers turn to new systems to seek answers to the general
questions.
Some of the areas in which further work is called for have been outlined in the individual sections, and there too some of the problems pertaining to each area have been acknowledged. In this section, a few additional points will be raised which bring the various sections into closer
context.
An aspect of this subject which merits systematic and intensive exploration is the variation of rate and mechanism with electronic structure.
To be considered is the influence of the distribution of electrons between
the dc and dy levels, the comparison of d electrons with f electrons, and
the comparisons for states of differing principal quantum number. It is
significant that the rate of electron exchange for the Eu++ - Eu+ + + system is lower, but not much lower, than for systems of this charge type in
which d electrons are involved in the redox reaction. There is a large difference in the radial extension of an electron in the 4f orbital of a rare
earth ion, as compared to a d orbital for a transition metal ion so that if
an outer-sphere activated complex is involved a much lower rate of electron transfer for Eu++ - Eu+++ would be expected. The fact that the
rate for the rare earth system is only slightly less suggests that the system
finds a path in which the factor of barrier penetration is less rate determining, as may well be the case in the bridged activated complex. The
higher activation energy for Eu++ - E u + + + as compared to Fe++ Fe+++ may reflect the smaller contribution of f electrons in the binding
of the binuclear complex. Experiments in which a variety of reducing agents
presenting different features of electronic structure-for example, Ti++,
V++, Cr++, Fe++, Eu++-are used in reaction with a group of substitution-inert oxidizing agents, such as a series of Co+++ complexes,
should be particularly instructive.
The different sensitivities of different reactions to the influence of
ligands is not understood and has, in fact, been little commented on. In
Table IX are collected some data illustrating the point.
The advantage of OH- over H 2 0 as electron mediator decreases in
order from Co(II1) to Cr(II1) to Fe(II1). This is also the order of increasing acidity of the corresponding aquo ions and the correlation suggests that Fe++ +, because of its greater capacity for polarizing the ligand,
benefits less from the substitution of OH- for HzO. The decreased sensitivity of the Fe+++aq. reaction compared to that of (NH3)&oOHz+++
to the substitution of DzO for HzO fits in with this suggestion-far less in
the way of stretching the OH bond is necessary to make the electrons
48
H. TAUBE
TABLE IS
COMPARISON OF
LIGANDEFFEms FOR SYSTEM8 OF COMMON

(COMPARISONS
FOR M ( I 1 I ) L
M++)
CHARQE
TYPE
Relative k for
Fe+++- Fe++
F e + + +- Cr++
(H?O)&r+++ - Cr++
(Ir;H&Co+++- Cr++
Eu++" - Eu++
8
Calculated assumiiig
1
1
1
1
1
1160
2800
> 8 X lo'
3 x 10'
11
18
>2
10'
>50
> loo"
for EuCi++ is < 1 (U),

and this is probably
a generous upper limit.
available in the case of F e + + + , as compared to Co(II1). A remarkable

result exposed in Table IX is the enormous change in the relative efficiencies of C1- and OH- even for cases of the same charge type. This again
may be a consequence of the differing capacities of the ions to cause dissociation of the proton in the activatcd complex. A comparison which
avoids this complicating factor, as for example of the halide ions, would
be very worthwhile.
The relative efficiencies of ligands in promoting electron transfer appear
also to be a strong function of charge type. Thus C1- is almost ineffective
in promoting electron transfer between Ce+ + + - Ce(1V), although it is
quite effective in systems of the charge type +2, +3 in which the innersphere activated complex is accessible and even more effective for Cu ( I ) Cu(111.The evidence (97) for the reaction of CrCl++ with M n + + + is
that M n + + + chooses a path involving OH- as bridging group in preference to CI-, even when the acidity is large. It may be fairly general that
when ions of large positive charge are involved the hydroxy path prevails. As has already been pointed out, the analysis of these effects is difficult and complex. The efficiency of the bridging group in making the
binuclear coniplex must be considered, the rate of decomposition of the
binuclear complex to products, and for the labile systems there is also the
question of whether the catalytic group is involved in a bridging or simple
ligand position, or both.
The study of binuciear intermediates is strongly called for. Thus far
such species have not been detected in a direct measurement, but it seems
likely that with properly chosen cations, ligands, and conditions, appreciable concentrations can be built up. This subject has been a matter
of concern in investigations of "interaction absorption," but in most of the
49
systems studied the halides were used as ligands, and these ions are not
particularly effective in making stable bridged species. Groups such as
OH-, OAc-, and perhaps ions with conjugated bond systems, may be
more effective. Another type of intermediate that merits investigation is
the one formed as the immediate product of the reduction of Co(II1) or
Cr(II1) complexes. No evidence for a Cr++ species different from Cr++aq.
was found in the experiments done by Ogard and Taube (98), but the
techniques used were not particularly searching. The Co(II1) system has
the additional interesting feature that the primary product of reduction
of a Co(II1) complex may be in a different electronic state from Co++aq.
The transition involves a change in spin, and the state of high energy may
persist sutEciently long to be detected. The anomalous dependence (76)of
the rate of the exchange between Co (en)3+ + + - Co (en) 3+ + on the concentration of Co (en) 3+ + + can be interpreted qualitatively on the assumption that the process
Co(ex&++ (quartet state) + Co(en)a++(doublet state)
ia to some extent rate determining.

A recent observation which may lead to an advance in understanding
the operation of the bridged versus outer-sphere activated complex is this:
C r ( d i ~ ) ~ +(143)
+ has been shown to react very rapidly with Co(II1)
complexes, including Co (NH3)S+ + + ; V (dip)Q + + reacts much less rapidly with the same Co(II1) complexes. In these reactions we are almost
certainly concerned with outer-sphere activated complexes. It will thus be
possible to compare rates for the two types of mechanisms for a common
group of oxidizing agents which can be formed in great variety.
An area which has been little touched on is the influence of cations in
promoting redox reactions between anions. There is of course the obvious
effect of inducing reactions by generating intermediates in a redox process,
and such effects have been the subject of considerable study. But there
should also be effects which arise from including both reactants in a complex with the metal ions. In such a case the intermediates would not necessarily be formed as discrete entities. The question at issue is this: Do
cations act as electron mediators for reactions of anions in the same way
that anions Berve for cations? There is no reason why this should not be
the case and, in fact, the work of Ward and Weissman (235) on cation
influence on the rate of electron transfer between naphthalene and naphthalenide salts is a beginning in this field.
Much can be learned from studies in nonaqueous solvents. Some of the
special features to be expected from work in NH3 and CH30H have been
pointed out. In addition, NH3 offers the interesting possibility of studying
reactions of solvated electrons, and some new kinetic features can be ex-
50
H. TAUBE
pected when reducing agents are used which approach the alkali metals
in reducing potential but perhaps do not yield an appreciable equilibrium
concentration of solvated electrons.
A vigorous development in these areas, in others which have not been
touched on, and in some not anticipated by the author, can be expected.
REFERENCES
1. cf. Abegg, R., and Auerbach, F., Handbuch der Anorganischen Chemie, Vol.
4, p. 76. Hirzel, Leipzig, 1921.
2. Adamson, A. W . , J .Phys. Chem. 55,293 (1951).
3. Adamson, A. W., and Vorres, K. S., J. Inorg. 14 Nuclear Chem. 3, 206 (1956).
4. Amis, E. S., Kinetics of Chemical Change in Solution, p. 11. Macmillan, New
York, 1949.
5. Amphlett, C. B., Quart. Revs. 8,219 (1954).
6. Anderson, A., and Bonner, N. A., J. Am. Chem. SOC.76,3826 (1954).
7. Anderson, J. S., Briscoe, H. V. A., and Spoor, N. L., J. Chem. SOC.1943, 361.
8. Armstrong, W. D., and Singer, L., Anal. Chem. 26, 1047 (1954).
9. Baker, F. B., unpublished; referred to in ref. 98.
10. Baldwin, H. W., and Taube, H., to be published.
If. Ball, D. L., and King, E. L., J. Am. Chem. SOC,80, 1091 (1958).
f2. Barker, F. B., and Kahn, M., J. Chem. Phys. 78,1317 (1956).
13. Barrett, J., and Baxendale, J. H., Trans. Faraday SOC.52,210 (1956).
14. Baxendale, J. H., and Hardy, H. R., Trans. Faraday Soc. 50, 808 (1954)
16. Baxendale, J. H., Hardy, H. R., and Sutcliffe, L. H., Trans. Faraday SOC.47,
936 (1951).
16. Bearcroft, D. J., Sebera, D., Zwickel, A,, and Taube, H., to be published.
17. Becker, W. E., and Johnson, R. E., J. Chem. Phys. 79,5157 (1957).
18. Betts, R. H., Can. J . Chem. 33, 1780 (1955).
19. Betts, R. H., Gilman, H. S. A., and Leigh, K., J. Am. Chem. SOC.72, 4978 (1950).
20. Bonner, N. A , J. Am. Chem. SOC.71, 3909 (1949).
2Ou. Bonner, N. A., and Hunt, J. P., J. Am. Chem. SOC.
74, 1866 (1952).
21. Bonner, N. A., and Potratz, H. A., J . Am. Chem. SOC.73, 1845 (1951).
B. Brady, I. G. W., and Krause, J. T., J. Chem. Phys. 27, 304 (1957).
23. Bromted, J. N., and Livingstone, R., J. Am. Chem. SOC.49, 435 (1927).
24. Brown, C. I., Craig, R. P., and Davidson, N., J. Am. Chem. SOC.73, 1946 (1951).
25.Brubaker, C. H., and Michael, J. P., J. Znorg. & Nuclear Chem. 4, 55 (1957).
$6. Carpenter, L. G., and Dodson, R. W., private communication.

fl. Cohen, D., Sullivan, J. C., and Hindman, J. C., J. Am. Chem. SOC.
76, 352 (1954).
28. Cohen, D., Sullivan, J. C., and Hindman, J. C., J. Am. Chem. SOC. 77, 4964
( 1955).
29. Cohen, D., Sullivan, J. C., Amis, E. S., and Hindman, J. C., J . Am. Chem. SOC.
78,1543 (1956).
30. Connick, R. E., private communication.

31. Dodgen, H. W., and Taube, H., J. A m . C h e m SOC.
71,3330 (1949).
32. Dodson, R. W., J . Am. Chem. SOC.72,3315 (1950).
33. Dodaon, R. W.,and Davidson, N., J. Phys. Chem. 56,866 (1952).
34. Duke, F. R.,I. Am. Chern. Soe. 70,3975 (1948).
35. Duke, F. R., and Parchen, F. R., J. Am. Chem. Soc. 78, 1540 (1956).
51
36. Dwyer, F. P., and Gyarfas, E. C., Nature 166,481 (1950).

67. Eichler, E., and Wahl, A. C., J . Am. Chem. SOC.80, 4145 (1958).
38. Ellis, P., Wilkina, R. G., and Williams, M. J. G., J . Chem. SOC.1957,4456.
39. Evans, M. G.,and Nancollas, G. H., Trans. Faraday SOC.
48,363 (1953).
40. Feder, H., resulk9 described by Taube, H., J. Phys. Chem. 58,523 (1964).
41. FUOSS,
R. M., J.Am. Chem. SOC.56,1027,1031 (1934).
42. Furman, S. C., and Garner, C. S., J.Am. Chem. SOC.74,2333 (1952).
43. Gates, H. S., and King, E. L., J . Am. Chem. Soc. 80, 5017 (1958).
44. George, P., and Irvine, D. H., J . Chem. SOC.
1954,587.
46. Gordon, B. M., and Wahl, A. C., J . Am. Chem. SOC.80,273 (1958).
46. Grossman, J. J., and Garner, C. S., J. Chem. Phys. 28,268 (1958).
47. Gryder, J. W., and Dodson, R. W., J . Am. Chem. SOC.73, 2890 (1951).
48. Gurnee, E. F., and Magee, J. L., J . Chem. Phys. 26,1237 (1957).
49. Halperin, J., and Taube, H., J . Am. Chem. SOC.74,35,380 (1952).
60. Harbottle, G., and Dodson, R. W., J . Am. Chem. SOC.70, 880 (1948); 73, 2442
(1951).
61. Hartmann, and
I.,
Schlafer, H. F., 2. phyaik. Chem. (Leipzig) BlW, 116
(1951) ; Z. Naturforsch. 6a, 754 (1951).
62. Hasted, J. B., Proc. Roy. SOC.A205,421 (1951).
66. Heal, H. G., and Thomas, J. G..N., Trans. Faraday SOC.
45,ll (1919).
64. Heidt, L. J., and Smith, M. E., J. Am. Chem. SOC.
70,2476 (1948).
66. Hindman, J. C., in preparation.
66. Hindman, J. C., Sullivan, J. C., and Cohen, D., J . Am. Chem. SOC.76, 3278
( 1954); 79,4029 ( 1957).
67. Holstein, T., J . Phys. Chem. 56,832 (1952).
68. Hornbeck, J. A., J . Phys. Chem. 56,829 (1952).
69. Horne, R. A., and Dodson, R. W., private communication..
60. Hornig, H. C., and Libby, W. F., J . Phys. Chem. 56,869 (1952).
61. Hornig, H. C., Zimmerman, G. L., and Libby, W. F., J. Am. Chem. SOC.72,
3808 (1950).
62. Hudis, J., and Dodson, R. W., J . Am. Chem. SOC.78,911 (1956).
68. Hudis, J., and Wahl, A. C., J . Am. Chem. SOC.75,4153 (1953).
64. Hunt, J. P., and Taube, H., J . Chem. Phys. 19,602 (1951).
66. Ilse, F.E., and Hartmann, H., Z . physik. Chem. (Leipzig) B197, 239 (1951);
2.Naturforsch. ba, 751 (1951).
66. Keenan, T. R., J. Phys. Chem. 78,2339 (1956).
67. Kern, D. M. H., and Orleman, E. F., J . Am. Chem. SOC.71,2102 (1949).
68. Kierstead, H. A., J . C h m . Phys. 18,756 (1950).
69. King, E. L., Sister M. John Mark Woods, and Gates, H. S., J . Am. Chem. SOC.
80, 5015 (1958).
70. King, W. R., Jr., and Garner, C. S., J. Am. Chem. SOC.74, 3709 (1952).
71. Kraus, C. A., J . Chem. Educ. 12,567 (1935).
78. Latimer, W. M., Oxidation Potentials, p. 50. Prentice-Hall, New York, 1952.
76. Latimer, W. M., Pitzer, K., and Slansky, C. S., J . Chem. Phys. 7 , 208 (1939).
74. Laurence, G. S., Trans Faraday SOC.53, 1316 (1957).
76. Lewis, W. B., Coryell, C. D., and Irvine, J., J. Chem. SOC.Suppl. 2, 5386 (1949).
76. Libby, W. F., J. Phys. Chem. 56,863 (1952).
77. Libby, W. F., J. Am. Chem. SOC.
62,1930 (1940).
78. Linhard, M., Z . Elektrochem. SO, 224 (1944).
79. Linnenbom, V. J., and Wahl, A. C., J . Am. Chem. SOC.71,2589 (1949).
52
H. TAUBE
80. McConnell, H. M., and Weaver, H. E., J. Chem. phys. 25,307 (1956).
81. Magnuason, L. B., and Huieenga, J. R., J . Am. Chem. SOC.73,3202 (1951).
88. Marcus, R. A., J. Chem. Phys. 24, W (19.56).
8.9. Marcus, R. A., J. Chem. Phys. 26,872 (1997).
84. Marcus, R. J., Zwolinski, B. J., and Eyring, H., J. Phys. Chem. 58, 432 (1954).
86. Meier, D. J., and Garner, C. S., J. phys. Chem. 56, 853 (1952).
86. Meyer, E. G., and Kahn,M., J . Am. Chem. Soc. 73,4950 (1951).
87. Meyer, E. G., and Melnick, A., J. Chem. Phys. 61,367 (1957).
88. Milburn, R., and Taube, H., to be published.
89. Milburn, R., and Vosburgh, W. C., J . Am. Chem. SOC.77,1352 (1955).
90. Mitchell, J. H., and Ridler, K. E. W., Roc. Roy. Soe. A l e , 911 (1934).
91. See Moffitt, W., and Ballhausen, C. J., Ann. Rev. Phys. Chem. 7, 107 (19561,
for references to the work of L. Orgel, C. J. Ballhausen, C. Klixbull-Jorgensen,

and J. Owen. See also Holmes, 0. G., and McClure, D. S., J . Chem. Phys. 26,
1886 (1957).
92. Murmann, R. I<., Taube, H., and Posey, F. A., J. Am. Chena. SOC.79, 262
(1957).
93. Muschlitz, E. E., Jr., and Simons, J. A., J. Phys. Chem. 56, 837 (1952).
94. Neumann, H. M., and Brown,R. H., J. Am. Chem. SOC.
78,1843 (1956).
96. Newton, T . W.,in preparation.
96. Newton, T. W.,J. Phys. Chem. to be published.
97. Ogard, A. E., and Taube, H., J. Phys. Chem. 62,357 (1958).
98. Ogard, A. E., and Taube, H., J. Am. Chem. Soc. 80,1084 (1958).
99. Orgel, L. E., Rept. X" Conseil inst. intern. Chim. Solvay p. 289 (1956).
100. Owen, J., D ~ ~ C Z LFaradny
S ~ ~ OSOC.
~ S19, 127 (1955).
101. Penna-Franca, E., and Dodson, R. W., J. Am. Chem. Soc. 77, 2651 (1955).
1 W . Penney, W. G., and Schlapp, R., Phys. Rev. 41, 194 (1932).
10.9. Plane, R. A., and Taube, H., J . Phys. Chem. 56.33 (1952).
104. Platzman, R., and Franck, J., 2.Physik 138,411 (1954).
106. Posey, F. A., and Taube, H., J. Am. Chem. Soc. 75,4099 (1953).
106. Posey, F .A., and Taube, H., J. Am. Chem. SOC.
79, 265 (1957).
207. Postmus, C., and King, E. L., J. Phys. Chem. 59,1208 (1955).
108. Prestwood, R. J., and Wahl, A. C., J. Am. Chem. Soc. 70, 880 (1948); 71, 3137
(1949).
109. Rabideau, S. W., J. Am. Chem.Soc. 79,6350 (1957).
110. Rich, R. L., and Taube, H., J. Phys. Chem. 58, 6 (1954).
111.
Richardson, J., private communication.
112. Reynolds, W. L., and Lumry, R. W., J. Chem. Phys. 23,3460 (1955).
119. Rona, E., J. Am. Chem. SOC.72,4339 (1950).
114. Rutenberg, A. C., and Taube, H., J. Chem. Phys. 20, 825 (19s).
1 1 4 ~Sebera,
.
D. K., and Taube, H., to be published.
116. Sheppard, J. C., and Wahl, A. C., J . A m . Chem.Soc. 79,1020 (1957).
116. Sherrill, M. S., King, C. B., and Spooner, R. C., J. Am. Chem. SOC.65, 170
(1943).
117. Silverman, J., and Dodson, R. W., J. Phys. Chem. 56,846 (1952).
118. Simons, J. H., and Cramer, W. H., J. Chem. Phys. 26,1272 (1952).
119. Slatcr, J. C., Phys. Rev. 36,57 (1938).
1gO. Smyth, H. D., Harnwell, C. P., Hogness, T. R., and Lunn, E. G., Nature 119, 85
(1927).
53
121. Stephanou, S.E.,Asprey, L. B., and Penneman, R. A., U. S. Atomic Encrgy

Commission Rept . AECU 925 (1950).
12la. Sullivan, J . C.,Cohen, D., and Hindman, J. C., J . Am. Chem. SOC.76, 4275
(1954).
12% Sullivan, J. C., Cohen, D., and Hindman, J. C., J . Am. Chem. SOC.79, 3672
(1957).
1.93.Sutcliffe, L. H., and Weber, J. R., Trans Faraday SOC.52, 1225 (1956).
1.84. Symons, M. C.R., J . Chem. SOC.1954,3676.
126. Taube, H.,
J . Am. Chem.Soc. 77,4481 (1955).
126. Taube, H.,Chem. Rev. 5469 (1952).
1U.Taube, H.,
unpublished observations.
12%. Taube, H., and King, E. L., J . Am. Chem. SOC.76,4053 (1954).
and Myers, H., J . Am. Chem. SOC.76,2103 (1954).
129. Taube, H.,
Myers, H., and Rich, R. L., J . Am. Chem. SOC.75,4118 (1953).
130. Taube, H.,
131. Taube, H., and Posey, F. A,, J . Am. Chem. SOC.78,15 (1956).
13g. Van Alten, L., and Rice, C. N., J . Am. Chem. SOC.70,883 (1948).
133. Van Vleck, J. H., Phys. Rev. 41,208 (1932).
134. Wahl, A. C.,and Deck, C. D., J. Am. Chcm. SOC.76,4054 (1954).
136. Ward, R. L.,and Weissman, S. I.,J. Am. Chcm. SOC.79,2086 (1957).
136. Weiss, J., J . Chem. Phys. 19, 1066 (1951).
18'. Weiss, J.,Proc. Roy. SOC.A222,128 (1954).
158. Weissman, S.I., and Garner, C. S., J . Am. Chent. SOC:
78, 1072 (1956).
139. Wertz, J. E., J. Chem. Phys. 24,484 (1956).
140. Weisendanger, H. U. D., Jones, W. H., and Garner, C. S., J. Chem. Phys. 27,
668 (1957).
141. Wolf, F.,Ann. Physik 34,341 (1938).
1.42. Zener, C., Phys. Rev. 82,403 (1951).
I@. Zwickel, A.,and Taube, H., to be published.
144. Zwolinski, B. J., Marcus, R. J., and Eyting, H.. Chem. Rev. 55, 157 (1955).
COMPOUNDS OF AROMATIC RING SYSTEMS AND METALS
E. 0. Fischer
. .
and H P Fritz
lnrtitut fur anorganische Chemie. Univerritat Miinchen. Munich. Germany and

Anarganirch-chemirches Laboratorium. Technirche Hochrchule. Munich. Germany
I . Introduction .
. .
I1. Complex Formation by Five-Membered Ring Systems . . . .

A . Dicyclopentadienyl Iron . . . . . . . . . . . . .
B. Preparative Methods for Metal Cyclopentadienyls . . . .
1 . Interaction of Metals and Cyclopentadiene . . . . . .
2. Interaction of Metal Salts and Cyclopentadienyl Magnesium
Bromide . . . . . . . . . . . . . . . . .
3 . Interaction of Alkali Metal Salts of Cyclopentadiene and Metal
Halides . . . . . . . . . . . . . . . . . .
4 . Interaction of Metal Halides and Cyclopentadiene . . . .
5 . Interaction of Metal Carbonyls and Cyclopentadiene . . .
6 . Special Methods . . . . . . . . . . . . . . .
a . For Ferrocene . . . . . . . . . . . . . . .
b . For Cyclopentadienyl Thallium . . . . . . . . .
C . Chemical Properties of Pure Cyclopentadienyl Metal Compounds
1. Cyclopentadienyl Compounds Known at the Present Time .
2. First Main Group . . . . . . . . . . . . . .
3 . First Subgroup . . . . . . . . . . . . . . .
4 . Second Main Group . . . . . . . . . . . . .
5. Second Subgroup
. . . . . . . . . . . . . .
6 . Third Main Group . . . . . . . . . . . . . .
7. Third Subgroup . . . . . . . . . . . . . . . .
8. Fourth Main Group . . . . .
. . . . . . .
9. Fourth Subgroup
. . . . . . . . . . . . . .
10. Fifth Main Group . . . . . . . . . . . . . .
11. Fifth Subgroup . . . . . . . . . . . . . . .
12. Sixth Subgroup . . . . . . . . . . . . . . .
13. Seventh Subgroup . . . . . . . . . . . . . .
14. Eighth Group
. . . . . . . . . . . . . . .
15. Indenyl Compounds of Eighth-Group Elements . . . . .
D . Physicochemical and Crystallographic Properties of Pure Metal
Cyclopentadienyl Compounds
1. Magnetic Investigations . . . . . . . . . . . .
2. Polarographic Investigations
. . . . . . . . . .
3 . X-ray Investigations
. . . . . . . . . . . . .
4 . Mam Spectrographic Investigations . . . . . . . .
5 . Spectroscopic Investigations . . . . . . . . . . .
6 . Thermodynamic and Other Data
. . . . . . . . .
55
56
.
. .
. .
. .
57
57
59
59
59
. .
60
61
. .
. .
. .
61
61
61
61
. .
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
. .
. .
62
63
64
64
64
65
66
66
67
68
68
69
70
70
71
73
75
76
77
79
80
80
. .
56
. .
E 0 FISCHm AND H P FBITZ
E . Ferrocene as an Organic Compound . .

1. Substitution Reactiom . . . . . .
2 . Basic Types of Ferrocene Derivatives .
F. Mixed Cyclopentadienyl Metal Complexes
1. Cyclopentadienyl Metal Carbonyls . .
a . Fifth Subgroup . . . . . . .
b . Sixth Subgroup . . . . . . .
c . Seventh Subgroup . . . . . .
d . Eighth Group . . . . . . . .
2 . Cyclopentadienyl Metal Nitrosyls . .
a . Sixth Subgroup . . . . . . .
b . Seventh Subgroup . . . . . .
c . Eighth Group . . . . . . . .
3. Physicochemical Investigations . . .
G . Molecular Structure and Bonding . . .
1. Molecular Structure . . . . . . .
2 . Bonding
. . . . . . . . . .
H . Technical Applications of Cyclopentadienyl
82
. . . . . . . . .
83
84
86
. . . . .
. . . . .
.
.
.
.
.
.
.
.
80
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
89
. 90
. . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . . .
Metal Complexes .
.
.
IV . The Existence of Complexes of Seven-Membered Aromatic Systems .

.
90
. 91
. 91
.
91
.
92
. 92
. 93
.
97
I11. Complex Formation by Six-Membered Ring Systems . . . . . .

A . Introduction . . . . . . . . . . . . . . . . . . .
B. Preparative Methods . . . . . . . . . . . . . . . .
C . Complexes Containing Two Six-Membered Rings . . . . . . .
1 . Fifth Subgroup . . . . . . . . . . . . . . . . .
2 . Sixth Subgroup . . . . . . . . . . . . . . . . .
3 . Seventh Subgroup . . . . . . . . . . . . . . . .
4 . Eighth Group . . . . . . . . . . . . . . . . . .
D . Mixed Complexes Containing Six-Membered Rings . . . . . .
1. Carbonyls . . . . . . . . . . . . . . . . . . .
2 . Mixed Five- and Six-Membered Aromatic Ring-Metal Complexef
E . Physicochemical Investigations . . . . . . . . . . . . .
F. Structure and Bonding . . . . . . . . . . . . . . .
G. The Elucidation of the Constitution of the Chromium Phenyl Compounds . . .
. . . . . . . . . . . . . . . . .
References
87
87
89
97
97
98
99
99
99
101
101
101
101
102
103
106
106
107
109
1. Introduction
Until 1951. the year in which dicyclopentadienyl iron. Fe (CaHa)2. was

discovered. the chemistry of complexes formed by metals and hydrocarbons constituted only a small part of either organic chemistry or the
chemistry of inorganic complex compounds . Most of the work on organometallic compounds was devoted to alkyl and aryl compounds of metals
of the main groups. such as cacodyl (the first organometallic compound
known). the Grignsrd reagents. Gilmans metallic derivatives of organic
compounds. and the metal carbonyls .
57
Unsaturated hydrocarbons were, however, known to be able to form

complexes with the halides of some of the platinum metals and with
those of copper, silver, and mercury. In such compounds the bonding
of the organic residue was sometimes taken to be purely coordinative,
as in (PtC12 C2H4)2, and sometimes a-bonding between the metal and
carbon, as in CUCeHs or Pt(CH3)4 was assumed.
Only a few examples of complexes involving aromatic nuclei were
then known. Menschutkins complexes of benzene and its homologues
with antimony halides had been discovered and thoroughly investigated ;
and in these, in the complexes such as AlzBr6 * Ca&(CHa)8 isolated by
Norris and Ingraham, and in compounds such as CsH6 AgCIOa, the
essential participation of r-electrons of the aromatic ring in bond
formation is now recognized. Coordination compounds in the Sidgwick
sense, i.e., with the making up of the electronic structure of the metal
atom to an inert gas configuration, were, however, unknown, and the
isolation of ferrocene provided the first instance of a substance in which
the presence of this type of bonding can be regarded as well established.
Since the field of metal complexes containing aromatic rings has in the
meantime expanded 80 extraordinarily, it seems best to discuss the
known experimental facts and results in terms of the size of the aromatic
ligands. Summaries of five-membered ring metal compounds have been
published in 1955 (42, 1 5 0 ~ ) .
II. Complex Formation by Five-Membered Ring Systems
Metal derivatives, all largely ionic in character, of the five-membered ring compounds cyclopentadiene, indene, and fluorene have long
been known, but there was, until now, no corresponding well-defined
organometallic compound of iron. It was therefore a really remarkable
event when, in 1951, Miller et al. (130)and, independently, Kealy and
Pauson (1141, reported the existence of an iron compound Fe(CaH6)2
containing two cyclopentadienyl groups attached t o one metal atom.
A. DICYCLOPENTADIENYL
IRON
The first group of workers observed the formation of this compound
by the action of cyclopentadiene on reduced iron a t 300. The latter
group described an entirely different approach: it was expected that a
coupling reaction of a solution of cyclopentadienyl magnesium bromide
containing ferric chloride would yield the (unknown) hydrocarbon fulvalene, CloHs; instead, the reaction
FeCL + 2CsHsMgBr Fe(CsH& 2MgBrCl

(1)
took place, with the formation of dicyclopentadienyl iron. Because of
58
E. 0. FISCHDR AND H. P. FRITZ
the similarity of its reactions to those of benzene, it was christened by

R. B. Woodwsrd ferrocene, by which name it is now usually known.
Ferrocene, Fe(Cd&)2, forms orange needles, mp 173-174;it is insoluble in water, but is readily soluble in all organic solvents. It is stable
in air, sublimes readily above loo, and is thermally stable up to 470.
By suitable oxidizing agents, such as ferric or silver salts, it is converted
into the blue dichroic cation [Fe(C6H6)2]+; this process is easily reversible. While ferrocene itself is diamagnetic-its susceptibility ~~~~~l~ =
-125 x
cm3/mole-the cation has a moment corresponding to the
presence of one unpaired electron (219).Polarographic studies show the
standard potential for the change to be -0.56 v (1@,219). Kaplan et al.
(112) investigated the thermodynamic properties of ferrocene and found
the heat of sublimation to be 16.81 kcal/mole, the heat of vaporization
of the liquid 11.3 kcal/mole, and the heat of fusion 5.5 kcal/mole; the
triple point was found to be 183 and the boiling point under normal
pressure 249.The heat of formation from the elements (only hydrogen
being in the gaseous state) was found by Wilkinson, from the heat of
combustion of ferrocene, to be 33.8 kcal/mole (81). Studies of the vibrational spectrum led, after taking into account the heat of sublimation
(112),to a value of 40.7 kcaI/mole (125). Infrared and Raman spectra
were measured by several workers, and values corresponding to aromatic C-H and C-C bonds were found (21, 112, 124) ; the dichroism of
fcrrocene crystals was examined by Yamada et at. (225). All the results
of the physicochemical researches on ferrocene may be correlated. X-ray
analysis, like the properties mentioned above, soon showed that the
original formulation by the discoverers (114) as
CH=CH
CH==CH
i /
CH=CH
CH==CH
CH-Fe-CH
CH==CH
CH=CH
CH=CH
CH- Fe++CH
CH=C
with partially ionic bonding and localized a-bonding of the iron to the
methylenic carbon atom, was untenable. A new structure was therefore
suggested independently by Wilkinson, Woodward, and their co-workers
(9191, and by Fischer and Pfab (70). The former group suggested a
pentagonal prism or antiprism, a so-called sandwich structure; the
latter, from X-ray measurements of unit cell dimensions, established the
presence of a center of symmetry and hence of the antiprismatic Doppelkegelstruktur (double-cone structure) . Fourier analyses completely
confirm this conception (SO, 31, 35) ; Fig. 1 shows a Stuart model of ferrocene.
RQ.1. Model of the ferrocene
59
atructure.
B. PREPARATIVE
METHODSFOR METALCYCLOPENTADIENYLS
The field of metal complexes of cyclopentadiene has subsequently
been extended mainly by the two groups of workers who elucidated the
structure of ferrocene. New methods for the preparation of such compounds were essential to further progress.
1. Interaction of Metals and Cyclopentadiene
By this method, which had formerly been employed only for the
alkali metals and iron, the magnesium derivative, Mg (C5H5)2, has recently been prepared in the vapor phase. Cyclopentadiene vapor was
passed over magnesium powder a t 500-600" and under suitable conditions
crystals of mp 176" were obtained ( 4 ) . Thallium and indium reacted
similarly with cyclopentadiene vapor a t 350, giving the compounds
T1C5H5 and InC5H5 (55).Attempts to develop a more general method
of preparation from cyclopentadiene and molten metals met with no
success (191).
6. Interaction of Metal Salts and Cyclopentadienyl
Magnesium Bromide
The reaction of the Grignard reagent CsH5MgBr with metal salts
moved to be the most widely applicable method. Depending on the re-
60
E. 0. FISCHER AND H. P. FRITZ
sulting oxidation state of the metal, either the uncharged cyclopentadienyl compound or a cationic complex is obtained (56, 207, 217), the
reaction taking the course
+ MeX,
(CsfI,)2MeX,-z
2CsHbMgBr
+ 2MgBrX.
(2)
For the preparation of uncharged complexes which are sensitive to oxidation, excess of the Grignard reagent (which acts simultaneously as a
reducing agent) is employed, and the experiment is carried out in an
atmosphere of nitrogen. The compound Mg(C5Ha)2,which is formed
as a by-product, is converted into the magnesium salt of cyclopentadiene
carboxylic acid by the passage of carbon dioxide through the ethereal
solution, and is thus precipitated and separated (51). As the source of
the metal in this reaction Wilkinson used the acetylacetonates as well
as the halides, and we have employed complex ammine thiocyanates
with advantage. Yields are 50-70%.
3. Interaction of Alkali Metal Salts
of Cyclopentadiene and Metal Halides
,4nother method for the preparation of the complexes was early developed by us. As is well known, Thiele, at the turn of the century,
obtained the white saline compound KC5H5 (the first metal derivative
of a five-membered ring system) by the action of metallic potassium
on freshly prepared cyclopentadiene in benzene solution (191). It was
found that a suitable method for the preparation of the alkali metal
salts consisted of the action of the metals on cyclopentadiene in liquid
ammonia, according to the equation
2CsHs
+ 2Me liq NHJ 2MeCJ16 + H, (Me = Cs, Rb,K, Na, and Li).
(3)
The hydrogen which is produced hydrogenates some of the cyclopentadiene (108, 232) :
C;Hs
+ Hx
-j
C6He.
(4)
It was shown that the alkali metal compounds increase steadily in solubility along the series given above, and that they are dissociated to an
extent sufficient to enable precipitation reactions involving the C5H6ion to be possible. On adding anhydrous metal salts which are soluble
in liquid ammonia, such as nitrates and thiocyanates, the following type
of reaction takes place:
Co(NHs)r(SCN)n$- 2fiCbHs-
liq NHa
[Co(NH,)e.](CsHs)z
+ 2LiSCN.
(5)
The saline ammine complexes lose ammonia when heated in vacuo, and
the uncharged dicyclopentadienyl compounds of Fe, Co, Ni, Cr, and Mn
61
are thereby obtained (50,58, 59). The method generally gives yields
of about 70% except for chromium and manganese. It has further been
shown that other strongly polar solvents (provided they do not contain
reactive hydrogen) can be used instead of liquid ammonia: Wilkinson
(211, 214) has introduced tetrahydrofuran as probably the most elegant
and versatile solvent, and we have shown that other liquids (dimethylformamide and tertiary butyl alcohol) may occasionally also be used
with advantage (48,49).
4 . Znteraction of Metal Halides and Cyclopentadiene

This possible route to the corresponding organometallic compound
arises from the use of bases such as triethylamine, diethylamine, or
pyridine (10), roughly according to the equation:
2C&
+ FeClr + 2 Base
--f
Fe(C&I&
+ 2 Baae - HCl.
(6)
This method, which has been described as giving only a low yield, has
been developed industrially to an extraordinary extent, in the case of
2.
Fe (CBHB)
5 . Znteraction of Metal Carbonyls and Cyclopentadiene
A method which has been used mainly for the preparation of

cyclopentadienyl metal carbonyls, but which is occasionally applicable,
is that represented roughly by the equation
MeWO).
+ 2CsHs
Me(CJl&
+ zC0 + Hs.
(7)
The yields are 10-300/0;for the nickel compound substantially lower

(159, 208).
6. Special Methods
a. FOTFerrocene. For ferrocene itself another synthesis, also involving

an intermediate organometallic compound, has recently been described
(109).By the action of mercuric chloride in methanol on cyclopentadiene
a mixture of the compounds C5H6HgCland Hg(C5Hs) 2 is obtained, which
with finely divided iron gives ferrocene in yields of up to 30%.
b. For Cyclopentadienyl Thallium. An unusual method for the preparation of this compound has been described recently (128). This is
similar to one reported earlier by Thiele (192)for the preparation of the
copper, silver, and mercury compounds; Thiele, however, did not fully
establish his claims. The thallium compound, in contrast to those of the
alkali metals, is obtained almost quantitatively when an aqueous solution
of the hydroxide is shaken with cyclopentadiene:
TlOH
+ C&+
TIC& 4- HtO.
62
E. 0. FISCHEX AND H. P. FRITZ
The reaction proceeds so nearly to completion that it has been suggested

as the basis of a method for the separation of thallium or of the hydrocarbon.
C . CHEMICAL
PROPERTIES
OF THE PUFLE
CYCLOPENTADIENYL
METAL
COMPOUNDS
The chemical properties of these new compounds will now be summarized, with special reference to their role in inorganic chemistry. It was
seen soon after the discovery of ferrocene, that these new compounds not
only present a host of new problems for the inorganic chemist (e.g., in the
variation of the metal and in the preparation of complexes containing
other ligands) , but because of the aromatic character of the cyclopentadienyl rings they are also of fundamental significance in organic
chemistry. The hydrogen atoms of the rings in ferrocene can be substituted and, as will be described later, undergo a large number of the
reactions characteristic of aromatic systems. I n determining the chemical
properties of the cyclopentadienyl compounds, the oxidation state of the
transition metal a t the center of the compound is the controlling factor.
If it is one which is usual in other complexes of the metal, the cyclopentadienyl compound is for the most part quite stable. If, however, it is
abnormally low, sensitivity towards oxidizing agents results, and in all
work with the compound air must then be excluded. The extreme
sensitivity to oxidation of Co (CaHs)2, in contrast to Fe (CaH,) 2, is typical
of the behavior of these substances as complexes, and will be discussed
later. The [Co(CaH5]
-I-ion, on the other hand, is extraordinarily resistant
to chemical attack. The uncharged dicyclopentadienyl complexes of the
ferrocene type are readily volatile, crystallize very well, and dissolve
easily in the common organic solvents. As a rule, oxidation and reduction
may be brought about without difliculty. For those dicyclopentadienyl
cations which are stable to hydrolysis, a large number of precipitation
reactions are available, the most important being those with the reineckate, [Cr (SCN) (NH3)23 - (70) ; tetraphenylborate, [ (CsHb)4B] - (67) ;
and picrate, (CsHzN30,)- (819)ions. Perchlorate, hexafluorophosphates,
fluoroborates, chloroplatinates and cobaltinitrites are also often only
sparingly or moderately soluble. Polyhalides (e.g., tribromides and
triiodides) are also slightly soluble, but common anions such as sulfate,
carbonate, nitrate, cyanide, fluoride, chloride, bromide, iodide, and phosphate yield no precipitate.
Alkali and alkaline-earth metal cyclopentadienyls are, in contrast to
the transition metal complexes, typically saltlike in character; so, too, are
63
the tricyclopentadienyl compounds of the rare-earth metals. Their ionic

structure is shown by their dissociation in nonacidic strongly polar
solvents such as liquid ammonia, tetrahydrofuran, N,N-dimethylfonnamide, and glycol dimethyl ether. The compounds of various main group
elements such as zinc, mercury, tin, lead, and bismuth occupy an
intermediate position; they show varying degrees of stability towards
hydrolysis, and their molecular structures are not yet fully understood.
1. Cyclopentadienyl Compounds K m n at the Present Time
In Fig. 2, in the framework of the periodic table, are shown the

elements (both transition elements and typical elements) for which
Metal
'
l 1
IH
COI
Cyclopentadienyl Compounds
I P I
PIII
Hn
lo'
I*
FIQ.2. Metal cyclopentadienyl compounds. Key: Cp, cyclopentadiene; dotted

squares, saltlike type ; shaded squares, penetration complexes; squares with hgonals,
transitional type; d, diamagnetic; p', pa, pa, paramagnetic with one, two, and three
unpaired electrons, respectively; col, colorless; yel, yellow; viol, violet; cryst,
crystalline.
64
cyclopentadienyl compounds have so far been isolated. The division into

saltlike compounds, true organometallic penetration complexes, and intermediate substances is also shown, though it is not considered that an
unequivocal classification is at present always possible. Oxidation states
of the metals (that of an aromatic cyclopentadienyl residue being taken
as -1) and net charges on complexes are also shown.
2. First Main Group
The white saline potassium compound KC5Ha was obtained by Thiele

as early as 1901 by the action of the metal on a solution of cyclopentadiene
in benzene (191) ; compounds of all the alkali metals except francium
have now been made by the interaction of the metal and the hydrocarbon
in liquid ammonia (38).They are without exception colorless salts which
decompose immediately, with consequent discoloration, on exposure to
the atmosphere. They are dissociated to a large extent in polar solvents,
and the Na-C bond has been estimated as 50% ionic (205). They are
insoluble in nonpolar solvents, and in solvents which contain active
hydrogen (e.g., water) they a t once undergo hydrolysis according to the
equation
MeC'Hr
+ H a -+ MeOH + C&.
(9)
3. First Subgroup
In this group, too, the first observations were made by Thiele (192),
who found that cuprous and silver ions are precipitated from aqueous
ammoniacal solution by cyclopentadiene, but gave no analytical data.
Thallium was later found to be precipitated similarly. Copper(I), however, also yields a white crystalline cyclopentadienyl-triethylphosphine
romples, C5H&uP(C2H5)3, which is stable to water (215) ; this was, in
fact, analogous to an earlier method described by van Peski and Melsen.
Attempts to make a complex by the Grignard reagent method or by
the action of cyclopentadienyl sodium on cuprous halides were unsuccessful.
4. Second Main Group

The dicyclopentadienyl derivatives of these elements which are known
at the present time (those of magnesium and calcium) are in many respects similar to those of the alkali metals, being colorless, saltlike compounds. The magnesium compound was successfully prepared by the
interaction of sodium cyclopentadienyl and magnesium bromide in tetrahydrofuran (914), by thermal decomposition of cyclopentadienyl magnesium bromide (214) and, recently, by the action of cyclopentadiene
65
vapor on magnesium powder a t 500-600' (4). It forms colorless needles,

mp 176-178", and reacts vigorously with water, with the formation of
cyclopentadiene and magnesium hydroxide. With carbon dioxide and
carbon disulfide, formation of the magnesium salts of cyclopentadienecarboxylic and -thiocarboxylic acids results. It is soluble in ether and is
immediately decomposed on exposure to the atmosphere.
The calcium compound is obtained by the reaction
CaCz
+ 2CSH6 + Ca(CsHs)r + GHz
(10)
but it has not yet been prepared in a pure state (2.91).

5. Second Subgroup
The colorless zinc co~pound,Zn (C&) 2, which sublimes at 160" under partial decomposition, is obtained in small yield from zinc chloride
and cyclopentadienyl sodium in diethyl ether; however, the less stable
cadmium compound decomposes, with separation of cadmium, under
these conditions ( 5 5 ) . The mercury compound, Hg(CaHs)2, is produced
in 20% yield by the action of the sodium derivative on mercuric chloride
in tetrahydrofuran (8151. The action of cyclopentadiene on the complex
K2(HgIa) in aqueous alkaline solution results in the precipitation of a
mixture of CsH6HgI and Hg (C5H6)*, from which the latter compound
may be obtained in good yield by extraction with a mixture of tetrahydrofuran and petroleum ether (62).It foriiis pale yellow crystals which
begin to decompose at about 60' and which melt a t 83-85'. The compound is readily soluble in most solvents; it decomposes slowly even when
kept in the dark at room temperature; it is insoluble in water and reacts
with neither water nor bases. On the other hand, decomposition occurs in
dilute hydrochloric acid, It converts ferric chloride to ferrocene quantitatively, and it yields an adduct with maleic anhydride (216).
By the action of monomeric cyclopentadiene on mercuric chloride in
methanol, in the presence of sodium acetate, a bright yellow crystalline
deposit, consisting of a mixture of CeHaHgCl and Hg(C&)2, is obtained
(109, 192). Neither treatment with solvents (e.g., tetrahydrofuran or
dioxan) nor vacuum sublimation effects a separation of this mixture,
since with slight increase in temperature the crude product decomposes
(109).
Cyclopentadiene reacts with mercuric cyanide in aqueous solution,
giving the insoluble cream compound C6HaH&N. Like the corresponding
iodide (which is also cream) this compound is fairly stable in air, but does
not melt sharply (62). By the reaction of indenyl lithium and mercuric
chloride in dry ether, indenyl mercuric bromide (the bromine being
66
E. 0. FISCHER AND H. P. FBITZ
derived from the indenyl bromide used at an earlier stage in the

synthesis) is obtained as a yellow substance of mp 118". Attempts to
make diindenyl- and difluorenyl-mercury have not been successful (27).
6. Third Main Group
Sodium cyclopentadienyl reacts with indium trichloride in ether,

tetrahydrofuran, or dioxan, forming the volatile compound InC5H5,
which may be obtained as pale yellow needles by sublimation ( 5 6 ) .This
substance does not melt sharply, but slowly decomposes and darkens at
110". It is very sensitive to oxidation and is slightly soluble in benzene,
in which decomposition takes place; it is, however, rather less soluble
in ether and petroleum ether. Water has no action on the compound,
but on addition of a little sulfuric acid it is vigorously decomposed.
The crystals are slightly sensitive t o light. In the preparation of InC5H5
an intermediate product, In (C5H5)31 can be isolated in very small yield
as golden crystals. On sublimation a t above 160" it decomposes to form
the monocyclopentadienyl compound.
Thallous hydroxide or thallous salts in the presence of alkali react
with cyclopentadiene to yield T1C5H5 (12.9). This substance, which
is stable in air, crystallizes a t 0" in colorless needles (in pale yellow
needles at 60"), and is practically insoluble in water; it is, however,
moderately soluble in methanol, acetone, pyridine, etc. It does not melt
when heated, but it sublimes a t 100" in a high vacuum (81).
7. Third Subgroup
All known cyclopentadienyl compounds of this group are prepared

by the action of the sodium derivative on the metal chlorides in tetrahydrofuran (9, 167, 212). With the exception of (C5H5)3UC1and
Th(C5H5)41 all are ionic compounds of general formula Me(C5H5)3. The
results of Wilkinson and his co-workers are summarized in Table I. All
of these substances have similar properties. All are unstable in air,
especially the cerium compound, which is noteworthy for being blue-green
in the vapor state. Water effects immediate decomposition; the uranium
compound is converted into the unstable green [U(C5H5)3]+ ion, which
may be precipitated as its reineckate; the thorium derivative yields the
still more unstable [Th(C5H6)2]++or [Th(C5H5)3]+ ion; the others are
decomposed immediately to cyclopentadiene and the metal hydroxide,
All the neutral cyclopentadienyl complexes are almost insoluble in petroleum ether, cyclohexane, and benzene, but are moderately soluble in pyridine, tetrahydrofuran, and dioxan. They are rapidly decomposed by carbon
disulfide, carbon tetrachloride, or chloroform. The compounds U (C5H5)a
and U(CJH5)2C1have also been made and analyzed, but because of the
67
TABLE I
CYCLOPENTADIENYL
COMPOUNDS
OF SUB-GROUP
THREE
ELEMENTS
Compound
Me(C5H&
Me =
Melting point,
closed tube
sc
240
295
395t
435 t
415t
380
365
350
302
285
273
260
Y
La
Ce
Pr *
Nd
Sm*
Gd
DY
Er
Yb
(C.Jh),UCl
(CsH&Th(?)
("(3
Color
Sublimation
temperature,
10-4 mm Hg
("C)
yellow
pale yellow
colorless
orange-yellow
blue-green
reddish-blue
orange
yellow
yellow
pink
dark green
dark red
colorless
200
220
260
230
220
220
220
220
220
200
150
245
250
Yield
(%)
75
85
25
72
83
78
75
84
85
88
82
82
1
* Ammoniates of these compounds have been observed.

t Decomposes.
smallness of the yields in which they were obtained they have not yet
been investigated further (16'7).
8. Fourth Main Group
Silicon compounds of formula (CaHa)Si (CHs) and (CsHs)2% (CH,) 2

are obtained by the action of cyclopentadienyl magnesium bromide on
trimethylchlorosilane and dimethyldichlorosilane; they are colorless
liquids which boil at 43-44' (19 mm pressure) and 73' (25 mm), respectively (89). The monocyclopentadienyl compound darkens on exposure to air, and it forms an adduct of mp 105" with maleic anhydride,
thus showing the retention of the conjugated system in the compound.
Stannous chloride reacts with cyclopentadienyl lithium in dimethylformamide to give the compound S ~ ( C E H (48).
~ ) ~ This substance is
unstable in air, forms colorless crystals (mp 105"), and is moderately
soluble in benzene, ether, and petroleum ether. Cold water does not effect
hydrolysis, but acids bring about decomposition with liberation of
cyclopentadiene. The interaction of aryl tin (IV) chlorides and cyclopentadienyl magnesium bromide leads to a series of dicyclopentadienyltin(1V) aryl compounds (93) which are unstable in air or in light and
are rapidly hydrolyzed by water.
68
The lead compound Pb (CaH5)2, which may be obtained by the action

of the sodium derivative on lead nitrate in dimethylformamide, has
similar properties. It occurs as well-formed, strongly refracting yellow
crystals which do not melt sharply, but which after sintering at 139" or
higher temperatures become deep orange in color. Since this color change
is reversed on cooling, the compound is thermochroic. The compound
dissolves in benzene, acetone, ether, and petroleum ether, but not in water,
and does not undergo hydrolytic fission, with liberation of cyclopentadiene, in the cold (49).
The interaction of indenyl- and fluorenyl-lithium and phenyl-substituted plumbic chlorides produces the corresponding phenyl lead
derivatives (B'), but attempts to prepare difluorenyl lead and diindenyl
lead were not successful.
9. Fourth Subgroup
Dark red crystals of the complex (C5H6)2TiBr2 (mp 314') are

obtained by the action of cyclopentadienyl magnesium bromide on
titanium tetrachloride in toluene; the compound is moderately soluble in
aromatic solvents, but water slowly decomposes it, producing a yellow
solution which gives the usual precipitation reactions for dicyclopentadienyl metal cations. The almost colorless complex (C5H5)2ZrBr2 (mp
260" with decomposition) is obtained by an analogous method (211,217).
By reduction with zinc a t 275", (CaHa)2TiC12is converted into the
green crystalline compound (C5H5)zTiCl (209). In addition, the compound Ti (C5H5)a is obtained in 40% yield by the interaction of titanium
dichloride and sodium cyclopentadienyl in tetrahydrofuran (37). Dark
green crystals are formed, which sublime at 120-180", and start decomposing at 130" without melting; oxidation is rapid in air. In benzene, ether,
and petroleum ether it forms green solutions; carbon disulfide and carbon
tetrachloride decompose it; oxygen-free water, in which it is insoluble,
very slowly oxidizes it to the compound (C5H5)2Ti (OH) 2. Two modifications of an etherate exist. By the action of (CaH5)2TiC12on lithium
aryls, a series of crystalline orange compounds of the type (c&)2
T i ( a r ~ 1are
) ~ obtained (188, 189, 190). Grignard reagents react similarly
(157), from which it is deduced that the aryl groups are a-bonded to the
metal.
10. Fifth Main Group
Mention must first be made of some interesting compounds in which

the C6H5- ion is bonded to nitrogen; these are pyridine cyclopentadienylide (125) and diazocyclopentadiene (98).Triphenylphosphonium
69
cyclopentadienylide (161) is a pale yellow crystalline compound (mp

230") which is stable to boiling alkali and which (in contrast to triphenylphosphine-metliylenes) does not react with ketones.
Tricyclopentadienyl bismuth, Bi (CsH6)3, which is obtained from
cyclopentadienyl sodium and bismuth trichloride in solvents which do not
contain reactive hydrogen, exists as an orange modification which a t
18" is converted irreversibly into a black form. It is hydrolyzed
by water (73).The analogous carmine compound Sb(CsHa)3, which is
also hydrolyzed by water, is prepared by a similar method a t low
temperature (7'3).
11. Fifth Subgroup
Dicyclopentadienyl metal halides of vanadium, niobium, and tantalum

may be made by the action of cyclopentadienyl magnesium bromide on
the appropriate metal salt in benzene-ether solution, or by interaction of
the sodium derivative and the salt in tetrahydrofuran or dimethylcellosolve (611,617).Starting from vanadium tetrachloride, the compound
of quadrivalent vanadium is obtained; the pentabromides of niobium and
tantalum yield derivatives of the elements in the quinquevalent state. The
vanadium compound, (C6Hd 2VC12, forms light green crystals which
decompose without melting at 250"; it is soluble in alcohol and in
chloroform, slightly soluble in ether, benzene, carbon disulfide, and
carbon tetrachloride, and insoluble in petroleum ether. It dissolves in
water to give an unstable green solution containing the [V(C3H5)z]++
ion, which in the Jones reductor can be reduced to the purple
[V(C,H,),]+ ion (211).
The uncharged compound V (CsH5) was first prepared from vanadium
tetrachloride and an excess of cyclopentadienyl magnesium bromide in
the absence of air (51); it was later obtained by reduction of the dichloride by lithium aluminum hydride (609).It forms readily volatile violetblack crystals of mp 167-168". It is very soluble in benzene, ether, and
petroleum ether, and is stable in the absence of air.
zNbBrs, forms fine,
Dicyclopentadienyl niobium tribromide, (C5H5)
dark reddish-brown needles which decompose without melting a t about
260". The compound is readily soluble in polar solvents, but does not
dissolve in petroleum ether. On exposure to the atmosphere it is slowly
hydrolyzed, with the formation of the orange-red compound (CaH5)
Nb (OH)Brz (211). The tantalum compound, (C3Ha)zTaBr3, forms rosecolored crystaIs (mp 280' with decomposition) which are extremely sensitive to moisture. It is soluble in polar solvents, the solutions resembling
those of the niobium compound.
70
12. Sixth Subgroup
Dicyclopentadienyl chromium was first made in very small quantity

by thermal decomposition of the ammine [Cr(NH3)B](C5H5)3 (50) ; it
was later also obtained by vapor phase interaction of chromium carbonyl
and cyclopentadiene (208) or by the action of the sodium compound on
chromous or chromic chloride in tetrahydrofuran ( 2 2 ) . The compound
sublimes a t 60-80' in a high vacuum, forming ruby-red needles (mp
173'). In the absence of oxygen it is stable up to 300'; it is not decomposed by light, and it is not attacked by oxygen-free water. Treatment with acids yields transient blue solutions which may contain cationic
binuclear monocyclopentadienyl-hydroxychromium compounds. Dicyclopentadienyl chromium reacts with iodine t o form the green iodide
(C6HJ2CrI, which is soluble in, e.g., tetrahydrofuran or dioxan to form
green solutions (6.9).By means of the third method given for the preparation of Cr (CsH5)2, cationic dicyclopentadienyl derivatives of molybdenum and tungsten may be prepared ( 2 2 ) .
When the Grignard reagent reacts with molybdenum pentachloride,
a red solution containing the [(C6H5)2MoCl]+ ion, which may be
precipitated as its reineckate or silicotungstate, is produced; this is converted by oxidation into the violet-red dichroic [ (C5H5)2MoC12]+ ion,
which yields an insoluble chloroplatinate. From molybdenum pentachloride and sodium cyclopentadienyl in tetrahydrofuran the compound
(C6H5)2M~C12
is obtained in ,the form of green needles ( 2 2 ) . Tungsten
hexachloride reacts similarly, and the compound [ (C5H5)2WC1212 (PtCl6)
may be isolated.
IS. Seventh Subgroup
Dicyclopentadienyl manganese was first obtained by thermal decomposition of the ammine [Mn(NH,),] (C5H5)2 (50, 1 1 1 ) . Shortly
afterward, it was made in good yield by the reaction between manganese
bromide and cyclopentadienyl sodium in tetrahydrofuran ( 2 1 4 ) , and it
may also be prepared by the interaction of manganese chloride and an
ethereal solution of the Grignard reagent, followed by sublimation (60).
At ordinary temperatures it forms brown crystals.
The compound is exceedingly reactive, and is vigorously decomposed
by deaerated water with the separation of cyclopentadiene. It is only
sparingly soluble in benzene or ether, forming brownish-yellow solutions.
When it is heated in a sealed tube, it becomes almost colorless a t 158-159',
and then melts sharply, without decomposition, a t 172-173' ( 2 1 4 ) . On
slow cooling the color change is reversed a t 159", but by chilling the
white form the transformation may be retarded. By an analogous reaction
71
of rhenium pentachloride and cyclopentadienyl sodium in tetrahydrofuran, a yellow crystalline substance which sublimes at 120-170,melts at
162O, and has the formula Re(C5Ha)zH,is obtained (213). The compound is unaffected by water, but reacts moderately rapidly with air. It
is readily soluble in benzene, ether, and petroleum ether. The hydrogen
atom is not acidic, and the compound does not react even with 6N sodium
hydroxide. Surprisingly, the compound acts as a base, and will dissolve in,
for example, hydrochloric acid, to form the complex ion [ (CsHd zReH2]+ ;
addition of alkali reverses this process.
14. Eighth
Group
The most important compound in this group is, of course, ferrocene;

since, however, it has been described in detail in Section IIA as the
parent substance of the whole group of five-membered ring complexes,
only a further brief reference to its properties will be made here. Although
the early attempts to hydrogenate it under ordinary conditions were unsuccessfuI, this operation was accomplished under extreme conditions
(144); under 280 atm of hydrogen at 300-340, using Raney nickel in
cyclohexane, pentane and cyclopentane are produced, but even after four
hours over half of the ferrocene remains unchanged. A more elegant
method, and one of greater significance in the preparative organic chemistry of five-membered ring compounds, is that of reductive cleavage by
lithium in ethylamine (194). This operation may also be effected by
alkali metals in liquid ammonia.
Because of the extreme stability of the cobalticinium ion,
[ Co (CaHs)2] +, other cyclopentadienyl compounds of cobalt were not
at first obtainable. The bromide, [ Co (C5H5)2]Br, was first made by the
action of cyclopentadienyl magnesium bromide on cobalt (111) acetylacetonate in benaene (20) ; it may also be obtained by oxidative hydrolysis
of the product obtained from the Grignard reagent and cobalt bromide in
ether (67). Both the bromide and the corresponding fluoride are very
+ ion is very stable to both acids and
hygroscopic. The [CO(CJHS)Z]
alkalis; it is unattacked even by aqua regia, alkaline hydrogen peroxide,
or ozone in acetic acid, and only fusion with potassium pyrosulfate or
treatment with fuming perchloric acid effects decomposition of the
complex.
The neutral complex C O ( C ~ Hhas
~ ) been
~
obtained by the action of
alkali metal cyclopentadienyls on the complex [Co (NH3)4] (SCN) in
liquid ammonia (69) and, more recently, by lithium aluminum hydride
reduction of the [Co(CaH5)z]+ ion ($09). It forms deep violet-black
needles, melts at 173-174, and is exceptionally sensitive to atmospheric
oxidation. In organic solvents such as benzene, petroleum ether, ether, or
72
alcohol it forms dark reddish-brown solutions which are stable if air is

compIetely excluded. Cold water which has been saturated with nitrogen
neither attacks nor dissolves the compound. After the discovery of the
tricyclopentadienyl compounds of the rare-earth metals, attempts to
prepare the compound C O ( C ~ Hwere
~ ) ~ made. By the interaction of
[Co (C5H6)2] Br and lithium cyclopentadienyl in liquid ammonia or
tetrahydrofuran a new type of binuclear derivative, the reddish-brown
compound C02(CaH5)5, was obtained (59); this substance is freely soluble
in solvents such as benzene and ether.
The nickel complex, iYi(CsHs)z, has been made by the action of the
Grignard reagent on nickel (11) acetylacetonate (217)or from potassium
cyclopentadienyl and the ammine [Ni(NHs) e] (SCN) 2 in liquid ammonia
( 5 8 ) .It forms dark emerald-green crystals which sublime at 80-90' and
which, when heated in nitrogen, melt, with decomposition and the formation of a nickel mirror, at 173-174'. It is only slowly oxidized in air, and
cold water neither attacks nor dissolves it; it is, however, readily soluble
in organic liquids. Oxidation of the compound yields an orange-yellow
solution containing the ion [Ni(CaHa)z]+,which is stable for a short
period in weakly acidic media, and which may be precipitated as the
reineckate or tetraphenylborate.
Ruthenium, the homologue of iron in this group, was also shown to
form complexes quite early. Ruthenocene, Ru (C5Hd2, is obtained by
treatment of the acetylacetonate of tervalent ruthenium with five times
the theoretical quantity of the Grignard reagent (206), or, better, by
the action of cyclopentadienyl sodium on ruthenium trichloride in tetrahydrofuran ( 4 7 ) .It forms pale yellow scales which sublime at 120' and
melt a t 200'. Its properties are closely similar to those of ferrocene;
it is soluble in organic solvents, and in the absence of air is not attacked
by bases or by sulfuric or hydrochloric acid. Oxidation converts it into
the pale yellow [ Ru (C5H5)2] + ion.
The compound Rh(C5H5)2has not yet been obtained, but the action
of cyclopentadienyl magnesium bromide on rhodium (111) acetylacetonate yields a derivative of the cation [Rh(C6H5) J + , which forms a yellow
solution in water ( 9 5 ) . It gives precipitates with the usual reagents and
is generally similar in properties to the [ Co (C5Ha)2 ] + ion.
Osmocene, O S ( C ~ H ~has
) ~ recently
,
been made, in 20% yield, by the
interaction of cyclopentadienyl sodium and osmium tetrachloride ( 4 7 ) .
It forms colorless crystals which melt at 230'; again, it dissolves in
organic solvents; water, however, has no action on it, nor does it dissolve.
The hydrogen atoms in the cyclopentadienyl rings, like those in ferrocene
and ruthenocene, can undergo substitution; this property has not been
described for other metal cyclopentadienyl compounds, except in one
73
reference to the [Co(CsH5)z]+ ion. Careful oxidation of osmocene by

ferric chloride results in the formation of the brown [0s(CsHa)20H]+
cation; a t low concentrations the color becomes yellow.
The [Ir (CaH5) J + ion, analogous to the [Co (CSH~)
21 + and
[Rh(C5Hs)a]+ ions, also has been obtained and investigated; as was
expected, it showed similar properties. The uncharged complex I r (CSHS)2,
like Rh(C5H5)2, has not yet been prepared. For palladium and platinum
only simple olefin complexes are, as yet, known. Among these there is
also a complex of platinous chloride with cyclopentadiene, but physical
evidence excludes the possibility of a "sandwich" structure in this compound (29).
15. Indenyl Compounds of Eighth-Group E Eements
For metals in this group several compounds have been made in which,
instead of cyclopentadiene, indene is the source of the five-membered
ring. Diindenyl iron, Fe(CsH7)2, may be prepared as violet-black crystals by the action of indenyl magnesium bromide (76) or indenyl lithium
(151) on ferric chloride. This compound sublimes at 90" in vacuo, sinters
at about 1W0,
and melts a t 179-180"(in an evacuated tube at 184-185').
The solid substance is relatively stable in air, but its solutions in organic
solvents such as benzene, ether, or alcohol (which are dichroic), decompose rapidly. Water neither dissolves nor attacks the compound, and it
is not possible to oxidize it to a stable cation. It may, however, be hydrogenated in alcoholic solution, using a platinum catalyst, and ditetrahydroindenyl iron, Fe (CeHI1)2, an orange-colored oily liquid (fp 18.519", distilling without decomposition at 125" at 10-3 m m pressure) is
thus obtained (7'7). This substance is stable to air for a short time; it is
insoluble in water, but dissolves readily in organic solvents. Suitable
oxidizing agents convert it into the rather unstable blue [Fe (CsHll) 21 +
cation.
Diindenyl cobalt, Co (CsH7)2, may be produced by the interadion
of indenyl potassium and the amrnine [CO(NHB)~]
(SCN12 in liquid
ammonia (87). It forms black lustrous crystals which can be sublimed
without decomposition; they slowly sinter at 160" and melt rather gradually at about 180". The compound is very soluble in benzene, ether, and
alcohol but less soluble in petroleum ether, forming brown solutions. It is
considerably less sensitive to oxidation than the dicyclopentadienyl,
C O ( C ~ H ~but
) ~ powerful
,
oxidizing agents convert it into the yellow
[Co(CeH7)2]+ ion (873, which may also be obtained by a Grignard
reaction (151).
A dark red-brown nickel compsund, Ni(CeH7)2, may also be prepared; this gives cherry-red solutions in organic solvents (87). It is
74
TABLE I1
MACXETIC
MOMENTS
OF CYCLOPENTADIENYL
METALCOMPOUNDS
Central
ion
Mg+z
In+
Ti+
No. of Moment
unpaired (Bohr
Cornpourid
elec- magnetons)
(Cp = [CsHs]-) trons
calc
MgCp2
InCp
TICD
SC+3
y + 3
La+3
Ce+J
Pr+3
0
0
0
0
0
0
0
0
0
0
0
0
Nd+3
Sm+3
Gd+3
Dy+3
Er+3
Y b+3
YbCp3
Cp3UCl solid
111 solution
SnCpz
PbCpz
TiCp2
(TiCp?)'
CpzTiBrz
BlCpx
Moment
(Bohr
magnetons)
exP
0
0
0
0
0
0
2.46
3.61
3.63
1.54
7.98
10.00
9.45
4.00
3.16
2.72
0
0
0
2.3
0
0
3.78 k 0.19
2.86 k 0.6
Nb+5
Cr+z
Cr+3
2
2
0
0
0
1
0
0
3
2
1
0
2
3
2.83
2.83
0
0
-0
1.73
0
0
3.87
2.E3
1.173
0
2.83
3.87
Mo+~
Mo+5
Mn+2
1
1
Fe+2
Fe+3
0
1
0
1.73
1.73
0
1.73
1.73
0
2.26
2.34 f 0.12
1.76 k 0.07
0
2
0
2.83
0
1.7
0
2.86 k 0.1
1
0
0
1.73
0
0
1.73
0
2.83
0
0
0
1.76 f 0.09
0
2.12 k 0.1
0
0
T3+4
Sn+2
Pb+2
Ti+P
Ti+3
Ti+'
Bi+3
\l+Z
v+3
y+4
Co+2
Co+3
Ni+a
N i+3
Fe+Z
Fe+Z
co+2
Co+3
hTi+2
Ru+2
Ru+2
cocp,
Co2Cpssolid
in solution
(COCPZ)+
NiCp,
0
20
0
1.95
0
3.02 rt 0.15
3.81
3.73
0
1.5 k 0.1
Refereuce
66
56
71
9
9
9
9
9
9
9
9
9
9
9
164
;y4
71
SY
211
217, 211
YS
66, 51,116
211,
2lY, 211
211
66
74
71
22
22
216, 116
219, YO, 34
219
34
g134
39
34,807
58, 66, I 1 6,
618, 34
58, 34,218
161, Y6, 66
66
87, 66
87
66
66
Yl
75
considerably more labile than the compound Ni (C6H.6) 2-an order of

stability which is common to all indenyl complexes. The orange ruthenium compound, Ru(C9H7)2,has also been isolated (185); like the iron
complex, on hydrogenation it absorbs only four atoms of hydrogen per
six-membered ring.
D. PHYSICOCHEMICAL
AND CRYSTALLOGRAPHIC
PROPERTIES
OF PURE
METALCYCLOPENTADIENYLS
1 . Magnetic Investigations
I n Table I1 are assembled the compounds for which measurements

have been made (so far by the Gouy method) and the results obtained.
The behavior of the manganese compound, Mn (CaH,) 2, is especially
interesting. This compound has been investigated over the temperature
range 4-530'K. (116, 216), and is found to be antiferromagnetic with a
N e d temperature of 134 2K. Earlier results are thereby superseded.
100
200
300
400 500
TOK-
Fro. 3. Temperature dependence of the susceptibility of Mn(Cs&), in the range

1.5-470"K.
The brown low-temperature form of Mn(Ca&)2 shows a sharp transition point a t 158-159K and undergoes an enantiotropic transition into
a pale pink modification. There thus takes place a sudden transition to
five unpaired electrons, see Fig. 3 (corresponding to a moment of 5.9
Bohr magnetons for 8 = 0) and these are also observed to be present from
the melting point a t 170-172" to 530K. We interpret this as being a
gradual transition of a penetration complex in which relaxation occurs
76
into a purely ionic coinpound with increasing temperature. However,

conditions in the lattice of the brown Mn(CsHs)2, which have
not yet been conclusively elucidated, may also play a decisive part.
In solid solution in Mg(C5H5)21as in benzene or ether solution, the
compound behaves as a purely ionic complex a t all temperatures (66,
216). Dicyclopentadienyl vanadium is also found to be antiferromagnetic at temperatures below 13K. (116).
An investigation of the parahydrogen conversion by the manganese
compound in tetrahydrofuran threw little light on the problem, for the
compound dissociates completely into ions, and so only the characteristic
results for the b l n + + ion were obtained (182).Some others of the
compounds listed in Table I1 have also been investigated by means of
the parahydrogen conversion (180, 181, 182).Except for the uncharged
complexes C O ( C ~ Hand
~ ) Ni(CBHd2,
~
they gave results in harmony with
those obtained by magnetic measurements. For the two compounds mentioned, there appear to be abnormally great orbital moments for the
unpaired electrons (175).
2. Polarographic Investigations
The redox potential has been determined for a number of the new
compounds; it is usually found to be similar to those of known compounds of the metal. An increased stability of the higher oxidation state
of the metal with higher atomic weight is observed. Table I11 sets out
the data. The second column gives the values measured against the
TABLE I11
POLAROGI~APHIC
HALF-WAVE
POTEIZTIALB
OF SOMEFIVE-MEMBERED
RING COMPLEXES
Half-wave potential (v)
Redos reactioii
in water
in alcohol
-0.44
-0.3'2
-0.44
-0.71
+0.16
-1.16
+0.3 1
+0.30
-0.08
-0.21
+0.11
-0.60
f0.26
+0.22
Reference
211
21 1
91 1
911
149
149
149
218
218
149
149
151
77
standard calomel electrode in aqueous solution; the third column the

values in alcoholic solution. For the dicyclopentadienyl cations of Rh
(111) and Ir(II1) (also present as their perchlorates) a half-wave potential of -1.53 v in water for the former showed that the neutral
Rh(C5Ha)zwas probably formed, whereas for the latter no indication of
Ir(C5Hs)2
was found (25).
3. X-ray Investigations
This method was applied even at an early date to elucidate the

molecular structure of the metal cy clopentadienyls. After the first investigations of ferrocene itself (30,33, 7'0,161), others, such as those
of Co(CSH5lz (16R), Ni(C,Hdz (S.8,159), V(CSHS)Z(199),Mg(C,HdZ
(199), Cr(C5Ha)2(ROO), and of the brown form of Mn(CaH6)z (198)
followed. Space groups and unit cell dimensions were determined: with
the exception of Mn(C5H5)2,which crystallizes in the rhombic system,
with the probable space group &h21 with 16 molecules in the unit cell,
all the compounds have monoclinic symmetry with space group C2h5,with
two molecules in the unit cell.
Contrary to the view first put forward ( H ) , Mg(CaHs)z in the solid
state also possesses the pentagonal antiprismatic structure first established for ferrocene (40).For Mn(CaH5)2, indirect evidence, such as the
volume occupied by a molecule and the formation of mixed crystals with
Mg(CaH5)2in space group C Z , ~
makes
,
it very likely that this compound,
too, has a before sandwich-configuration. Table I V presents the results in
TABLE IV
LATTICECONBTANTS
Substance
Fe(Cd%)t
cO(csH$,
Ni(CsHd2
(Mn(CsH&)
Cr(CbH3,
V(CsHr)z
Mg(CsHs)r
OF SOME CYCLOPENTADIENYL
ci,
$1
5.91
5.90
5.88
(9.95)
5.92
5.88
5.98
7.59
7.71
7.86
(11.70)
7 .a8
8.02
8.04
10.51
10.60
10.68
(28.07)
10.72
10.82
10.98
METALCOMPOUNDS
B
(degrees)
Volume of unit
cell (As)
121.1
121.1
121.2
201.8
206.4
211.1
(204.1)
213.8
218.0
224.1
121.2
121.3
121.9
comparative form. The form of the unit cell shown diagrammatically in

Fig. 4 is valid for all of the compounds investigated except the manganese
complex, and rests on Fourier analyses of ferrocene (80,31). These
showed that in ferrocene the CCC distance in the rings (1.403If: 0.02 A)
is uniform and slightly greater than that in benzene, the Fe-C distance
78
E. 0. F I S C H W AND H. P. FEITZ
is 2.045 0.01 A, and the distance between the rings 3.32 A. In Ni (CaHa)
the C - C distance is somewhat shorter, the Ni-C distance 2.20 A, and the
rings are 3.4 A apart (92).On the basis of the three-dimensional Fourier
analysis, rotation of both molecules in the unit cell or of both rings in individual molecules must be ruled out (91). The pentagonal antiprism
FIG.4. The packing in the unit cell of metal dicyclopentadienyls.
configuration of the molecule is thus proved. Electron-diff raction studies

at 400" give Fe-C as 2.03 zk 0.02 A and C--C as 1.43 ? 0.03 A. At this
temperature free rotation of the rings appears to occur (183).It seems
significant in this context to note that for free cyclopentadiene, values for
the C=C distance of 1.34 A and for the C-C distance of 1.54or 1.44 A
had previously been found (179). From Table IV it is clearly seen that,
in general, with increase in metallic radius from iron to nickel and
iron to vanadium only h and c increase, while a, corresponding to the
orientation of the molecules, remains approximately constant. The abnormal size of the unit cell of the magnesium compound is striking and
a radius of the magnesium atom of about 1.6 A can be extrapolated,
although the compound is in every way essentially ionic, so that a magnesium ion with radius 0.78 A would be expected.
These circumstances suggest that the *-electrons of the five-membered ring are not incorporated in the magnesium ion, which already has
an inert gas configuration, and that they therefore act as a buffer and
make the interatomic distance greater.
Numerous compounds have also been investigated by means of the
X-ray absorption edge method (11, 12, 13, 119). This allows one to
identify the lowest free p-orbitals which can be occupied by excited 1s
electrons. The results obtained for ferrocene in the solid state, in solution in solvents of different polar character, and in the vapor phase, are
79
shown in Fig. 5. For comparison, K4[Fe(CN)e] and Fe(CO)6, as cornplexes in which the krypton configuration has been attained, are also
included. The weak first absorption band probably corresponds to a 413
+i
+I0
~
~-
FIG.5. X-ray absorption edges.
sntibonding absorption (Ruch) and shows therefore that the 4p-levels

are involved in the ring to metal bonding.
4. Mass Spectrographic Investigations

These were made on the uncharged dicyclopentadienyls of vanadium,
chromium, iron, cobalt, nickel, ruthenium, manganese, and magnesium,
80
and on the compound (C6H6)*ReH (88),and it was thus possible to

show the close relationship of Mg (CaH6) and Mn (CaH,)2. The atability
of the nonionic dicyclopentadienyl compounds, as compared with thc
ionic ones, stands out clearly in this work,
5 . Spectroscopic Investigations
The first infrared investigations were made on ferrocene and ruthenocene (123). Together with the Raman measurements also made on
both compounds, they constitute a relatively simple spectrum for compounds containing so many atoms, and there must thus be a high degree
of symiiietry in the structure. Apart from a C-H frequency a t 3075 cm-l
(observed using a potassium bromide disk), i.e., in the aromatic C-H
stretching region, only four strong bands are found in the whole range
of the rock-salt prism. On the basis of the spectrum the symmetry group
DBdwas proposed (123) and later confirmed (1.24). More extensive investigations, among thcm those on Fe (C5D5)2, led to reliable frequency
assignments for Fe (CBHJ 2 r Ni (C6H5)2, and Ru (C5HB)0.
On the basis of these results and our own studies on numerous other
metal dicyelopentadienyls, it was possible to define various bands in the
infrared spectrum as characteristic of ferrocenelike Me (C5H5)2 structures: (1) a C-H stretching frequency a t 3100 cni-I (WH) ; (2) a C-C
stretching frequency at 1410-1430 em-' ( V C C ) ; (3) an antisymmetrical
ring-stretching frequency a t 1100-1110 cm-' (vcC) ; (4) a C-H bending
frequency at lo00 cin-l (Sen) ; ( 5 ) a frequency at 770-830 cm-', which'
can be interpreted either as a G - H bending (124),or as a metal-carbon
stretching frequency ( 8 2 ) . Frequency ( 4 ) , for example, is so characteristic for unsubstituted five-membered rings, that with its aid it can
be decided whether mono- or di-substitution has taken place in products
of substitution reactions. Most measurements of ultraviolet absorption
spectra were made in parallel with infrared and Raman measurements.
Table V shows for which compounds spectroscopic investigations have
been published. In the infrared region the spectra of the neutral and
cationic complexes of the same metal are essentially similar, but in the
ultraviolet region they are distinctly different.
6 . Thermodynamic and Other Data
The possibility of calculating thermodynamic functions from infrared and Raman frequencies has been exploited for Fe(C5H5)2,Ni(C5Ha)2,
and Ru(C6H6)2 (124). The vapor-pressure curve of ferrocene was also
reported soon after its discovery ( 1 1 2 ) .Vapor-pressure measurements for
dicyclopentadienyl manganese are also available t.216). The heats of
formation of Fe(C&)2 and Ni(CaH5)2 from the C5H6 radical and the
81
TABLE V
SPECTROSCOPIC
STUDIES
OF CYCXDPENTADIENYL
AND DIINDENYL
METALCOMPOUNDS
References
Substance
Mg(CsHoh
WCsHsh
(CaHs)Cu[P(CaHs)r]
(Lanth) (CSH&
(CsH&UCI
(CaHs)Si(CHsh
(CsH&TiBr,
(Cd-IMrBrz
(CsHshvC12
(CsHsIzNbBr3
(CsHdfaBra
[(CoHs)zMoCl]+
[(CsHsl~M~Cl?l+
[(C~Hs>zWClzl+
Mn(CsH&
Fe(CsH&
[Fe(C6HM+
[C~(CsHs)zl+
Ni'C6Hs)z
Ru(CsHsh
[Ru(CsH6)zlf
[Rh(CsH&I+
~Ir(CsHsM+
Fe(COH&
Fe(CoHdz
Co(CoH&
[CO(C~H~)~+
Maxima found in
ultra violet (mp)
238, 290
247
Ultraviolet
816
816
9,212
Infrared
816
816
816
17 maxima in
range 400-800
164
164
240
816
115
dl1
811
211
21 1
211
88
88
88
816
325, 440
219,118
253, 619
260,310,410
308, 700
275, 320
320
230
26,907'
68,218
206
806
114, 118
12.9, 124
$6
96
76
77
87
151
87
151
below 200,300
270,430,665
214 320,435
274,322, 410
322
819,111,
183, 184
819
* Absorption like that by the free ions of the lanthanides.

gaseous metal were found to be -147 and -123 kcal/mole, respectively,
and those from elements in their standard states to be 33.8 and 62.8
kcal/mole ($18). Spectroscopic measurements led to values of 40.7 and
62.8 kcal/mole, respectively (194). Conductivity measurements have
been made on a series of compounds (164,816). Measurements of dipole
moments of cyclopentadienyl metal complexes have so far been made
for only a limited number; the results are collected in Table VI. From
measurement of the dipole moments of mono- and diacetylferrocenes
in benzene solution (3.03 and 4.23 Debye units, respectively) it was
82
H.
E. 0. FISCHEB AND
P. FRITZ
TABLE VI
DIPOLE
MOMENTSOF S o n COMPLEXES
IN BENZENE
Substance
Dipole moment
(Debye units)
~~
Reference
~~
1.02
1.63
0
1.28
0
0
0
0
0
0
197
197
74
74
197,219
74
197
74
74
&, 197
concluded that both rings are rotating freely (165).The dipole moment
of di- (p-chlorophenyl-cyclopentadienyl) iron in benzene is 3.12 0.03
Debye units. This eliminates free rotation in this compound, since the
moment is in closest agreement with those calculated for the 1,l prismatic and 1,l antiprismatic structures (184).
The method of proton or nuclear magnetic resonance has been applied to the compounds (CBH5)2ReHand Fe(C5H6)2(215,15,2).Furthermore, attempts were made by spectrophotometric methods to establish
the formation of a thermodynamically stable molecular complex of ferrocene (14). According to these measurements, ferrocene and iodine together in solution show no charge transfer complex, but are merely in
thermal equilibrium with ferricinium triiodide.
E. FERR~CENE
AS AN ORGANIC
COMPOUND
The inorganic problems in the chemistry of the cyclopentadienyl
metal compounds (i,e., the varying of the central metal atom) were dealt
with in section I1.C. Those on the organic side (Le., the behavior of
the five-membered ring bound in the complex) will now be briefly reviewed. The outstanding results here are those showing the surprisingly
easy substitution of the hydrogen atoms in ferrocene. So much work has
been published on this aspect of ferrocene chemistry that within the
pattern of this aurvey, which is orientated more toward inorganic chemistry, only a condensed summary can be given.
Any idea of describing each compound that has been prepared must
be abandoned. In the following, only the most important types of ferrocene derivatives and reactions are dealt with; as for the rest, however,
83
an attempt will be made to give as complete a review as possible of the

literature which has appeared so far.
1. Substitution Reactions
Woodward was the first to show that ferrocene, in its reactivity, is to

be considered as an aromatic system ($$4), and it was by him that, on
analogy with benzene, the name ferrocene was given. He was the first
to carry out a Friedel-Crafts reaction, in which on treatment with acetyl
chloride and aluminum chloride ferrocene gave diacetyl ferrocene, which
was converted by hypoiodite into the dicarboxylic acid, esters of which
could easily be obtained. At first, it was thought that this reaction is not
generally applicable to ferrocene, and that, for example, oxalyl chloride,
ethyl bromide, isopropyl chloride, or benzyl chloride in the presence of
aluminum chloride did not react (166). Later, however, it was shown
that from the last three of these compounds and ferrocene, alkyl ferrocenes may be prepared under suitable conditions (1S5). It was also found
that ethylene dichloride and ferrocene can be converted into a highly
polymeric resin (135). Alkyl ferrocenes can also be made directly from
alkyl halides and ferrocene (1S5),by reduction of acyl ferrocenes (3:
137, 195)) or by reduction of carboxylic acids (1S9). The reactions of
the acyl compounds obtained by the Friedel-Crafts synthesis have been
investigated (94, 95, 100, 148, 165, $04). They give many of the characteristics of carbonyl groups. Many comparative studies of the influence of conventional and of ferrocenyl substituents on organic systems
have been made (140, 145); orientation on substitution has also been
investigated (158, 166, 171).
The preparation and structure of homoannular diacetyl ferrocenes
(164), and intramolecular cyclizations, both homoannular (I68) and
heteroannular (148, 168), have been exhaustively investigated. As part
of this work, ferrocene o-carboxylic acids have been prepared (170).A
rearrangement with ortho-substitution of a quaternary ammonium ion
derived from ferrocene is also noteworthy (89). Work has been done
on the isolation of optical isomers of substituted ferrocene derivatives
(6, 166) and on cis-trans isomerism (169).
The usefulness of catalysts other than aluminum chloride, e.g., anhydrous hydrogen fluoride ($09, $23), boron trifluoride (100), and phosphoric acid (94), has been examined. The mercuration reaction which
is characteristic of aromatic systems takes place also for ferrocene, as has
been described by Nesmeyanov and his co-workers (1.63, 1.47). Through
the mercury compounds these workers have also obtained the halogenoferrocenes.
84
E. 0. FISCHEE AND H. P. FRITZ
Interaction of ferrocene and diazonium compounds, which is followed

by loss of nitrogen, leads to a large number of derivatives (17, 141, 142,
143, 20s); most of them, naturally, are aryl-substituted compounds.
Ferrocene cannot only be mercuriated, but can also be metalled by
lithium butyl or similar compounds such as sodium phenyl. A great
variety of possibilities for substitution is afforded by the interaction
of ferrocenyl lithium and halogen-containing compounds (6, 128, 143,
162). Carboxylic acids, in particular, may easily be obtained by formation of the lithium derivative of ferrocene and treatment with carbon
dioxide.
It was soon discovered that ferrocene condenses with aldehydes:
under suitable conditions either condensation with elimination of water
or carbinol formation may occur (136, 166,202). Further, base-catalyzed
aldol and Claisen condensations involving the a-hydrogen atom of acetylferrocene can be carried out (100).
In addition to all these possible reactions, substituted ferrocenes
can also be prepared directly from the corresponding cyclopentadiene
derivatives (160). Most of these, it is true, are difficult to obtain or
even, at present, unknown. It seems, therefore, that the reverse way of
making substituted cyclopentadiene derivatives through the ferrocenyl
compounds, followed by abstraction of the metal, is of greater practical
significance. The splitting off of the ring may be brought about by hydrogenolysis or by the action of halogens; the application of a pressure
of 281) atm of hydrogen at s temperature of 300" makes the former
process difficult to handle, however, and a much more elegant way is by
reductive cleavage with lithium in ethylamine (194).
The simplest substitution, the replacement of hydrogen by deuterium,
has not yet been achieved, though under the conditions used all six hydrogen atoms of cyclopentadiene are replaced (115).
2. Basic Types of Ferrocene Derivatives
The number of derivatives now known is so great that, in addition

to those mentioned in the preceding section, some other basic types can
be distinguished among the multitude of acyl, alkyl, and aryl derivatives of ferrocene. Except for a nitro-compound all the general aromatic
derivatives are known. This has been the case for the longest time for
the carboxyiic acids and the di- and monosulfonic acids obtained by the
action of sulfuric acid (in acetic acid or acetic anhydride) on ferrocene.
The disulfonic acid forms large hygroscopic crystals. Salts and methyl
esters of both acids are known (146,902).
Ferrocenylamine, Fe ( C I O H ~ N His~ )obtained
,
by the Kosheshkov reaction from ferrocenyl lithium and the benzyl ether of hydroxylamine
85
(140, 24.8) or, alternatively, from ferrocene monocarboxylic acid by the

Curtius degradation (3).It is an orange crystalline substance which has
a transition point at 110" and which melts a t 153-155". It can be sublimed in vacuo, is easily soluble in organic solvents, and forms acetyl
and benzoyl derivatives (mp 170.5-172" and 177-178' respectively). The
amine hydrochloride is readily soluble in water but only slightly soluble
in hydrochloric acid. At 20" in 80% alcohol it has the following dissociation constant and pK values (140), which may be compared with
those of aniline.
Kdim
Ferrocenylamine
Aniline
10-9
7.2 X 10-"
1.55
PK
8.81
10.14
Ferrocenylamine is therefore a considerably stronger base than aniline.

The preparation of ferrocene aldehyde, Fe(CloHBCHO), may be
achieved by direct formylation of ferrocene with N-methylformanilide
(16,94,95,172) ; by this method only one -CHO group is introduced. The
aldehyde is also formed by the action of hexamethylenetetramine on
ferrocenylmethyltrimethylammonium iodide (18). Further methods are
by the conversion (120) of N,N-dimethylaminomethylferrocene (obtained by aminomethylation of ferrocene) into the aldehyde and via
the transformation of the methiodide of this compound (109). The aldehyde is a red crystalline substance which is seen under the polarizing
microscope to have a transition point at 45"; it melts at 124.5'. It forms
a bisulfite compound, an oxime, a semicarbazide, a 2:4-dinitrophenylhydrazone, and an acetal. It is soluble in aqueous acids and organic solvents, and it undergoes the Cannizzaro reaction, but, surprisingly, it
does not give the corresponding acid on oxidation (94). The typical
aldehyde reactions of formylferrocene have been extensively investigated (101)) e.g., the aldehyde can be converted through its oxime into
ferrocenyl-cyanide, Fe(CloH&N) (18).
The first phenolic derivative of ferrocene to be reported, di(1-hydroxy-3-methyl-cyclopentadienyl) iron, was prepared by the reaction
between 3-methyl-2-cyclopentanone,sodamide, and ferrous chloride in
liquid ammonia. This compound is remarkably unstable, being extremely
sensitive to air. It is a yellow substance which sublimes at 130-140O at
0.1 m m pressure, and is readily soluble in methylene dichloride but
insoluble in carbon tetrachloride and in n-heptane. In water saturated
with nitrogen a yellow solution is produced which rapidly becomes biuegreen on exposure to air. The compound forms a benzoate (mp 121123.5') which is stable in air. An analogous derivative of the dicyclopentadienyl cobalt (111) cation has also been obtained (6).
86
Many derivatives have been prepared from ferrocene monocarboxylic

acid (3). Acetylferrocene was reduced by lithium aluminum hydride to
the carbinol, and this was then converted into vinylferrocene. From this,
polymers and copolymers with other polymerizable substanEes have
been obtained. The polymers are easily obtainable in the cationic form
and in the reduced, uncharged form, which are interconvertible (3).
Urethanes (3), amino acids and urea, hydantoin and pyraeoline derivatives with ferrocenyl substituents have also been prepared (100, 178).
Halogenoferrocenes were isolated, as already mentioned, via the mercury compounds (1.47). The iodo compound, Fe (CloH91) was described
as not forming a Grignard reagent.
For the preparation of unsymmetrical ferrocenes, two ways by which
only mono-substituted derivatives are produced have been suggested.
One route starts from iron tetracarbonyl and a substituted cyclopentadiene; the other from monocyclopentadienyl iron dicarbonyl bromide,
which on treatment with a substituted cyclopentadienyl lithium is finally
converted into the corresponding mono-substituted ferrocene. Experience
shows the first method to be more suitable for the preparation of aryl,
and the latter method for the preparation of alkyl, derivatives (97).
Corresponding work already carried out on substitution in Ru (C5Ha)
and O S ( C ~ Hhas
~ )also
~ been fruitful. It is found that in the FriedelCrafts reaction with acetyl or benzoyl chloride there is a distinct predominance of mono- ovcr disubstitution as the atomic weight of the
central atom increascs (47, 72).
F. MIXEDCYCLOPENTADIENYL
METALGOMPLEXES
In addition to the complexes of cyclopentadiene and indene, described already, in which the metal usually lies between two rings, a
great number of complexes in which the central metal atom is surrounded
on one side by a ring and on the other by conventional ligands are also
known.
1. Cyclopentadienyl Metal Carbongls
The group of monocyclopentadienyl metal carbonyls is the most extensive. In this, in contrast to the pure cyclopentadienyl compounds, it
is found that complexes are formed only by transition elements of groups
5 to 8, i.e., metals which can enter into true ferrocene-type bonding.
There is a regular relationship for this series: transition elements of
odd atomic number yield readily volatile compounds of formula
C5H5Me(CO),, while the even-numbered members form dimers of composition [C5HSMe(CO),l2. The actual ralue of 5, the number of ligand
carbon monoxide molecules, corresponds to tlie number of pairs of elec-
87
trons required for the attainment of the next inert gas configuration if
it is assumed (a) that the five-membered aromatic ring is again bonded
to the central metal atom with all six of its ?r-electrons, and (b) that a
metal-metal bond contributes in the case of the dimeric compounds (86).
In accordance with these principles the following series of diamagnetic
complexes may be adduced for the first period of transition metals:
CsHsV(C0)4, [CdbCr(C0)3]2, C6HSMn(CO)3, [CsHsFe(C0)212,
C5H5Co(CO) 2, and [CsH6Ni(CO) ] 2. The cyclopentadienyl carbonyl
complexes known a t the present time are described more fully in the
following sections.
a. Fifth Subgroup. In this, only the complexes of vanadium have to
be considered. The longest-known monomeric compound of the mixed
type is CaHsV(CO) 4, which is obtained by the action of carbon monoxide
under pressure on dicyclopentadienyl vanadium a t elevated temperatures (61). It is an orange-colored substance of mp 138" which is readily
sublimed at 60-70" and is stable for a short time in air. Solutions in
organic media decompose more rapidly. From this compound it proved
possible to prepare the first derivative of a monomeric cyclopentadienyl
metal carbonyl. By decomposition of CsHsV (CO) with sodium in liquid
ammonia the yellow sodium salt, Na2[C5H5V(GO) 3], which is soluble
in ammonia, is obtained. From this, also in lig. NH3, the corresponding
potassium, rubidium, and caesium salts, which are sparingly soluble and
diamagnetic, may be precipitated on addition of their iodides. In all these
compounds vanadium has the oxidation number -1 (80).
b. Sizth Subgroup. The green diamagnetic compound (CaH5)2Cr2
((30)s was the first chromium compound of this type to be mentioned.
It was described as being produced by interaction of chromium hexacarbony1 and cyclopentadiene in the liquid phase (22). According to our
own detailed study of the action of carbon monoxide under pressure on
the cyclopentadienyl Cr (C5H5)2 at various temperatures, there results
first, at 100-110", mainly brown-black crystalline (CsHs)3Cr2(CO) 3,
which sublimes at 130" and melts with decomposition at 190-193'. It is
practically insoluble in nonpolar solvents but dissolves freely in some
polar compounds such as dimethylformamide. It is assumed to have the
saltlike structure [Cr (C5H5)21 + [CaHaCr(CO)81 -. On increasing the
temperature to 150-170, the chief product becomes the blue-green
crystalline compound [CsHsCr (CO) 312, which sublimes at 90-100" and
melts at 163-168' with decomposition. It forms solutions which are deep
green in nonpolar solvents such as benzene and yellow-green in alcohol,
and which rapidly decompose in air (62).If hydrogen is added in the
pressure reaction, the readily volatile yellow monomeric hydride, CaHoCr
(CO)3H, is obtained a t about 75", and this readily decomposes, with loss
88
E. 0. FISCHE1B AND H. P. FRITZ
of hydrogen, to form the dimeric cyclopentadienyl carbonyl compound.

When pure, the carbonyl hydride, which is golden yellow, melts at 58-58' ;
it gives yellow solutions in organic solvents which are stable if oxygen is
excluded. Its acidic character is shown by the ready transformation with
aqueous alkalis to give salts containing the air-stable anion [C5H6Cr
(CO)a] - ; these yield precipitates with suitable cations. When such solutions are acidified, the sparingly soluble weak acid is reprecipitated (5@.
Alkyl esters are obtained by treatment of the sodium splt with alkyl
halides (157)or dimethyl sulfate (69).
The first mixed cyclopentadienyl metal acetylacetonate t o be made is
the dark green compound CBH5Cr(CH3COCHCOCH3)Br, which may be
sublimed, and which results from the reaction of cyclopentadienyl magnesium bromide with chromium (111) acetylacetonate in benzene. (193).
The homologues of chromium have also been very fully studied.
Thus, the dark purple compound (C5H5)2M02(CO)a (mp 215-217') was
first described as formed from molybdenum carbonyl and cyclopentadiene at about 300' (908). It is readily soluble in chloroform, less so
in carbon tetrachloride, benzene, carbon disulfide, and alcohol, and
scarcely soluble in ligroin. Probably, however, the substance was in fact
the dimer [C5H3Mo(CO)3]3 (154).The hydride C5H6Mo(CO)3H was
obtained as a yellow, readily volatile compound, decomposing at 50-52",
by the interaction of molybdenum carbonyl and cyclopentadienyl lithium
in dioxane and acidification of the resulting solution (59).As in the case
of the chromium compound, dimerization with loss of hydrogen occurs,
and alkyl esters may be prepared by the action of the alkali nietal salts
on alkyl halides (154,155, 157). The corresponding chloride, C6H5MO
(CO)&i, was obtained in the form of orange crystals, which decomposed a t about 145O, from the hydride and carbon tetrachloride. The
crystalline iodide, C5H5M~(CO)31
(mp 134') can be made as red crystals from the mercury salt [C5H5Mo(CO)3]Hgand iodine (154).Oxidation of the carbonyl hydride gives the cyclopentadienyl carbonyl
[C6HsMO(C03)12 (154).
In the case of tungsten, reaction of the carbonyl with cyclopcntadiene
gives the purple-red crystalline compound [ C5H5W(CO)3] (mp 240242" with decomposition) , which is readily soluble in chloroform, much
less soluble in other organic solvents, and insoluble in ligroin. It is not
attacked by acids or bases. If tungsten carbonyl reacts with cyclopentadienyl lithium in dimethylformamide, a yellow solution of the
lithium salt is formed, from which the hydride, C5H5W(CO)3H,may be
isolated on acidification as a readily volatile yellow substance melting
at 66-67'. I n this case no loss of hydrogen occurs up to 180". Reaction
89
of the hydride with diazomethane gives the methyl ester (86),while

other alkyl derivatives are formed by the action of the sodium salt on
alkyl halides (154,155).
c. Seventh Subgroup. Yellow CaH5Mn(CO) may be obtained in
good yield by the interaction of the cyclopentadienyl compound,
Mn (C5H5)a, and carbon monoxide at elevated temperatures. It sublimes
above 60",melts at 77", smells like camphor, is completely stable in
air, and dissolves readily in organic solvents without decomposition (60).
The compound CH3CaH4Mn(CO) has recently become of technical interest as a fuel additive ( 2 ) .
On treating the compound (C5H6)2ReH,which has been mentioned
earlier, with carbon monoxide at 90" and 250 atm it is possible to obtain
the carbonyl (CaH5)2ReH(CO)2, which has neither acidic nor basic
properties (82). It sublimes at 90-100", melts a t 111-112", and is freely
soluble in all the usual solvents such as benzene, ether, petroleum ether,
and acetone. Unlike cyclopentadienyl rhenium hydride, it is stable in air.
d. Eighth Group. The compound [CKHKFe(CO)2]2
(22, 43,98, 1591,
which was originally formulated (22) as (C5Hs)2Fe2(CO)a, is obtained by interaction of iron pentacarbonyl and cyclopentadiene dimer
in the liquid phase. It forms dark red-brown crystals (mp 194") which
are readily soluble in pyridine, chloroform, and alcohol, less soluble
in carbon tetrachloride and carbon disulfide, and almost insoluble in
petroleum ether. It is neither attacked nor dissolved by water, but it
can be transformed by oxidation in an aqueous medium into what is
presumably the hydrated [CKH5Fe(CO) 2H20]+ cation. In organic solvents an anhydrous chloride (159),cyanide (169),bromide (98),and
iodide (156)may be obtained. From the bromide CKHKFe(CO)2Brand
cyclopentadienyl sodium, the fairly labile dicyclopentadienyl metal carbonyl, (CaHs)2Fe(CO)2,may be isolated in the form of deep red crystals
of mp 46". The existence of the [ (C6Hs)2Fe(CO)2]+ cation is judged
to be probable (96). The dimeric compound [C6HKFe(CO)2] 2, which
has already been mentioned, may be reduced in methanol solution by
O ) With
~ ] mercuric
sodium amalgam to a yellow anion, [ C K H ~ F ~ ( C -.
cyanide this yields the golden yellow mercuric compound, Hg [CsHeFe
(CO)2]2, which is stable in air and is soluble in all the usual organic
solvents. When it is sublimed at 80-90" partial decomposition occurs,
with loss of mercury and formation of the dimer once more (43). A
variety of alkyl and aryl compounds of the type C&d?e(CO)2R can
be obtained from the halides (156).
The compound C&$'e (CO)2% (CH3)3 was prepared as orange
needles, subliming at about 60" and melting at about 70", by the interaction of the sodium salt [ CsH5Fe(CO) 21 Na and trimethylchlorosilane
90
E. 0. FISCHER AND H. P. FBI"'&
(160). This compound is thermally stable up to about 200" and is soluble in the usual organic solvents.
Interaction of dicyclopentadienyl cobalt and carbon monoxide at
90-150" and 200 atm yields the red-brown compound C6H5Co(C0)2 as
a readily volatile liquid (mp -22') bp 139-140/710 mm with partial
decomposition) which is decomposed in air but forms stable eolutions
in organic solvents if air is excluded. With water neither solution nor
rapid hydrolysis is observed (60, 61, 159).
The first cyclopentadienyl metal carbonyls to be discovered have the
formulas (C6H6)&02 (GO) 4 and (C6H6)&oFeH (CO) , but are saltlike
compounds having the structures [Co (CsH,) 21 + [Co (CO) - and
[Co (CaH5)2] + [FeH (CO)4 ] - (57). For nickel also, the dimer
(CaH6Ni(CO)] has recently been discovered ; it is a red substance which
is obtained by the reaction between nickel carbonyl and dicyclopentadienyl nickel in benzene (69). By raising the reaction temperature, in
experiments on cleavage with sodium amalgam in methanol, or a t a
higher sublimation temperature, a very stable complex having the surprising composition (C5H5)3Ni3(CO) results.
2. C.yclopentadieny I Met a1 Nitrosy Is
a. Sixth Subgroup. An almost quantitative yield of the nitrosyl

C5H6Cr(CO)2N0 may be obtained from the cyclopentadienyl carbonyl,
[ C5H5Cr(CO)3] 2, and nitric oxide in benzene ( 8 5 ) .This orange-red-colored compound (nip 67-68') is soluble in all organic solvents, and is stable
in air a t room temperature. Xhen chromic chloride is allowed to react with
cyclopentadienyl sodium in tetrahydrofuran and the resulting green solution is treated with nitric oxide, the compound C5H5Cr(N0)&1 is obtained (156). It forms yellow-green crystals which decompose at about
140" without melting; with sodium thiocyanate, the corresponding thiocyanate complex is obtained. With methyl magnesium iodide it gives a green
crystalline compound (mp 83") of composition C5H5Cr(NO)2CH3.With
diaeomethane the chloride gives C5H5Cr(NO) ,CH2C1, while with ethyland phenyl-magnesium bromides the products are C6H5Cr(NO) & z H ~
and C5H5Cr(NO) 2C6H5, respectively (153). The cation [CSH5Cr
(NO) 2 ] +, which can be obtained in aqueous solution, will react in suitable solvents with cyclopentadienyl sodium to form the compound
(C6H6)2Cr(N0)2
as amber-colored crystals of mp 65' (153).
Nitrosyl derivatives of the homologues of chromium may be made
directly as insoluble precipitates by decomposing the complex anions
[C6H5Mo(CO)31 - and [CsHsW (CO) 3] - with nitric oxide in aqueous
media (85). After high-vacuum sublimation at 55-60' the orange-red
91
crystals of the molybdenum compound, C5H5Mo(CO) 2N0, which is

somewhat lighter in color than its chromium analogue, melt a t 85" ; the
tungsten compound melts at 105-107". The former compound may also
be obtained from the carbonyl hydride, e.g., by decomposition with nitrogen dioxide (154).
b . Seventh Subgroup. The cation [ C5H5Mn(CO)aNO]+ is obtained
in the reaction of the carbonyl C5H5Mn(C0)3with GM-nitric acid in
ethanol, and is isolated as its chloroplatinate, which is fairly soluble in
water and which decomposes without melting a t 165-175" (159). Reaction of nitric oxide with dicyclopentadienyl manganese in tetrahydrofuran gives a compound (C5H5)3Mn2(N0)3in the form of purple-black
crystals which decompose a t 100" and are moderately soluble in organic
media. They are insoluble in water and do not react with it (166). When
the compound is treated with sulfur in carbon disulfide solution, brownblack crystals of the sulfide C5H5MnNOS2,which have a metallic glance
and decompose on warming without melting, are obtained. This substance probably exists as a tetramer or hexamer (158).
c. Eighth Group. The interaction of dicyclopentadienyl nickel and
nitric oxide in petroleum ether yields the compound C5H5NiN0 as a
stable red-brown liquid, boiling a t 144-145" a t 715 mm, and miscible
with the usual organic solvents (60,85, 159).
3. Physicochemical Investigations
Apart from the compounds (C5H5)3CrZ(C0)3( 5 2 ) , which has three

unpaired electrons, and (C5H5)3Ni3(CO)2 (69), which has one, all the
complex compounds described here are diamagnetic. For most of them
the infrared spectrum has also been described in outline. Essentially two
kinds of C-0 bands are found. First, there is the normal carbonyl frequency at 1900-2000 cm-l, which also occurs in the pure metal carbonyls,
but sometimes in the cyclopentadienyl metal carbonyls with a small displacement towards longer wavelengths. Secondly, frequencies still more
displaced towards longer wavelengths are observed a t 1700-1800 cm-1,
and these are associated with bridging carbonyl groups. A further frequency, situated between these two regions, has been attributed by some
authors to carbonyl groups forming a sort of closed ring between the
metal atoms of binuclear cyclopentadienyl metal carbonyls ; but an X-ray
study of the structure of [C5H5Mo(CO)3] 2 (221) has very recently
shown that this interpretation cannot be correct. Crystallographic data
are available for the compounds [C5H5W(CO) 3]2 (22U), [C5HsMo
(CO) 312 (820, 2221, [ C ~ H S(COSI
F ~ 2 (98, 220), C5HtMn (CO)3 ( V ,
and C5H5NiN0 (8).
92
E. 0. FISCHER AND H. P. FRlTZ
G . MOLECULAR
STRUCTURE
AND BONDIXG
1. Molecular Structure
The configuration of the solid ferrocene-like dicyclopentadienyl derivatives is well established by X-ray methods as a pentagonal antiprism. In
ferrocene the rings are rigidly fixed in the staggered positions, but they
become freely rotating a t about 360" (18.3). It has, accordingly, proved
possible to isolate stereoisomeric ring-substituted derivatives of these
substances (6, 167). Alone among the dicyclopentadienyls so far examined, the tin and lead compounds Sn(CBH5)2and Pb(CjH5)z show
pronounced dipole moments in benzene. In view of the great similarity
between the infrared spectra of these two compounds and those of ferrocene and ruthenocene, the sandwich arrangement of the rings must be
present in the tin and lead derivatives; the rings may, however, be
slightly inclined towards one another. A certain degree of polarization,
i.e., departure from the completely symmetrical sandwich structure, is
also considered by E. Ruch to be possible from theoretical considerations.
An interesting suggestion has been made (118) for the structure of
dicyclopentadienyl rhenium hydride, (C5H5)2ReH, which might also be
considered for the cyclopentadienyl hydrides of chromium, molybdenum,
and tungsten. According to this, the proton resides within one of the
cones formed by the metal atom and the ring. The second proton which
is readily taken up by the complex, with formation of the [ (C5H5)2
ReH2] + ion, is then assumed to go into the other cone.
For the saline mono-, di- and tricyclopentadienyls [e.g., KCaHa,
Mg(C5H5)2,and Bi(C5Hjj3] it is probable that the metal atom occupies
a central position with respect t o the rings and that these are symmetrically arranged; this has already been found t o be so for the magnesium compound.
A complicated structural problem arises in the case of the compound
C O ~ ( C ~ (39).
H ~ )It~ may be assumed that there is one ring between the
cobalt atoms; of the two outer rings, one may be 0-bonded to the metal
and the other attached centrally with r-electrons. I n accordance with the
zero dipole moment, the two like rings should occupy trans positions
relative to one another. An earlier structure, in which all three rings were
placed centrally with respect to the two metal atoms, seems less likely.
For mixed compounds of the type CsHaMn(C0)3 or (C6H5)NiN0 the
structure should certainly be one having a symmetrical geometrical arrangement. The location of the hydrogen atoms in the cyclopentadienyl
carbonyl hydrides of chromium, molybdenum, and tungsten presents a
93
problem, however ; a central position between the carbonyl groups seems

probable (86,118).
For dimeric compounds of the type C6HaMe(CO),&feC6H6 a cylindrical (triple-decker-sandwich) structure was originally proposed. According
to Wilson and Shoemaker ( H I ) , however, the compound [CSHBMO
(CO) 312 possesses a completely different structure: each molybdenum
atom lies at the center of a cube, one side of which is occupied by the
five-membered ring; the three carbonyl groups are directed towards three
of the corners of the opposite side of the cube; and the metal-metal bond,
which forms the only link between the two halves of the molecule, lies in
the direction of the remaining corner. In the compound (CaH,) 3Cr2(C0)3
there is a major transition to the ionic structure [Cr(CaHa)2]+ [Cr(CbHB)
(CO)s]- (52).In this way the anion acquires the krypton configuration,
while the cation, by likewise including all of the n-electrons of the ring,
should have three unpaired electrons, a conclusion in keeping with the
results of magnetic measurements.
In the rhenium compound (C5H6)zReH(CO)z,one of the cyclopentadienyl rings is assumed to be a-bonded, and the other r-bonded to
the rhenium atom. The hydrogen atom should then, as in the case of
dicyclopentadienyl rhenium hydride, (C5H5)2ReH (118), lie in the cone
formed by the metal and the centrally bonded ring (82).For dicyclopentadienyl iron dicarbonyl, (c6HS)2Fe(CO)2, a very similar structure
is proposed in which the ring that is d-bonded and that which is r-bonded
to the metal are inclined to one another, and both lie on the side of the
metal atom opposite the two carbonyl groups (97). For the alkyl and
aryl derivatives of cyclopentadienyl metal carbonyls of chromium,
molybdenum, tungsten, and iron (large numbers of which are known) , as
well as for some similar derivatives of the compound C5H6Cr(NO)2C1, a
metal-carbon a-bond specially stabilized by the five-membered ring has
been assumed (97,153).
2. Bonding
The nature of the bonding in the metal cyclopentadienyls and their

derivatives is still the subject of much discussion, although the compounds
have been known for some years. It is now quite certain that in the
saltlike alkali, alkaline earth, and rare-earth metal cyclopentadienyls the
bonding between the ring and the metal is essentially ionic. Compounds
such as dicyclopentadienyl vanadium and chromium also show hydrolytic
cleavage. On the other hand, however, cyclopentadienyl thallium, for
which the recent investigations of Cotton and Reynolds (20) suggest an
ionic bond, may be sublimed and is precipitated from an aqueous medium
94
as an insoluble product. In our judgment this clearly indicates an essentially covalent bond. I n the fifth main group of the periodic table we find
in antimony and bismuth elements which tend towards an intermediate
type between purely ionic and covalent ferrocenelike bonding; zinc and
mercury are probably similar. For the compounds Sn(CsHa)2 and
Pb(C5H,) 2 x-complex bonding must probably be assumed, a t least in part,
while for mixed cyclopentadienyl compounds of silicon normal v-covalent
bonding has been proposed (89, 615).
There arc, in essence, two concepts of the type of bonding in the
transition metal dicyclopentadienyls. Both are based on a sterically
central position of the metal with respect to the rings. It is well known
that true organometallic derivatives of the transition metals, in which the
metal and carbon each provide one electron towards the formation of a
single bond, are (except in the case of elements of the second subgroup)
extraordinarily labile, if indeed they can be made a t all. In the first place,
such a bond may be stabilized by hybridization of the d-orbitals (which
by themselves are not suitable for bonding) with the s- and p-orbitals.
The coordinative covalencies formed by electron pairs then bring about
Stabilization of metal-carbon bonds, e.g., in carbonyls, cyanides, isocyanides, acetylides, and fulminates. The decisive factors in this process
are the filling, or approximate filling, of the orbitals of the metal to the
next inert gas configuration, and the double-bonding component of the
metal-carbon bond.
After the elucidation of the real molecular structure of ferrocene, it
was at first generally assumed that the metal attained the inert gas
structure (70, ,219).Wilkinson, Woodward, and their co-workers (219)
associated five r-electrons with each five-membered ring, and these,
together Kith the eight electrons of the central F e ( 0 ) atom, built up the
electron shell of the latter to eighteen. According to our concept (70)
ferrocene is a completely new type of penetration complex. The two
parallel staggered aromatic cyclopentadienyl anions each possess three
pairs of r-electrons, and the central Fe++ ion interacts with each pair,
forming six coordinate covalencies, distributed around it in a trigonal
antiprism, i.e., almost octahedrally arranged. Figure 6 shows this concept
clearly; according to it, the central F e + + ion retains its characteristic
coordination number of six.
'l'hcorctical studies by E. Ruch (173-177) proved the correctness
of this concept of the bonding, with which the retention of the aromatic
character of the ring is in harmony. The nature of the bonding in the
dicyclopentadienyl metal compounds was discussed by him in terms of
molecular and equivalent orbitals, and on the basis of calculated overlap
integrals for ferrocene an energy-level diagram was suggested which
95
FIQ. 6. The electronic configuration of ferrocene.
indicates the orbitals occupied (denoted U, W , or 6 according to their

symmetry) in, for example, compounds of titanium, vanadium, niobium,
chromium, molybdenum, iron, cobalt, rhodium, and nickel, shown in
Fig. 7. There are two other proposals for the bonding in ferrocene which,
although differently developed, also lead t o the krypton configuration
(110,186).
Against this view are comprehensive theoretical studies which put only
a single covalent d,-p, bond between the metal and the ring in all the
transition metal compounds (23,26, 32, 119, 131). On this basis, ferrocene, for example, should show only a closed 3d shell. Four r-electrons
would then remain in the rings, which could be considered as radicals,
without taking part in the bonding.
Experimental results, such as those obtained by X-ray absorption
edge measurements, lead us to prefer the penetration complex concept,
which ixivolves the characteristic tendency of transition metals to complete their electron shells and assume their typical coordination numbers
(41).
In the course of our own work, which had as its object the further
96

18-electron shell
confirmation of our idea of the bonding, we were able to show, among

other things, that diindenyl compounds may be hydrogenated in the
six-membered ring, but that only four hydrogen atoms per ring were
absorbed ( 1 8 5 ) . The T electrons of the third double bond of the benzene
ring are thus drawn into the five-membered ring to form the complex. Our
concept of the type of bonding also receives important support from the
surprising reaction for the preparation of cyclopentadienyl thallium from
aqueous solution, as well as from the isolation of cyclopentadienyl indium.
In these cases the assumption that the three pairs of r-electrons belonging
to the cyclopentadienyl anion are attracted into the free 5 p or 6p orbitals
of In(1) or Tl(I) also leads to the attainment of the xenon or radon
f towards ozone
configuration. The stability of the cation [Co(CaHa)a]
in acetic acid again, in our opinion, proves that all the a-electrons are
drawn upon t o form the metal complex (41). The gradation in complex
stability is consistent with our theory and is, indeed, to be expected.
Maximum stability must occur in the uncharged type of compound having
an inert gas configuration. This is in fact the case for ferrocene, ruthenocene, and osmocene, and for cations such as [Co(CsH5)2]
+ which are
COMPOUNDS O F AROMATIC RING SPSTEMS AND METALS
97
isoelectronic with them. Decrease in the extent to which the metal

electron shells are filled is paralleled by an increase in sensitivity to
hydrolysis, as, for example, in the case of the dicyclopentadienyl compounds of chromium and vanadium.
For the almost invariably diamagnetic mixed complexes of transition
metals with cyclopentadienyl and other ligands, such as carbon monoxide
and nitric oxide, completion of the inert gas configuration is the controlling factor to an even greater extent (41). It is characteristic that in
this case maximum stability is encountered in the monomeric cyclopentadienyl metal carbonyls [for example, C6H5Mn(CO)3] , corresponding to the preferred coordination number of six for the metal. In keeping
with this idea, particularly for the last compound, the presence of a shell
of 18 electrons has been concluded on theoretical grounds (122, 126).
Linnett (1.22) showed that the two theories are not incompatible, for he
also established, from consideration of molecular and equivalent orbitals,
that the attainment of the inert gas configuration is essential.
H. TECHNICAL
APPLICATIONS
OF CYCLOPENTADIENYL
METALCOMPLEXES
Metal cyclopentadienyls have already proved to be of considerable
technical interest. According to investigations carried out mainly in the
United States, ferrocene is able to bring about smokeless combustion of
fuel oil when added in amounts of 0.05% (1).Its use as an oil additive for
jet motors and as an antiknock has also been proposed. Liquid ferrocene is
said to have advantages as a thermally stable heat-transfer medium.
Dicyclopentadienyl nickel also has possible applications, and dicyclopentadienyl titanium-dichloride has been studied in association with
aluminum alkyls for the polymerization of ethylene (15, 133, 134). Finally, the carbonyl CH3C5H4Mn(C0)3has been considered as a fuel
additive in conjunction with tetraethyl lead (2). Other information is
given in the patent literature ( 4 2 ) .
111. Complex Formation by Six-Membered Ring Systems
A. INTRODUCTION
Shortly after the discovery of ferrocene and the elucidation of its
molecular configuration, one of us (70)put forward a suggestion for the
nature of the bonding in this compound, according to which ferrocene
was a penetration complex, the essential feature being the filling of the
electron shells of the central atom to the krypton configuration by the
three pairs of *-electrons of the aromatic ring. On this hypothesis it should
be immaterial, a t least as far as the bonding is concerned, whether the
rings are charged or uncharged, provided that the condition of the
98
availability of the r-electron sextet is fulfilled. Since the same view was
put forward by E. Ruch from the theoretical standpoint, experiments
aimed a t the preparation of a dibenzene chromium (0)seemed to be not
without some prospect of success. The investigation was made with
chromium since in this case it was hoped that the addition of two rings
would give the inert gas configuration, and with it maximum stability. The
problem then became experimental in character.
In 1954 a new synthesis for chromium carbonyl was developed in
collaboration with IT. Hafner. The system CrCl3/A1Cl3/A1/CaHs/CO
used in this case &as found, when the reaction was more closely studied,
to open the way to the compound C r ( c ~ H 6 ) (t 5 3 ) .It was shown that in
an autoclave a t 180, without the carbon monoxide, a mixture resulted
which contained the [Cr (C6He)2]i- cation, produced as foliows:
3CrCL
+ 2.M + AlCI, + 6CsHs-+ 3[Cr(CJTs)?](A1Cl,).
(11)
The cation could be isolated by precipitation as the picrate or the

reineckate. By reduction with sodium dithionite i t could be converted
directly into brown-black dibeneene chromium, Cr (CeHs)2, which can
be sublimed (53,54):
2[Cr(COH&+
+ S&- + 40H-
-+
2Cr(CgH&
+ 2H20 + 2S03.
(12)
This process is analogous to the preparation of ferrocene from ferricinium

ions. It was soon possible to extend the work to the preparation of a
N
23, -
TI
..
1 L"V
UCo3,U
V%.,
00Ar2
Tfe!4r3,1"
fco%rzlJ
fVJ&J
L0JAr21'
4ur
&4
dAr*
- -nHI
VlIl
M
z(r
cspd
t a u
dRu%r>l"
CRhJJh,>l"
fdArZl'
a a
-w
74
nRc
a r
Wolr,
I~rJlrll'
To~'J&~J"
DJJ'A$*'
CWhJ
FIQ.8. Complexes of uncharged aromatic compounds with transition metals
large number of aromatic metal complexes containing two six-membered

rings, with a wide variety of both metals and aromatic ligands. Figure 8
shows the metals which behaire in this way.
B. PREPARATIVE
METHODS
The synthesis from metal chloride, aluminum chloride, the aromatic
compound, and aluminum powder (in those cases where reduction was
99
necessary), which was developed in the case of chromium, proved to be

extraordinarily successful and widely applicable (53).In the course of a
detailed study of the preparation of dibenzene chromium it was soon
found possible to introduce important simplifications. Although initially
an autoclave was essential in order to reach the necessary reaction
temperature of 140,it was found that addition of catalytic amounts of
mesitylene enabled the synthesis to be carried out simply under reflux.
At an earlier timc, by adding an eightfold excess of aluminum chloride
[calculated on the basis of Eq. (ll)]it was possible to attain an almost
quantitative yield ( 7 5 ) . Alkyl and aryl homologues of benzene, on the
other hand, reacted under reflux from the start to form the corresponding
chromium complexes.
A further synthesis of metal complexes containing two six-membered
aromatic rings, which was limited to chromium, was based on the interaction of phenyl magnesium bromide and chromic chloride, and subsequent hydrolysis of the reaction mixture to the [Cr (CeHs) 23 + cation
(107, 227, 228).
C. COMPLEXESCONTAINING
Two SIX-MEMBERED
RINGS
1. Fifth Subgroup
In this group it was possible after a long series of trials to obtain the
yellow cation [V(C6H6)21 + from the system VC1~/AlCl~/Al/C&6, this
time, surprisingly, only under reflux; unlike its chromium analogue, the
[V (CsHs) 2] + ion is unstable to hydrolysis. However, just as in the case
of the molybdenum and tungsten compounds, red-brown dibenzene
vanadium, V(C6H6)2, can be obtained by a rapid hydrolytic disproportionation. In the form of large crystals it is black; it sublimes a t
120-150",tends to sinter a t about 120,melts sharply a t 277-278", and
decomposes above 330': It is instantly oxidized in air, but gives redbrown solutions in organic solvents if air is excluded. It is only
slightly soluble in methanol. With water which is saturated with
nitrogen there is neither dissolution nor hydrolysis, but on allowing
access of air the complex is oxidized without formation of a stable
cation (64).Red-brown dimesitylene vanadium, V[sym- (CHs) ~ C ~ H2, S ]
which is extremely sensitive to air, may likewise be obtained. Attempts to
reduce V (CaH6)2 to the [V (C&) 2] - anion by means of alkali metals in
liquid ammonia were unsuccessful.
2. Sixth Subgroup
Dibensene chromium is prepared by the methods which have already

been mentioned (63, 107, 227, 228). The compound may be sublimed in a
100
E. 0. FISCHER AND H. P. FRJTZ
high vacuum at 150, and deposits as brown-black crystals of mp 284285'. It is only moderately soluble in organic solvents such as benzene
(16 mg/ml) , forming yellow or, a t higher concentrations, brown, solutions.

Its solubility in ether is lower; it is relatively rapidly oxidized in air.
Water neither dissolves nor attacks the compound. Thermal decomposition sets in a t about 3OO0,with formation of a metallic mirror.
The uncharged compound is the best starting point for the preparation
of salts, since it may easily be oxidized by atmospheric oxygen in a twophase system comprising water and an organic solvent. The iodide, tetraphenylborate, reineckate, picrate, and perchlorate are typical insoluble
salts. Various compounds of the cation are sensitive to light. Neutral or
weakly alkaline solutions may be kept for weeks in air without appreciable decomposition, but acidic solutions soon decompose. The
perchlorate and similar compounds with oxidizing anions are explosive
in the dry state. Instead of benzene, many of its alkyl and aryl hoinologues, such as toluene, o-, m -, and p-xylenes, mesitylene, pseudocumene,
hexamethylbenzene, diphenyl, and tetralin, may be used as ligands in the
complex. In all these cases the aluminum chloride method may be employed.
Dibenzoic acid chromium is reported as produced when the mixture
formed in the preparation of dibenzene chromium by the Grignard
reaction is treated with carbon dioxide (29'7). Substitution reactions for
the hydrogen atoms in the rings of the complex, such as are known to
occur readily for ferrocene, do not occur in this case (90).
Complex formation for molybdenum, the first homologue of chromium,
was brought about by means of the aluminum chloride method (79). Redbrown [MO(CsHs)z]+ is formed, and is then reduced to the green uncharged compound, Mo(CsH6)2, by sodium dithionite. A more elegant
method is the recently discovered disproportionation in an alkaline
medium (64),which occurs according to the equation
6[Mo(CnHe)2]"
+ SOH-+
5Mo(CeH~.)s MOO,-
+ 4Hz0 + 2CeHe.
(13)
Dibenzene molybdenum forms glistening green crystals which sublime

at 90-100",decompose without melting a t about 115", with separation of
a metallic mirror, and are readily soluble in organic liquids. I n keeping
with its much greater sensitivity to oxidation than the chromium compound, dibenzene molybdenum ignites in air. Uncharged dimesitylene
molybdenum and additional salts of the cation have also been characterized (79, 187).
The preparation of uncharged yellow-green dibenzene tungsten,
W(CsH@),was recently carried out by using a similar disproportionation
with the mixture resulting from the interaction of tungsten hexachloride,
101
aluminum chloride, aluminum, and benzene (187).This compound, which

is much more sensitive to oxidation than dibenzene molybdenum, sublimes
in a high vacuum at 50" and decomposes a t 110". In this case the orangeyellow iodide of the cation may also be prepared.
3. Seventh Subgroup
Although attempts made so far to prepare complexes of the type

Mn (aromatic) or [Mn (aromatic)2 ] + have been unsuccessful, it has been
possible to isolate a bright yellow tetraphenylborate of the [Re(CsHc)a] -Iion (as fine needles) and a red-brown reineckate of the [Re (sym- (CH3)3
CeH3) 23 + ion by the aluminum chloride method. The free bases are stable,
but attempts a t reduction t o the uncharged complex of the element were
unsuccessful (83).
4. Eighth Group
I n this case complex formation was first observed for iron. It was
necessary, however, to use mesitylene in place of benzene as a ligand
more resistant to hydrolytic fission. The orange-red [Fe (sym- (CH3)
CeH3)2]+ ion, which is relatively stable in weakly acidic media, was
obtained and characterized as the fire-red iodide, the orange tetraphenylborate and the light red reineckate. Tetralin and alkyl benzenes
may likewise enter into complex formation (4).In the case of the
homologues ruthenium and osmium, the light yellow cations [Ru(sym(CH3)8C6H6)23 + + and [0 s (sym- (CH3) 3CaH3) -I- + were found in the
same way. They are considerably more stable than the iron complex, and
the same increase in stability is found in the series cobalt, rhodium, and
iridium. The red-brown [Co (sym- (CH3)3C6H3)2] 3 + ion is considerably
more labile than its lighter-colored rhodium analogue, and the light
brownish-yellow iridium cation is the most stable. All three of these
complex ions may be isolated as reineckates, and that of rhodium also as
the iodide (84). In the case of platinum(1V) experiments carried out up
to the present lead to the hope that the corresponding complex will be
isolated.
D. MIXEDCOMPLEX=CONTAINING
SIX-MEMBERED
RINGS
1. Carbonyls
The preparation of a compound of this type, C6H&r(CO)3, was first

achieved by the interaction of chromium carbonyl and dibenzene chromium in benzene in an autoclave (67). This substance was also obtained
more simply by the reaction (68)
Cr(CO)s
+ CsHs
--f
CsH6Cr(CO)r
+ 3CO.
(14)
102
E. 0. FISCHER AND H. P. F R I T Z
The compound forms yellow crystals (mp 162-163') which may be sublimed and which dissolve in the usual organic solvents. It is diamagnetic
and, in contrast to dibenzene chromium, is stable in air. This reaction,
which opened up fundamentally new possibilities for the addition of sixmembered aromatic rings to transition metals, soon proved to be extraordinarily widely applicable. Thus, by means of it, yellow molybdenum
and tungsten compounds CoHsMo(CO) and CGH6W(CO) 3 (decomposition temperatures 120-125" and 140-145", respectively) were prepared directly. In place of benzcnc itself, a great number of its alkyl
derivatives can take part in complex formation. Mention may be made
of the compounds sym- (CH,) 3 C G H 3 M(CO)
~
(decomposition temperature 130-140), sym- (CH3)3CeH3W(CO) (mp 160"), and the exceptionally stable (CH3)&&r(CO) :%(mp 232"). The compound CH3C6H5
Cr(CO)3 (mp 80"),which is to be rcgnrdcd as isoclectronic with the
cyclopentadienyl derivative CH3CjH4Mn(CO):%,is, in surprising contrast
to the latter, no longer liquid. The isomers 0- (CHJ 2C6H4Cr(CO) (mp
88-90'), m- (CH3)2C6H4Cr(CO) (mp 104-105"), and p - (CH,) 2C6H4
Cr(CO13 (mp 97-98') differ only slightly in stability.
Surprisingly, it was found possible by this method to prepare complexes of benzene derivatives which had not previously given diaromatic
complexes by the aluminum chloride method. Thus, by decomposing chromium earbonyl with chlorobenzene a quite stable yellow compound of
formula ClCsH5Cr(CO)3 (nip 96-98") Kas obtained; with aniline the
protiuct was the basic yellow substance (H2NCoH5Cr(CO), (mp 161") ;
with phenol, acidic water-soluble yellow HOC6H5Cr (CO) (mp 114') ;
wit11 n ~ , : h y lbenzoate, orangc CH30COCoH6Cr(CO) (mp 93-95") ; and
with anisole, the compound CH30C6H5Cr(CO) (mp 80-82'). Naphthalene. thc simplest condensed six-membered ring compound, reacts with
chroniium carbonyl to give the complex CloHsCr(CO) (which decomposes a t 150-160'). All of the complexes mentioned are readily soluble in
organic media and, in the solid state, are relatively stable in air. For the
neighboring elements of the seventh subgroup only the cation [sym(CH3)3C6H3Mn(C0)3]
-t has been described (19) ; it is isolated as the
cream-colored iodide after reaction of the carbonyl bromide Mn (CO)J3r
with boiling mesitylene in the presence of aluminum chloride.
2. Mixed Fire- and Six-Membered Aromatic Ring-Metal Complexes
After the green complex [ c r ( C s H a ) ~-k] [CsH&r(C0)3]- (which can

be sublimed) had been obtained by precipitation (65), showing aromatic
five- and six-membered ring systems as ligands on two different metal
atoms, the combination of two such ligands with the same metal atom
was achieved only quite recently. The compound CH3C5H4MnCsHs was
COMPOUNDS O F AROMATIC RING SYSTEMS AND METALS
103
obtained, as ruby-red crystals (mp 116-118") which could be sublimed

and which were relatively stable in air, by interaction of phenyl
magnesium bromide and methylcyclopentadienyl manganese chloride or
dicyclopentadienyl manganese in tetrahydrofuran (19). Another uncharged complex, C5H5CrCaHc (mp 227-229' with decomposition), was
obtained as a volatile orange solid by the reaction between chromic
chloride and equivalent quantities of phenyl magnesium bromide and
cyclopentadienyl magnesium bromide (85). I n Group VIII, the cation
[ C5H5FeCaH3(CH3)3] + was obtained by treatment of cyclopentadienyl
iron dicarbonyl chloride, C5H6Fe(CO)&1, with boiling mesitylene and
aluminum chloride, and was isolated as the ivory-colored iodide (19).
E. PHYSICOCHEMICAL
INVESTIGATIONS
X-ray work was particularly important in settling the problem of the
structure of dibenzene chromium, and showed that this compound, unlike
the dicyclopentadienyl metal compounds studied earlier, possessed a
FIG.9. Model of the structure of dibenzene chromium.

cubic lattice with the same space group as carbon dioxide, Pa3-The.The
unit cell contained four molecules, and had a side of 9.67 rfi. 0.03 A: the
volume occupied by one molecule was found to be 226 A3. The hexagonal
prismatic structure of dibenzene chromium is shown clearly in Fig. 9.
104
Distances in the centrosymmetric molecule have been determined provisionally as C-C = 1.38 0.05 A and Cr-C = 2.19 i- 0.1 A ($001) Dibenzene vanadium and dibenzene molybdenum do not have a cubic lattice.
V (CsH6) crystallizes like the dicyclopentadienyl metal complexes in the
monoclinic system with the space group Czr5 and a = 6.27 A, b = 7.74 A,
e = 11.06 A, p = 119.3' (197).
A polarographic study of the Cr ( CaHa)2- [Cr (CaHe)2] + transition has
been made (91I . The dipole moments of the compounds Cr (C6H6) (196),
Cr(sym-(CH3)3C6H3) (196), and, surprisingly, C5H5CrsH6 (74) are
all zero.
Magnetic measurements have been made on a large number of the
new six-membered ring complexes with a view to elucidating the nature
of the bonding. The values found for the metals forming uncharged
complexes or cations are shown, together with their oxidation states, in
Table VII. It is striking to note how the formation of complexes is
TABLE VII
MAGNETIC
MOMEKTSOF COMPLEXES
CONTAINING
Two SIX-KEMBERED
RINGS
always controlled so as to lead to diamagnetism and the closest possible

approach to the inert gas configuration, showing that all the w-electrons
are being drawn into the electronic shell of the metal. The hypothesis of
the bonding of the aromatic system by threefold coordination again
seems to be the best explanation. Isoelectronic series may be constructed
on the basis of these relationships; they also give a survey of the whole
system and of the possibilities of its extension, e.g., W(CaH6) 2, [Re (sym(CH3)3C6H3)2]+, [Os(sym-(CH3)3C~H3)2]++,
and [Ir(sym-(CHds
Co&)aI3+-
105
It is also characteristic that the diversity of the oxidation states exhibited, in contrast to those in the corresponding five-membered ring
complexes, is very limited. This must be attributed to generally weaker
bonding of the uncharged aromatic ligand relative to the charged one,
for which there is additional electrostatic attraction between the ring and
the metal ion.
Just as in the case of cyclopentadienyl metal complexes, six-membered
ring aromatic complexes have infrared absorption bands in characteristic
regions. On the basis of compounds examined up to the present time, unsubstituted six-membered rings bound in complexes are characterized by
intense bands in five frequency ranges in the rock salt prism region ( 4 5 ) .
These, given below, are also valid for mixed complexes such as the compounds CaH5CrC6H6and C6H6Cr(C0)3:(1) a C-H stretching frequency
a t 3010-3060 cm-l (ycH) ; (2) a C-C stretching frequency at 1410-1430
cm-l ( V W ) ; (3) a C-C stretching frequency a t 1120-1140 cm-l (vcC);
(4) two or three C-H deformation frequencies at 955-1000 cm-l ( 6 ~ ;~ )
(5) one or two C-H deformation frequencies at 740-790 cm-l (Aca).
From recent investigations, including those on Cr ( C6D6)2, the number of
intense infrared absorption bands for dibenzene chromium leads to the
conclusion that the compound has no sixfold symmetry, but very prob~
These results provide convincing
ably the symmetry group D S C(187).
evidence for our view of the structure, with approximately octahedrally
distributed metal-carbon bonds and perhaps weakly puckered rings which
certainly exhibit threefold symmetry. On geometrical grounds, an analogous variation of symmetry in cyclopentadienyl rings in a complex is
impossible. Measurements of electron spin resonance in the cation
[Cr(CeH6),] + have been made (36,36): they show that the one unpaired electron in the ion is distributed over the whole complex.
The enhanced tendency to form only complexes in which an inert
gas configuration is approached more closely comes to light in an impressive way for the mixed six-membered ring-metal complexes. Without
exception, the aromatic hydrocarbon-metal carbonyls of the sixth subgroup elements are diamagnetic; so, too, are the complexes of formulae
[ (CH3)3C6H3Mn(CO)3] +, C H B C S H ~ M ~ C ~and
H ~ , [C6HsFeC6H3
(CH3)31 +. The absorption of all the r-electrons is therefore established
for these cases, too. The fact that a shell of thirty-five electrons provides
sufficient stability for the formation of a r-complex (as in diaromatic
complexes such as those of Cr(I), Mo(I), or V(O), which have already
been mentioned) is shown by the existence of the compound C6H&rC&,
the only instance of its kind known at present. This complex has a
moment of 1.70 Bohr magnetons, corresponding to a single unpaired
electron,
106
F. STRUCTURE^

AKD
BONDING
On the basis of chemical properties and physicochemical investigations

dibenzene chromium is considered t o have the same spatial distribution
of bonds as ferrocene. X-ray studies have shown that the molecule is
ccntrosymmetric, and that all the C-C and Cr-C distances are practically equal. One may therefore decide with certainty on the double
cone structure (see Fig. 9 ) . As already mentioned, it may be concluded
that the metal-carbon bonds have an approximately octahedral arrangement and that the benzene rings have threefold symmetry.
It is not yet possible to say whether the rings exhibit free rotation.
From the zero dipole moment of ditoluene chromium (2001), it can be
concluded that under certain circumstances there is unhindered rotation.
Electron-diffraction studies a t various temperatures might throw light
on this point. I n keeping with the theory of bonding in ferrocene, we
assume a central chromium atom of zero oxidation state; it attains the
krypton configuration by accepting coordinate links from the three pairs
of r-electrons of each aromatic ring. This accords with the diamagnetism
of the compound, since otherwise the chromium in zero oxidation state
would have six unpaired electrons. The existence of the complex cation,
which is exactly analogous to the dichroic red-blue ferricinium ion
(which also contains one unpaired electron), shows that an electron can
be removed from the three doubly-occupied 3d-orbitals that are not involved in ring-metal bonding; this is just like the ferrocyanide to ferricyanide ion transformation ( 5 3 ) . This view of the bonding is also
applicable without difficulty to the diaromatic complexes of other transition metals, as well as to the mixed types of compound. A molecular
orbital theory of the covalent bonding between chromium and the two
benzene rings, based on that for fcrrocene, has also been developed (119).
G. THEELUCIDATIOS OF THE CONSTITUTION OF THE

CHROMIUM
PHESYL
COMPOUNDS
All of the developments so far described are specially significant in
relation to the chromium phenyl compounds obtained by interaction of
chromic chloride and phenyl magnesium bromide, which were investigated
by F. Hein after 1919. These were originally considered to be true
organometallic compounds, possessing chromium-carbon u-bonds, of
general composition Cr(CaH5), (71. = 3 or 4 ) , and Cr(C6H3)*X (n= 3, 4,
or 5 ) , where X represents anions such as iodide and anthranilate. J.
Zernike in 1939 first considered them to be some sort of hypercomplexes,
and for the saltlike compounds the chromium was assumed to be in the
unipositive state (230).In 1953 a ferrocene-like structure was discussed,
107
though an incorrect oxidation state was attributed to the chromium ( 9 2 ) .

At about this time H. H. Zeiss had begun a critical repetition of Heins
work and had confirmed his analytical results completely (2%). On the
basis of these, L. Onsager and Zeiss proposed a basic structure for the
compounds which was in accord with all the experimental results (2.86,
229).
I n the chromium (I) salts, the metal ion should lie between two rings
in a double cone structure and have a shell of 35 electrons, including all
the r-electrons of both rings. According t o this view, Cr (CeH6)8 may be
written as CeH&r (CS& * C&K) , cr (CeHK)4 as Cr (C6H6 /CI&) 2,
and Cr(CeHa)60Has a phenate of the [Cr(Ca& * CeHs)z]+ cation.
This idea received strong support from hydrogenation experiments with
lithium aluminum deuteride, and from the recovery of deuterated benzene
and diphenyl on reduction of the complex which results in the Grignard
reaction with lithium aluminum deuteride, though not on treatment with
lithium aluminum hydride and working up with deuterium oxide (228).
In 1955 the entirely independent synthesis of dibenzene chromium and
the X-ray proof of its structure afforded the opportunity for an investigation in collaboration with Hein. The compound Cr(CsH5) 41 was prepared
by the Grignard reagent method, and the complex [Cr(CsHs CsHj)2]I
was obtained from the system CrCl~/A1C13/Al/C6H6 CeHK and precipitation with iodide ion. The identity of these products, shown by comparison of their infrared and ultraviolet absorption spectra, X-ray absorption edges, and magnetic properties, provided unambiguous proof of the
aromatic-complex nature of the chromium phenyl compounds (78, 103).
In a similar way, the compound Cr(C&6)s was later shown t o have the
constitution CaHeCrCaHa C6H6 (104). An old problem in the chemistry
of organometallic compounds was thus solved. The presence of the
[Cr(C6H6)2]+ ion was also shown by separating the products of the
Grignard reaction on a chromatographic column (106, 107).The mechanism of formation of the complex by the Grignard method (.8.87,298)
, and
the thermal degradation of triphenyl chromium (106)have also been
studied. This work provides further support for the r-complex theory.
IV. The Existence of Complexes of Seven-Membered Aromatic Systems
Now that complex formation with suitable metals has been established
for anionic aromatic (CaH6)- rings and for uncharged aromatic compounds like benzene, the question arises as to whether the aromatic
(C7H7)+ cation, which has become known in recent years, can function
similarly. We have made investigations with trcpylium bromide, (C7H7)+
Br-, obtained from benzene by expansion of the ring with diazomethane,
108
E. 0. FISCHEB AND H. P. FRITZ
bromination of the resulting tropylidene and abstraction of hydrogen

bromide, according to the method of E. Doehring (45).
Carbonyls such as those of chromium, molybdenum, or iron were first
allowed to react with tropylium bromide, but, unlike those with benzene
or cyclopentadiene, these experiments were unsuccessful, even when an
autoclave was used. Attempts were then made to obtain a sevenmembered ring complex by the thermal decomposition of tropylium saits
of carbonyl hydrides, but it was found that the great sensitivity of
tropylium bromide to reduction, coupled with the strong reducing power
of the carbonyl hydrides, militated against success. Thus, the interaction
of the sodium salt N a [C6H6Cr(CO),] and tropylium bromide, followed
by sublimation of the resulting green product, led, by the redox reaction
ZNa[CsHrCr(CO)J
+ 2C7H7Br+ [CsHsCr(CO)l]s+ C,HI, + ZNa Pr
to dimeric cyclopentadienyl chromium tricarbonyl and ditropyl instead of

proceeding according to the equation
[C~HSC~(CO)~]
(CrH7)--+ C6H,CrCt&
+ 3CO.
In compounds of the type Cr (CO) (NH,) , too, it is impossible to replace

the labile ammonia ligands by the tropylium ring. The method of Grignard reagent formation and simultaneous reaction with metal halides
(e.g., starting from cyclopentadiene or bromobenzene) was applied to
bromocycloheptatriene (obtained by ring expansion of bromobenzene)
and chromic chloride, but proved unsuccessful. The same was true of
attempts to expand the ring in dibenzene chromium by reaction with
diazomethane and ultraviolet irradiation.
The next step was to prepare suitable salts of the tropylium ion and
to try to obtain the complex from these. In order to demonstrate the
existence of the tropylium ion in solution, a precipitating agent was
sought, and was found in the tetraphenylborate ion, which in dilute
acetic acid solution gives a bright yellow precipitate of composition
(C7H7)[ B (c6H5)*I. Red-brown plates of composition (C7H7)H3
[Fe(CN) 61 are precipitated by potassium ferrocyanide from solutions
containing hydrobromic acid, and an analogous reaction occurs with
potassium ruthenocyanide. Using energetic thermal conditions, an attempt
was made to introduce the ring into a complex, but only decomposition
products resulted, and the cyanide ligands were not displaced. By interaction of tropylium bromide and cuprous bromide the diamagnetic
tropylium salt of the acid H(CuBr2) was obtained. Tropylium bromide
and Hz(PtBr6) in glacial acetic acid likewise gave only a saltlike orangered compound (C7H7)z(PtBr6).The reaction of C7Hson platinic bromide
or Hz(PtBra) surprisingly led to a diamagnetic compound of the com-
109
position C7HBPtBr2or CiH7PtBr2, which forms orange needles of mp

154.5-155, is almost insoluble in water as well as in usual organic
solvents and cannot be sublimed under usual conditions. Although some
of the chemical properties are similar to those of olefin complexes of the
type (olefin * PtBr2)2, the infrared spectrum of the compound clearly
shows the existence of an equilateral hydrocarbon ring, so that we assume
the structure (C7H7PtBr2)2.Because of the fact, that more than four
infrared (IR) active normal frequencies, as predicted for (C7H7)+ in
symmetry group DTh,can be found, we regard the original group D7h
as changed to symmetry group C7,,. The selection.rules of C7,,predict seven
IR-active fundamental frequencies for (C7H7)+. The absorption bands
typical for (C7H7)+ in (C7H7)Brappear at nearly the same position in
the spectrum of (C7H7PtBr2)2.Just as in a likewise orange-colored compound of the approximate composition C7H7PtBr3, the assumption of
d2sp3-configurated metal atoms would lead to genuine r-complexes of the
aromatic seven-membered ring, the possible existence of which was discussed earlier (174). Our work is still in progress, but some results will
be published soon (46).
Very interesting in connection with the above-mentioned work is the
recently reported (810) preparation of a compound C7HBMo(CO)3.
Formation of a seven-membered ring complex with metals in a negative
oxidation state, e.g. Re- in the compound KRe, is opposed by the relatively strong oxidizing action of the tropylium ion. In tropylium compounds of main group elements, such as S(C7H7)2, the rings do not (according to Doering) form a sandwich complex but are bonded normally,
as in a thioether.
An important contribution to the problem of *-electron bonding with
seven-membered rings is probably made by measurements of the chemical
shifts of ring protons as studied by the nuclear magnetic resonance
method. It has recently been shown (117) that the displacement of the
electron density of the ring protons increases along the series (CaH5)-,
C6H6, (C7H7)+. Perhaps this expresses the decreasing tendency to enter
stable *-bonds.
REFERENCES
1 . Anonymous, Chem. Week 76, 54 (19%); Ind. Eng. Ghem. 47, 13A (1955).
2. Anonymous, Chem. Eng. News 3 5 1 9 (1957).
3. Arimoto, F. S.,and Haven, A. C., J . Am. Chem. SOC.77,6295 (1955).
4. Barber, W. A., J . Znorg. & Nuclear Chem. 4,373 (1957).
6 . Benkeser, R. A., Goggin, D., and Schroll, G., J . Am. Chem. Soc. 76, 4025 (1954).
6. Benson, R. E., and Lindsey, R. V., Jr., J . Am. Chem. SOC.79, 5471 (1957).
7. Berndt, A., and Marsh,R. E., Chem. and Chem. Eng. at Caltech. Inst., Pasadena, California, Rept. 1958/57, No. 76.
110
8. Berndt, A., and Schomaker, V., Chem. and Chem. Eng. a t Caltech., Rept.
1956/57, No. 57.
9. Birmingham, J. M., and Willrinson, G., J. Am. Chem. SOC.78,42 (1956).
10. Birmingham, J. M., Seyferth, D., and Wilkinson, G., J. Am. Chem. SOC.76,
4179 (1954).
11. Boke, K., 2.physik. Chem. (Frankfurt)B.1 1 4 4 5 (1957).
12. Boke, K., 2.physik. Chem. (Frankfurt) S.1 10,59 (1957).
13. Boke, K., 2.physik. Chem. (Frankfurt) CNS.1 11,326 (1957).
14. Brand, J. C. D., and Sneeden, W., Trans. Faraday Soc. 53,894 (1957).
16. Breslow, D. S., and Newburg, N. R., J. Am. Chem.Soc. 79,5072 (1957).
16. Broadhead, G . D., PhD. Thesis, University of Sheffield, ShefEeld, England, 1956.
17. Broadhead, G. D., and Pauson, P. L., J. Chem. SOC.1954,367.
18. Broadhead, G. D., Osgerby, J. M., and Pauson, P. L., Chem. & Znd. 1957, 209.
19. Coffield, T. H., Sandel, V., and Closson, R. D., J. Am. Chem. Soc. 79, 5826
(1957).
20. Cotton, F. A,, and Reynolds, L. T., J. Am. Chem. Soe. 80,269 (1958).
$1. Cotton, F. A., and Wilkinson, G., J. Am. Chem. h e . 74,5764 (1952).
22, Cotton, F. A,, and Wilkinson, G., 2.Naturforsch.9b, 417 (1954).
23. Cotton, F. A., and Wilkinson, G., 2.Naturforsch. 9b, 453 (1954).
24. Cotton, F. A., Liehr, A. D., and Wilkinson, G., J. Inorg. & Nuclear Chem. 1,
175 (1955).
26. Cotton, F. A., Whipple, R. O., and Wilkinson, G., J. Am. Chem. SOC. 75, 3586
(1953).
D. P., Macoll, A,, Nyholm, R. S., Sutton, L. E. R., and Orgel, L. E.,
J. Chem. SOC.1954,332.
27. DAns, J., Zimmer, H., and Brauchitsch, M.v., Chem. Ber. 88, 1507 (1955).
28. Doering, W. von E., and De Puy, C. H., J. Am. Chem. SOC.75, 5955 (1953).
29. Doyle, J. R., and Jonassen, H. B., J. Am. Chem. SOC.78,3965 (1956).
$0. Dunitz, J. D., and Orgel, L. E., Nature 171, 121 (1953).
31. Dunita, J. D., Orgel, L. E., and Rich, A., Acta Cryst. 9,373 (1956).
32. Dunitz, J. D., and Orgel, L. E., J. Chem. Phys. 23,954 (1955).
3.9. Eiland, P. F., and Pepinsky, R., J. Am. Chem. SOC.74,4971 (1952).
34. Engelmann, F., 2. Naturforsch. Bb, 775 (1953).
36. Elschner, B., and Herzog, S., 2.Naturforsch. 12a, 860 (1957).
36. Feltham, R. D., Sogo, P., and Calvin, M., J. Chem. Phys. 26, 1354 (1957).
57. Fischer, A. K., and Wilkinson, G., J. Inorg. & Nuclear Chem. 2, 149 (1956).
38. Fischer, E. O., unpublished work.
39. Fischer, E. O., Tagungsber. Chem. Ges. DDR. Akademieverlag, Berlin, 1956.
p. 132.
40. Fischer, E. O., 2. Naturforsch.9b, 619 (1954).
41. Fischer, E. O., Rec. trav. chim. 75,629 (1956).
48. Fiseher, E. O., 2. angew. Chem. 67,475 (1955).
43. Fischer, E. O., and Bottcher, R., 2.Naturforsch. lob, 600 (1955).
4.4. Fischer, E. O., and Bottcher, R., Chem. Ber. 89,2397 (1956).
46. Fischer, E. O., and Fritz, H. P., unpublished work.
48. Fischer, E. O., and Fritz, H. P., 2. physik. Chem. (Frankfurt) fN.S.1 17, 132
$6. Craig,
(1958).
47. Fischer, E. O., and Grubert, H., unpublished work.
48. Fischer, E. O., and Grubert, H., 2.Naturforsch. l l b , 423 (1956).
49. Fischer, E. O., and Grubert, H., 2. anorg. allgem. Chem. 286, 237 (1956).
111
60. Fischer, E. O., and Hafner, W., 2. Naturfmsch.Sb, 444 (1953).

51. Fischer, E.O., and Hafner, W., Z. Naturforsch.9b,503 (1952).
52. Fischer, E.O., and Hafner, W., 2. Naturforsch.lob, 140 (1955).
53. Fischer, E.O., and Hafner, W., 2. Naturforsch.lob,665 (1955).
54. Fischer, E.O., and Hafner, W., Z. anorg. allgem. Chem. 286,146 (1956).
55. Fischer, E.O., and Hofmann, H. P., unpublished work.
56. Fischer, E. O., and Hofmann, H. P., 2. angew. Chem. 69,639 (1957).
67. Fischer, E.O., and Jira, R., Z. Naturforsch. Sb, 1 (1953).

58. Fischer, E.O., and Jira, R., Z. Naturforsch.8b. 217 (1953).
69. Fischer, E.O., and Jira, R., 2. Naturforsch.8b,327 (1953).

60. Fischer, E.O., and Jira, R., 2. Naturforsch, 9b,618 (1954).
GI. Fischer, E.O., and Jira, R., 2. Naturforsch. lob,355 (1955).
6.2.Fischer, E. O., and IZogler, H. P., Diplomarbeit KBgler, Tech. Hochschule,
Munchen (1956).
63. Fischer, E. O., and Kogler, H. P., 2. angew. Chem. 68,462 (1950).
64. Fischer, E, O., and Kogler, H. P., Chem. Ber. 90,250 (1957).
66. Fischer, E. O., and Kogler, H. P., 2. Naturforsch. 13b, 197 (19%).
66. Fischer, E.O., and Leipfinger, H., 2. Naturforsch.lob,363 (29%).
67. Fischer, E.O., and Ofele, K., Chem. Ber. 90,2532 (1957).
68. Fischer, E.O., and Ofele, K., unpublished work.
69. Fischer, E. O., and Palm, C., unpublished work.
70. Fischer, E. O., and Pfab, W., 2.Naturforsch.7b,377 (1952).
71. Fischer, E.O., and Piesbergen, U., 2. Naturforsch. llb, 758 (1956).
78. Fischer, E.O., and Rausch, M., unpublished work.
73. Fischer, E. O., and Schreiner, S., 2. angew. Chem. 69,205 (1957).
74. Fischer, E.O., and Schreiner, S., unpublished work.
76. Fischer, E. O., and Seeholzer, Diplomarbeit, Tech. Hochschule, Munchen. (1957).
78. Fischer, E.O., and Seus, D., 2. Naturforsch.Sb,694 (1953).
7'7. Fischer, E.O., and Seus, D., 2. Naturforsch.9b,386 (1954).
78. Fischer, E.O., and Seus, D., Chem. Ber. 89, 1809 (1956).
79. Fischer, E.O., and Stahl, H. O., Chem. Ber. 89,1805 (1956).
80. Fischer, E.O., and Vigoureux, S., unpublished work.
81. Fischer, E. O., and Wirrmuller, A., unpublished work.
89. Fischer, E. O., and Wirrmuller, A., 2. Naturforsch. 12b,707 (1957).
83. Fischer, E.O., and Wirrmiiiler, A., Chem. Ber. 90,1725 (1957).
84. Fischer, E.O., and Zapf, W., unpublished work.
86. Fischer, E. O., Beckert, O., Hafner, W., and Stahl, H. O., 2.Naturforsch. lob,
598 (1955).
86. Fischer, E. O., Hafner, W., and Stahl, H. O., 2. anorg. allgem. Chem. 282, 47
(1955).
87. Fischer, E. O., Seus, D., and Jira, R., 2. Naturforsch.Sb, 692 (1953).
88. Friedman, L., Ira, A. P., and Wilkinson, G., J . Am. Chem. SOC.77, 3689 (1955).
89. Frisch, K.C., J . Am. Chem. SOC.75,6060 (1953).
90. Fritz, H.P., and Fischer, E. O., Z. Naturforsch.12b,67 (1957).
91. Furlani, C.,and Fischer, E. O., 2. Ekktrochem. 61, 481 (1957).
99. GaLeter, H.,Dissertation, University of Kiel (1953).
93. Gilman, H., and Gist, L. A., Jr., J . Org. Chem. 22, 250 (1957).
94. Graham, P.J., Lindsey, R. V., Jr., Parshall, G. W., Peterson, M. L., and Whitman, G. M., Contributioq No. 404, Chem. Dept., Expt. Sta., du Pont and Co.
112
P. J., Lindsey, R. V., Jr., Parshall, G. W., Peterson, M. L., and Whitman, G. M., J . Am. Chem. SOC.79, 3416 (1957).
96. Hallam, B. F., and Pauson, P. L., Chem. & Ind. 1955,653.
9.7, Hallam, B. F., and Pauson, P. L., J. Chem. SOC.
1956,3030.
98. Hallam, B. F., Mills, 0. S., and Pauson, P. L., J . Znorg. & Nuclear Chem. 1,
96. Graham,
313 (1955).
99. Hauser, C. R., and Lindsay, J. K., J. Org. Chem. 21,382 (1956).
100, Hauser, C. R., and Lindsay, J. K., J . Org. Chem. 22,482 (1957).
101, Hauser, C . R., and Lindsay, J. K., J . Org. Chem. 22,906 (1957).
102. Hauser, C. R., Lindsay, J. K., Lednicer, D., and Cain, C. E., J. Org. Chem. 22,
717 (1957).
103. Hein, F., Chem. Ber. 89, 1816 (1956).
104. Hein, F., and Eisfeld, K., 2.anOTg. allgem. Chem. 292,162 (1957).
106. Hein, F., and Fischer, K. W., 2. anorg. ailgem. Chem. 288,299 (1956).
106. Hein, F., and Kurras, E., 2.anorg. allgem. Chem. 290,179 (1957).
107. Hein, F., and Muller, H., Chem. Ber. 89, 27B (1956).
108. Hiickel, W., and Schwen, R.,Chem. Ber. 89,150 (1956).
109. Issleib, K., and Brack, A., Z . Natwforech. llb, 420 (1956).
110. Jaffe, H. H.,J. C h m . Phys.21,156 (1953).
111. Jira, R., Dissertation, Tech. Hochschule, Miinchen (1955).
112. Kaplan, L., Iiester, W. L., and Katz, J. J., J. Am. Chem. SOC.74, 5531 (1952).
113. Kauer, E., Z. physik. Chem. (Frankfurt)8.3 6, 105 (1956).
I i i . Kealy, T . J., and Pauson, P. L., Nature 168, 1039 (1951).
116. Kursanov, D. N., and Parnes, Z . N., Doklady Akad. Nauk SSSR 109, 315 (1956).
116. Leipfinger, H., Z. Naturforsch. 13b, 53 (1958).
117. Leto, J. R., Cotton, F. A., and Waugh, J. S., Nature 180,978 (1957).
118. Liehr, A. D., Naturwiss. 44,61 (1957).
119. Liehr, A. D., and Ballhausen, C. J., Acta Chem. S c a d . 11,207 (1957).
120. Lindsay, J. K., and Hauser, C. R., J. Org. Chem. 22,355 (1957).
191. Lindstrom, E. G., and Barusch, M. R., Contrib. Calif. Research Corp. 131st
Meeting Am. Chem. SOC.,April 1957.
12.9. Linnett, J. W., Trans. Faraday SOC.52,904 (1956).
123. Lippincott, E. R., and Nelson, R. D., J . Chem. Phys. 21,1307 (1953).
124. Lippincott, E. R., and Nelson, R. D., J . Am. Chem. SOC.77,4990 (1953).
125. Lloyd, D., and Sneezum, J. S., Chem. & Znd. 1955,1221.
126. Longuet-Higgins, H. C., and Orgei, L. E., J . Chem. SOC.
1956, 1969.
1%7. Liittke, W., and Fritz, H. P., unpublished work.
19s. Mayo, D. W., Shaw, P. D., and Rausch, M., Chem. & Ind. 1957,1388.
129. Meister, H., 2.angew. Chem. 69,533 (1957).
130. Miller, S. A., Tebboth, J. A., and Tremaine, J. F., J . Chem. SOC.1952,632.
131. Moffitt, W., J. Am. Chem. SOC.76,3386 (1954).
13%. Mulay, L. N., Rochow, E. G., and Fischer, E. O., J. Inotg. & NucZear Chem. 4,
231, (1957).
133. hatta, G., Pino, P., Mazzanti, G., and Giannini, U., J. Am. Chem. Soc. 79, 2975
(1957).
134. Natta, G., Pino, P., Mazzanti, G., Giannini, U., Mantica, E., and Peraldo, M.,
Chem. Abstt. 51,7049 (1957).
156. Nesmeyanov, A. N., and Kochetkova, N. S., Doklady Akad. N m k SSSR 109,
543 (1956).
136. Sesmeyanov, A. N., and I<ritskays, I. I., Chem. Abstr. SO, 13856 (1956).
COMPOUNDS OF AROMATIC BING SYSTEMS AND METALS

137. Nesmeyanov, A. N., and Volkenav,
113
N. A., Dokkzdy Akad. Nauk SSSR 107, 157
(1956).
138. Nesmeyanov, A. N., and Volkenav, N. A., Doklady Akad. Nauk
SSSR 111,
605 (1956).
Z. A., Doklady
Akad. Nauk SSSR 112, 439 (1957).
140. Nesmeyanov, A. N., Perevalova, E. G., and Golovnya, R. V, Doklady Akad.
Nauk SSSR 103,81 (1955).
141. Nesmeyanov, A. N., Perevalova, E. G., and Golovnya, R. V., Doklady Akad.
Nauk SSSR 99,539 (1955).
142. Nesmeyanov, A. N., Perevalova, E. G., Golovnya, R. V., and Schilovtseva, L.S.,
Doklady Akad. Nauk SSSR 102,535 (1965).
143. Nesmeyanov, A. N., Perevalova, E. G., Golovnya, R. V., and Nesmeyanov, 0. A.,
Doklady Akad. Nauk SSSR 97, 459 (1954).
1.64. Nesmeyanov, A. N., Perevalova, E. G., Golovnya, R. V., Nikitina, T. V., and
Simukova, N. A,, Izvest. Akad. Nauk SSSR Otdel. Khim. Nauk 1956, 739.
146. Nesmeyanov, A. N., PereValova, E. G., Golovnya, R. V., Simukova, N. A., and
Starovskii, 0.V., Izvest. Akad. Nauk SSSR Otdel. Rhim, Nauk 1957, 638.
146. Nesmeyanov, A. N., Perevalova, E. G., and Hhuranov, S. S., Doklady Akad.
Nauk SSSR 113,1293 (1957).
147. Nesmeyanov, A. N., Perevalova, E. G., and Nesmeyanov, 0. A , Dokhdy Akad.
Nauk SSSR 100,1099 (1955).
1-48. Nesmeyanov, A. N., Volkenav, N. A., and Vilchevskaya, V. D., Dokkzdy Akad.
Nauk SSSR 111,362 (1956).
149. Page, J. A., and Wilkinson, G, J.Am. Chem. SOC.74,6149 (1952).
160. Pauson, P. L., 1. Am. Chem. SOC.76,2187 (1954).
160a. Pauson, P. L., Quart. Revs. (London) 9, 391 (1955).
161. Pauson, P. L., and Wilkinson, G., J. Am. Chem. SOC.76, 2024 (1954).
158. Pfab, W., and Fischer, E. O., 2.anorg. allgem. Chem. 274,316 (1953).
16.9.Piper, T. S., and Wilkinson, G., Chem. & Znd. 1955,1296.
164. P per, T. S., and Wilkinson, G., Natumks. 42,625 (1955).
166. Piper, T. S., and Wilkinson, G., Naturwiss. 43,15 (1956).
156. Piper, T. S., and Wilkinson, G., J . Znorg. & Nuclear Chem. 2,38 (1956).
167. Piper, T. S., and Wilkinson, G., J. Znorq. & Nuclear Chem. 3, 104 (1956).
168. Piper, T. S., and Wilkinson, G., J . Am. Chem. SOC.78,900 (1956).
169. Piper, T. 8.,Cotton, F. A., and Wilkinson, G., J. Znorg. & Nuclear Chem. 1,
139. Nesmeyanov, A. N., Perevalova, E. G., and Beinoravichute,
165 (1955).
1GD. Piper, T. S.,Lemal, D., and Willtinson, G., Naturwiss. 43,129 (1956).
161. Ramirez, F., and Levy, S., J . Am. Chem. &c. 79,67 (1957).
162. Rauech, M., Vogel, M., and Rosenberg, H., J. Org. Chem. 22,900 (1957).
163. Rausch, M., Vogel, M., and Rosenberg, H., J . Org. Chem. 22,908 (1957).
164. Reynolds, L. T., and Willrineon, G., J. Znorg. & Nuclear Chem. 2, 246 (1956).
166. Richards, J. H., and Curphey, T. J., Chem. & Ind. 1956, 1456.
166. Richmond, H.H., and Freiser, H., J. Am. Chem. SOC.
77, 2022 (1955).
167. Riemschneider, R., and Helm, D., Chem. Ber. 89, 165 (1956).
168. Rinehart, K. L., and Cupby, R. J., J. Am. Chem. SOC.79,3290 (1957).
169. Rinehart,K,L., and Motz, K. L., Chem. & Znd. 1957,1150.
170. Rinehart, I<. L., Curby, R. J., and Sokol, P. E., J. Am. Chem. SOC.79, 3420
(1957).
114
171. Rinehart, K. L., Motz, K. L., and Sung
Moon, J . Am. Chem. Soc. 79, 8r49

(1957).
172. Rosenblum, M.,Chem. & A d . 1957,72.
1.73. Ruch, E.,Sitzungsber. buyer. Akad. Wiss. Sonderdruck No. 20 (1954).
174. Ruch, E.,Tagungsber. Hauptjahrestagung 1955, Chem. Ges. DDR, p. 125.Akademieverlag, Berlin, 1956.
176. Ruch, E., 2.physik. Chem. (Frankfurt) S.1 6,356 (1956).
176. Ruch, E.,Rec. trau. chim. 75, 638 (1956).
177. Ruch, E.,and Fischer, E. O., 2.Naturforsch.7b,676 (1952).
178. Schlogl, K.,Monatsh. Chem. 88,601 (1957).
179. Schomaker, V.,and Pauling, L., 1.Am. Chem. Soc. 61,1769 (1939).
180. Schwab, G.M., and Voitlander, J., Z. physik. Chem. (Frankfurt) CN.S.1 3, 341
(1955).
181. Schwab, G.M., Fischer, E. O., and Voitlider, J., Naturwias. 41,228 (1954).
182. Schwab, G. M., Voitlider, J., and Baumel, A., 2. physik. C h a . (Frankfurt)
ENS.] 10,343(1957).
183. Seibold, E.A., and Sutton, I. E., J . Chem. Phys. 26,1967 (1956).
184. Semenow, D. A., and Roberts, J. K., 1.Am. Chem. SOC.79,2741 (1957).
186. Sew, D.,
Dissertation, Tech. Hochschule, Munchen (1956).
186. Simonetta, M., Ricerca sci. 23,1234 (1953).
187. Stahl, H.O., Dissertation, Tech. Hochschule, Miinchen, 1957.
188. Summers, L., Rept: Congr. International Union for Pure and Applied Chemistry, Paris, 1956.
189. Summers, L., and Uloth, R. H., J. Am. Chem. SOC.76,2278 (1954).
190. Summers, L.,Uloth, R. H., and Hoimes, A,, J. Am. Chem. SOC.77, 3604 (1955).
191,Thiele, J., Ber. 34,68 (1901).
192. Thiele, J., Ber. 34,71 (1901).
193. Thomas, J. C., Chem. & Znd. 1956, 1388.
194. Trifan, D.S.,and Nicholas, L., J. Am. Chem. Suc. 79,2746 (1957).
196. Vogel, M., Rausch, M., and Rosenberg, H., J. Org. Chem. 22,1016 (1957).
196. Weiss, E.,
2.anorg. allgem. Chem. 287,236 (1956).
197. We&, E.,unpublished work.
198. Weiss, E.,and Fischer, E. O., 2.Naturforsch. lob, 58 (1955).
199. Weiss, E.,and Fischer, E. O., 2.anorg. allgem. Chem. 278,219 (1955).
$00. Weiss, E.,and Fischer, E. O., 2.anorg. allgem. Chem. 284,69 (1956).
201. Weiss, E.,and Fischer, E. O., 2.anorg. aZZgem. Chem. 286,142 (1956).
202. Weinmayr, V.,J. Am. Chem. SOC.77,3009 (1955).
203. Weinmayr, V . , J .Am. ChemSoc. 77,3012 (1955).
2O4. Weliky, N.,
and Gould, E. S., J. A m . Chem. SOC.
79,2742 (1957).
206. Wiese, H. K., and Cohen, C . A., Chem. Abstr. 51, 8664 (1957).
206.Wilkinson, G.,J. A m , Chem. SOC.74, 6146 (1952).
207. Wilkinson, G.,J. A m . Chem. SOC.74,6148 (1952).
208. Wilkinson, G.,J. Am. Chem. SOC.76,209 (1954).
209. Wilkinsori, c., Snturu~iss.42,96 (1955).
210. Wilkineon, G.,133rd Meeting Am. Chem. SOC.,
Sun Francisco (1958).
211. Willcinson, G., and Birmingham, J. M., J . d m . Chem. Soc. 76,4281 (1954).
212. Wilkinson, G.,and Birmingham, J. M., J. Am. Chem. SOC.76,6210 (1954).
$13. Wilkinson, G., and Birmingham, J. M., J. Am. Chem. SOC.77,3421 (1955).
214. Wilkinson, G., and Cotton, F. A., Chem. & Znd. 1954,307.
215. Wilkinson, G.,and Piper, T. S., J. Znorg. & Nuclear Chem. 2,32 (1956).
115
21$. Wilkinson, G., Cotton, F. A., and Birmingham, J. M., J . Inorg. & Nuclear Chem.
2,95 (1956).
217. Wilkinson, G., Pauson, P. L., Birmingham, J. M., and Cotton, F. A., J . Am.
Chem. SOC.
75,1011 (1953).
2f8. Willcimon, G., Pauson, P. L., and Cotton, F. A., J . Am. Chem. SOC.76, 1970
(1954).
219. Wilkinson, G., Rosenblum, M., Whiting, M. C., and Woodward, R. B., J . Am.
Chem. SOC.74,2125 (1952).
220. Wilson, F. C., and Shoemaker, D. P., Naturwiss. 43,57 (1956).
221. Wilson, F. C., and Shoemaker, D. P., J . Chem. Phys. 27,809 (1957).
222. Wilson, F. C., and Shoemaker, D. P., Intern. Unwn Cryst. 4th Intern. Congr.
Montreal, 1967 Abstract p. 54.
823. Wolf, L., and Beer, M., Naturwiss. 44,442 (1957).
294. Woodward, R. B., Rosenblum, M., and Whiting, M. C., J . Am. Chem. SOC.74,
3458 (1952).
226. Yamada, S., Nakahara, A., and Tsuchida, R., J . Chem. Phys. 22,983 (1954).
226. Zeiss, H. H., Yale Sci. Mag. 29,14 (1955).
227. Zeiss, H. H., and Herwig, W., J . Am. Chem. SOC.78,5959 (1956).
228. Z e k , H. H., and Herwig, W., Ann. 606,209 (1957).
229. Zeiss, H. H., Tsutsui, M., and Onsager, L., Abstr. 186th Meeting Am. Chem.
SOC.,New York (1954). Pp. 2930.
230. Zernike, J., Chem. Weekblad. 36,24 (1919).
,831.Ziegler, K., Froitzheim-Ktiblhorn, H., and Hafner, K., Chem. Ber. 89,434 (1956).
232. Ziegler, K., Z. angew. Chem. 66,209 (1954).
RECENT STUDIES OF THE BORON HYDRIDES
William N Lipscomb
School of Chemistry. Univenity of Minnesota. Minneapolis 14. Minnesota
I . Introduction
. . . . . . . . . . . . . . . . . . . 118
I1. The Boron Hydride Structures

I11. The Valence Theory . . .
A . Electron Deficiency . .
B. The Three-Center Bond
C.Diborane
. . . . .
D Higher Hydrides . . .
E . Molecular Orbitals . .
. . . . . . . . . . . . . 118
. . . . . . . . . . . . . . 121
. . . . . . . . . . . . . . 121
. . . . . . . . . . . . . . 122
. . . . . . . . . . . . . . 124
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
126
128
IV. Nuclear Resonance (NMR) Studies of the Boranes, Substituted Boranes.

and Related Compounds . . . . . . . . . . . . . . . 131
A . BH4- . . . . . . . . . . . . . . . . . . . . .
131
B. AI(BHJ3 . . . . . . . . . . . . . . . . . . . 132
C.&Es-.
. . . . . . . . . . . . . . . . . . .
.
.
.
.
D . BaH, Addition Compounds . . . . . . . . . . .

E.Bp&. . . . . . . . . . . . . . . . . . .
F.B4Hm. . . . . . . . . . . . . . . . . . . .
G.BGHo. . . . . . . . . . . . . . . . . . .
H.BsHii. . . . . . . . . . . . . . . . . . .
I.BaHm. . . . . . . . . . . . . . . . . . .
J . BioHi4 . . . . . . . . . . . . . . . . . .
K . BioHm- . . . . . . . . . . . . . . . . . .
L . BioHJ, . . . . . . . . . . . . . . . . . .
M: BloHJ (mp, 116") and BluHt?Br (mp, 105") . . . . .
N . BloHJ (mp, 72") . . . . . . . . . . . . .
0. BsHRBr (mp, 32-34") and BsHJ (mp, 53") . . . . .
.
132
. . 133
.
134
.
134
.
134
. .
. .
.
.
.
.
.
.
135
135
.
136
.
136
.
136
. . 137
. . 137
. . 138
V. The Equations of Balance . . . . . . . . . . . . . . . 138

VI. A Topological Theory of Boron Hydrides . . . . . . . . . . . 138
VII . Possible Boron Hydride Ions .
VIII . The Boron Halides
Notes Added in Proof .
References
. 144
. . . . . . . . . . . . . . . . . 146
IX Reactions among Boron Hydrides
Acknowledgments
. 148
. 151
. 154
. .
117
154
118
WILLIAM N. LIPSCOMB
1. Introduction
Of the monumental early work by Stock on the boron hydrides, so

ably summarized by him in his Baker Lectures (68), Sidgwick ( 6 . 5 ~re)
marks that All statements about the boron hydrides earlier than 1912,
when Stock began to work upon them, are untrue. In a remarkable series
of studies, Stock characterized B2He, B4Hlo, B5H9, B5Hl1, BBHIO,and
BloHlr and gave tentative evidence for a dodecabor&ne and possibly an
additional hexaborane. The fact that it was over twenty years before an
additional boron hydride (B9Hl5) was well characterized is only one of
many tributes that can be made to the thoroughness of Stocks chemical
researches.
The systematic principles of boron hydride structures and chemistry
are the principal subjects of the present review. There are several reasons
why these principles became clear such a long time after the discovery
of these compounds: (a) most of the compounds must be handled in
grease-free vacuum line systems; (b) some of the boron hydrides are
unstable at ordinary temperatures, explosive on contact with air, and
toxic; ( c ) the structures are based on principles, still incompletely
developed, of electron-deficient compounds; and (d) location of the
hydrogen atoms is a crucial part of the structure determinations, unlike
the situation in hydrocarbons, and had to be done for the most part in
X-ray diffraction studies of single crystals grown a t low temperatures.
The present review will be restricted to the boron hydrides themselves
and their relatively simple substitution derivatives. Even the closely
related, and very important, borohydrides are not reviewed and are
mentioned here only insofar as they are relevant to the boron hydrides.
Also, in order to restrict the review to manageable dimensions, I have
omitted references to nearly all of the early structure investigations and
early systematic valence theories now generally believed to be incorrect,
and to claims made for the existence of new compounds when the evidence
seems to be not sufficiently extensive or convincing.
The several sections of this review include the structures, the valence
theory of known structures with emphasis on the charge distribution, the
nuclear resonance studies of hydrides, derivatives and related compounds, the topological extension of the valence theory and its use in making predictions, and finally some brief remarks on the interconversions of
the hydrides.
II. The Boron Hydride Structures
The various times when a given structure is guessed correctly, given

some support from experiment, convincingly proved by experiment, and
119
understood, are rarely the same. An attempt to establish the latter three
objectives has been made (37, 18) for B2H6, BlOH14, BsHs, BIH10, and
B5Hl1 in chronological order. Later studies have established the structure
(17, 16) of
and the structure (16,&, 93) of BsH1o.l The certainty
with which these structures are known results from spectroscopic techniques for &HeJ and from X-ray diffraction studies of single crystals for
all of the other hydrides ; but independent though not conclusive evidence
exists for the structures from concurrent electron diffraction studies of
B5Hg and B4HI0 and from subsequent electron diffraction studies of
BzH6 and B10H14;also, a microwave study of B5H9 and several recent
nuclear magnetic resonance studies, to be described in Section IV, have
now appeared.
The structures and significant details of the molecular geometry are
summarized in Figs. 1 to 7. Since the previous summary, the full
microwave study of B5H9 has appeared (97). The slightly low X-ray
FIG.1. The BrH, structure. Electron diffraction parameters are R-B = 1.77 A,
B-H = 1.19 A, B-H (bridge) = 133 A, L HBH = 1215.
B
FIG.2. The B,Hm structure. Revised parameters are taken from the recent
reinvestigation (60) of the X-ray data. Electron diffraction values of B-H (terminal) = 1.19 A, and for B-H (bridge), 133 and 1.43 A are chosen; the X-ray study
established that the bridge Hs are closer to the central borons.
Fxa. 3. The BSH, structure. Interatomic distances, 1.69 A and 180 A are taken
from the microwave study (a).
What I tell you three times ia true (12).
120
WILLIAM N. LIPSCOMB
1-70
ti
FIG.4. The BaEI,structure, with revised X-ray parameters (60) not averaged over
the apparent molecular symmetry plane.
FIG.5. The BJX, molecule. The unusually short B-B distance of 1.80 A has a
standard deviation of f 0.01 A in the X-ray study.
FIG.6. The BeHm structure (16).
H
FIG. 7. The &HI, structure. Parameters are from the reinvestigation (60) of
the original data and have not been averaged over the presumed CPVmolecular
symmetry.
121
values of the B-B distances in the X-ray study of B5H9 have been
shown (58) to be caused by a torsional motion of the molecule in the
crystal, but the probable errors previously assigned do include the
revised values. A reinvestigation (50) of the X-ray data for B4H10,
B5Hll, and B1&4 has led to slightly revised distances, as shown in the
figures and discussed in Section 111 on valence theory. The position of the
unique hydrogen in the apparent symmetry plane of the BaHll molecule
improved in this revision to a distance of 1.09 A from the central, apex B,
and to distances of 1.77 and 1.68A (kO.19 A) from the two outer boron
atoms.
The B-H distances in the X-ray diffraction studies are not listed here.
They not only have large probable errors, but also have recently been
shown ($3)to be very sensitive to the choice of X-ray scattering curves
for boron. The electron diffraction values for B-H distances, although
they have large probable errors, are not subject to large systematic errors,
and hence are probably fairly reliable for BzHe and B4HI0. There is
some theoretical evidence ($7) that the B-H distances should increase
slightly as the H/B ratio decreases, but this effect has not been tested.
The same theoretical discussion has predicted the direction of asymmetry
of the hydrogen bridges.
The remarkable geometrical similarities of these hydrides is very
striking and will become more so as the valence theory is developed.
Superposition of the ball and stick type of molecular models on one
another brings out the diborane type of geometry in the open parts of
these molecules, and is very suggestive of the kinds of hybridization that
may be chosen for convenience about the boron atoms.
111. The Valence Theory
A. ELECTRON
DEFICIENCY
The theory of valence of these structures is of interest for several
reasons. The hydrides themselves have an unusual set of formulas, and
one might hope that a theory would correlate these and predict other
members of the series. But more important, because these hydrides are
electron deficientin the sense that there are more orbitals than electrons,
one might hope that their electronic and geometrical structures will aid in
the understanding of the large number of intermetallic compounds, and
of the border line between metals and nonmetals. This interpretative
problem is comparatively simpler for boron, which uses only the 2s and
the three 2p orbitals, and hydrogen, which uses only the Is orbital, in
the approximation discussed here. This approximation is fairly good in
122
WILLIAM N. LIPSCOMB
the smaller hydrides but may need some modification in the larger
hydrides ant1 infinite solids.
In saturated hydrocarbons, for example, CzHe, each carbon supplies
four orbitals and each hydrogen supplies one. Bonds are formed by
overlapping of one of the four tetrahedral sp3 orbitals of carbon with
either the 1s orbital of hydrogen to form a C-H bond, or another sp3
orbital of carbon to form a C-C bond. When the total number of electrons
exceeds the number of available atomic orbitals, as in NH3 or NzH4,
excess pairs occur in hybrid orbitals. In these compounds the chemist
writes chemical bonds, and unshared pairs if necessary, and regards the
orbitals as filled if each element of the first row has an octet of electrons
and if hydrogen has two electrons.
Kow in BrHg there are only 12 electrons for 14 orbitals. Any molecule
in which the number of valence orbitals exceeds the number of valence
electrons we shall call electron deficient. But the chemists intuition tells
him that a polyatomic molecule with unfilled valence orbitals should not
exist for long. If such a molecule were formed, it should adopt a new
geometry capable of description in terms of closed shells. Such a description is indeed possible for B2Hol and extension of the principles to the
higher boron hydrides is also possible. The chemical generalization is an
electron-pair three-center bond, which can be regarded as a localized
molecular orbital extending over three atomic nuclei with no change of
sign of its wave function in the region between the nuclei. This description
is a simple extension of an ordinary electron-pair bond, which can be
regarded as a localized two-center orbital extending over two atomic
nuclei with no change in the sign of its wave function between the
nuclei.
B. THE THREE-CENTER
BOND
The usual hybrid orbitals for boron in its better-known chemistry are
tetrahedral (e.g., BHc-) or trigonal (e.g., BF3). We shall start our
description with these hybrids and with the 1s orbital of hydrogen (Fig.
Fm. 8. (a) The spherically symmetrical 1s orbital for H ; (b) tetrahedral hybrid
orbitals for B; (c) trigonal hybrid orbitals for B, showing the rr-orbital extending
above and below the plane of the spa hybrids ( 4 2 ) .
123
(4
(b)
(0)
Fm.9. Examples of three-center bonds (41). Boron orbitals from BIand Bn,and
from Bs of (c) may be sp' or spa. The simplest LCAO calculations distribute the
bonding electron pair equally among the three atoms of the central three-center
bond in (c), and place --e on H or on Ba of the open three-center bonds, (a) or (b).
8 ) . Then we may form the types of three-center bonds shown in Fig. 9.
Intermediate types also may occur.

The results of a simple LCAO discussion (18), in which the molecular
orbital is formulated as a linear combination, @ A
&$B
a&, from
the atomic orbitals +A, qB, and $0 of atoms A, B and C, respectively,
yield only one lowest energy state for reasonable assumptions of the
geometries and interaction parameters. Thus for the ground state, only a
single electron pair is required. The detailed discussion (18) will not
be repeated here but the results are summarized in Fig. 10. As an
example, however, consider the bonding level of the open three-center
and qC
bond where overlap is neglected. Symmetry requires that
enter equally, and normalization requires that the s u m of squares of
coefficientsshall be unity. Hence the form of this orbital must be
+=
a($A
+ +
$c)
2-
$B.
Now neglect the interaction between the nonadjacent pair $A and +c,
and call $$.4HWr = .f$BH$Adr = .f$cH$& = $$BH$cdr = 8, and
J$AH$&
= .f$&$Bdr = J$cH$cdr = Ho.Then the energy E = J+H$dr
becomes
E = HO 4 a d l - 2a2j%
The energy minimum can be found by setting dE/da = 0, from which we
find a = 3, so that E = Ho di 8, and therefore
124
WILLIAM N. LIPSCOMB
-i.o)-l.o
-20
G-
-%
-2.0
0
FIG.10. Results of simplified LCAO discussion of the three-center orbital (41).

Orthogonality of the AOs and constancy of the Coulomb integrals have been
assumed. The ratio of resonance integrals - y / p = O corresponds to an open threecenter bond, that for -y/B = 1 to a central three-center bond, and that for ,8/y = 0 to
an ordinary electron-pair bond. The lowest energy states is E+a.
A simplified interpretation of the (12s as probability functions then places

the electron pair with probability 1/4 on atoms A and C , and 1/2 on atom
B in the open three-center bond (Fig. 9).
We now outline two approaches to a description of valence in the
boron hydrides. The first employs three-center bonds. This particular kind
of localized molecular orbital seems most suitable for the smaller, more
open hydrides. Its use in the more complex hydrides will require delocalization of the bonding electrons either by a molecular orbital modification
or a resonance description. The second approach is simply that of
moleculsr orbitals, which is particularly effective in the more condensed
and symmetrical hydrides. These approaches merge as the discussion
becomes more complete. It is an important result that filled orbital
descriptions are obtainable for the known boron hydrides. Also some remarks about charge distribution in the boron hydrides are possible. But the
incompleteness of this valence theory in this nontopological form is
indicated by the lack of a large number of unknown hydrides, whose
existence would be consistent with these assumptions.
C. DIBORANE
The molecular geometry and three-center bonding scheme (45, 18) for
BBHBare shown in Fig. 11. If we assume that two of the sp3 hybrid
orbitals of boron have maxima directed toward the singly bonded hydro-
125
FIG.11. Bonding in B I H ~(41).
gen atoms with L HBH = 121.5', then the equivalent (14) hybrid
orbitals in the three-center bond should make an angle HaBHb = 102'.
The observed value of this angle is 97", in excellent agreement, considering the experimental uncertainties, and the possibility of bent bonds.
Although the atoms are neutral in this simplified valence theory,
there is experimental evidence from a study of the proton magnetic
resonance spectrum (53) that the two bridge protons are more negative
than are the four terminal protons. Such differences in the diamagnetic
shielding have not yet been put on a satisfactory quantitative basis.
However, a thorough theoretical study of the molecular orbitals, in a
complete LCAOSCF discussion of the 4-electron problem involving
the bridge hydrogen atoms and the boron orbitals directed toward them,
indicates (21) that the bridge protons have about -0% excess charge,
a result confirmed by a more extensive study (76). A recent study of
the infrared vibrational intensities indicates a large change in electric
moment as the bridge hydrogens move toward the same boron apd a
small change when a terminal hydrogen moves (63).In our discussion
of the more complex boron hydrides, we shall make some use of these
indications of charge distribution about bridge and terminal hydrogen
atoms in discussions of their dipole moments (Section I11 E).
Comparison of the bonding in diborane and ethylene is illuminating.
In the usual description, each C atom of ethylene has planar sp" orbitals,
two directed toward H and the third directed toward the other C atom.
Overlap of the two sp2 orbitals directed toward each other, and of the
remaining r-orbital normal to the plane of sp2 hybridization with the
corresponding orbitals of the other C atom, give the two molecular orbitals of the double bond. If we call these molecular orbitals +a and +*, we
,., withmay then form equivalent orbitals XI = $a qr, and x2 = - +
out altering the description of the ground state or of the molecular wave
function. But if we remember that has the same sign above and below
the plane of the molecule, while
changes sign in this molecular plane,
we then abserve that XI is almost completely above the molecular plane,
and x2 is almost completely below this plane (Fig. 12). The equivalent
molecular orbital, X I , for example, can be thought of as being formed
from two atomic approximately sp3 hybrids, one from each C atom. This
126
WILLIABI S . LIPSCOMB
x, =
9"
x,=
+-
*#
J.,
qT
FIG 12. EquivaIent orbitals in the ground state of ethylene (41).
two-center orbital can be thought of as an electron-pair bond. In order

to obtain the three-center electron-pair bond in B2H6, one has only to
superimpose the 1s orbital of hydrogen and replace C by B. Thus the
general appearance of the orbitals is not greatly changed, and, in particular, the symmetry properties and number of nodes remains exactly
the same in CzH6and B2&.
D. HIGHER
HYDRIDES
The use of three-center bonds in describing bonding in the higher
hydrides has been discussed in some detail (18)and hence only 8 brief
summary is given here with a few modifications.
As a first illustration of the procedure, consider B4H10 (Fig. 13).
Electron-pair bonds to six hydrogen atoms, and three-center bonds to
the four bridge hydrogens, employ ten electron pairs out of the total of
eleven. These ten pairs also use up all of the hydrogen orbitals and all
but two of the boron orbitals, one on each BI. Thus we have one electron
pair for two boron orbitals which are directed toward each other and
hence form a single electron-pair bond between atoms BI.
As a second illustration, consider B5Hll. Bonds to the eight terminal
hydrogen atoms and to the three bridge hydrogen atoms require all but
two electron pairs and use up all but six boron orbitals. Hence, we must
form two three-center bonds to utilize the remaining two electron pairs
in these six orbitals; and the two methods of doing so, consistent with
the molecular plane of symmetry, are shown in Fig. 13.
Similarly for framework bonds, B6Hlo is required to have two threecenter bonds and two two-center bonds, BsHls is required to have four
127
H
H
-H
FIQ.13. A topological valence bond description of the known higher hydrides of

boron. Resonance hybrids are indicated by the symbol t).If the bonding geometry
at each boron atoni is recognized as tetrahedral or trigonal, the correct threedimensional structures can be realized.
three-center bonds and two two-center bonds, and BloHlr must have
six three-center bonds and two two-center bonds. Except for B ~ H Qfor
,
which a more general approach is more informative, these bonds can
be drawn consistent with the molecular symmetry (Fig. 13). Invariably
the simplest method of arriving a t the appropriate hybrid orbitals for
the boron atoms is to start from the bridge hydrogen regions where the
128
WILLIAM N. LIPSCOMB
directions of hybrid orbitals are most clear, and then work toward the
inner parts of the molecule.
As a predicti'on, consider BJHo, a reaction intermediate. This molecule has nine electron pairs, all of which would be required for bonds
to terminal hydrogen or to bridge hydrogen atoms. The most satisfactory
model has three BH2 groups linked by bridge hydrogen atoms.
E. MOLECULAR
ORBITALS
Useful as the three-center bond concept is in forming a common basis
for describing the known boron hydrides in terms of filled orbitals, and
in making some predictions, it seems clear that the electrons in the more
complex and symmetrical boron hydrides must be much more delocalizied.
Consider the three-center bond approximation in BgHo, in which three
electron pairs are left to fill molecular orbitals from seven boron orbitals
after all bonds to H have been formed. We would then expect one threecenter bond and two two-center bonds. The resulting bonds (Fig. 13)
indicate that the apex boron atom BI forms three electron-pair bonds
(one to H and two to BII) and one open three-center bond to two other
boron atoms. Now, if we divide the electron pairs equally and use the
simplest LCAO approach (Fig. 9) for the charge distribution in the
open three-center bond, we find that the apex boron atom has a net
charge of -1, and hence the four other boron atoms have a charge of
+1/4. This result, which comes from the assumption (18) of constant
single-electron Coulomb integrals, will evidently be modified if variation of the magnitudes of these integrals is allowed. An attempt to introduce this variation into the ordinary secular equation method with
the use of Moffitt's linear relation (49) between the Coulomb integral
and the number of electrons on the atom in question has been made by
Eberhardt, Crawford, and Lipscomb (18). The net result has been to
reduce the total formal charges on these atoms by a factor of about two,
and hence to obtain dipole moments in rough agreement with experiment,
assuming that the excess charge is centered on the atoms in question.
A more detailed consideration (39) of the charge distribution in the
hybrid orbitals about boron atoms in B5H9 and B5Hll indicates that the
excess charge centered on the atoms in question probably needs to be
reduced somewhat further in order for the expected dipole moment value
to approach more closely the experimental value. The qualitative charge
distribution for BJHe is as follows: Assume, as the bridge-hydrogen
geometry around BII (Fig. 13) suggests, that the hybridization is tetrahedral. The two boron orbitals of each BII atom directed toward the
bridge protons are deficient in electrons, compared with the other two
BII orbitals which are directed toward, respectively, the terminal hydro-
129
gen and BI. This situation can be thought of as arising from the relative
deficiency of electrons in the boron orbitals of the three-center bonds
involving bridge hydrogens, as compared with the electron pair in the
bond toward the terminal hydrogen or the framework electrons in the
bond toward BII. This relative deficiency has been shown (39) to be
almost independent of hybridization about the apex boron atom, BI,
and can be attributed to the unbalance of the relatively large atomic
moments in hybrid orbitals about BII. Then, if we assume neutral hydrogen atoms and neutral boron atoms, the resulting dipole moment of
B6Howould be about 6 Debyes, where the negative end of the molecule
is toward the apex. As the first correction to this argument, one would
anticipate that the bridge hydrogens carry a slight excess of negative
charge (64). If we then place about -0.le on each bridge hydrogen, this
calculated value will be in reasonably good agreement with the experimental value (88)of 2.13 Debyes. As a second correction, the apex boron
atom probably does carry a small negative charge, relative to the basal
boron atoms, and quite possibly the bridge hydrogens carry slightly more
than -0.le of negative charge. Further corrections would include a discussion of electron correlation effects, usually of importance in charge
distribution ;but the higher hydrides have yet to be studied theoretically
in this much detail. A similar unpublished analysis of charge distribution in B4H10 by C. W. Kern and W. N. Lipscomb has led to a value of
< 2 Debyes in reasonable agreement with the as yet unpublished value
of 0.6 Debyes observed by J. R. Weaver, C. W. Heitech, and R. W.
Parry.
The symmetry characteristics of the filled molecular orbitals (18) of
B6H0 are similar in some respects to those of benzene. The orbital of
lowest energy does not change sign throughout the molecule, and can
be thought of as a five-center orbital. The orbitals of next lowest energy
change sign at the apex boron atom and each has one node which corresponds to the node in the apex boron T) orbitals perpendicular to the
symmetry axis of the molecule. These three orbitals take up the six
FIQ.14. Topological representation of the lowest framework orbital (a) in BsHp,

and the doubly degenerate pair, (b) and (c). Six bonding electrons go into these
three orbitals (41).
130
WILLXAM N. LIPSCOMB
bonding electrons (Fig. 14) in agreement with the Pauli exclusion principle. This molecular orbital description is certainly the most natural
one and the relation between this description and the three-center bond
description has been discussed in detail (18).
--
FIO.15, Fimt approximation to the charge distribution in the known boron

hydrides. The abbreviated notation omits all terminal hydrogens except when a
B& group is present. These results are obtained when the bond charge distribution in Fig. 9 is applied to the structures in Fig. 13. The third valence structure
for M U(extreme right) correlates relatively poorly with the observed bond
distances and with the other hydride structures. Charges in B,ROare averaged over
the resoname hybrid.
131
Discussion of the orbitals in the more complex hydrides, for example,
B5Hl1, BaHlo, and B10H14, follows similar lines but their symmetries are
lower and hence there are more parameters in the molecular orbitals.
The results are similar to those for B5H9, however, and the resulting
charge distributions appear in general to place the negative end of the
molecule as a whole a t or near the apex, away from the bridge hydrogen
atoms. As in B5H9,the dipole moments can be thought of as arising from
the atomic moments about boron atoms, if one chooses to discuss atoms
in terms of hybridization.
The results of the simplest three-center bond discussion of charge
distribution in the boron hydrides are of some interest in connection with
electrophilic substitution, and hence are summarized here for the known
hydrides (Fig. 15). These charge distributions are more detailed than
those given (39) by a free-electron model of charge distribution in a hemisphere or hemispherical shell, but both of these treatments agree in a
general way. Negative charge tends to concentrate in the inner regions
of the molecule because the wave functions vanish toward the exposed
regions. If one superimposes .this general effect on the LCAO results
shown in Fig. 15, one may well expect that the two outer borons attached
to bridge hydrogens in BloH14 would be more positive than the innermost
pair of borons. If one adds the further conclusion that the bridge hydrogens carry about -0.le to -0.2e of charge, these two outermost borons,
labelled with 0 excess charge in Fig. 15, would become even more positive. It then becomes difficult to say whether this outer pair of borons
or the next inner group of four equivalent borons is the more positive.
On the other hand, the identity of the most negative pair of borons in
BloH14 is reasonably certainly that shown in Fig, 15, as has been confirmed by experiments described in Section IV.
IV. Nuclear Resonance (NMR) Studies of the Boranes, Substituted
Boranes, and Related Compounds
The recent applications of both H and Bll resonance techniques to

the boron hydrides and related compounds has led to the confirmation
of some of the structures, elucidation of many of the substituted structures, and an impressive amount of detail about the charge distribution
in these molecules and ions.
A. BH4The alkali borohydrides contain the tetrahedral BH4- ion, as has

been established by the early infrared work of Price (56) and powderdiffraction study by Soldate (67). A confirmation of the tetrahedral
132
WILLIAM N. LIPSCOMB
structure was obtsined by high-resolution NMR techniques by Ogg (53)

who Also studied B2H6 by this method. A study of the second moment
of the comparatively broad proton resonance in the solid alkali borohydrides (19) has led t o the B-H distance of 1.256 20.02 A, which is distinctly greater than one might extrapolate from the well-known distances
of 1.002 A in CHI and 1.035 A in NH4+. This effect of the negative
charge in expanding the hydrogen and boron orbitals is discussed in
Section IV C in relation to internal exchange in some of the negative
hydride ions.
B. Al(BH4)3
The trigonal, planar arrangement of 3B about a central A1 has been
established in an electron-diffraction study (6, 66, 4) which succeeded
in disproving the correct hydrogen arrangement, later established from
the infrared spectrum (66). The remarkable nuclear resonance (A12', Bll,
and H) spectra (64) produces a time equivalence of the protons in each
BH4 group by a tunnelling type of mechanism. In this process, collapse
of the hydrogen fine structure of the Bl' resonance does not occur but
rather all H's become equivalent so that the quintet of relative intensities
1,4,6,4, 1 is produced by interaction of four equivalent H's with the B"
spectrum. The hydrogen resonance is broad. The temperature independence of these spectra and the non-collapse of the H fine structure argue
strongly for this internal rearrangement, as opposed to some dissociation
type of equilibrium.
C. BaHB-
The Bl1 resonance (55) of a solution of NaB3H8 (25) in ether or DzO

yields an apparent septet with observed relative intensities of 1.0:3.5:
6.3:7.8:6.3:3.5: 1.0. The equality of spacings is very striking, but agreement is poor with the ratios 1:6:15:20:15:6:1 expected for a septet
resulting from a group of six H atoms. However, if all B's are equivalent, and all B's see all eight H atoms, the relative intensities of the
nonet are 0.1:0.9:3.2:6.4:8.0:6.4:3.2:0.9:0.1. The two very weak outer
lines are predicted to be one-eighth of the intensity of the weakest members of the septet, and hence are presumably lost in the background. The
proton resonance is broad, as is the case in A1(BH4)3.A static model in
agreement with the observations seems impossible, and hence this author
has suggested a model in which the electronic structure remains nearly
invariant and internal exchange takes place as in Al(BH4)3. The two
possible structures shown in Fig. 16 are both derivable from the predicted
B3Ha structure by removal of either a bridge or terminal proton. Presumably this type of internal exchange is facilitated by negative charge
(a 1
133
(b)
FIQ.16. Possible BaHa structures. Preference for structure (a) is suggested by the
valence theory.
transfer toward the boron and hydrogen by the A1 in A1(BH4)s and by

the negative charge in the ions. The resulting expansion of the interatomic distances, similar to that observed in BH4-, probably makes this
internal exchange much more likely in these compounds than in the
hydrides themselves.
D. B3H7 ADDITIONCOMPOUNDS
An X-ray diffraction study (62) of (CH3)3NB3H7,and the preparation (26)and preliminary Bll resonance study (55) of (CzH5)z0* B3H7
have been reported. The crystallographic evidence that the molecule has
a three-fold axis, and the somewhat rapid fall-off of intensities with
increasing angle of scattering, together with the valence-theoretical observation that a single hydrogen on the three-fold axis will be nonbonding to the Ba molecular orbitals, have led to the suggestion (23) of a
disordered crystal structure, in which there is a vacant orbital available
for formation of a single bond to the lone pair of NH3 or N(CH3)3
(Fig. 17a).
FIQ.17. Possible S& structures. The B atom at the top has a vacant orbital.
Formal charges are +1 and +% on the apex B atoms of (a) and (b), respectively.
134
WILLIAM N. LIPSCOMB
The B resonance in (CzH5)20 B3H7 is apparently a symmetrical

sextet which, on the grounds of a similar argument to that given in Section IV C above for B3Hs-, is very probably the more intense group
of an octet with theoretical line ratios of %:1:3:5:5:3:1:4/7. Again, the
two outermost lines are presumably lost in the background. If so, we can
expect internal exchange in B3H7-ether complexes, partly because of
transfer of negative charge toward the B3H7.These etherates were first
prepared by Edwards and co-workers (26).
In B3H7 * NH3 a crystal disorder similar to that found in B3H7
N(CH3)3 has been observed (51) above about -16C. Nevertheless,
C. E. Nordman and co-workers have obtained ordered single crystals
of the low temperature form, and in an excellent and detailed X-ray
diffraction study have shown that the B3H7residue shown in Fig. 17(b)
is joined to the NH3 by a single bond t o the boron atom having a vacant
orbital. It may, of course, not be correct t o deduce the structure of the
isolated B3H7molecule from these results, but the structure of Fig. 17(b)
has, in fact, a more favorable charge distribution than that of Fig. 17(a).
Note that addition of H- to the vacant orbital of either of these possible
B3H7 structures gives the B3Hs- ion of Fig. 16(a).
6
E. B2He
The several NMR studies (32, 53,64) of diborane confirm the bridge
structure. In general, among all of the boron hydrides, the bridge hydrogens tend to show H resonance at higher fields than the terminal hydrogens. This result is striking confirmation of the type of charge distribution expected on chemical grounds and predicted from molecular orbital
theory ( 2 1 ) . The detail with which the NMR spectrum can be analyzed
is shown by Schoolery (64) especially in the case of B2Hs. In this compound and in the higher hydrides, it has proved possible to simplify the
observed NMR spectra very greatly by nuclear stirring techniques,
in which a given nucleus (e.g., B) is induced to undergo many transitions during an observation. Under these circumstances, a neighboring
nucleus sees only the average over these orientations, and the fine structure due to the nucleus being stirred collapses to a single line.
J?.B&IO
The NMR spectrum has been run ( 3 2 ) .
G. B5HB
The use of Bl stirring techniques (7) has not only confirmed (64)
the tetragonal pyramidal structure including the ratio of bridge to terminal hydrogens, but has further indicated that the apex boron atom
135
shows resonance at a higher field than the basal borons. This result
agrees with the simplified valence theory of this compound, in which
the apex boron is the more negative. Somewhat improved spectra obtained later (61) confirm these results. The hydrogen resonance (6.4)
indicates that the apical hydrogen atom is slightly more negative than
the basal terminal hydrogens, but not so negative as the bridge hydrogens.
H. B&ii
The H and B1l resonances have been analyzed (61) to assign the most
negative boron (type I, apex) of the valence theory to the boron resonance at highest field. In the Bll-saturated hydrogen NMR spectrum, the
peaks reading from low to high field are said to be of approximate relative intensities of 6:1:3:1. Schaeffer, Schoolery, and Jones (61) assign
the six terminal H, except the apex H, to the large peak; then the terminal apex H to the next peak, the three bridge hydrogens to the next
peak, and the unique hydrogen on the apex boron atom to the single
peak a t highest field. They emphasize the lack of certainty about this
assignment, and further investigation is desirable by a combination of
NMR, infrared, and isotopic exchange techniques. In particular the relative peak weights may be 6:1:2:2,the bridge hydrogens may not all give
peaks a t the same place since two of the hydrogen bridges are probably
unsymmetrical (37'), and the unique hydrogen may not be so greatly
different from the other terminal hydrogens that it should show resonance a t the highest field.
The position of the unique hydrogen almost in the basal plane of the
four boron atoms has been reinvestigated twice (50) since the original
structure study (36).Among all of the boron hydrides, this is the only
extra hydrogen attached to a boron atom connected to more than two
other boron atoms. The valency rules would be somewhat simpler if this
hydrogen were a bridge hydrogen, and a satisfactory structure can be
drawn in such a case (W), although the hydrogen arrangement does not
have the symmetry of the boron arrangement. However, a very recent
unpublished reinvestigation by Dr. M. G. Rossmann and myself of these
X-ray diffraction data by the least-squares method, in which all atoms
have been assigned different temperature factors, indicated no randomness in the hydrogen positions; and hence it is still very likely that this
unique hydrogen lies in the apparent symmetry plane of the molecule
and that the published structure is correct.
1. BsHio
Pyeliminary H resonance results on a somewhat small sample (58)
are in better agreement with the hydrogen arrangement established (16,
136
WILLIAM N. LIPSCOMB
25) by X-ray diffraction than with the alternative arrangement suggested earlier (40, 41) on the basis of valence theory. The presence of
four bridge hydrogens rather than three thus receives some support from
the NMR data, but some caution is appropriate in interpreting these
NMR results in a molecule of low symmetry.
J. B10H14
The low symmetry of BloHla does not readily permit a complete assignment of all peaks to specific boron or hydrogen atoms. Two assignments, however, are reasonably certain. That there are four bridge hydrogens is indicated by the H resonance peak a t highest field (64, 61)
which, relative to those peaks a t lower field, has the correct relative integrated intensity. Also, the B1l resonance shows a doublet a t high field
which is no doubt due to the two equivalent borons of type I11 (Fig. 13)
split by the terminal hydrogens on each. This latter assignment is predicted by the charge distribution and strongly supported by the results
described below (Section I V L) for B10H1212.The doubt that one must
Iiave about the remaining assignments will probably be removed as
additional isotopic or otherwise substituted boron hydrides are studied
by NMR techniques.
K- BioHi3The BloH14 molecule in 50% ethanol-water behaves (20) as a strong

monoprotic acid. The second ionization constant has a pK > 10, out. of
range of the studies made so far. The doubled peak at high field in tlic
B" NMR spectrum shows that the initial stages of ionization do not
involve the 111,111' hydrogen atoms ( 5 8 ) .Infrared spectra of deuteriumhydrogen exchange of BloDI4 indicate (20,24) that the four bridge hydrogens exchange first, followed by further exchange or internal rearrangement involving a t least four of the terminal hydrogen positions.
There is also some as yet unpublished (20)infrared evidence suggesting
a BH2 group in B10H13-, and possible structures for this ion are discussed
in Section VII.
1,. B10H1212
The assignment of substitution of I in B10H1212to the 111, 111' positions is based on a determination (59) of the intramolecular I . . , I
distance by projection methods from X-ray diffraction data. The observed distance of 6.25 A compares favorably with that expected from
the known molecular structure. If the B-H results of Kaspcr, Lucht, and
Harker (SO) are extrapolated to an expected B-I distance, the result is
5.92 A for the 111, 111' substitution, and 7.28 A and 7.56 A for the other
RECENT STUD]=
OF THE BORON HYDRIDES
137
possibilities. If the further refinement of the BloHla structure by Moore,

Dickerson, and Lipscomb (60) is used, 111, 111' substitution predicts a
distance of 6.21 A, in somewhat better agreement with the observed
value. The other possible positions lead to predictions of 7.88 h and
7.60 A, and hence are rejected as possibilities. In view of the importance
of this structure determination in the following argument, the location
in some future study of the boron framework in the X-ray diffraction
study would be important as confirmatory evidence of the position of
substitution.
The Bll resonance study (62) of B10H1212establishes that the I atoms
are substituted on the two B's giving Bl' peaks at highest field. It is
important that in this and the following Bll resonance studies of substituted boron hydrides, the positions of boron resonances are but little
shifted from their relative positions in the pure hydrides. This discovery
has made the field of substituted boron hydrides easily studied by NMR
techniques, and should lead to the identification of presently unassigned
resonance peaks in the pure hydrides themselves and provide a basis
for studies of internal exchange reactions. Thus, we can expect more
detailed analyses of the charge distributions in both the hydrides and
their substitution products to appear in the near future.
This identification of the I substitution at the 111, 111' positions in
BloH1212 leads also, upon comparison with the charge distribution in
B10H14, to the idea that the dominating mechanism of substitution is
electrophilic attack of the halogen atom, as was predicted earlier (40,
41). Perhaps this is true in a majority of cases, but the existence of two
different B10H131's leads one to the caution that other processes may be
involved, perhaps a competing reaction by the same or a different
mechanism (58).
M. B10H131(MP, 116') AND BlOHISBr(MP, 105")

The H and B" resonance spectra indicate (62) that substitution takes
place in both of these compounds on the boron giving resonance a t highest field. By the arguments given in Section IV L, this is the boron atom
of type 111.
N. BloHlsI (MP, 72")

The H and Bll resonance spectra show clearly that the doublet at
highest field is unaffected, and hence that substitution does not occur a t
the I11 position. A detailed argument (68) suggests that I1 substitution
has taken place and that the previously published assignment (61) of
the unresolved peaks in B10H14itself needs to be revised. An investigation
of this structure by X-ray diffraction methods is needed.
138
0. B5HsBr
WILLIAM N. LIPSCOMB
(MP,
32-34')
AND
B6HJ
(MP,
53')
The position of substitution is clearly indicated by both H and Bll

resonance spectra to be the more negative apex boron atom of B ~ H o .
It is, therefore, remarkable that the apparent rates of deuteration of
the apex and basal terminal hydrogen atoms of B4Hg by B2D6 are the
same (35) to within +lo%. Koski, Kaufman, and Lauterbur (35) remark that they have not eliminated mechanisms in which only terminal
apex (or base) H atoms exchange, followed by rapid intramolecular
redistribution of H and D atoms among all terminal positions. While
they considered this mechanism unlikely and hence favored direct exchange, it may be important to reconsider this mechanism in view of the
strong preference of halogens for the apex position.
Mention of the terminal-hydrogen substituted B2H5Br seems desirable to complete this section, and perhaps to remind us that no bridge
halogens have been found. But the existence of C1 bridges in BeC12, of
CH3 bridges in A12(CH3)6, and of N bridges in BzH5NH2 and
BzH5N(CH, j suggest that these or similar possibilities may exist for
substituted boron hydrides or boron hydride ions.
V. The Equations of Balance
The relative numbers of orbitals, electrons, hydrogen, and boron

atoms, and bonds of various types can easily be expressed in a systematic
way. The procedure developed by Eberhardt et al. can be simplified as
follows. For a boron hydride, B,H,+,, containing s H bridges, x extra
B-H groups, t three-center B-B-B bonds and y two-center B-B bonds,
the equations of balance may be written in a simplified form as follows.
The hydrogen atom balance is simply s x = q. Since each boron supplies four orbitals but only three electrons, the total number of threecenter bonds in the molecule is the same as the number of boron atoms,
s
t = p. Finally, if we consider B H as the bonding unit, each of which
then supplies one electron pair, these p pairs are used up as three-center
B-B-B bonds, two-center B-B bonds, and in supplying half of each
pair to each extra hydrogen; thus p = t y q/2.
The results, summarized in Table I, will be essential in the topological theory which will now be outlined.
+ +
VI. A Topological Theory of Boron Hydrides
Without the introduction of some formal arguments which abstract

the common structural features of the known hydrides, the three-center
139
TABLE I
TABLES
OF s, t,
AND
- &HI
- BrHa
BaHr
BaHr
BaHi
BaHs
----
BrHz
2010
1101
2002
2210
1301
4020
3111
2202
BSHL
BaHi
BrHe
-
0600
2310
1401
41x1
3211
2302
0200
BrH4
z FOR BORON
HYDRIDESTRUCTURES,
B,H,+,+
0300
2110
1201
3011
2102
3003
B4Hs
0400
4012
3103
4004
5021
4112
3203
5013
4104
5005
BaHu
BaHir
&Hi6
BsHia
BiHa
Ba&a
---4220
2410
1601
3311
2402
6030
5121
4212
3303
6022
5113
4204
6014
5105
6006
BiH7
BtHs
BiHii
BiHia
BiHir
BrHir
B7IIis
BiHn
-------0700
2510
I601
4320
3411
2502
6130
5221
4312
3403
7031
6122
5213
4304
7023
6114
5205
7015
6106
7007
BaHu
BaHia
BaHm
BaHm
BaHn
BaHu
BaHg
&HIP BsHu
----4420
6230
8040
8032
OSOO
2610
8024
8016
8008
1701
3510
2602
5321
4412
3503
7131
6222
5313
7123
6214
5305
7115
6206
7107
4.404
BaHie
BaHn
BSHZJ &Hai
-- -- -----BoHe
BsHu
&Hi:
&Hu
BsHe
om
2710
1801
4520
3611
2702
6330
5421
4512
3603
8140
7231
6322
5413
4504
0.10.0.0
2810
1901
4620
3711
2802
6430
8240
7331
6422
5513
4604
z
a
-- +- -
5521
4612
3703
9033
8124
7215
6306
9041
8132
7223
6314
5405
9025
8116
7207
9017
8108
&Hn
9009
BioHn BioHu BioHa BiaHr

BioHa
-----
10.0.5.0 10.0.4.2 10.0.3.4 10.0.2.6 10.0.1.8 10.0.0.10

9133
9125
9117
9109
9141
8232
8224
8216
8208
7323
7315
7307
6414
6406
5506
numberof bonm-bomn rinole bods, B-B

number of sxtra H atamr on B-H;Lo.. BHr ~ o u p s
t
9..
t
P. P t + Y + 9 / 2 .
bond approximations described above lead to a large variety and nuniber of possibie hydrides. The topological ideas to be described here are
those of Dickerson and Lipscomb (16) and are capable of very considerable extension to other systems. The major principles are concerned
140
WILLIAN N. LIPSCOMB
with the ways in which various types of bonds are able to connect the
framework B-H units that are adjacent to one another:
(a) Assume that each boron must be connected to each adjacent boron
atom by a t least one two-center or three-center bond of some kind.
(b) Define the connectivity of a boron atom as the number of adjacent boron atoms in the molecule. Define the excess connectivity ( E C )
as connectivity minus three.
( c ) Define the excess negative connectability of a bond arrangement
a t the site of a given boron atom as the number of orbitals which it requires from that boron atom minus the number of other boron atoms
which are connected to the given one by that bond arrangement. A bond
arrangement may be several bonds considered as a unit.
Definition ( b ) , applied to the more compact icosahedral and octahedral fragments to which our predictions are limited for convenience, is
illustrated in Fig. 18. Definition (b), applied to various bonds and bond
arrangements, leads to the excess negative connectabilities ( E N C ) shown
in Fig. 19. Note that a B-B single bond or a B-H-B three-center bridge
bond, when isolated from other bonding arrangements, connects two
borons and uses one orbital from each so that addition of these bonds
*9
FIG.18. Excess connectivities of possible boron-hydride frameworks.
-I
-I
-I
-I
141
-I
n
0
0
+-,
-I
I+ I
-I
FIG.19. Excess negative connectabilities of various bonds. Note that the extra
B-H, which forms a BHI group, removes a boron orbital without connecting that
boron to any other boron.
produces no change in the EC of a framework. Also note that if every
boron were of connectivity 3, i.e., EC = 0, the three orbitals left after
forming the B-H bond would suffice to form a framework of simple
covalent bonds, provided enough electron pairs are available. Hence,
any EC > 0 must be removed by addition of three-center B-B-B bonds,
and any EC < 0 must be removed by adding extra Hs to form BH2
groups. It is also permissible, as shown by the so far unique example
of a boron atom in BaHI1, to add an extra H to a boron of EC > -1,
thereby raising the EC of that boron atom by 1. Hence the procedure to
be followed is:
(a) Make a topological sketch of the assumed boron framework, indicating the EC at each boron site.
(b) Add extra B-H bonds to every boron of EC = -1, and possibly
to certain borons of EC = 0 or 1, but not to a boron of EC = 2.
( c ) Add three-center B-B-B bonds or arrangements of them, such
m shown in Fig. 19, in such a way that the EC at every boron site is
reduced to zero.
(d) Complete the molecule with B-B and B-H-B bonds, subject
to the equations of balance (Table I).
It must be emphasized that this type of topological theory is not
limited to three-center bonds. Its extension to other types of semilocalized multi-centered molecular orbitals might well prove useful in an
extensive study of valence in certain intermetallic compounds in which
142
WILLIAM N. LIPSCOMB
the nearest neighbors can easily be recognized, especially if the number

of nearest neighbors is not too large. This type of theory is, moreover,
not limited to finite molecules, single polyhedral fragments, or icosahedral fragments.
In order to eliminate from consideration certain structural features
not present among the known boron hydrides, some further principles
may be stated in order of decreasing reliability:
(a) All known boron hydrides have at least a plane of symmetry.
Low symmetry appears to provide reactive points which are averaged
over when symmetry is introduced. A new hydride probably would have
a t least a plane or some other twofold element of symmetry.
(b) No two adjacent boron atoms can be connected by both a central and an open three-center bond, because such an arrangement gives
a very unfavorable angle between valence orbitals.
(c) Borons of EC = 2 do not contribute to a B-H-B bridge or to
an extra B-H, for reasons relating to unfavorable valence angles between orbitals. This assumption restricts B-H-B bridges to the edge of
the framework, and BH2 groups to the outer boron atoms.
(a) Each boron of EC = -1 will probably be bonded by one hydrogen bridge to a t least one of its two adjacent borons. Again this principle
is based upon the requirement that angles between valence orbitals shall
not be too small.
(e) An edge boron of EC = 1 will probably not be connected t o two
other borons by hydrogen bridges. This is.probably the weakest of our
assumptions.
As an example, consider the B3 hydrides based on a B3 triangle. Only
the 3003 B3Hs structure mentioned earlier (18)is satisfactory (see Section IIID). The reader may find it instructive to try to draw the B3H7
and B3Hs possibilities (B3H3is impossible), and show that all have unsatisfactory angles about boron for reasonable hybrid orbitals. The least
unsatisfactory of these is the 2102 B3H7with C, symmetry in which the
single bond shown in Fig. 17s has been converted to a central threecenter bond.
As a second example, consider the possible Bs hydrides (15) based
on the framework shown in Fig. 18. Two adjacent borons of E C = 2
require a t least four three-center B-B-B bonds and hence t 2 4. There
is at least one B-H group because of the one boron of EC = -1, and
hence x 3 1. At least one bond from this boron of EC = -1 to the other
borons is a hydrogen bridge, and with the plane of symmetry we require
s 2 2. These restrictions eliminate most of the possibilities in Table I,
leaving 3611, 2702, 5421,4512, 3603, 5413, 4504, and 5405 as possibilities,
and restricting the number of hydrogen atoms to between 13 and 19. A
143
inore detailed study of these possibilities yields only the valence structures shown in Fig. 20. Note that the first three have the same nuclear
arrangement and correspond merely to different ways of drawing valence
bonds. The last structure is distinct, however, and has three BH2 groups.
When the structure of Fig. 20(d) was introduced into the least-squares
I
c)
(d)
FIQ.20. Possible W hydride valence structures.
refinement of the X-ray data (16),the two hydrogen atoms in the additional BH2 groups refined to the bridge positions shown in Fig. 20(a),
b),or (4.
The importance of this topological extension of the valence theory
is that it yields so few predictions (15), which may be summarized as
follows:
B4 (icosahedral fragment) gives the established BaHIo structure
uniquely.
B5 (icosahedral fragment) yields uniquely the established geometrical arrangement for B5H11.
B5 (octahedral fragment) yields the established B5H9 structure if a
fourfold axis is required.
Bs yields two possible
structures, one with no BH2 groups, and
one with one BH2 group. The former is the structure determined in an
X-ray diffraction study (W). Be also leads to a B6Hl2 structure, with
two BH2 groups on borons of EC = 0, probably a reactive situation.
B, yields no satisfactory structures based on an icosahedral fragment.
144
WILLIAM N . LIPSCOMB
Bs yields two BsH12 structures and one geometrical arrangement for

BsH14.
BIOyields only the established &OH14 structure.
It is possible that some new hydride, perhaps less stable than the
known ones, will violate some of these assumptions. This is particularly
likely for reaction intermediates. Hence an examination has been made
(23) of some of the more likely possibilities which may occur if the
symmetry restriction is removed. It is further quite likely that the present form of the valence and topological theory will be of aid in the
elucidation of reaction mechanisms.
VII. Possible Boron Hydride tons
It is easy to extend the foregoing topological arguments to boron hydride ions (4.2). Consider ions of the charges c = -2, -1, 1, and 2 in
the form B,Hcp+,+,. The modified equations of balance are s + x =
c, and t y = p - c -q/2. The structures derived
q +c, s +t = p
by the procedures described in Section VI can easily be drawn in topological form from the four numbers s, t , y, and x, which precede the
formula. These possible ions are classified as the following types:
(a) Ions similar in geometrical structure to known hydrides: The
4450 BloH14-2 ion of C2,, symmetry is the only ion of this type, and is
probably the most interesting of all of the ions. Except for a probable
contraction of the two 2.0-A.B . . . B contacts towards a more normal
value, the geometry of this ion is probably very similar to that of the
well-known 4620 BloHlr structure, or is possibly a tautomeric form.
(bj Ions similar in electronic structure to known hydrides: The prediction (43) of the BsH9-like ions, B4H7- and BGH11+, was based upon
analogies of these compounds to CeHe, C5H5- and C7H7+.Equivalent
orbital transformations reduce these to 3030 BIH7- and 5210 BGHlt+
ions in the present description.
(c) Ions derived from known hydrides by removal of H f from a
B-H-B bridge bond: These structures satisfy our rules if the electron
pair is used to form a B-B bond. It is obvious how a list of these ions
can be made from the known and hypothetical neutral hydrides, and
hence such a list will not be given here. However, thc 2013 BsHa- ion
(25) is probably either a member of this class or of class (d).
(d) Ions derived from known hydrides by removal of H + from a
B-H (terminal) bond: Terminal hydrogens tend to be less negative
(21) than bridge hydrogens and might ionize off more readily, although
the resulting ion usually does not have as satisfactory a valence structure as the ion produced in class (c). The various terminal hydrogen
145
atoms in the known hydrides are attached to borons of different formal

charge (18) and those attached to the more positive boron atoms will
presumably ionbe off more readily. The four borons of type I1 in BloHii
are more positively charged than the others, and hence possible ionization of one or more of the attached hydrogens may be suggested; presumably the pair of borons of type I are as positive or next most positivc
and should also be considered as ionizable. There is some evidence that
BloH14 has an ionizable hydrogen ($0)and the series of salts NaBIOH13,
Na2BI0H12, etc., may be suggested. When this suggestion is combined
with the class (a) type, Na2Bl0HI4,we expect another series beginning
with Na3B10H13, etc.
(e) Ions derived from addition of H+ to a single B-B bond on the
edge: The B&l+ ion derived by addition of a proton to the strained
single B-B bond in the known ($3) 4220 B6H10 structure is expected to
be stable. Although no other of the known hydrides based upon icosahedral fragments have this structural feature, several of the possible hydrides and ions do, and might become stabilized by the addition of one
or two protons.
(f) Other ions based upon icosahedra or icosahedral fragments: Ions
ion may also exist, as
like the 3122 B6Hlo- ion and the 3100
may the 2113 B4Hs- and 2440 B8H10-2, which have one or more single
B-B bonds along the edge; see class (e). Probably the most promising
ion (47)of this type is B12H12-2, which is an 0.10.3.0resonance hybrid
.in our localized-bond description.
(g) Ions based upon tetrahedra, octahedra, or their fragments: Our
bonding scheme applies to any geometrical framework in which one can
say that only close contacts represent valence interactions, and hence
includes 0230 B4H4-2, (0400 B4H4),2040 B4H6-2, 3030 B4H7- also listed
in class (b) , (4020B4H8), 5010 B4H9+, and 6O00 B4H10+2, all based on
a tetrahedral boron arrangement; 2240 B ~ H B - ~3230
,
B8H9-, 0430
BsH~-2 (see Eberhardt et al., 18), 1420 B&-, (2410 B6He), and 3400
BaHe+, all based on an octahedral boron arrangement; and a similar
series based on the B6HB arrangement. Within each of these groups, these
ions are interrelated by the process describing class (e) as the charge
increases, or class ( c ) as the charge decreases. Ions involving features
not present in the known hydrides have been omitted.
While the above classifications and lists probably do not exhaust all
possibilities, it may be hoped that the omissions do not overlook any
obviously promising stable ions. For example, we have not considered
ions based on linked polyhedra or linked fragments of polyhedra. We
have not tried to list completely the results of applying the processes
of classes ( c ) , (d), and (e) to all ions listed above. Also, we have not
146
WILLIAM N. LIPSCOMB
taken complete account of a very suggestive isomerisation mechanism

indicated in the topological theory (16) and shown in Fig. 21 for the
B10H13- ion derived by process (c) above. For BaHlo the left side is
favored ; for B10H13- apparently the right side, which places negative
charge on the BH2 group, is favored. A similar tautomerism can be drawn
FIG.21. Tautomeric forms for B,H,-. Most of the ambiguities remaining in the
topological theory are related by a similar tautomerism (16).
between the 4450 ( C W ) and 2632 ( C W ) and 0814 (Cz)B10H14-2 structures. Finally, these principles are applicable to types of polyhedra other
than those discussed here.
The available experimental evidence for ionic boron hydride fragments suggests that much careful experimental work is desirable. Stock
(68) mentions NazBzH6and Na2B4H10,from which the original hydrides
B2HB and B4H10 are a t least partly recoverable. He also describes (69,
70) Na2B,Hs, and NazBIHs as a decomposition product of Na2B4Hlo.
Recent work (25,31) has shown that the diborane reaction gives NaBH4
and NaB3He. The remarkable B2H7- ion ( l o ) ,prepared by reaction of
BH4- with BZH6, may possibly be formed only in solution or only when
BH4- can react with the double hydrogen bridge which occurs in B2Hs
but not in the higher hydrides. This ion probably consists of a single BHB
hond between two BH3 groups.
In summary, the more stable ions are probably B1zH12-2, B10H14-',
B10H13-7 B6H11+, BsHs-, B4H7-, BgH6-2, B5H10-, B3He+. The fact that
both positive and negative ions might exist leads to the suggestion (43)
of purely ionic hydrides, possibly prepared by reactions of salts of these
ions with one another.
VIII. The Boron Halides
Aside from the welLknown trihalides, BX3, the structures of only

B2F4 (7I), B2C14 (S),and B4C14( 1 , W ) are established. At least three more
halides (73,79)(BClo,B)
=,soluble in organic solvents; (BClo.6)$,insoluble
in solvents; and a probably two-dimensional poIymer, (BCI)=, have been
147
described. Although it is probably unwise to formulate a general theory

from these few examples, at least some of thc aspects of boron halide
structures may be stated.
The importance of back coordination of the extra electrons on C1
in stabilizing BCIB relative to its dimer has long been recognized.
Presumably the primary B-Cl bonds are planar a-type sp2 hybrids.
The remaining dorbital of boron perpendicular to the molecular plane
can accept an average of an electron pair from orbitals of similar symmetry from the C1 atoms. This type of electron transfer is not possible in
BH3 because no orbitals of H have the symmetry of the T orbital of B,
and hence BH3 dimerizes. The fact that B2F4 and B2C14 are planar, at
least in the solid state, also suggests that # orbital interactions are of
importance either within the molecule or between different molecules in
the solid. Again this explanation accounts for the non-existence of B2H4.
A similar explanation in molecular orbital language appropriate to the
tetrahedral B4C14molecule has been given by Longuet-Higgins (46) in
order to account for the stability of B4C14 and the nonexistence of the
c ompar able B4H4.
In deriving the principles of boron chloride structures, we shall make
the following assumptions:
(a) Terminal BC12 groups can occur, with three-coordinated boron.
(b) No C1 bridges occur; if this assumption is to be modified, we must
expect that the C1 bridge involves single bonds to each boron, similar to
the situation (57) in BeC12.
(c) BCl2 are terminal groups, and hence any B-C1 bond can be replaced
by a B-BC12 group. Thus we may limit consideration to polyhedra or
fragments with terminal B-C1 bonds which can later be replaced by
terminal B-BC12 groups.
(d) If the formula of a boron chloride is represented as B,Cl,+,, then
the equations of balance (18) are for the electron balance
p
+ q + 2t + 2y = 3 p
and for the orbital balance

p
+ q + 3t + 2y = 4p.
Note that the assumption of no bridge C1 atoms and no BCls groups at this
stage requires that s = x = 0. The above two equations give t = p , from
which we find y = -q/2 from either of the above equations. Values for
s, t , y, and x are thus 0, p , -q/2 and 0. But y 2 0, and hence q 0;
moreover, y is integral, so that q is an even negative number or zero.
Suppose we illustrate these principles with a discussion of a molecule
<
148
WILLIAM N. LIPSCOMB
of approximate composition BsCls or BloCls. The possible formulas are,

according t,o the above rules: B4C4,B5Cl5, BaCls, B7C17, BsCls, BsC14,
B7Cls, B8C16,BOCl7,and BloCls. All such molecules are to have terminal
C1 groups replaced, if necessary, by BClp groups in order to bring the total
number of C1 atoms to eight. A BSCl8 possibility based upon a B4
tetrahedron is tetrahedral B&l, with each C1 replaced by BC12. A B&l6
model with two tetrahedral B4C13 groups joined by a single B-B bond,
after which we replace two C1 atoms with BCL groups, leads to a BloCls
molecule of the form C12B-B4ClzB4C1z-BC1~.
These two models do
not, by any means, exhaust all of the possibilities but they do illustrate
the ideas.
If the polymers are based on B4 tetrahedra, a suitable linear polymer
might be -B4Clr--B4CIrB4Cl2-,
and a suitable sheet-like polymer
might be B4Cllinked by three bonds to three neighboring B&l groups, etc.
The molecule (BClo.J (where 2 10) is currently under investigation by X-ray diffraction methods ( 2 9 ) .
IX. Reactions among Boron Hydrider
Although BE3 has not been observed experimentally, the kinetics of

the initial stages of several reactions of diborane (13,9) render the
steady-state equation (6),
BzH6
2BHS,
followed by the reaction

BH,
+ B2H6--+
(higher products)
(2)
as the generally accepted initial reactions. The discussion will be

simplified if we consider first this and the later work relating to these
initial reactions, after which we shall return to a consideration of the
further reactions in the synthesis of the higher hydrides.
The initial stages of B2Ho polymerization ( I 3, 9 ) , of exchange (48)
between BzHe and I)*,of exchange (48) between B2Hs and BIDs, of
exchange (65) between B2Ho and BZ"H6, and of reaction (74) of BzHs
with CzH4, all support this mechanism. Experimentally, all of these
reactions are 3/2 order with respect to diborane concentration, as may be
derived from the mechanism by writing Eq. (1) as a steady state,
K = (BHa)2/ (BzHs), so that
d(higher products)
dt
- k(BH,)(B,Ha)
-
is the rate-determining step of the initial stage.
k6(B2H6)3/2
149
RECENT STUDIES O F T H E BORON HYDRIDES
In summary, the B2Hg-B2De exchange reactions are:

BZHs Ft 2BH3
B2De a 2BDa
+ BsDa + BHBDs + BHa

BDa + BsHs + BHaBDa + BHa
(BDa + BHa @ BHaBDs)
BHa
the third and fourth being rate determining. The B2H6-D2

summarized by the reactions:
BgHa
BHa
exchange is
2BHa
+ De + BHlD + HD (surface)
BHD
+ BzHs * B A D + BHa,
where the last equation is rate determining.

The B2Hg--C2H4 reaction is of importance partly because it gives
support to the idea that the initial higher product is B ~ H or
B B&
HP.
The initial stages of the mechanism suggested (74) are:
B2Hs Ft 2BHs
Be&
BlHp
+ BHa
+ CB4
BsH9
+
BeHsEt
+ BHa.
The last two reactions are then repeated starting with B2H5Et in place of
B& to obtain B2H4E&,which reacts with BH3 to give B3H7Etz.At this
stage an autocatalytic reaction appears,
BaH.rEt,
+Bas
+ 2BsHsEt
+ BHa,
so that B2Ha and CzH4 compete for these later products. The autocatalytic effect increases considerably as the final product BEts appears,
and if 2 (&He) > (C2H1), the reaction becomes explosive.
The present literature (IS,9, 60) on the polymerization to higher
hydrides around lW0C may be summarized in the sequence of reactions
B&s
2BHa
(1)
150
WILLIAM N . LIPSCOMB
Reaction (1) is apparently very rapid. Combination of (1) and (2) gives
the reaction its 3/2 power dependence on diborane concentration. Reaction (3) or (3) presumably is rate controlling. Combination of (4) and
(5) is the recognizable reaction, B3H7 B2Ho + B5Hll HP, which is
questioned by R. Schaeffer and others as a single step and is currently
being investigated. Reaction (5) is nearly an equilibrium, discovered by
Burg and Schlesinger (11) ; it has been invoked (GO) to account for the
sharp braking of the rate of Hz formation as the B5Hl1 concentration
rises. Reaction (6), which niay not be a single step, is the slowest reaction
in the above list. Reactions (5) and (6) account nicely for the later
cleavage of B5H11 to B2H6 which is formed along with BBHs, apparently
with the same activation energy and a t an apparent rate about 2.5 times
the rate of formation of B5Hs (60).Reaction (31, which may be two steps
with B5H13as an intermediate, avoids formation of the electron deficient
intermediate, B3H7.The slow reaction, B4H10 Hz + 2B2Hs should also
be included as a minor reaction ( 1 1 ) . Nevertheless, hydrogen inhibition
appears to merit further study in these reactions.
The importance of further isotope-exchange studies of the boron
hydrides and simple reactions among them can hardly be overemphasized.
Such systematic studies as those by W. S. Koski and his co-workers will
continue to clarify possible mechanisms of these reactions. It is conceivable ( 2 3 ) that all of the hydrides containing BH2 groups, B2Htl,
B4H10, B5Hll, and BSH15, undergo reversible dissociation into BH3 and
intermediates such as BH3, B3H7, B4Hs, and BsH12 which will add BH3,
Hz or react with one another or other hydride molecules. It seems unlikely,
on the other hand, that the more stable relatively highly condensed
B,H,,4 hydrides undergo associations with BH3 to form a B,H,+e
hydride (34). The relation of these rather general speculations to the
pyrolysis is that a t higher temperatures, near 100C, the reactions of
&HI0 may be those of B3H7, and the reactions of BjHll may be those of
B4Hs which could add H2 to give an unstable form of
or
B3H7 H2, as suggested by the reverse of reaction ( 5 ) . Thus the
mechanism of the pyrolysis can hardly be said to be understood clearly,
but the stage of understanding of the structures of various possible
reaction intermediates from the valence theory is sufficiently advanced
that one can hope for some help from the structural concepts in thc
elucidation of the mechanisms and intermediates.
So far the important results of the isotopic-exchange experiments may
be summarized as follows ( 3 3 ) : (a) Exchanges occur in which BH::
remains an entity, as in the BzHs self-exchange; the rate of deuterium
exchange is three tinies the rate of boron exchange. (b) Exchanges occur
151
where BHs exchanges with terminal hydrogen and not with bridge
hydrogen, as in BsH9 and B10H14. (c) Intramolecular exchange appears
to occur in B5H11 and in B4H10,as was first noted in the A1 (BH4)s and
later in BsHs- and BsH7 addition compounds. Very little is known of the
further reactions which lead to B10H14 and polymers but a mass spectroscopic study ( M )of the reaction at 78OC for 24 hr of isotopically enriched
and B5I1H9indicates that half of the boron atoms in the
resulting B10H14 come from BBHg.Also, exchange between B2H6 and
B5H9occurred, possibly involving a pair of B atoms at each step.
Finally, the efficient interconversions of certain boron hydrides into
others by chemical reagents well below room temperatures and under
conditions adaptable to large-scale production is a very significant discovery (8). The starting material, B5H11, which is relatively easily
obtainable from pyrolysis reactions, can be converted in 98% yield to
B4Hlo by water at about 0C according to the reaction
BdIii
+ 3Hz0+ BJLo + 2Hz + B(OH)s.
Also [ (CHs)2N]2BH below 0C converted 0.7 mole of B5Hll into 0.6 mole
of B5Hs, some B2H6 and B4H10, 0.04 mole of B6HlO and small amounts of
other products. This same reagent produces a high yield of B5H9 from
even in small yields by these methods is
B4H10. Production of
especially interesting because it is difficult to obtain by the pyrolysis
method. The low temperatures in these processes give some promise that
new hydrides, less stable than the presently known ones, might possibly
be prepared.
Notes Added in Proof
1. A NEWBORON
CHLORIDE
B8Cls
The structure of the boron chloride referred to at the end of Section

V II I has now been solved, as shown in Fig. 22. The molecular symmetry
is D2#,
and the polyhedron is a new one among small molecules containing
boron, As has been shown (18) in other examples, the three-center bond
description is not appropriate in highly condensed molecules unless resonance is invoked, and hence a molecular orbital description is preferred. A
convenient description starts with sp hybrids along each B-CI bond. The
resulting eight sp hybrids pointing toward the center of the molecule may
then be formed into linear combinations which give two bonding orbitals.
Then the'local T orbitals on each B tnay be also made into linear combinations requiring six bonding pairs. These eight pairs, together with the
eight other pairs in B-CI bonds, and the lone pairs on C1 then give a filled
152
WILLIAM N. LIPSCOMB
FIQ.22. The &CL structure, showing the positions of axes of symmetry. Small
circles represent boron and large circfes represent chlorine.
orbital description of the molecule. As in B4Cl4, the lone pairs on C1 may

interact with low-lying unfilled molecular orbitals of the boron framework (46).
A systematic investigation of molecular orbitals in related B8 polyhedra is now under way in order to see whether this structure is unique
for all reasonable Bs possibilities.
2. T O P O ~OFYB,HbO, COMPOUNDS
The recent report by W. H. Bauer and S. E. Wiberley [Abstr. 1SSrd

Meeting Am. Chem. Soc., San Francisco, p. 13L, (1958)l and by J. A.
Hammond (private communication) of a possibly electron deficient
compound B4Hl20 has suggested the investigation of valence rules
for this and similar compounds. Assume that Bp-,,Hp--n+pOnsatisfies the
same bonding rules as the boron hydrides. Then the equations of orbital
153
halancc, electron balance, and hydrogen balance (18) are
4p = p
- n + 28 + x + 3t + 2y + z
+ 6n = p - n + s + s + 2t + 2y + 22
3 ( p - n)
s+x=q
where z is the number of unshared pairs. From these equations a complete

topological representation of possible formulas may be derived. If desired,
the assumptions that there are no BHs or OH groups can easily be removed, and more general formulas may be obtained.
As a simple pertinent example, suppose that we examine formulas of
the type B4H4+*0.Since there are more orbitals than electrons, it is very
likely that z = 0. Hence these equations become
16 = q s 3t 2y
14 = q 2t 2y
s=x=q
and the s t y 5 numbers are, for all possible q,
q=
14
12
10
8
6
4
2
0
2.0.0.12 1.1.0.11 0.2.0.10 0218 0226 0234 0242 0250
2.0.1.10 1119 1127 1135 1143 1151
2028 2036 2044 2052 2060
Note that q is restricted to the range 0-14, inclusive. If x > 4, there is
an OH group, which seems unlikely, and hence q probably lies between
0 and 6, inclusive. A reasonable type of formula would then require the
oxygen at the center, and hence we surmise that t = 0. With these fairly
drastic assumptions, we then arrive at a 2044 B4Hro0,a 2052 B4Ha0, and
a 2060 B4He0 as likely possibilities. Of these, the first seems most preferable and the second, the least. Thus, we venture a guess that the compound is B4Hlo0with four single B-0 bonds and two bridge hydrogens
each between a pair of BH2 groups.
It is worth noting that in the formula B4H,0 there are two more
orbitals than electrons for all values of g, and hence two three-center
bonds are required.
An example of a compound with more electrons than orbitals is the
hypothetical B2H203 (J. A. Hammond, private communication) for
which n = 3, p = 5 and q = 0. The equations give s = x = 0 and z t = 4, but it is unlikely that there are both unshared pairs and threecenter bonds, and so the most likely bond numbers are s = O , t = 0,
y = 7, x = 0, and z = 4. Two BH groups each joined to three oxygen
atoms are plausible, but there must also be one 0-0bond; but formulas
in which boron is three-coordinated are also possible.
+ + +
+ +
154
WILLIAM N. LIPSCOMB
3. AMMONIATESOF BORONHYDRIDES
In a series of seven papers, R. W. Parry, D. It. Schultz, S. G. Shore,

P. R. Girardot, G. Kodama, R. C. Taylor, and A. R. Emery [J. Am.
Chem. SOC.80, 4 (1958)] have shown, among other things, that B2Ho.
2NHa is H2B(NH,) 2+BH4-. These studies follow an earlier set of studies
by S. G . Shore, R. W. Parry, E. L. Lippert, W. N. Lipscomb, and E. W.
Hughes [ J . Am. Chern. SOC.77, 6084 (1955); 78, 502, 503 (1956)l in
which the structure of the iiionomer was proved to be H3NBH3, now
shown to have a dipole moment of 4.9 Debyes [J. R. Weaver, S. G. Shore,
and R. IT. Parry, J . Chem. Phys. 29, 1 (1958)l.
These studies lend themselves to some generalization. R. W. Parry
(private communication) has suggested that B4HI0-2NH3might conceivably be H2B(NH3)2+B3Hs-. On the other hand, B6H9-4NH3 does not
seem to be based upon known positive or negative ions, and hence an investigation of its structure, as well as investigations of the other ammoniates, would be well worth undertaking.
A further generalization of these and other reactions of boron hydrides
has led Parry and Edwards (private communication) to classify reactions of boron hydrides on the basis of their known structures. The types
of reactions are (a) symmetrical cleavage (to BH3) of a double bridge
bond, (b) unsymmetrical cleavage (to BH2+) of a double bridge bond,
( c ) loss of Hzr and (d) loss of H+. Some of these reactions have been
suggested before on the basis of isotopic exchange studies, and intramolecular rearrangement is not included among these reactions. However, these types of reactions appear to systematize the interpretations
of nearly all of the reactions of the boron hydrides.
ACKNOWLEWMESTS
It is a pleasure to acknowledge discussions of many aspects of boron chemistry
with Professor Riley Schaeffer and to call attention to my principal collaborators,
Professors 1%. H. Eberhardt and Bryce L. Crawford, Jr., and Dr. Richard E. Dickerson in the theoretical papers, as well as Dr. W. J. Dulmage, Dr. C. E. Nordman,
Dr. L. R. Lavine, Dr. I<. Eriks, Dr. F. L. Hirshfeld, Dr. R. E. Dickerson, and Dr.
R. A. Jacobson in the structure determinations.
REFEREXCES
1. Atoji, M., and Lipscomb, W. N., J . Chem. Phys. 21, 172 (1953).
M., and Lipscomb, W. N., Acta Cryst. 6, 547.(1953).
3. Atoii, M., Rheatley, P. J., and Lipscomb, W. N., J. Chem. Phys. 27, 196 (1957).
4. Bauer, S. H., J. Am. Chem. SOC. 72, 622 (1950).
6. Bauer, S. H., Shepp, A., and McCoy, R. E., J . Am. Chem. SOC.75, 1003 (1953).
6 . Beach, J. Y., and Bauer, S. H., J . Am. Chem. SOC.62, 3440 (1940).
2. Atoji,
155
7 . Bloom, A. L., and Schoolery, J. N., Phys. Rev. 97, 1261 (1955).
8. Boone, J. L., and Burg, A. B., J. Am. Chem. SOC.80, 1519 (1958).
9. Bragg, J. K., McCarty, L. V., and Norton, F. J., J. Am. Chem. SOC.73, 2134
(1951).
10. Brown, H. C., Stehle, P. F., and Tierney, P. A., J. Am Chem. SOC. 79, 2020
(1957).
11. Burg, A. B., and Schlesinger, H. I., J. Am. Chem. SOC.55, 4009 (1933).
12. Carroll, Lewis, Bellman in The Hunting of the Snark, p. 11. Peter Pauper Press,
Mt. Vernon, N. Y.
13. Clarke, R. P., and Pease, R. N., J. Am. Chem. SOC.73, 2132 (1951).
14. Coulson, C. A., V. Henri Memorial Volume, Contribution B 1Bude de la
Structure MolBculaire, p. 25. Desoer, Liege, 1948.
16. Dickerson, R. E., and Lipscomb, W. N., J. Chem. Phys. 27, 212 (1957).
16. Dickerson, R. E., Wheatley, P. J., Howell, P. A., and Lipscomb. W. N., J. Chem.
Phys. 27, 200, Footnote 2 (1957).
17. Dickerson, R. E., Wheatley, P. J., Howell, P. A., Lipscomb, W. N., and Schaeffer,
R., J . Chem. Phys. 25, 606 (1956).
18. Eberhardt, W. H., Crawford, B. L., Jr., and Lipscomb, W. N., J. Chem. Phys.
22, 989 (1954).
19. Ford, P. T., and Richards, R. E., Discussions Faraday SOC.19, 230 (1955).
20. Guter, G. A., and Schaeffer, G. W., J. Am. Chem. SOC.78, 3546 (1956); also
private communication from G. W. Schaeffer.
21. Hamilton, W. C., Proc. Roy. SOC.(London) A235, 395 (1956).
22. Hillman, M., Mangold, D. J., and Norman, J. H., Abstr. 133rd Meeting Am.
Chem. Soc, &n Francisco p. 18L (1958).
23. Hirshfeld, F. L., Eriks, K., Dickerson, R. E., Lippert, E. L., Jr., and Lipscomb,
W. N., J. Chem. Phys. 28, 56 (1958).
24. Hawthorne, M. F., and Miller, J. J., J. Am. Chem. SOC.80, 754 (1958).
26. Hough, W. V., and Edwards, L. J., J. Am. Chem. SOC.78, 689 (1956).
26. Hough, W. V., Ford, M. D., and Edwards, L. J., Paper presented before the
Division of Inorganic Chemistry, Natl. Meeting Am. Chem. Soc, New York
(1957).
27. Hrostowski, H. J., and Myers, R. J., J. Chem. Phys. 22, 262 (1954).
28. Hrostowski, H. J., Myers, R. J., and Pimentel, G. C., J. Chem. Phys. 20, 518
(1952).
29. Jacobson, R., Atoji, M., and Lipscomb, W. N., to be published eventually; also
Urry, G., Kerrigan, J., Parsons, T. D., and Schlesinger, H. I., J. Am. Chem.
SOC.76, 5999 (1954).
30. Kasper, J. S., Lucht, C. M., and Harker, D., Acta Cryst. 3, 436 (1950).
31. Kasper, J. S., McCarty, L. V., and Newkirk, A. E., J . Am. Chem. SOC.71, 2583
(1949).
32. Kelly, J., Jr., Ray, J., and Ogg, R. A., Jr., Phys. Rev. 94, 767 A (1954).
33. Koski, W. S., Abstr. 1 S r d Meeting Am. Chem. Soc, San Francisco p. 18L (1958).
34. Koski, W. S., Kaufman, J. J., Friedman, L., and Irsa, A. P., J. Chem. Phys. 24,
nl (1956).
36. Koski, W. S., Kaufman, J. J., and Lauterbur, P. C., J. Am. Chem. SOC.79,
2382 (1957).
36. Lavine, L., and Lipscomb, W. N., J. Chem. Phys. 22, 614 (1954).
37. Lipscomb, W.N., J . Chem. Phys. 22, 985 (1954).
38. Lipscomb, W.N., Discussions Faraday SOC.19, 249 (1955).
39. Lipscomb, W. N., J. Chem. Phys. 25, 38 (1956).
156
WILLIAM N. LIPSCOMB
40. Lipscomb, W. N., J . chim. phyr., 53, 515 (1956).

4 . Lipscomb, W. W., f. Phys. Chem. 61, !23 (1957).
48. Lipscomb, W. N., J. Phys. Chem. 62, 381 (1958).
.@. Lipscomb, W. N., J . Chem. Phys. 28, 170 (1958).
44. Lipscomb, W. N., Dickerson, R. E., Hirshfeld, F. L., and Trefonas, L., Acta
C w s t . 10, 777 (1957).
46. Longuet-Higgins, H. C., J. chim. phys. 46, 275, discussion (1949).
46. Longuet-Eggins, H. C., Quart. Rev. (London) 11, 121 (1957).
47. Longuet-Higgins, H. C., and Roberts, M. de V., Proc. Roy. SOC. (London)
A230, 110 (1955).
48. Maybury, P. C., and Koski, W. S., J. Chem. Phys. 21, 742 (1953).
49. Moffitt, W., Proc. Roy. SOC. (London) A1%, 510 (1949).
60. Moore, E. B., Jr., Dickerson, R. E., and Lipscomb, W. N., J. Chenz. Phys. 27,
209 (1957).
61. Nordman, C. E., Acta Cryst. 10, 777 (1957).
52. Norment, H. G., Abstr. 4th Intern. Congr. Crystallography, Montreal Paper
6.4 (1957).
63. Ogg, R. A., Jr., J . Chem. Phys. 22, 1933 (1954).
54. Ogg, R. A., Jr., and Ray, J. D., Discussions Faraday SOC.19, 239 (1955).
66. Phillips, W. D., Chambers, W. J., Miller, H. C., and Muetterties, E. L., private
communication.
66. Price, W. C., J. Chem. Phys. 17, 1044 (1949).
67. Rundle, R. E., J . Chem. Phys. 20, 132 (1952).
68. Schaeffer, R., private communication.
69. Schaeffer, R., J. Am. Chem. SOC.79, 2728 (1957).
60. Schaeffer, R., J. Chem. Phys. 26, 1349 (1957).
61. Schaeffer, R., Schoolery, J. N., and Jones, R., J. Am. Chem. SOC.79, 4606
(1957).
68. Schaeffer, R., Schoolery, J. N., and Jones, R., J. Am. Chem. SOC.80, 2670 (1958).
63. Scherer, J. R., and Crawford, B. L., Jr., J. Chem. Phys. to be published.
64. Schoolery, J. N., DiScussiom Faraday SOC.19, 215 (1955).
66. Shapiro, I., and Keilin, B., J. Am. Chem. Soc. 77, 2663 (1955).
66a. Sidgwick, N. V., The Chemical Elements and Their Compounds, p. 338. Oxford Univ. Press, London and New York, 1950.
66. Silbiger, G., and Bauer, S. H., J. A m . Chem. SOC.68, 312 (1946).
67. Soldate, A. M., J. Am. Chem. SOC.69,987 (1947).
68. Stock, A., Hydrides of Boron and Silicon. Cornell University Press, Ithaca,
N. Y., 1933.
69. Stock, A., and Laudenklos, H., Z. anorg. Chem. 228, 178 (1936).
70. Stock, A., Sutterlin, W., and Kurzen, F., Z. anorg. Chem. 225, 225, 243 (1935).
7 1 . Trefonas, L., and Lipscomb, W. N., J. Chem. Phys. 28, 54 (1958).
72. Wartik, T., private communication.
73. Wartik, T., Urry, G., and Schlesinger, H. I., 1. Am. Chem. SOC.74, 5809 (1952).
74. Watley, A. T., and Pease, R. N., J . Am. Chem. SOC.76, 835 (1954).
76. Yamazaki, M., J . Chem. Phys. 27, 1401 (1957).
LATTICE ENERGIES AND THEIR SIGNIFICANCE

IN INORGANIC CHEMISTRY
. .
T C Waddington
The Uoiversity Chemical laboratories. Cambridge. England
I. General Introduction and Account of the Born-Haber Cycle . . . . . 158

I1. The Calculation of Lattice Energies . . . . . . . . . . . .
A. The Extended Clasaical Calculation . . . . . . . . . . .
1. The Madelung Term. U X . . . . . . . . . . . . . .
2 . The Repulsive Energy Term. -UR . . . . . . . . . .
3. The Dispersion Energy (London) Term. U L . . . . . . . .
4 . The Zero-Point Energy Term. UE . . . . . . . . . . .
5. The Permanent Electrical Multipole Term. UQ . . . . . . .
B . The Kapustinskii and Templeton Calculations of the Lattice Energies
of Ionic Crystals . . . . . . . . . . . . . . . . .
1. Kapustinskii's Formula and the Thermochemical Radius . . . .
2. Templeton's Prediction of Madelung Constants . . . . . .
C . The Quantum Mechanical Prediction of Lattice Energies . . . .
D . The Calculation of Lattice Energies from Hydration Enthalpies . .
I11. Individual Values of the Lattice Energies of Alkali Metal and Alkaline
Earth Salts . . . . . . . . . . . . . . . . . . . .
A. The Alkali Metal Halides . . . . . . . . . . . . .
B . The Alkali Metal Hydrides . . . . . . . . . . . . .
C . The Alkali Metal and Alkaline Earth Chalcogenides . . . . .
D The Alkali Metal Hydroxides . . . . . . . . . . . .
E . The Alkali Metal Hydrosulfides . . . . . . . . . . .
F. The Alkaline Earth Imidea . . . . . . . . . . . . .
G. The Alkali Metal Amides . . . . . . . . . . . . .
H . The Alkali Metal Cyanides . . . . . . . . . . . .
I. The Alkali Metal Borohydrides . . . . . . . . . . . .
J . The Alkali Metal Superoxidea . . . . . . . . . . . .
K . The Alkaline f i r t h Peroxides . . . . . . . . . . . .
L . The Alkaline Earth Acetylides . . . . . . . . . . .
M . The Alkali Metal h i d e s . . . . . . . . . . . . . .
N . The Alkali Metal Bifluoridea . . . . . . . . . . . .
0. The Alkali Metal Cyanates . . . . . . . . . . . .
P The Alkali Metal Thiocyanates . . . . . . . . . . .
Q . The Alkali Metal Nitrates . . . . . . . . . . . . .
R The Alkaline Earth and Other Metal Carbonates . . . . . .
S. The Alkali Metal Borofluoridea . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
162
162
162
171
174
176
176
177
177
179
181
186
188
189
189
192
192
195
196
196
196
197
197
198
198
198
199
199
200
201
202
203
IV. Uses of Calculations of Lattice Energies . . . . . . . . . . . 203

A. AHt%-(g) and the Determination of Electron Ailhities . . . . . 203
.B. The Determination of Absolute Values of Proton Affinities . . . . 204
157
I 58
7. C. WADDINGTON
C. The Absolute Enthalpies of Formation of Complex,Ions

. .
D . Lattice Energies and Nonionic Contributions to Bonding
in Crystal Lattices . . . . . . . . . .
. . . .
1. The Halides of Univalent Metals (Other than the
Alkali Metals)
. . . . . . . . . . . . . . . .
2. The Halides of the Divalent Metals . . . . . . . . .
3. The Chalcogenides of the Monovalent Metals (Other than the
Alkali Metals)
. . . . . . . . . . . . . . . .
4. The Chslcogenidcs of the Divalent Metals . . . ; . . . .
5. The Oxides of Some Trivalent Metals . . . . . . . . .
6. The Oxides of the Tetravalent Metals . . . . . . . .
E. The Determination of the Stability of Hypothetical Compounds .
F. Lattice Energies and the Effect of Fluorides and Oxides on the
. . . .
. .
Osidntion State of Metal Salts . . .
References
206
206
207
208
.
.
.
.
.
211
211
213
213
214
218
218
I. General Introduction and Account of the Born-Haber Cycle
This review article gives an account of the calculation of the lattice

energies of ionic solids, the results of such calculations for a large number of salts and the application of these results and calculations to 8ome
prohlems in inorganic chemistry. No attempt has been made t o deal with
the many important topics, involving lattice energy calculations or calculations of a similar type, which lie outside the region of inorganic
chemistry. Examples of such topics are the calculation of the surface
energies of crystals, the energies of formation and of movement of point
defects in ionic solids and the crystal spect58 of ionic solids.
The last comprehensive review of lattice Gnergy calculations was produced by Sherman (114) in 1932. Since then no extensive review of results and calculations has been undertaken, though there have been
several accounts of various aspects of the subject. Pritchard (108) in
his review of values of electron affinities has discussed a number of lattice energies. Kapustinskii (67) has given a brief and uncritical account
of the work of the Russian school in this field. There is an excellent
chapter on the calculation of the lattice energies of the alkali halides
in Seitzs (113) book, and Partington (105) has given an account of
some methods of calculating Madelung constants. Born and Huang (17)
have given a very full description of the Ewald (37) method of obtaining
these constants.
Historically the quantitative theory of ionic crystals was developed
between 1918 and 1924 by Born (14), Born and Lande (19, ,%
Made-I),
lung (88) and Haber. In particular Born devised formulae which permit
the calculation of the lattice energy of an ionic crystal. The lattice energy
of such a crystal may be defined as the increase in internal energy a t ab-
LATTICE ENERGIES
159
solute zero accompanying the separation of the constituent ions to positions where they are infinitely removed from one another. In order to
account for the stability of crystals known to be composed of ions it is
necessary to introduce forces between the ions that are noncoulombic,
for no stable equilibrium is possible in an electrostatic system of charges
unless other forces are present.
The history of the development of the theory of lattice energies is
largely an account of the development of the ideas about these noncoulombic forces. Though they are discussed in detail in Section I1 it
is convenient to summarize the various expressions for the lattice energy
involving these forces that have been developed and to designate them
by the names that will be used for them in later sections.
(a) Born-Lande or Born Equations
Here N A is Avogadros number, M is the Madelung constant,, z1 and z2

are the valencies of the ions, e is the charge on the electron, r is the distance between unlike ions, and B is a constant. If the fact that at the
equilibrium distance r0 of the ions in the crystal (dU/dr),,, = 0 is utilized
then
(b) Simple Born-Mayer Equation
Here B is another constant and p has the dimensions of a length. Again

if (8U/8r)r=To
= 0 is utilized
(c) Extended Born-Mayer Equation
Here iVAC/r6 and (9/4)Ndhvmnr are terms introduced to allow for the
induced dipole-induced dipole forces in the lattice and for the zero point
energy of the lattice respectively.
(d) Huggins Treatment
This treatment is similar to the extended Born-Mayer exccpt that
adjusted crystal basic radii are used to obtain B and optical data on
the crystal itself to obtain C .
160
T. C. WADDINGTON
(e) Ladd and Lee's Equation

Because of the difficulty of assigning basic radii in many salts,
(aU/ar),,, = 0 is used and the expression
is derived.
Quite apart from its theoretical calculation, by the use of one of the
expressions developed above, it is possible to relate the lattice energy of
an ionic crystal to various measurable thermodynamic quantities by
means of a simple Hess's law cycle. This cycle was first proposed and
used by Born (15) and represented in its familiar graphical form by
Haber (48). It is now usually referred to as the Born-Haber cycle. The
cycle is given below for a uni-univalent salt in terms of enthalpies.
+X in. n o r m o l ~ - ~N9)~- o
chemical stote
'4
xiof
ot 2 5 O
All the magnitudes refer to 25C and 1 atm.

AH,OMX(S) is the enthalpy of formation of the crystalline salt.
AH,OMX(aq) is the ent,halpy of formation of the salt in aqueous
solution at unit activity.
.%Hao is the standard enthalpy of solution of the crystalline salt to
unit activity.
U is the lattice energy of MX at 298.3"K (25C).
U 2RT is the corresponding lattice enthalpy.
H is the enthalpy of hydration of the two gaseous ions to unit activity.
E is the electron affinity of the X radical.
161
LATTICE ENEBGIES
I is the ionization potential of the metal atom.

AH,OX(g) is the standard enthalpy of formation of the X radical in
the gas phase.
L is the enthalpy of sublimation of the metal M.
From this cycle the following independent relations may be derived.
(a) Between the lattice energy and the combined ion-hydration enthalpies
U = -H+AHa0-2RT
(7)
(b) Between the lattice energy and the electron affinity
AHfoX(g) - E = U 2RT - I - L AHf'MX.

(8)
The quantity AHtoX(g) - E is often called AHjoX-(g) and the
quantity I L, AH,OM+ (g) as can be seen from the equation
AH = U 2RT
MX(s) + M+(g) X-(g),
(9)
whence U 2RT = AH,OM+ (g) AH,OX- (g) - AHjoMX(s). It must
be remembered that U here is U298-I(,the lattice energy at 25"C, and that
this is related to the lattice energy UOa t 0K by
+l
T
UT = Uo
+ C,(X-)
[C,(M+)
- C,(MX)]-dT
- 2RT.
(10)
It is usual to assume that the specific heats C,(M+), C,(X-) are 5/2 R
cal/mole in which case the relation becomes
UT = Uo
LT
[3R - C,(MX)]*dT.
(11)
It must be remembered that the lattice energy given by the BornHaber cycle is an experimental lattice energy and is not .dependent upon
the nature of the assumptions made about the bonding in the crystal. The
classical theoretical calculations are of course dependent upon the assumption of the ionic nature of the bonding in the lattice. Because of
this the Born-Haber cycle has been used mainly for three purposes.
(a) When it seems reasonable to assume that the lattice is ionic the
cycle is used to calculate AHtoX-(g), = AHtoX(g) - E , from U, and
hence to calculate either E or AH,OX by the use of subsidiary data. An
indirect check on the validity of the assumption of ionic nature and on
the accuracy of the calculation is provided by the constancy of the
value of AHf0X(g) - E from salt to salt.
(b) When the lattice may not be totally ionic the cycle is used to
derive an experimental value for the lattice energy, which may then be
compared with that given by the ionic theory. If the experimental lattice
162
T.
C . WADDINGTON
energies are systematically higher than those obtained from the classical
ionic theory there is probably an appreciable nonionic contribution t o
bonding in the crystal.
(c) If an approximate ionic structure can be assigned to an unknown
or hypothetical compound then its lattice energy can be calculated and
the Born-Haber cycle often used to derive a value for its enthalpy of
format ion.
ti. The Calculation of lattice Energies
A brief account will be given in this Section of the various methods

that have been employed to calculate the lattice energies of ionic crystals, starting with a discussion of the extended classical calculation. Each
term in this calculation will be discussed separately.
A. THEEXTEXDED
CLASSI('AL
CALCCLATION
1. The Madelung Term, U.v
This term arises from the electrostatic interaction of the point charges
with which the ions have been replaced in the theoretical consideration
of the problem. The interaction energy of two point charges zle and z2e,
distance r12 apart is z1z2e2/r12. Similarly, the total electrostatic energy
tTAfof n such charges of magnitude zd (i = 1,2,3,. . . , n) is
in which the summation extends over all pairs of charges, each pair being
considered. This may also be written in the form
where the summation is now a double sum over all charges and the superscript prime indicates that the cases i = j are to be excluded. For
binary crystals such as sodium chloride, sodium nitrate, and calcium
fluoride, the results can always be expressed in the simple form
Here I'af is the energy per mole; z+e and z - e are the absolute values
of the charges on the positive and negative ions; 1 is one of the characteristic crystal dimensions; K.4 is the Avogadro number and M i is the
Madelung constant, a pure number characteristic of the crystal structure
and independant of the dimensions of the lattice.
LATTICE ENERGIES
163
The series for the Madelung constant cannot be evaluated by uncritical summation because it converges with extreme slowness. Two
general methods are available for summing the Madelung series; the
first is to arrange the terms so that summation takes place over electroneutral layers and the series converges sufficiently rapidly to be summed
directly; the second method is to replace the point charges with a distributed charge and then use a mathematical manipulation to obtain a
quickly convergent series. Both methods have been used though the first
is only convenient in the case of crystals of high symmetry. Kendall (73)
was the first worker to obtain a Madelung constant by the first method
and later Evjen (36) evaluated the Madelung constants of NaCl, CsCI,
ZnS, and CaC03 by considering the lattice to be built up of cubical or
rhombohedral shells around a central ion and by summing over each in
turn. However in the case of CsCl and CaC03, de Boer (25) and Krishnan
and Roy (75) have pointed out that a correcting term for the residual
electrical double layer has to be added because, since these crystals are
body centered, the cubes and rhombohedra around the central ion are not
electrically neutral but alternate in sign. The magnitude of this correcting term has been calculated by the above authors. Hojendahl (52) used
a similar method of summation to Evjen, to obtain the Madelung constant of NaCl and of NaCl lattices distorted along a trigonal axis. By
treating CsCl and CaC03 as distorted NaCl lattices he summed over
neutral rhombohedral shells and avoided Evj ens difficulty. The method
of summation recently described by F. C. Frank (39) is essentially of
this type.
The first actual evaluation of a Madelung constant was made by
Madelung (88) who used the second method. He represented an infinite
row of charges with periodic repetition by a Fourier series. The potential for a point outside the row has the same periodicity as the charge
distribution. By extending the above method, an expression for the potential of a plane of alternating charges may be derived. The potential
for the entire crystal is found by decomposing the lattice into neutral
rows of points and lattice planes and then calculating the potential due
to each plane and row. Madelungs method though giving a rapidly converging series cannot be applied in many cases because of the impossibility of decomposing msny lattices into neutral lines and planes.
Born (16) developed a method of computing the Madelung series of
more general applicability. The unit cell was divided into s subcells by
means 9f systems of equidistant planes parallel to the sides of the cell
60 that each ion in the unit cell was located at one of the sublattice
points (PI/s, Pg/s, Ps/s) where PI, P2, and Pa are integers which may
assume the values 0, 1, 2, 3, . . . , 5-1. For R cubic crystal the Madelung
164
T. C. WADDINGTON
constant is then given by
where p is the number of stoichiometric molecules in the unit cell. S p is

defined by
Born called n ( P / s ) E n(Pl/s, P2/s, P 3 / s ) the Grundpotential.

It is the potential a t the origin of the unit cell of a charge distribution
cos 2*(mP1+ mzP2
+ mZ3)
and is defined by the formula
The superscript prime indicates that the cases tl = t z = t s are to be

excluded. The Born Grundpotential is a special case of the generalized
zeta-function developed by Epstein (33, 34) and Emersleben (30,31)
has evaluated the zeta-function and hence the Grundpotential II(zyz)
for a cubic lattice for all cases from n(O00)to n(1/21/242)in steps of %.
He used the ~ ( Z V Z so
) obtained to calculate very accurately the Madelung constants for the cubic lattice types NaC1, CsC1, ZnS, CaFz and
CueO. Relations between various grundpotentials have been shown to
exist by Hund ( 5 5 ) , by Emersleben (32)and by Sakamoto (111) who
has summarized them and recalculated a number of Emerslebens original values.
Ewald (S7) developed a method for evaluating the Madelung constants and other sums for crystal lattices which depends on the use of
a 6-function transformation. This is probably the method of most general application though it suffers from the disadvantage that the convergence of the series obtained is less rapid than that obtained by the
Emersleben method.
An account of the 6-function transformation and the evaluation of
Madelung constants by this method has been given recently by Born and
Huang ( 1 7 ) . Bertaut (8, 9) has shown that if the atoms of an ionic
crystal are replaced by spherically symmetrical charge distributions,
f(r), which do not overlap, the Madelung constant can be expressed as
a single, absolutely convergent infinite series whose terms are a function
of the lattice vectors in reciprocal space. Ewalds series emerges as a
165
LATTICE ENERGIES
particular case of Bertaut's approach to the problem. For the case where
f ( r ) = a ( R - T)" an expression of the form
is obtained.
Here M1 is the Madelung constant based on 1 as unit distance, p is
the number of molecules in the unit cell, zIis the charge number of atom
j, V is the volume of the unit cell and h is the magnitude of the vector
(hl, hz, hs) in reciprocal space or the reciprocal of the spacing of the
planes (hlhzh3). The coordinates of atom j are zlj, zzj, ZSj. The sums
over j are taken over all the atoms in the unit cell. F(h) is the Fourier
transform of the Patterson function and cp(h) is the Fourier transform
of the charge distribution f(r).F ( h ) is given by
Templeton (117) has examined the rate of convergence of Bertaut's series

for various functions f(r) and the errors obtained by termination after
varying numbers of terms. Values of g and +(h) are given in Table I
fgr various functions f ( r ) . From Templeton's. results it is clear that the
assumption of a linear charge distribution gives the most rapid convergence down to an accuracy of 0.1% in the Madelung constant.
TABLE I
CHARQEDISTRIBUTIONS
AND THEIR
FOURIER
TRANSFORMS
Atomic shape
Uniform
Linear
Parabolic
Cubic
I@), r < R
3
4rR3
?rR4
-' 2rR6
uR6
+(h)
7!
3/5
18(sin a
- a cos a)**
us
- 2 cos a - 2)'
26/35
288(a sin a
25/38
7200(a COB a
a10
23/22
- 3 sin a + 24'
a12
259,200(a1 ia sin a
+ 4 cos a - 4)*
ffl'
* a = 27rhR.
If, instead of one of the simple functions given in the Table, which
are equal to zero for all T > R,a normalized Gaussian distribution is taken,
) kS exp (-k2 d),
the Bertaut equation must be corrected
that is, f ( ~ =
to allow for the overlap of the charge distributions. In this case Bertaut
166
T. C. WADDINGTON
TABLE I1
NUMEUICAL
VALVES
C r p t a l type
OF THE h1ADELUNG COXSTANTS FOR VARIOUS ClZYSTAL
Space Group
LATTICES
df,,
hfdo
M",
References
30, 51,SZ?, 73, 88,5 ,
NaCl
Fin3m
1.74756
2.20178
3.49513
CS"1
Pm3m
F43m
C6mc
Fm3m
Pn3m
1.76267
1.63805
1.64132
5.03878
4.44248
2.03536
2.38309
2.03536
3.78295
2.386
11.63657
10.25946
109,111,
Sphalciite
Wurtzit c
CaF2
CU?O
7.33058
50, 31,6,111
SO, s1,111
47, 54,55,6,111
SO,S l , 77,6,13,111
SO, 31, 19, 115, 56,
111
Hutile
Anntase
CdI,
&Quartz
A1203
Perovskite*
Cnlcitc
Arngonitc
CaSO4
BaSO,
SrSOr
PbSOI
I(X3
NnX3
PtS(C,cJ = 1 )
PdO(c/u = 2)
PdO(e/u = 2 )
NbO t
NaTl
cu,vs,
Ca+(W04)( c / u = 1)
(ClU = 2 )
(ci. = 2 )
A3+B3-$
BiF&
ZrI,
Y OF
LaOCl
Spinels*
P4/mnin
C4/amt
HIlm
H622
R3c
Pmtm
See following text

See following text
See following text
See following text
See following text
See following text
Fmsm
Fmam
I4/mmnt
R3c
39.99634 63.49026
33.39098 53.00488
See following text
See following t e x t
12.84724
13.50363
13.18779
13.15651
13.16377
See following text
See following text
PmCm
Bhm
Pnma
Pnmu
Pnma
I 4 / m
R3m
1.580'206
12
12
55
58
112
22,5i,55,111,114,
118
114,8i
114
$0
5.9, SG
109
109
109
109
109
60,186,127, 128
56
51
P4,'ntttin
Fd3m
1.604935
1.504265
1.513429
10.398386
111
51
51
111
I4Ja
1.594364
1.613972
1.633580
9.566187
9.57905
13.75454
1.615503
51
51
51
61
51
51
See following text

18.75
See following text
118,61
118
'6,27
* Space group for the undeformed, cubic lattice given.
t NaC1-type, vacancies at 0, 0, 0 (Nb) and 4, 4, + (0).

4 Cubic. A at 0, 0, 0; B I at 3, 4, 0; B I I at 4, 0, ?j;B I I Iat 0, +,f.
5 Cubic face-centeled, Bi at 0, 0, 0: FI a t f , f, f ; F I I at f, f , f ; F T IaI t 4, 0, U.
167
LATTICE ENERGIES
has shown that
--czi2-kl
/ -
2rvp
exp ( - r h 2 / k Z )
h
L
-22
ij
3
erfc (d;krij).
r,j
(20)
This expression is equivalent to the double series of Ewald (S7). The

choice of the constant k is arbitrary, zi and z j are the charge numbers
of the ions i and j and rij is the distance between them.
Recently Benson (6) has described a new and relatively simple formula for calculating the Madelung constants of a NaC1-type crystal
and Benson and.Van Zeggeran (6) have extended this method to CsCltype crystals. The method appears to be capable of extension to generalized CsCl and NaCl type crystals.
The results of the application of these calculations to various types
of crystal are discussed below. Table I1 summarizes the values of
Madelung constants obtained for different crystal lattice types. The
general formulae which have been developed to express the Madelung
constants as iunctions of various lattice parameters and the variation
of Madelung constants with crystal parameters are discussed in more
detail below.
The results of Bollnow (19) for the Madelung constant of the rutiletype structure are expressible in the form
- 4.11(0.721 - a)2
(21)
for values of the axial ratio, a = c/a, in the neighborhood of 0.721. For
anatase the Madelung constant may be written (1.2) as
M , = 4.800 - 0.707(2.620 - a)*
(22)
M,
= 4.816
where a is again the axial ratio, c/a.

For &quartz, which has an hexagonal lattice, Hylleraas (58) has tabulated M,, as a function of two parameters, the axial ratio cla and a parameter u, the distance of the oxygen ions from the hexagonal axis divided
by a. His values are given in Table 111.
TABLE I11
VALUES OF k f r 0 FOR THE
__
u=&
u=+
= 3%
u=+
QUARTZ
STRUCTURE
~~~
~~
c/a = 1.0
c/a = 1.1
c/a = 1.2
4.4216
4.4261
4.2209
4.1754
4.4275
4.4303
4.2540
4.4288
4.4317
4.2862
cla = 1.414
c/a = 1.732
4.4248
4.3998
168
T. C. WADDINGTON
Hund (66) has evaluated M , of the hexagonal cadmium iodide structure as a function of the axial ratio c/a and u, the z-coordinate of the
iodide ion divided by c. The values he obtained are given in Table IV.
TABLE IV
VALUESOF M,
=o
u = T1B
u = i
1L
CADMIUM
IODIDESTRUC~URE
FOR THE
cla = 0.815
c/a = 1.12
c/a = 1.61
c/a = 2.00
4.55
4.53
4.58
4.68
4.63
4.65
4.72
4.73
4.64
4.68
4.74
4.71
4.64
4.72
4.73
3.90
cla = 4.00
4.64
4.68
3.87
-2.06
Schmaeling (112) has evaluated the Madelung constant of the rhombohedral corundum, Alz03, structure as a function of two parameters,
the approximation having been made that each aluminum atom is
equidistant from six oxygen atoms. The two independent variables are
a = c/a and w = v/a, where a is the diagonal of a rhombohedra1 face,
c is one-half the length of the body diagonal, and v is the distance
between adjacent aluminum atoms. The Madelung constant is then
given by
- 65.250(0.5454 - w ) ~
+ 30.70(1.312 - a)(0.5454 - w).
311, = 25.0312 - 5.930(1.312 - a)'
(23)
Yerovskites are compounds of the general formula ABX3, having a

cubic lattice with the atoms a t the following positions: A at O,O,O; B at
1/2,1/2, 1 / ; and 3X a t 0, 1/, 1/;1/2,0, 1 / ; 1/, 1/2,0.
The type of perovskite
with a charge of + 1 on A, +2 on B, and -1 on X will be called a 1-2
type. If the charges on the cations are interchanged it is called a 2-1
type. In an analogous way one can define 1-5,5-1,3-3, 2-4, and 4-2 types
(in which X has a charge of -2). It will be obvious that the Madelung
constants of 2-4,4-2 and the 1-2,2-1 types merely differ by a factor of 4.
Compounds do occur in which no A atoms are present but in which the
B and X atoms occupy their places in the perovskite lattice. Such compounds, for example, Re03 and ScF3 can be considered as 0-6 and 0-3
perovskites. The Madelung constants of the perovskites have been discussed by Sherman (114), by Hund (55), by Brauer (99), by Templeton
(118), by Hoppe (51) and by Sakamoto (111). Their results are concordant and are summarized in Table V.
169
LATTICE ENERGIES
TABLE V
MADELUNG
CONSTANTS
M,
FOR
DIFFERENT
TYPESOF PEROVSKPTE
~~
Type
Example
Brauer
Hoppe
0-6
1-5
ReOs(ScFj)
NaTaOs
SrTiOs(KZnFs)
LaAIOs
LiBaFa
71.616
58.536
49.508
44.544
53.672
46.856
54.116
71.6316
2-4
3-3
4-2
5-1
6-0
Templeton
Sakamoto
____
71.6315
58.53535
12.37744*(1-2) 12.377468 (1-2)
44.65489
10.91768 (2-1)
46.85727
* Shermans value; Templetons values are based on Shermans value and M, for
CsCl and NaC1.
For the calcium-carbide type of lattice Kazarnovskaja (YO) has expressed the Madelung constant, Msoas a power series in a( =c/a).
Ma, = 2.2018
+ 1.467(a - l)z- 5.919(a - 1)3+ 18.96(a - 1)4
- 24.3(a - 1).
(24)
For the KHFz structure, which is a CsCl structure compressed along

a tetrad axis, Madelung constants have been calculated for a number of
different values of the axial ratio c / a in the pseudo-unit cell (60, 126,
197). However the author (128) has developed a general expression,
M,,= 2.0354 + 0.33923(1 - a) - 1.619(1 -
(25)
for the Madelung constant with the value of a. in the pseudo-unit cell
containing only one molecule as the reference distance. This formula can
be generalized to the simple body-centered orthorhombic lattice to give
an expression,
M,
= 2.0354
+ 0.33923[(1 - a)+ (1 - a)] + 0.3547(1 - a) (1 - p2)

- 1.619[(1 -
+ (1 - @)]
(26)
where a = c/a, p = b/a and a is the largest cell side. M a,is the Madelung
constant based on a.
Hojendahl (69) has obtained general expressions for the Madelung
constants of crystals of the calcite type, which he regards as NaCl lattices compressed along one trigonal axis, and of crystals of the sodium
bifluoride type, which he regards as NaCl lattices elongated along one
trigonal axis. The standard distance in the crystal lattice is taken as the
distance, Lo,between neighboring layers along the trigonal axis. The nearest distance between ions in the same layer is L1.The lattice can then
170
T. C. WADDINGTON
be determined by one parameter p = L1/Lo.The Madelung constant is

given by
JfL, =
$0.3573
1
+ 2.085 -1 - 1.9185 71 + 0.7754 71 - 0.1286 7'
P2
(27)
The spinels are compounds of the type ABzOl in which the O= ions
are arranged in cubic close packing, with the A atoms occupying tetrahedral holes and B atoms octahedral holes. Verwey, de Boer, and Van
Santen (26, 27) have calculated the Madelung constants, by the Evjen
method for the various values of the parameter u given and the various
charges p and q on A and B. The values p = 3, q = 234%represents the
case where A is A3+ and the B positions are alternately filled with 3 f
and 2+ ions, for example, Fe304. The Madelung constants are given in
Table VI.
TABLE VI
~IADELCXG
COXSTASTS,lIJo0, FOR
Charge
Parameter
u
0 3i.j
0.380
0.385
0.300
4
2
3
4
2
3
4
2
3
4
->
3
SPINELS*
31
__---
_.
~~
THE
P
~
2
3
2:
2
3
'
-)
2I
>
3
2;
2
3
- 2I
'>
I1
~~
138.1
128.6
130.7
135.8
131.1
131.0
133.6
134.0
131.2
131.5
136.5
131.1
138.2
128.6
130.8
* I values by Evjen method, I1 values by Ewald method

for the ideal case (i.e. a perfwtiy cubir Inttice) u = 0.375.
Yttrium oxyfluoride and several similar compounds have the fluorite

type of structure with the oxygen and the fluorine distributed among
the anion positions. Both Templeton (218) and Hoppe (51) have considered the Madelung constants of the crystals. There are three forms,
tetragonal, rhombohedral, and cubic. Their Madeluiig constants arc
given in Table VII.
171
LATTICE ENERGIES
TABLE VII
THE MADELUNG
CONSTANTS,
Mi,
1
OF THE
YOF LATTICES
(4)a.o
Structure
Mt
Tetragonal
Rhombohedral
Cubic, with La at 0, 0,O
0 at t , t , f and F at S, t, t
Cubic, with the 0 and the F
arranged statistically
11.504
11.526
1I .71568
~~
2. The Repulsive Energy Term,
11.33725
-UR
In order to account for the stability of ionic crystals it is necessary

to introduce repulsive forces between the ions in addition to the ordinary
coulombic forces. Born and Lande (201,in their original treatment of
the problem, represented these forces as varying as the inverse nth power
of the distance and treated the potential energy of the crystal as the sum
of two terms, given by the expression
Initially they tried to calculate the value of n from the Bohr model
of the atom and then from the Lewis-Langmuir model. However the
inapplicability of these models was soon realized and the value of n was
determined empirically from the compressibility.
The constant B can be eliminated by remembering that at the equilibrium separation in the crystal, roJ (aU/ar) = 0, so that
Hence
B=
Mzlzze2rOn-l
n
(28)
and
uo = N ~ M z l z 2 e(1 - l/n).
To
(2)
The development of wave mechanics led to the realization that ions

with completed subgroups were spherically symmetrical and that, for
the outermost shell the electron density fell off exponentially with distance. Unsold (120) and Pauling (105) pointed out that the application
172
T. C. WADDINGTON
of quantum mechanics led to an expression for the repulsive potential

T. Born and Mayer (21) therefore
expressed the repulsive energy, ur, between two ions in a crystal as
which involved exponential terms in
u, = a exp (-/PI
(29)
where u and p were constants and T was the distance apart of the ions.
They found that for the alkali halides they could adopt a constant value
for p ( = 0.345 x
, if they determined a from the equation
where b was a constant for all the alkali halides, zu and zA were the
valencies of the two interacting ions (equal to -1 and +1, respectively,
for the alkali halides), nA and nc were the number of electrons in the
outer shells of the ions and rAand To were the ionic radii, as determined
by Goldschmidt (40). Mayer and Huggins (94) later used the condition
that a t the equilibrium internuclear distance in the crystal (dU/dr) = 0
to determine a set of self consistent basic radii which when put into
the equation for the lattice energy produced the equilibrium internuclear
distance. By using 9 such radii they calculated the 20 lattice constants
of the alkali halides to nearly the probable experimental error of their
determination. Huggins (53) later recalculated the lattice energies of
the alkali metal halides using these basic radii. The expression for the
repulsive energy U Rwas given by
bc+F[k
exp
(2rc - k t r ~ ) ] (31)
where b is given the value 10-l2 ergs; c+-, c++, and c-- are the expressions (1 Z C / ~ C Z A / ~ A )introduced by Pauling (106); M and M
TABLE VIII
VALUESOF
TEE
Ion
Basic radius
10- cm
Li+
Na+
K+
Rb
Csf
0.475
0.875
1.185
1.3m
1.455
HUGGINS
BASICRADII
Ion
FC1-
BrI-
Basic radius
lo-*em
1.110
1.475
1.600
1.785
LATTICE ENERGIES
173
are the number of nearest like and unlike neighbors respectively; ka is

the ratio of the shortest distance between like neighbors to that, TO between unlike neighbors and rc and rA are now the Huggins "basic radii."
The values of Huggins basic radii are given in Table VIII. If only the
coulombic and repulsive forces are considered then the lattice energy
may be written as
and utilizing (aU/ar) = 0, B1 can be eliminated to give an expression

for U at the equilibrium distance in the crystal.
These two expressions are sometimes called the simple Born-Mayer expressions for the lattice energy of a crystal.
Because of the dificulty of assigning basic radii in some crvdals,
notably those of the halides of silver and thallium, Ladd and Lee (76)
have extended the above expression, which eliminates B' and hence the
basic radii, to the case where other forces, that is, dispersion energy terms,
are included. They obtain for the lattice energy UO
The values and meaning of C and D and hv,,, will be discussed in the
next section.
Ladd and Lee's expression leads to an interesting observation about
the accuracy of the simple Born-Mayer expression. If it is used in the
form
then errors are introduced by the neglect of the dispersion energy terms
NAC/T" and NAD/t.8. These may be as large as 10 kcal/mole. However
if the simple Born-Mayer expression is used in the form
then the errors arise from the neglect of (1 - 6p/ro) NAC/rO6 and so
forth. Now p = 0.345, so that for most salts, particularly those of the
heavier alkali metals (1 - 6 p / ~ o ) is less than 0.4 so that the error has
been reduced to 4 kcal/mole or less.
174
T. C. WADDINGTON
All the methods discussed above apply only to spherical ions, and
are dependent on the assumption of spherical forces and isotropic compressibilities. When the ions are nonspherical, some assumption has to
be made about the shape of their repulsion envelopes so that radii in
different directions can be assigned. Gray and Waddington (43) and
Waddington (127)have discussed this problem with relation to the nonspherical azide, N 3 - , and bifluoride, HF2-, ions. The calculation of the
repulsive energy is also dependent upon some knowledge. of the coefficient
of compressibility of the crystal. It is true that for the alkali halides the
assignment of basic radii enables a single value of the constant p t o be
chosen but the difficulty of calculating repulsive energies in even so
similar a system as that of the alkaline earth chalcogenides and the
magnitudes of the errors introduced when the compressibilities are not
known has been demonstrated by Huggins and Sakamoto ( 5 4 ) .
3. The Dispersion Energy (London) Term, U L
London (83)has shown that there exists an attractive force between

all molecules or ions caused by a synchronization of the oscillations of
their electrons. Any two molecules or ions have an attractive energy tiL,
given by
where el and e2 are characteristic energies.of the two molecules or ions

and a1 and a2 are their polarizabilities. The total dispersion energy for
a crystal lattice is then
where ulz, all and aZ2 are respectively the coefficients of l/rB for the
positive-negative, positive-positive, and negative-negative ion pairs. So
and Soare respectively the sums over all unlike and like ions of l / ( ~ ~ / r , , ) ~
where rL is the distance between a given ion and the ith ion in the lattice
and ro is the shortest ion-ion distance. Such sums have been evaluated by
Lennard-Jones and Ingham (82) and their values are given in Table
IX. Born and hfayer (21) in the initial calculation of the dispersion
energy for the alkali metal halides assumed tl and 2 to be the ionization
potentials of the free ions.
In the case of a singly charged positive ion this is the second ionization potential of the element. In the case of a negative ion it is the electron affinity. Born and Mayer ( 2 1 ) and Mayer and Helmholtz (93)
initially used the polarizabilities of the gaseous ions calculated by Paul-
175
LATTICE ENERGIES
TABLE IX
~:OEFYICIENTY FOR THE INDUCED
DIPOLE-DIPOLE
AND
I)IIWl,E-~UADRUOLE
Lattice type
ss
Se
ASS
8s
NaCl
CSCl
6.5952
8.7088
4.354
1.8067
3.5445
0.762
6.1457
8.2007
4.104
0.8002
ZnS
VORCEB
2.1476
0.253
ing (104). Mayer (91) later concluded that the values for the London
dispersion energy obtained in this way were too small and that the error
arose from taking polarizabilities and energies characteristic of the
gaseous ions; he also pointed out that a term representing the induced
dipole-induced quadrupole interactions should be included. He used the
known ultraviolet absorption spectra of NaC1, KC1 and K I to calculate
the dispersion energy exactly and then examined the validity of various
approximations. He found that to a reasonable approximation the contributions of the negative and the positive ions could be separated and
that the polarizability of the positive ion from crystal to crystal was
fairly constant. For the positive ion he adopted the value of a given
by Pauling (104) and took E equal to 0.75 c0 where c0 was the ionization
potential of the gaseous positive ion. As a fair approximation he took c
for the negative ion to be 1.30 hvo where vo is the frequency of the first
sharp line in the crystal ultraviolet spectrum. a for the negative ion was
found from the expression
where Q0O2 was constant for any given negative ion and independent of
the crystal. The values of the dispersion energy were more than doubled
by using the data from the ultraviolet spectra of the crystals. The induced dipole-quadrupole energy for the crystal is given by the expression
Here the d are the analogs of the a in the expression for the induced
dipole-dipole interaction and have been discussed by Margenau (89, 90).
Ss and Ss are sums of the type ~ / ( T ~ / Tover
~ ) ~all like and all unlike
ions in the crystal lattice. They have been computed by Lennard-Jones
and Ingham (88) for various lattice types and are given in Table IX.
The actual value of the induced dipole-quadrupole term is about one
tenth that of the induced dipole-dipole term.
176
T. C. WADDINGTON
4. The Zero-Point Energy Temn, Cz

The terms so far discussed all contribute t o the potential energy of
a spherical ion in a crystal lattice. The lattice energy at absolute zero
is obtained by adding to these terms a term equal to the energy retained
by the lattice a t absolute zero when it is regarded as an assembly of
coupled oscillators. This energy, Uz,is the zero-point energy and is
given by the Debye theory as being
9
Uz = NA-hv,,
4
(36)
where h is Plancks constant and vmaX is the Debye maximum frequency.

The value of this maximum frequency may be estimated with sufficient
accuracy by using infrared absorption data, assigning to vmPI the frequency of the residual rays or Reststrahlen. If the infrared data are
not available Waddington (1%) has pointed out that the relation beand the Debye
tween the absolute entropy at 298C of the crystal, Soabs,
maximum frequency given by the expressions
So.bs
R[lnf 4- T(dlnf/dT),]
(37)
and
lnf = In g -
()
8 kT
--
Vmas
In [1 - e--hv/kT]v2 dv
vamSIX
(36)
may be used. Here f is the partition function per molecule and g is the
electronic partition function. The value of Soab.is plotted against the
zero point energies of the alkali halides and the zero-point energies of
other salts interpolated from their absolute entropies.
5. The Permanent Electrical Multipole Term, UQ
The account of the term-by-term calculation of the lattice energy

given above is only correct for an array of spherical ions. If nonspherical
ions are present in the lattice then the initial assumption made in calculating the coulombic energy, that is, that a charged ion may be replaced
by a point charge at its center, is no longer correct. However, this difficulty may be resolved if the charge distribution on the ion is known or
can be calculated. Then the charge on the ion may be represented in
terms of a point charge plus a permanent electrical inultipole and an
immediate simplification effected. The electrostatic energy can now be
heated as the s u m of two terms, the first, and by far the larger, being the
simple Madelung energy due to the interaction of the point charges and
the second being the multipole energy.
The multipole energy itself is made up of two terms, the first being
177
LATTICE ENERGIES
the multipole-multipole interaction energy and the second the energy of

interaction of the multipole with the surrounding point charges. Of these
two terms the first is very much greater than the second. In the case of
symmetrical cylindrical ions, such as the azide, Ns-, and the bifluoride,
HF2-, ions, the multipole is in fact a quadrupole and expressions for the
energy of interaction have been developed by Gray and Waddington
(43) and by Waddington (12T).The energy of the quadrupole-quadrupole interaction is given by
Where q is the quadrupolar charge and d the length of the quadrupole,

R is the distance between the two quadrupoles, B is the angle between
the line joining the centers of the quadrupoles and the axis of the first
quadrupole, and 5 is the direction of the axis of the second quadrupole
relative to those of the first. It is usually only necessary to compute
quadrupole-quadrupole interactions for nearest and next to nearest
neighbors. The quadrupole-point charge interaction energy has been
summed over the whole lattice in the case of tetragonal, body-centered
crystals of symmetry DN,and shown to be
u
4
2
p =
1.30 f 0.40)(1 - ar2)qd2e.

(a'>
(39)
Here a3 ( = c'/a'), is the axial ratio in the pseudo-hit cell, q and d

have the same signficance as before and e is the electronic charge.
B. THEKAPUSTINSKII
AND TEMPLDWN
CALCULATIONS
OF THE
LATTICE
ENERGIES
OF IONICCRYSTALS
1 . Kapustinskii's Formula and the Themnochemical Radius
The method of extended calculation of lattice energies described in

the previous sections is restricted to salts the crystal structures of which
are exactly known. For less simple structures than those of the alkali
halides the extended calculation of lattice energies becomes increasingly
laborious as the symmetry decreases. If however only the first two terms
in the lattice energy are considered, that is, U is taken to be U M - UR,
then adopting either an inverse power law or an exponential law for the
repulsive term, U may be written as either
or
If the number of ions in the chemical molecule is
Y,
the number in a
178
T. C. WADDINGTON
mole is S A v . The above equations may therefore be rewritten
(40, 41)
Where a = Jl/(v/2\. The Madelung constant, M, is proportional to
the number of ions in the chemical molecule and hence a is independent
of this. Although a is not identical for different lattice types, Kapustinskii
(6.4,65) found empirically that in passing from one lattice type to another the change in the constant a was proportional to the change in the
interatomic distance, a result shown by Sherman (115) to be in agreement with the correct calculation of the Madelung constant for cuprite.
Every crystal may then be considered to transform into a rock-salt
lattice without change of lattice energy if the coefficients a and ro are
simultaneously modified so as to have the values corresponding to the
ions in a rock-salt lattice. tio can now be found either by measuring
ro and the crystal structure and calculating M and hence a or by putting
TO = rg
rc, the sum of the Goldschmidt ionic radii for coordination
number 6 , and taking a = 1.745, its value for a rock-salt lattice. The
substitutions ro = rA + rc in Angstrom units, p = 0.345 A, a = 1.745,and
N A e 2 = 329.7 kcal/A give Kapustinskiis formula
Kapustinskiis equation, of course, cannot be expected to produce

values for the lattice energies as exact as those produced by extended
calculation. Essentially the Kapustinskii formula ignores all contributions to the lattice energy save U R and Uar.If the ions are nonspherical
there is considerable difficulty in choosing a repulsion envelope for the
ion and assigning a radius. This difficulty also arises in the extended
calculation of the lattice energy but only in calculation of U R , which is
an order of magnitude less than U,. In the Kapustinskii equation this
difficulty arises in Uaaas well.
Finally the most serious objection is the replacement of rot the interatomic distance, by rd + re. There are many salts in which the unlike
ions are not touching, for example, NaI, where the lattice spacing is
cfetermined by I- - I- contacts. Also, the assignment of ionic radii can
only be regarded as approximate w e n in the alkali metal halides. In
spite of these objections the Kapustinskii formula remains a useful guide
to approxiniate lattice energies, especially in those cases where the structure is unknown. In general, as the examples discussed later will show,
the values for lattice energies given by the Kapustinskii equation fall
on the low side of the correct lattice energy.
179
LATTICE ENERGIES
One of the difficulties that arises in the use of the Kapustinskii calculation is that of assigning a value to the ionic radius of the anion when
a complex salt is being considered. Yatsimirskii (133) has pointed out
that if the reaction
MX(s) = hl+ig)
+ X-(g)
is considered, then the enthalpy of the reaction, U

U
(9)
+ 2RT, is given by
+ 2RT = AHfoM+(g)+ AHfoXx-(g)- AHfOMX(s)
(43)
and if now two salts with the same anion but different cations are considered, MI and M2, then
UM~X
- UM*X= AHjoMi+(g)- AHj0M2+(g)
- [AHfoMIX(S)- AHf0M2X(s)] (44)
and
+ rx- ( 1 -
m,+
rM,+
+ rx-)]
(45)
using Kapustinskiis equation. If the enthalpies of formation of the solid

salts MIX and MZX and AHjoM1+(g) and AHfoM2+(g) are known and
the Goldschmidt radii of MI and M2 are known, rx- can be found. The
radius of X found in this way is termed by Yatsimirskii the thermochemical radius and it can of course be inserted into the Kapustinskii equation
to give the lattice energies of the salts MX and also AHj0X-(g). Yatsimirskii (133, 134) lists a large number of thermochemical radii and he and
Kapustinskii (68) have also calculated a large number of thermochemical
radii for tetrahedral ions and shown that they are fairly independent of
the other ion in the lattice. The values they obtain for the thermochemical
radii of various anions are given in Table X.
2. Templetons Prediction of Madelung Constants
It is perhaps useful to discuss in this section the work of Templeton

(116) on the empirical prediction of Madelung constants. He has defined
a reduced Madelung constant a by a = 2M,/zlzzv for binary lattices.

Here M,, is the Madelung constant with the nearest-neighbor distance ro
taken as the standard distance in the lattice, z1 and z2 are the anion and
cation charge numbers and v is the number of atoms in the chemical
molecule. Templetons reduced Madelung constant a is just Kapustinskiis
LY divided by the product of the charges on the ions. Templeton has shown
empirically that a can be represented by the equation
a = 1.89
- 1.00/m
TABLE X
THETHERMOCHEMICAL
RADII(A) OF SOME COMPLEX
ANIONS
Ion
Radius
NH*-
OH-
NO,-
HCOO-
CNO-
CHICOO-
HCOZ-
IOa-
CN-
1.30
1.40
1.55
1.58
1.59
1.59
1.63
1.82
1.82
__
--
__
Ion
Radius
NOaI .89
Br03-
HS-
CNS-
c104-
MnO4-
1.95
1.95
c1032.00
cd&(NOz)&
1.91
2.23
2.36
2.40
Ion
Radius
10,-
BF4-
so,*-
Crof-
2.28
021.80
coat-
2.49
1.85
2.30
2.40
Ion
Radius
Ion
Radius
MOO,'-
sC04'-
Tc04?-
BeF4'-
B03'-
P0a3-
2.54
2.43
2.54
2.45
1.91
2.38
AsO43-
SbOia-
Bi04*-
Siol4-
2.48
2.60
2.68
2.4
5
d
G
El
3H
8
LATTICE ENERGIES
181
where m is the weighted harmonic mean of the coordination numbers, with

an accuracy of better than 1% for a dozen cases of binary structures.
To permit application to more complex structures, such as ternsry salts,
Templeton (118)redefined the reduced Madelung constant as a = 2M,,/
2z2. When the nearest neighbors of different atoms are a t different distances, one can expect a to be less than the value predicted by the above
equation. The prediction will correspond more closely to a,,, defined as
a, = arm/ro, where r,,, is the harmonic mean of the distances, weighted
according to their frequency of occurrence and according to the product
of the charges of the ions involved.
The accuracy of the prediction for the ternary salts is less than that
for the binary salts, the average deviation from the accurately calculated value being about 2% for the salts with equal distances and about
3.5% for the salts with two distances (12 cases being considered). The
worst case was in error by 10%. Templeton provides a theoretical justification for his equation and points out that it wi11 not yield accurate
predictions for any structure whatever. Rather it represents approximately the largest value which is possible for a given coordination number. If a substance is a salt, it can only be stable in a structure with
Madelung constants not too far below the largest one possible, since the
crystal energy is determined principally by the electrostatic potentiat.
C. THE QUANTUMMECHANICAL
PREDICTION
OF LATTICEENERGIES
An entirely different approach to the calculation of lattice energies

was initiated by Hylleraas (69) who applied a general quantum mechanical treatment to the calculation of the lattice energy of lithium hydride.
He used one-electron wave functions of the hydrogenic type with nuclear
screening so that the entire computation could be performed analytically.
The wave functions were
with z = 1 for hydrogen and 3 for lithium; ah is the Bohr unit distance
(0.58 A ) . The chief objection to his calculation was that his approximate
wave functions did not give very good binding energies for the free H.and Li+ states. The actual method of calculation was similar to that
employed by later workers and will be discussed in connection with their
calculations. His actual results were that the binding energy was 218
kcal/mole and that a. = 4.42 A. These may be compared with empirical
values of 218.5 kcal/mole and 4.084 A.
Landshoff (78,79) later calculated the lattice energies of NaCl by
this general quantum mechanical method and the same method was
adopted by Lowdin (84, 85, 86) who calculated the cohesive energies of
182
T. C . WADDINGTON
LiCI, NaC1, KCl, and NaF. Later Jiro Yamashita (130, 131) calculated
the lattice energy of LiF and Benson and Wylie (7) also calculated the
lattice energy of LiF. Recently Shoichi Kobayashi (74) has calculated
by quantum mechanics the lattice energy of CUZSand Lundquist (87)
has recalculated the lattice energy and other parameters of LiH. The
basis of all these calculations is the same though Benson and Wylies
represents a simpler approach to the problem. The Hamiltonian operator
for the system of N ions is written down. The total energy of the ground
state of the crystal is then given by the lowest eigenvalue E of the
Schrodinger equation
H# = E#
(47)
for the antisymmetric eigenfunctions $ = $(XI, X Z , . . . , zi, . . . , XN)
where x4 are the space and spin coordinates of the electron i. The solution of the equation is the lowest value of E in:
To solve this equation the one electron approximation is used. The differences in the treatments of the various authors lie in the differences
in their assumptions about orthogonality in the wave functions used and
in the use of either a variational or perturbation method to find the minimuin energy.
It is convenient to split the energy of cohesion U up into four parts
uc = UMad + v c o m
+ Uo
(49)
The Madelung term is as in the classical theory; the Ucor,represents a
correction to the Madelung term arising from the fact that the ions are
not point charges at the lattice sites but overlapping charge clouds; the
term Uexchange is a straight exchange energy arising from the exchange of
electrons on two atoms. Only the exchange terms between neighboring
atoms are important. The first three terms can be considered as resulting
from only pure two-body interactions. However, this is not true for the
U , term and it does include quantities which arise from the interaction
of many-body forces.
Landshoff made two calculations of the lattice energy of NaC1, the
first with wave functions in which the exchange of electrons within the
ion was neglected, the second with Hartree-Foch wave functions in which
this exchange was taken into account. He calculated the lattice energy,
the lattice parameter, and the compressibility. The values he obtained
are listed in Table XI and compared with the experimental ones. He
treated a set of linear combinations of the wave functions of the free
ions as though they were exactly orthogonalized and normalized; but in
Uexchange
salt
LiH
TABLE XI
u MECHANICAL
~
CALCULATION
OF LATTICEENJEBQIES,
EQUILIBRIUM
DISTANCES
AND COMPRESSIBILITIES
Calculation
Lattice
energy
(kcal/mole)
Equilibrium
distance
(A)
Compressibility
bar
Hydrogenic wave functions
Hylleraaa
S. 0. Lundquist (1954)
Experimental value
218
205
217 f 7
4.084
Lowdin
Benson and Wylie
Yarnathits (1)
Yamaehita (2)
Huggins calculation
199.5
242
247
239
243.6
4.79
4.02
4.00
4.00
4.02 (exp) 300K
Landshoff (1)
Landshoff (2)
Lowdin
Huggins
165
182
183.2
183.5
5.55
8.2 x lo-"
5.56
4.35 x 10-12
5.54
4.5 x lo-'*
5.628 (exp) 300K
LiCl
Lowdin
Huggins
187.7
200.2
5.37
4.2
5.144 (exp) 300K
x 10-la
KCl
Lowdin
Huggins
166.9
167.9
6.17
6.0
6.278 (exp) 300K
N&F
Lowdin
Huggins
205.1
215.4
4.58
4.620 (exp) 300K
cuss
Kobayashi
480
5.1'4
3.95
LiF
NaCl
Remarks
4.42
4.0
x lo-'*
1.7 x lo-"
1.3-1.7 X 10-'2
1.96 x 10-12
Older Hartree wave function

Hartree-Foch wave function
Hartree-Foch wave function
32
Slater's wave functions used

and method of Landshoff
employed
C-L
%
!
184
T. C. WADDINGTON
reality this was only true if second and third order terms were neglected.
Lowdin derived a formula for a set of linear combinations of the free
ion wave functions that were exactly normalized and orthogonalized and
deduced values for the lattice energy, lattice parameter, and compressibility of LiF, NaF, LiCI, NaCl, KCI. His values for these quantities for
sodium chloride are very close to those of Landshoff and both are in
good agreement with the experimental values. His value for KC1 is in
good agreement but the values for NaF and LiCl are pot in very good
agreement with experiment, and his values of the lattice constants for
LiF are a good deal out. No other calculations apart from Lowdins have
been made for NaF, LiCl and KCI but two other workers have computed
the lattice energy and parameters for LiF. Yamashita assumed a wave
function of the Hartree type for the 1s state of the Li+ and for the 1s
and 2s states of the F- ion. For the trial function for the 2pelectrons
for the F- ion he used Slater wave functions for the crystal with three
adjustable parameters. He minimized the total energy of the crystal for
various lattice distance by the use of the variational method. He initially
reported the values (1) (Table XI) but later reported the values (2).
He also calculated the diamagnetic susceptibility of the F- ion in the
crystal and found it to be in good agreement with the experimental measurements. His values are seen to be a good deal better than Lowdins,
probably because of the method he adopted to deduce the wave function
of the F- ion in the crystal, allowing for the effect of the crystal field
on the wave functions of the free ion.
Benson and Wylie proceeded by a rather different method in calculating the lattice energy of LiF. Their assumptions are a good deal more
drastic. The Li+ ion is treated as a unit positive charge and account of
the 1s electrons is taken only indirectly in the orthogonalization relations. All details of the inner structure of the fluorine ion were discarded
and the ion was treated as a nucleus of charge +5, surrounded by a shell
of six 2p-electrons. The Hartree self-consistent field calculation of the
radial density distribution of the two 2p-electrons of a free fluorine ion
was used. The function was suitably normalized and fitted to an
analytic expression of the Slater type. A simple perturbation calculation
was then carried out. The results obtained for the lattice energy, lattice
constant, and compressibility are given in Table XII. The agreement
obtained was fair, but it was clearly a drastic approximation to use
orbitals appropriate to the free ion in the interior of the crystal. As a
way of overcoming this, the outer orbitals of the fluorine ion were orthogonalized with respect to one another and made very nearly orthogonal
with respect to the 1s orbitals of the neighboring lithium ions. The values
of the lattice parameter, cohesive energy, and compressibility were de-
185
LATTICE ENERGIES
TABLE XI1
SUMMARY
OF BENEON
AND WYLIE'SRESULTS
FOR LiF
(i)
(ii)
(iii)
'
(iv)
(V)
(4
Simple perturbation
calculation
Wave functions orthogonalized
Wave functions adjusted to fit diamagnetism
Theoreticalresults of
Lowdin
Theoretical resultsof
Yamashita
Empirical values
Lattice
parameter
(A)
Cohesive
energy
(kcal/mole)
Compressibility
(10-1: bar)
3.78
247.4
1.8
5.16
188.1
4.8
4.02
242.0
1.7
4.79
199.5
4.0
4.00
239
4.02
243.6
(at 300K) (Hugginscalculation)
1.96
duced and are given in Table XII. These results show large deviations
from the empirical values, larger deviations than those of the simpler
treatment. It is interesting to note that these results are comparable
with the more exact calculations of Lowdin, who used an essentially
similar treatment. The next step of Benson and Wylie was empirical.
They adjusted the wave functions of the fluorine ion to give a diamagnetic susceptibility in agreement with the experimental results. The
values of the lattice energy etc. obtained in this way are given in Table
XII.
It appears from the above discussion that the most satisfactory
approach to the quantum mechanical calculation of lattice energies is
that developed by Yamashita, in which the parameters of the outer wave
functions of the ions are adjusted by a variational method to minimize
the total energy of the crystal. Orthogonalization of the simple free ion
wave functions seems to produce a result rather worse than that achieved
by ignoring the correction. No doubt with the availability of electronic
computors Yamashita's method will be extended to crystals in addition
to LiF, where it may be necessary to adjust the wave functions of both
the ions by a variational method, to allow for the effect of the crystal
field. This will produce an exceedingly tedious calculation. Yamashita
(1%) has slso used the method described above to show that the 02ion
is stable in the MgO crystal, though not in the gas phase.
An example of the application of quantum mechanical calculations
to more complicated systems is the calculation of the lattice energy and
186
T. C . WADDINGTON
lattice constant of P-CuzS by Kobayashi. The numerical calculation was

carried out to determine the positions of the Cu+ ions in the p-phase
since this could not be determined by X-ray analysis. Slaters wave
functions were used for Cu+ and S2- and the method developed by
Landshoff was applied. The potential energy minimum was determined
to fix the Cu+ ions. The results of the calculation me given in Table XI.
The lattice energy of LiH has recently been recalculated by S. 0.
Lundquist. He has not used the old hydrogenic wave functions used by
Hylleraas nor has he used the free ion orbital approach because he
maintains that the deformation of the diffuse H- ions in the field of the
crystal must be considerable. The total energy was minimized with respect to the effective nuclear charge of the H- ion.
D. THECALCULATION
OF LATTICE
ENERGIES
FBOM HYDMTION
ENTHALPIES
Between the lattice energy U of a crystal and the enthalpy of hydration H of the gaseous ions of the crystal to unit activity in water, there
exists the relation
U = -H+AHQ8-2RT
(7)
where S is the standard enthalpy of solution of the crystal to unit ac-
,001
I.o
1.25
1.50
2.0
1.75
Crystal radius of the anion
(n)
FIG.1. Enthalpies of hydration as functions of anion radii.
2.25
187
LATTICE ENERGIES
tivity. The enthalpy of hydration of the ion pairs is well known for the
alkali metal halides and two empirical correlations have been observed
and substantiated. The first is between the enthalpy of hydration of the
ion pairs and the crystal radii of the anion and cation. If the enthalpies
of hydration of the alkali halides are plotted against the crystal radii
of the halide ions as given by Pauling (107) smooth curves are obtained
(Fig. l ) , and thus if the ionic radius of any other anion is known it is
possible to obtain from the curves by interpolation the enthalpies of
hydration of the anion with the alkali metal ions.
The second empirical correlation that has been observed is between
the ion pair enthalpy of hydration and the lyotropic number (Biichner,
23,84). The lyotropic number was developed by Biichner as a quantitative expression of the position of the anion in the Hofmeister series. It
loo:
i ; ;
Ib
1'1
cs
I;' I;
Ib
I;
Lyotropic number of the anion
FIG.2. Enthalpies of hydration as functions of the lyotropic numbers of the anions.

is essentially a measure of the power of an anion to bind water to itself
at the expense of hydrated colloids in suspension. It is found experimentally that enthalpies of hydration of ion pairs vary linearly with the
lyotropic number of the anions (see Fig. 2).
Buchner has determined the lyotropic numbers of the following
188
T. C. WADDINGTON
anions F-, 103-, HzPOz-, Br03-, C1-, NOz-, C103-, Br-, Nos-,
CIO*-, I-, CNS-, Na-.
If the lyotropic number of an anion is known or a crystal radius can
be assigned to it, then the hydration enthalpies of its ion pairs with the
alkali metal ions can be calculated and hence if the enthalpies of solution of its alkali metal salts are known their lattice energies can be
found. Gray and Waddington (49) have used both methods to calculate
the enthalpies of hydration of the alkali metal aeides. The results are
mutually consistent and the lattice energies derived from them are also
in excellent agreement with the lattice energies found by extended calculation. Waddington (126) has used the Buchner lyotropic number to
derive hydration heats and lattice energies for most of the anions listed
above and the results obtained are found to be, in the cases where they
can be checked, consistent with the thermochemical data. The individual
results will be discussed in Section 111.
The disadvantages and limitations of the two methods given above
are the following. In the case of the crystal radius interpolation it is
necessary to be consistent and use one set of crystal radii. Gray and
Waddington and Waddington accepted the Pauling set. Another difficulty
arises if the anion is nonspherical. If this is the case it is necessary to
assign a mean radius. In the case of the azide ion the difficulty was
overcome by assuming that the ion was an ellipsoid of revolution and
using the change of structure and coordination number in passing from
potassium to sodium azide to calculate the major and minor axes of the
ellipsoid. Fortunately the curves of hydration enthalpy against anion
radius are shallow ones, the sum of the hydration enthalpies changing
by only about 6 kcal per 0.1 A change in the anion radius.
Again in the case of the lyotropic number interpolation the accuracy
is dependent on the accuracy of the measurement of the lyotropic number, though the sum of the hydration heats only changes slowly with
lyotropic number. It must also be stressed that the lattice energies obtained in this way are not theoretical lattice energies in that they are
not based on any model of the crystal. Rather they are empirical or
experimental in that they are based on a combination of empirical hydration enthalpies and experimental enthalpies of solution.
ill. Individual Values of the Lattice Energies of Alkali Metal
and Alkaline Earth Salts
Before discussing the uses to which the values obtained for lattice
energies can be put it seems worthwhile to give an account of the individual calculations that have been made of the lattice energies of various
LATTICE ENERGIES
189
alkali metal and alkaline earth salts and also to give some indication of
the accuracies of the various calculations. A very large number of lattice
energies can of course be computed by the use of Kapustinskiis formula
(64, 66)and Yatsimirskiis (13S, IS4) thermochemical radii listed in
Table X. These values are not in general given when they are the only
values available but they are listed for comparison with the results of
other workers.
A. THEALKALIMETALHALIDES
The lattice energies of the alkali metal halides have been calculated
by a large number of workers with approaches of increasing sophisticatios. The principal calculations by the classical ionic theory since Shermans (114) review have been made by Mayer and Helmholtz (M),
Verwey and de Boer (125),and by Huggins (5s). Seitz (113) has also
listed a series of values which differ somewhat from those given by
Huggins, probably because he does not use Huggins basic radii in
computing the repulsion energy. More recently Ladd and Lee (76) have
ueed a method to compute lattice energies which eliminates the need for
basic radii and their values are included for comparison. The values of
the lattice energy predicted (66) by the Kapustinskii equation are also
included for comparison in Table XIII.
The values of Huggins are probably. the most accurate lattice energies
obtainable and agree with the Born-Haber cycle values to within the
experimental accuracy of the cycle terms. The values given by the
Kapustinskii equation will be seen to be rather low. The Born cycle
values are obtained from the values of hHf0M+(g) and AH,OMX(S)
given by the U. S. Bureau of Standards, circular 500, and the values of
hHfOX- (g) decided upon by Pritchard (lot?), as a result of a review
of all the experimental data.
B. THEALKALIMEPALHYD~ZIDES
Apart from the quantum mechanical calculations of the lattice energy
of LiH, due to Hylleraas (59) and to Lundquist (871,the only calculation has been made by Bichowsky and Rossini (10) who derived the
lattice energy of LiH using a Born-Lande expression. Accurate values
of the lattice energies are available from thermochemical data and
Kazarnovskii (72) has used these values to calculate the exponent in
the Born repulsion coefficient. Using a simple Born-Mayer expression,
ignoring Van der Waals terms, the present author, taking p = 0.345 A,
the value found for the alkali halides, has obtained values for the lattice
energies of all the alkali metal hydrides. These values are compared with
the thermochemical data in Table XIV. It will be seen that the theoreti-
TABLE XI11
LATTICE ENERGIES
(KCAL/MOLE)
OF
Salt
LiF
LiCl
LiBr
LiI
NaF
NaCl
NaBr
NaI
KF
KC1
KBr
KI
RbF
RbCl
RbBr
RbI
CsF
CSCl
CsBr
CSI
THE
ALKALIMETALHALIDES
Born
cycle
Sherman
1932 (114)
Kapustinskii
1943 (66)
Mayer and
Helmholtz
1933 (99)
de Boer and
Verwey
1936 (99)
241.2
198.2
188.5
175.4
216.0
183.8
175.9
164.5
191.5
166.8
160.7
151.0
183.6
162.0
155.2
146.5
171.0
153.2
148.3
140.3
238.9
192.1
181.9
169.5
213.8
179.2
170.5
159.6
189.2
163.2
156.6
147.8
180.6
157.7
151.3
143.0
171.6
147.7
142.3
134.9
227.7
192.1
189.5
170.4
211.5
179.9
175.5
161.0
188.5
162.7
161.3
146.8
181.7
158.2
149.7
141.0
170.4
149.4
143.9
134.7
240.1
199.2
188.3
174.1
215.4
183.1
174.6
163.9
189.7
165.4
159.3
150.8
181.6
160.7
153.5
145.3
173.7
152.2
146.3
139.1
243.0
201.9
192.0
178.7
216.1
186.0
178.3
168.9
193.5
168.9
163.4
154.6
186.0
164.7
158.6
150.3
Huggins
1938 (63)
1940 (ff.9)
243.6
200.2
189.5
176.1
215.4
183.5
175.5
164.3
192.5
167.9
161.3
152.4
183.0
162.0
156.1
148.0
175.7
153.1
149.6
142.5
242.0
200.2
189.7
176.2
214.5
184.4
176.5
166.1
191.8
167.5
161.7
153.1
184.2
162.7
-156.7
148.4
175.8
155.6
150.0
142.7
Seitz
Ladd arid
Lee
1958 (76)
<
P
U
r?
192
165
159
149
TABLE XIV
THELATTICEENERGIES
(KCAL/MOLE)
OF
THE
ALKALIMETALHYDRIDES
Y
Thermochemical i
e
salt
LiH
NaH
KH
RbH
CsH
Quantum mechanical
value
218 (Hylleraas)
205 (Lundquist)
Kazarnovskii
value for n
Bichowsky
and Rossini
4.40
5.24
5.85
6.63
6.97
218.2
Waddington
234.0
202.0
177.2
168.6
162.0
vduefrom
Born cycle
216.5
193.8
170.7
164.4
156.1
8
M
m4
m
td
E:
H
rn
192
T. C. WADDINGTON
cal values are too high and this is ascribed to the difference between the
repulsion energies and compressibilities of the hydrides and those of the
halides.
c. THEALKALI3IETAL .4XD
ALKALINE
EARTHCHALCOGENIDES
The lattice energies of the alkaline earth oxides, sulfides, selenides,

arid tellurides have been calculated by a number of workers, though
most attention has been focused on the oxides. Sherman (114) used the
simple Born formula. Later $layer and Maltbie (96) used the BornMayer expression for the lattice energies, calculating the London dispersion energies from the polariaabilities of the free ions. van Arkel and
de Boer (123) also calculated the lattice energies of some of the chalcogenides. Later de Boer and Verwey (29) recalculated the lattice energies of the oxides, because the Mayer and Maltbie interatomic distances
had been based on density measurements and differed significantly from
those obtained from the X-ray data. Other calculations have been made
by Fowler (58) and by Kapustinskii (66).More recently Kapustinskii
and Yatsimirskii (69) have recalculated a large number of lattice energies and Huggins and Sakamoto (54) have recalculated the lattice
energies of all the alkaline earth chslcogenides. They used the Huggins
basic radii and improved expressions for the dispersion energy based
on the optical properties of the crystals. Unfortunately satisfactory compressibility data were not available and so the repulsion energy term
remains the major source of error in their calculation. Morris (99) has
recently calculated the lattice energies of BaO, LizO, Na20, K20, RbzO
and Cs20; he considers these salts to be more ionic than the alkaline
earth oxides. The values obtained by the different workers are collected
in Table XV.
Probably the results of Huggins and Sakamoto and of Morris represent the best available lattice energy data. Huggins and Sakamoto
give values of aH,002- = 221 =t 15, AHfoS2- = 152 -t 15, AH?Se2- =
165 k 15 and AHfoTe2- = 145 20 kcal/mole. hiorris obtains AHPOZ= 211 7 kcal/mole.
+_
D. THE-4LfiALI J~ETAL
HYDROXIDES
The position with regard to the lattice energies of the alkali metal
hydroxides is by no means satisfactory. At room temperature they have,
with the exception of LiOH which is tetragonal, orthorhombic lattices,
in which the OH- ions are obviously not rotating. A t higher temperatures they undergo a transition to cubic NaCl lattices. A number of
calculations of their lattice energies have been made, not all of them
based on a knowledge of the crystal structure. Born and Kornfeld (18)
TABLE XV
LATTICEENERGIES
(KCAL/MOLE) OF
Huggina
and
Sakamoto
Salt
Morris
BeO
BeS
Be&
BeTe
MgO
MnS
SrO
NasS
KIO
KIS
RbO
R M
(b)
1026
854
1082
893
855
795
938
788
757
699
841
726
701
662
792
692
671
635
746
659
639
608
8B
CaSe
CaTe
SrS
SrSe
SrTe
BaO
Bas
BaSe
BaTe
Li90
Li2S
NasO
(4
21.3
763
907
764
734
678
816
707
683
644
769
675
654
620
724
643
624
594
Mayer
and
Maltbie
THE
CWLCOGENIDES
van Arkel
and Boer;
de Boer and
Verwey 3
Fowler
Kapustinskii
and
Yatsimirskii
1080,1030
(136)
Sherman Kappustinskii
907
939
77 1
914
800
936.920
790
976
766
743
934
807
831
737
Go,825
710
845
740
704
841
72I
698
719
684
766
686
784,780
680
789
696
664
790
686
666
769
676
644
727
647
740,735
640
876
714
700
648
829
688
675
626
780
662
652
609
751
666
638
746
655
636
731
646
617
!Y39
601.7
532.8
513.0
* Values of C. D. Weat (199).
Zhanov
$ In this column when two values are given they represent upper and lower limits.
801
786
5!
ij
m
E!
2
695
576:
516
565*
446:
+
co
w
TABLE XVI
THE LATTICEENERGIES(KCAL/MOLE)
OF THE ALKALIMETALHYDROXIDES
Salt
LiOH
NaOH
KOH
RbOH
CsOH
Goubeau
Born and
Van Arkel
and Klemm
Kornfeld
and de Boer
Goubeau
U AHfoOH- U AIZfoOH- U AHfOOH- U AHfOOH215
205
181
172
158
63.1
41.8
42.7
44.0
48.1
205
166
145
137
130
73.1
80.8
78.7
79.0
76.0
205
73.1
234
43.6
Kapustinskii
as given by
Unzhakov
Waddington
West
U A H ~ O H - U AHfOOH- U AH,"OH229.5
209.0
184.0
174.1
165.1
48.1
35.8
39.7
42.0
41.1
197
175
50
49
201
46
Hydration
heat
U AHjOOH229.3
196.1
172.7
165.7
156.6
48.8
50.7
51.0
50.3
49.5
5
?'
4
<
&
s
LATTICE ENERGIES
195
calculated the lattice energies of the salts, as did van Arkel and de Boer
(123),without a proper knowledge of the crystal structure. Later Goubeau (41) calculated the lattice energy of LiOH from its crystal structure; the values of the lattice energies of the other alkali metals he lists
are based upon this calculation and are not direct determinations. Goubeau and Klemm (49) later attempted to correct Goubeaus original
value for LiOH to allow for the permanent dipole moment of the OHion, which is not rotating. The values for the lattice energies given by
Juza (62) are based on fairly crude assumptions about the molecular
volume of the hydroxide ion, arrived a t from specific gravity measurements.
West (129) gives a value for the lattice energy of the high temperature
form of NaOH using a simple Born-Lande expression. Kapustinskiis
(65, 66) formula has also been used to calculate a thermochemical radius
of the alkali metal hydroxide ion of 1.40 A, and inserting this into the
Kapustinskii equation the lattice energies given in Table XVI are obtained. Unzhakov (121) quotes a slightly different set of lattice energies
derived by the Kapustinskii equation. Unfortunately none of these calculations yield consistent values of AH,OOH- (g) = AH,OOH(g) - EOH
from hydroxide to hydroxide, nor values in agreement with the experimental value for E O H given by Page (102).
The present author has used the Born-Mayer expression to calculate
the lattice energies of the high temperature cubic forms of NaOH and
KOH, and has also calculated the lattice energies from the hydratiop
enthalpies obtained from interpolating on the crystal radius-hydration
enthalpy plot of Fig. 1, the crystal radius of 1.47 A found by Goubeau
(41). The values of West, the authors Born-Mayer calculation for
NaOH and KOH and the hydration enthalpy values are in good agreement with each other and lead to a value of AHFOH- (g) of 50 kcal/mole
in good agreement with Pages experimental value for EOH.
E. THDALKALI METALHYDROSULFIDES
The lattice energies of the high temperature forms of the alkali metal
hydrosulfides were calculated by C. D. West (129)using a simple BornLande expression. Yatsimirskii (133) has assigned a thermochemical
radius of 1.95 A to the SH- ion. He does not list values for the lattice
energies in his paper but only the derived quantity AHFSH- (g) = 27.8
kcal/mole. The values obtained by using his thermochemical radius in
the Kapustinskii equation are given in Table XVII together with Wests
values for-the high temperature, cubic, forms, and with the authors
values using a simple Born-Mayer expression. The value of AH,*SH- (g) ,
obtained from each lattice energy are also listed and they are in good
agreement and lead to a value of 30 kcal/mole being adopted.
196
T. C. WADDINGTON
TABLE XVII
THE LATTICEENERGIES(KCAL/MOLE)
OF THE ALKALIMETALHYDROSULFIDES
West
Waddington
--Salt
G'
IdSH
NsSH
KSH
RbSH
CsSH
178
168
155
AH,%H~
150
143
33.3
30.1
29.5
28.8
Y atsimirskii
__
AH,%H-
AIT~SH-
33.1
29.8
30.5
32.8
183.5
172.8
156.5
150.0
144.0
28.5
28.6
29.5
27.8
~~~
181.4
168.2
155.3
149.0
139.0
F. THEALKALINEEARTHIMIDES
The alkaline earth imides have a cubic, NaC1-type lattice and their
lattice energies have been evaluated by A. P. Altshuller ( 4 ) . Since the
conipressibilities mere unknown he used the Born-Lande equation and
adopted the value of n used by Sherman for the corresponding oxides.
He did not incorporate a term to allow for the dispersion energy and
found for the lattices energies of CaNH, SrNH and BaNH values of
787, 752 and 711 kcal/mole, respectively. Only the heat of formation of
BaNH was available and from this a value of AH,ONH- of 261 kcal/mole
and hence a double electron affinity of -184 kcal/mole was obtained.
f:.
THE ALKALIMETALAMIDES
The lattice energies of the alkali inetal ainides were initially crudely
computed by Juza (@), who used density measurements to calculate
a molecular volume and hence a radius of 1.75 A for the amide ion. He
used this radius to obtain the lattice energy. Yatsimirskii (133) obtained
a thermochemical radius for the amide ion of 1.30 A. This value is widely
at variance with both Juza's original value and the crystallographic value
of 1.67 A obtained by Juza and Opp ( 6 3 ) .Neither Juza nor Yatsimirskii
allow for the dipole moment that the amide ion possesses nor do their
lattice energies produce very consistent values for hHf0NH2-. If the
mlue of 1.67 A for the radius of the ainide ion is used in Kapustinskii's
forniula the values for the lattice energies obtained are close to those
given by Juza. However, the values for aH,0NH2- are still not consistent and the best that can be said is that the values for the lattice
energies are doubtful.
H. THEALKALIMETALCYANIDES
The lattice energies of the alkali metal cyanides, which a t room temlierature crystallize with the KaCl and CsCl structures, were calculated
197
LATTICE ENERGIES
by Sherman (114). Yatsimirskii (133) has also calculated a thermochemical radius of the cyanide ion of 1.82 A and the values obtained
from the Kapustinskii equation using this radius are given in Table
XVIII. The present author has calculated the lattice energies of these
salts, using a simple Born-Mayer expression, ignoring the dispersion
energy and also using the more complicated expression of Huggins. In
both p was taken as 0.345 A. The values obtained by the latter method
are surprisingly near those given by Sherman and are probably as good
as any available. They lead to a value of AH,OCN- (g) = 7 kcal/mole.
The values of Yatsimirskii are less reliable.
TABLE XVIII
THE LATTICE
ENERGIES
(KCAL/MOLE)
OF THE ALKALIMETALCYANIDES
Waddington
(9
Sherman
Yatsimirskii
Simple
Born-Mayer
U AHjCN-
Salt
AHjOCN-
AHjaCN-
LiCN
NaCN
KCN
RbCN
CsCN
169.4
154.9
149.1
141.3
4-3.1
4-6.1
$6.3
4-6.6
192
179
162
155
149
+12.7
+13.2
4-12.2
f13.3
174
158
151
139
4-7.7
4-9.2
+8.4
4-3.3
(ii)
Huggins
U
AHfOCN-
170.2
156.5
150.3
142.4
$3.9
$7.7
+7.5
$7.7
I. THEALKALI METALBOROHYDRIDES
The lattice energies of the alkali metal borohydrides, which have
the NaCl structure, have been calculated by Altshuller (3). He employed a simple Born-Lande expression and chose the values of n. appropriate to the corresponding fluoride, which is isoelectronic with the
borohydride. He added a dispersion energy term. The values he obtained
for the lattice energy were: NaBH4, 168; KBH4, 159; RbBH4, 155;
CsBH4, 150 kcal/mole. These values lead to a value of -23 f 5 kcall
mole for AH,OBH~-(g) and hence to a value of -75 f 5 kcal/mole for
the enthalpy of the gaseous reaction
BHs
+ H-
= BHd-
J. THEALKALIMETALSUPEWXIDES
The lattice energies of the alkali metal superoxides have been calculated by Evans and Uri (56) and by Kazarnovskii (71).Evans and Uri,
198
T. C. WADDINGTON
unlike Kazarnovskii, did not calculate accurate Madelung constants for

the lattice and their values are inferior to his for that reason. The results of both calculations are given in Table XIX.
TABLE XIX
THELATTICEENERGIES
(KCAL/MOLE)
OF THE
Evans and Uri
Salt
~
KO?
RbOz
CSOZ
ALKALIMETALSUPEROXIDES
AHjOOC-
168
162
156
-22.1
- 19.4
-16.2
- 17.9
- 13.3
-7.7
172.2
168.2
164.5
AHjO02-
~-
Kazarnovskii
K. THEALKALIXE
EARTH
PEROXIDES
The lattice energies of the alkaline earth peroxides have been calculated by Evans and Uri (35) and by Vedeneev et al. (124). There is
considerable discrepancy in the two sets of results but that of the Russian workers is preferred for the reasons given in Section J.
TABLE XX
THELATTICEENERGIES
(KCAL/MOLE)
OF THE
ALKALINEEARTHPEROXIDES
Evans and Uri
Russian workers
Salt
L'
A HjOOP
AHfOO2'-
CaO?
Sr02
Ba02
735
698
647
+116.7
+100.8
752.9
717.0
+170.8
+113.9
+171.6
L. THEALKALINEEARTHACETYLIDES
The lattice energies of the metallic carbides have been calculated
by L. I. Kazarnovskaja (70) who obtained values of CaC2, 813.7; SrC2,
769.4 and BaC2, 720.8 kcal/mole. I. Hodes (50) has repeated Kazarnovskaja's calculation and obtained values of AH,OCZ2- (g) by means
of a Born cycle. He does not give values for the lattice energies but
reports a value of -245 kcal/mole for
(g) .
31. THE LATTICE
ENERGIES
OF
ALKALIMETALAZIDES
The lattice energies of the alkali metal azides have been evaluated
by Gray and Waddington ( 4 3 ) , ti) by direct computation, using the
THE
199
LATTICE ENERGIES
method of Huggins and allowing for the effect of the quadrupole moment
of the ion, (ii) by using the equation of Kapustinskii, and (iii) by computation of hydration heats and measurement of the heats of solution.
The values obtained by the different methods are given in Table XXI.
The values from (i) and (iii) are in good agreement but the values from
(ii) are low. An average value of hH,0Ns-(g) = 35 kcal/mole is obtained.
TABLE XXI
OF
THE LAITICEENERGIES
(KCAL/YOLE)
THE
ALBALI METALAZIDES
Hydration heat
Azide
LiN3
NaNa
KNa
RbNs
CsNs
Kapustinskii
formula
184
168.5
153
147
140
Term-by-term
calculation
(a) Ionic
radius
191
173
155
149.5
143.5
157
150
143
(b) Lyotropic
number
194
175
157
151.5
145.5
N. THEALKALIMETALBIFLUOBIDES
The lattice energies of the alkali metal bifluorides have been recently
computed by Waddington (1.37) by extended calculation. Values obtained are: KHF2, 153.4; RbHF2, 147.3; CsHF2, 143.3 kcal/mole. They
lead to a value for AHfOHF2- (g) of -150
5 kcal/mole.
0. THBALKALIMETALCYANATES
Yatsimirskii (134) has computed a thermochemical radius of 1.59 A
for the cyanate ion and from this the lattice energies of the alkali metal
cyanates may be derived by substitution in the Kapustinskii equation.
However the value 1.59 A for the radius seems to be much too low. The
alkali metal cyanates have lattice parameters almost identical to the
alkali metal azides and Gray and Waddington have deduced an average
radius for the azide ion of 2.04 A. A similar value would be expected for
the cyanate ion. Adopting the value 2.04 A, the author has (i) used it
in the Kapustinskii equation to deduce lattice energies and (ii) used it
in the hydration enthalpy versus crystal radii interpolation to deduce
hydration enthalpies for the cyanates. Unfortunately only the lattice
energies of sodium and potassium cyanates can be deduced because the
enthalpies of solution of the other cyanates are unknown. The charge
distribution on the cyanate ion is not known and this makes the ex-
200
T. C. WADDINGTON
tended calculation of cyanate lattice energies yery difficult. However

a preliminary calculation has been made by the author for potassium
cyanate. The results obtained are summarized in Table XXII. A value
of -63 kcal/mole is adopted for AH,ONCO-(~).
TABLE XXII
THELATTICEENERGIES
Yatsimirskii
Salt
LiNCO
NaNCO
KNCO
RbNCO
CsNCO
G
2003
193
173.5
165.5
158
(KCALiMOLE) OF THE
Iiapustinskiis
formula with
radius of NSO= 2.04 A
AH~NCO- c
-48.6
-48.1
184
169
I53
147
140
ALKALI1IETAL
Hydration hesc
from radius of
2.04 A
CYANATES
Extended
calculation
___
AHjoNCO-
AHJONCO-
AHfONCO-
-71.0
-67.4
176
156
-64.4
-64.2
158
-62.4
P.THEALKALIMETALTHIOCYASATE~
The lattice energies of the alkali metal thiocyanates can be calculated from Yatsimirskiis (133) thermochemical radius of 1.95 A or, in
the case of NaCNS and KCNS, from the hydration enthalpies obtained
ironi the Biichner interpolation by Waddington (126). The results arc
given in Table XXIII. There is a considerable discrepancy between
them. Waddingtons results lead to a value of -25.9 kcal/mole for
IH,OCKS- (g) and Yatsimirskiis t o a value of - 1 1.5 kcal/mole.
TABLE XXIII
THELATTICEENERGIES
(KCAL/MOLE) OF THE
ALKALIMETALTHIOCYANATES
Salt
1,iCX;S
NaCNS
KCNS
RbCNS
CsCNS
1-atsimirskii
L-
Value from
Iiydration heat
1&7.5
173
156.3
I 50
144
161
143
20 1
LATTICE ENERGIES
Q. THE ALKALIMETALNITRATES
The only direct calculation of the lattice energy of an alkali metal

nitrate was made by Topping and Chapman (119) in 1927. They calculated an electrostatic potential of an array of Na+, N6+, and 02ions, not including the self energy of an N6+032-group, so that this
term corresponds t o a Madelung energy for an array of point charges
at the centers of the NOS- and Na+ ions plus a multipole energy term,
the multipole having the charge distribution indicated below.
-2e
-2e
0
+N6e
II
-2e
This is obviously too high a charge distribution, so that the size of this
term would be expected to be too large. The repulsive energy term was
represented as l/P,n being taken as equal to 10; a value of 178 kcal/mole
was obtained. Hojendahl (6.2)has calculated an electrostatic energy of
180 kcal/mole and using a simple Born-Lande expression the present
author finds a value of 162 kcal/mole. This value is obviously too low,
as no allowance has been made for the dispersion energy or for the
multipole energy; these two terms together may well be of the order of
10-15 kcal/mole. Yatsimirskii (133) has calculated a thermochemical
radius of the nitrate ion of 1.89 A and the values obtained for the lattice
energies from this value are given in Table XXIV. Waddington (1%)
has also obtained the lattice energies from the value of the hydration
heats of the ion pairs obtained by the Buchner extrapolation. Yatsimirskii (153) gives a value of AHt0No3-(g) = 80 kcal/mole while Waddingtons results lead to a value of -81.4 kcal/mole.
TABLE XXIV
THELATTICE
ENERGIES
(KCAL/MOLE)
Salt
LiK03
NsNOa
KNOs
RbNOa
CsNOs
Kapustinskii
formula
U
187
176
160
153
148
OF THE
ALKALIMETALNITRATES
Chapman
and Topping
Hydration
heat value
Hojendahl
178
195
176
159
155
145
162
202
T. C. WADDINGTON
R. THEALKALINEEARTHAND OTHER METALCARBONATES

The lattice energies of the carbonates of several divalent metal ions
were calculated by Lennard-Jones and Dent (80). They calculated the
lattice energy of an array of M2+,
C4+,and 02- ions, not including the
self energies of the C4+082- groups. This is equivalent to calculating a
Madelung energy for an array of point charges at the centers of the
M2+ions and the centers of the C032-ions plus a multipole energy, the
multipoles having the distribution
-2e
0
-2e
0
+6e
-2e
This is obviously too high a charge distribution so that one would expect the total electrostatic energy to be too large. The repulsive energy
was represented as l/rn, n being taken equal to 10-11. However no term
was included for the dispersion energy. Hojendahl (52) has calculated
Madelung energies for a number of divalent metal carbonates with the
calcite structure, and using the repulsive energy term l/r" with n = 10
the author has calculated the lattice energies given in Table XXV from
Hojendahl's data. These are all probably too low as no permanent multipole or dispersion energy terms have been included. Kapustinskii and
Yatsiniirskii have calculated a thermochemical radius of 1.85 A for the
carbonate ion and the values obtained on inserting this into the Kapustinskii equation are also given in Table XXV. It will be seen from the
TABLE XXV
THELATPICE
ENERGIES(KCAL/MOLE)
OF THE ALKALINE
EARTH
AND OTHER METAL
CARBONATES
Kapustinskii and
Yataimirskii
Lennard-Jones
and Dent
Hojendahl plus
repulsive term
Salt
AHfCOa'-
AHf"C0s"
AHjOC03"
MgCOs
760
714
650
625
746
700
728
746
-67
-37
64
-61
-112
- 105
-86
-72
771
701
-56
-50
696
644
- 107
775
715
747
762
-83
693
652
665
G86
CaCOs
SrC03
BaCOs
ZnCOa
CdCOs
hfnCOa
FeC03
-90
-67
-66
-131
- 166
- 153
- 150
-142
203
LATTICE ENERGIES
table that the values for aHPC03-(g) are not very consistent in any
of the sets of data, nor do the sets of data agree with each other very well.
The internal inconsistency may be a reflection of nonionic bonding in
some of the heavy metal carbonates.
S. THE ALKALI METALBOROFLUOIUDES
The lattice energies of the cubic, high temperature forms of NH4BF4
and KBF4 have been calculated by Altshuller ((2). The lattice energies
of KBF4, RbBF4, and CsBF4 were also calculated approximately by de
Boer and Van Liempt ((28)
. Kapustinskii and Yatsimirskii (68) have
calculated a thermochemical radius of 2.28 A for the BF4- ion from
which the lattice energies of the alkali metal salts may be found. The
values obtained are given in Table XXVI. It will be seen that they are
all fairly concordant and lead to a value for aH,0BF4-(g) of -406
kcal/mole.
TABLE XXVI
TEE LATTICE
ENERGIES
(KCAL/MOLE)
Ruesian workers
Salt
AHj'BF4-
OF THE
ALKALIMETALBOROFLUORIDES
de Boer and
Van Liempt
AHf'BF4-
Altahuller
AHf'BF4-~
LiBF4
NaBF4
KBF4
RbBF4
CsBR
NHIBF,
167
157
143
138
133
-408
-406
-408
-403
-405
148
135
129
-403
-406
-409
144
-407
136
IV. Uses of Calculations of Lattice Energies
A. AHtoX- (g) AND THE DETERMINATION

OF ELECTRON
AFFINITIES
The application of the Born-Haber cycle to a theoretically determined lattice energy actually gives a value for AHtoX-(g), or
AHIoX(g) - E , and a knowledge of AHfox(g) is required before a
value can be assigned to the electron affinity. Alternatively, if a value
of the electron affinity is available from other sources a value for the
enthalpy of formation of the free radical AHtoX(g), can be obtained.
However in some cases such as NO3, 0 2 , NO*, X is not a free radical
but a stable molecule whose enthalpy of formation is known and then
the electron afhity can be found directly. In some other cases, for example, the alkali metal halides and the alkaline earth oxides, a bond
204
T. C. WADDINGTON
dissociation energy is needed in addition to the lattice energy and then

a value for the electron aftinity can be found. The values obtained for
AH,OX- (g) from the lattice energies discussed in the preceding section
are listed in Table XXVII.
TABLE X X V I I
VALVESOBTAISEDFOR A H / X - ( g )
Value
_____-
(KCAL/MOLE)
Group
Vnlae
15
26 1
i. 0
-23
217
152
16F
145
50
-31
___
c
2
1
-
N3HFZ-
CNON03BF,-
-19
171
Value
-245
35
150
-63
-81
-406
B. THEDETERMINATIOS
OF ABSOLUTEVALUES
OF PROTON
AFFIXITIES
The classical example of the use of lattice energies to determine proton affinities is the case of the ammonium ion. Grimm (&) pointed out
that the following cycle could be used to obtain the proton affinity of
ammonia.
The proton affinity of ammonia is therefore given by the relation
+ 2RT) + I,
Ex AHjoX&(g) AHjoH(g)
AH/OX(g) - AHjONH,X(s)
= - (I'
2RT) t AHj"H+(g) AHjoNHa(g) AHjoX-(g)
- AH,ONH&
where I , is the ionization potential of the hydrogen atom.
Values for the lattice energies of the ammonium halides were originally calculated by Griinni ( 4 5 ); later van Arkel and de Boer (122)
gave a value of 148 kcal/mole for the lattice energy of NH4Cl. Probably not too much reliance can be placed on these values. Rleick (11)
P?;H*=
-(
205
LATTICE ENERGIES
using the extended Born-Mayer method, calculated the lattice energies

of all the ammonium halides. His values together with the resulting
proton affinities are given in Table XXVIII. The actual value obtained,
209 kcal/mole, differs by about 3 kcal/mole from that given by Bleick
because more recent thermochemical data have been employed. It will
be seen that there is a discrepancy in the lattice energy of ammonium
fluoride, which has the wurtzite structure (4:4 coordination) ; the reason is that the actual lattice energy of NH4F contains an appreciable
energy term due to hydrogen bonding.
TABLE XXVIII
THELATTICEENERGIES
(KCAL/MOLE) OF THE AMMONIUM
HALIDES
AND THE PROTON
AFFINITYOF AMMON~A
NH$
NH&I
175.2
161.6
154.0
-162.8
- 176.4
PNH,
NH4Br
- 155.2
NHif
145.5
- 146.7
+111.6
367.08
-64.8
-11.04
+75.38
367.08
-59.2
-11.04
f64.61
367.08
-54.9
-11.04
+48.3
367.08
-49.1
-11.04
227.6
209.4
210.5
208.5
The method outlined above is capable of extension to a number of

groups though no very wide use has been made of it. Sherman (114)
calculated the proton affinity of water by assuming that the lattice
energy of hydroxonium perchlorate (H@-C1o4-) was the same as that
of NH4+C101- since the unit cells of the two solids are isomorphous
and very nearly the same size. He obtained a value of 182 kcal/mole.
The proton affinities of the methylamines could be obtained by this
method if their lattice energies were known. Certainly_ approximate
values could be obtained by the use of the Kapustinskii equation.
TABLE XXIX
PHOTON
AFFINITIES (KCAL/MOLE)
Group
Proton aflinity
NHa
NHe-
NH*
209
393
613
Group
Proton sfsnity
HsO
OH-
182
375
Group
Proton a i t y
'0
554
SH-
S'-
342
550
206
T. C. WADDINGTON
The lattice energies of the alkali metal hydroxides, the alkali metal
amides, the alkaline earth chacogenides and the alkaline earth imides,
give values for aHf0OH-(g) etc. and su enable the proton affinities of
these groups to be calculated. The values obtained are given in Table
XXIX.
C.
THEABSOLUTEENTHALPIE~
OF FORMATION
OF COMPLDX
IONS
In aqueous solution the factors which affect the enthalpy of formation of a complex ion are complicated by the presence of an effect caused
by a change in the enthalpy of solvation. If we consider the reaction
BFdaq)
+ F-(aq)
+ BFd-(aq)
we see that a fluoride ion has lost its solvation sheath and that a borofluoride ion has been solvated. In the similar reaction in the gas phase
BFdg)
+ F-(g)
+ BF4-k)
there are no solvation effects and the only effects are due (1) to a
change of hydridization of the original bonds and (2) to the formation
of a new bond. The calculated lattice energies of the alkali metal salts
enable values for aH,OF-(g) and aHf0BF4-(g) to be assigned and
hence the enthalpy of the reaction can be calculated. So far there are
data available for only a few complex ions, for example, HFz-, BH4-,
BF,-, and the values for their enthalpies of formation are given in
Table XXX.
TABLE XXX
THE ENTHALPIES
OF FORMATION
OF SOMECOMPLEX
IONS
Complex
ion
Reaction
Enthalpy
(kcal/mole)
BF4BH4HR-
BFI + F - +BF4BHI H- + B H I
HF + F - - + H F s -
-76
-75
-58
D. LATTICE
ENERGIES
AND NONIONIC
CONTRIBUTIONS
TO BONDING
I N CRYSTAL
LATTICE
Sherman (114) and other workers have compared crystal energies
obtained by the Born-Haber cycle from experimental thermochemical
data with theoretical values calculated assuming strict ionic character.
The differences obtained have been used to indicate deviations from
strict heteropolarity. Sherman (114) in his review gives results for 50
crystals, the computations being made with the Born-Lande equation.
207
LATTICE ENERGIES
Later workers have used the Born-Mayer equation and the extended
Mayer-Huggins method, allowing for the Van der Waals and zero-point
energies. The computations made and their comparison with thermochemical data, utilizing the most recent data, either from the U . S.
Bureau of Standards Circular 500 or from later sources, will be considered below.
1. The Halides of Univalent Metals
(Other than the Alkali Metals)

After Sherman (114), the lattice energies of the argentous, the thallous and the cuprous halides were recalculated by Mayer (9g) and by
Mayer and Levy (95). Later Altshuller ( 1 ) and Morris and Ahrens
(100) corrected the earlier experimental values for the lattice energies,
using new values for the heats of sublimation of the metals. Ladd and
Lee (7'6) have recently recalculated the lattice energies of the silver and
thallous halides by a method avoiding the use of the Huggins basic radii,
which are difficult to fix for these salts. The values obtained, together
with the best values for the thermochemical lattice energies are given in
Table XXXI.
It will readily be seen that the deviations between the theoretical
TABLE XXXI
LATTICEENERGIES
(KCAL/MOLE) OF SOMEUNIVALENT
HALIDES
Cycle values
Salt
AgF
AgCl
AgBr
Ad
T1F
TlCl
TlBr
T1I
CuF
CuCl
CuBr
CUI
AuF
AuCl
AuBr
AuI
Sherman
Mayer;
Mayer
and Levy
Ladd and
Lee
207.9
187.3
181.3
175.9
219
203
197
190
220
199
195
186
159.3
154.9
148.4
167
164
159
164
159
152
206.1
197.7
187.7
216
208
199
Morris and
Ahrens
23 1
219
217
214
198
176
173
168
234
232
229
249
249
251
Ladd and
Lee
228
216
214
211
175
172
166
208
T. C. WADDINGTON
and experimental lattice energies for the halides of a given metal are
least in the fluoride and increase through the chloride and bromide and
are greatest in the iodide, in accordance with the increasing polarizability and decreasing electronegativity of the halide. Again for the
metal salts of a given halide the deviations increase in passing from
the thallous through the argentous to the cuprous salts. This is, as Morris
and Ahrens point out, in the order of increasing ionization potential and
decreasing radius of the ion. It is interesting to note that aurous iodide,
the only aurous halide whose structure has been report,ed! has a lattice
consisting of chains of --i\u--I-Au-I-Au-.
2. The Halides of the Divalent Metals
Sherman (11.6) originally presented a good deal of data on the theoretical and thermochemical lattice energies of the divalent metal halides,
the calculations being made with the Born-Lande equation. Morris (98)
has extended the theoretical calculation, again using the Born-Lande
equation and has recalculated the thermochemical data. McClure and
Holmes (97) have also recalculated the thermochemical data for the
lattice energies of the transition metals. The results of these calculations
are given in Table XXXII.
Before these results are discussed however the occurrence of a new
energy term in those crystals in which the metal ion has an unfilled
d shell must be considered. Under the influence of the electrostatic field
of the neighboring negative ions the otherwise degenerate orbitals of the
metal ion are split; the nature of the splitting depends on the symmetry
of the crystal field. This splitting produces a stabilization energy. For
an account of the theory and references to earlier work the reader is
referred to the review articles by Orgel (101) and Griffith and Orgel
(4.41. Orgel (101) pointed out that if the thermochemical lattice energies
of the divalent transition metal halides were plotted against their atomic
number a minimum occurred a t the manganous salt, which has the d5
configuration. Recently Hush and Pryce (57) have repeated the Orgel
plot and drawn a smooth curve through the values of the lattice energies
of the calcium, manganous, and zinc salts. Here the metal ions have the
configurations 3d0, 3d5, 3dI0, so that in these salts the crystal field
stabilization energy is zero. The difference between points on this smooth
curve and the actual thermochemical values of the lattice energies gives,
to a first approximation, the crystal field stabilization energies. More
refined calculations carried out on the oxides of the transition metals
suggest that the approximation yields values which are rather low (57).
The variation of thermochemical lattice energy with atomic number is
209
LATTICE ENERGIES
70
55
50
Ca Sc Ti V Cr MnFe C o N i Cu
Zn
Atomic number
FIQ.3. Crystal field stabilization of dihalides of first row transition metals.
shown in Fig. 3 and the values of the crystal field stabilization energy
are given in Table XXXII.
It will be seen from Table XXXII, that after allowing for the crystal
field stabilization energy, there is very little discrepancy in the case of
TABLE XXXII
LATTICEENERGIES
(KCAL/MOLE) OF DIVALENT
HALIDES
-~
Slllt
Structure
MgF2
CaF2
TiF?
VFZ
CrF?
Mn
FcF?
COF?
NiF2
CUP?
ZnF?
BaF2
SrF?
PbF2
CdF?
HgF?
SrC12
MgBr?
MnBr,
FeBr?
CoBr?
MRIZ
CaI?
nI,
MnI?
FeI,
COI?
PbI2
CdIz
A
Rutile
Fluorite
Itutile
Itutile
Rutile
Rutile
Rutile
ltu tile
Fluorite
Fluorite
Fluorite
Fluorite
Fluorite
Fluorite
CdI?
CdI,
CdIz
CdI2
CdI?
CdIZ
CdI2
CdIz
CdI?
CdI?
CdIr
CdI,
Ionization
Dotential ev.
15.03
11 37
15.64
16.18
17.05
18.15
20.29
17.96
10.00
10.03
15.03
16.90
18.751
11.03
15.03
15.64
16.18
17.05
13.57
11.87
13.57
15.64
16.18
17.05
15.03
16.90
Sherman
696.8
617.7
- 1.8
+6
656.3
657.7
$6
697.1
689.6
+3 1
+37
+100
556.4
587.5
580.7
628.7
+14
+33
493.6
+12
461
477
+49
+Q9
Morris
684
622
+11
+2
663
681
688
694
627
688
561
592
584
634
626
496
513
522
530
536
486
455
486
481
493
502
460
477
0
+15
+a
+34
++22
100
+5
-4
+11
+28
+29
10
+62
+61
+77
+88
+62
+37
+62
+82
96
+lo3
+50
+99
Born-Haber
cycle
695
624
657 15f
671.5 f 20t
688 f 3
662
696
708
728
727
710
566
588
595
662
655
506
575
583
607
624
548
492
548
563
589
605
510
576
Crystal field
stabilization
enerw
0
0
17++
23+
30++
0
21.8
24.8
35.0
27.0+
+
0
0
0
0
0
0
0
0
0
16++
20f +
0
30.6
0
14.8
26.5
0
0
Represents the differencebetween the Born-Haber cycle and the calculated values of the lattice energy.
* There seems to be some doubt as to the structure of CuF,, which perhaps explains the large discrepancy between Morris and Sher-
mans vdues.
These values of the crystal field stabilization energy have been calculated by the author, the rest are from the paper by Hush
and Pryce (16).
$ Estimate only.
++
211
LATTICE ENERGIES
the fluorides between the calculated lattice energies and those obtained
from the Born-Haber cycle. This indicates that the fluorides are predominantly ionic. The small discrepancies which arise in the case of
CdFz and HgFz are probably no larger than could be accounted for by
a London dispersion term, which has been neglected in the simple calculation. Practically no data are available for the chlorides but by the
time the bromides have been reached the discrepancies are large and
cannot be explained by the crystal field stabilization energies, which are
too small. The discrepancies are even larger in the case of the iodides.
Within a given series of halide salts the size of the deviation increases
with increase in the ionization potential of the metal ion, as was pointed
out by Morris (98).
3. The Chalcogenides of the Monovalent Metals
(Other than the Alkali Metals)

A few lattice energies of these compounds have been calculated by
Sherman (114) and his values together with the values given by the
Born-Haber cycle are given in Table -111.
It will be seen that the
TABLE XXXIII
(KCAL/MOLE) OF SOME CHALCOQENIDES
THELATTICEENERGIES
OF MONOVALENT
METALS
Salt
Structure
Sherman
Cycle
Ago0
Cuprite
cuprite
Fluorite
Fluorite
7.574
7.724
7.724
7.724
585
682
612
599
+129.1
+95.7
+61
+94.3
714.1
777.7
672.7
693.3
cue0
cues
CusSe
differences between the thermochemical and the calculated values are

very large, so that these crystals cannot be considered to be primarily
ionic.
4. The Chalcogenides of the Divalent Metals

The lattice energies of the chalcogenides of the alkaline earth metals
and beryllium have already been discussed in Section I and it was noted
that even in these compounds there appeared to be a non-ionic contribution in the sulfides, selenides, and tellurides. Values for the lattice
energies of other chalcogenides are given in Table XXXIV, being taken
from Sherman (114) and from Hare and Brewer (4.8) for some oxides.
The size of the crystal field stabilization energy, given under (i) as a
first approximation and under (ii) as the value obtained by a more
!N
2
ThBLE XXXIV
LAWICEENERGIES
(KCAL/YOLE) OF SOME DIVALENT
METALCHALCOGENIDES
Cryntal field
stabilization
Salt
Structure
I
(ev)
CaO
Ti0
NaCl
NaCl
NaCl
NaCl
NaCl
NaCl
NaCl
Wurtzite
NaCl
11.87
13.57
14.71
15.64
16.18
17.05
18.15
17.96
16.90
Sphalerite
Wurtzite
Wurtzite
Sphalerite
NaCl
NaCl
Sphalerite
Sphalerite
Sphalerite
NaCl
17.96
17.96
16.90
18.75
15.03
15.64
17.96
16.90
18.75
15.64
vo
MnO
I'eO
COO
NiO
ZnO
CdO
HgO
Sherman
912
944
950
-0.6
-6.4
4-4.1
977
867
- 12.6
4-37.2
819
816
770
774
705
788
4-45.5
+45.3
+42.7
+79.5
f35.4
f13.3
+73.1
f73.3
+117.9
4-27.6
Hare and
Brewer (i)
846.7
959.1
990.1
915.9
937.6
957.6
974.4
-18.9
-31.3
-54.0
-3.5
0
-3.5
-0.2
Hare and
Brewer (ii)
Cycle
(i)
833
916
940
890
907
917
934
-4.8
+11.8
-3.9
+21.4
+30.6
+37.1
+40.2
828.2
927.8
936.1
911.4
937.6
954.1
974.2
964.4
904.2
0
62.5
55.2
0
13.1
18.9
26.2
0
0
PbO
ZnS
ZnS
CdS
HgS
PbS
MnS
ZnSe
CdSe
Hgse
MnSe
790
745
749
762
864.5
861.3
812.7
853.5
740.4
801.3
863.1
808.3
866.9
789.6
0
0
0
0
0
0
0
0
0
(ii)
74.1
63.9
13.6
20.3
29.1
213
LATTICE ENERGIES
refined treatment, as calculated by Hush and Pryce (57) is also given

in Table XXXN.
It will be readily seen from the table that the values for the oxides
given by the simple Born-Lande [Hare and Brewer (i)] and the simple
Born-Mayer [Hare and Brewer (ii)] expressions show quite large divergences, as much as 50 kcal/mole. The differences between these values
and the cycle values in no way parallel the crystal field stabilization
energy and it is quite evident that the accuracy of these simple calculations is probably less than the size of the crystal field stabilization
term. All that can be said is that, within the accuracy of the calculations there appears to be no evidence for any appreciable nonionic character in bonding. It is interesting to compare this with the case of the
fluorides of the same elements discussed in the previous section where
the same appears to hold true. The values for a few sulfides and selenides
are also given in Table XXXIV and it will be seen that here the differences between the calculated and the thermochemical values increase in
size in passing from sulfide to selenide and in passing from zinc through
cadmium to mercury, as would be expected from considerations of decreasing electronegativity in the anions and increasing ionization potential in the cations.
5 . The Oxides of Some Trivalent Metals
Sherman (114) has calculated the lattice energies of A1203 and Crz03.
No other calculations appear to have been made. In Table XXXV the
calculated values are compared with those obtained thermochemically.
TABLE XXXV
THE LATTICEENERGIES
(KCAL/MOLE)
Salt
Structure
OF
OXIDESOF
Sherman
~
AlzOs
Cr.08
Corundum
Corundum
THE
1307.4
1310
3708
3655
TRIVALENT
METALS
Cycle
A
~
~~
-45
-113
~~~
3663
3542
It will be seen that in both cases the calculated values are higher than
the thermochemical ones and that the differences are relatively small.
This indicates the essentially ionic nature of these compounds.
6. The O d e s of the Tetravalent Metals
Sherman (11.6) has calculated the lattice energies of SnOz and PbOz.
These calculated values, together with the thermochemical values are
given in Table XXXVI. The calculated value for SnOz is very close to
214
T. C. WADDINGTON
TABLE XXXVI
THEL A ~ I CENERGIES
E
(XCAL/MOLE) OF T H E oXIDi%SOF
Salt
Structure
Sherman
SnOr
PbOz
Rutile
Rutile
2143
2264
2734
2620
T H E TETHAVALENT M E T A L S
- 19
$153
Cycle
2715
2763
the thermochemical value, while for PbOz the difference between the
two is about 6%. In view of the uncertainty of the calculation this cannot be taken as very definite evidence of nonionic character.
E. THEDETERJIINATIOX
OF THE STABTLITY OF
IIYPOTHETIChL COMPOL-XDS
In 1923 Grimm and Hersfeld (46) calculated the lattice energies of

a whole series of hypothetical compounds. From these lattice energies
they obtained, by the use of the Born-Haber cycle, the enthalpies of
formation of the compounds. To obtain the lattice energies of some of
the hypothetical compounds they used the Born-Lande equation. For
some of the other compounds they assumed their lattice energies to be
equal to those of neighboring stable compounds in the periodic table,
for example
UNeC1
UNaC11 UIiaCh = UMgClt, UWgCl
UNaCll UAlClz = UMgC12.
Now from the Born-Haber cycle
+ + 1)RT = AHfM+(g) + nAH,OX-(g) - AHfoMX,(~)

(50)
L ~ ~+
~ ~ x+
m 1)RT = AH,OM+(g) + nAHjX-(g) - AHf0MXn(s)
U ~ t s . (n
(71
(51)
and so, if Lwx,
CM~~X.
AH,~MX,(S) - AHfoRIIX, (s) = AHfOM+(g) - AH,OM+ (g)

or
AHj0MXn(s) = AHfoM+(g)- AHjohln+(g)
+ AH~MX,(S).
(52)
(53)
The assumptions inherent in this are (i) that the lattices considered,
both real and hypothetical, are totally ionic, or if not that the nonionic
contribution is the same in each, and (ii) that the lattices have the
same energy, that is that the ions M and M have the same radius. Neither
of tlicse assumptions can be expected to be completely true, for cxample N a f , Nef and M g + are not isoelectronic, but the deviations resulting from the assumptions should be small. Though the results of Grimin
and Herzfeld need recalculating, because of the change in the values of
215
LATTICE ENERGIES
the thermochemical quantities employed, their underlying conclusions

remain unchanged. They calculated by the method outlined above the
enthalpies of formation of the hypothetical monohalides and corresponding chalcogenides of the inert gases and of the alkaline earth metals, and
of the dihalides and corresponding chalcogenides of the alkali metals
and of aluminum as well as of other compounds. The results given in
Table XXXVII follow their treatment, except that more modern thermochemical data are used. In Table XXXVII the monohalides are also
considered.
TABLE XXXVII
ENTHALPIES
OF
(KCAL/MOLE) OF MONOHALIDES
AND OXIDES
GASESAND THE ALKALINEEARTH METALS
FORMATION
OF THE INERT
Salt
M
MF
MC1
MBr
MI
Mz0
He
Ne .
A
Kr
Xe
Be
Mg
Ca
Sr
255
+208
+I00
+68
+39
- 14
-68
-69
-78
-73
+303
+246
+316
+258
+141
+lo7
+72
+48
- 18
-28
-39
-41
-37
-33
-44
-32
-27
-30
+336
+275
+156
+121
$660
+589
Ba
Zn
Cd
Hg
Ga*
In*
A1
-58
-68
-72
-94
+130
+97
62
$34
-30
-38
-49
-50
-44
-40
-48
-45
-41
-45
+a
+67
-1
- 13
-24
-27
-28
-24
-39
- 13
- 13
-11
+382
+320
$316
+122
+36
+44
+28
+31
-63
-26
* The values for Ga and In are based on an assigned ionic radius.

It will be seen from this table that the monohalides of the inert gases
all would be unstable to decomposition into their elements and so obviously could not exist in an ionic lattice. However this is not so for the
other monohalides all of which, with the exception of BeCl, BeBr, and
BeI, are stable with respect to decomposition to the elements. In these
compounds the question arises as to their stability with regard to their
disproportionation. In the case of the alkaline earths, zinc and cadmium,
the disproportionation reaction is
2 MC1-t M
+ MC12.
216
T. C. WADDINGTON
In the case of Hg, the transition 2HgX- Hg2Xnmust be considered and

in the case of indium and gallium the disproportionation to the trihalide
will be considered. It is interesting to note in passing that the value
AHfoInCl= 41 kcal/mole is quite close to the thermochemical value
quoted in the U. S. Bureau of Standards, circular 500. The enthalpies
of disproportionation are given in Table XXXVIII. The conclusions that
TABLE XXXVIII
ENTHALPIEY
OF DISYHOPORTIONATION
(KCAL/MOI.E)
OF THE
MONOHALIDES
Salt
c1
MF
xrg
- 148
- 152
- 134
- 140
-114
-100
- 106
- 49
-11
-13
Cll
Sr
Bll
(1) 2MX+
Zn
Cd
-93
Br
l m 2
-88
- 105
-93
-98
-4
-9
(2) 2HgX -+ HgZXT

-7
(3) 3MX -+ 2M
Ga
In
A1
+10
-5
-29
-31
-84
- 102
-92
+6
+4
-5
+ MX?
+6
- 16
-36
-12
- 16
42
can be drawn from the table are very interesting. As would be expected,
it appears that no monohalide of the alkaline earth metals is stable to
disproportionation. However, when the monohalides of zinc and cadmium
are considered, the enthalpies of disproportionation are much smaller
and in fact positive values are obtained for iodides. However, the accuracy of the calculation does not permit a firm conclusion to be drawn
and, for exact statements about equilibria, free energies are required.
With both gallium and indium the enthalpies of disproportionation of
the monohalides are small and in some cases positive. These are of course
just the elements in which unipositive states, of low stability, do appear
to occur. The experimental evidence is fairly strong for indium but
weaker for gallium. The low enthalpies of disproportionation of. the
aluminum halides are interesting.
G r i m and Herzfeld also pointed out the instability of the hypothetical dihalides of the alkali metals. The values of the enthalpies of
217
LATTICE ENERGIES
formation of these compounds together with those of some other compounds of divalent states is given in Table XXXIX.
TABLE XXXIX
THEENTHALPIES
OF FORMATION
(KCAL/MOLE) OF SOMEHYPOTHETICAL
DIHALIDES
AND THE CORRESPONDING
OXIDES
Salt
M
Li
Na
K
Rb
CS
A1
CU
Ag
M F?
MCl,
MBrz
MI*
MO
+403
104
+39
-30
- 185
- 127
-49
+lo61
$513
+204
+131
+51
-65
-52
+23
+lo95
+533
233
+158
+76
-35
-34
+40
+1133
+580
+267
+I037
+522
+243
188
+I23
-55
-37
+55
+113
+2
-5
67
It will be seen from the table that all the dihalides of the alkali
metals with the exception of CsF2, are actually unstable to their formation from the elements. CsF2 is of course unstable to disproportionation, the enthalpy of the reaction
CSF~= CSF
+ )Fp
being -96.9 kcal/mole. All the hypothetical aluminum dihalides are

markedly unstable to the disproportionation reaction
3 AlX? = 2 AlXs
+ Al.
The well-known thermochemical values for the cupric halides are included for comparison with the values for the hypothetical argentic
halides, which are seen, with the exception of AgF2, to be unstable with
respect to their elements. The value for Ago is rather surprising in view
of the stability of this compound. This is obviously an example of the
approach breaking down because of the assumptions (i) and (ii) above
not being fulfilled. Ago is probably not mainly ionic and is not isomorphous with CdO. AgF2 is seen to be not only stable with respect to its
elements but also thermochemically neutral with respect to disproportionation. AH for
is approximately zero.
218
T. C. WADDINGTON
F. LATTICE
ENERGIES
AND
ON THE
OXIDATION
THE
EFFECT
OF FLUORIDES
A N D OXIDES
STATE OF
METALSALTS
It is a well-known truism of inorganic clicmistry that most metals

show their highest oxidation states in their oxides or fluorides. A consideration of the lattice energies of these compounds and the Born-Haber
cycle Kill reveal x h y this is the case. Consider a halide MX,, then
+ +
U
(72
1)RT = AHfoMn+(g) nAH,OX-(g) - AHfoMX,(s)
or rewriting the expression
AHjOMX,(s) = AHfoMn+(g) nAHj0X-(g) - U - (n 1)RT.

If the compound is to be formed the reaction has to be exothermic so
that, neglecting the small term (n + l ) R T , if AHfoMX,2(s)is to be < 0
then
C
> AHj0M"+(g) + nAHfoX-(g).
(54)
For all the halogens the term A H j o X - (g) = AHjoX - I3 is not very
different from -60 kcal/mole so that the factor determining n from
halide to halide will be U . The quantity U is always higher for the
fluorides than for the chlorides; it will be seen from Fig. 3 that the
lattice energies of the transition metal difluorides are all about 60
kcal/mole greater than those of t,he corresponding dichlorides.
A similar argument may be applied to the oxides. Consider the coinpound
then
AHfoM0,/2 = AHfoMn+(g) iAHfo02-(g)- U
(55)
so that if the compound is to be exothermic
> AHfoMn+(g)i-
AHj0O2-(g).
(56)
Here the lattice energy U is usually greater than that for the corresponding fluoride, for example U;\l,o = 938 kcal/mole, U M ~ F
= ,695 kcal/mole,
because of the double charge on the oxide ion. However this is offset by
the value of AHf0O2-(g),which is +217 kcal/mole, as compared to approximately -64 kcal/mole for Nf0F-(g).
REFEXEKCES
1 . Altshuller, A . P., J. Am. Chem. Soc. 77,5455 (1955).
2. Altshuller, A. P., J. Am. Chem. SOC.77,6187 (1955).
3. Altshuller, A. P., J. Chem. Phys. 22, 1136 (1954).
4. Altshuller, A. P., J. Chem. Phys. 23, 1561 (1955).
LATTICE ENERGIES
219
6. Benson, G. C., Can. J. Phys. 34,888 (1956).

6. Benson, G.C., and Van Zeggeran, F., J. Chem. Phys. 26, 1083 (1957).
7. Benson, G. C., and Wylie, G., Proc. Phys. SOC.AM, 276 (1951).
8. Bertaut, E. J., J. phys. radium 13,499 (1952).
9. Bertaut, E. J., U.S.A.F. Rept. AD-22696,July, 1953.
10. Bichowsky, F. R., and Rossini, F. D., Thermochemistry of Chemical Sub-
stances. Reinhold, New York; 1936.

11. Bleick, W. E., J. Chem. Phys. 2, 160 (1934).
12. Bollnow, 0.F., Z . Physik 33, 741 (1925).
13. Bormann, F.,Z . Physik 1,55 (1920).
14. Born, M., Atomtheorie des Festen Zustandes. Teubner, Leipzig, 1923 ;
Handbuch d . Physik 24, 420 (1927).
16. Born, M., Verhandl. deut. physik. Ges. 21, 13 (1919).
16. Born, M., Z . Physik 7,124 (1921).
17. Born, M., and Huang, K., Dynamical Theory of Crystal Lattices. Oxford
University Presa, London and New York, 1954.
18. Born, M., and Kornfeld, H., Physik. Z.24,121 (1923).
19. Born, M., and Lande, A., Ber. Preuss. Akad. Wb.,Berlin, 45, 1048 (1918).
20. Born, M., and Lande, A., Verhandl. deut. physik. Ges. 20,210 (1918).
21. Born, M.,
and Mayer, J. E., Z . Physik 75.1, (1932).
22. Brauer, P., Z . Naturforsch. lOa, 420 (1955).
W.Buchner, E. H., Kolloid-Z. 75, 1 (1936).
8.4.Buchner, E. H., Rec. trav. chim. 69, 329 (1950).
86. de Boer, F., Rec. trav. chim. 67,697 (1948).
26. de Boer, F., Van Santen, J. H., and Verwey, E. J. W., J . Chem. Phys. 16, 1091
(1948).
87. de Boer, F., Van Santen, J. H., and Verwey, E. J. W., J. Chem. Phys. 18, 1032
(1950).
28. de Boer, J. H., and Van Liempt, J. A. M., Rec. trav. chim. 46, 124 (1927).
29. de Boer, J. H., and Verwey, E. J. W., Rec. trav. chim. 55, 443 (1936).
30. Emersleben, O.,Physik.Z . 24,73 (1923).
31. Emersleben, O.,Physik. Z . 24,79 (1923).
32. Emerslebpfi, O.,Z . physik. Chem. (Leipzig) B204, 43 (1955).
33. Epstein, P.,Math. Ann. 6,615 (1903).
34. Epstein, P., Math. Ann. 63,205 (1907).
36. Evans, M. G., and Uri, N., Trans. Faraday Soc. 45,224 (1949).
36. Evjen, H. M., Phys. R e v . 39,675 (1932).
37. Ewald, P. P., Ann. Physik 64,253 (1921).
38. Fowler, R. H., Statistical Mechanics, 2nd., ed., p. 330. Cambridge, University
P r m , London and New York, 1936.
3.9. Frank, F.C., Phil. Mag. [71 41, 1287 (1950).
40. Goldschmidt, V. M.,Skrijter Norske Videnskaps-Akad. Oslo M.-N. 1, 2, (1926).
@. Goubeau, J.,Z . physik. Chem. (Leipzig) B34,432 (1936).
48. Goubeau, J., and Klemm, W., Z . physik. Chem. B36,362 (1937).
43. Gray, P., and Waddington, T. C., Proc. Roy. SOC. A235, 481 (1958).
&. G r S t h , J. S., and Orgel, L. E., Quart. Revs. (London) 11, 381 (1957).
46. Grimm;H. G.,Handbuch d . Physik 24,518 (1927).
46. Grimm, H. G., and Herrfeld, K. F., Z . Physik 19,141 (1923).
47. Gross, B., Z.Krist. 76,565 (1931).
48. Haber, F., Verhandl. deut. physik. Ges. 21,750 (1919).
220
T. C. WADDINGTON
49. Hare, P. E., and Brewer, L., Report of Contract W-7405-eng. -48, University of
California Radiation Lab., 1956.
50. Hodes, I., Phys. Rev. 73,539 (1949).
51. Hoppe, R., 2.anorg. Chem. 283,196 (1966).
58. Hojendahl, K., Kgl. Danske Videmkab. Selskab, Mat.-fys. Medd. 16, No. 2,
133 (1938).
53. Huggins, M. L., J . Chem. Phys. 5, 143 (1937).
54. Huggins, M. L., and Sakamoto, Y.,
J. Phys. SOC.Japan 12,241 (1957).
55. Hund, F., 2.Physik 34,883 (1925).
66. Hund, F., Z. Physik 9 4 , l l (1935).
57. Hush, N. S.,and Pryce, M. N. L., J . Chem. Phys. 28,244 (1958).
68. Hylleraas, E. A., Z . Physik 44,871 (1927).
69. Hylleraas, E. A., 2.Physik 63, 771 (1930).
60. Jacobs, P. W. &I., PhD. Thesis, University of London, London, England, 1951.
61. Jones, R. E., and Templeton, D. H., J. Chem. Phys. 25,1Og2 (1956).
6.8. Juea, R., 2.anorg. Chem. 231,121 (1937).
68. Juea, R., and Opp, K., 2.anorg. Chem. 266,313 (1951).
64. Kapustinskii, A. F., 2.physik. C h a . (Leipzig) B22,257 (1933).
66. Kapustinskii, A. F., J. Phys. Chem. (USSR) 5, 59 (1943).
66. Kapustinskii, A. F., Acta Physicochim. URSS 18,370 (1943).
67. Kapustinskii, A. F., Quart. Revs. (London) 10,283 (1956).
68. Kapustinskii, A. F., and Yataimirskii, K. B., Zhur. Obshchei Khim. 19, 2191
(1949).
69. Kapustinskii, A. F., and Yatsimikkii, K. B., Zhur. Obshchei Khim. 26, 941
( 1956).
70. Kazarnovskaja, L. I., J. Phys. Chem. (USSR)20, 1403 (1946).
71. Kazamovskii, I. A., Doklady Akad. Nauk SSSR 59,67 (1948).
79. Kazamovskii, J., 2.Physik 61,236 (1930).
73. Kendall, J., J. Am. Chem. Sbc. 44,717 (1922).
74. Kobayashi, S., Busseiron Kenkyu 68,ll (1953).
76. Krishnan, K. S.,and Roy, S. K., Proc. Roy. SOC.A210,481 (1951).
76. Ladd, M. F. C., and Lee, W. H., Trans. Faraday SOC.54,34 (1958).
77. Lande, A., Verhandl. deut. physik. Ges. 20.217 (1918).
78. Landshoff, R., Z. Physik 102,201 (1936).
79. Landshoff, R., Phgs. Rev. 52,246 (19371, Abstract only.
80. Lennard-Jones, J., and Dent, B. M., Proc. Roy. SOC.A113,673,690 (1927).
82. Lennard-Jones, J., and Dent, B. M., Phil. Mag. t71 3, 1204 (1927).
88. Lennard-Jones, J., and Ingham, A. E., Proc. Roy. SOC.AlW, 636 (1925).
83. London, F., 2.physik. Chem. (Leipzig) B l l , 222 (1930).
84. Lowdin, P. O., Arkiv. Mat. Aslr. Fysik 35A. No. 9 (1947).
85. Lowdin, P. O., Arkiv. Mat. Aslr. Fysik 35A, No. 30 (1947).
86. Lowdin, P. O., Thesis, Uppsala, Almquist and Wiksell, 1948.
7. Lundquist, S. O., Arkiv Fysik 8,177 (1951).
88. Madelung, E., Physik. Z.19,524 (1918).
89. Margenau, H., Phys. Rev. 38,747 (1931).
90. Margenau, H., Revs. Modem Phys. 1 1 , l (1939).
91. Mayer, J. E., J. Chem. Phys. 1,270 (1933).
98. Mayer, J. E., J . Chem. Phys. 1,327 (1933).
93. Mayer, J. E., and Helmholtz, L., 2.Physik 75, 19 (1932).
94. Mayer, J. E.,and Huggins, M. L., J. Chem. Phys. 1,643 (1933).
LATTICE ENERGIES
221
95. Mayer, J. E., and Levy, R. B., J . Chem. Phys. 1,647 (1933).
96. Mayer, J. E., and Maltbie, M. McC., 2.Physik 75, 748 (1932).
97. McClure, D.S., and Holmes, 0. G., O.N.R. Rept,, Contract N 6 ori. 211-111,June,
1955.
98. Morris, D. F. C., J . Znorg. & Nuclear Chem. 4,s (1957).
99. Morris, D.F. C., Proc. Roy. SOC.
A242,116 (1957).
100. Morris, D. F. C., and Ahrens, L. H., J . Znorg. & Nuclear Chem. 3, 263 (1956).
101. Orgel, L. E., Proc. 10th Solvay Conf. Chem., Brussels p. 289 (1956).
102. Page, F.M., Discussions Faraday Soc. 19,87 (1955).
103. Partington, J. R., An Advanced Treatise on Physical Chemistry, Vol. 111, The
Properties of Solids. Longmans, Green, 1952.
104. Pauling, L.,Proc. Roy.Soc. A114,181 (1927).
105. Pauling, L.,J . A m . Chem. SOC.49,765 (1927).
106, Pauling, L.,Z . Krist. 67,377 (1928)
107. Pauling, L.,Nature of the Chemical Bond. Cornell University P r w , Ithaca,
N.Y., 1948.
108. Pritchard, H. O., Chem. Revs. 52,529 (1953).
109. Sakamoto, Y., J . Sci. Hiroshima Univ. A16, 569, 573, 577, 581, (1953).
110. Sakamoto, Y., Bull. Chem. Soc. Japan 26,283 (1953).
111. Sakamoto, Y., J . Chem. Phys. 28,164 (1958).
112. Schmaeling, V., Z . Physik 47,723 (1928).
l13.wSeitz, F.,Modern Theory of Solids, Chapter 11, McGraw-Hill, New York,
1940.
11.4. Sherman, J., Chem. Revs. 11,93 (1932).
115. Sherman, J., Z . Krist. 81,342 (1934).
116. Templeton, D. H., J. Chem. Phys. 21,2097 (1953).
117. Templeton, D. H., J. Chem. Phys. 23,1629 (1955).
118. Templeton, D.H., J. Chem. Phys. 23,1820 (1955).
119. Topping, J., and Chapman, S., Proc. Roy. SOC.A113,658 (1927).
120. Unsold, A.,2.Physik. 43,563 (1927).
121. Unzhakow, G. M., Doklady Akad. Nauk SSSR 90.63 (1953).
122. Van Arkel, A. E., and de Boer, J. H., Physiea 1,354 (1927).
123. Van Arkel, A. E., and de Boer, J. H., Chemische Bindung a h Electrostatische
Erscheinung. Hirzel, Leipzig, 1931.
124. Vedeneev, A. V., Kazarnorskaja, L. I., and Kazarnovskii, I. A., Zhur. Fiz. Khim.
26,1808 (1952).
196. Verwey, E. J. W., and de Boer, J. H., Rect. trav. chim. 55, 431 (1936).
126. Waddington, T. C., Ph.D. Thesis, University of Cambridge, England, 1955.
127. Waddington, T.C., Trans. Faraday SOC.
54, 25 (1958).
128. Waddington, T. C., unpublished results.
129. West, C . D., J . Phys. C h a . 39,493 (1935).
130. Yamashita, J., Rept. Znst. Sci. Technol., Univ. Tokyo 4, 269 (1950).
131. Yamashita, J., J . Phys. Soc. Japan 7,284 (1952).
132. Yamashita, J., and Kojima, M., Rept. Znst. Sci. Technol., Univ. Tokyo 4,
273 (1950).
133. Yatsimirskii, K. B., Zzvest. Akad. Nauk SSSR, Otdel, Khim. Nauk p. 453 (1947).
13.4. Yatsimirskii, K.B., Zzvest. Akad. Nauk SSSR, Otdel, Khim. Nauk p. 398 (1948).
136. Zhanov, V., Erschow, A., and Galachow, G., 2.Physik 94, 241 (1935).
GRAPHITE INTERCALATION COMPOUNDS
W Rudorff
University of Tiibingen.
Tuebingsn. Germany
. . . . . . . 224
I Crystal Structure and Mode of Reaction of Graphite

I1. Graphite Compounds with Homopolar Bonding . .
A. Graphite Oxide . . . . . . . . . . .
1. Properties and Composition . . . . . .
2. Constitution and Structure . . . . .
B . Graphite-Fluorine Compounds . . . . .
1. Carbon Monofluoride . . . . . . .
a . Conditions of Formation . . . . .
b . Crystal Structure . . . . . . . .
2. Tetracarbon Monofluoride . . . . . .
a . Preparation and Properties . . . .
b . Constitution and Structure . . . .
c . Comparison of the Structures of (GF) and
. . . . . . . 226
. . . . . . . 226
.
. 226
. . . . . . . . 228
. . . . . . . . 230
. . . . . . . . 230
. . . . . . . . 230
. . . . . . . . 231
. . . . . . . . 233
. . . . . . . . 233
. . . . . . . . 234
(CF) . . . . . . 235
I11. Graphite Compounds with Polar Bonding . . . . . . . . . . 236

A . Potassium- Rubidium- and Caesiurn-Graphite . . . . . . . . 236
1. Composition and Properties . . . . . . . . . . . . . 236
2. Structure of the Alkali Metal-Graphite Compounds . . . . . 237
3 . Intercalation of Sodium and Lithium . . . . . . . . . . 241
4 . Ammonia and Amine Derivatives of Alkali Metal- and Alkaline
Earth Metal-Graphite Compounds . . . . . . . . . . . 242
5. Magnetic Properties of Alkali Metal-Graphite Compounds . . . 245
6 . Bonding in the Alkali Metal-Graphite Compounds . . . . . . 246
. . . . . . . . 248
B . Gra.phite Salts . . . . . . . . .
C . Halogen-Graphite Compounds . . . . . . . . . . . . . 251
1. Graphite Bromide . . . . . . . . . . . . . . . . 251
2. Graphite Chloride . . . . . . . . . . . . . . . . 253
3. Iodine and Graphite . . . . . . . . . . . . . . . 254
4 . Iodine Monochloride and Graphite . . . . . . . . . . . 254
D . Intercalation of Metal Chlorides in Graphite . . . . . . . . 254
1. Preparation and Properties . . . . . . . . . . . . . 254
. . . . . . . . . . . . . . . . . . 258
2. Structure
3. Bonding in the Metal Halide-Graphite Compounds . . . . . 258
E. Other Intercalation Compounds . . . . . . . . . . . . . 259
1. Intercalation of Metal Oxides and Sulfides in Graphite . . . . . 260
2. Molecular Compounds of Boron Nitride . . . . . . . . . 261
IV . Residue Compounds
v . Comparative
References
Survey
. 262
. . . . . . . . . . . . . . . . . 263
. . . . . . . . . . . . . . . . . . . .
223
264
224
B ~ R F F
1. Crystal Structure and Mode of Reaction of Graphite
Of the two modifications of elementary carbon, diamond and graphite,

the latter is not only more abundant and of greater technical importance, but it is also more versatile and interesting in its reactions. In
its reactions diamond depends on the removal of successive carbon atoms
from the three-dimensional net and the formation of compounds of low
molecular weight, as, for example, in combustion to carbbn monoxide and
dioxide. Graphite, on the other hand, is able, under certain conditions, t o
take up atoms, ions or molecules in its lattice while leaving its structure
largely unchanged. The formation and constitution of these intercalation
compounds may be explained in terms of the structure of graphite and the
special bonding relationships encountered in it.
Graphite crystallizes in a layer lattice. The carbon atoms form regular
sheets of linked hexagons which are displaced relative to one another. In
the stable hexagonal modification (3,25), which has the space group
Deh4-C6/mmc, alternate sheets are situated above one another, giving a
sequence ABAB . . . , so that the identity period in the direction of the
c axis embraces two carbon layers (see Fig. 1 ) .
-i
2
I
'
7.41A
2.45A
FIG.1. The graphite crystal lattice.
In addition to the hexagonal modification, graphite has a second form

(17,&) which, according to Lipson and Stokes ( 4 7 ) , has a rhombohedra1
structure with the space group D3&R sm.In this, the carbon layers are
displaced in such a way that each third layer is superposed, giving a
sequence ABCABC. New investigations by Hofmann and Boehm
(94) and by Laves (46) show that the hexagonal form may be
225
transformed into the rhombohedra1 by mild mechanical treatment involving shear, for example, by grinding. In dressed natural graphite the
proportion may be as high as 30%, but it is seldom encountered in artificial graphite or in single crystals. The rhombohedral reverts to the
hexagonal form on heating above 1300.The two forms have not been
observed to differ in their ability to form graphite intercalation compounds. Since, moreover, the distances of the carbon atoms and the
bonding relationships are the same in the two modifications, we may
neglect the rhombohedral modification in what follows.
Within the layer planes each carbon atom is surrounded by three
other atoms at 1.41 A. The distance between layers is 3.355 A and is
thus substantially greater. Corresponding to these different distances
there are different types of bonding. Within the layers each carbon atom
is linked by a strong homopolar bond with its three neighbors. On the
other hand the electrons representing the fourth valency of the carbon
atoms are not localized in definite bonds but, as in the case of the r-electrons of a large aromatic molecule, are shared by the whole layer.
Graphite owes its high thermal and electrical conductivity to these socalled iree electrons, which form the conduction band in the graphite.
A relatively energetic reaction is needed to break the homopolar bonds
and remove separate carbon atoms from the layers, but a suitable reactant is often able under quite mild conditions to change the state or
concentration of electrons in the conduction band and thus to become
intercalated between the carbon planes. When this occurs the carbon
planes remain intact and only their distance from one another increases.
For this reason the formation of intercalation compounds is always
linked with swelling of the graphite in a direction perpendicular to the
basal planes.
The bonding of reactants in the graphite lattice may be very varied
in type and, in some cases, is not yet fully understood. This will be
discussed in detail when the separate compounds are described. At this
stage, however, it may be said that the graphite compounds fall into two
groups, according to the character of the bonding. In compounds of the
first class there is homopolar bonding between the intercalated atoms and
the carbon atoms. To this belong graphite oxide, carbon monofluoride,
and tetracarbon monofluoride. In the second class, to which the majority
of graphite compounds belong, the bonding has a more or less pronounced
polar character. It is characteristic of these compounds that occupation
of the lattice occurs in a series of steps or stages. Thus the reactant
may enter the space between each set of basal planes, or that between
each second, third, fourth, and so on. The separate compounds or phases,
which may be clearly distinguished from one another by analysis or by
w.
226
RUDOEFF
X-ray examination are called the stages 1, 2, 3, . . . . The alkali metalgraphite compounds and the graphite salts afford an opportunity for
the detailed study of this typc of stepwise intercalation.
This review deals with the present position of our knowledge of these
graphite compounds. This requires also that brief reference be made to
compounds which have been known for a long time and repeatedly
studied. I n so far as this work is covered by earlier reviews (33, 58, 63,
577, it has been omitted here. On the other hand, research done in the last
fifteen years, in so far as it has led to the discovery of new compounds
or new results on the constitution and structure of graphite compounds,
is discussed somewhat more fully. Kumerous recent studies of physical
properties of graphite compounds, such as conductivity, magnetism, and
the Hall effect are mentioned here only when they are significant in
elucidating the type of bond in the compounds dealt with.
II. Graphite Compounds with Homopolar Bonding
A. GRAPHITE
OXIDE
1. Properties and Composition
Graphite oxide, which is also known as graphite acid, was described

by Brodie (6) as early as 1860 and is probably the most fully skudied
graphite compound (for bibliography see refs. 33, 5 3 ) . I n spite of this,
however, its constitution is not yet fully elucidated. It is best prepared
by oxidizing graphite with a mixture of potassium chlorate and concentrated sulfuric and nitric acids (81).Thorough washing with water yields
products which vary in color from yellow to dark brown. Hofmann and
his co-workers (8) recently were able t o prepare light-colored preparations, which darkened on warming or in light, by washing in the dark
with water saturated with chlorine dioxide and containing hydrochloric
:wid. The light-colored graphite became black a t once xhen introduccd
into dilute sodium hydroxide solution. No correlation between color and
composition was found, however.
Graphite oxide is thermally unstable. When rapidly heated to about
200" it deflagrates with formation of carbon monoxide, carbon dioxide,
water, and soot-like carbon. This graphite oxide soot is shown by the
electron microscope to be different from ordinary soot. It consists of thin
plates, 20-30 A thick and up to loo00 A2 in extent (75,37). When warmed
slowly, graphitr oxide is broken down to graphite-like products, which,
however, still contain oxygen. Carbon dioxide predominates in the gaseous
decomposition products. Graphite oxide possesses oxidizing properties and
227
is reduced to graphite by reducing agents, though the reaction is never

quite complete.
The composition of graphite oxide is most suitably characterized by
the C:O ratio, after taking into account the hydrogen found in combustion (33). If all four of the valencies of each carbon atom were satisfied
by bonding to oxygen this ratio would be two, but this value is never
reached. Values for C:O of 2.4 to 2.9 are found for well-oxidized preparations and values of 2.7 t o 2.8 are particularly common. Less oxidized
preparations with C:O ratios of 3.5 to 4 are, however, homogeneous
phases, and show the same characteristic properties as more highly oxidized specimens.
Among the characteristic properties of graphite is the ability to undergo swelling. Water and organic liquids with polar groups may be
taken into the lattice with a reversible change in the interplanar distance. This is 6.35 A and 9 A in preparations dried over phosphorus
pentoxide and in the air, respectively, and it increases in liquid water
to 11 A. I n weakly alkaline solutions (0.01 N ) swelling is so great that
the interplanar distance can no longer be measured with X-rays.
Hydrogen found by analysis to be present in graphite oxide cannot
be attributed solely to water taken up in the swelling process, since it
is always present in well-dried specimens. According to Thiele (85) and
to Hofmann et al. (35,367, graphite oxide always. contains hydrogen
ions capable of exchange, which are probably associated with weakly
acid hydroxyl groups and more strongly acid carboxyl groups. The
presence of carboxyl groups in graphite oxide is to be expected from its
mode of preparation, for mellitic acid is produced in the oxidation of
carbon with potassium chlorate in the presence of sulfuric and nitric
acids.
I n connection with the constitution of graphite oxide, the question
of the number and nature of the acid groups and the point of their attachment is especially interesting. Comprehensive studies by Hofmann
and his co-workers (36, 35, 8, 74) on the extent of base exchange and
on the methylation of graphite oxide have made a substantial contribution towards solving this problem. The results of these investigations
may be summarized as follows.
I n well-oxidized specimens the fraction of -OH groups, as calculated from measurements of sodium ion exchange from sodium ethylate
solution, is found to be about 1100 meq per 100 gm of graphite oxide
(dried over phosphorus pentoxide). Roughly the same value is obtained
by eluting from a well-dried specimen all the water left from swelling
by means of dioxane, and attributing the remaining hydrogen to -OH
or -COOH groups.
228
w.
RUDORFF
Only a part of the -OH groups-about 700 meq per 100 gm-can
be methylated with diazomethane or acetylated with acetic anhydride.
That it is the -OH groups attached to the carbon planes which react
in this way follows from the fact that, after methylation or acetylation,
the interplanar distance in the dry graphite oxide is increased by 3-4 A.
Methylated graphite oxide preparations retain the property of base
exchange with sodium methylate solution to an extent corresponding
with the number of -OH groups not involved in the methylation reaction.
The number of carboxyl groups has the comparatively small value of
30-100 meq per 100 gm.These groups may be determined by methylat,ion Kith methanol and hydrochloric acid. In this case the weakly acid
-OH groups of the layer planes do not react. Carboxyl groups can
probably be attached only a t the edges of the crystal or a t the edges of
cavities in the layer planes, for preparations methylated with methanol
show the same interplanar distance as unmethylated specimens.
2. Constitution and Structure
Neither the infrared spectrum nor X-ray analysis enables a detailed

picture of the structure of graphite oxide t o be obtained. According to
Hadzi and Novak (24) the -OH stretching and bending frequencies
and the C=O bond frequency occur in the infrared spectrum, but it is
not possible to determine whether the C=O group is present only in
carboxyl groups, or in keto groups as well. A band a t 9 . 3 ~
is attributed
by the authors to the C-OH
bond vibration, but the assignment of
further weak bands is uncertain. The position and bonding of the atoms
cannot be determined with certainty by X-ray analysis since, apart
from the (001) reflections, from which the distance between the basal
planes may be calculated, only ( h k ) cross lattice reflections occur.
In a structure proposed by Ruess ( 7 4 ) , see Fig. 2, hydroxyl groups
and oxygen atoms are bonded to the carbon layers according to a statistical distribution. The carbon layers are believed to have lost their
aromatic character and are no longer planar, as they are in graphite,
but puckered (see the structure of carbon monofluoride under Section
IIBlb). Puckering of the rings is supported by the expansion of the carbon hexagonal network which, if tetrahedral bonding of the carbon
atoms is assumed, leads to a G C distance of 1.52 A. Ruess assumes an
ether-like bonding of oxygen atoms in the 1:3 position above and below
the layer planes. The mode of bonding cannot be established experimentally. All attempts to cleave the oxygen bridges and increase the number
of 4 H groups have failed. If a reaction occurs as it does, for example,
229
GRAPHITE INTERCALATION COMPOUNDb
C-Atoms
O = 0-Atoms
6,OH-Groups
FIQ.2. Layer of graphite oxide with tertiary OH-groups (Rueas, 74).
with hydriodic or hydrobromic acid, immediate reduction to graphitelike products ensues.

Hofmann (8) supposes that the layers in graphite oxide are under
strain because of the oxygen bridges and that the aromatic state of the
carbon planes in graphite itself is relatively more stable. This is supported by the fact that the heat of formation of graphite oxide from
graphite, oxygen, and water is almost zero (58). On breaking the oxygen
bridges, therefore, reduction is favored with the formation of C=C bonds
and a gradual transition to the aromatic conditions of graphite itself.
Riley (2, 7) suggests another mode of bonding of oxygen. He supposes
that graphite oxide contains peroxide groups, +O-,
which bind the
graphite layers together. Contrary t o this idea is the fact that free oxygen is never observed in the thermal decomposition of graphite oxide.
Hofmann and Ruess (SS, 74) also suggested that oxygen bridges are
broken when graphite oxide swells and that the number of -OH groups
increases. This, however, is not in keeping with what is found experimentally.
The acidic character of - O H groups in graphite oxide may be explained if the --OH groups are attached to a carbon atom with a C=C
double bond, that is to say, if they are enolic in character. The presence
of C=C bonds in the layers would then be characteristic of graphite
oxide. This would explain why, in the oxidation, a C:O ratio of 2 is
never reached.
Doubly bound carbon atoms with an enolic group can only be bound
w.
230
B~~OBPF
to one further carbon atom. It follows that in those positions where the
enolic group is attached the hexagonal network must be broken. Figure
3a shows the section of a graphite oxide basal plane according t o a new
proposal by Hofmann (8). This structure leads to an idealized formula
C ~ O Z ( O Hwith
) ~ a C : O ratio of 2.67 and 1240 meq of OH per 100 gm,
which is close to the observed value.
(0
(b)
FIG.3. Segment of a layer of graphite oxide (Hofmann, S). (a) Enol groups. (b)
lieto groups.
In addition to this enolic form, a keto form is also structurally possible (Fig. 3b) which may predominate in the light-colored graphite
oxide. The change of graphite oxide from the light to the dark color in
dilute alkali could then be associated with the transition to the enolic
form.
It is possible that tertiary hydroxyl groups are present in graphite
oxide in addition to enolic groups. This is shown in Fig. 2. This would
explain why, with aqueous alkali metal hydroxides and with diazomethane, only a fraction of the -OH groups which are present undergo
reaction.
A totally different structure for graphite oxide was proposed by R. E.
Franklin (18).He concluded from X-ray studies that there was a structural continuity between the oxygen atoms of the graphite oxide and
those of water intercalated in the lattice. Both form together a negatively charged polylayer (O,H,) with respect to which the carbon
layers are cationic. The chemical prollerties of graphite oxide are not,
however, readily reconciled with an ionic structure and negative hydroxyl ions.
B. GRAPHITE-FLUORINE)
COMPOUXDS
1. Carbon Monofluoride
a. Conditions of Formation. Whereas soot and active carbon burn
in fiuorine when exposed t o it a t room temperature, coarse graphite is
231
stable to fluorine up to about 400'. Between 500' and 700' reaction is

for the most part explosive and above 700" the graphite burns to a mixture of fluorocarbon such as CFr and C2F6. Only in the narrow temperature range 420-460' did Ruff and Bretschneider (77) obtain a gray
product of the composition CFo.92which they called carbon monofluoride. In a further investigation of carbon monofluoride W. and G. Rudorff
(69)obtained homogeneous preparations the composition of which was
between CFo.ssand CFo.ge5.While the fluorine-poor preparations are still
black, the color lightens with increasing fluorine content. The compounds
with the highest fluorine content are pure white and transparent in their
layers. At the same time the electrical resistance increases very much
and the white preparations are practically nonconducting.
Formation of carbon monofluoride is catalyzed by gaseous hydrogen
fluoride (63).There are characteristic differences according to the crystalline condition of the carbon. With coarse crystalline graphite in the
form of flakes, formation of carbon monofluoride is determined almost
entirely by the partial pressure of hydrogen fluoride and shows little
dependence on temperature. At an HF partial pressure of 200 mm, reaction occurs at 250'. When, however, the carbon forms a very dense aggregate of fine crystals, which is the case with retort graphite, reactivity
is determined less by hydrogen fluoride pressure and much more by
temperature. Retort graphite is very resistant at 250') even a t high
hydrogen fluoride concentrations, but it becomes more reactive than flake
graphite above 320'.
Catalysis by hydrogen fluoride explains the formation of a carbon
monofluoride layer on carbon and graphite anodes in the electrolysis of
potassium fluoride-hydrogen fluoride melts, which was observed by
Ruff (76).This unwanted reaction can lead to destruction of the anodes
in the space above the electrolyte and, according to RudoriT and Hofmann (71))is also responsible for the so-called aging of carbon and
graphite anodes, which manifests itself by an increase in the anode
voltage on prolonged electrolysis.
Carbon monofluoride is strikingly unreactive, being unwetted by
water and unattacked even by concentrated acids and alkalis. No reaction occurs up to 400" on heating in hydrogen. When rapidly heated
the compound deflagrates with production of a flame, reaction being
more vigorous the higher the fluorine content. It gives a cloud of very
finely divided soot together with carbon tetrafluoride and higher fluorocarbons.
b. Crystal Structure. The structure proposed by Ruff is based on
the assumption that fluorine is present in the graphite lattice as ions.
The ratio of the radii of fluoride ion and carbon does not, however, allow
232
w.
RUDORFF
the fluoride ions to occupy a layer above and below each graphite plane,
so that Ruff had to arrange the fluoride ions in six layers between two
carbon planes. This structure is, however, very unlikely on electrostatic
grounds. Quite apart from this, the great inertness of the monofluoride
is not indicative of ionic bonding of the fluorine, for all graphite compounds which are ionic in type, and especially the graphite salts (compare Section IIIB), are very reactive and are decomposed by traces of
water, with reversion to graphite. By contrast, carbon monofluoride is
hydrophobic and cannot be changed back to graphite.
According to W. and G. Rudorff (68)carbon monofluoride is a homo-
'ol
b
*C OF
FIG.4. Crystal latt,ice of carbon monofluoride (W. and G. Rudorff, 62).
polar compound in which fluorine is bonded in the same way as in

organic fluorides. Figure 4 shows the new structure proposed. The fluorine atoms form a layer above and below the carbon layer but, because
of the sp3 bonding, the carbon atoms no longer lie in a plane, as in
graphite. Instead, the layers are puckered and the C-C distance (1.54 A)
is that for a single bond. Each composite layer is comparable with a giant
hydroaromatic molecule.
Total absence of (hkl) reflections in the X-ray diagram shows that
the separate CF-layers are no longer regularly arranged with respect
to one another, and are only roughly a t the same distance apart. This
distance depends on the fluorine content: with the highest fluorine con-
GRAPHITE INTEXWALATION COMPOUNDS
233
tent it is 6.6 A and it increases as the fluorine content decreases, to 8.8 A

at CFo.6e.A clear explanation for this unusual dependence of the distance
on the fluorine content has not been given.
Unfortunately the X-ray powder diagram of carbon monofluoride
shows only a few reflections, so that the structure suggested above cannot be based on an exact structural analysis. The increase in the C-C
distance relative to that in graphite follows unambiguously from the
fact that (hkO) reflections, which occur as cross lattice reflections,
are displaced to smaller diffraction angles. These reflections in the case
of a puckered layer do not give the C - C distance directly but only the
projection on the basal plane. The position of the reflections gives this
projection as 1.467 A, while calculation based on a tetrahedral angle and
a C-C distance of 1.54 A leads to the value 1.453 A. If the size of the
C-C-C
bond angle is increased somewhat the rather higher experimental value may be correctly reproduced.
A sure proof of the presence of a C-F bond has recently been furnished by recording the infrared spectrum of the compound (59). In the
region 2-25p there is a single strong broad band at 1215 cm-l. The C-F
frequency in perfluorinated organic compounds such as c6F12lies at
1200-1320 cm-l (86),so that the observed band may be assigned with
certainty to a C-F vibration-probably an unsymmetrical stretching
vibration. The above model readily explains the low reactivity, the
absence of electrical conductivity, and the formation of carbon fluorides
of low molecular weight on thermal decomposition.
2. Tetracarbon Monofluoride
a. Preparation and Properties. In addition to carbon monofluoride,

graphite and fluorine are able to form yet another compound which is
like carbon monofluoride in many properties, but differs from it clearly
in composition, color, conductivity, and structure. This is tetracarbon
mpnofluoride, which was prepared by W. and G. Rudorff (64).
The compound is readily formed by passing a fluorine-hydrogen fluoride mixture over graphite at room temperature. The rate of reaction
depends on the composition of the gas mixture and drops rapidly with
decreasing hydrogen fluoride partial pressure. If the graphite is held at
80-100", the reaction does not occur, from which it follows that the
temperature range in which the compound is formed is relatively closely
limited. Using powdered graphite, the compound has a velvet black appearance, sometimes with a bluish tinge. The composition may vary in
the range CaF-Cs.6Fand clearly depends on the conditions of preparation.
Like carbon monofluoride, tetracarbon monofluoride is also very inert,
though it is appreciably less stable thermally. It is completely stable in
234
w.
RUD~RFF
air, and is unattacked by dilute acids and alkalis, even in the warm.
Strong reducing agents such as ferrous salts or hydriodic acid produce
no change, and even chromic-sulfuric acid has no action in the cold.
From the last observation it follows that graphite, in combining with
fluorine, has lost its ability to form graphite bisulfate (compare Section
IIIB) .
When warmed above 100" tetracarbon monofluoride preparations
begin t o decompose and when rapidly heated they deflagrate-sometimes
with flame-and leave a very finely divided soot. Decomposition also
occurs when the compound is gradually heated to 200-300" in a stream
of fluorine or of a fluorine-hydrogen fluoride mixture. Thus it is not possible to convert the compound to carbon monofluoride by further fluorination.
Unlike carbon monofluoride, the compound retains an electrical conductivity which, however, is about two powers of ten lower than that
of the original graphite. W. and G . Rudorff found a specific resistance
of 2-4 ohm * cm a t a pressure of 750 kg/cm2. Measurements by McDonnell
et al. (48) also showed a similar strong decrease in conductivity on going
from graphite to C4F.
The compound is diamagnetic, .the molecular susceptibility referred
to CF0.15
being -9.03 x
(48). Although it may be shown that the
compound contains no hydrogen fluoride, the latter plays a decisive role
in the synthesis for, as already mentioned, the rate of reaction is strongly
dependent on the hydrogen fluoride concentration. I n this connection it
is of interest that graphite when oxidized with fluorine forms a salt-like
graphite bifluoride with the approximate composition CZ4+HFz- * 4HF
(58).In this compound the C:F ratio is also 4:1, but it is formed only
in the presence of liquid anhydrous hydrogen fluoride and has quite other
properties and a different constitution from tetracarbon monofluoride. It
is not impossible, however, that the bifluoride occurs as a short-lived
primary product, since a skin of liquid hydrogen fluoride may form on
the graphite crystals a t room temperature from the gas stream. Further
oxidation by fluorine would then lead from this compound to tetracarbon
monofluoride. This assumption that the bifluoride occurs as an initial
product would explain the observation that a t 80' graphite is unchanged
in a fluorine-hydrogen fluoride stream since, a t this temperature, no
liquid film can form.
6. Constitution and Structure. The properties mentioned above indicate clearly that in this compound, as in carbon monofluoride, a covalent
C-F bond is present. Since, however, only every fourth carbon atom
gives an electron for bonding fluorine, it is understandable that the
compound still conducts electricity. The structure is shown in Fig. 5.
The network of carbon hexagons remains, with a

C-C distance, as in graphite. There is, however,
graphite structure in that the carbon planes are
one above the other. Their distance apart is 5.34 A
235
practically unaltered
a difference from the
in identical positions
for preparations with
5.5A
0 F- A toms
=
FIG.5. Crystal lattice of tetracarbon monofluoride (W. and G . Riidorff, 6 3 ) .
the composition C4F and this increases t o 5.50 A for the composition
C3.6F.
The fluorine atoms are arranged in two layers, one above and one
below each carbon plane. The C-F distance cannot be determined by
X-ray diffraction, but is assumed to be 1.4 A, as in organic fluorine compounds. For C4F, if the fluorine atoms are regularly distributed, they
form in each layer a triangular network with an F-F distance of 4.9 A.
c. Comparison of the Structures of C P and CF. The structure
of tetracarbon monofluoride differs from that of carbon monofluoride in
that the distance between layers is smaller and, especially, in that the
carbon planes are not puckered. It seems unlikely that those carbon
atoms to which fluorine atoms are bonded project out of the plane and
that the carbon planes are buckled, for this would distort the hexagonal
network and would be recognizable in the X-ray powder diagram, both
in the position of and in a broadening of the (hkO) reflections. These
reflections are, however, relatively sharp, as in graphite itself.
Structural studies made up to the present lead to the generalization
that in a graphite compound all C-C distances in the carbon layers,
apart from lattice imperfections, are the same and that the layers are
either plane or puckered, even when some of the carbon atoms have B
different type of bonding. In carbon monofluoride and tetracarbon monofluoride the two possible structures for the carbon layers are realized.
236
w.
RUEQRFF
So long as the proportion of carbon atoms involved in bonds with fluorine

remains small-in CrF it is one quarter-the aromatic character of
graphite with plane carbon sheets is retained. Bonding of fluorine, however, certainly produces tension within the layers and this may be the
reason for the low thermal stability. The carbon planes become puckered
when, as in carbon monofluoride, the majority of the carbon atoms become involved in Sps bonds. The fact that no further compound occurs
in the region between CF0.28 (= C3.&) and CFo.sa and that there is no
continuous transition from the one compound to the other is in keeping
with this view. For the monofluoride the most stable preparations are
those which approximate most closely to CFl.o, as the puckered carbon
planes are then free of strain. The further the composition deviates from
the ideal formula the greater is the strain, as is shown by the observed
increase in thermal instability with decreasing fluorine content.
111. Graphite Compounds with Polar Bonding
-4.POTASSIUMRUBILXUMAND CESIUM-GRAPHITE
1. Composition and Properties
Graphite reacts with striking readiness with the molten alkali metals
potassium, rubidium, and cesium, or their vapors, to form alkali metalgraphite compounds. Fredenhagen et al. (19, 20) showed that for all
three alkali metals, depending on the temperature and alkali metal vapor
pressure, products were formed which differed from one another in color.
The first compound has a color between bronze and copper-red and has
the approximate composition CBMe(Me = K, Rb, Cs) . Such a compound
is prepared readily by carefully distilling off in a vacuum the excess of
alkali metal used in the preparation. On further heating, CBMe loses
alkali metal and passes over into a steel-blue compound. Finally, all the
alkali metal may be driven off by heating strongly.
The composition of the blue product was not established with certainty in Fredenhagens work and Schleede and Wellmann (79) first
derived the formula Cl&e from crystal-structure measurements. However, further analytical and X-ray investigations by Riidorff and Schulze
(66,67) showed conclusively that the blue compounds contained less
alkali metal and had the formula C24Me.This conclusion is supported by
HQroldswork (32) which showed that in the isobaric breakdown curve
of CSK the first clear break occurs a t C24K. This product gave the same
X-ray powder pattern as the blue compound studied by Schleede (31).
Fredenhagen found calorimetrically a value of 12 kcal/mole for the
heat of formation of CSK formed by introducing graphite into an excess
237
of molten potassium. A careful redetermination by Quartermann and

Primak (51) led to the lower value of 7.8 kcal/mole. HBrold (31) calculated from vapor-pressure measurements that AH for the reaction
CzaK 2Kli, = 3C8K was -6.8 kcal. The corresponding AH values for
the Rb and Cs compounds are -11.6 kcal and -20.2 kcal, respectively.
With Fredenhagens value for the heat of formation of CBK we then
obtain
24C
+ Kliq + C2,K;
AH = -29.2 kcal.
If Quartermanns value is taken, AH = -16.6 kcal.

The alkali metal-graphite compounds are extremely reactive. They
ignite in air and may react explosively with water. In the controlled
reaction with water or alcohol only alkali hydroxide and hydrogen result: there is no acetylene or any other hydrocarbon. Fredenhagen concluded from this that the compounds could not be carbides. Mercury
dissolves the alkali metal out of the lattice. When treated with liquid
ammonia, C8Me gives up only a third of the alkali metal and takes in
its place two molecules of ammonia (see Section IIIA4).
2. Structure of the Alkali Metal-Graphite Compounds
The crystal structure of these compounds was first determined by

Schleede and Wellmann (79). In the compound C8Me a layer of alkali
metal atoms is present between each pair of carbon planes, whereas in
the blue alkali-poor compound this occurs between every other pair. Introduction of the alkali metal increases the interplanar distance to 5.41 A
for potassium, 5.61 A for rubidium, and 5.95 A for cesium. According to
the definition given in the introduction these two compounds are referred
to as the first and second stage.
In the first stage the alkali iiietal atoms form a triangular net, the
edge of which has the value a = 4.91 A and is exactly twice as long as
the hexagonal a axis of the hexagonal carbon net (Fig. 6a). When the
layers are fully occupied the first stage has the composition C&e. The
arrangement of atoms in this compound, as found by new structural
determinations (67), is shown in Fig. 7. The new model differs from that
of Schleede and Wellmann only in orientation of the carbon and alkali
metal planes relative to one another. According to these authors, the
carbon planes are arranged as they are in the hexagonal modification
of graphite (ie., in the sequence ABAB . . .) and the metal atoms lie
between one carbon atom and the middle of a hexagonal ring of carbon
atoms situated above or below it. In the new structure all the carbon
planes are in identical positions (i.e., they have a sequence AAAA . . .)
and the alkali metal atoms always occupy positions above or below the
238
w.
R ~ R F F
FIG.6. (a) The triangular packing of the alkali atoms in CsMe. (b) The hexagon
net of the alkali atoms in CsMe, C%Me.Both (a) and (b) are shown in relation to
the hexagon net of carbon atoms
middle of a hexagonal ring of carbon atoms. This arrangement seems

intrinsically more likely for, in it, every metal atom has twelve carbon
atoms at the same distance from it (3.07 A for C8K or 3.24 A for C8Rb),
whereas in the first model each metal atom has as neighbors one carbon
239
l+a=4.90KX-
0 K-Atoms
FIQ.7. Crystal lattice of potassium graphite CiK (Rudorff, 67).
w.
240
RUDORFF
atom belonging to the layer above, and six belonging to the layer below
it, at different distances,
Since a metal layer in stage 2 is separated from the next by two
carbon planes, the Me:C ratio should be half as big as in stage 1 if the
alkali metal layers were populated equally in the two stages. On this
assumption Schleede arrived at the formula CleMe. Since, however, the
composition is actually CZ4Me, the metal atoms must be less densely
packed in the second stage. Such a result is obtained if every third Me
atom is removed from the triangular network as it exists in stage 1. The
resulting hexagonal network fits the carbon network with the same
Ale-Me distance as does the triangular network (Fig. 6b). This arrangement gives the exact composition CsaMe for stage 2.
The arrangement of metal atoms in a hexagonal network cannot be
rigidly proved by X-ray methods, since no further superstructure reflections which require a greater identity period than the carbon network
possesses can be recognized for stage 2. Presumably the metal atoms in
the wide-meshed net are much more mobile in the direction of the layer
plane than at right angles to it, so that the (h0.l) and (hlcl) reflections
are more disturbed than the (001).
In addition to these two stages there are other alkali-poor stages to
which the author drew attention in 1944 (57). A systematic study of the
systems graphite-potassium and graphite-rubidium (67) showed that in
all, five stages occur in the range between CsMe and CsoMe. Their compositions are shown in Table I. The formulae of stages 2 to 5 are whole
TABLE I
~~
~~
Lattice constants, c
for potagsium graphite
(A, *
Composition
1st &age GMe
2nd stage G&fe
3rd stage CSMe
4th stage CeMe
5th stage CGMe
5.41
8.76
12.12
15.48
= 5.41
= 5.41
= 5.41
= 5.41
18.83 = 5.41
+ 1 X 3.356
+ 2 X 3.355
+ 3 X 3.355
4-4
X 3.35s
* The published values in kX have been converted

into A.
number multiples of 12:l from which it follows that the same hexagonal
network of metal atoms is present in all these stages as has already been
described above for stage 2.
The formulae correspond to idealized compositions, though each stage
241
has a definite homogeneity range within which the characteristic sequence of carbon and metal layers remains undisturbed. When there are
large deviations from the ideal formula two stages are obtained together
as, for example, when, in attempting to make preparations with the composition C16Me, stages 1 and 2 result. The transition of the separate
phases into one another thus occurs discontinuously.
Introduction of the metal layer leads, for all the stages, to the same
increase in the distance between the adjacent carbon planes, as may be
seen from the identity periods for potassium-graphite (see Table I). In
addition the carbon planes next to the metal layers always have identical
positions. Thus entry of the alkali metal into the lattice is linked not
only with an expansion lattice but also with a lateral displacement of
the carbon planes. Metal atoms are then able to arrange themselves so
that they lie over and under the carbon hexagons and are surrounded
symmetrically by twelve carbon atoms. Figure 8 shows the sequence of
I ----
C-Layw
---- K- Layer
FIQ.8. Layer-plane sequence in stages 1-5 of alkali graphite (section perpendicular to the layer planes).
carbon and metal layers and the orientation of the carbon planes to one
another for stages 1 to 5.
9. Intercalation of Sodium and Lithium
While potassium, rubidium, and cesium enter the graphite lattice very
readily, sodium and lithium are much less reactive towards graphite and
do not appear to form the corresponding compounds. Fredenhagen (19,
a0) found that sodium reacts with soot but not with graphite. Tammann
242
W. BUDORFF
and Sworykin (82) observed that little rods of retort carbon are attacked by sodium vapor between 500" and 900' and disintegrate, giving
products which ignite in air. They concluded that sodium, like potassium,
is able to form a solid solution in carbon at higher temperatures. I n spite
of this, attempts to prepare definite compounds were a t first unsuccessful
( 3 2 ) . HCrold (31) was then able t o intercalate sodium and cesium together by allowing a sodium-cesium alloy to react with graphite, but the
Na:Cs ratio in the product was only 0.12.
In recently published work Asher and Wilson ( 1 ) report that when
graphite is heated with about 3% of sodium to 400" a deep violet product
results; the X-ray powder diagram differs characteristically from that
of graphite. Quantitative X-ray analysis gave values for the composition
between CsoNa and Cs8Na. The authors believed that a stage 8 compound had been formed, but, in spite of wide variations in the experimental conditions, they were unable to obtain any further compounds.
It follows that there are certain differences in behavior towards graphite
for sodium on the one hand and the heavy alkali metals potassium, rubidium, and cesium on the other.
The reaction between lithium and graphite has been studied by HCrold
( 3 1 ) .At 500" and above, after long periods of reaction only the carbide
Li2C2 results, but a t intermediate stages products occur which differ
distinctly in their properties from the carbide. With an over-all composition CILi a product was obtained which was harder than graphite and
gave a homogeneous golden-yellow powder when broken up. On increasing the Li:C ratio first black, then grey and finally white substances
result. The graphite compounds containing lithium are, according to
HCrold, stable in dry air. Hydrogen and acetylene are produced when
they are decomposed with water, the proportion of acetylene increasing
with the Li:C ratio in the sample and reaching 100% for Li:C = 0.5. No
X-ray study has yet been made of these striking compounds which differ
in properties and composition from the other graphite compounds.
4. Ammonia and Amine Derivatives
of Alkali Metal- and
Alkaline Earth Metal-Gruphite Compounds

These compounds resemble alkali metal-graphite compounds both in
their structure and in their properties. The method used in their preparation was the result of a chance observation in the author's laboratory
(80). During an investigation of electrolytic reactions in liquid ammonia
a solution of sodium amide in liquid ammonia was electrolyzed with
graphite electrodes, and it was found that the cathode swelled and disintegrated. Since the electrodes were not attacked in the solution when
there was no electrolysis it was clear that the change cbuld only have
243
been brought about by a reaction of sodium separated a t the cathode in

the presence of ammonia.
Rudorff et al. (6'6, 72, 54) obtained the ammoniates in a pure state
by the action of the blue solution of the alkali or alkaline earth metals
on graphite. If the metal is in excess a deep blue graphite compound with
the structure of stage 1 results. The strontium and barium compounds
are violet a t low temperatures. After washing with liquid ammonia and
removal of the absorbed ammonia in a vacuum a t 0", the composition
corresponds approximately to the formula C12Me (NH3)2. The potassium,
rubidium, and cesium compounds may also be obtained by treating CsMe
with liquid ammonia.
3CsK
+ 4NH3 + 2CizK(NHs)z iK.
If the graphite is in excess, the ammonia solution is completely decolorieed and one obtains black or blue-black preparations which are less rich
in metal. At a ratio of about 1Me:28C only stage 2 occurs.
TABLE I1
AMMONIATES
OF ALKALI
AND ALKALINE-EARTH
COMPOUNDS
Second stage
First stage
Composition
1, (A)
6.62
6.63
6.56
6.5s
6.5s
6.6
6.36
6.36
Composition
1, (A)
9.93
9.9,
9.94
9.9s
9.87
9.79
I n Table I1 analytical data are given for preparations which appear,

on X-ray examination, to be single substances. Deviations from the
ideal formulae may be caused by some phases occurring over a certain
range of composition. It is, however, clear from the data that on passing
from stage 1 to stage 2 the ratio Me:C decreases by more than half, as
in the case of the ammonia-free compounds. In stage 2 of the alkaline
earth metal compounds it seems that up to four molecules of ammonia
may be bound per metaI atom.
All these compounds are exceedingly sensitive to air and moisture,
but they are no longer spontaneously inflammable as the alkali metalgraphite aompounds are. They are decomposed by water with vigorous
244
w.
RUDORFF
evolution of hydrogen and formation of metal hydroxide solutions. In

the absence of air the cesium, rubidium, and potassium compounds may
be kept unchanged for years in a sealed tube a t room temperature, but
they decompose at elevated temperatures with evolution of hydrogen and
the formation of metal amide. The lithium and alkaline earth metal
compounds are much less stable and decompose even at ordinary temperatures in the course of a few days.
By the simultaneous intercalation of the metal and ammonia the
graphite lattice is expanded more than for pure alkali metal-graphite
compounds. The distance between the carbon planes is 6.6 A for the
ammonia derivatives of all the alkali metals and of calcium: for those
of strontium and barium a value of 6.4 A has been found.
The approximately constant expansion of the graphite lattice, which
is practically independent of the size of the intercalated metal atom,
appears at first sight unusual for, in the ammonia-free compounds, the
distance between the layer planes increases, as expected, with increasing
size of the alkali metal atom from potassium to cesium. The constancy
of the expansion for the ammoniates is perhaps attributable to the effect
of the positions of the ammonia molecules in the lattice in determining
the increase in the interplanar distance. If this were so, metal atoms or
ions could perhaps find sufficient room in holes in the ammonia lattice.
In support of this view it may be added that the expansion becomes
greater if, in place of ammonia, a layer of amine such as methylamine or
ethylamine is intercalated.
In the deep blue first stage of the lithium-methylamine compound,
CI2Li(CH3NH2)2,which may be prepared in the same way as the ammonia compound, the distance between the carbon planes is 6.9 A. The
corresponding potassium compound, which contains only one molecule
of methylamine per atom of potassium and which is prepared by prolonged treatment of C8K with methylamine, has exactly the same interplanar distance. Graphite reacts with lithium and ethylene diamine to
form a stage 2 compound CzeLi(en), in which the interplanar distance
has increased even to 8.5 A ( 5 4 ) .
Tetramethyl- and tetraethyl-ammonium or -phosphonium can take
the place of an alkali metal ( 5 4 ) . These compounds are formed a t
graphite anodes on electrolyzing solutions of the corresponding salts in
liquid ammonia. The tetraethyl ammonium compound is best prepared
by allowing the blue solution obtained by electrolysis of the azide in
liquid ammonia to react with finely powdered graphite. In this case
intercalation leads only to stage 2, as mhy be seen from the ratio
1 (C2H5)+N:25C determined by analysis. Khereas the blue solution starts
245
to decompose at -60, the graphite compound may be kept unchanged

under liquid ammonia at -40" for several hours, Apparently ammonium
is also able to enter the graphite lattice. Dzurus and Hennig (14) have
electrolyzed solutions of ammonium nitrate in liquid ammonia with
carbon cathodes and have found that the resistance of the latter decreases. Similar results were obtained with solutions of methylamine
hydrochloride in methylamine.
6. Magnetic Properties of Alkali Metal-Graphite Compounds1
Unpublished measurements by Klemm and Abitz (44) have shown

that the compound CeK possesses a weak temperature-independent paramagnetism (xmole= +9.7 x
which corresponds to the order of
magnitude of the paramagnetism of the electron gas in a metal. Ubbelohde (48) found a somewhat higher value (xmole= +16.2 X 10-9
for the compound C7.78Kand this was also independent of temperature.
Rudorff and Schulze (67) have made measurements on stages 1 to 5 of
the potassium and rubidium compounds. Surprisingly, the difference in
magnetic properties for the various stages was quite small. Down to
stage 5 all compounds are weakly paramagnetic. The values for stages
1 to 5 of potassium graphite at room temperature are: +10.0, +8.7,
+8.5, +5.7 and +4.0 (all x lo-*). Values for rubidium graphite were
between +5.0 x
(stage 1) and +3.7 x lo-" (stage 5). It is seen that
the paramagnetism of the alkali-poor stages is not made up additively
of the paramagnetism of stage 1 and the diamagnetism of pure graphite.
It follows that introduction of alkali metal atoms influences the special
state of the electrons in the graphite lattice which is responsible for the
anomalous diamagnetism of pure graphite, and that this effect extends
throughout the lattice even when the metal atoms occupy more widely
spaced planes, as in stage 5.
Measurements made in the author's laboratory by W. Sprissler on
stage 1 of the ammoniates of potassium and rubidium graphite and on
stage 1 of the lithium-methylamine compound gave susceptibilities of the
same order of magnitude as th0s.e of pure alkali metal-graphite compounds. Values of the susceptibilities, corrected for the diamagnetism of
2, +5.68; C13.2Rb(NH3)1.0,
ammonia or methylamine, are: CI~.OK(NH~)
+KO; C12.0Li(C&NH2)l.o, +8.35 (all X
The approximately equal
values for paramagnetism of the alkali metal-graphite compounds and
their ammoniates leads to the conclusion that their constitutions are
similar.
'All susceptibility values given in this section refer to 1 gm-atom of carbon.
w. RUDORFF
246
6. Bonding in the Alkali Metal-Graphite Compounds
These may be considered as intermetallic compounds. However, they

possess a certain polar character, as Zintl (90) has shown tobe always
the case for intermetallic compounds between a noble and a very base
metal. The alkali metals are base and show a transition in favor of cation
formation whereas for graphite, which ranks as a noble metal, the transition is to an anion: in other words, the bond has a limiting structure
Me+ Graphite-.
The stereochemical interpretation2 of the alkali-graphite compounds
may also be explained in this way (57). If the volume of graphite is
deducted from the molecular value of the compound C8Me, the alkali
metal increments shown in Table 111, column 2, are obtained. CoinpariTABLE 111
ALKALIhfETALS IN THE
ALKALI-GRAPHITE
COMPOCNCS
H ITH ATOMICAND IONICVOLUMES
~OMIARISON OF S P . i < E IIVCREMEKTS OF THE
Compound
CSK
CsR b
CsCs
Increment of
alkali metal in
the graphite
rompour.ds
A4tomicvolume
Me in
metals
Ionic volume
Me+
in salts
25.5
30.3
33.6
43.4
53.1
66.0
16
20
26
sons with the corresponding atom and ion volumes show that the increments approximate to the values for the Me+ ions, as they are found in
salts. Formation of the compound CeMe is therefore associated with an
appreciable contraction which is greatest for cesium, probably because
in that case the polar character is most pronounced.
Ubbelohde and his co-worker8 have made an experimental contribuThe alkali increments were first calculated by K. Meisel (see ref. 4 ) , but the
incorrect formula CmMe was used for stage 2. The values can, however, as is done
here, be referred to only the first stage, CaMe, since the alkali atoms in stage 2 are
less densely packed. Meisel comes to almost the same values as are given here
because he assumed the =me density of occupation of the alkali layers.
* Ubbelohde puts forward the view that in C X a pseudometallic bond is present
between the potassium atoms and certain carbon atoms. He uses as a basis the
structure proposed by Schleede and Wellmann. On the new structure localized K-C
bonds are very unlikely. The distance of the carbon planes is erroneously given in
this work as 7.76 A instead of 5.41 A. Arguments on the nature of the bonding, based
on the higher value, are therefore weakened.
247
tion to the question of the bonding in these compounds by measuring the

electrical conductivity of potassium graphite of various compositions
(48). The values in Table IV, which were obtained with polycrystalline
material a t a pressure of 20 kg/cm2, are taken from Ubbelohde's work.
The intake of potassium thus brings about a considerable increase in the
TABLE IV
ELECTRICAL
CONDUCTIVITY
OF POTASSIUM
GRAPHITE
Conductance (ohm-' cm-')
Composition
90K
298K
Kish
Graphite
KCs.*
KCi2.a
KCm.8
110
128
35.2
26.5
1302
1073
_.
980
870
565
conductivity. The temperature coefficient of conductivity is also negative, as it is for a true metal. This signifies that the alkali metals give
up electrons to the conduction band of graphite, thus increasing the
number of charge carriers.
The polar character of the alkali-graphite compounds also explains
readily why the regular sequence of Me and C layers is maintained, even
in the alkali-poor stage, for the electropositive alkali layers will always
tend, because of mutual repulsion, to distribute themselves as uniformly
as possible over the whole graphite lattice.
Similar bonding relationships should also exist in the ammoniates
and amine derivatives of the alkali graphite compounds. This is indicated by Hennig's discovery (14), which has already been mentioned,
that the conductivity of carbon electrodes increases in the electrolysis of
ammonium salts in liquid ammonia. Moreover, the ammoniates have
practically the same temperature-independent paramagnetism as the
ammonia-free alkali graphite compounds (see Section IIIAB) . It remains,
however, to be explained why sodium and the alkaline earth metals,
which are taken up very incompletely or not at all in the molten or
vapor state, react as readily as do potassium, rubidium, and cesium
if the metals are dissolved in liquid ammonia. This is certainly related
to the fact that the state of all these metals in the blue solutions is very
similar. They are extensiveIy or compIetely ionized, while the dissociated
electrons are loosely bound in solvate complexes. Uptake of electrons by
graphite and the associated absorption of the metal can occur much more
w.
248
RUWRFF
readily from these solutions than in a reaction with the metals in the
fused or vaporized state. In the latter case the ionization energy of the
metal is involved (48, 29) and this increases from cesium to lithium
and the alkaline earth metals. The ammonia molecules taken up with
the metal in stoichiometric proportions presumably form complexes and
stabilize the interstitial compound. It is relevant in this connection to
think of Zintls polyanionic compounds, such as Na4Pb9 and others ( 9 1 ) ,
which are stable only in the form of ammoniates and decompose when
the ammonia is removed.
The decrease in stability of the ammoniates from cesium to lithium
and the alkaline earth compounds does, however, make it less probable
that in these compounds a true ionic bond, corresponding with the formula C12-K+ * 2NH3, is present. If this were so the stability of the
ammonia complex would increase with decreasing radius and increasing
charge, as it does for a saltlike compound. For the ammoniates, therefore, the ionic formula represents only a limiting case.
B. GRAPHITESALTS
In the presence of concentrated acids, graphite may be oxidized by
strong oxidizing agents or by anodic oxidation to the so-called graphite
salts (22, 83).In these compounds, according to Hofmann et al. (22, 60,
$8, 6 8 ) , anions such as HS04-, NOs-, C104-, HFz-, and CF3COOare incorporated in the graphite lattice together with molecules of acid.
The separate graphite salts may be transformed reversibly into one
another by treatment with other concentrated acids. Thus, the bisulfate
may be produced from the nitrate by treatment with concentrated sulfuric acid, and this in turn made into the perchlorate with perchloric
acid (60). These changes may also be reversed.
The name graphite salt is justified by the fact that in the production of these compounds graphite behaves as a true metal. The oxidizing
agent takes electrons from the graphite, leaving the carbon layers positively charged and able to bind anions,
Graphite
+ 3n 0 + nHpSO,
Graphiten+(HSO4)n-
-i
+ n2 H@.
Oxidation and the associated incorporation of the anions occurs stepwise. The first stage for the bisulfate, which is deep blue in color, like
the first stage of all other graphite salts (blue graphite) (5, 78) has
the approximate composition C24+HSOa- 2H2S04. Each anion takes
two molecules of acid with it into the lattice. With this relationship a
close and regular packing of anions and acid molecules is obtained which
fits the hexagonal network of carbon atoms well ( 6 0 , 5 5 ) .
The second, third, and following stages, which are no longer blue but
249
black, result on incomplete oxidation or by reduction of the first stage.

In stages 2 and 3 the ratios anion:C atoms are 1 : s and 1:81, respectively, from which it follows that the anion net in the second and following stages is less dense than in the first stage (compare the similar
relationships for the alkali graphite compounds in Section IIIA) .
The stage to which the graphite is oxidized in the presence of a particular acid depends not only on the strength and quantity of the oxidizing agent but also on the acid concentration. As an example of the
influence of acid concentration it may be mentioned that in 83% sulfuric
acid and with excess of oxidizing agent, the formation of the first stage
of the bisulfate does not occur and only the second stage is obtained.
In 63% sulfuric acid oxidation goes to the third stage and a t concentrations below 50% uptake of bisulfate can no longer be detected by X-ray
examination.
The nature of the acid is also important. With sulfuric, nitric, perchloric, and selenic acids the corresponding first stage is readily obtained.
Phosphoric and arsenic acids lead only to stages corresponding to a low
degree of oxidation and organic acids such as acetic acid give no graphite
salts. The incidence of reaction is apparently related to the strength or
self-dissociation of the acid. This is especially well seen by comparing
the behavior of graphite anodes on electrolyzing concentrated or anhydrous acetic, dichloroacetic, trichloroacetic, and trifluoroacetic acids.
Unlike acetic acid, the halogenated acids attack graphite increasingly
in the order given above, which corresponds to that of increasing acid
strengths. In trifluoroacetic acid and also in the strongly acidic boron
trifluoride-diacetic acid compound graphite electrodes swell so much as
a result of graphite salt formation that they disintegrate in a short
time (68).
Graphite salts are instantly decomposed by water and by polar organic solvents. Consequently the isolation of these compounds is extraordinarily difficult and attempts to remove adhering acid by washing
with other liquids almost always leads to decomposition of the graphite
compound. They may be reduced to graphite under certain conditions
by such reducing agents as ferrous or stannous salts or by cathodic reduction, and these reactions may be used in evaluating the degree of
oxidation.
DISTANCE.
(b)BETWEEN
TABLE V
CARBONPLANESIN
FIRST-STAGE
GRAPHITESALTS
~~
~~
7.85
7.95
7.99
8.06
8.19
8.2B
250
w.
RUD~RFF
The graphite salts were the first case in which the stepwise uptake
and the associated change of the graphite lattice was followed exactly
by X-ray methods up to the fifth stage (55). Discussion of details may
be dispensed with as the structure, including the arrangement of the
carbon planes, is the same as in the alkali metal-graphite compounds. A
picture of the graphite salt structure may be obtained from Fig. 8 by
replacing the alkali metal layers by anion and acid layers. The distance
between the carbon layers when acid layers are incorporated in the lattice is naturally greater than with alkali layers. It amounts to about
8 A. From the values given in Table V it may be deduced that the degree
of lattice expansion increases somewhat with the size of the acid residue
from nitrate to biselenate.
The regular distribution of the anionic layers and the discontinuous
transition from one stage to another, which is associated with a completely new distribution of the anion layers in the lattice, as well as the
ready anion exchange, may be explained readily in terms of the ionic
character of the compound. Finally, this ionic structure may be recognized most clearly from the fact that the transformation of graphite
into a graphite salt may be carried out reversibly by electrochemical
means. The emf of the first stage of the bisulfate has been determined
as 0.6 volt against graphite and 0.49 volt against a platinum-hydrogen
electrode (60).
The work of Hennig (27) on the change of electrical properties of
graphite on forming graphite bisulfate is consistent with the constitution
of graphite salts. It must be mentioned a t the outset that the investigation was limited to Compounds with a very low degree of oxidation (the
maxinium was 4 meq/gm-atom of C as opposed to 47 meq/gm-atom in
stage 1) . The bisulfate compounds were prepared by anodic oxidation
in concentrated sulfuric acid. Resistance measurements during the oxidation showed that a decrease of resistance of the graphite was associated with the uptake of bisulfate ions. The Hall coefficient also acquires
a positive sign.
These results may be interpreted on the basis of the band model for
graphite (49, 88). Because of the loss of electrons during the oxidation,
holes are produced in the lower almost completely occupied band of
graphite and as a result the number of charge carriers increases. Since
this occurs predominantly in the lower band, the Hall constant is positive. Graphite bisulfate is thus a conductor of the p-type.
When the bisulfate compound is reduced cathodically, the conductivity sinks once more, though the initial value for the graphite taken
is not reached again. When about two-thirds of the compound is reduced,
the conductivity does not change further even when further current is
251
passed. Hennig concluded from this that reduction was not completely
reversible and explained this by the occurrence of stable residue compounds (compare Section IV) .
C. HALOGEN-GRAPHITE
COMPOUNDS
The halogens show characteristic differences in their behavior towards graphite. Investigations up to the present have failed to show a
reaction of iodine with graphite. Bromine and chlorine give unstable
graphite compounds with the limiting composition CsX, which are stable
only under the saturation pressure of the halogen. On the other hand
fluorine forms the two compounds C4F and CF which are very
inert chemically and are homopolar in character. Iodine monochloride
behaves towards graphite similarly to bromine.
1. Graphite Bromide
Graphite bromide is formed by the action on graphite of gaseous or
liquid bromine or of bromine dissolved in organic solvents. About 0.8 gin
of bromine per gram of graphite is taken up from an atmosphere saturated with bromine vapor a t room temperature, corresponding with an
atomic ratio of 1Br:8C (21, 5 6 ) . This value, which is calculated from
the weight increase, is found for natural graphites of very different particle size and has been verified by HBrold (31) from tensimetric measurements. The same author finds the somewhat lower value of 1Br:lOC
for Acheson graphite.
The bromine which has been taken up is for the most part rapidly
given up on diminishing the external pressure of bromine vapor. However, a residue, which can amount to 3-10% according to the crystal size
of the graphite and the rate of decomposition, remains behind in the
graphite. Even at a red heat it is not possible to get graphite completely
free of bromine. These bromine-poor products, which are called residue
compounds, have been more closely investigated by Hennig (28) (compare Section IV).
From X-ray investigations, pyknometric measurements under liquid
bromine, and from Znicroscopic measurements of the volume increase of
graphite crystals it is found that, for the composition C8Br (or ClsBrz)
alternate interplanar spacings of the graphite lattice are widened to
7.05 A by the entry of bromine. Intensity calculations for the (001) interferences agree best with the assumption that the bromine forms a
layer between two carbon layers (56).I n addition to this second stage
there probably exists a third stage poorer in bromine (89, 31).
The constitution of graphite bromide has been discussed many times.
An indication of a partly ionic structure is derived from the fact that
w. RUWRFF
the spacing of the carbon planes (7.05 A) is greater than the sum of the
252
effective cross section of a graphite and a bromine layer. I n accordance

with this one obtains in a calculation of the spatial requirement of bromine too high a value (23.5), which approximates to the value for
bromide ions in salts (25.0) and is considerably above the atomic volume
of bromine (19.2).
Magnetic measurements have not made any substantial contribution
to the question of the bonding of bromine. A weak diamagnetism has
been found by several authors (41,$3, 48, 30). Ubbelohde finds for
It follows from all the measurements that
C S . S ~ Bxgm
~ , = -0.40 x
the anomalous diamagnetism of graphite is lowered by the uptake of
bromine and is almost lost at high bromine concentrations.
Conductivity measurements carried out on preparations with high
arid low bromine contents show that the conductivity of graphite increases considerably with the incorporation of bromine (48, 28). Hennig
(28) finds that the Hall constants of graphite bromide and bisulfate,
after treating the graphite so as to produce the same fractional decrease
in the resistance, are approximately of the same magnitude and he concludes that the two compounds are very similar in constitution. From
these measurements Hennig calcuIates the relationship of the ionized to
the un-ionized fraction in graphite bromide as 1Br- to 3Br2. The ideal
formula of stage two is then Css+Br- 3Br2. According to this view the
compound is a graphite bromide and belongs to the saltlike graphite
compounds. It is, however, characteristic of these that they may be
changed reversibly into one another by ion exchange. One would therefore expect that, under concentrated sulfuric acid, graphite would be
transformed into the second stage of the bisulfate. This reaction does
not, however, occur a t room temperature even after weeks. The bisulfate
is first formed a t 80-100C and then only the third stage results with a
ratio 1HS04-:81C. It thus appears that there are certain differences
between graphite bromide and graphite bisulfate.
Graphite bromide is of interest in relation t o the carbon-catalyzed
formation of hydrogen bromide from the elements. This synthesis occurs
without a catalyst as a homogeneous reaction a t 500' but has a measurable rate at 120" if carbon is used as a catalyst. It is not very likely
that under the reaction conditions (higher temperature, low partial pressure of bromine) bromine molecules can be introduced between the carbon planes. A much more likely assumption is that bromine is absorbed
on the basal faces of the graphite crystals and is bound there in the
same way as it is bound between the planes in graphite bromide. This
view is supported by measurements of the catalytic behavior of graphite
oxide soot in the synthesis of hydrogen bromide (75). This soot, the
253
surface of which is predominantly made up of basal faces (compare

Section I I A l ) , shows so high a catalytic activity that it seems impossible to attribute this effect to the small surface of the prismatic faces.
The activation of halogens by carbon may probably be thought of
in the same way as the activation of molecular oxygen by semiconductors (see summaries under refs. 16, $6). According to this View, chemisorption of bromine on the basal faces of the graphite crystal is linked
with an electron displacement or transfer of the free electrons of the
graphite to bromine, We may then expect that when the electron concentration in the carbon planes is lowered, as it is by the incorporation
of metallic halides (see Section IIID3), the catalytic properties of the
graphite will deteriorate. This has been fully substantiated in the author's laboratory by a comparison of the rate of formation of hydrogen
bromide with graphite or indium trichloride-graphite as catalyst.
2. Graphite Chloride
Whereas graphite bromide has been known for a long time, the lower
rate of reaction of chlorine with graphite delayed the discovery of
graphite chloride, the existence of which has been proved only recently.
In the course of an investigation of the conductivity of graphite under
liquid chlorine a t -33" Hennig (B)
observed that the resistance decreased in the course of three days to one-tenth of its initial value. He
concluded that chlorine, like bromine, is able to enter the graphite lattice.
Juza and Schmidt (49, 40) came to the same conclusion when, in studying the sorption of chlorine and bromine on active carbon, they noticed
a fall in the anomalous diamagnetism of the carbon. Juza and his coworkers (49, 98), in a detailed investigation, used the change of susceptibility as a criterion for interstitial compound formation. The rate
of reaction between chlorine and graphite proved to be very dependent
on particle size. With particles of diameter 0.5-1 mm at -78' the anomalous diamagnetism disappeared in about 500 days whereas, for a diameter of 0.03 mm, it did so in about 100 days. The rate of entry of chlorine
into the graphite lattice increased with rising temperature and was a
maximum a t -12". At 0
' and above a reaction could no longer be detected by the magnetic method.
In order to determine the composition of the saturated graphite chloride a susceptibility-concentration diagram was drawn, from which it
was found that 0.10-0.136 gm-atoms of chlorine per gram atom of carbon d c e d to destroy the anomalous. diamagnetism completely. These
values are consistent with the formula C&1, which is analogous to the
formula of graphite bromide. Isobaric degradation of the graphite chloride at low temperatures yields results which also support this formula.
w.
254
RUDORFF
Graphite chloride prepared at low temperatures is stable at 0", above

which temperature it decomposes, even under the saturation pressure of
chlorine. Juza refers to the fact that graphite bromide also decomposes
at increased temperatures under the saturation pressure of bromine.
Uptake of chlorine may be substantially increased if the decomposition product of graphite bromide, i.e., the residue compound, is used in
place of pure graphite, and the amount of chloride absorbed rises with
the bromine content of the graphite. On thermal decomposition, specimens which contain bromine retain more chlorine than does bromine-free
graphite. An increased uptake of chlorine may also be obtained by using
graphite made by decomposing the bisulfate in the presence of hydrogen
chloride. Summarizing, it may be said that the behavior of chlorine and
bromine is similar, The reaction rate is lower for chlorine, which must
be related to the fact that the chloride is stable only at much lower
temperatures.
3. lodine and Graphite
No change in the conductivity or susceptibility of graphite can be

observed when it is allowed to react with iodine at -78" to 183" (d8,39).
Hdrold (31) was also unable to detect any reaction up to 500". On the
other hand Juza (39),using graphite containing bromine, observed a
pronounced lowering of the susceptibility. The amount of iodine was
about three times as great as the bromine content of the graphite so that
it could not have been taken up only as iodine monobromide.
4. Iodine Monochloride and Graphite

Absorption of iodine monochloride by graphite has been detected
both by the conductivity change and from X-ray observations (98).
Riidorf and Sils (73)found that graphite increases in weight by about
250% in iodine monochloride vapor at 30, while in single crystals saturated with the monochloride there was more than a threefold increase
in volume. The identity period, I,, was calculated to be 21.2 A and embraced two carbon planes. It was not, however, clear if a first or second
stage was present. Croft ( l l a ) found that in the reaction of iodine trichloride with graphite, the product prepared a t 90" swells when heated
strongly. After washing with hydrochloric acid it retains 30% of trichloride.
D. INTERCALATION
OF METALCHLORIDES
IN GRAPHITE
1. Preparation and Properties
These compounds, which have become known in large numbers only

in recent years, form a specially interesting class of graphite compounds
255
because of problems associated with their constitution. Their investigation probably cannot yet be regarded as complete, for a number of experimental observations still await an explanation. The compounds result
when graphite is heated with the anhydrous metal chloride. The intercalated chloride is given off again a t a sufficiently high temperature. Up
to the present only metal chlorides have been observed to form these
compounds, but compounds with other halides should also be formed.
The first compound of this sort was ferric chloride graphite (21, 84).
Rudorff and Schulz (65) obtained from graphite and ferric chloride
a t 200-300" a stage 1 compound with 60-70lyo of ferric chloride. I n this
compound separate layers of ferric chloride, as they occur in the lattice
of the pure FeC13, were intercalated between the carbon planes, which
were unchanged, though the distance between them was increased to
9.4
A.
In attempts to separate ferric chloride and aluminum chloride by

intercalation in graphite, Croft (10) found that aluminum chloride
could be intercalated either d o n e or together with ferric chloride. Further work (11, I l a ) led to the surprising result that numerous metal
chlorides are able to react with graphite. The compounds studied are set
out in Table VI, which is taken from Croft's work. It includes thirty
TABLE VI
SUMMARY OF INTERCALATION
TESTRESULTS(CROFT,I l a )
Substances intercalated
in graphite
TaC16
FeCll
CrC13
cacl?
CrOtF?
MoCla
wcle
UClr
UO*C12
ReC14
COc&
RuCl;
Substances not intercalated

in graphite
CUCl
BeCl?
MgCle
CaC1:
BaC12
ZnCh
CdClz
CdIz
HgBCln
HgClt
CClr
SiCL
GeCh
TiCll
SnCla
SnCt
PbClz
PbCL
ThCL
ThIi
PCla
PC16
ASClr
SbCls
BiCls
VCL
SOlCl,
SOCl,
SeOCIz
TeCL
MnCl?
COClZ
NiCll
chlorides and two oxychlorides for which intercalation was detected.

I n general, reactions were carried out a t temperatures uear t o the
melting or sublimation point of the particular halide. After heating the
graphite and halide together for several hours, the graphite was washed
with hydrochloric acid to remove excess halide. Intercalation was rec-
256
w. RUDOEFF
ofired in most cases by swelling of the graphite when it was heated

rapidly to 700. The quantity of halide taken up was determined by
evaporating to fuming with concentrated sulfuric acid and weighing the
residue after burning off the graphite. Alternatively, readily volatile
halides were determined. by thermal decomposition of the graphite compound. The results are scattered over a wide range of compositions. Thus
Croft finds for CuC12,4% ; for AlC& and MoC15, 25% ; for TaClk, 44% ;
and for T1C13, 70%. Two things are noteworthy in Crofts work: first;
intercaIation is not restricted to readily volatile chlorides of low melting
point and, second, cobalt trichloride and palladium tetrachloride, which
do not exist in the free state, are stabilized by intercalation.*
Croft refers to the use of the different behavior of metal chlorides
towards graphite as a means for effecting their separation. Experiments
of this sort have been made by Vickery and Campbell (87) for the rareearth halides. Crofts results on the reactivity of the halides are not
fully borne out. In an independent study of the action of aluminum chloride on graphite it was found that intercalation occurs only if free chlorine is present during reaction. This condition appears to have been
satisfied in Crofts experiments, since the halides were prepared by
chlorinating the metal and were sublimed djrectly onto tbe graphite.
Rudorf and Zeller (70)were able, by carefully subliming off the excess
halide, to prepare homogeneous aluminum chloride-graphite preparations
corresponding to stages 1, 2, and 4, among which stage l., with 5245%
AlC& (1A1Cl3:9C) was remarkable because of its deep blue color. The
same color was shown by stage 1 of the gallium chloride graphite with
60% of GsCl,. On the other hand, only the second stage for indium
trichloride could be prepared, with 53% InCb (61). These two halides
also react only in the presence of free chlorine. The same observation
was made in the reaction of graphite with uranium pentachloride, cadmium chloride, and mercuric chloride. The last two examples show that
chlorides for which Crofts experiments gave negative results are able
to become intercalated if the reaction is carried out in a chlorine atmosphere.
According to Dzurus and Hennjg (15) the free chlorine necessary for
the occurrence of reaction between graphite and aluminum trichloride
does not act catalytically, as Rudorff and Zeller first assumed, but is
partly taken up in the lattice. Rudorff and Landel (61) found this to be
true of the gallium trichloride-graphite compound also. For both compounds the ratio Me:Cl is between 1r3.2 and 1:3.4 and the compounds
therefore oxidize potassium iodide solution and cause iodine to separate.
These intercalation compounds were obtained by prolonged heating of graphite
with the lower chlorides CoCL or PdCL in a stream of chlorine.
257
By contrast the quantity of active halogen is extremely small for the

compounds of graphite with indium chloride, mercuric chloride, and
cadmium chloride, and lies a t the limit of analytical detection.
The compounds show striking differences in their behavior t o water,
acids, and organic solvents. Whereas the blue aluminum chloride- and
gallium chloride-graphite compounds are extremely sensitive t o water
and are largely decomposed by it, it seems that indium chloride- and
ferric chloride-graphite are stable for, even after boiling with concentrated hydrochloric acid, one finds high values for the halogen content
(InC13, 50%; FeC13, 5 5 % ) . New investigations by Rudorff and Landel
show, however, that these compounds are also decomposed by water
and organic solvents. This follows unambiguously from the dependence
of the MeC13 content on the size of the graphite crystals, the nature of the
solvent, density determinations, and X-ray investigations. Probably all
metal halide-graphite compounds, as well as the other polar graphite
compounds, are decomposed by water, though there are great differences
in the extent of decomposition for the separate compounds.j
These differences may be explained as follows: the solvent undoubtedly attacks the edges of the layer planes of the graphite crystal. When
the halide is leached out from the edge, the c interplanar distance of
about 9.4 b can no longer be maintained in this region. Before the solvent can penetrate further into the crystal or the halide can diffuse from
the interior, the interplanar distance at the crystal edges reverts t o about
the value in graphite itself. Leached preparations are thus not homogeneous, but the crystal consists of a nucleus of undecomposed compound which is protected from further decomposition by a more or less
thick skin of graphite. Analysis shows that the smaller the crystals, the
greater the weight of the halide-free aone. The more rapidly the leaching
agent acts (e.g., water > acetone) the more extensively does decomposition penetrate into the crystal.
There is only a difference in degree between graphite compounds of
aluminum trichloride or gallium trichloride on the one hand and indium
trichloride or ferric chloride on the other as regards their behavior in
water; for the first two, the rate of decomposition of the graphite coinpound, or the mobility of the halide within the crystal, is so big that,
for sufficiently small crystals, decomposition extends over the whole
crystal. The lack of homogeneity in leached ferric chloride-graphite
preparations has also been detected by Cowley and Ibem (9) by X-ray
and electron diffraction. It was found that a preparation containing
Crofts analytical remlta therefore probably are not valid for the cumpoeition ,of
the pure compounds, since all samples were treated with hydrochloric acid before
analysis.
258
w. RUDORFF
55% of ferric chloride contained 18% of graphite and 82% of a ferric

chloride-graphite compound with 66.9% of the halide.s
2. Structure
An exact structure determination has up to the present been made

only for ferric chloride-graphite (9). The separate ferric chloride layers,
which lie between two carbon layers, are only 92% occupied. The arrangement of the ferric ions and their distance from one another is the
same as in ferric chloride, the hexagonal a axis of the network of ferric
ions forming an angle of 30" with the a axis of the graphite. The arrangement of chloride ions and their distances, on the other hand, are
different from those in free ferric chloride. They have preferred sites,
which reveals a certain relationship to the carbon atoms of the neighboring graphite layers. In indium trichloride graphite certain additional
lines in the powder diagram indicate that the trichloride is also intercalated as layers with almost the same In-In distance as in free indium
trichloride. From the coniposition an approximately 80% occupation of
the indium trichloride layers is deduced ( 6 1 ) .
For other compounds only the lattice expansion has hitherto been
determined by the X-ray method by calculating the identity period
from the (001) reflections. The distance apart of the carbon planes is
approximately the same for the various halides and equal to 9.5 A.
3. Bonding in the Metal Halide-Graphite Compounds
The bonding has been discussed in detail by Croft (11b) and Dzurus
and Hennig ( 1 5 ) . These authors concluded that the compounds were
ionic. Croft supported this view by considering the differences which he
found in the reactivity of the halides. He points out that, except for
elements of Group IIIA, the intercalated compounds belong to the transition elements, and that, for polyvalent elements, only the chloride of the
higher oxidation state is intercalated. Thus cupric and ferric chlorides
react whereas cuprous and ferrous chlorides do not. Intercalation occurs
only when two criteria are satisfied: it must be possible to transfer electrons to the cation from the conduction band of the graphite, and the
cation must have a correspondingly high electron affinity. Croft considers
the presence of unoccupied orbitals in the cation to be necessary. For
aluminum chloride-graphite and the homologous compounds in which
the cation cannot change its charge, free valency orbitals of the cation
"The lack of homogeneity was attributed by the authors to decomposition of the
ferric chloride-graphite compound formed initially into graphite and a phase richer
in ferric chloride as it cooled.
259
are occupied by electrons from the 2 p a-orbitals of graphite, when a

bond like that in outer orbital complexes is thought to result.
Daurus and Hennig have put forward a modified and more detailed
theory. Measurements of the electrical conductivity and Hall effect on
halide-poor FeC13-graphite and A1C13-graphite compounds with a ratio
1MeC13:200-500 C atoms show that these substances belong to the acceptor or p-type compounds. Since the changes in electrical properties
are the same as in graphite bisulfate and graphite bromide, Dzurus and.
Hennig conclude that the metal halide compounds are similarly constituted and adopt such formulae as, for example, C,+CIFeClz 3FeC13
for ferric chloride-graphite. Only a quarter of the ferric ions are thought
to act as electron acceptors. Earlier magnetic measurements made on
ferric chloride graphite (65) are consistent with this view. For other
metal halides capable of existing in several oxidation states, corresponding formulae were proposed. Calculations of free energy made on Hennigs electrostatic model (29) give the differences in reactivity of the
separate halides in roughly the correct order.
I n the case of aluminum chloride-graphite the halogen which is simultaneously intercalated in the lattice acts as an acceptor: C,+Cl3A1C13. Strong support for this view comes from the fact that the
quantity of aluminum chloride intercalated is proportional to the concentration of acceptor. In this the acceptor may be-in
addition to
chlorine-bromine, iodine, or a halide of one of the transition elements.
Thus aluminum chloride is intercalated in the absence of free halogen
if ferric chloride is added as acceptor. For very low concentrations the
ratio of intercalated aluminum chloride to ferric chloride lies between
2 and 3, so that the formula C,+ClFeCL * 3A1C13 suggests itself for
this compound. Whether, and to what extent, this view still remains to
be amended can be shown only by further experiments. Magnetic measurements on suitable compounds might be particularly revealing.
E. OTHERINTERCALATION
COMPOUNDS
I n the view of the constitution of metal chloride-graphite compounds
put forward by Croft and Hennig (see Section IIID), the controlling
factor in intercalation is the electron affinity and electron configuration
of the cation, while the nature of the anion is of secondary significance.
Croft considers this theory to be supported by the fact that oxides and
sulfides of polyvalent elements in higher oxidation states are also able
to react with graphite (compare ref. I d ) . In a further publication, Croft
( I S ) generalizes the information gaiined on graphite compounds and
comes to the conclusion that other crystal lattices, and particularly layer
w.
260
RWDORFF
lattices, should be able to take up suitable compounds, provided an

electronic interaction can occur between the host lattice and the intercalated partner, and that the host lattice has room to take up the latter.
Croft has attempted t o verify this prediction experimentally in the case
of boron nitride, which is discussed in a later section.
1. IntercuEatim of Metal Oxides and Sulfides in Graphite (12)
The investigation embraces the oxides Sb204, CrO3, and Moos and
the sulfides Sb2SS, Ti$, CuS,, FeS2, Cr2S3,V2S3, MoS2, WS2, and PdS.
The compounds were heated in quartz ampoules with graphite in molecular ratios between 1 : l and 2:l. The temperatures were between 200"
(TlZS or CrOs) and 57OOC (WS2). In reactions with the sulfides, free
sulfur was also added. Reaction products from sulfides were extracted
with carbon disulfide to remove free sulfur, separated from finely divided metal sulfide which had not reacted by sieving and sorting under
the microscope, and washed with nitric acid. Excess of molybdenum or
chromium trioxide could be removed by sublimation or by washing with
acid,
Analysis gave the values in Table VII for the oxide or sulfide contained in the graphite. Croft left'open the question as to whether the
quantities found represented the greatest possible amounts which could
TABLE 1711
I'ERCEXTAGE
Oxide
sbzoc
Cr03
Md)3
Percentage
1.6
55.0
10.75
OF OXIDE OR SULFIDE IN GRAPHITE
Sulfide
Percentage
Sulfide
Percentage
Sb?&.S
TlS?
cus1.3
FeS2
9.4
27.0
16.2
12.2
Crdh
VzS3.s
4.0
3.7
14.1
8.8
WS2.4
PdS?.?
be intercalated. Some of these products showed properties similar to

those which are characteristic of the metal chloride compounds, e.g.,
swelling when strongly heated. It was possible to sublime out intercalated molybdenum trioxide unchanged by heating the graphite compound. The iron sulfide and copper sulfide compounds decomposed when
treated with dilute acids. X-ray powder photographs for the sulfidic
products were very complicated but were different from those of the
lower sulfides. Interplanar distances of 11-12 A were measured for the
compounds of the oxides with graphite. Croft interprets these experiments as showing that metals whose chlorides are intercalated in graphite
261
are also able to react as oxides or sulfides provided the metal is in a high
oxidation state.
2. Molecular Compounds of Boron hritride
The crystal structure of boron nitride resembles that of graphite.

The boron and nitrogen atoms form plane regular hexagonal nets which
are arranged parallel to one another at a distance of 3.33 A. An essential difference between graphite and boron nitride is that in the latter
there are no free electrons. Pure boron nitride is white and does not
conduct electricity.
The similarity of the two structures led Croft (13) t o attempt the
preparation of boron nitride intercalation compounds. He used the metal
halides SbCI3, SbCI5, AsC13, CuCI, CuCl2, FeC13, AlC4, and also BF3,
BC13, Br2, ICl, liquid ammonia, and N2H4. The same procedure was
used as in the preparation of metal chloride-graphite compounds.
Croft concluded froin the weight increase of the boron nitride as well
as from analysis of the products after washing with dilute hydrochloric
acid that, of the compounds studied, SbCla, AsC13, CuC1, FeC13, AlCla,
and NzH4 were intercalated in boron nitride in amounts between 2%
(SbCl3 and N2H4) and 13% (AIC13). As in the case of graphite compounds, these products swell considerably when heated. The photographs
published by Croft are very striking. They show that boron nitride single
crystals containing 2% of hydrazine swell similarly to ferric chloride
graphite containing 56% of ferric chloride.
Croft explains the results in terms of the acceptor property of the
boron atoms. They tend to take an electron pair from a donor molecule
to form sp3 bonds. This is why only halides in lower oxidation states
react with boron nitride (e.g., CuCl and SbC13 but not CuClz or SbCI6).
This is different from the case of graphite in which the carbon planes
are a source of electrons. The intercalation of aluminum and ferric
chlorides is attributed by Croft to bonding by the electron pairs of the
nitrogen atoms of boron nitride. It is surprising, however, that boron
trifluoride and trichloride, which are particularly good electron acceptors, appear unable to become intercalated.
A closer investigation and characterization of these interesting reaction products seems called for, since some of Crofts results are contrary
to investigations made in the authors laboratory on the behavior toward
boron nitride of FeCIa, CoCL, NiC12, NiBrz, AICla, and other compounds (69).No change in the boron nitride could be observed and, especially, there was no expansion of the boron nitride lattice, which would
be expected as the result of intercalation. An explanation for the dis-
262
w. R ~ ~ D ~ R F F
crepancy can probably be sought only in a difference between the starting materials. I n the experiments made in the Tubingen laboratory, two
pure white finely divided preparations were available which had been
prepared in different ways a t 800". Determinations of boron and nitrogen gave 98-99% of BN. The specimens used by Croft were obtained
from an amorphous product by heating to 2000". H e described the crystalline and partly laminar product as black and lustrous. This description
is not, however, consistent with the properties of boron nitride, so that
Croft's results probably do not refer to the pure compound.
IV. Residue Compounds
Analytical investigations of graphite compounds have repeatedly

shown that when the pure compound is decomposed or reduced, a small
proportion of the intercalated substance is obstinately held in the lattice,
and is often recovered only after complete destruction of the graphite.
According to Hennig (27) decomposition always leads to "residue compounds" which differ from the actual graphite compounds in various
properties, e.g., conductivity and Hall effect. Residue compounds may
be reconverted into the same lamellar compounds from which they were
produced. The residue compounds from graphite bisulfate and graphite
bromide have been studied in detail (27, bra, 28, &a, 50).
The quantity of intercalated material retained depends on the composition of the corresponding lamellar compound and increases with the
original concentration in the host lattice. Differences between the two
compounds sre associated with the ionized part of the intercalated component. Hennig has calculated from the conductivity of the residue
compound of graphite bromide a ratio of lBr-:2.2BrZ and for graphite
bisulfate 1HS04- :4H2S04,whereas in the saturated lamellar compounds
It is difficult t o find an
the ratios are 1Br-:3Brz and 1HS04-:2HzS04.
explanation for the increase of the fraction ionized for graphite bromide
and for the decrease in the case of the bisulfate.
The two compounds differ structurally in the way in which the reaction partner is distributed in the graphite lattice. Whereas in the lamellar
compounds the intercalated layers have a periodic distribution-i.e., are
interspersed stepwise-between the carbon planes, distribution in the
residue compounds is irregular. Precision measurements of the lattice
constants for the residue compounds of graphite bromide and bisulfate
gave only a trifling increase in the c axis of the graphite, most of the
deviations being within the limits of error. Hennig concluded that impurities in residue compounds were situated mainly in holes and imper-
263
fections in the graphite lattice, the fraction remaining intercplated between the carbon planes being under 1%.
Chemically, the residue compounds are of interest because they possess a greater chemical reactivity than pure graphite. Thus. Hennig was
able to show that the residue compound from graphite bromide (1.4Br:
1OOC) was able to react to a small extent with iodine or aluminum chloride (2A1C13:1OOC), whereas there is no reaction for pure graphite. For
the influence of the residue compound of graphite bromide on the intercalation of chlorine in graphite see Section IIIC2.
Many properties of residue compounds are also consistent with the
assumption that the separate crystals of a graphite compound are not
homogeneous after decomposition, and still contain undecomposed compound in their interiors. This is surrounded by a more or less thick outer
layer of graphite which protects it from further decomposition. This
idea has been discussed already in Section IIID in connection with the
behavior of metal halide-graphite compounds towards water. It will be
necessary to wait for further experiments in order to explain more precisely the mode of decomposition of graphite compounds.
V. Comparative Survey
The new compounds found in the last ten years, CaF, the ammonia and amine derivatives of alkali- and alkaline earth metal-graphite
compounds and the large group of metal halide intercalation compounds,
constitute further examples of the surprisingly diverse reactivity of
graphite. The study of physical properties of the graphite compounds,
together with elucidation of structures by X-rays, has led to a deep
insight into the mode of reaction of graphite and the constitution of intercalation compounds.
The formation of graphite compounds is linked in every case to a
change in the electron concentration in the graphite. The elements with
the highest electron affinity, i.e., fluorine and oxygen, produce the greatest
change, in that they form covalent bonds with the carbon atoms. Linked
with this we find a change in the structure of the carbon layers in carbon
monofluoride and graphite oxide, as well as a decrease or disappearance
of conductivity. Strictly speaking, these are no longer graphite compounds, for the characteristic properties of graphite, namely the presence
of free electrons and the arrangement of carbon atoms in planes, have
been lost.
Chemically, these homopolar compounds differ from the other graphite
264
w.
RUDORFF
compounds in their stability to water and other solvents7 and in their

relatively low reactivity, which is particularly striking for the fluorine
compounds. All other reactions of graphite which lead to intercalation
compounds may be associated with an electron transition between the
reaction partner and the carbon planes. When this occurs only the electron concentration in the conduction band is altered and the layer planes
remain as they were in graphite. The compounds are therefore polar in
character. They differ from the homopolar compounds in their crystal
structure, their conductivity, which is greater than that of pure graphite,
and also in their greater reactivity. The intercalation of the reaction
partner in a series of stages is characteristic of these compounds. This
is certainly to be attributed to the tendency for the layers with a charge
of the same sign to distribute themselves a t equal distances in the lattice.
Graphite acts as an amphoteric substance in producing these polar
compounds: it is an electron acceptor towards the strongly electropositive alkali and alkaline earth metals and a donor to atoms or ions with
a sufficient electron affinity. However, the number of electrons accepted
or donated always remains small in relation to the total electron concentration of the individual carbon planes. I n the graphite salts, for
example, in which the ionic cfiaracter is undoubtedly most strongly
developed, there is only one positive charge to 24 carbon atoms of a
carbon layer. This follows from the formula for stage 1 of the bisulfate,
CZ4+HS04- 2H20. A similar ratio is encountered for other compounds,
for example, c27CI- * 3A1C13 or CP7+A1C14- 2AlCl3; CZ7+GaCI4 *
2GaC1,.
It is interesting that apparently only small deviations from this ratio
are possible in each stage, which, although it possesses a certain range
of homogeneity, goes over discontinuously to other stages with greater
changes in composition, These have new charge distributions in their
lattices. I n relation t o the charge of the carbon atoms, appreciably more
atoms or molecules are intercalated, i.e., the greater part are not bound
ionically, and probably serve to fill in gaps, PO that the intercalated partner can form coherent layers between the carbon planes. All compounds
of this sort studied to the present have been found to be reactive and to
he more or less completely decomposed by water.
REFERENCES
1. Asher, R. C., and Wilson, S. A,, Sature 181,409 (1958).
2. Balfour, A. E., Riley, H. L., and Robinson, R. M., J . Chem. SOC.1936, 456.
3. Bernal, J. D., Pioc. Roy. Soc. A106, 749 (1924).

4. BiItz, W., Raumchemie der Festen Stoffe. L. V o s , Leipzig, 1937.
Swelling of
of oxidation.
graphite oxide in water is reyersible and does not change the degree
265
5. Brodie, B. C.,Ann. Chem. Phys. [31 45, 351 (1855).

6.Brodie, B. C., Ann. Chem. 114, 6 (1860).
7. Carter, A. H., De Moulds, V., and Riley, H. L., J. Chem. SOC.1937, 1305.
8. Clauss, A., Plass, R., Boehm, H. P., and Hofmann, U., 2.anorg. u. allgem. Chem.
291,205 (1957).
9. Cowley, J. M., and Ibers, J. A., Acta Cryst. 9,421 (1956).
10. Croft, R. C.,J. Appl. Chem. 2,557 (1952).
11. Croft, R.C., Nature 172, 725 (1953).
Ila. Croft, R. C., Australian J. Chem. 9, 184 (1956).
Zlb. Croft, R. C., Australian J. Chem. 9, 194 (1956).
12. Croft, R. C., Australircn J. Chem. 9,201 (1956).
13. Croft, R. C., Australian J. Chem. 9,206 (1966).
14. Dzurus, M., and Hennig, G. R., J . Chem. Phys. 27,275 (1957).
15. Dzurus, M., and Hennig, G. R., f. Am. Chem. &c. 79,1051 (1957).
16. Fensham, P. F., Quart.Revs. 11,227 (1957).
17. Finch, G. J., and Wilman, H., Proc. Roy. SOC.A155, 345 (1936).
18. Franklin, R.E., J. chim. phys. 50, C 26 (1953).
19. Fredenhagen, K., and Cadenbach, G.,2. anorg. u. aUgem. Chem. 158, 249 (1926).
20. Fredenhagen, K., and Suck, H., 2.anorg. u. allgem. Chem. 178,353 (1929).
82. Frenael, A., Dissertation, Berlin, Techniiche Hochaehule, 1933
93. Frenzel, A.,and Hofmann, U., 2.Etektrochem. 40,511 (1934).
$3. Goldsmith, M., J. Chem. Phys. 18,523 (1950).
i?d. Hadzi, D., and Novak,A., Tram. Faraday SOC.51,1614 (1955).
$6. Haasel, O.,and Mark, H., 2.Physik 25,317 (1924).
1.Hauffe, K., Angew. Chem. 67,189 (1955).
97. Hennig, G.R., J. Chem. Phys. 19,922 (1951).
flu. Hennig, G. R., J. Chem. Phys. 20, 1438 (1952).
98. Hennig, G. R., J . Chem. Phys. 20, 1443 (1952).
$9. Hennig, G. R., Proc. 1st and $nd Con!. on Carbon, Univ. of Buffalo, p . 103
(1956).
30. Hennig, G. R., and McClelland, J. D., J . Chem. Phys. 23,1431 (1955).
31. Hkrold, A.,Bull. SOC. chim. France 1955,999.
39. Hkrold, A.,Compt. rend. 232,838 (1951).
33. Hofmann, U.,Ergeb. ezakt. Naturw. 18,229 (1939).
34. Hofmann, U., and Boehm, H. P., 2.anorg. u. aUgem. Chem. 278,58 (1955).
35. Hofmann, U.,and Holst, R., Chem. Ber. 72,754 (1939).
36. Hofmann, U.,and Konig, E., 2.anorg. u. allgem. Chem. 234,311 (1937).
37'. Hofmann, U.,and Ohlerich, G., Angew. Chem. 62,16 (1960).
38. Hofmann, U.,Frenzel, A., and Csalan, E., Ann. Chem. 510, 1 (1934).
59. Juza, R.,and Schmeckenbecher, A., 2. unorg. u. aUgem. Chem. 292, 46 (1957).
40. Juza, R., and Schmidt, P., Chem. Ber. 89,428 (1956).
41. Juza, R., Liibbe, H., and Heinlein, L., 2.anorg. u. allgem. Chem. 258, 105 (1949).
&. Juza, R.,Schmidt, P., Schmeckenbecher, A., and Jiinck, P., Naturwks. 42, 124
(1955).
43. Juza, R., Jonck, P., and Schmeckenbecher, A., 2. anorg u. allgem. Chem. 292,
34 (1957).
4. Klemm, W., and Abitz, ref. in FZAT Rev. German Sci. 19N-lOqS 9,II, 14 (1948).
&u. Kmatfro, E.A , J . Chem. Phys. 21,2162 (1953).
46. Laidla, D. S., and Taylor, A., Nature 146, 130 (1940).
&. Laves, P., and Baskin, Y.,2.K+t. 107,338 (1956).
266
W. E h W F
47. Lipeon, H., and Stokes, A. R., Proc. Roy. SOC.A181,93 (1942).
48. McDonnell, F. R. M., Pink, R. C.,and Ubbelohde, A. R., J. Chem. Soc. 1951,
191.
Mrozowski, S., Phys. Rev. 78,644 (1950).
Mrozowski, S., J. Chem. Phys. 21, 492 (1953).
Quartermann, L., and Primak, W., J. Am. Chem. SOC.74, 806 (1952).
Riley, H. L., Fuel 24, 8 (1945).
63. Riley, H. L., Fuel 24, 43 (1945).
64. Rubisch, O., Diasertation, Tubingen, 1956.
66. RudorfT, W., 2. physik. Chem. (Leipzig) B45, 42 (1938).
66. Rudorff, W., Z. anorg. u. allgem. Chem. 245, 383 (1941).
67. Rudorff, W., Wien. Chem. 2. 47, 172 (1944).
68. Riidorff, W., 2. anorg. u. allgem. Chem. 254, 319 (1947).
69. Rudorff, W , and Brodersen, K., Z. Naturjorsch. 12b, 595 (1957).
60. RudorfT, W ,and Hofmann, U., 2.anorg. u..crUgem. Chem. 238,l (1938).
61. Rudorff, W., and Landel, A., 2.anorg. u. a%em. Chem. 293,327 (1958).
62. Rudorff, W , and RudorE, G., 2. anorg. u. allgem. Chem. 253,281 (1947).
63. Rudorff, W , and RiidortT, G., Chem. Ber. 80,413 (1947).
64. RiidorB, W , and Rudorff, G., Chem. Ber. 80,417 (1947).
65. Rudorff, W., and Schulz, H., 2.anorg. u. allgem. Chem. 245,121 (1940).
66. Rudorff, W , and Schulze, E., Angew. Chem. 66,305 (1954).
67. Rudorff, X.,and Schulze, E., 2.anorg. u. allgem. Chem. 277,156 (1954).
68. Rudorff, W.,and Siecke, W.F., Chem. Ber. 91, 1348 (1958).
69. Rudorff, W . and Stumpp, E., 2. Naturjorsch. 13b, 459 (1958).
70. RudorfT, W ,and Zeller, R., 2. anorg. u. allgem. Chem. 279,182 (1958).
71. Riidorff, W ,Hofmann, U., Rudorff, G , Endell, J., and R u e s , G., Z. anorg. u.
allgem. Chem. 256, 125 (1948).
78. Riidorff, W.,
Schulze, E., and Rubisch, 0 , Z. anorg. u. allgem. Chem. 282, 232
(1955)
78. RudorfT, R , Sils, V., and Zeller, R., 2.anorg. allgem. Chem. 283,299 (1956).
74. Ruess, G., Monatsh. Chem. 76,381 (1946).
76. Ruess, G., and Vogt, F., Monatsh. Chem. 78,222 (1918).
76. Ruff, O., Chem. Ber. 69A. 181 (1936).
77. Ruff,O., and Bretschneider, O., 2. anorg. u. &gem. Chem. 217, 1 (1934).
78. Schafheutl, J . prakt. Chem. 21, 155 (1841) ; 76,300 (1859).
79. Schleede, A., and Wellmann, M., 2.physik. Chem. (Leipzig) B18, 1 (1932).
80. Schulze, E , Dissertation, Tubingen, 1954.
81. Staudenmaier, L., Chem. Bet. 31, 1394 (1899).
82. Tammann, G., and Sworykin, A., 2. anorg. u. allgem. Chem. 168, 218 (1927).
83. Thiele, H., Z. anorg. u. allgem. Chem. 206, 407 (1932); 2.Elektrochem. 40, 26
(1934).
84. Thiele, H., 2. anorg. u. allgem. Chem. 207,340 (1932).
86. Thiele, H., Kolloid-2.80, 1 (1937).
86. Thompson, H. W., and Temple, R. B., J . Chem. SOC.1948,1432.
87. Vickery, R. C , and Campbell, N. L., J . A m . Chem. SOC.79, 5897 (1957).
88. Wallace, P. R., Phys. Rev. 71,622 (1947).
89. Zeller, R., Dissertation, Tubingen, 1954.
90. Zintl, E., Angew. Chem. 52, 1 (1939).
91. Zintl, E., Goubeau, J., and Dullenkopf, W., 2.physik. Chem. (Leipzig) A154, 1
(1931).
49.
60.
61.
68.
THE SZILARD-CHALMERS REACTION IN SOLIDS'

Garman Harbottle2 and Norman Sutin
Department of Chemistry. Brookhaven National laboratory. Upton. New York
I . Introduction . . . . . . . . . . . . . . . . . . . .
268
A. Elastic-Collision or Billiard-Ball Model . . . . . . . . . . 268
B. Elastic-Inelastic Collision Model . . . . . . . . . . . . 269
C. Epithermal-Reaction Model . . . . . . . . . . . . . . 269
D . Brush-Heap or Random-Fragmentation Model . . . . . . . . 270
E . Hot-Zone Model . . . . . . . . . . . . . . . . . . 270
I1. Chemical Effects of Nuclear Recoil . . . . . . . . . . . . 271
A . Difficulties in the Interpretation of Experiments . . . . . . . 271
B . Ionic Crystals . . . . . . . . . . . . . . . . . . 272
C. Molecular Crystals . . . . . . . . . . . . . . . . . 278
I11. Post-Recoil Annealing Effects . . . . . . . . . . . . . . 281
A. Introduction . . . . . . . . . . . . . . . . . . . 281
B. The Interpretation of Thermal Annealing . . . . . . . . . 283
1. Annealing of, Initially Correlated Pairs . . . . . . . . . 283
a. Simple Recombination of Correlated Pairs . . . . . . . 283
b. Recombination of Correlated Pairs Modified by Electrostatic
. . . . . . . . . . . . . . . . . 285
Interactions
c. Recombination of Correlated Pairs with Electron Transfer as the
Rate-Determining Step . . . . . . . . . . . . . . 287
d . Recombination of Correlated Pairs after a Random-Walk Procew 288
e. Recrystallization of the Hot Zone . . . . . . . . . . 290
2. Annealing of Initially Randomly Distributed Pairs . . . . . . 292
a . Diffusion-Controlled Bimolecular Recombination . . . . . . 292
3. Thermal Exchange Model . . . . . . . . . . . . . . 293
4 . Phenomenological Approach . Annealing Functions Involving Mixed
293
Kinetics . . . . . . . . . . . . . . . . . . . . .
5 . Processes Distributed in Activation Energy . . . . . . . . 294
6 . Activation Energy Varying with the Number of Defects . . . . 297
C . The Interpretation of Radiation Annealing . . . . . . . . . 298
1 . Gamma Rays . . . . . . . . . . . . . . . . . . 299
. . . . . . . . . . . . . . . . . 300
2. Fast Neutrons
D . Discussion and Comparison with Experimental Data . . . . . . 301
IV. Conclusion . . . . . . . . . -. . . . . . . . . . . .
309
V. Appendix . . . . . . . . . . . . . . . . . . . . .
309
A . Rate of Increase in Retention Due to Recrystallization of Spherical
Zones, Eq . (16) . . . . . . . . . . . . . . . . . . 309
'Research performed under the auspices of the U . S. Atomic Energy Commission.
* John Simon Guggenheim Memorial Foundation fellow .
267
268
GARMAN HARBOTTLE AND NORMAN SUTIN
B. Derivation of Rate Law for Exchange, Eq. (20) . . .

Acknowledgments . . . . . . .
. . .
. .
List of Symbols . . . . . . . . . . . . . .
References
. . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
. . 310
. ,311
. . 311
. .
312
1. Introduction
The chemical effect of a nuclear transformation was observed by

Ssilard and Chalmers (78) in 1934. They irradiated liquid ethyl iodide
with neutrons and found that radioactive iodine could be extracted into
water. The effect was attributed to the rupture of the carbon-iodine
bond by the mechanical recoil imparted to the iodine nucleus by the
incident neutron. Subsequently Fermi et aZ. ( 1 ) showed that the recoil
energy given to the nucleus by the emission of gamma rays following
thermal neutron capture was sufficient to break the bonds holding the
capturing atom to the remainder of the molecule.
The chemical behavior of atoms which have undergone nuclear transformations is as yet imperfectly understood. Numerous investigations
have shown that the radioisotope produced by thermal neutron irradiation of a compound in dilute solution or in the gas phase is usually
obtained in a chemical form other than that of the capturing molecule.
The mechanical recoil (45) imparted to the nucleus by the capture
gamma rays, or the molecular excitation resulting from the internal conversion of the capture radiation must therefore be large enough to ensure the disruption of the chemical bonds holding the capturing atom
to the remainder of the moIecule in practicaIly all neutron captures.
However, when pure liquids or solids are irradiated, an appreciable
fraction of the total activity is found in the parent compound. This
activity presumably arises from a secondary reentry process.
The various theories proposed to explain the reentry process will be
briefly considered.
A. ELASTIC-COLLISIOS
OR BILLIARD-BALL
MODEL
An elastic-collision model was proposed by Libby (45) and extended
by Miller, Gryder, and Dodson (57). The essential feature of this model
is that the recoil atom loses its energy predominantly in billiard-ball
collisions with the atoms of the surrounding molecules. If sufficient energy
is transferred to an atom in a given collision, the molecule containing the
struck atom will dissociate into free radicals. If the recoil atom has a
kinetic energy less than a critical amount after a collision, it will be
trapped in a reaction cage and react with a radical trapped with it;
otherwise it will escape from the cage. It is assumed that such processes
return the radioactive atom to a nonseparable form.
SZILARD-CHALMERS REACTION
IN SOLIDS
269
While this model can qualitatively explain the retentionS as the parent
compound in the alkyl halides it cannot account for the formation of
relatively large yields of radioactive products resulting from hydrogen
substitution and carbon-carbon bond rupture. For if the recoil halogen
atom can impart sufficient energy to a hydrogen atom to break the
carbon-hydrogen bond in an elastic collision, the halogen atom should
be left with sufficient energy after the collision to enable it to escape
from the reaction cage. These considerations led Libby to modify the
elastic-collision model.
B. ELASTIC-INELASTIC
COLLISION
MODEL
In the modification of the elastic-collision model Libby (29) still
maintained that retention as the parent alkyl halide results from the
replacement of an inactive halogen atom by the recoiling atom in an
elastic collision as described above. This process occurs while the recoil
atom is still moving with about 100 ev of kinetic energy and is called a
hot reaction. However, Libby now proposed that in a lower energy
region the recoil atom can undergo inelastic collisions with the alkyl
halide molecule as a whole, and this can lead to carbon-hydrogen and
carbon-carbon bond rupture. Such inelastic processes are supposed to lead
to the formation of products corresponding to hydrogen-substitution and
synthesis. These processes are probable in the 10-ev energy region and
are designated epithermal reactions.
This model is unable to account for the distribution of radioactive
products resulting from the reactions of recoil halogens in liquid hydrocarbon media (66). In these systems it is found that reactions leading
to retention as the parent compound occur in an energy region below that
leading to hydrogen substitution.
C. EPITHERMAL-REACTION
MODEL
In order to explain the behavior of recoil chlorine atoms in liquid
hydrocarbon media, Miller and Dodson (56) proposed that every recoil
atom forms an excited intermediate complex with the hydrocarbon
diluent, and that this complex then decomposes by a number of different
paths, leading to the various radioactive products. They also showed that
the concept of elastic atom-atom collisions could be entirely relinquished
3RetentionJJis a term widely used in the literature to denote either qualitatively, the appearance of radioactive atoms in a particular molecular species, or
quantitatively, the percentage of the radioactive atom in such a species, or loosely,
the sum total of all such combinations. Thus one speaks of retention of iodine
as ethyl iodide or organic retention of iodine in irradiated ethyl iodide. The
latter means the sum of all the organic species containing iodine such as ethyl and
methyl iodide, methylene iodide, etc.
270
GARMAN
I
-
AND NORMAN SUTIN
in interpreting their results and instead they suggested a more familiar

chemical approach. On this model the distribution of products is determined by competing chemical reactions in the epithermal region.
D. BRUSH-HEAP
OR RANDOM-FRAGMENTATION
MODEL
Willard (82) has also questioned the validity of the elastic-collision
model. He points out that since the molecules in a condensed phase are
not isolated, the assumption of elastic atom-atom collisions is not justified.
On the random-fragmentation model (82) the recoil atom loses its energy
by inelastic collisions with the surrounding molecules and breaks bonds in
rather indiscriminate fashion. When its energy has been reduced below
bond-breaking energies, it may combine either with the radicals in its
immediate vicinity, or with radicals it encounters in the course of its
diffusion in thermal equilibrium with the medium. While the appearance
of a wide variety of radioactive products, as found in the alkyl halides
(28, 29, 43, 48, 68, 7 2 ) , lends support to this model, their distribution is
generally not that expected from a random fragmentation of bonds.
E. HOT-ZONEMODEL
A model based upon the displacement spike concept of radiation
damage in solids has recently been proposed by the authors (37). An
essential feature of this model is that the energy of the recoil atom is dissipated in producing displaced atoms and in heating a small region of
the crystal (the (displacement spike). This hot zone eventually coinprises about lo00 atoms which remain above the melting temperature of
the crystal for about lo- sec; during this time chemical reactions have
a reasonable chance of occurring. Such reactions probably do not have
sufficient time to go to completion but are quenched by the cooling of the
hot zone. Subsequent heating (annealing) of the irradiated crystals may
give these reactions an opportunity to proceed further, and these effects
are generally termed thermal-annealing reactions.
The main respects in which the hot-zone model differs from the other
models previously described will be considered next.
The distance between successive collisions of the recoil atom on the
hot-zone model depends upon the energy of the recoil atom. On the
other hand the mean free path of the recoil atom on the billiard-ball model
is independent of the recoil energy, and is of the order of atomic radii,
Thus at 1000 ev the mean free path calculated on the hot-zone model is
about ten times larger than that given by the billiard-ball model. The
Iatter model accordingly gives a relatively higher temperature and longer
relaxation time of the hot zone than that given by the authors model. It
has been pointed out (89) that the hot zone would reach such high
SZILARD-CHALMERS REACTION I N SOLIDS
271
temperatures on the billiard-ball model that extensive decomposition of

polyatomic species would result.
The hot-zone model differs from the random-fragmentation model in
an important respect. The random-fragmentation model is based upon
free radical recombinations. The authors model suggests that in view of
the relatively low concentration of free radicals present in the hot zone,
chemical reactions with the parent species may also play an important part. The picture may be further complicated by ion-molecule
reactions (71).
The concept of chemical reactions during the lifetime of the hot zone,
which is an important feature of the authors model, is similar to the
epithermal-reaction concept proposed for liquid systems by Miller and
Dodson. However, the latter authors did not attempt to estimate such
parameters as the size, temperature, and lifetime of the hot zone.
In the next section we will consider some recent results of SzilardChalmers studies against the background of the various models and in
the third section we will treat annealing phenomena in some detail.
II. Chemical Effects of Nuclear Recoil
A. DIFFICULTIES
IN THE INTERPRETATION
OF EXPERIMENTS
Although hot-atom effects have been observed in many systems, most
of these appear in isolated experiments, frequently for the sole purpose
of preparing isotopes of high specific activity. Some few systems have,
however, been rather more extensively investigated with a view to understanding the fundamental processes involved. For the purposes of this
discussion, we shall consider two broad classes: ionic crystals and molecular crystals. The application of any model to predict accurately the hotatom chemistry of a particular compound would require a detailed
knowledge of such factors as capture gamma-ray spectra, internal conversion coefficients, bond energies, intermolecular forces, crystal structure,
orientation and mobility of molecules, thermal diffusion coefficients,
energy transfer mechanisms, activation energies and entropies, and will
not be attempted here.
In interpreting the experimental results, a major difficulty arises. This
is the question of the possible alteration of the original distribution of
radioactive atoms among the various species present in the crystals by
the particular method of analysis employed. This difficulty may be
illustrated by considering potassium chromate. If chromium recoil leads to
species such as Cr04=, CrOs, Cr022+, Cr04+, and Cre+ in the crystal
($I), dissolution of the sample in water will result in the production of
chromate by all but the last two, at least, since chromic anhydride and
272
GARMAN HAREUYITLE AND NORMAN SUTIN
chromyl ion are known to hydrolyze to chromate (or dichromate) under

all conditions of pH. Howevcr, one can only guess at the reactions of
Cr04+ and Cra+ in water. Similar problems of interpretation arise in the
other oxyanions.
Attempts have been made by several investigators to limit these
difficulties. In the arsenic (52) and antimony triphenyls (SS), chromatographic separations were employed; in the cobaltic trisethylenediamine
(91) experiments alternative fractionation methods were used, while in
the cobalticyanide (66) and chloriridate (22) cases paper electrophoresis
was found useful in separating a number of species produced by dissolution in various media. I n the organic halides (4.8)and organic (carbon-14)
compounds (70) carriers for many possible products were added and
then fractionated. The same technique has recently been applied in
studies of potaasium periodate (5),in which iodide, iodate, and periodate
fractions were separated. However, it is difficult to develop completely
reliable analytical procedures; and, in general, inferences as to the
original distribution of the radioactive species are drawn from the
observed data and the assumed reactions during analysis of hypothetical
entities present in the solid.
B. IONICCRYSTALS
If we consider under this classification the broad group of oxyanions,
we find that there are certain general trends in the data (Table I).
Although there are seen to be substantial differences among varidus
metallic salts of a given anion (for example KMn04 and NaMn04), it
appears that there is an inherent tendency of permanganate retentions to
fall around lo%, chromates about 70-&0%, iodates roughly the same,
bromates about lo%, chlorates the same or lower, phosphates about
50-60%, perrhenate lOO%, and arsenates fairly high. Where lower oxidation states of the same elements have been studied, it has been found that
phosphites ( 6 ) and arsenites (44) produce predominantly the parent form
with a little of the oxidized (17% and lo%, respectively) state, while
rhenium alone shows a great tendency toward the production of oxidized
recoils. However, recent experiments of Maddock and de Maine (61)
on Ah03 and Sb203 showed that within experimental error none of the
higher oxidation states was produced on recoil.
Libby (44) has proposed an explanation of these data. He suggests
that the distribution of the bonding electrons between the recoiling atom
and its attached oxygens will be the same as that normally existing in the
covalent bond. Depending on the relative electronegativities of the
capturing atom and oxygen, either oxygen atoms or oxide ions will be
lost in the recoil. In the former case the oxidation number of the recoil-
273
SZILARD-CHALMERS BEACTION IN SOLIDS
TABLE I
RETENTIONS
IN IRRADIATED
CRYSTALLINE
OXYANION
COMPOUNDS
Compound
Recoil
atom
Retention
(%I
Reference
Mn66
8.8
9.1
22.5
8.0
6.0
12.8
47
47
47
47
47
47
66
54.5
73.6
79.9
60.8
89.9
69
17.5
31.9
55.3
34.6
56
56
1148
Bra2
Cl38
cia4
Re186188
Pa2
66
67
67
86103- 410453
60
22
10
9
12
10
1.5
9
56
36
56
56
36
36
20
20
20
6
20
$0
90
41
21
41
41
46
46
100
39
58
45
6
6
6
50
A2376
36
36
36
75-90
76
60
90
76
76
274
ing atom will be lower than its value in the parent molecufe, and a
successful Szilard-Chalmers separation should be possible. When oxide
ions are lost, a successful Szilard-Chalmers separation will be possible
provided the recoil species rare reduced by water more rapidly than they
undergo hydration reactions.
These ideas have been extended by Green, Harbottle, and Maddock
( I S ) to cases where oxide ions are lost. They supposed that the important reaction of the recoil species a t the end of its track was the loss of
one or more oxide ions, or conversely, incomplete readdition of oxide ions
to the recoil atom. Then each partially-reconstituted species (for example,
Mn04-, hln03+, M n 0 2 + ++, etc.) will suffer either hydration (which will
reconstitute the parent) or reduction on dissolution in water. The argument runs that the greater the oxidation potential of the parent anion,
the sooner will a species be reached (through successive loss of oxide ions)
which will inevitably oxidize water on dissolution, and consequently the
lower the retention. This rough relationship is indicated by Fig. 1 in
-02
-0.4
-0.6
-0 8
-1.0
-I 2
-1.4
-1.6
STANDARD POTENTIAL VOLTS
FIG. 1 . Relation between retention in oxyanion salts and oxidation potential.

Point I , permanganates; 2, bromates; 3, chromates; 4 , chlorate; 5 , iodate; 6 .
periodate; 7, pewhenate; 8. arsenxte
which typical retentions are plotted against oxidation potentials. One

would predict from this that vanadates( Eo ,- - 1.0 volt), selenates
i E o - 1.2 volts) and sulfates ( E , , - 0.2 volt) would all have fairly
high (> 50%) retentions.
The foregoing considerations sliow that i t is possible to explain many
of the observations on the oxyanions in ternis of recombination reactions
between the recoil atom and oxygen atoms or ions produced in collisions.
275
SZILARD-CHALMERS REACTION IN SOLIDS
Although the activation energy for the recombination of fragments is

fairly low, the fragments may not be nearest neighbors and the activation
energy for diffusion may be a few electron volts (80).
It is also possible that the recoil atom could gain one or more
oxygens in an activated exchange with a neighboring oxyanion, during the
lifetime of the hot zone. I n the authors' paper cited above (37), equations
were given which permit an order-of-magnitude calculation of the
probability of this process. If one assumes an activation energy of 3 ev for
oxygen transfer, in a hot zone of about 50 ev, the results indicate that
oxygen transfer could occur several times before the hot zone cools down.
The theory that the oxygenated recoil species are formed in reactions
between recoil atoms and surrounding oxyanions or dislodged oxygens is
supported by the results of mixed crystal experiments. Mixed crystals
(solid solutions) of potassium chromate in potassium sulfate (35) and
fluoberyllate ( 5 0 ) ,and potassium permanganate in potassium perchlorate
(67) have been investigated. The results are given in Table 11. These
TABLE I1
RETENTIONS
IN MIXEDCRYSTAL
SYSTEMS
Recoil
source
Diluent
K&d4
K&d4
KzCd,
K&d4
KzCrO4
KMn04
KMnOl
KMnOl
KzSO4
KzSOi
K2BeF4
K2BeF4
None
KCIO4
Kclo4
None
Source molecule Retention

(Mole %)
(%)
Reference
2.77
0.49
1.56
-1.
100.
40.
1.
100.
69
70
25
26
61
49
62
22
35
35
35
50
35
67
67
67
data indicate that whereas recoil chromium in a sulfate crystal can find
an oxyanion or debris oxide ions with which to react, producing a high
retention, the same thing cannot happen in the fluoberyllate crystal. It
is conceivable that the retention observed here (25%) indicates the
probability that the recoil chromium does not move far enough away
from its original site to avoid a short-term annealing in the hot zone.
The thermal annealing of these mixed crystals, described by Maddock and
de Maine (50), bears out this supposition, and these results will be
discussed in Section 111. One would predict that the activation of
chromium in fluoberyllate crystals by fast neutrons would lead to still
lower retentions, of the order of a few per cent, and much lower annealing
rates. The increased pennanganate retention in the mixed crystals
276
GARMAN HARBOlTLE AND NOBMAN SUTIN
perhaps is due to oxidation of the recoil species in displacement spike

reactions. According to the Libby hypothesis fragmentation of C1-0
bonds will liberate oxygen atoms while the Mn-0 bonds will liberate oxide
ions. In the mixed crystals the recoil species will find themselves in an
environment of relatively higher oxidizing power. Similar considerations
could account for the higher retention in chromate-sulfate mixed crystals.
The results with chromate-sulfate and permanganate-perchlorate mixed
crystals are in sharp disagreement with a billiard-ball model, which would
predict the highest retention in the pure chromate or permanganate
crystal.
The oxidation of the recoil species in the mixed crystals may be
regarded as a specific chemical effect. Similarly, the anomalously low
retentions observed in the two salts, ammonium iodate (22%) (20)and
ammonium chromate (17.5%) (55) are in striking contrast with those of
the corresponding alkali metals, and suggest that in the hot zone ammonium ion may act as a reducing agent, which the alkali metals cannot
do because of their high ionization potential. There is also some evidence
that water of crystallization may act as a reducing agent in the permanganate salts ( 4 7 ) .
Oxidation reactions in the hot zone are probably responsible for the
large yield of Cr(V1) obtained in the positron decay of Mn51 to Cr5I in
crystals of cesium permanganate ( 1 1 ) . Manganous ion in solution yielded
only chromic ion, whereas in crystals of MnC03, 3070 of the chromium
recoils appeared as Cr(V1). This probably occurred as a result of
reactions between chromium recoils and debris oxygen atoms or carbonate
ions.
Further evidence of specific chemical effects in the hot zone appears
in the n-ork of Aten, who investigated the distribution of oxidation states
of P32 formed by the n,p reaction in irradiated inorganic sulfur compounds (3). When he irradiated solid potassium sulfate or sodium sulfate
decahydrate, only a few per cent of the recoils were in lower (nonphosphate) oxidation states. With sodium sulfite (hydrated) the percentage rose to about SO%, while for sodium sulfide 65% to 95% were
in lower oxidation states. He found similar correlations when the P32was
produced by the C135 ( n p ) process. Here, the irradiation of KC10,
or KC10, produced 99% phosphate, while NaCl gave only 35%.
Within the category of ionic crystals, there are two similar complex
ions of cobalt (111) which have been extensively studied: cobaltic
trisethylenediamine nitrate (91) and potassium cobalticyanide (66).
Both these crystals show fairly low inherent retentions (approximately
570 and 1970,respectively) and both give evidence of fragments having
277
unfilled coordination spheres present after recoil. Both show typical

thermal- and radiation-induced annealing reactions. In the cobaltic
lrisethylenediamine studies Zuber irradiated the dextrorotatory complex,
added levorotatory carrier and then resolved the mixture by standard
methods. He found that in the irradiated d crystal, of the 57% retention
4.5% was in the d form and only 0.5% in the 2 form. This extremely interesting result may indicate that the displacement spike zone remains
hot long enough for the reconstituted molecule to assume the correct
stereoisomeric configuration. The fact that such a decided preference
is shown for the parent d form also sets a limit upon the amount of
damage that could have been done in those displacement spikes which
lead to retention. This observation is also consistent with a billiard-ball
model, where one would expect that the original configuration would be
preserved. It would be of great interest to repeat Zubers experiment with
the d complex, activating the cobalt with fast neutrons to produce CoS8
recoils of high energy, giving larger displacement spikes.
Mention should be made of another type of recoil experiment with
ionic crystals, namely the production of S35and P32in the neutron irradiation of alkali chlorides (42, 13,23). Here the recoil atoms are generated
through heavy-particle emission from the compound nucleus, and recoil
energies are quite high (approximately 15 and 50 kev, respectively). The
proportions of the different oxidation states of sulfur obtained are apparently not very reproducible and whereas one investigator has found
the results to depend on pretreatment of the crystals [degassing (&)I
another does not confirm this (23).The main difference between the two
sets of experiments lay in the temperature at the irradiation site and the
quality and intensity of the concomitant fast neutron and gamma
radiation. The proportion of the different oxidation states of P32formed
in recoil in KC1 also appeared to depend on degassing before bombardment. Here the effect of concomitant and subsequent ioniaing radiation
was examined and found to be very marked. In the light of recent
observations on thermal and radiation annealing and the temperature
dependence of the latter, it would be desirable to repeat these experiments.
Summarizing the observations on the ionic crystals it may be said that
there is abundant evidence that reactions in the hot zone play an important part. Thus, any theory, such as the elastic-collision model which
neglects specific chemical effects, e.g., reduction by NH4+ and HzO, or
oxidation by c104-, under the influence of the high local temperature,
cannot give a complete explanation of the data. In none of the studies of
the hot-atom chemistry of a series of oxyanion salts, e.g., the permanganates, have correlations been established between the retention and
278
GARMAN HARBOTTLE A N D NORMAN SUTIN
crystal properties such as the density or atomic weight of the cation. These
factors should have a marked effect on the efficiency of energy transfer
in a billiard-ball process, and on the cage energy.
C. MOLECULAR
CRYSTALS
Recent calculations of the authors (37) have shown that about 16 displacements will be produced in the slowing down process in molecular
crystals, and that the volume raised above the melting temperature
includes about 4000 atoms. A similar volume for the hot zone can be
obtained from a consideration of the latent heat of fusion. For a recoil
energy of 300 ev, and a bond energy of 4 ev, about 240 ev will be left for
melting the crystal. Taking a latent heat of fusion of 0.1 ev, this corresponds to a final molten zone of about 2400 molecules.
The simplest molecular crystal which has been studied is sulfur.
Xlsson investigated the distribution of P32, P33 and S33 in neutrouirradiated sulfur (61). The first two isotopes are formed by n,p reactions
and have energies of recoil of 45 and 15 kv, respectively. The third is
formed by neutron capture and would have a much lower recoil energy, of
the order of a hundred electron volts or possibly even smaller. Nilssons
results give the most direct evidence yet obtained for a molten-zone displacement spike surrounding high-energy recoils in molecular crystals, for
he found that a large amount of the P32and P33, but hardly any S3j
followed the ihsoluble Sp fraction of the sulfur samples on dissolving in
carbon disulfide. Since the S p modification of sulfur is formed by melting
the orthorhombic crystals, these results are strong evidence for a molten
displacement spike engendered by a high-energy recoil.
The molecular crystals which have been most extensively studied are
those in which the recoil atom is initially organically bound, for example,
the alkyl halides (28,30,43,48,72), and organometallic compounds such
as ferrocene (77) and triphenylarsine (52). The complexity of the hot
reactions in which several different types of radicals are present is well
illustrated by the alkyl halides. Products corresponding to the isomeriaation, polymerization, substitution, and degradation of the parent alkyl
radical are formed in good yield. For example, the recent work of Evans
and Willard (26), in which gas chromatographic techniques were used to
separate the various radioactive products, has shown that no less than 16
different species containing radioactive brominc are present following
neutron irradiation of n-propyl bromide. Although this particular work
was concerned with a liquid phase, other experiments have shown that
recoil reactions in solid alkyl halides are equally complex (48).
A recent investigation of the radiolysis of alkyl iodides in the crystalline and glassy states has revealed differences in the yields of products
2 79
SZILARD-CHALMERS REACTION I N SC&lUS
which were ascribed to the differing orientation of the molecules in the

two states (40). Such effects are included in the hot-zone model through
an entropy of activation term in the calculation of reaction probability.
Similar differences have been found in the Szilard-Chalmers reaction in
glassy and crystalline propyl bromide (68).
A comparison of the results of investigations on alkyl halides and
organo-metallic compounds suggests that the yield of radioactive parent
compound obtained in Szilard-Chalmers investigations may be 'related to
TABLE I11
R.ELATIONOF RETENTION
TO NUMBER
OF GROUPS
COVALENTLY
BONDED
TO THE CAPTURING
ATOM
Number of groups covalently

bonded to the capturing atom
in the parent molecule
Yield of
radioactive
parent
molecule (%)
Reference
one
Carbon tetrabromide
Carbon tetrachloride
Methyl iodide
Ethyl iodide
n-Propyl iodide
Impropyl iodide
n-Butyl iodide
Sec-butyl iodide
n-Propyl bromide
Isopropyl bromide
Tritium in glucose and galactose
Tritium in benzoic acid
Two
Ferrocene
Three
Triphenylarsine
Triphenylstibine
Four
Carbon-14 in orgarlie compounds
Cobalt phthalocyanine
93
51
44
27
21.7
39.2
2 1.
13.9
26.9
11.7
-12
25
12
SO
so
@
48
@
48
48
48
@
48
69
88
Y7
2.1
4.4
-1-3.5
-3
85, 86, 87
16
the extent of covalent bonding of the capturing atom. This is shown in

Table I11 which refen to irradiations in the solid phase. The dependence
of yield upon number of covalent bonds shown in this table may be
rationalized by reference to the energy and entropy of activation terms in
the hotAzone reaction probability expressions mentioned above. The
280
GARMAN HARBOTIZE AND NORMAN SUTIN
greater the number of bonds, the larger the activation energy and the less
favorable the steric factor.
In all the alkyl halides investigated the organic yield in the solid phase
is greater than 30% with the exception of secondary butyl iodide for
which it is 25%. The excited molecule of this compound is known to be
unstable with respect to decomposition into the olefin and hydrogen
iodide,.and this could account for the lower organic yield.
In view of the large number of reactions which can occur in organic
systems, the chemical nature of the environment is much more important
than in the case of the oxyanions. Indeed Willard has concluded that in
the alkyl halides, the chemical characteristics of the medium are more
important than physical factors such as the energy, charge, strength of
cage walls, and masses of the atoms involved (8g).
A recent investigation (48) of the chemical effects of the n,2n activation of iodine in alkyl iodides indicates that molecular structure plays an
important part in determining the yield of minor products. Results with
n-propyl and isopropyl iodides show that the yield of methyl iodide is
larger in the irradiation of the former compound. A random-fragmentation model would predict a larger yield of methyl radicals from the
isopropyl iodide. In several other systems yields are not consistent with
a random-fragmentation model. A possible explanation for these observations on the basis of the hot-zone model is that before the hot zone cools
the recoil atom can undergo reactions which require moderate activation
energies.
The ratio of the organic yield in the solid phase to that in the liquid
has been measured for a large number of organic halides. The ratios are
generally greater than unity and bear no apparent relation to the structure
of the halide or the density change on freezing (82). This may be ascribed
to the different steric factors, stronger reaction cage, and the longer lifetime of the hot zone in molecular crystals. It may be possible to account
for those few cases in which the organic yield is larger in the liquid phase
on the basis of a relatively larger contribution from thermal reactions.
In liquid systems the thermalized atoms have an opportunity to diffuse
freely through the medium and enter into the more familiar chemical
reactions ( 5 8 ) .A useful technique has been developed to determine which
products are formed in hot reactions (34, 84). Before the irradiation of
the organic halide, low concentrations of substances known to react
readily with thermalized halogen atoms are added. The part of the organic
yield which is insensitive to added halogen is attributed to hot reactions.
In this way it lias been found that both the organic and inorganic yields
are formed by hot as well as thermal reactions. Such a distinction between
hot and thermal reactions applies equally to all the models discussed.
281
The liquid alkyl chlorides exhibit several interesting properties. In

contrast to the alkyl bromides and iodides, t,he alkyl chlorides show no
scavenger effect ( 1 5 ) . This is ascribed by Willard to the ease with which
chlorine atoms react with carbon-hydrogen bonds to form stable hydrogen
chloride. The fate of the radioactive chlorine atom is thus determined
before it comes to thermal equilibrium with the medium. Willard further
points out that the organic yield of the chlorides is independent of chain
length and thus of the ratio of carbon-chlorine bonds to other bonds (82).
On the other hand, such structurally similar molecules as C2HaCl,
C2H5Br and C Q Hhave
~ organic yields of 2176, 32%, and 41%, respectively. The random-fragmentation and billiard-ball models predict
that all three compounds should have similar organic yields. This supports
Williards conclusion that the chemical characteristics of the medium are
more important than physical factors in these systems and finds a ready
explanation in the hot-zone model. As mentioned in Section I a similar
conclusion had been reached by Miller and Dodson (56) on the basis of
experiments with recoil chlorine atoms in hydrocarbon media.
111. Post-Recoil Annealing Effects
A. INTRODUCTION
The annealing of recoil atoms was first reported by Williams (83),
who noticed that the specific activity of .radioactive antimony produced
by the pile irradiation of antimony pentafluoride and ammonium fluoantimonate decreased with increasing time of irradiation. This effect
was ascribed to two phenomena: the gradual return of initially separable antimony atoms t o t b ; parent form during irradiation, and the simultaneous macroscop.. decomposition of the compounds. It was also
observed that the retention as the parent compound in these two substances w a ~increased by exposing them to ionizing radiation following
neutron irradiation.
Green and Maddock (32) subsequently found that the retention of
chromium-51 in potassium chromate could be increased by heating the
neutron-irradiated crystals. Thermal and radiation annealing have since
been observed in a number of systems ( 4 , 5, 9,20, 21, 31, 35, 39, 41, 50,
52, 53, 54, 60, 66, 67, 74, 77,91).In general, annealing returns a part of
the initially separable recoil atoms to the form of the parent compound.
The annealing of Szilard-Chalmers recoil atoms finds a striking parallel in the annealing of radiation damage produced in solids by neutron,
charged-particle, or photon irradiation ( 2 5 ) .In both cases one is dealing
with solids which have been altered by the production of defects in a
matrix of otherwise normal crystal. In both cases the defects are the
282
result of a particular atom having been given a sudden impulse, sufficient

to displace it from its normal position in the lattice. The Szilard-Chalmers
method consists of observing the defects through the change in the chemical state of the atom responsible for the defects, while radiation-damage
studies have usually depended upon changes in the optical absorption,
resistivity, Youngs modulus, or semiconducting properties of the irradiated solids. One advantage of the Sailard-Chalmers method is that the
number of atoms annealed in a particular experiment is immediately
given by the result of the experiment, whereas in radiation-damage annealing studies a hypothesis must be made to connect the change in the
physical property to the change in the number of atoms in the defect
state. A second useful consequence of a Szilard-Chalmers investigation
is that it may give directly the fraction of atoms which have suffered
displacement and then have been reincorporated into normal lattice sites.
Physical methods yield no direct measure of this quantity. By using the
appropriate nuclear reaction one can often cause the recoil atom to be
projected with either very high (-lo5 ev) or very low (@lo2 ev) kinetic
energies. Another important difference in the chemical and physical approaches is the relatively greater sensitivity of the former. In typical
Szilard-Chalmers studies only very small doses of radiation need be
given. In favorable cases, integrated thermal neutron fluxes of 1OO neutrons/cm2 suffice, and it is likely that the defects or displacement spikes
produced by recoil atoms under these conditions are isolated events. In
typical solid-state studies of annealing of radiation damage by measurement of X-ray diffraction, electrical resistivity, or Youngs modulus, significantly greater damage must be done to the system to produce measurable effects. An exception is the study of semiconductors, whose electrical properties are very sensitive to the introduction of imperfections.
In a sense, Szilard-Chalmers annealing studies are complementary to
those of radiation damage annealing, for while the former are concerned
with fairly complex crystals, containing several types of atoms in different molecular or ionic combinations, the latter have generally been
concerned with simpler substances (e.g., monatomic metals, alkali halides,
diamond, graphite, silica, etc.). Thus it is necessary to study the damage
in the former systems from a fundamentally chemical viewpoint, taking
into account breaking and re-formation of bonds, synthesis of new
chemical entities, and changes in electronic structure ; while a more
physical approach, invoking a mechanical, hard-sphere electrostatic
model of the displacement (and annealing) processes has generally been
adopted in dealing with the latter systems. Solid-state physical studies
have largely ignored the influence of the chemical bond.
283
B. THEINTERPRETATION
OF THERMAL
ANNEALING
The recombination of pairs of interstitials and vacancies has usually
formed the basis of models for thermal annealing. It has sometimes been
assumed that the vacancy is mobile (as in the vacancy model for selfdiffusion) and sometimes that the vacancy is a fixed trap or sink, having
a capacity of one interstitial atom, and the interstitial is mobile. The
form of the kinetic equations is usually not dependent on the choice, and
this is true even if both are mobile provided there is no initial correlation
of interstitials and vacancies.
However, in typical cases of Szilard-Chalmers annealing, a fundamental asymmetry appears. If we consider a recoil bromine atom in crystalline potassium bromate, the atom may come to rest in an interstitial
position, but the conjugate vacancy is then an array of unbonded
oxygen atoms or oxide ions. To assume recombination by vacancy diffusion might imply that these move through the crystal as a unit, which
is difficult to accept. For our purposes a vacancy in a complex crystal
may be regarded as a trap consisting of a fragment or any species capable of reacting chemically with the interstitial atom. In some cases this
might consist of an oxygen atom or oxide ion, and these simple entities
would of course be capable of diffusion. In others the trap might be an
F center, electron deficient hole, etc.
It is convenient to make a broad distinction between models on the
basis of the initial separations of the species which undergo reaction on
annealing. If, as is usually the case with atoms activated by n,y reactions, the range of the recoil atom is short [of the order of a few lattice
spacings (90)],
there will be a high degree of correlation between interstitials and vacancies, or, to put it another way, the radioactive recoil
atoms and the fragments with which they combine will have small initial
separations. As annealing proceeds reactions may take place between
species which were initially further apart. We will also consider a phenomenological approach by which mixed kinetics may be handled and
schemes in which the activation energy is permitted to vary and the
kinetics are not specified.
1. Annealing of Initially Correlated Pairs
a. Simple Recombination of Correlated Pairs. As the simplest case

we will consider that the annealing process consists of the recombination
of the recoil atom with a collision fragment or with one of its originally
bonded partners and that the pairs have small initial separations. Such
a process in a simple monatomic crystal has been visualized by Brown
284
GABMAN HARBOITLE AND NORMAN SUTIN
et al. (10) in the following way (similar arguments probably apply to

more complex systems). Close to the vacancy there is elastic deformation of the lattice which results in a lowering of the potential barrier to
the jumping of the interstitial atom from one interstitial position to the
next as it approaches the vacancy. The jump toward the vacancy will
have the lowest potential barrier, and the barrier will be lower the closer
the interstitial is to the vacancy. Since the initial separation of the interstitial and vacancy is small the rate of recombination will be determined by the jump of the interstitial from its original site in the deformed region (which may be approximately identified with the displacement spike zone). If this jump frequency is vj, then
(1)
where v0 is a typical frequency for the vibration of atoms in the lattice
and E is the energy required by the interstitial to jump to an adjacent
interstitial site. Brown et d.state that v0 lOls see-I. Although thie is
a normal crystal frequency, it is worth noting that ions or atoms close
to vacancies do not have normal vibration frequencies. In particular,
interstitial atoms as Frenkel defects have frequencies greater than those
of the normal lattice (76).
The rate of recombination will then be given by
y1
= v,g-ElkT
-dn/dt = v;n
(2)
where n is the number of uncombined pairs correlated as described above.
This equation leads to the simple first-order law that
nf =
w-
(3)
in which nt and no are the number of uncombined pairs a t times t and
zero, respectively.
Provided the annealing process described above leads to retention
(the re-formation of the parent compound, or a species which produces
the parent compound on chemical analysis), Eq. (3) can be expressed
in terms of retentions as follows
In ( R ,
V I ~
- R t ) = In ( R , - Ro) - v,t
(4)
in which Rt and R , are the retentions at time t and infinite time of isothermal annealing, respectively. The quantity (Rm - R t ) / ( R m- Ro) is
the fraction of pairs still uncombined a t time t . These quantities are
shown graphically on the generalized annealing curve in Fig. 2.
In order to simplify the equations we will consider only the increase
in the number of recombined atoms on annealing. Although as we have
seen in Section I1 the initial retention ( R o ) is usually not zero, and may
OF
NUMBERS
F TS
1
rb
TIME OF ISOTHERMAL ANNEALING
FIG.2. Generalized annealing curve illustrating the quantities Rr, Ro,R 00 and
the numbers of uncombined defects n
be quite large, it may be taken as the base line from which annealing
proceeds. This procedure is equivalent to taking Ro = 0 (Fig. 2) and
does not alter the form of the kinetic expressions derived.
In the event that there are several classes of fragment loci in the
same, or different deformed zones, having different activation energies
for annealing E l , E2, . . . then
+ .. .
(5)
where no = nol no2
. . . and vjt is the jump frequency of the atom
in the zxb locus from its initial position. For each vjt there will be a characteristic activation energy Ec see Eq. (1) , and the over-all approach of
the quantity (R, - R t ) t o zero will then follow a succession of exponential periods analogous to radioactive decay of a mixture of isotopes of
differing half-lives.
If the annealing process is of this nature, and there is a finite number of types of recombination loci, then in principle the annealing curve
can be resolved into that many exponential components. It is clear that
by this procedure any arbitrary monotonic experimental curve can be
fitted provided a sufliciently large number of exponential terms, Eq. ( 5 ) ,
are taken; however, a necessary condition is that extrapolation of the
linear portions of the curves thus resolved for different temperatures
must give the same intercept values (at t = 0 ) , since the quantities nol,
no2, . . . are not affected by the temperature of annealing.
b. &ecmbination of Correlated Pairs Modified b y Electrostatic Interactions. An interesting model which involves diffusion of interstitial
atoms in the neighborhood of a correlated defect has been proposed by
Maddock and de Maine (49, 60). Although they first treated their own
nt =
+ +
Qle-vl
286
GARMAN HARBOTTLE h X D NORMAN SUTIN
data on potassium chromate by this model, it has been found possible

to apply it with some success to other systems as well. I n the case of
potassium chromate, the basic annealing reaction is assumed t o be
c'ro'+ + 0- -+ cro**+
(6)
and the oxide ion is assumed to be mobile. After irradiation but before
annealing, the distribution of Cr04+ - O= distances ( X ) in the crystal
is assumed to be as shown in Fig. 3. There are no distances less than X o ,
X.
FIQ.3. Distribution of CrO"

thc Maddock-& Maine model.
X(THEcr04+- o= DISTANCE)
- 0: distances in
xw
irradiated potassium chromate on
because partners having this distance of separation are so close that they
recombine even a t room temperature, and thus add to the Ro term in the
retention (see Fig. 2). On the other hand there are none further separated than X,, a distance equal to the maximum range of the recoil atoms.
Between these two distances the distribution is assumed to be uniform.
The rate of recombination of partners is given by
-dn/dt = nye- i 7 l k T e V ~ k T S
(7)
where n is the number of possible pairs, U is the activation energy barrier opposing recombination, Y is the frequency of vibration of the defect
oxide ion, T is the absolute temperature, X is the smallest separation of
partners a t time t , and V is an electrostatic potential term containing, in
this case, the charge on the ions, dielectric constant, and crystal parameters. The quantity V / X is a potential gradient. Thus the kinetic expression is similar to the first case considered, Eq. (1) and ( 2 ) , modified
by the presence of a field term, in this case the coulombic attraction of
oppositely-charged ions, which tends to assist recombination.
The annealing process is considered analogous to the growth of very
thin oxide films treated by Mott ( I d , 5 9 ) . On this picture the increase in
287
retention on annealing involves the recombination of partners having

increasingly greater values of X ; this may be visualized in Fig. 3 as the
slow movement of the vertical line at X , towards X,, the area lying to
the left of the line representing the pairs that already have recombined
and the area to the right those still uncombined at time t. Thus the
change in retention AR is a linear function of X and enters the rate
expression as follows:
dR = , - U / k T e V / c k T ( A R f S )
(R, - R ) dt
where S is a constant and c is a constant relating the retention to X.
When annealing is begun AR is small and the second exponential
term in Eq. (8) is large and dominates the rate of reaction. This corresponds to the rapid initial growth of thin films of oxide. However, as
annealing proceeds the second exponential term becomes effectively unity
and the rate of increase in retention is then governed by the first exponential. This corresponds to the slowly-rising plateau portion of the
annealing curve (50,53,54).Maddock and de Maine take the value AR
(equal to A R ~ )for which the second exponential term becomes effectively unity from the expression
V
(9)
ckT(AR, S) =
where CY is a small number such that ea is arbitrarily close to unity. This
state of affairs corresponds to the formation of an oxide film of limiting
thickness, at which point rapid growth ceases and further growth becomes very slow (12,5 9 ) .
We then see that values of AR, characteristic of the plateaus a t different temperatures T satisfy the relation
(ARC S ) = C/T
or
AR, = C/T where C is an arbitrary constant, equaling V/cka.
This important result, Eq. (lo), states that on the Maddock-de Maine
model the increase in retention on annealing to the plateau should be a
linear function of 1/T, and such a relation is in fact observed in a number of cases (50, 53, 54). We shall discuss this further in the section on
relevant experimental data.
c. Recombination of Correlated Pairs with Electron Transfer as the
Rate-Determining Step. In many Szilard-Chalmers reactions the recoil
fragment acquires a high positive charge as a result of internal conversion effects. The partial or complete neutralization of this
288
GARMAN HABBOITm AND NORMAN SUTI19
charge may be necessary before the recoil fragment can recombine with
other fragments on annealing, and this electron transfer process may be
the rate-determining step. In such systems the fragments which re-form
the parent species may be trapped Bufficiently close to one another for
recombination to occur once the recoil fragment has gained the required
number of electrons from its surroundings. (Alternatively the parent
species might be re-formed from the reduced recoil fragment during the
subsequent chemical analysis.)
Although such a mechanism could very well account for the initial
retention (Ro in Fig. 2), it is difficult to see how the apparent energies
of activation for the annealing process can be reconciled with this view.
There are numerous examples of annealing reactions which do not proceed at all, or very slowly, at 0C but which go rapidly a t temperatures
of a few hundred degrees centigrade. Such behavior implies, if frequency
factors are normal, activation energies of the order of magnitude 1 to
2 ev. This is the order of magnitude of the activation energy for diffusion of ions in crystalline ionic lattices but seems too large for electron
transfer. However, it should be noted that Pringsheim et al. (14) found
that the bleaching of F centers in LiF required temperatures of 100200C, corresponding to reasonably large activation energies for this
particular type of electronic effect.
d. Recombination of Correlated Pairs after a Random- Walk Process.
For the interstitials which are located outside the deformed region the
probability of jumping in any direction is the same. These interstitials
wander according to a random-walk process; those that wander back
into the deformed region are subsequently captured by their original
partners, and the rest wander off through the crystal.
Fletcher, Brown et al. (10, d 7 ) have applied this model to the annealing of radiation damage in germanium. The deformed region is considered to be a trap or sink of radius a such that the approach of a
mobile entity to within the distance a inevitably produces recombination.
The medium is assumed to be infinite and isotropic. All fragment pairs
are correlated and have identical initial separations b ; the number of
uncombined pairs nt still present after a random walk of duration t is
related to the number present at the beginning, no, by
nt
(b [ 3 + ;erf (2(Dt)1/2
)]
U)
no 1 -
where D is the diffusion coefficient of the diffusing entity. The variation

in D with temperature will involve the activation energy E in the usual
way
= D,,e-E/kT
(12)
289
where Do is a constant, roughly independent of temperature. The relation

between the diffusion coefficient and the corresponding jump frequency
vj is given by the well-known expression (73, 7b)
D = 12vj
=
$2v*-EfkT
(13)
where vo is defined as in Section Bla above and 1 is a lattice constant.

In terns of retentions the Fletcher-Brown expression, Eq. (11), becomes
(14)
again disregarding the initial retention.

This equation has several interesting features. First, it produces a
plateau after long annealing times, the corresponding retention being
given by R = u/b. This is the liberation phenomenon, i.e., (1 - u/b)
is the fraction of pairs liberated, or not eventually recombined with
their initial partners. As Fletcher and Brown point out, liberated diffusing entities can still combine with traps in other parts of the lattice.
This will be governed by bimolecular kinetics, if the number of diffusing
entities and traps are comparable (see below). To illustrate the FIetcherBrown expression, we have plotted .in Fig. 4 the quantity (1 - erf
FIG.4. The Fletcher-Brown random-walk error function expression plotted against

Dt/a*, a function of the time of isothermal annealing.
[a/2 ( D t )lI2] } against D t/u2; this is equivalent to arbitrarily taking

b /u = 2. The abscissae of Fig. 4 are then proportional to the time of
annealing and the five curves could represent one particular curve on
five different time scales differing from one another by factors of 10.
290
The second interesting feature of the Fletcher-Brown expression is clearly

seen on the second of these curves, and that is the point of inflection near
the origin,
Although we could consider the five curves as rcprcscnting the annealing behavior on five different time scales for one value of D / a 2 , that
is to say, one temperature, we could equally well consider that all five
relate to one time scale, but differ in having five values of D/a2, that is
to say, five values of D ,the diffusion coefficient, differing by factors of
10. The family of curves then bear considerable similarity to the sets of
isothermal annealing curves obtained in typical Szilard-Chalmers studies,
except for the point of inflection near the origin, which has not been
observed experimentally. It is worth noting that the Fletcher-Brown
plateaus are all a t the same value of R , namely a / b and independent
of temperature. Even if we take the point a t which the rapidly rising
portion begins to level off, as we did in the case of the Maddock-de Maine
model, we do not obtain the linear relation between AR and 1/T observed
experimentally.
e. Recrystallization of the Hot Zone. If, as we notcd above, the hot
zone on cooling contains disorder of various types (i.e., broken bonds,
free radicals, interstitial atoms, and vacancies) we could regard the annealing process as one of recrystallization, where the disordered region
would be a second phase of the same crystal, thermodynainically unstable with respect to the first phase, which is the normal crystal. Since
the two phases are in contact, the normal matrix serves as the nucleqs
for recrystallization. Although recrystallization of the hot zone would
decrease the entropy, there would still be a large decrease in free energy
owing to the highly exothermic nature of the free-radical and interstitialvacancy recombinations, and relief of strains in the lattice. Thus the
process should be spontaneous once a sufficiently high temperature is
reached. A recent discussion of the kinetics of recrystallization has
been given by de Boer (8). Unfortunately the theoretical studies deal
with two perfect crystal phases in contact through a disordered iaycr,
while in our systems the rapid cooling of the hot zone is likely to produce
a glassy phase (which can give up a very large heat of crystallization)
in contact with a perfect crystal. However, one may with certain assumptions derive a kinetic expression for the change in retention with
time of isothermal annealing based upon a recrystallization model. These
assumptions are: (1) There are initially present iV disordered regions
of roughly spherical shape, each containing one annealable SzilardChalmers recoil atom. (2) The locations of the recoil atoms within the
disordered regions are random. (3) Recrystallization proceeds from the
boundary inward a t a constant linear velocity v. (This is generally as-
29 1
sumed in mathematical treatments of recrystallization.) (4) Each recoil

atom is annealed when the phase boundary passes over it. The dependence of the velocity of boundary movement on temperature is given
by Hartshorne (38) as
1
2
where A , is a frequency term, E, is an activation energy, q is the heat
of transformation of the disordered zone to an ordered one, and To is the
absolute temperature of the transition point. Since in this case the ordered zone is stable up to temperatures which are large compared to the
annealing temperature T,the quantity l/To is probably small compared
to 1/T. As mentioned above, the quantity q is probably large compared
to the heats of typical crystalline phase transitions.
Subject to the assumptions listed above, one may derive (see Appendix A) the isothermal kinetic law
v = -A,exp (-E,/RT)
fraction annealed = Rt = 1
R,
- (1 -
where Rt and R , are the retentions at time t and a t infinite time, v is the
velocity given above, Eq. (15), and ro is the initial radius of the disordered zone.
In Fig. 5 are plotted the functions Rt/R,, Eq.. (16) , for various values
TIME IN UNITS OF rJv
FIG.5. The function R t / R = plotted against a function of the time. Simple

recrystallisation model. Curves for several velocities v are shown.
292
GARMAN HAaSOTTLE AND NORMAN SUTIN
of 2, against the time t , in units of ro/v. It is interesting that although

these curves have, for larger values of v, shapes not unlike annealing
curves at relatively high temperatures, the curves for smaller values of
v do not show the characteristic plateau or pseudo-plateau usually seen
in plots of experimental data. Although this treatment could be modified,
viz., by adopting variable radii ro or velocities v for different zones or
nonspherical zone shapes, it seems of little use to do this until more refined and extensive experimental data are at hand.
2. Annealing of Initially Randomly Distributed Pairs
a. Diffusion.-Controlled Bimolecular Recombination. The recombination of randomly distributed fragments follows second-order kinetics. If
the number of fragments of the two recombining types is the same, then
the rate expression is
dn -- kn2
dt
where n is the number of uncombined fragments a t time t, and k is

a bimolecular rate constant. The variation of k with temperature is
given by
= kge-E/kT
where ko is a constant roughly independent of temperature and E is the

activation energy.
These kinetics apply whether one or both of the recombining entities
are mobile. In the case of correlated pairs produced by radiation damage
Fletcher et al. (10, 27) maintain that after the first stage of recombination (governed by the unimolecular and randomiwalk kinetics) was
complete, the liberated species would continue to recombine, much
more slowly, by a bimolecular path. Only one activatioq energy, that
governing the diffusion coefficient D in Eq. ( l l ) , would apply to both
the random-walk and bimolecular kinetics. If bimolecular processes are
important in a Szilard-Chalmers system it seems likely that the recombination rate would depend on the extent of radiation damage in the
crystal before and during annealing. Experimental data bearing on this
point will be discussed below.
Bimolecular kinetics lead to the following retention law:
where the symbols have their usual meanings and the retention is measured from Ro.
SZILABD-CHALMERS REACTION IN SOLIDS
293
3. Thermal Exchange Model
Another mechanism whereby activity can be returned to the parent

form is through exchange reactions. For example, we can visualize the
following exchange reaction occurring on heating neutron-irradiated solid
organic halides ( 5 4 ) .
X*
'
+ RX G X + RX*
where X* is the radioactive recoil atom and RX the parent species. It

may be argued that the activation energy for such a process is too high
to account for annealing at moderate temperatures. But just as the activation energies for diffusion are thought to be lowered in the presence of
defects, so may activation energies for exchange.
If exchange between X * and any of its c neighboring groups is equally
probable it can be shown (Appendix B) that the retention should follow
the law
(R,
- R J / ( I - R,)
c exp
[+ c + l
Yt]
where v is the frequency of the exchange process. This predicts that the
logarithm of AR should yield a straight line when plotted against the
time t. It may also be seen that the retention can never increase to 1000/0
but only to c/ f c 1).
The temperature dependence of the annealing of such a system would
enter through the frequency V , which would presumably be governed by
an activation energy as in Eq. (1).
4. Phenomenological Approach. Annealing Functhns

Involving Mixed Kinetics
It has been shown by Fletcher and Brown (27) that provided the
vacancy moves in a crystal by jumping to an adjacent site, the annealing should depend upon the average jump time T and the time of isothermal annealing t only through the ratio t / r . Since the temperature
dependence of the annealing appears almost entirely through T, it should
be possible to combine data obtained from isothermal annealing experiments at different temperatures into a single curve of equivalent annealing a t a single temperature. This is accomplished by multiplying
the time scale of each isothermal curve by an appropriate factor, and
replotting all such curves on a single graph. Then the resultant composite annealing curve may be fitted by unimolecular, bimolecular or errorfunction expressions (see Sections Bla, Bld, and B2 above) or any
combination of any number of these. Several examples of this type of
294
GARMAN HARBOTTLE) AND N O W A N SUTIN
curve fitting are given in a recent paper by Clarke (17) including

Fletcher and Brown's results on germanium. When the jump times T
have been obtained, the crystal frequency yo may be calculated from
provided the energy of activation E is known. This quantity may be

obtained by one of the methods described by Parkins, Dienes, and Brown
(6S),which do not require any knowledge of the mechanism of the recombination reaction. In the discussion below, we will consider the
application of this method to typical Seilard-Chalmers annealing data.
5 . Processes Distributed in Activation Energy
In all of the foregoing cases it has been assumed that the energy of
activation for the process leading to recombination has one or a few
discrete values. Vand ('79) has suggested that the kinetics of annealing
of metal films could be interpreted in terms of recombination of defects
distributed in activation energy. We will outline the mathematical treatment of this concept of Vand, as amplified by Primak (64),and show
how, beginning with a set of isothermal annealing curves, one can in
principle obtain the spectrum of activation energies.
It is assumed that the annealing processes, though distributed in
activation energy, all follow kinetics of the kind
dn/dt = knr
(22)
where ?i is the number of annealable defects, y is the order of the reaction, and k is n rate constant of the usual form
k = Ae--E/kT.
(23)
Next, it is necessary to assume that the property p, which is actually
measured, is linearly related to the number of defects n,
p = f.
(24)
In the case of Szilard-Chalmers annealing this requirement presents no
problem since the quantity ( R , - Rt), where the symbols have their
previous meanings (see Fig. 2), is a linear function of the number of
annealable defects which remain. Equation (22) is then
dpldt = f A [exp ( - E l k T ) I ( d f ) 7 .
(25)
I t is convenient to consider first the situation if E is a single, discrete
activation energy, then to proceed to distributions in E . If E is discrete,
we can integrate Eq. (25) to get p,, the measured property a t time t in
295
terms of its initial value

p,
pt =
PO.
One obtains
p u [I - (1 - y) Bt exp (-h'/kT)]*'1-7
PO
[exp ( - A t exp (-E/kT))I
yf
y = 1
(1%)
(27)
pt = Po&
(28)
defining 6,, the characteristic annealing function. The constant B is compounded of other constants
=
A (po/f)y-'
(29)
= AQr-1
(30)
of
is
the
initial
number
of
annealable
defects.
Rearrangement
where
Eqs. (26) and (27) leads to the following:
r=l
which show that for a fixed period of isothermal annealing a plot of
In [In ( p o / p ) ] against 1/T should give a straight line only for y = 1. As
mentioned above, the property p can be identified with the retention
Similarly, for a fixed degree of annealing, p / p o
showing that the logarithm of the time necessary to produce a given

degree of annealing, p / p o , plotted against 1/T should yield a straight
line of slope E / k for any order of reaction y. This is one of the Parkins
et d. (63)results. mentioned in Section B4 above. For Szilard-Chalmers
studies a fixed degree of annealing means a fixed increase in retention, A R .
As mentioned above, Eq. (26) to (35) refer to a discrete activation
energy. If the annealing processes are distributed in activation energy
the measured value of the property P a t time t will be the integral over
all activation energies of all the discrete processes occurring a t the annealing temperature a t any instant. These will tend to lie within a narrow
band of activation energies dE, centers with lower activation energy
already having been annealed out, and those with higher energy as yet
296
untouched. Thus
where p o ( E ) is the initial distribution of observable properties p with

activation energy E, and 6,(E,t) the value of the characteristic annealing function at time t .
The annealing process may be visualized mathematically as the movement of the function 8, along the E axis of the p o ( E ) versus E distribu-
ENERGY OF ACTNATION
FIG.6. Graphical representation of the annealing process in terms of the movement of the characteristic annealing function e.,
tion (see Fig. 6 ) . The position of the point of inflection in the 6, function,
Eo, at any time t is given by

E,, = kT In (Bt)
any y
(37)
and as annealing proceeds at temperature T the function 6, sweeps across

the distribution p o ( E ) , at first rapidly, then moving ever more slowly
because of the dependence on In t. The quantity Eo is referred to as the
characteristic activation energy. A t any given time t the portion of the
initial spectrum to the left (i.e., a t activation energies lower than E,)
of Or has been annealed out and the portion to the right has not yet been
annealed.
An important simplification results if the function B is approximated
by a step-function rising at Eo. This replacement is justified for small
values of y as 6, in these cases rises steeply. The integral, Eq. (36), then
297
becomes
Pt
=jE:POW)dE
(38)
from which one obtains, by differentiation and substitution
Ldp
= kTd(lnt)
(39)
=POm
where the designation Eo rather than E indicates that the step-function
approximation has been taken. One is now in a position to obtain a plot
of po(Eo) against Eo.
The first step is to plot the measured value of the property, P , (in
the case of Szilard-Chalmers studies, ( R , - Rt)) against In t . From the
slopes of these curves (one for each temperature) one obtains the quantities (l/kT) dp/d(ln t). These quantities are then plotted against kT
ln(Bt) = Eo. The factor B is arbitrarily chosen to give the best overlap
between successive isothermal annealing curves: a change in B does not
alter the shape of the initial spectrum of .activation energies, but only
the position on the Eo axis. If one takes only the average slope of each
isothermal curve and then plots the values of (l/kT)dp/d(ln t ) so obtained against Eo one obtains a segment of a horizontal straight line, lying at the appropriate value of the slope on the po(Eo) axis and extending from Eo = kT In ( B t l ) to Eo ='kT In (Bt2) on the Eo axis where
tl and tz are the annealing times over which po(Eo)was evaluated. This
type of plot gives a rough indication of the initial spectrum of activation
energies. If the annealing data are sufficiently precise, a second stage of
refinement may be reached in which, instead of an average slope, one
reads off the actual slopes point by point and then plots (l/kTj
dp/d(ln t ) calculated from the actual slopes against kT In ( B t ). One obtains from this a better representation of the initial spectrum, but the
method suffers from the difficulty inherent in all procedures in which
slopes are obtained from experimental data, namely, that the data must
be of a high order of precision. Unfortunately, such precision is difficult
to obtain in annealing studies.
Several examples of analysis of actual data by this method are given
in Clarke's (17) paper, and we shall discuss below its application to
Szilard-Chalmers annealing data.
6. Activation Energy Varying with the Number of Defects
In the interpretation of isothermal-annealing data for radiation-damaged metals, M a n et al. (66) and Overhouser (62) suggested that the
a r t i n t i m energy for recombination might be a function of the number
298
GARMAN HABBOTTLE AND NORMAN SUTIN
o defects present. Dienes (24) has amplified this point in terms of a

linear dependence
Eo
- an
(40)
where E is the apparent activation energy in an actual damaged sample,

containing n defects, Eo would be the corresponding sctivation energy
of annealing isolated defects, and a is a constant. This leads to a general
rate law
Such a treatment has somewhac the same effect as the Primak model,
ie., as annealing proceeds, n grows smaller and the apparent activation
energy increases (24, 62).
Dienes points out that for kinetics obeying Eq. (41), a plot of In t
against 1/T for fixed degree of annealing still gives a straight line, but
the slope changes with the degree of annealing. As might be expected, a
relatively small perturbation in E leads to a large change in the time
scale.
We thought it worthwhile to mention this approach and to call attention to the fact that the Szilard-Chalmers method represents a unique
way of testing this hypothesis. Annealing of recoil atoms may be studied
in crystals which have received anything from infinitesimal (a few seconds in the pile) to massive doses of radiabion, corresponding t o large
changes in the quantity n ; Eq. (40). Some experimental results bearing
on this point will be discussed below.
c. THE IXTERPRETATION
OF
RADIATION
ANNEALING
It has been observed that recoil atoms undergo annealing when exposed to ionizing radiation from radioactive sources such as cobalt-60,
from X-ray generators or from nuclear reactors (9, 20, 21, 55, 66, 8 3 ) .
In the reactor not only gamma rays but presumably fast neutrons also
play a role. A possible explanation of this effect is that in crystals exposed to radiation, the interactions of the radiation with the crystal give
rise to thermal spikes, and that these occasionally occur sufficiently
close to a recoil atom to give it the activation energy necessary to anneal.
Such a view implies that radiation annealing is fundamentally similar to
thermal annealing. Another possibility which permits somewhat more
efficient use of the energy put into the crystal by radiation is that the
lattice vibrations excited by the fast electrons produced by the radiation pass through the crystal until they encounter a defect. There they
give up some of their energy, again suppljring the recoil atom with suffi-
299
cient energy to permit it to surmount its potential barrier to recombination. Finally, the recoil atoms may combine with fragments or electrons
produced by the radiation. For the purposes of this discussion we will
consider separately annealing by gamma rays and by fast neutrons.
1. Gamma Rays
Moderately energetic gainma rays lose their energy entirely by interacting with electrons. At low energies the most important interaction is
the photoelectric effect, while a t energies -1 MeV, Compton scattering
dominates. The maximum energy which a Compton electron of energy
E can transfer to an atom of mass M is
where in is the electron mass and c the velocity of light. Provided that
the energy necessary to displace an atom, E d , 25 ev, it can be seen that
a Compton electron of energy 0.41 Mev can displace an atom of mass 50.
This is about the average energy of the Compton electrons from interactions of -1 Mev gamma rays (for example, from a cobalt-60 gammaray source).
However, the cross section for such displacements is small and thus
the number of bonds ruptured in this manner is small compared to the
number-ruptured by electronic excitation and ionization of the molecules
by Compton electrons. A Compton electron is pictured as losing its energy in local hot spots, corresponding to the production of secondary
electrons of energy dl00 ev. These secondary electrons dispose of their
energy in a volume containing of the order of 100 molecules. At the beginning of the range of the Compton electron, the hot spots will be
spaced several thousand Angstroms apart, but by the time the energy has
dropped to 1 kev, the spacing may be only 50A.
The problem suggested by these data is whether radiation annealing
is due to the local high temperatures in the hot spots or to the enhanced
probability of recombination owing to the increased supply of ions, free
radicals, or electrons produced by the radiation.
The first of these alternatives may be approached in the following
manner. While part of the energy of the secondary electron is used in
bond rupture, most of it ultimately appears as heat. We can calculate
the average energy dissipated per molecule when the solid has received
a sufficient dose of radiation to anneal about half the recoil atoms. In
typical cases this requirement is about 5 X lo7 r. Such a dose corresponds
to an energy dissipation of 3 x 102 ev/gm or about 1 ev/mol. Under
these conditions every radioactive atom is included once, on the average,
r~
300
GdaMAN HABB(YITLE AND NOBMAN SUTIN
a hot spot of 100 ev. Since 1 ev is the order of magnitude of the activation energy to be expected for typical annealing reactions, it is seen
that this simple thermal spike mechanism probably would suffice.
The second mechanism is recombination of the recoil atom with ions
or radicals produced by radiation. Because of the weaker intermolecular
forces in molecular crystals, individual excited molecules are relatively
more isolated and will decompose into fragments (usually free radicals)
more readily than in ionic crystals. Assuming that 10 ev are necessary
to produce a free radical pair in a molecular crystal, about ten such pairs
would be produced per 100-ev hot spot. Since we have seen that for doses
of 5 X lo7 r every radioactive atom is included once, on the average, in
a hot spot, it follows that every radioactive atom will have a good probability of being found in the vicinity of a free radical a t least once.
Although most of the free radicals would recombine with one another
their concentration would probably still suffice to make this second
mechanism an attractive possibility.
A third mechanism involves the movement of electrons resulting from
the ionization produced in the crystal by gamma rays. Electrons released
in this way may be captured relatively far away, and capture by defects
could be of importance in ionic crystals. In general radiation, like thermal, annealing leads to an increase in the observed oxidation state of the
radioactive atoms whereas capture of electrons would reduce the oxidation state. However, if partial neutralization of excess positive charge
was a necessary requirement before the recoil species could anneal, electron movement might still play a role.
The rate of radiation annealing in potassium bromate (22) and
chromate (35) has been found to depend on the temperature. Ordinary
thermal annealing was negligible over the range of temperatures employed. The results suggest that there is an energy of activation, of the
order of 8 kcal/mol, associated with the radiation annealing process.
The fact that this process is temperature-dependent shows that the
energy put into the crystal by gamma rays is effective in annealing when
it has been degraded to quite a low level, or diffused widely. This conclusion supports both the second and third mechanisms proposed above,
since the activation energy could apply in either case to the free radical
recombination in the vicinity of the hot spot. If the radiation annealing
occurred in a near-hit, high-temperature process, the rate of such annealing would not be affected by a small change in the ambient temperature.
2. Fast Neutrons
I n contrast to gamma rays, fast neutrons lose their energy entirely

by nuclear interactions, producing knocked-on atoms of maximum energy
301
Em,, given by
where M is the mass of the knocked-on atom and E the neutron energy.
Thus a neutron of only 325 ev can displace an atom of mass 50 (if the
displacement energy is 25 ev) . These displaced, knocked-on atoms, unless very light, lose energy by collisions and not by ionization. Each
energetic atom will initiate a hot zone with its attendant thermal effect,
as described in Section I, and in such zones annealing could take place
by any of several mechanisms. A process of particular importance in
organic Szilard-Chalmers systems is the knocking-on of protons. These
will lose energy by ionization as well as collision, and this should produce effects similar to gamma-ray annealing. Unfortunately there are as
yet no experimental data on the annealing effects of a pure fast-neutron
flux although such experiments are, in principle, easily carried out.
D. DISCUSSION
AND COMPARISON
WITH EXPEJUMENTAL
DATA
If we are to compare the results of recoil atom annealing experiments
with studies of the annealing of radiation damage it should be recognized
at the outset that the chemical studies, by reason of the valence and
bonding changes involved, are examining somewhat more complex systems. Still, the analogous isothermal curves of the annealing of SzilardChalmers recoils on the one hand and the annealing of optical centers,
resistivity, or density changes on the other, all as a function of time of
annealing, show such a surprising similarity in shape, viz. a rapid initial
rise followed by a slowly rising plateau, with the extent oT initial rise
governed by the temperature, that one is tempted to apply analogous
theoretical treatment to the data, and to assume that one is observing
in both cases something characteristic of crystals in general.
Several studies have revealed the complexity of typical chemical
crystalline systems in regard to annealing. The work of Aten et al. (6)
showed that recoil of iodine in KIOl led to a distribution of activity in
unheated samples as follows: I- - IZ activity lo%, 103- 86%, and IOaabout 4%. However, on annealing it was found that the I- - It activity
passed over into 103- below 100C, with no further production of active
periodate. This process was essentially complete for annealing times of
15 min at 100C. In the range 100-24OoCthe iodate activity began t o
pass over to periodate, showing the usual shape of annealing curve with
temperature-dependent plateau.
Similarly, in the case of Sharman and McCallums (74)studies of
recoil reactions of carbon in anhydrous sodium carbonate it was found
302
GARMAN HARB0"LE
AND NORMAN SUTIN
that different products (produced by dissolving the irradiated crystals

in water) began to disappear a t quite different annealing temperatures.
Such results as these are a clear indication that several types of fragment
loci, having different annealing characteristics (very likely different
activation energies), exist in the irradiated crystal. The case of potassium
periodate cited above is rather good evidence for stepwise addition of
oxide ions. Evidence for similar effects a t a single annealing temperature
was found by Maddock and Sutin (52) (stepwise addition of phenyl
radicals to arsenic fragments) and by Rauscher and Harbottle (66)
(addition of cyanide to cobalt fragments).
The most extensively studied crystalline Szilard-Chalmers system is
potassium chromate, where the annealing of recoil chromium has been
measured for a number of temperatures, over wide ranges of time (31, 50,
53, 5 4 ) . Although stepwise addition of oxide ions probably also occurs
here, there are arguments, based upon the chemistry of hexavalent
chromium, for thinking that only one of these influences the observed
retention, and hence that the kinetics refer to a single elementary annealing process; Eq. (6). This, combined with the great thermal stability of
K2Cr04 and long lifetime ( t l I P= 27 days) of the radioactive chromium,
make it a most valuable system for study.
In various sections above we have discussed the simple kinetic
schemes, some of which have been applied to experimental data on annealing of recoil atoms. For example, it was thought a t first (91) that the
kinetics of cobaltic trisethylenediamine annealing obeyed a linear combination of two unimolecular terms. Further analysis (53) has shown that
this interpretation was probably in error, and also that the potassium
chromate and bromate data could not be fitted by a small number of
biinple exponential terms. We have also attempted to fit the potassium
bromate and cobaltic trisethylenediamine data by the error-function
expression suggested by Fletcher and Brown (10, 273, see Section Bld
above, and although the fit is reasonably good over portions of the
isothermal curves, the point of inflection near the time-zcro axis in the
error-function expression is not observed experimentally.
The easiest analytical test for annealing data for which discrete
activation energies are assumed is that given in Section B4 above. One
constructs a composite annealing curve from all the data and then
attempts to fit this curve by the appropriate expressions for the chosen
order of reaction y, or by the error-function expression, or combinations
of these. Examples of such fits are given for radiation damage in
germanium (Fletcher et al., 27) and in magnesium oxide (Clarke, 17, 19).
Both these cases are shown in a recent paper by Clarke ( 1 7 ) . We have
applied this method to some of Zuber's cobaltic trisethylenediamine data
303
(91) : thc composite annealing curve is shown in Fig. 7 and is there fitted
fairly accurately by a single bimolecular (7 = 2) expression, Eq. (19).
The interpretation of this result might be that the annealing rate, -dn/dtJ
is proportional to fhe density of interstitials times the density of
vacancies multiplied by the volume of the spherical region containing
In (TIME. IN H&RS)
FIG.7. Composite annealing curve of recoil atoms in cobsltic trisethylenediamine, data of Zuber (91). Line is a bimolecular fit.
the first shell of groups around the region of disorder. Such an interpretation suffers from the fact that it is difficult to visualize what physically
constitutes an interstitial and a %acancy in such a complex system.
It is, however, interesting that bimolecular kinetics have been found in
several other cases of thermal annealing (7,6R).
I n the case of the potassium chromate data the composite annealing
curve approach does not work a t all. One cannot even approximately get
all the annealing data onto a single curve of retention against In t by
merely shifting the l n t scale. This difficulty was observed by Mrs. de
Maine in her thesis (55) for short-term annealing data and we have
found thk same result if we consider only the long-term annealing curves
(54).Also, a plot of In t versus 1/T for a constant degree of annealingsee Eq. (34) and (35)-does not yield a straight line when applied to
304
GARMAN HABB<MTLE AND NORMAN SUTIN
the chromate data. These results indicate that some other interpretation
should be sought.*
We shall first discuss the explanation given by Maddock and de Maine
(49, 60). Their proposal, which is described in Section B l b above,
satisfactorily accounts for the linear relationship that has been observed,
not only in the case of potassium chromate but for several other systems
as well, between the change in retention on annealing to the plateau,
A R , and 1/T. Their formalism also explains the observaJion that, when
one has annealed to the plateau characteristic of temperature TI,an
increase in temperature from TIto T2causes a new rapid growth to
occur, tapering off at the new plateau characteristic of temperature Tp.
This model may be criticized, however, on the following grounds: (1)
There is an assumed uniform distribution of fragment spacings .between
the minimum and maximum separations (X,and X, in Fig. 3). This
assumption is necessary to establish the linear relationship between AR
and X (Section B l b above). (2) There is also the question of the potential
V , Eq. (7). Although it is not necessary for V to be electrostatic, as was
argued in the case of potassium chromate, it is required that the potential
lead to a term 1/X in the second exponential term in Eq. (7). In the
annealing of molecular solids, it seems probable that neutral atoms and
radicals recombine. One can still assume that the potential V comes from
the strain field of the crystal, in such a way that a l / X dependence is
obtained, but in view of the large repulsive forces acting on an interstitial
in a lattice such a dependence would be fortuitous, particularly since
such terms usually go as a rather high power of X . (3) The recent
results on the slow part of the annealing process in potassium chromate
by de Maine et al. (54) show that, following the initial rapid rise, the
retention increases slowly and linearly to 100%. According to Eq. ( 8 ) ,
with the second exponential term effectively unity the final slow process
should be an exponential fall of the quantity (100 - R t ) to zero with
time. What was experimentally observed was a kinetic law
-dn/dt = k
(44)
where k is a zero-order rate constant with an apparent activation energy

of about 4 kcal/mol. It seems to be inherent in the Maddock-de Maine
model that -dn/dt = kn, and it is difficult to see how zero-order kinetics
could be accommodated.
The Vand-Primak treatment of annealing kinetics, in which the energy
The discrete activation energy case EEq. (31) and (32)l predicts that if +y= 1
a plot of In [In(p./p)l against l / T should yield a straight line, for fixed times
of annealing. Such R linear retationship is not obtained when potassium chromate
data are used.
305
of activation is assumed to have a continuous distribution of values, has

been discussed in general terms in Section B5 above. As pointed out by
Clarke (18) it can also be invoked to explain the potassium chromate
annealing data. On this view, the rapid rise to the pseudo-plateau corresponds to the rapid initial movement of the function 0, over that portion
of the activation-energy spectrum which contains the lowest values of E
(see Pigig.
6). The dependence of the position of the plateau on temperature
arises from the linear dependence of Eo, the position of the point of
inflection of Qy, on T,the absolute temperature as expressed in Eq. (371,
for constant time of annealing. The portion of the spectrum annealed
first is presumably identified with close interstitial-vacancy pairs, whose
energies of activation are materially reduced by the attendant lattice
strain. Then the slowly rising portion of the pseudo-plateau represents
events of only slightly larger activation energy. If we compare the curves
reproduced in Clarkes paper for the annealing of radiation damage in
germanium (27), silica (65), and magnesium oxide (17, 19) with the
potassium chromate data (50, 53, 54) the same general features, a rapid
rise to a pseudo-plateau depending on temperature, followed by a slow
upward movement are seen. The only possibly unique feature of the
latter data is the zero-order approach to complete annealing. The data
of Overhouser (62) on the annealing of radiation damage in copper from
-123C to 33C also show a number of these plateaus.
On this view, it would seem that in general Seilard-Chalmers experimentation has been directed at the theoretically less interesting
feature of the annealing curve, namely the rapidly rising part; whereas
solid-state research has dealt with the kinetics of what the chemists have
considered as the plateaus of their curves. It is unfortunately not always
easy to obtain reliable data for long-term annealing in Seilard-Chalmers
systems. The half-life of the radioisotope employed is frequently too short,
and in addition the crystalline compound investigated may be thermally
unstable. As mentioned above, potassium chromate is a particularly
favorable case in that neither of these difficulties interfere.
We have applied the Primak treatment to the potassium chromate
annealing data of de Maine et al. (54). The value of 1nB was chosen
as 18.5 to give the best fit. The spectrum obtained (Fig. 8) does not at
first sight look very promising, but this is in part due t o the great
sensitivity of the method to experimental error and to the rather close
spacing in temperature of the isothermal annealing curves, since the
resolution of the method is only of the order of 3 kT.The steeply rising
slopes at the low-energy side should be compared, for example, with the
similar portion of Primaks spectrum for silica (65): it will be seen that
the behavior of the slopes is similar. About 10% of the spectrum of
306
GARMAN HARB(YTTLE AND NORMAN SUTIN
600
500
400
4% 3oo
-1%
200
I00
0.5
1.0
E, IN ELECTRON VOLTS
1.5
FIG.8. The initial distribution of activation energies in the Szilard-Chalmers

reaction in potassium chromate; po ( E , ) against Eo in electron volts.
annealable chromium ii t . , of those centers whose annealing reactions can

be followed) lies to the left of the curves shown in Fig. 8. The data seem
to indicate a cutoff in the neighborhood of 1.4 ev, that is, there are no
processes of higher activation energy.
At this point i t might be of interest to turn the arguinent about, assume
a simple distribution of activation energies and see to what this leads.
We shall assume that in potassium chromate about 90% of the activation
energies of the annealing process are uniformly distributed between 1.10
and 1.42 ev, i.e. the distribution of activation energies is flat between those
two values. Then the increase in retention AR for a constant time of
annealing will vary linearly with Eo, the position of the point of inflection
of the annealing function 0, [Eq. (37) and Fig. 61 and the increase in
retention to the plateau (for example, at a constant 20 hours heating) will
vary linearly with T, the absolute temperature. A plot of AR against
l',where the values are those taken from Maddock and de Maine, is given
in Fig. 9. A flat distribution of activation energies is not entirely inconsistent with the Primak-spectrum plot of Fig. 8, and as an ad hoc
choice may be compared to the assumption of a uniform distribution of
fragment spacings X in the Maddock-de Maine treatment (Section B l b
above).
Maddock and co-workers (53, 64) have also studied annealing
processes in mixed crystals of potassium chromate-fluoberyllate in which
SZILARD-CHALMERf5 REACTION IN SOLIDS
307
AR
30
5
P
I-
e
g
2
20
z
W
I-
z
w
"Ea
10
ANNEALING TEMPERATURE
OK
FIQ.9. The increase in retention to the plateau,

AR, plotted against T, the
'
absolute temperature, for potassium chromate.
the potassium chromate was very dilute (- 1 mol per cent). The annealing curves are quite similar to those of pure potassium chromate, but
rise more steeply near the origin of the time scale. This could be interpreted to mean that the spectrum of activation energies is similar in
shape but that a portion of it lies shifted to lower values of .Eo. The fact
that the recoil chromium species anneal so readily in the dilute mixed
crystals is perhaps evidence that the annealing reaction is largely a local
phenomenon ; i.e. that' original partners recombine. It would obviously be
of interest to study other similar systems, for example in which potassium
bromate OF iodate were diluted.
It has recently been observed (41) that atoms of two different isotopes
of the same element do not exhibit identical annealing behavior. In
particular, it was found that BrS2recoil atoms annealed more rapidly than
308
BrBoma toms in neutron-irradiated alkali bromates. Presumably this

difference arises from a slight shift of the spectrum of activation energies
to lower values in the case of recoiling Brsz atoms. Whether this shift is
due to differences in the recoil energy or in the average charge of the
radioactive bromine, the interesting fact remains that the spectrum of
activation energies is not uniquely determined by the crystal and the
interstitial atom, and that isotopic atoms of the same element, a t rest
in the same crystal, can be associated with two different spectra. The
Maddock-de Maine model can also account for an isotopic annealing
difference, by assuming that the two isotopes have different average
fragment spacings X [Eq. (7) and Fig. 31, for example, by virtue of
different recoil energies.
When samples of crystalline compounds are subjected to pile neutron
irradiation it has frequently been found that the larger the integrated
flux the higher the initial retention. Such an effect is the result of radiation
annealing (see above). However, it has now also been observed (53,54,
77) that on thermal annealing of such crystals, the more heavily damaged
anneal the more rapidly, arriving at a higher retention plateau for the
same isothermal annealing time, or reaching 100% retention a t a shorter
time. Such effects are perhaps the'result of a general downward perturbation of the spectrum of activation energies of the radioactive recoil atoms
in the presence of the large number of additional defects introduced by pile
irradiation. This effect has been mentioned in Section B6 above, and
might operate with even first-order annealing reactions. If the kineties
were higher than first order then annealing would of course be accelerated
even without perturbation in the activation energy. The Primak treatment of the potassium chromate annealing data indicated that for a
sample bombarded one week at a flux of 1O'O neutrons/cm2/sec thc top of
the activation energy spectrum fell at about 1.42 ev. Our calculations taking the annealing data of de Maine indicate that for a sample which was
irradiated one week at 5 x lo1' neutrons/cm2/sec the top of the spectrum
would be displaced downward to about 1.24 ev. It would be of great
interest to carry out similar studies in which the crystals were heavily
radiation-damaged, for example by fast neutron or other heavy-particle
bombardment, before introduction of the Szilard-Chalmers atoms.
Although we have not distinguished between ionic and molecular
crystals in the above discussion, for the reason that the latter also show
typical thermal annealing effects, it is possible that their annealing
reactions may be discovered to arise from quite different mechanisms.
If we take the ratio TJT,,,
as a measure of the tendency of a crystal to
anneal recoil atoms, where T, is the absolute temperature a t which an-
309
nealing is half-complete in 10 min, and T, is the absolute melting point,

we find that whereas most inorganic, ionic crystals have values of Ta/T,,,
in the range of 0.4 to 0.6 (including several cases of annealing of simple
radiation damage) the three organic crystals that have been studied fall
in the range 0.8 to 0.95. It seems significant that this range of Ta/Tm
(0.4 to 0.6) in ionic crystals brackets the ratio of the Tamman temperature to the melting point (defined as 0.5). The Tamman temperature is
frequently taken as a rough indication of the temperature at which lattice
diffusion becomes appreciable (81). It was found by Andrew ( 2 ) in his
nuclear magnetic resonance studies of self-diffusion in molecular solids
that the latter showed several stages of increasing agitation below the
melting point, culminating in a form of molecular movement which could
be identified as diffusion of whole molecules throughout the lattice. In
cyclohexane, this occurred a t 220'-240K, which temperature is about
0.82 of the melting point. A loosening of the lattice suficient to promote
diffusion of whole molecules would certainly be expected to have a
profound effect on the annealing by recombination of recoil atoms or
fragments in molecular solids.
IV. Conclusion
We have hoped in this article to accomplish a twofold purpose. On

the one hand we wished to make available to investigators of the SzilardChalmers reaction and associated recoil effects some of the results of
recent work in solid state physics, in particular that part of the latter
dealing with the study of radiation damage. I n this way a fresh light could
perhaps be cast on some of the work progressing in this now classic field
of radiochemistry, On the other hand we wished to call to the attention
of workers in the field of radiation damage the existence of chemical data
bearing upon, and in some cases giving unique answers to, questions of
interest to them.
V. Appendix
A. RATEOF INCREASE
IN RETENTION
DUETO RECRYSTALLIZATION
OF SPHERICAL
ZONES, EQ. (16)
Subject to the assumptions (1) to (4) listed in the text (Section le)
let ro = initial radius of the disordered spherical zone, r = radius at time
t , v = linear velocity of motion of phase boundary, N = number of zones
(each containing one recoil atom) present initially, and n = number of
recoil atoms uncombined at time t. The probability of recombination in
310
GARMAN HABBOTTLE AND NORMAN SUTIN
any given disordered zone, P, per second is then
and since T
= r0
- ut.we have
Since -dn/dt = (l/K)dZ?/dt = NP, where K is a constant and R the

retention, we have
3V
@ = K N - (ro - vi)z = R , 2 (r,, - ut)*.
dt
To3
Toa
Inteeratine
at t = 0, Ro/R, = 0, and a t t = r0/u, * i.e., complete recrystallization,

R / R , = 1.
B. DERIVATIOX
OF RATELAW FOR EXCHANGE,
EQ. (20)
The radioactive recoil atom X" is assumed to be trapped in the midst
of c inactive neighbors A X . All sites are identical. As annealing proceeds,
X" can exchange with any neighboring A X to give AX* and X , and the
process is reversible. Let Y be the frequency of exchange per group. If
the number of active ( X * ) atoms be n(X*), etc., then
+
+
+
n(X*)
n(AX)* = n (conservation of active atoms)
n(X)
n(X*) = n
n ( A X ) n(AX)* = can
where n is the total number of sites. The frequency with which exchange
processes occur through the sample is then nv.
The rate of disappearance of X" is then
which reduces to
Integrating this and putting in boundary conditions we obtain
n(X*) =
i.e.
cn
(c + 1)
n ( X * ) - n(X*),
c+l
c - n ( X * ) , exp
where the subscript
oo
311
refers to infinite time. In terms of retentions
R,
--- Rt
1 - R,
- c exp [
vt].
ACKNOWLEDGMENTS
The authors acknowledge with thanks helpful conversations with
R. W. Dodson, M. Wolfsberg, A. G. Maddock, F. P. Clarke, and J. H. 0.

Varley. One of the authors (G. H.) is indebted to the John Simon Guggenheim Memorial Foundation.
LISTSOF SYMBOLS
Constant relating the number of defects to the change in activation
energy
Order of reaction
Jump frequency of interstitial atoms
Frequency of vibration of atoms in a lattice
; v j above)
Average jump time ( l / ~ j see
Characteristic annealing function, Vand-Primak model
Frequency terms
Sink radius, Fletcher-Brown model
Constant containing the frequency A
Initial pair separation distance, Fletcher-Brown model
Number of neighboring groups, exchange model; also, velocity of light
Diffusion coefficient
Activation energy, also, energy of a neutron or Compton electron
Activation energy, Hartshorne expression
Energy to displace an atom to an interstitial position
Maximum energy of recoil from a Compton electron, or from a neutron
Characteristic activation energy, Vand-Primak model ; also, activation
energy for isolated defects, Dienes treatment
Constant relating measured property due to defects, and number of defects
Boltzmanns constant (usually as kT); also, a rate constant
Lattice constant
Atomic mass
Number of interstitial atoms, or uncombined pairs
Same aa n, but at time zero and t , respectively
Same a~ p t , but relating to variable energy of activation case
A property whose magnitude depends on the number of defects, and the
same at time zero and t , respectively
Initial distribution of properties p
Same aa PO(&),but denoting that a step-function approximation l i ~ ebeen
taken for 8,,
Heat of tramformation of crystalline phases
312
R
Ro,Rt,R
AR.
r,ro
T
T.
T,
t
U
J7
GARMAN HUBWTLE AND NOBMAN SUTIN
Retention (see footnote 3)

Retention at time zero, t and the final, limiting value
Change in retention from start of annealing to the plateau, Maddock-de
Maine model
Radius of spherical disordered zone, and the initial value of the same
radius
Temperature ( K )
Characteristic annealing temperature (OK) (temperature necessary to
anneal one half the defects in ten minutes)
Melting point of the crystal (OK)
Time
Activation energy, Maddock-de Maine model
Electrostatic potential, Maddockde Maine model
Velocity of movement of phase boundary, recrystallization model
Distance of separation of recombining pairs, Maddock-de Maine model
REFERENCES
I. Amaldi, E., DAgostino, O., Fermi, E., Pontecorvo, B., Rasetti. F., and SegtC, E.,
Proc. Roy. SOC.(London) A149, 522 (1935).
8. Andrew, E. R., in Report of the Bristol Conference on Defects in Crystalline
Solids. The Physical Society, London, 1954.
3. Aten, A. H. W., Phys. Rev. 71, 641 (1947).
4. Aten, A. H. W., and van Berkum; J. B. M., J . Am. Chewt. SOC. 72, 3273 (1950).
6.Aten, A. H. W., Koch, G. K., Wesselink, G. A,, and de Roos, -4. M., J. Am.
C h a . SOC.79, 63 (1957).
6. Aten, A. H. W., van der Straaten, H., and Riesebos, P., Science 115, 267 (1952).
7. Bemski, G., and Augustyniak, W. M., Phys. Rev. 108, 645 (1957).
8. de Boer, J. H., Disczlssions Faraday Sac. 23, 171 (1957).
9. Boyd, G. E., Cobble, J., and Wexler, S., J. Am. Chem. SOC.74, 237 (1952).
10. Brown, W.L., Fletcher, R. C., and Machlup, S., Phys. Rev. 90, 709 (1953).
11. Burgus, W.H., and Kennedy, J. W., J. Chem. Phys. 18, 97 (1950).
1.2. Cabrera, N , and Mott, N. F., Repts. Progr. in Phys. 12, 163 (1948).
13. Carlson, T. A., and Koski, W. S., J. Chem. Phys. 23, 1596 (1955).
14. Cader, R., Pringsheim, P., and Yuster, P., J. Chem. Phys. 18, 887 (1950).
15. Chien, J. C. W., and Willard, J. E., J. Am. Chem. SOC.75, 6160 (1953).
16. Christman, D.,private communication.
17. Clarke, F. P., Discussions Faraday Sac. 23, 141 (1957).
18. Clarke, F. P., Discussions Faraday SOC.23, 170 (1957).
13. Clarke, F.P., Phil.M a g . [Sl 2,607 (1957).
M. Cleary, R. E., Hamill, W. H., and Williams, R. R., J. Am. Chem. Soc. 74, 4675
( 1952).
21. Cobble, J., and Boyd, G. E., J. Am. Chem. SOC. 74, 1282 (1952).
U.,Giacomello, G., and Maddock, A. G., Ricerca sci. 21, 1788 (1951).
23. Croatto, U.,and Maddock, A. G., J. Chem. SOC.1949, 351; Nature 164, 613
( 1949).
24. Dienes, G. J., Phys. Rev. 91, 1283 (1953).
96. Dienes, G.J., and Vineyard, G. H., Radiation Effects in Solids, p. 129. Interscience, New York, 1957.
26. Evans, J. B., and Willard, J. E., J. Am. Chem. SOC.78, 2908 (1956).
2.2.Croatto,
SZILARD-CHALMEha REACTION IN SOLIDS
B. Fletcher, R. C., and Brown, W. L., Phys. Rev. 92, 585 (1953); Fletcher,
Brown, W.L., and Wright, I(.A., ibid. 92, 591 (1953).
313
R. C.,
98. Fox, M., and Libby, W. F., J. Chem. Phys. 20, 487 (1952).
29. Friedman, L., and Libby, W. F., J. Chem. Phys. 17, 647 (1949).
30. Goldhaber, S., Chiang, R. S. H., and Willard, J. E., J. Am. Chem. Soc. 73,
2271 (1951).
31. Green, J. H., Harbottle, G., and Maddock, A. G., Trans. Faraday SOC.49, 1413
(1953).
32. Green, J. H., and Maddock, A. G., Nature 164, 788 (1949).
3.9. Hall, R. M. S., and Sutin, N., J. Znorg. & Nuclear Chem. 2, 1M (1956).
34. Hamill, W. H., Williams, R. R., and Schwarz, H. A., J. Am. Chem. SOC.72,
2813 (1950).
36. Harbottle, G., J. Chem. Phys. 22, 1083 (1954).
36. Harbottle, G.,and Maddock, A. G., J. Znorg. & Nuclear Chem. 5, 249 (1958).
37. Harbottle, G.,and Sutin, N,, J. Phys. Chem. 62, 1344 (1958).
38. Hartshorne, N. H., Discussions Faraday SOC.5, 149 (1949).
39. Herr, W. P., 8. Elektrochem. 56, 911 (1952).
40. Hornig, E. O., and Willard, J. E., J. Am. Chem. SOC.79, 2429 (1957).
4.2. Jach, J., and Harbottle, G., Trans. Faraday SOC.
54, 520 (1958).
4.9. Koski, W., J. Am. Chem. SOC,71, 4042 (1949).
4.3. Levey, S., and Willard, J. E., J. Am. Chem. SOC.74, 6161 (1952).
4. Libby, W. F., J. Am. Chem. SOC.62, 1930 (1940).
46. Libby, W. F., J. Am. Chem. Soc. 69, 2523 (1947).
46. McCallum, K. J., and Holmes, 0. G., Can. J. Chem. 29, 691 (1951).
47. McCallum, K.J., and Maddock, A. G., Trans. Faraday SOC.
49, 1150 (1953).
48. McCauley, C. E., Hilsdorf, G. J., Geissler, P. R., and Schuler, R. H., J. Am.
Chem. SOC.78, 3246 (1956).
49. MadClock, A. G., BuU. inform. sci. et tech. (Paris) 2, 3 (1957).
60. Maddock, A. G., and de Maine, M. M., Can. J. Chem. 34, 275 (1956).
61. Maddock, A. G., and de Maine, M. M., Can. J. Chem. 34, 441 (1956).
6.2.Maddock, A. G., and Sutin, N., Trans. Furaday Soc. 51, 184 (1955).
63. de Maine, M. M., Thesis, Department of Inorganic Chemistry, Cambridge
University, England, 1955.
64. de Maine, M. M., Maddock, A. G., and Taiigb61, I<., Discussions Faraday SOC.
23, 211 (1957).
66. Marx, J. W., Cooper, H. G., and Henderson, J. W., Phys. Rev. 88, 106 (1952).
66. Miller, J. M., and Dodson, R. W., J. Chem. Phys. 18, 865 (1950).
67. Miller, J. M., Gryder, J. W., and Dodson, R. W., J. Chem. Phys. 18, 579 (1950).
68. Milman, M., and Shaw, P. F. D., J. Chem. SOC.1957, 1303, 1317, 1325.
69. Mott, N. F., Tra&. Faraduy SOC.
43, 429 (1947).
60. Nath, A., Venkateswarlu, K. S., and Shankar, J., Proc. Indian Acad. Sci. AM,
29 (1957).
61. Nibon, G., Acta. Chem. Scand. 10,94 (1956).
6.8.Overhouser, A. W., Phys. Rev. 90,393 (1953);94, 1551 (1954).
63. Parkins, W. E., Dienes, G. J., and Brown, F. W., J. Appl. Phys. 22, 1012 (1951).
64. Primak, W.,Phys. Rev. 100. 1677 (1955).
66. Primak, W., and Szymanski, H., Phys. Rev. 101, 1268 (1956).
66. Rauscher, H.,and Harbottle, G., J. Znorg. & Nuclear Chem. 4, 155 (1957).
67. Rieder, W., Broda, E., and Erber, J., Monatsh. Chem. 81, 057 (1950).
68. Rowland, F. S., and Libby, W. F., J. Chem. Phys. 21, 1495 (1953)
314
69. Rowland, F. S., Turton, C. N. and Wolfgang, R., J . Am. Chcm. Soc. 78, 2354
(1956).
YO. Schrodt, A. G., and Libby, W. F., J . Am. Chem. SOC.76, 3100 (1954).
71. Schuler, R. H., J. Chem. Phys. 26, 425 (1957).
Y3. Schuler, R. Ef., and McCauley, C. E., J. Am. Chem. Soc. 79. 821 (1957).
73. Seitz, F., The Modern Theory of Solids, p. 494. McGraw-Hill, New York,
1940.
74. Sharman, L. J. and McCallum, K. J., J . Am. Chem. SOC.77, 2989 (1955).
75. Stone, F. S., in Chemistry of the Solid State (W. E. Garner, ed.). Butterworths, London, 1955.
76. Sue, P., J. chim. phys. 45, 177 (1948).
77. Sutin, N., and Dodson, R. W., J . Znorg. & Nuclear Chem. 6, 91 (1958).
78. Szilard, L., and Chalmers, T. A., Nature 134, 462 (1934).
79. Vand, V., Proc. Phys. Soe. (London) 55, 222 (1943).
80. Wang, J. H., in Radioactivity Applied to Chemistry (A. C. Wahl and N. A.
Bonner, eds.), Table 4A. Wiley, New York, 1951.
81. Welch, A. J. E., in Chemistry of the Solid State (W. E. Garner, ed.), p. 307.
Butterworths, London, 1955.
82. Willard, J. E., Ann. Rev. Nuclear Sci. 3, (1953).
83,Williams, R. R., J . Phys. & CoZEoid Chem. 52, 603 (1948).
84. Williams, R. R., Hamill, W. H., Schwarz, H. A., and Burell, E . J., J. Am. Chem.
SOC.74,5737 (1952).
86. Wolf, A. P., and Anderson, R. C., J. Am. Chcm. SOC.77, 1608 (1955).
86. Wolf, A. P., Gordon, B., and Anderson, R. C., J. Am. Chem. SOC.78, 2657 (1956).
57. Wolf, A. P., Redvanly, C. S., and Anderson, R. C., Nature 176, 831 (1955).
88. Wolfgang, R., Rowland, F. S., and Turton, C. N., Science 121, 715 (1955).
89. Yankwich, P. E., Can. J . Chcm. 34, 301 (1956).
90. Yosim, S., and Davies, T. H., J . Phys. Chcm. 56, 599 (1952).
91. Zuber, A., Thesis, Faculty of Pure Science, Columbia University, New York
(1954) ; USAEC Document NYO-6142 BNL.
ACTIVATION ANALYSIS
D. H. F. Atkins and A. A. Smales
United Kingdom Atomic Energy Research Establishment, Harwell, Didcot. Enoland
I. Introduction
. . . . . . . . . . . .
11. Irradiation . . . . . . . . . . . . . .
111. Radiation Sources . . . . . . . . . . .
IV. Experimental . . . . . . . : . . . .
A. Preparation of Samples . . . . . . . .
B. Radiochemical Purification . . . . . . .
C. Source Preparation . . . . . . . . .
D. Activity Measurements . . . . . . . .
E. Gamma Spectrometry . . . . . . . .
F. Measurements During the Activating Collision .
G. Precision and Accuracy . . . . . . . .
V. Conflicting Nuclear Processes . . . . . . .
VI. Applications of Neutron Activation Analysis . .
VII. Activation Analysis with Charged Particles . . .
VIII. Conclusion . . . . . . . . . . . . . .
References . . . . . . . . . . . . . .
. .
. .
. .
. .
.
.
.
.
.
.
.
.
. . . 315
. .
316
. . . 319
. . . 321
. . . . . . . 321
. . . . . . . 322
. . . . . . . 323
. . . . . . . 324
. . . . . . . 326
. . . . . . . 327
. . . . . . . 328
. . . . . . . 331
. . . . . . . 334
. . . . : . . 339
. . . . . .
341
. . . . . .
342
1. Introduction
The determination of elements by means of their natural radioactivity

has been an established analytical method for a number of years. While
it possesses distinct advantages in certain cases, the method is limited
to those naturally occurring radioactive elements which show a readily
perceptible level of activity.
The discovery by Curie and Joliot (19) that radioactivity could be
induced in a number of light elements by bombardment with alpha
particles enabled the range of this method to be extended. Fermi (23)
greatly increased the number of elements that could be activated artificially by using neutrons, which had been slowed down by passage through
water or paraffin wax, as projectiles. It is primarily the method of
activation first investigated by Fermi that has enabled activation analysis
to develop into the highly sensitive analytical tool it is today.
Activation analysis consists in the determination of the amount of an
element present in a sample by measuring the intensity of activity induced
in that element by bombardment with suitable nuclear projectiles. With
some reservations which will be discussed below, the activity induced
315
316
D. H. F. ATKINS AND A. A. SMALES
is proportional to the amount of element present and is independent of

its chemical state. In practice a standard is usually irradiated with the
sample and the measured intensities of their radiation compared. This
technique, known as comparative activation analysis, has the' advantage
that the absolute value of the flux of activating particles and the activation cross section for the nuclear reaction induced need not be known.
In certain cases the activities may be compared without chemical
separation, but frequently the method contains steps designed to isolate
the required element in a pure state prior to measuring its activity.
The necessary chemical operations are normally carried out in the
presence of an inactive isotopic carrier added after the activation.
Whether separation is carried out or not, the mass of the element
S determined in the sample, is given by the equation:
m w of X in sample - radiation intensity from X in sample .
maim of X in standard - radiation intensity from X in standard
Activation analysis was first applied by von Hevesy and Levi t w o

years after the discovery of artificial radioactivity ( 3 6 ) . Determination
of 0.1% dysprosium in rare earth mixtures was made by activation with
neutrons from a 300-millicurie radium emanation-beryllium neutron
source. The 2.3-hr half-life induced activity due to Dy165 was compared
with that induced in mixtures of known dysprosium content. A similar
method was used to determine europium in gadolinium.
Further advances were necessarily limited until the development of the
high-voltage accelerating machine, and, of greater importance, the advent
of the nuclear reactor. The high charged particle or neutron flux obtainable from these sources has extended the range of elements susceptible to
activation analysis and greatly increased the sensitivity of the method.
At present more than 60 elements have been determined successfully,
many of them in the submicrogram region.
The bulk of the published work on activation analysis has been concerned with neutron activation and this, the more general technique,
will be dealt with in some detail. Aotivation with charged particles will
be considered in Section VII.
11. Irradiation
If the sample containing the element to be determined is placed in a

homogeneous flux of neutrons, some may be captured by the target nuclei
to produce radioactive isotopes. These compound nuclei are unstable,
having a definite probability of decay, and some will disintegrate during
the bombardment. As a result, concentration of the radioactive species
ACTIVATION ANALYSIS
317
will increase until it is such that the rate of formation is equal to the
rate of decay.
The rate of formation is given by ~ u Nwhere
,
f is the activating flux
in particles per square centimeter per second, u is the activation cross
section in square centimeters for the reaction, and N the number of nuclei
of the isotope involved. The rate of decay is AN*,A being the decay
constant of the radioactive species of which N* atoms are present. The
over-all rate of growth is then given by:
dN*/dt = fuN - AN*.

Integration over the irradiation period t gives:
N*
= fuN(1
(1)
- e-&')/X.
(2)
At the end of the irradiation the activity At in disintegrations per second
from the N" unstable atoms present is A"', and therefore:
A t = xN* = fuN(1 - e+).

(3)
The decay constant is related to the half-life of the nuclide, l'%,
by
the expression :
X = 0.693/Tlp.
(4)
Substituting for A in Eq. (3)
A t = fgjV(1 - e-O.G93t/TI/t).
(5)
The expression within parentheses is termed the growth or saturation
factor and reaches a maximum value of unity for infinitely long irradiations when
At = At,, = fuN.
(6)
A value of 0.5 for the saturation factor is obtained for irradiations of
one half-life and is proportionately smaller for shorter irradiations.
Irradiation for ten half-lives gives a value of the saturation factor of
0.999, but except in the case of very short-lived isotopes it is seldom
worthwhile to irradiate for periods longer than one half-life, as thereafter
long-lived activities formed by neutron capture in extraneous impurities
in the sample build up at a greater rate than the desired isotope.
Equation 5 can be simplified by replacing the number of atoms N by
W + / M , where W is the weight of the element present, + is the abundance
of the isotope activated, and M the chemical atomic weight. If U, the
activation cross section, is expressed in barns (1 barn is equal to 10-24
square centimeters) it reduces to
A t = W+uf 6.02 X 1028X 10-24(1- e-0.'3w/T1/*)/M
(7)
all variables being related to the final radioactivity of the nuclide formed.
318
It must be stressed that whcn this formula is used to find the total
radioactivity of an irradiated element it, must be applied to each isotope
of the element in turn.
Values for the growth factor are conveniently obtained from a
prepared graph, such BS that of Sniales (8.93,or from a nomogram (96).
As soon as the sample is removed from the activating flux, its initial
activity A. will commence to decay exponentially, the activity A t , , after a
time t, being given hy
-lt,= A&-0.693L/Ti/r.
(8)
With elements of short half-life, the sensitivity of the method, besides
being dependent on the flux and activation cross section, will also depend
on the time which must elapse between the termination of activation and
the measurement of the activity. With short-lived nuclides this period
frequently determines the over-all sensitivity of the method.
Fluctuations in the neutron flux can lead to errors caused by an unequal degree of activation in the samples and standards. The effects may
be minimized by irradiating samples and standards very close together.
More important are variations caused by progressive reduction of the
effective neutron flux through either samples or standards due to the
presence of elements having a high capture cross section for neutrons of
the particular energy range in the irradiation position. T o some extent this
self-shielding effect can be eliminated by taking small samples and
diluting the standards with material of low cross section (e.g., water,
silica, alumina, or sucrose). Some idea of the extent of self-shielding to
be expected can be obtained by means of the equation
(9)
where fo and f are the initial and attenuated flux, N the number of
absorbing centers per cubic centimeter, T the radius of a spherical sample,
and u the total neutron absorption cross section for the complete energy
spectrum.
In a discussion of the errors in neutron activation analysis Plumb and
Lewis (72) deal with the effect of self-shielding by resonance neutrons.
Neutrons emitted during the fission of U235have energies ranging from
0 to about 15 MeV. The energy of the faster neutrons is reduced by
collision with a moderating material (graphite or light or heavy water)
until they have a Maxwellian distribution of energy around 0.025 ev
and are described as thermal. Most of the reactions utilized in activation
analysis are of the ( n , y ) type and are induced by thermal neutron capture.
As the neutron energy is increased from the thermal region, the capture
cross section for the (n,?)reaction decreases as the inverse of the neutron
f = &-N.r
ACTIVATION ANALYSIS
31 9
velocity (1/V law), although a very high cross section is often obtained
over a small energy range (40). These resonance peaks as they are called
are superimposed on the otherwise linear variation of cross section with
1/V, and occur in the intermediate or epithermal region of the energy
spectrum, 0.1 ev to about 50 kev. At higher energies the 1/V law no
longer holds, and the cross section does not change with energy. However, at these higher energies reactions such as (n,p) and (np) become
of increasing importance and these can cause complications, as will be
considered below.
If the element being determined or a constituent of the sample matrix
material has a high capture cross section for thermal or resonance
neutrons, possible errors due to self-shielding cannot be ignored.
The level at which self-shielding errors become significant may be
measured by irradiating increasing weights of the material under investigation either as solids of similar general shape or in solution. After
irradiation the specific activity due to the unstable nuclides in each of the
samples is measured. If this is constant and does not increase with
dilution self-shielding is insignificant. A progressive decrease in specific
activity with increasing sample weight is indicative of self-shielding.
Cabell and Smales (12) used this technique in an investigation of the
self-shielding effect in the caesium and rubidium chloride standards used
in the determination of these elements in materials of geological interest.
In the determination of sodium a t the submicrogram level in lithium
Smales and Loveridge (85) measured the extent of self-shielding resulting
from the high cross section (67 barns) of the matrix material. With
samples weighing 50-100 mg a mean negative error of between 10 and
20% was observed.
The contribution to neutron activation by resonance neutrons alone
can be determined by irradiating two identical samples of the element of
high cross section, one wrapped in thin cadmium foil. Cadmium is practically opaque to thermal neutrons (U = 20,000 barns), so that any
activity induced in the wrapped sample must be due to neutrons of
energies other than thermal.
Self-shielding of resonance neutrons can be minimized by irradiating
in the graphite-loaded column (thermal column) of the pile. Here the
contribution by neutrons of energies greater than thermal is considerably
reduced.
111. Radiation Sources
The eensitivity of activation analysis is directly proportionai to the

particle flux employed. In consequence it is the aim of the analyst inter-
320
D. H. F. ATKINS A N D A. A. SMALES
ested in trace element determinations to obtain the highest possible flux,

which for neutron activation is found together with certain other advantages in the nuclear reactor. Table I contains a list of some research reacTABLE I
NEUTRONFLUXAVAILABLE
FROM SOME
NUCLEARREACTORS
THERMAL
~~
Reactor
Fuel
~~
Flux
Moderator (neutrons/cm*/eec)
BEPO (England)
3NL (Brookhaven,
Natural uranium
Graphite
10'9
U.S.A.)
X-10 (OakRidge,
U.S.A.)
N R X (Canada)
MTR (Idaho,U.S.A.)
LITR (OakRidge,
U.S.A.)
DIDO (England)
Natural uranium
Graphite
10"
Natural uranium
En,riched uranium
Enriched uranium
Graphite
DsO
HsO
6.8 X 10"
5 x 1014
Enriched uranium
Enriched uranium
Hi0
DsO
1.1
10"
4 x 10'8
1 x 1014
tors, with their fuels, moderators, and the thermal fluxes obtainable from
them (13).The value for DIDO has been added.
Particle accelerating machines may also be used in the production of
neutrons by seoondary reactions. In general the steady flux is lower than
that listed for nuclear reactors. If thermal neutrons are required, moderation is usually carried out by allowing the neutrons to pass through paraffin. Table I1 includes some results for a number of typical particle
acceler&om (34, 101).
TABLE I1
NEUTRONFLUXAVAILABLE
FROM ACCELERATINGMACHINES
Type
Cyclotron
Van de Graaff
Cockcrofb
Walton
Betatron
Linear
accelerator
(Ramell)
Particle and
target
Be9
Beg
(d,n)B'O
(d,n)B10
T (d,n) He4
e-U
e-TJ
Neutron
emission
(neutrons/sec)
Approximate
thermal flux
in moderator
(neutrons/cm*/sec)
l m a , 8Mev
Mpa, 2Mev
3 X lOla
1 X 1O1O
1 x 10"
5 x 108
100@,250kev
3pa, 80Mev
1 X lo*
2 x 1010
1 x 108
1 x 108
Ion current
and particle
energy
25 ma, 14 Mev
4X
lOI4
321
ACTIVATION ANALYSIS
Neutron activation analysis using low-level laboratory sources of neutrons has received some attention, although because of the low fluxes available the sensitivity of the method is greatly reduced. Meinke and Anderson
(57') have explored the possibility of using them for precise work. Table
I11 contains results for neutron emission from some laboratory sources
(2,103).
The figures in Table I11 refer to unmoderated neutrons. Sources conTABLE I11
NEUTRON
EMISSION
FROM LABORATORY
NEUTRON
SOURCES
~
Reaction
Pu'JD-Be
Ra**E-Be
PoalO-Be
Sb1"Be
BeO(a,n)C"
Be%,T%)C1*
&@(CX,n)C'f
Be@(r,n)Be*
Half-life
24,400 yr
1,622 yr
138 days
60 days
Neutron emission
(neutrons/curie/sec)
1.4
x 107
1.3 x 107
2.6 X lo6
3.2 X lo6
taining radium and antimony require shielding, because of the intense

gamma radiation emitted. P U ~ ~ ~and
- , 'B
oz1O-Be
~
sources have the advantage in that their decay is practically free from gamma emission, and
consequently much less shielding is required. Po2losuffers from the disadvantage of a short (138 day) half-life. The same disadvantage applies
to the photoneutron source listed, although this may be reactivated by pile
irradiation.
IV. Experimental
A. PREPARATION
OF SAMPLES
In activation analysis, as in all trace element determinations, care
must be exercised to avoid contamination. One of the advantages of activation analysis is that the manipulation prior to irradiation of the sample
is minimbed. Contamination by inactive material after irradiation is unimportant, provided it does not take place in amounts sufficient to affect
the chemical yield. The restrictions imposed on neutron activation by selfshielding mentioned above may necessitate the use of small samples and
either liquid or solid dilutions of the pure standard material to overcome
the di5culty.
In research reactors where the operating temperature is low, short irradiations may be performed in polythene provided that iiquids are not
used. Where long irradiations are necessary, or for small volumes of liquids, it is customary to seal samples and standards in silica ampoules.
322
D.
H. F. ATEINS AND A. A. SMALES
B. RADIOCHEMICAL
P~FICATION
Normally, as soon as possible after completion of the irradiation the
sample is subjected to radiochemical purification designed to isolate the
required product free from foreign activities.
With products of short half-life the time necessary for achieving the
radiochemically pure final source for activity measurement must be as
short as possible if the method is to show high sensitivity. As the activated
nuclide is frequently present a t the microgram or submicrogram level, a
quantitative separation from many times its weight of the sample matrix
material would present great difficulty. Instead it is customary to add inactive isotopic carrier at the milligram level and, after ensuring that chemical exchange takes place between it and the active nuclide, to isolate the
required element by specific chemical steps. These steps need not be quantitative; the percentage recovery of the carrier can be determined a t the
end of the experiment and the measured activity corrected to 100% chemical yield. Care must be taken that exchange between the carrier and the
irradiated material is initially complete.
Radiochemical purification of the induced activity in the presence of
carrier involves chemical operations such as precipitation, distillation,
solvent extraction, chromatography, and ion exchange. While the chemistry performed on the carrier and sample should be designed to isolate the
material in a pure state, a useful operation frequently carried out is scavenging. Strongly adsorptive precipitates such as ferric hydroxide, lanthanum fluoride, and antimony sulfide may be formed in the solution. These
precipitates, by coprecipitation, occlusion, and surface adsorption can be
used to remove unwanted traces of activity. A scavenging agent should be
chosen that will not carry down appreciable amounts of the carrier from
solution. An alternative method is to add hold-back carriers for the
unwanted traces of activity and precipitate the required element in their
presence.
Activation analysis has the advantage over other methods of trace
element determination, in that after irradiation it is insensitive to contamination by inactive material. However, where nuclides of long, or
moderately long, half-life are estimated in a radiochemical laboratory
precautions must be taken to prevent contamination with residual activity from previous analyses.
A number of accounts of methods used in the radiochemical purification
of a variety of elements has been published (17,51, 52, 55, 5 6 ) .
In the chemical separation i t is advisable for the standards to be taken
into solution with carrier, precipitated, weighed for chemical yield deter-
ACTIVATION ANALYSIS
323
mination, and the radioactivity measured in exactly the same form as the
sample. Where pure standard materials are employed it may not be necessary to subject the standards to the entire radiochemical purification. It
is important, however, that the final source should be in the same form
and contain approximately the same weight of material as those derived
from the sample in order to minimize self-scatter and self-absorption errors. If the chemical yields and consequently the source weights differ
considerably, corrections for self-absorption may be made from an experimentally determined curve.
C. SOURCE
PRDPABATION
After the final precipitation stage in the radiochemical purification,
the compound of activated element and carrier is mounted in a form suitable for the measurement of the radiation emitted from the unstable nuclei
present. If the final assay of activity is by beta counting, care must be
exercised in the selection of a suitable compound for the source and in
ensuring that it is evenly spread on the counting tray. The choice of a
compound for the source material is governed by the chemistry of the element and in cases such as the alkali metals, niobium, and tantalum the lack
of suitable compounds is an added complication. Where a number of compounds are available the final choice should be made after consideration
of the following points. The compound selected should be stoichiometric
and should have 8 low solubility in order to prevent excessive losses in the
final precipitation. Elements of high atomic number should be avoided in
the source as far as possible and this is of particular importance with
nuclides emitting soft beta particles (those with a maximum energy of not
greater than 0.5 MeV) . Since it is frequently czonvenient to dry the source
under an infrared heating lamp the compound selected should be or become
thermally stable, a t least to the maximum temperature likely to be encountered. Finally, the dry source should not be loose and powdery so that
it is di5cult to handle or so coarse-grained that an evenly spread sourc6
cannot be obtained.
A number of techniques have been employed for source mounting. The
source material may be slurried on to a weighed circular aluminum tray
with a little alcohol or water. The process need not be quantitative. Solvent
is removed from the tray, conveniently by surface heating under an infrared heating lamp. In another widely used technique the source precipitate
may be deppsited onto a m a l l circle of filter paper by suction, dried, and
fixed to a suitable backing material of low atomic number before counting.
The source may also be deposited onto a metal disk or filter paper circle
fitted into a demountable centrifuge tube. A good method which is un-
324
fortunately not generally applicable is electrodeposition directly anto

the counting tray. Details of these and other techniques will be found in
text books on radiochemistry (15, $4).
L).
ACTIVITY~ ~ E A S U R E M E N T S
It has already been mentioned that difficulties are to be expected in
counting sources of nuclides emitting soft beta particles. These difficulties

increase as the maximum energy decreases, and with beta particles whose
maximum energy is less than 0.2 &lev special mounting and counting techniques are required. However, the majority of determinations can be carried out with comparatively simple and inexpensive equipment. The minimum essential consists of a lead-shielded end-window Geiger-Muller tube,
a power pack able to deliver up to 2000 volts, a preamplifier, and a scaling
unit.
It has been assumed that the chemical procedure used to isolate the
element after the addition of carrier has been successful, and the final product is free from all extraneous activity. Frequently the reliability of this
assumption can be verified without a great deal of difficulty, by the application of some or ali of the following checks: (1) determination of the
half-life; (2) determination of the maximum beta energy; (3) examination of the gamma spectrum.
1. If short-lived nuclides are produced during activation, measurements of the count rate can be made for a period of several half-lives and
decay curves constructed. After corrections for the background and dead
time of the counter, successive counts can be plotted on a logarithmic scale
against time expressed linearly. Provided that the activity is pure and no
daughter activities grow into the sample, the resultant plot will be a
straight line whose slope is related to the half-life of the nuclide, those
derived from samples and standards being parallel. However, frequently
more than one isotope of an element is produced during irradiation. Thus
during the neutron irradiation of antimony, beta activities of half-life 2.8
days due to Sb122and 60 days due to Sb124are formed and a plot of the count
rate against time results in a composite decay curve. In comparative activation analysis, where pure standard activities are obtained, the resolution
of such a decay curve is not usually necessary, but occasionally the presence of foreign activity may necessitate resolution in an attempt at identification. In a limited number of cases encountered in activation analysis
daughter activities may grow into the pure activity isolated and a composite decay curve will result. Smales (81) used fission product Ba14*in the
by neutron activation in materials of geological indetermination of U235
terest. This isotope has a half-life of 12.8 days and decays by beta emis-
ACTIVATION ANALYSIS
325
sion (maximum energy 1.02 MeV) to daughter La140,half-life 40.2 hr,

maximum beta energy 2.20 MeV. The characteristic nature of the growthdecay curve in such cases may actually aid identification.
2. With long-lived isotopes or where the half-lives of two nuclides are
sufficiently close to make resohftion difficult, absorption measurements are
frequently made.
Positive or negative beta particles emitted from unstable nuclides (as
opposed to conversion electrons) have energies ranging from a maximum,
EMAX,to zero with an average energy of approximately one-third of the
maximum. In view of the complex energy spectrum the maximum energy
is usually determined by absorption methods. A number of techniques have
been described, depending on the activity present in the sample and the
required accuracy (21).An additional method may be applied if a check
on the radiochemical purity of the sample rather than an energy determination is sufficient. The absorber curve for the sample may be superimposed over that for the radiochemically pure standard obtained under
the same conditions. If no gross contamination is present the two plots will
be identical in shape.
3. Emission of a beta particle or positron from a parent element often
leaves the daughter in an excited state, excitation energy being lost by the
emission of a monoenergetic gamma ray photon or photons. Measurement
of the characteristic gamma spectrum which is associated with the nuclear
event is often a help in the identification or determination of the nuclide.
Alternatively, examination of the gamma spectrum may be used as an additional check on the purity of the source.
Owing to the low efficiency of detection (rarely more than 1%) of
gamma radiation by a Geiger-Muller tube, a scintillation counter is normally used for this purpose. Scintillation counters employing large crystals
of organic materials such as naphthalene or anthracene have been used as
phosphors with detection efficiencies of better than 10%. A phosphor of
higher atomic number such as sodium iodide, activated by the presence
of about 1% of thallium iodide, will give better efficiency and is usually
employed. The phosphor crystal is mounted in optical contact with a
photomultiplier tube, pulses from which are amplified and fed into a
scaler.
The energy of gamma radiation can, like that of beta particles, be
determined by means of an absorber curve using a scintillation counter
as detector. However, on account of the greater penetrative powers of
gamma radiation, lead rather than aluminum absorbers are used. If sufficient activity is present in the sample a single-channel gamma spectrometer may be employed (14, 68, YO, 71) and the energy spectrum compared
326
with that obtained from standards. If the activity level from the sample
is too low for it to be examined on the single channel gamma spectrometer,
a multichannel instrument can be used (95,@, 104).
The application of some or all of these checks on the purity of the isolated activity coupled with the specific raaiochemical purification render
the over-a11 specificity of activation analysis high.
I n the past the greatest sensitivity attainable in activation analysis has
been found in beta counting. The background of a conventional shielded
beta counter is of the order of ten counts per minute, because of cosmic
radiation and the materials used in the construction of the counter and
castle. The large crystals used for gamma scintillation counters still suffer
from a high background caused by traces of potassium in the crystals and
by the difficulty of shielding them from extraneous radiation. .However,
gamma counting in a single channel of a gamma spectrometer can rival beta
counting in sensitivity and should become of importance in activation
analysis as these instruments become increasingly available.
E. GAMMASPECTROMGTRY
After irradiation it is quite feasible to take the sample into solution in
the presence of carriers and isolate a series of radiochemically pure compounds for counting. This in fact has been done by Smales et al. (86), who
determined 2.5-hr half-life NP5, 12.9-hr half-life Cua4and 5.3-yr half-life
CoB0in rocks, marine sediments, and meteorites by neutron activation.
These workers are currently estimating ten elements in similar materials
(89)
The advantage of this method is obvious where it is essential to obtain
as much information as possible about trace element abundances on limited amounts of irreplaceable material such as meteorites.
Direct estimation of a number of elements without separation is a
perfectly feasible method provided that the activity induced in the matrix
material during irradiation is not excessive. Resolution of decay curves
alone is not likely to be widely applicable, but if this is accompanied by
gamma spectroscopy the utility of the method is considerably extended.
Morrison and Cosgrove (63) determined ZnSB,As76,T V 7 , Fe", and NaZ4
in neutron irradiated silicon by direct gamma scintillation spectrometry.
Smales (84) neutron irradiated samples of a sodium-potassium alloy and,
after allowing two weeks for decay, direct gamma spectrometry revealed
the presence of Cs134(0.59, 0.80 Mev y ) , Ag"O (0.88, 1.36, 1.48 Mev y ) ,
and Rbs7 (1.09 Mev 7 ) . Quantitative analysis gave 100 ppm of silver and
5 ppm of rubidium. I n a separate analysis which included a radiochemical
purification, 0.1 ppm of caesium was obtained. A two-crystal gamma
spectrometer was employed by Iredale (44) in an examination of %uper-
ACTIVATION ANALYSIS
327
pure" aluminum after pile irradiation. With this instrument he was able
to detect and determine manganese, copper, sodium, antimony, and scandium.
Salmon (7'8) has extended the applicability of the method of direct
gamma spectroscopy by the determination of cobalt in iron using gammagamma coincidence measurements. Febg, half-life 46 days, emits gamma
photons of energy 1.12 and 1.30 MeV, while 5.3 yr half-life Coeoemits
gamma photons of 1.16 and 1.32 MeV. Clearly it would be impossible to
differentiate between the two nuclides on a conventional spectrometer.
Fese can decay in two distinct ways, each way being associated with its
own gamma ray photon. Co*O, however, has only one mode of decay and
the gamma photons associated with it are emitted within a very short time
of each other. By counting the coincident pairs of gamma rays Salmon was
able to differentiate between cobalt and iron, and by this method he determined cobalt at the 0.02 to 0.25% level in steels after pile irradiation.
F. MEASUREMENTS
DUEING
THE ACTIVATINQ
COLLISION
The measurement of radiations emitted during the activating collision
is a method that has been applied in a number of cases where the sensitivity of neutron activation analysis is otherwise poor.
One method, of particular interest for the light elements, utilizes the
prompt photoneutron emission which follows gamma irradiation. This
method is not generally applicable without some apparatus for producing
high-energy gamma rays. But, in the cases of beryllium and deuterium, the
thresholds for the (y,n) reaction are sufficiently low to enable a laboratory
source of gamma radiation to be used. Gaudin and Pannell (26) determined beryllium a t the 1-2 ppm level in low-grade ore8 by irradiating the
samples with gamma ray photons from SblZ4(maximum energy 2.11 MeV,
half-life 60 days) and counting the neutrons emitted after moderation in
paraffin wax in boron trifluoride-filled neutron detectors. The threshold
value for the reaction Bea(y,n)2He4is 1.63 MeV. Haigh (S1) utilized the
reaction D2(y,n) H1, gamma threshold 2.2 MeV, for the determination of
deuterium in relatively pure liquids, using a NaZ4Bource of 15.0-hr halflife and maximum gamma energy 2.76 MeV. After moderation in paraffin
the neutrons were counted in boron-lined ionization chambers. Any beryllium, which has a lower gamma threshold, in the samples would emit neutrons under these conditions.
The extension of the techniques of measuring the radiation emitted
during the activating collision to pile irradiation requires a method of detection sensitive to the nuclear event but relatively insensitive to the pile
neutron and gamma flux. Stewart and Bentley (97)estimated uranium
328
extracted from sea water by irradiating the sample in a neutron beam in

the counting chamber of a fission fragment counter.
A more common technique employs a nuclear emulsion to detect the
radiation. The sample is irradiated in close proximity to a sensitive emulsion, which is subsequently developed, fixed, and examined under the
microscope. In this way it is possible to distinguish tracks due to alpha
particles, fission fragments, etc. Faraggi et al. (22) and Mayr (54) used
this technique to determine boron by the B10(n,a)Li7 reaction down to a
level of 2 x lo-%gm. Lithium at the 10-ll-gm level was determined by
Picciotto and van Styvendael (69) by the reaction Lia(n,a)H3 and Curie
and Faraggi (18) studied the distribution of uranium in the surface of
polished minew1 specimens by the U236( n f ) reaction.
G. PRECISION
AND ACCURACY
Before any discussion of the precision and accuracy of activation analysis, it is important that the distinction between them should be appreciatcd. Precision is an index of the reproducibility of repeated individual
analyses, whereas accuracy is a measure of the closeness of the mean value
to the truth.
Although many papers give data from which the precision of the
method they describe can be calculated, with the exception of that of
Seyfang ('79) little work has been done specifically with the object
of attaining high precision in activation analysis. The isotopic content of
depleted and enriched uranium had been determined by Seyf ang and
SmaIes by means of the fission product Ba"O produced during pile irradiation (80). Seyfang, using the same method, showed that the precision
could be improved by paying more attention to the weighing and counting
procedures. The following points were considered :
1. By weighing the uranium oxide samples and barium sulfate sources
to the nearest 0.05 mg the error was '0.1% or less.
2. Precautions taken with the counting equipment enabled repeat
counts from a pulse oscillator to be recorded with a coefficient of variation
of 50.1%.
3. The paralysis time of the counting set up was adjusted to 300 2 2 p
>ec, thereby enabling accurate coincidence corrections to be made.
4. Hysteresis in the Geiger counter was eliminated by placing the
source under the window for several minutes before counting started.
Seven portions of natural uranium as the oxide UaOswere irradiated,
processed chemically after the addition of barium carrier, and the barium
sulfate sources counted for 4 min (approximately 50,OOO counts). The
coefficient of variation on a single result was found to be +0.53% of which
0.44% was due to statistical variations in the count rate.
329
ACTIVATION ANALYSIS
The precision of activation analysis is, when all other variables have
been eliminated, limited by statistical variations in the counting rate.
Consequently the precision will tend to decrease as the amount of element
estimated and the activity present in the final source derived from it becomes smaller. However, statistical variations in the counting rate are
rarely of importance except at extreme sensitivity and then the effect can
be minimized by increasing the counting period.
Determinations of the accuracy of activation analysis are complicated
by the lack of suitable standard materials whose trace element content has
been unequivocally established. Smales et al. (86), in the determination
of nickel, cobalt, and copper in iron meteorites and other materials, tested
the accuracy of their method by the analysis of commercially available
standard steel samples. Satisfactory agreement was obtained between the
TABLE IV
COBALTR E S ~ ON
S STANDARDSTEELS
Co content by
absorptiometry
Alloy steel
number
Published
avemge
CO content
Co content
by activation
Tetraphenyl
m n i c acid
Nitroso-R
salt
25 1
252
253
254
255
256
0.018
0.015
0.012
0.010
0.006
0.016
0.070
0.043
0.031
0.027
0.019
0.031
0.073
0.043
0.070
0.042
0.032
0.029
0.020
0.034
0.030
0.027
0.020
0.031
activation results and the published values for nickel and copper, but in
many cases the cobalt content by activation differed widely from the published figure (see Table IV). Results for the cobalt content determined
by an absorptiometric method using tetraphenyl arsonic acid which confirmed the activation results were published in the same paper. More recently Bagshawe (4) independently confirmed the activation values for
cobalt with results obtained with an absorptiometric method using nitroso-R salt.
While steel samples may be used as reasonable alternatives to iron
meteorites in accuracy determinations they wbuld not be as suitable in
comparisons with silicate rocks. A granite G1 and a diabase W1 have been
suggested as international standards for the determination of the major
and minor constituents of igneous rocks. A number of workers using several separate methods including neutron activation have determined the
330
TABLE V
RUBIDIUMIN G1
W1
AND
Rb found (ppm)
Analyst
Smsles and Webster (94)
Cabell and Smales ( I d )
Herzog and Pinson ($6)
Horstman ($7)
Method
G1
Wl
Stable isotope dilution

Neutron activation
Flame photometry
214
219
216
205
204
21.3
21.4
28.5
19
25
rubidium content of G1 and W1. Table V taken from Smales and Webster
(9.9)contains some published results for the rubidium content.
The accuracy of the activation analysis method when applied to sea
water has been checked by comparison with other methods in the case of
rubidium and strontium. The rubidium content (Table VI) of a sample of
TABLE VI
RUBIDIUMIN SEAWATER
Anal?-st
Method
Rb content
(pglliter)
Smales and Webster

Smales and Salmon

Neutron activation
121.4 f 5
120 f 10
sea water has been investigated by Smales and Salmon (92) using neutron
activation and by Smales and Webster (94) with an isotope-dilution
method.
Hummel and Smsles (42) determined the strontium content of sea
water by neutron activation, isotope dilution using Sr8*, and a tracer
technique with SrBS(see Table VII) .
TABLE VII
STROWTIUM IN S E A W A T E H
Method
Strontium content
(mg/liter)

Neutron activation
Tracer technique
8.1 f 0.1
8.1 f 0.3
8.0 f 0.1
ACTIVATION ANALYSIS
331
From the published results for the accuracy of activation analysis applied to three widely different matrix materials it will be seen that the
accuracy compares very favorably with other modern techniques. The case
of cobalt is particularly interesting, inasmuch as the results obtained by
activation analysis indicated the unreliability of conventionaI analytical
methods when applied to this estimation at low levels.
V. Conflicting Nuclear Processes
Limitations may be imposed on activation analysis by conflicting nuclear processes. These have been mentioned above as becoming of increasing importance in neutron activation when the particle energy increases
above that necessary to bring about the required interaction.
I n the nuclear reactor the normal reaction used in activation analysis
is the ( n , y ) induced by thermal neutrons but there are several routes by
which a nuclide A of atomic number 2 and mass M may be produced:
1. A F - l
2. AF+l
AF
AF
5. AF$.
(n,p)
(n,a)
A$
4. AF+l
(n,2n)
A2
5. A F + l
(r,n>
A?
6.
(7,~)A F
The following are indirect methods of forming the radio-nuclide:
7 ~ A$??;
.
( n , ~ ) A$?-1 + A$
7 b . AFr:
(n,?)
AF-i' -% AF-'
(n,r>
AF
Isotopes of elements of intermediate mass number may be formed by
nuclear fission of heavy elements.
8. Heavy elements (n,f) fission products
The more important of these reactions will be considered below. Reactions 2 , 3 , and 4 are brought about by the fast neutron flux in the pile,
which flux Mellish et al. (59) calculated to be 0.17 of the slow flux in the
center of the Harwell pile (BEPO) . Fortunately, however, the cross sections for these reactions are usually considerably lower than those for
normal ( n , y ) reactions induced by thermal neutrons. Contributions
brought about by (n,p) and ( n p ) reactions can often be greatly reduced,
as mentioned previousIy, by irradiating in the thermal column of the reactor, with some loss of sensitivity. Reaction 4, (n,2n),is produced by
332
energetic neutrons and is not of great importance since its occurrence does
not lead to a change in atomic number. The same argument will apply to
reaction 5, (y,n),which will result in the same mass change as reaction 4.
Reaction 6, ( y , p ) , is potentially more serious as it will result in a change
in atomic number and as in the case of reactions S and 3 it must be recognized as a potential source of interference when one element is being determined in the presence of a considerable excess of an element one or two
units higher in atomic number.
Smales et al. (86)in the determination of nickel, cobalt, and copper by
neutron activation deal with possible interference by neutron-induced reactions other than ( n , y ) . They conclude that interference is unlikely except in the case of cobalt, where the cross section Ni58(n,p)Co58is known
to be high ( 5 8 ) . CoS8is a 71-day half-life positron emitter (0.48 MeV) ,
while Co60 is a 5.3 yr half-life beta emitter (0.31 Mev), so that if a t the
end of the experiment the isolated cobalt activity is counted without energy discrimination, any C O present
~ ~ will be counted as CoSoand lead to
a spurious result. For irradiations not exceeding 70 days, during which
period the growth of activity with time is linear for both isotopes, they
calculate that the ratio of the measured beta activity from CoSs and Coao
derived from equal masses of nickel and cobalt is 0.0007. Such interference
will be within the limits of experimental error, but when nickel to cobalt
ratios are high, interference becomes serious and the authors describe the
following methods for overcoming the difficulty:
(a) The activities of the cobalt isolated from standard and sample are
compared by measuring the areas under the CoSo1.16- and 1.32-Mev peaks
on the gamma spectrometer.
(b) Irradiation in the thermal column of BEPO reduces the CoGoactivity by a factor of about ten, but due to the very much smaller energetic
component of the neutron flux that for Cog8is reduced to a much greater
extent.
The formation of different isotopes of the same element by (n,y) and
(n,p) reactions considered in the previous paragraph presents a more
straightforward problem than that where the same nuclide is formed by
these conflicting reactions. An example of this is the determination of the
A40 content of potassium minerals, the ratio K40/A40 being of importance
in age determinations. Attempts were made to determine the A40 content
by neutron activation, utilizing the reaction A40( n , y )All, but it was found
that the method was subject to prohibitive interference by the (n,p) reaction K*l(n,p)A". The difficulty was overcome by Moljk et al. (60),
who
separated the argon prior to activation. The purity of the separated argon
was checked by irradiating a monitoring sample of the pure gas. The latter
contains the isotope A3Gwhich undergoes (n,y) reaction to give 35-day
333
ACTIVATION ANALYSIS
half-life A37 which decays by K and L electron capture. Any atmospheric

argon present in the gas separated from potassium minerals can be detected by this long-lived activity after the 1.82-hr A41 activity has decayed. With this method it was possible to measure 30-40 X
gm of
A40 at the lower limit.
The determination of sodium in pure aluminum by the reaction NaZ3
(n,y)Na24is complicated by the reaction A127(n,a)Na24brought about by
the fast component of the neutron flux spectrum. Salmon (77) determined
the contribution to the sodium content by this reaction in BEPO by irradiating identical samples and standards normally and in the thermal
column. The difference in the sodium content determined by these two
methods gives the apparent sodium content derived from the (np) reaction. Salmon found this value to be 81 ppm under the conditions of the
experiment.
Although it would be expected that the combined yields of (n,y;P) and
(n,y ;P;n,y) reactions would be very low, interference in highly unfavorabIe cases has been recorded. SmaIes and Pate (91) during the determination of arsenic in germanium dioxide by the A ~ ~ ~ ( As76
n , y )reaction
showed that the following sequences of reactions occur:
GeM
(n,~)
Gegg+ A&
(n,~) A&
and
b
+ A&.
The production of As7'' from germanium does not introduce appreciable
error in the determination at the 0.05 ppm arsenic level when irradiations
are made for 15 hr a t a flux of 10l2 neutrons/cm2/sec. As77, which is also
a fission product, can be differentiated from As16 by counting through an
absorber (approximately 800 mg of aluminum per square centimeter) because of the large difference in their maximum beta energies--AP, 2.97
MeV; A P , 0.7 MeV. However, the sensitivity of the method is somewhat
reduced because of the high background of Bremsstrahlung produced by
the interaction between the AsT70.7-Mev beta particles and the absorber.
A further example of interference caused by a succession of (n,y) reactions followed by beta decay is found in the determination of phosphorus
in silicon by the P31(n,y)PS2reaction when an apparent phosphorus content is caused by
Geg!
(n,~)
Ge:;
:i
S
(n,?)
b
si!: +
:
P
(n,r)
eg.
Kant et al. (60)and James and Richards (47) calculated the phosphorus
that would be produced under the conditions of irradiation. Kant et al.
found that corrections were negligible for irradiation periods of 15 days if
the phosphorus initially present is greater than 0.1 ppm.
334
The presence of uranium in a sample exposed to a flux of thermal neutrons can cause errors if the nuclide or nuclides determined are fission
products or are isotopic with them. Hudgens and Dabagian (39) determined zirconium in zirconium-hafnium mixtures by separating the Nbga,
daughter of Zrs6 formed by n,y reaction, after the addition of carrier.
Contributions .from fission product niobium (Nbe6) can be allowed for
by irradiating a further sample, isolating fission product Ba140 and from
the fission yield curve making allowance for radioactive niobium derived
from any uranium impurity.
A further example of interference by fission is given by Seyfang and
Smales (80)who were faced with the problem of determining the U236
content of mixtures of the naturally occurring isotopes. Contributions to
the fission product Ba140 used in the estimation from the fast fission of
U238
were greatly reduced by irradiating samples in the thermal column of
BEPO. Any decrease in sensitivity was compensated for by increasing the
irradiation period.
VI
Applications of Neutron Activation Analysis
Applications of activation analysis are to be found in all the major

fields where analytical chemistry is employed. Some of the unique properties of the method have rendered its contributions in the field of trace
element analysis invaluable.
Perhaps the greatest single difference between activation analysis and
other trace element methods lies in the virtual freedom of the activation
method from contamination after irradiation. While this is of obvious importance in all fields of trace element analysis, it is of particular importance in the biological field. If conventional analytical methods are used
the first stages in the analysis often consist in the destruction of large
amounts of organic matter with the attendant risk of loss and contamination.
In the determination of sodium, potassium, and phosphorus in milligram samples of human muscle, Stone and Reiffel (98) avoided the possibility of contamination by irradiating intact samples after drying at
100C for about 20 hr. After irradiation no chemical separation was employed, the sodium 2.76-Mev gamma photons and the potassium 3.6-Mev
beta particles being counted through suitable absorbers to eliminate foreign activities. Eight to fourteen days were allowed for the sodium and
potassium activities to decay before P32,half-life 14.3 days, was counted.
A similar technique of irradiating the intact dried sample had been used
by Smales and Salmon (92) in a determination of the rubidium and caesium content of seaweeds and other materials, and a particularly striking
ACTIVATION ANALYSIS
335
example is the determination of arsenic jn broad beans where the pod was
irradiated unopened (90).
Information aa,to the fate of strontium in the animal and plant organism is of particular interest a t the present time. Harrison and Raymond
(33) developed a rapid method for determining the metabolism of strontium and barium without interference from calcium. After short irradiation in BEPO, samples were chemically treated and the isotopes were
gamma counted using a scintillation counter. More recently strontium
and barium have been determined in bone by Sowden and Stitch (96).
Strontium at the 5-60 ppm level has been determined in animal tissue by
Brooksbank, et al. (10) while Bowen (8) has determined strontium and
barium in sea water and marine organisms.
The determination and the investigation of the metabolism of both
toxic and nontoxic elements in the animal and plant organism have been
investigated by activation methods and because of the medical and legal
aspects associated with it, that of arsenic has received much attention. The
distribution of arsenic in hair was first investigated by Griffon and Barbaud (28,$9). No separation was employed by these workers, who determined the location of the ingested element by traversing the length of the
hair with a well-screened Geiger counter. This methbd has the advantage
of leaving the sample intact a t the end of the investigation. However, it is
not entirely free from objection, due to the presence of other induced activities in the material and a quite general technique was described by
Smales and Pate (90,91) who determined the element in biological materials such as hair, nails, urine, and blood from normal human subjects,
and the internal organs of a mouse.
Activation analysis may be regarded as an alternative to tracer techniques in the study of trace element metabolism. The activation method
does not suffer from the disadvantage of possible radiation damage to the
biological system inherent in the use of radioactive tracers. A further disadvantage in the use of tracers has been pointed out by Harrison and
Raymond (33).They state that while giving valuable information as to
the relative retention and distribution of trace elements, no absolute information as to their excretion rates is obtained.
The sensitivity of analytical methods for trace elements needs to be
high, and ideally the reaction or property utilized should be entirely specific. The sensitivity of activation analysis is dependent on the element
estimated and in many cases it is extremely high. Specificity is covered by
three parameters, half-life of the nuclide produced, its maximum beta energy, and its gamma spectrum.
The analytical requirements of the electronics industry engaged in the
production of semiconductor materials such as silicon and germanium and
336
D. H. F. ATKINS AND A. A. SMALFS
the problems associated with production of the high-purity materials required by the atomic energy and other projects, necessitate analytical
methods of high sensitivity and specificity. The characteristics of semiconductor materials are highly dependent on the presence of traces of impurities such as copper, gallium, antimony, and others. Smales (83)has
reviewed some of the analytical methods available for the determination
of trace elements present as impurities in semiconductors. No one method
is suitable for all the problems in this field, but owing to its high sensitivity
and specificity activation analysis has been able to make a valuable contribution.
Arsenic at the 0.1-ppm level in germanium dioxide has been determined by Smales and Pate (91) after neutron irradiation in BEPO, while
the uptake of impurities from a graphite crucible used in the melting of
germanium has been studied by Morrison (62). Traces of copper in germanium were measured by Szekely (100).The standards employed in this
work were artificially prepared mixtures of copper (1 ppm) in germanium.
Copper, zinc, antimony, molybdenum, and the rare earths in germanium
and its oxide at the 0.1-ppm level have been determined by Jakovlev ( 4 5 ) .
Phosphorus in the range 0.004 to 2.0 ppm and arsenic have been determined by James and Richards (46, 47) in the transistor material silicon.
Kant et aZ. (50) also determined iron, copper, zinc, gallium, arsenic, silver,
cadmium, indium, antimony, thallium and bismuth in the range
to
10-6 ppm by neutron activation. A combination of radiochemical separation and direct gamma spectrometry has been used by Smales et al. (87)
for the determination Gf arsenic, antimony, and copper as well as other
elements, at the parts per thousand million level in pure silicon.
Hudgens and Cali (38) have described a determination of antimony at
the parts per million level in zirconium oxide. Distillation and solvent
extraction were used in the radiochemical purification and antimony was
mounted, prior to counting, by electrolysis onto platinum foil. Atchison
and Beamer (3) have determined arsenic, phosphorus, copper, sodium,
strontium, calcium, and chromium in pure magnesium.
A method for the determination of the oxygen content of powdered
metals, and in particular beryllium, has been described by Osmond and
Smales (67). The metal powder is mixed with lithium fluoride and irradiated with nefltrons whcn the reactions below take place:
1. Liz
B.
0'2
--f
+ H;+
+ Hi
F ~ :+ R;
He:
Fl*, which is a 0.65-Mev positron emitter, is isolated, purified, and

mounted as lead chlorofluoride. Good agreement between this and conventional methods was obtained for the oxygen content of beryllium.
ACTIVATION ANALYSIS
337
Activation analysis has been applied to some of the problems connected

with the study of luminescent solids and phosphors. Grillot (30)determined
gm of copper in solids at different stages in the preparation
of luminescent substances and Bancie-Grillot trnd Grillot ( 6 ) determined
chlorine in zinc sulfide. The thallium content of potassium iodide crystals
has been measured by Delbecq et al. (20),and an accuracy of 5% at the
microgram level claimed.
Trace quantities of uranium in aluminum were determined by Mackintosh and Jervis (63).Two methods were described, one utilizing fission
product barium while the other used the reaction UB8(n,y)UaS, half-life
23.5 min. In the latter case, after simple chemistry, uranium was mounted
as the diuranate. Fairly good agreement was obtained between the two
methods, that using U238being the more sensitive and requiring shorter
irradiation periods.
While high sensitivity has been obtained in the examination of pure
materials, a far more rigorous test of the activation method is found in
its application to materials of a more complex matrix. Emission and X-ray
spectrometry and direct spark source mass spectrometry are all restricted
by the lack of suitable standards when applied to materials of complex
composition. Provided that precautions are taken to avoid self-shielding
errors radioactivation is largely independent of the nature of the matrix
material. It is this advantage which has enabled activation analysis to
prove such an invaluable tool in geochemistry.
Cabell and Smales (la) have described the application of neutron
activation to the determination of rubidium and caesium in international
standard rocks, lepidolites, samples from the Skaergaard Intrusion of
East Greenland, and some stony meteorites. Smales and Salmon (92) had
previously determined the same two elements in sea water after a preliminary concentration on an ion exchange resin, and directly in other materials of geochemical interest.
Strontium determinations in soils, sea water, etc., have the same topicality as similar determinations in biological materials. The strontium
and barium content of a number of soils and plants has been determined
by Bowen and Dymond ( 9 ) .They found that strontium was preferentially
absorbed with respect to calcium by plants from most of the soils considered while barium was taken up much less readily.
Neutron activation was used by Smales et aE. (86) to determine the
nickel, cobalt, and copper contents of rocks, marine sediments, and meteorites. In a more recent paper these workers discuss the application of this
method to cosmic and other magnetic spherules. Radioactivation was
particularly suitable in this case as the available weight of samples lay
within the range of 10-250 pg (88).From the nickel/copper, nickel/cobalt,
338
copper/cobalt, and nickel/iron ratios observed they were able to establish

a close similarity between meteoritic material and the spherules obtained
from deep-sea cores. Spherules obtained from River Danube sand and
from the roof of a collecting station off the coast of Sweden gave quite
different ratios and indieated a different origin for these materials.
Gold contents which varied from about 400 pg/meter3 for English
coastal water to about 15 pg/meteF for waters from the Bay of Biscay
were obtained by Hummel using neutron activation (41). The amount of
gold and rhenium in iron meteorites has been determined by Goldberg and
Brown (27). The use of milligram amounts of the elements as standards
necessitated corrections for self-shielding due to the high neutron capture
cross sections of these elements.
The gallium and palladium content of iron meteorites has been investigated by Brown and Goldberg (11). The elements were determined on
separate samples after irradiation in the Argonne pile. Morris and Brewer
(61) modified the method of Brown and Goldberg for the determination
of gallium in blende.
Neutron irradiation has been employed in a number of cases to determine isotope abundance or to investigate the existence of isotopes where
theory has indicated that this is possible.
The determination of isotopic abundance by means of an ( n , p ) reaction
induced by thermal neutrons has been described by Coon (16). Variations
in the He3 content of helium gas depending on its source had been recorded
and Coon was able to confirm these with gas obtained from wells and
from air by means of the reaction He3(n,p)H3whose cross section for
thermal neutrons is approximately 5,000 barns. A search for Si32in natural
silicon was conducted by Turkevich and Tomkins (102).Theory had indicated that Si32might be a beta-stable isotope and occur in natural silicon
in small, undetected amounts. Neutron-irradiated quartz was examined
radiochemically for 25-day half-life P33,the daughter of Si33 formed by
( n , y ) reactions on the sought isotope. However, only P32,
probably formed
from impurities, was detected and assuming a cross section for the Sia2
( n , ~ ) Sreaction
i ~ ~ of 0.05 barns an upper limit of 4 X
results for
the abundance of Si32in natural silicon. Subsequent work has shown Si32
to be an approximately 300-yr half-life beta emitter.
The geochemistry of uranium and thorium has excited considerable
interest on account of their strategic importance. Smales determined uranium in rocks by neutron activation followed by isolation of fission product BaI4O (81). Interference from the fast fission of any thorium present
in the sample and from beta-emitting barium isotopes formed by ( n , y )
reaction is discussed and methods of overcoming the difficulties are described. The uranium content of two iron meteorites was determined by
ACTIVATION ANALYSIS
339
Reed and Turkevich (73). Uranium was estimated as fission product BaI4O
and as the beta-emitting NpZasformed by the reaction U233R
(n,y)U2aafollowed by beta decay to Np23e.Approximately 100 cpni of the long-lived
alpha emitter Np237were added to enable a calculation of the neptunium
chemical yield. Stone meteorites were examined by Hamaguchi et al. (32)
using the same technique. In addition barium was determined in the
sample.
Small quantities of thorium in geoiogical samples have been determined
by Jenkins (48). Thorium was estimated as the beta-emitting isotope
Th232,half-life 22.1 min., formed by irradiation in BEPO. In view of the
short half-life a rapid procedure for chemical purification had to be developed if adequate sensitivity was to be obtained. Chromatographic
separation on an aluminum oxide-cellulose column was the method successfully applied. Thorium in stone meteorites has been estimated by Bate
et al. (7). Pa233,the 27.4-day half-life daughter of Th233,was used in this
estimation. Although this resulted in a reduced sensitivity, greater time
was available for the chemical processing. Rona (75) measured the isotopic ratio Th232/Th230in uranium minerals. Thorium was separated from
the uranium mineral and the Th232content was determined by activation
analysis. Th230was measured by alpha-particle counting.
VII. Activation Analysis with Charged Particles
It has been stated that the bulk of the published work on activation
analysis has been concerned with the more general technique of neutron
activation. A low probability of neutron activation or an overlong or overshort half-life of the activated species produced can, however, result in
poor sensitivity for the method. This is the case with a number of the
lighter elements such as beryilium, aluminum, carbon, nitrogen, and
oxygen. I n a number of these cases a satisfactory alternative may be found
in activation by charged-particle bombardment (65).
The problems in this field of activation analysis are in many respects
more complicated than those found in neutron activation. The positive
potential barrier around the atomic nucleus prevents interactions below a
certain minimum threshold value. Above this value a law of the 1/V type
which holds for neutrons is not operative; the cross section of the reaction
increases rapidly with particle energy until, as with neutrons of higher
energies, new competing reactions become significant and the reaction
yield falls off. Reactions with charged particles which involve the ejection
of a further charged particle from the compound nucleus usually have a
lower probability of taking place than those which result in the production
of neutrons, since the charged particles formed from such interactions must
340
D. H. F. ATKINS AND A. A. SMALEG
penetrate the electrostatic barrier to escape from the parent nucleus.

Deuteron reactions of the (d,n)type usually have a higher cross section
than those induced by protons or alpha particles.
The simultaneous occurrence of several different nuclear reactions
which lead to a variety of products is a feature of activation by charged
particles. This may be exemplified by the following reactions which may
occur in deuteron bombardment:
A2
(d,t)
AF-'
2. A F
(d,p)
AZ+'
3. AF
(dp)
A~Z?
1.
(d,n)
AFZ
5 . AF
( 4 2 ~ 1 ) AF+I
For a given energy of bombardment all these reactions tend to take
place to a certain extent. At low energies the (d,n) reaction frequently
predominates, while the ( d , p ), (d,a), (d,t) , and (d,2n) reactions become
increasingly important as the deuteron energy is increased. This complexity compares unfavorably with thermal-neutron activation where
frequently one reaction occurs to the virtual exclusion of all others.
Self-shielding during irradiation is of much greater importance with
charged particles, whose range in matter is short, than with neutrons. It is
customary to use targets suf6ciently thick to absorb the whole of the
radiation, but even if this technique is empl6yed the physical shape and
composition of samples and standards is of great importance. Because the
whole of the particle beam is absorbed in the top few millimeters of the
sample large amounts of heat are released and an efficient system of cooling must be devised if the sample is not to be affected ( 7 4 ) .
The protons, deuterons, or alpha particles used are obtained from accelerating machines, although lower fluxes of alpha particles may also be
obtained by the decay of suitable nuclides. The particle flux obtainable
from an accelerating machine may be estimated from the ion beam currents quoted in Table 11. One microampere of protons or deuterons will
correspond to 6.3 x 10l2 ions per sec, while the same beam current of
alpha particles will contain one-half of this number.
The large amount of energy released in charged-particle bombardment
tends to exclude this method where thermally labile organic or biological
materials are employed as targets. However, Odeblad (64) was able to
determine sodium in urine and other elements in biological materials by
means of the weak source of alpha particles (160 millicuries) derived from
the decay of Poll0. A similar source of alpha particles has been used to
2.5 min (66).
determine aluminum by the reaction A12' (~t,n)P~~-half-life
4. A F
ACTIVATION ANALYSIS
341
Boron, which under these conditions gives 10.3-min half-life N13; and
magnesium, which gives 2.4-min half-life A128, should be absent.
Due to the long half-life of C14 it is not feasible to determine carbon
by neutron irradiation. Sue has used the reaction C12(d,n) N13 for its determination in organic matter (99).Samples of approximately 100 pg were
sealed in silica microtubes and irradiated for periods of 5 min in the
deuteron beam from a cyclotron when the following reactions occurred:
1. 016(d,n) F" half-life 1.1 min
2. Si30(d,p)SiS1 half-life 2.65 hr
3. C12(d,n)N13 half-life 10 min
Fifteen minutes were allowed for the fluorine activity to decay before
plotting the decay curve of N13 and S P . The curve was resolved and it
was found that 100 cpm of N13 corresponded to 0.31 pg of carbon. Activity
due to $53' was used as a measure of beam strength in this experiment.
Albert et al. (I), in a determination of carbon in steel by this method,
separated the nitrogen formed by Kjeldahl distillation and liquid-counted
the nitrogen in the distillate. They state that the sensitivity is improved
by this method.
Deuterons of 3 Mev were used by Sagane et al. (76) to reveal the presence of 10 ppm of sodium in aluminum.
The potential value of high-energy electron-producing machines such
as the linear accelerator for activation analysis must not be overlooked.
Here photonuclear reactions (y,n) can be used, either to produce a high
neutron flux (as most charged-particle machines can by choice of a suitable reaction), or directly on samples. This may well be valuable, particularly for some light element determinations.
The direct determination of oxygen in organic materials and metals
by the reaction OlB(y,n)Ol5,half-life 2.1 min, 1.7 Mev positron, using a
22 Mev betatron has been described and the possibility of extending the
method to carbon and nitrogen discussed by Basile and co-workers (6).
Six milligrams of oxygen in organic materials and 0.1% in aluminum has
been quoted as a lower limit for the sensitivity of the method.
VIII. Conclusion
Activation analysis is a powerful addition to the armory of modern

analytical chemistry. Its sensitivity for many elements is already very
high, as can be seen from a few examples listed in Table VIII (and of
course the higher fluxes now becoming available with the development of
more advanced reactors means even higher sensitivity) adding a new dimension to the older methods and being matched at present only by the
342
D. H. F. ATKINS A N D A. A . SMALES
TABLK VIII
ESTIMATED SENSITIVITISFOR SOME ELEMENTS IN
Element
DS
Eu
Au
In
&In
AS
La
Cu
Ga
Ta
h'a
CS
co
P
Rb
Ba
Hg
c1
U
M0
Ki
Cr
Sr
cd.
S
Radionuclide
formed
Half-life
2.3 hr
9.2 hr
2.7 day
54 min
2.6 hr
26.5 hr
40 hr
24.1 hr
12.8 hr
14.1 hr
111 day
1.5 hr
2.3 yr
5.3 yr
14.3 day
18.6 day
85 min
47 day
37.3 min
12.8 day
68 hr
2.6 hr
27.8 day
50 day
164 day
87.1 day
NEUTRON
ACTIVATION
Estimated
sensitivity*
(gm)
1x
1x
5x
1x
1x
5x
5x
5x
1x
1x
10-12
10-12
10-12
10-1'
10-11
10-11
10-11
10-11
10-10
5x
5x
5x
5x
1x
1x
5x
5x
1x
1x
I x
1x
1x
5x
10-10
10-10
10-10
10-10
10-9
10-9
10-9
10-9
10-5
10-10
1 x 10-10
1 x 10-10
10-8
10-7
10-7
10-7
10-7
* Estimated using slow neutron flux of lO1f neutrons/cmg/sec, irradiation period

of one month or saturation, whichever is less, a Z h r period for radiochemical processing,
and 100% chemical yield (49).
mass spectrometer-isotope dilution method. But there is even another advantage over this latter method, an advantage which becomes increasingly
important with increasing sensitivity ; freedom from contamination and
reagent "blank" difficuities. Only those who have actual experience with
this problem when trying to operate a t the submicrograni level can really
appreciate the importance of the contribution made by activation analysis.
REFERENCES
1. Albert, P., Chaudron, G., and Sue, P., Bull. SOC. chim. France. 20, C97 (1953).
2. Anderson, H. L., Nuclear Science Series, Prelim. Rept. No. 3. National Research
Council, Washington D. C, 1948.
ACTIVATION ANALYSIS
343
3. Atchison, G. J., and Beamer, W. H., Anat. Chem. 24, 1812 (1952).
4. Bagshawe, B., Proc. Congr. 012 Modern Anal. Chem. in Znd, 1967 p. 21, Heffer,
Cambridge ( 1957).
6. Baneie-Grillot, M., and Grillot, E,Compt. rend. 237, 171 (1953).
6. Basile, R., Hurk, J., Lbvhque, P., and Schuhl, C., Compt. rend. 239, 422 (1954).
7. Bate, G. L., Huienga, J. R., and Potratz, H. A., Science 126, 3274 (1957).
8. Bowen, H.J. M., J . Marine Biol. Assoc. UX.35, 451 (1958).
9 . Bowen, H.J. M., and Dymond, J. A., Proc. Roy. SOC.Bl44, 355 (1955).
10. Brooksbank, W . A., Leddicote, G. W., and Mahlman, H. A., J. Phys. Chem. 57,
815 (1953).
11. Brown, H., and Goldberg, E., Science 109,347 (1949).
12. Cabell, M. J., and Smales, A. A., Analyst 82,390 (1957).
13. Catalogue of Nuclear Reactors Atomic Energy of Canada, Ltd., Rept. No.
C.RB.490. 1955.
14. Connally, R. E., and Leboeuf, M. B., Anal. Chem. 25, 1095 (1953).
15. Cook, G. B., and Duncan, J. F., Modern Radiochemical Practice. Oxford
Univ. Press, London and New York, 1952.
16. Coon, J. H., Phys. Rev. 75,1355 (1949).
17. Coryell, C. D., and Sugarman, N., The Fission Products, Book 3, Nat. Nuclear
Energy Ser., Vol. IV-9.McGraw-Ell, New York, 1951.
18. Curie, I., and Faraggi, H., Compt. rend. 232, 959 (1961).
19. Curie, I., and Joliot, F.,Compt. rend. 198, 254, 559, 2089 (1934).
20. Delbecq, C.J., Glendenin, L. E., and Yuster, P. H., Anal. Chem. 25, 350 (1953).
21. Duncan, J. F., and Thomas, F. G., Nucleonics 15, (lo), 82 (1957).
99. Faraggi, H., Kohn, A, and Doumerc, J., Compt. rend. 235, 714 (1952).
23. Fermi, E.,Amaldi, E., and DAgostino, O., Proc. Roy. SOC.Ale, 483 (1934).
24. Friedlander, G., and Kennedy, J. W., Nuclear and Radiochemistry, Wiley,
New York, 1955.
25. Gallagher, J. D., McKibben, J. L., and Lang, H. J., U. S. Atomic Energy Comm.,
Rept. No.LA-1917 (1965).
26. Gaudin, A. M., and Pannell, J. H., Anal. Chem. 23,1261 (1951).
37. Goldberg, E., and Brown, H., Anal. Chem. 22,308 (1950).
98. Griffon, H.,
and Barbaud, J., Ann. pharm. franp. 9,545 (1951).
29. Gri!Ton, H.,
and Barbaud, J., Compt. rend. 232, 1455 (1951).
30. Grillot, E., Compt. rend. 234,1775 (1952).
31. Haigh, C. P., Nature 172,359 (1953).
3%Hamaguchi, H., Reed, G.W., and Turkevich, A., Geochim. et Cosmochim. Acta
12,337 (1957).
33. Harrison, G. E., aJld Raymond, W. H. A., J . Nuclear Energy 1,290 (1955).
34. Havens, W. W., Jr., Proc. Intern. Congr. ma Peaceful Uses of Atomic Energy,
Geneva, 1955 4,74 (1956).
36. Herzog, L. F., and Pinson, W. H., Geochim. et Cosmochim. Acta 8, 295 (1955).
36. von Hevesy, G., and Levi, H., Kgl. Danske. V i d m k a b .Sebkab. Mat.-fys. Medd.
14, 5 (1936).
37. Horstman, E. L., A n d . Chem. 28,1417 (1956).
38. Hudgens, J. E., and Cali, P. J., Anal. Chem. 24, 171 (1952).
39. Hudgens, J. E., and Dabagian, H. J., Nucleonics 10, (5),25 (1952).
.do. Hughes, D.J., and Harvey, J. A,, Neutron Croas Sections, Rept. No. BNL 325.
U. 8. Govt. Printing Office, Washington, D. C., 1955.
41. Hummel, R. W., Analyst 82, 483 (1957).
344
42. Hummel, R. W., and Smales, A. A., Analyst 81,110 (1956).

49. Hutchinson, G. W., and Scarrot, G. G., Phil. Mag. [71 42, 792, 957 (1951).
&. Iredale, P., Atomic Energy Research Establishment, Harwell. Rept. No.
A.E.R.E. EL/M 98,1957.
46. Jakovlev, J. V., Proc. Intern. Conf. on Peaceful Uses of Atomic Energy, Geneva,
1966 1 5 , s (1956).
46. James, J. A., and Richards, D. H., Nature 175, 769 (1955).
47. James, J. A., and Richards, D. H., Nature 176, 1026 (1955).
48. Jenkins, E. N., Analyst 80, 301 (1955).
49. Jenkins, E. N., and Smales, A. A., Quart. Revs., 10, 83 (1956).
60. Kant, A., Cali, P. J., and Thompson, H. D., AnaLChem. 28, 1867 (1956).
61. Kleinberg, J., U. S. Atomic Energy Comm., Rept. No. LA-1566 (1953).
68. Kleinberg, J., U.S. Atomic Energy Comm., Rept. No. LA-1721 (1964).
53. Mackintosh, W. D., and Jervis, R. E., Atomic Energy of Canada Ltd., Rept. No.
CRDC-704, 1957.
54. Mayr, G., Nucleonics 12, (5), 58 (1954).
66. Meinke, W. W., U. S. Atomic Energy Comm., Rept. No. AECD-2738 (1949).
66. Meinke, W. W., U. S. Atomic Energy Comm., Rept. NO. AECD-3084 (1951).
57. M e i k e , W. W., and Anderson, R. E., Anal. Chem. 25, 778 (1953).
68. Mellish, C. E., and Payne, J. A,, Yature 178, 275 (1956).
59. Mellish, C. E., Payne, J. -4., and Otlet, R. L., Intern. Conf. on Rndioisctopes in
Industry, Paris. Pre-print UNESCO/NS/RIC/189, 1957.
60. Moljk, A., Drever, R. W. P., and Curran, S. C., Nucleonics 13 (2), 44 (1955).
61. Morris, D. F. C., and Brewer, F. M., Geochim. el Cosmochirn. Acta 5, 134 (1951).
6% Morrison, G. H., Nucleonics 11 (1),28 (1953).
63. Morrison, G. H., and Cosgrove, J. F., Anal. Chsm. 27, 810 (1955).
64. Odeblad, E., Acta Radwl. 42,391 (1954).
65. Odeblad, E., Acta Radwl. 45,396 (1956).
66. Odeblad, E., and Odeblad, S., Anal. Chim. Acta IS, 114 (1956).
67. Osmond, R. G., and Smales, A. A,, Anal. Chim. -4c'n 10, 117 (1C51).
68. Owen, R . B., Atomics 4 ( l ) , 5; 4 (2),34 (1953).
69. Picciotto, E., and van Styvendael, M., Compt. ,end. 232, 855 (1951).
70. Pierson. D. H., Brit. J. Appl. Phys. 6,444 (19%).
7f. Piereon, D. H., Nature 173,990 (1954).
76. Plumb, R. C., and Lewis, J. E., Nucleonics 13, ( 8 ) , 42 (1955).
73. Reed, G. W., and Turkevich, A,, Nature 176, f91 (1955).
74. Reld, A. F., Rev. Sci. Znstr. 18,501 (1947).
76. Rona, E.. Trans. Am. Geophys. Union 38,754 (1957).
76. Sagane, R., Eguchi, M., and Shigeta, J., J. Phys.-Math. SOC.Japan 16, 383 (1942).
77. Salmon. L., Atomic Energy Research Establishment, Harwell, Rcpt. No.
A.E.R.E. C/R 1324,1954.
78. Salmon, L., Atomic Energy Research Estab!irhment, Harwell, Rept. X o . A.E.R.E.
C/R 2377(3), 1958.
79. Seyfang, A. P.. Analyst 80.74 (1955).
SO. Seyfang, A. P., and Smales, A. A., Analyst 78,394 (1953).
Sf. Smales. 4 . A., Atialyst 77, 778 (1952).
8g. Smales, A. A., Atomics 4, 55 (1953).
83. Smales, A. A . , J . Electronics 1, 327 (1955).
S4. Smales, A . A.. A o c . Intern. Conf. on Peaceful I'ses of Atomic Energy, Geneva,
1955 9,273 ( 1956).
ACTIVATION ANALYSIS
345
86. Smales, A. A., and Loveridge, B. A., Anal. Chim. Acta. 13, 566 (1955).
86. Smales, A. A., Mapper, D., and Wood, A. J., Analyst 82, 75 (1957).
87. Smales, A. A., Mapper, D., Wood, A. J., and Salmon, L.,Atomic Energy Research
Establishment, Hamell, Rept. No. A.E.R.E. C/R 2254, 1957.
88. Smales, A. A., Mapper, D., and Wood, A. J., Geochim. et Cosmochim. Acta 13,
123 (1958).
89. Smales, A. A., Mapper, D., and Wood, A. J., Private communication. 1958.
90. Smales, A. A., and Pate, B. D., Analyst 77, 196 (1952).
91. Smales, A. A., and Pate, B. D., Anal. Chem. 24,717 (1952).
9.2. Smales, A. A., and Salmon, L., Analyst 80, 37 (1955).
93. Smales, A. A., and Webster, R. K., Anal. Chim. Acta. (in press) ; Atomic Energy
Research Establishment, Harwell, Rept. No.A.E.R.E. C/R 2204, 1957.
94. Smales, A. A., and Webster, R. K., Geochim. et Cosmochim. Acta 11, 139 (1957).
96. Sowden, E. M., and Stitch, S. R., Biochem. J . 67, 104 (1957).
96. Stehn, J. R., and Clancy, E. F., Nucleonics 13 (4), 27 (1955).
87.Stewart, D. C., and Bentley, W. C., Science 120, 50 (1954).
98. Stone, C. A., and ReitTei, L., J . Lab. C&n. Med. 49, 286 (1957).
99. Sue, P., Compt. rend. 237, 1696 (1953).
100. Szekely, G., Anal. Chem. 26, 1500 (1954).
101. Taylor, T. I., and Havens, W. W., Jr. in Physical Methods in Chemical Analysis
Vol. 111. (W. G. Berl, ed.). Academic Press, New York, 1956.
109. Turkevich, A., and Tomkins, A., Phys. Rev. 90, 247 (1953).
103. Wattenburg, A., Phys. Rev. 71, 497 (1947).
104. Wilkinson, D. H., Proc. Cambridge Phil. SOC.45, 3 (19B)
THE PHOSPHONITRILIC HALIDES AND THEIR DERIVATIVES
N . L. Paddock
. .
and H T Searle
Albright and Wilson (Mfg.) Ltd., Oldbury. Birmingham. England
I . Introduction
. . . . . . . . . . . . . . . . . . . . .
I1. The Preparation of the Phosphonitrilic Halides

A. General
. . . . . . . . . . .
B. Cyclic Polymers . . . . . . . . .
C . Linear Polymers . . . . . . . . .
D . The Course of the Reaction . . . . .
I11. Properties .
.
.
.
.
.
.
.
.
.
.
I V . Structure . . . . . . .
A . General
. . . . . .
B . Experimental . . . . .
1. Trimeric Compounds .
2 . Tetrameric Compounds
3 . Medium Polymers . .
a . General . . . . .
b . Spectroscopic Data .
4 . High Polymer
. . .
C . Structural Theory . . .
. . . . . . 349
. . . . . . 351
. . . . . . 352
. . . . . . . . .
. . . . .
. . . . .
. . . . .
. . . . .
Amines . .
. . : . .
. . . . .
357
. . . . 357
. . . . 359
.
.
.
.
.
.
.
.
.
.
. . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
359
359
360
362
363
364
. . . . . . . . . . . . . . 365
. . . . . . . . . . . . . . 365
. . . . . . . . . . . . . . 366
. . . . . . . . . . . . . . 366
.
.
.
.
.
.
.
.
.
.
V . Discussion . . . . . . . . .
A. Further Aspects of the Theory .
B. Charge Distribution Effects . .
C . Geometrical Effects . . . . .
D . A General Elimination Reaction
. .
. .
. .
. .
. .
. .
. . .
. .
. .
. .
.
.
.
.
.
.
.
.
.
.
. . . . . . 367
. . . . . . 368
. . . . . . 368
. . . . . . 369
. . . . . . 372
. . . . . . 372
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . 375
. . . . . 375
. . . . . 375
.
. 377
. . . . . 378
. . . . . . . . . . . . . . . . . . . . 379
List of Symbols
References
. . . . . . 348
. . . . . . 348
. . . . . . . . . . . . . . . . . . . . . 353
. . . . . . . . . 353
. . . . . . . . . 354
A . General
. . . . . . . . . . .
B . Physical Properties . . . . . . . . .
C. Chemical Properties . . . . . . .
1 Polymerization
. . . . . . . .
2. Reactions with Donor Molecules . .
a . Hydrolysis . . . . . . . . .
b . Esterscation . . . . . . . .
c . Ammonia and Primary and Secondary
d . Tertiary Amines and Donor Solvents
3 . Alkyl and Aryl Derivatives . . . .
4 . Reactions with Acceptor Molecules .
V I . Conclusion
.
.
.
.
.
348
. .
347
. 380
381
348
N. L. PADDOCK AND H. T. SEARLE
1. Introduction
The existence and stability of homologous series of compounds lie a t

the foundations of organic chemistry. The systematic variation of properties within the saturated and unsaturated series, both linear and cyclic,
provides a firm basis for the development and criticism of stereochemical and thermochemical concepts, and for theories of chemical reactivity.
In the development of inorganic chemistry, on the other hand, homologous series have played a less important part. The only well-defined
series which have been studied are the siloxanes (R2SiO), and the phosphonitrilic chlorides (PNC12)", based respectively on the isoelectronic
repeating units SiO and PN. They each form extensive series of both
cyclic and related linear polymers. Each includes an element of principal quantum number 3, a feature which accounts for some resemblances
between the two groups, and which requires an extension of the ideas
derived from the organic field.
The siloxanes have been dealt with in detail elsewhere (15, 65) ; this
review is concerned almost exclusively with the phosphonitrilic halides.
The chlorides were first recognized as a homologous series by Stokes
(85, 89),who isolated and investigated the lower members, from trimer
to heptamer. This, and more recent work, was reviewed in 1943 by
Audrieth et al. (3).
The object of the present review is to correlate and interpret B
growing body of chemical and physical information, to show how the
more recent developments provide a context in which the older work
can be understood, and to relate the chemistry of the phosphonitrilic
series to other groups of organic and inorganic compounds.
II. Preparation of the Phosphonitrilic Halides
A. GENERAL
Of the various preparative methods described by Audrieth et al. (3)!

that originally developed by Schenk and Romer (71) has proved most
generally useful. It involves the ammonolysis of phosphorus pentachloride according to the equation
PCls
+ NH,C1+
PNCli
+ 4HC1
(1)
The reaction is carried out by heating phosphorus pentachloride with

a small excess of finely divided ammonium chloride in refluxing s-tetrachloroethane, a solvent which boils a t a suitable temperature (146") and
which is comparatively inert to chlorination. The initially vigorous re-
PHOSPHONITRILIC DERIVATIVES
349
action slowly subsides, until after 6-8 hr the evolution of hydrogen

chloride is barely noticeable. The excess of ammonium chloride is filtered
off, and the solution concentrated under reduced pressure a t as low a
temperature as possible. The residue is a mixture of oil and crystals,
the bulk of which, typically 75-90%, is soluble in low-boiling petroleum
ether, and consists of a mixture of polymers (PNCL).. The insoluble
portion, a product of incomplete reaction, has the composition (PNC12),
PC15, and depending on reaction conditions may vary in texture from a
light oil to a plastic rubber.
The reaction may also be carried out in monochlorobenzene or
o-dichlorobenzene (18), though complete reaction in the former solvent
may require more than 24 hr. High yields of the lower polymers may
also be obtained from the dry reaction of ammonium chloride and
phosphorus pentachloride (84).
The question of whether the phosphonitrilic chlorides have cyclic or
linear structures has been frequently discussed in the literature. The
structural evidence will be considered in Section IV, where it will be
seen that the cyclic formulation is established beyond doubt for the
trimer and tetramer, and is highly probable for the higher soluble polymers; it will meanwhile be assumed for all compounds of formula
(PNX2),. For reasons considered in Section 11, C the petrol-insoluble
fraction (PNClz), PCls is believed to consist of a mixture of linear
polymers. The two series are related, in that the linear compounds of
low molecular weight are intermediates in the formation of the cyclics
by the ammonolysis of phosphorus pentachloride.
B. CYCLICPOLYMJGRS
Most attention will be paid here to the well-defined series (PNC12).
and related halides. The relative proportions of the individual chloride
polymers in the soluble fraction vary with reaction conditions, but are,
typically, trimer 40-45%, tetramer 15-20%, pentamer 15-20%, hexsmer
1-3%, heptamer and upwards 25% (average). They may be separated
by a combination of fractional crystallization and fractional distillation
in vacuo. This method was first used by Stokes (89),who isolated the
trimer to hexamer as crystalline solids, and the heptamer as an oil. The
undistillable residue obtained after removal of the heptamer consists
of a mixture of polymers, approximately (PNCI?)10-11. This has similar
properties to the pure compounds of lower molecular weight, and is probably composed of molecules with similar structures.
One 'other related compound was isolated by Stokes (89),of formula
P,N,Cle, almost certainly polycyclic and with the structure shown in
formula I. It occurs with the hexameric chloride, which is approximately
3Fio
N. L. PADDOCK A S D H. T. SEARLE
twenty times as abundant. Its low polarity has been demonstrated by

Krause (46).
CI
C1
P
''
(1)
A comparable preparative method can be usea for the phosphonitrilic

bromides and chlorobromides. The trimeric and tetrameric. bromides
(PNBr2)a,rmay be prepared by the action of ammonia ( 6 ) or of ammonium bromide (8) on phosphorus pentabromide, though the dissociation of the latter a t any convenient reaction temperature reduces the
yield considerably, and the two polymers are difficult to separate. The
mono- bromo-compound P3N3C15Br, from ammonium bromide and phosphorus pentachloride, is the first derivative in which the chlorine atoms
are substituted other than in pairs (64). The dibromo and tetrabromo
derivatives were obtained in the same investigation, by the use of
phosphorus chlorobromides.
The phosphonitrilic fluorides do not rcsuIt from the interaction of
ammonium fluoride with phosphorus pentachloride ; ammonium hexafluorophosphate is formed instead (47). They may be prepared indii ~ x t l yby the reaction of potassium fluorosulfite with the chlorides at
10oo-12o0(75),
(PXClz),
+ SKSO?F
--t
(PNFe),
+ 8KC1 + 8SO2
(2)
Tields are good, and the metliod can be extended to the higher polymers.
The pentameric fluoride, for instance, has been obtained by D. H. Paine
(2)
Fluorination of the trimeric chloride with lead fluoride yields mixctl
derivatives of the tetramer, P4N4F&12 (74) and P,N,F,Cl4 ( 7 3 ) . Both
compounds polymerize under pressure at 300" to rubbers; pyrolysis of
the high polymer a t atmospheric pressure gives the mixed chlorofluorides
P3N3F2C14and P3N3F4C12 ( 7 3 ) .
No iodides have been reported. From a solution of triphosphonitrilic
chloride in acetone, sodium iodide precipitates, in the course of months,
all the chlorine as sodium chloride ( 7 6 ) .The color of iodine is, however,
9
351
seen in the solution almost from the beginning, and the phosphonitrilic
iodides are evidently unstable, possibly for steric reasons.
C. LINEARPOLYMERS
Incomplete ammonolysis of phosphorus pentachloride gives a high
yield of the petrol-insoluble oil of empirical composition (PNCIz),, PCls
referred to in Section 11, A. A similar mixture of linear polymers may
be prepared by the action of phosphorus pentachloride on the cyclic
polymers at 350" ( 1 ) . The reaction proceeds a t a comparable rate t o
the polymerization which would occur in the absence of the pentachloride,
and presumably depends in the same way on thermal opening of the
ring. The product is a pasty solid for n = 4 or less; higher members are
oils increasing in viscosity with n, until rubbery solids are eventually
formed (H. T. Searle, 8).The individual members of the series have not
yet been obtained pure, since their thermal stability is not high, and
phosphorus pentachloride easily splits off in even slightly polar solvents.
They are, however, distinguished as a class from the cyclic polymers
by a high polarity, by a reactivity to water of the same order as that
of phosphorus pentachloride itself, and by the appearance of two extra
peaks in their ultraviolet absorption spectra (Fig. 1).These peaks vary
4.(
3.(
EI%
I cm
2.(
I.(
250
300
350
400
450
Wavelength ( m p )
FIO.1. The ultraviolet absorption spectra of two samples of insoluble oil, of

composition PCI,(NPCId,,CI, (of different molecular weights), compared with the
spectra of (PNCId,.
352
N. L. PADWCK AND H. T. SEARLE
in intensity for samples of different molecular weights, but their position

remains constant.
The linear structure of these polymers is strongly suggested by the
conditions required for their formation from the cyclic compounds and
phosphorus pentachloride. That the ring is actually broken in the reaction is shown by the fact that a mixture of polymers, not the original cyclic compound, is obtained by the ammonolysis of the reaction
product. It is natural, therefore, to formulate these compounds as
PC1, (NPCi,) ,I * Cl, the chlorides of the polyimidophosphoric acids.
They may be regarded as the prototypes of other linear phosphonitrilic derivatives. For instance, the cyclic trimeric chloride reacts with
damp lead fluoride (74) to give the compound P3N3FB 2HF * 2H20
which is probably linear, as are some of the phenylated products (10).
The linear deriyatives are usually highly reactive, and are therefore
difficult to isolate and characterize. It is probable that they play a much
inore important part in phosphonitrilic chemistry than would be suggested by the sinall number which have been accurately described.
D. THE COURSEOF
THE
REACTION
The n i a ~ ~ n einr which the linear and cyclic polymers are formed in
the ammonolysis of phosphorus pentachloride is not known in detail,
but seems likely to involve a series of condensation reactions. The first
stage is thought to be the formation of the unstable ammonium hexachlorophosphate:
SH&l
+ PCL -+ NH,PC&
(3)
This then breaks down, possibly to NH2PC14 or more probably to

?iH:PC13, the true nitrogen analogue of P0Cl3. This may react either
with itself, or with more PCl, to give PCll N:PC13. These two compounds are the first members of the two series H(NPClZ),Cl and
PCl, (NPC12).C1; higher members can result from more than one condensation process involving the original reactants and the intermediates.
So mcnibers of the first series have been obtained, although they are
presumably the immediate precursors of the cyclic polymers, formed
from them by intramolecular elimination of hydrogen chloride. The
second series has been dealt with in Section 11, C.
There is both direct and indirect evidence for parts of this proposed
reaction path. Tetramethylammonium chloride dissolves in a solution
of phosphorus pentachloride in acetonitrile to give tetramethylammonium
hexachlorophosphate (33). As stated in Section 11, B the reaction of ammonium fluoride with phosphorus pentachloride yields the very stable
ammonium hexafluorophosphate. Ammonium chloride, though virtually
353
insoluble in nitrobensene, dissolves if phosphorus pentachloride is present, to give a conducting solution (60).The suggestion that ammonium
hexachlorophosphate is formed as a first step in the reaction therefore
appears to be reasonable.
Although no members of the linear series H (NPC12).C1 end-stopped
with the elements of hydrogen chloride have been .isolated, the methyl
analogue CH3N:PC13 of the lowest member has been obtained in high
yield, from the reaction of methylammonium chloride and phosphorus
pentachloride, as a hygroscopic crystalline compound resembling phosphorus pentachloride in reactivity (W. S. Holmes, 2 ) . The corresponding
phenyl derivative was prepared by Gilpin (35).
Mixtures of the lower members of the second series, PC1, (NPC12),C1,
in which a phosphonitrilic chain is terminated by the elements of phosphorus pentachloride, can be isolated from a solution in which reaction
between ammonium chloride and phosphorus pentachloride has not been
allowed to go t o completion. They are identical in type with those made
by the direct reaction of the cyclic polymers with phosphorus pentachloride at high temperatures (Section 11, C ) . Both products are converted to mixtures of cyclic polymers on treatment with ammonium
chloride, provided n is small enough. If n. > 2, chain growth predominates
over ring closure.
General support for this type of reaction scheme is afforded by investigations on the effect of reactant concentrations. An excess of phosphorus pentachloride gives a high yield of the linear end-stopped bodies;
the proportion of these petrol-insoluble compounds can be reduced almost
t o zero by extracting the phosphorus pentachloride slowly into a refluxing suspension of ammonium chloride in tetrachloroethane ( I ) .
The hydrolysis of, for example, dimethyl dichlorosilane to cyclic
dimethylsiloxanes and linear diols can be regarded as following a similar pattern, if OH3+CI- is taken as the analogue of NH4+Cl-. Increase
of acid concentration, or the use of a mutual solvent, increases the
proportion of cyclic polymers; their proportion is decreased by the use
of insufficient, alkaline or scarce water in the form of hydrated salts
(see pp. 80 ff. in ref. 65).
111. Properties
A. GENERAL
The main features of the chlorides are comparative chemical inertness and thermal stability. Many derivatives have been obtained, but
reaction is often slow and incomplete, and the products have not always
been completely characterized. Although they are acid chlorides, the
354
phosphonitrilic chlorides are not hydrolyzed rapidly ; the trimer in particular can be steam-distilled without serious loss. As will be seen in
more detail in Section 111, C, 1, polymerization takes place near 350"
to a rubberlike high polymer.
The patent literature on the phosphonitrilic chlorides is extensive;
no attempt will be made to deal with it here. Most proposed applications
have been to the high-polymer field, particularly where resistance to
high temperatures is required. Phosphonitrilic compounds have proved
valuable in flame-proofing treatments for fabrics and have been suggested as plasticizers, as oil additives, and as catalysts in silicone manufacture.
The lower polymers are beautifully crystalline solids, the trimer
having a faint camphorlike smell. It has marked irritant properties.
Inflammation of the eyes and difficulty in breathing is experienced some
hours after exposure to the vapor; the symptoms may take a day or two
to disappear. Great care should be taken in the manipulation of these
compounds, particularly the trimer. They are dangerous in that the
irritant eEect is not immediate; it is also easy t o become sensitized to
their action, so that small doses may eventually have large effects. The
effect is mainly one of discomfort; no permanent injury appears to result.
B. PHYSICAL
PROPERTIES
The physical properties of the phosphonitrilic halides are presented
in the three groups, general physical properties, crystallographic data,
and solubility data. The general physical properties are summarized
below. Each literature reference refers to all values following the preceding reference.l
Trimeric Phosphonitrilic Halides
P3N3F6( c ) d2,, = 2.21 (F. R. Tromans, 2), mp (triple point) 27.1', vp
(27.1") 293 mm Hg (75), vp (c) login P ( " , ~ , =
) 11.252 - 2630.9 (l/T)
(from -50' to 27'), AHsnbl12.0 kcrtl/mole (W. S. Holmes, %), bp 51.8'
(75)*
VgN3F4Cl2 (1) bp 115-117" (73):

Y3N3F2Cld (1) bp 140-142" (73).
P3N3C16 ( c j d20 = 1.99 (85) (A. Wilson, d ) , mp 114' (86),vp (c) log,,
p(,,) = 11.187 - 3978 (l/T) (from 75.2" to 114.9'), AHf,, 5.0 kcal/mole
(84), bp 256.5' (85, 55 but see 8 4 ) , vp (!) log,, p(mm)= 8.357 - 2880
(l/T) (from 124.9' to 189.3'), AHvap 13.2 kcal/mole (84), p = 0.51
D (46).
P3N3C16Br (c) d20 = 2.27, mp 123.5' (64).
Y3N3CllBr2 (c) ~ Z =
O 2.44, mp 136.5" (64).
A list of symbols appears on p. 380.
355
P3N3C12Br4(c) dzo = 2.84, mp 167' (64).

P3N3Br6 (c) dZo= 3.18 (calc.) ( 8 ) ,mp 191" (6, 8).
Tetrameric Phosphonitrilic Halides
P4N4F8(c) d20 = 2.22 (F. R. Tromans, 2 ) , mp (triple point) 30.4", vp
(30.4") 67.7 mm Hg, bp 89.7" (75).
P4N4F6C12(1) d13.5= 1.8742, mp -12.1', bp 105.8", vp (1) loglo
p(,",,,) = 7.923 - 1911 ( l / T ) , AHTap8.7 kcal/mole (74).
deg-l (calc), mp
P4N4F4C14(1) do = 1.9568,coef. cu. exp. 0.71 X
-24.9", bp 130.5", VP (1) log10 p(mm)
= 7.615 - 1911 ( l / T ) , AHrrp 8.7
kcal/mole (7.9).
P~N~C
(G)~ d20
S = 2.18 (85), mp 123.5", bp 188"/13 min Hg, 328.5"
(89, but see 55), vp (1) (from published graph) loglo P(~,,,) = 8.51 3360 ( l / T ) , AHrap15.5 kcal/mole ( 5 5 ) , p = 0.39 D (46).
P4N4Brg (c) d20 = 3.44 (calc), mp 202' (8).
Pentameric Phosphonitrilic Halides
P5N6Flo (1) bp 122'/751 mm Hg (D. H. Paine, 2 ) .
P5N5Cl10 (c) d ~ =
o 2.02 (H. McD. McGeachin, 2?), mp 41, bp
223-224'/13 mm Hg (89), p = 0.47 D ( 4 6 ) .
Hexameric Phosphonitrilic Halides
PBN6F12 (1) bp 147" (D. H. Paine, 2 ) .
P~N&112 (c) dm = 2.05 (A. Wilson, 8 ) , mp 91", bp 261-263'/13 mm
Hg (89), p = 0.48 D ( 4 6 ) .
Heptameric Phosphonitrilic Halide
P7N7CIl4 (1) mp reported (89) to be below -18O, but almost certainly at least 10" higher, bp 289-294"/13 mm Hg (89),p = 0.54 D (46).
Polycyclic Phosphonitrilic Halide
PsN,Cls (c) mp 237.5', bp 251-261'/13 mm Hg (89),p = 0.72 D (46).
The crystallographic data for the phosphonitrilic halides are summarized in Table I.
The solubilities of the phosphonitrilic chlorides in various solvents
have been determined by de Ficquelmont (29) and L. G. Lund (2). The
results for 20, expressed in gm solute/100 gm solvent, are given in Table
11.
The solubilities of the trimeric chloride over a range of temperature
correspond, within a few percent, to those expected by the theory of
regular solutions (38).They therefore depend on the relative internal
356
TABLE I
CRYSTALLOORAPHIC DATAFOR TEE
PHOSPHONFPRILIC
HALIDES
Unit cell dimensions
(A)
Compound
Crystal
system and
space group
Probably
orthorhombic
Optical
constants
(J. K. Leary, 2)
Q
= 1.395
@ = 1.440
y = 1.47
2V = 75-80'
14.00 6.16 12.94 4 Orthorhombic .(//a)

= 1.619
14.30 6.25 13.03 4 P , u m
B(/c)
= 1.621
14.15 6.20 13.07 4
y ( / b ) = 1.631
2V = 26"
14.24 6.28 13.00 4' Orthorhombic
14.27 6.34 13.02 4 Orthorhombic
Pnaz,, or P.,
14.29 6.48 13.33 4 Orthorhombic
7.48 13.83 5.16 2 Monoclinic

Pz,/n
(a = 112O)
10.79
5.93 2 Tetragonal
P42/n
44
63
A. Wilson, 2
64
14.38 6.64 13.35 4 Orthorhombic Q(/c)

P-
Other
references
1.720
= 1.742
y ( / b ) = 1.812
2v = 54"
B(//a)
= 1.379
B ( / b ) = 1.428
y = 1.463
7C = 62"
2v = go"
Q
= 1.675
w =
1.678
F. R. Tromans, d
44
45
10.82
5.95 2
11.18
6.29 2 Tetragonal
P42/n
19.37 15.42 6.23 4 Orthorhombic

P213121
Q = 1.629
D. E. C. Corbridge, d
B = 1.644
y undetermined
(optically -we)
eee footnote"
a Hexameric phosphonitrilic chloride exists in at least two, and probably three,
crystalline forms. An orthorhombic modifkation was described by Stokes (89) and
examined by Tawin (92). who found axial ratios a:b:c::0.5482:1:1.757;the crystals
were optically positive, the plane of the optic axes being (100). The cell size of a triclinic form has been determined by A. Wihon (2);a = lO.d, b = 10.7& c = 11.4A,
(I = 93.5",
0 = 90, y = 117";its optical constants are Q = 1.628,B = 1.649,y =
1.650,2V = 34'. A second, less stable, triclinic form has appreciably higher refractive
indices.
357
TABLE I1
SOLUBILITIES
OF THE PHOSPHONITRILIC
CHLQRIDES
IN VARIOUS
SOLVENTS
AT 20" EXPRESSED
AS GM SOLWTE/IOO GY SOLVENT
(PNCI*)*
Solvent
Ref.29
n-Pentane
n-Hexane
n-Heptane
Ref. 2
(PNCM,
Ref. 29
6.1
7.0
25.5
25.5
27.9
8.4
(sCrs0"C
petrol)
(80-90"C
petrol)
Carbon
disulphide
Carbon
tetrachloride
&m-tetrac hloroethane
Benzene
Toluene
Xylene
Diethyl ether
Dioxane
55.0
47.3
38.8
46.4
29.6
(PNC1s)s
Ref. 2
-2000
(PNCl&
Ref. 8
27.8
34.0
lo30
22.0
52.1
38.9
Ref. 2
38.3
33.4
56.5
16.6
21.4
17.8
13.9
12.4
8.2
15.5
98
7.9
23.5
145
pressures of solvent and solute, and show the comparative unimportance

of chemical interaction between them. In particular, the solution in s y m tetrachloroethane is almost ideal ; the temperature coefficient of solubility corresponds to a latent heat of fusion of the trimeric chloride of
4.8 kcal/mole, slightly lower than the figure, 5.0 kcal/mole, determined
from the vapor pressure curves (84).
C. CHEMICAL
PROPERTIES
The reactions of phosphonitrilic compounds may be conveniently considered in four groups: polymerization, reactions with donor molecules,
alkyl and aryl derivatives, and reactions with acceptor molecules. The
relation between structure and reactions will be considered in Section V.
1. Polymerization
All the lower phosphonitrilic chlorides polymerize on heating to 250"350C (89),though if the materials are carefully purified the reaction
is slow. As pormally prepared, the product retains small amounts of the
lower cyclic polymers, which may be removed by extraction with a
solvent (27). The high polymer, having a molecular weight of a t least
20,000 (78,79), is stable at room temperature, swells in benzene and
other organic solvents, and gives a characteristic X-ray fiber diagram on
358
stretching (53), see Section IV, B, 4. It also resembles natural rubber

in its mechanical properties (61, 52, 5 S ) , though it hardens and breaks
up in damp air. Depolymerization begins above 350", with the formation of a mixture of lower polymers; quantities of a black, insoluble,
infusible material are obtained on slow heating of the polymer to about
500".
The equilibrium has been studied in a preliminary way by SchmitzDumont (721, who heated the trimeric chloride in the vapor phase at
600". The mixture of polymers obtained included trimer, tetramer, and
higher crystalline and liquid polymers. The existence of equilibrium was
established by showing the same mixture of polymers to be obtained,
whether the starting material was trimer, tetramer, or high polymer.
The proportion of higher polymers was decreased by increasing the temperature; polymerization is therefore exothermic. The interconversion
of the lower polymers has also been studied by de Ficquelmont (988).
The phosphonitrilic fluorides and chlorofluorides behave similarly,
though the latter compounds are less thermally stable than the chlorides.
Since tetrameric chlorofluorides are formed by fluorination of the triincric chloride with lead fluoride, the eight-membered ring is evidently
the more stable. Dissociation of the chlorofluoride P4N,F6C12was observed in the range 201-302". At the higher temperature the vapor density
corresponded nearly to the dimer PaN2F3Cl, and Schmitz-Dumont and
Kulkens (74)interpreted their results in terms of simple dissociation of
the tetrameric molecule into two equal parts. P4N4F4C14also dissociates
in the same temperature range, but to a smaller extent (73). The chlorofluoride high polymers are stable at room temperature, but depolynierization at atmospheric pressure occurs a t lower temperatures than
with the chlorides.
Qualitatively, the results taken together indicate the increase of stability with polymer size in the series dimer, trimer, tetramer, and high
polymer. They are, however, suggestive rather than exhaustive, and
detailed determinations of heats of formation of the separate members
.\r.ould be valuable.
The kinetics of the polymerization of tri- and tetraphosphonitrilic
chlorides in solution and in bulk have been studied by Patat and Kollinsky (58) and Patat and Frombling (57). Hydrocarbons are unsuitable
solvents, since they react to give hydrogen chloride; successful results
were obtained in carbon tetrachloride. The proposed mechanism involves
unimolecular initiation, either by oxygen (in solution) or another phosphonitrilic molecule (in bulk). A bimolecular propagation step is followed by unimolecular termination. Traces of water were found by
Renaud to have a significant effect on the polymerization process (62,63).
359
t2. Reactions with Donor AIolecules
The phosphonitrilic chlorides react with many donor molecules, particularly those carrying an active hydrogen atom, though the products
are not always well-defined. (For references to work on mercaptans and
thiophenols, see Audrieth et al. (S).) Attention will be directed here
mainly to hydrolysis, esterification, and reactions with amines.
a. Hydrolysis. The hydrolysis of the chlorides to the cyclic imidophosphoric (phosphonitrilic) acids is slow. From the trimer the interniediste product P3N3C14(OH)2has been obtained (86, 86). The acid
Y33306H6 is normally tribasic, though a hexasilver salt has been obtained (87), and the structure is therefore usually written (NH *
P(0)OH)3. (The tetrasodium salt obtained by Stokes (87) is probably
a derivative of the linear acid HO [PO (OH)NH] sH). Further hydrolysis of the acid to the imidophosphoric acids and finally to ammonia and phosphoric acid takes place easily (7, 86, 87). Narath et al.
156) have found that replacement of the NH groups in the sodium salt
of triphosphonitrilic acid by oxygen atoms takes place without the formation of large amounts of chain imidophosphates. It may be that the
ready interconversion of ring and chain forms suggested by Stokes for
the pentameric and hexameric acids occurs also with the trimer.
The tetrameric chloride is more readily hydrolyzed to the very stable
acid (NH* P(0)OH)4 - 2H20 (85, 86, 88). Two octasilver salts were
obtained by Stokes (88); the cyclic acid is otherwise tetrabasic. The
unit cell sizes of the dihydrate, the dipotassium and dirubidium salts
have have been obtained by Corbridge (19). The tetrameric acid is also
formed by decomposition of the pentameric acid (90). There is little
definite information on the higher acids (90,91).
The phosphonitrilic fluorides (PNF,) 3,4 may also be hydrolyzed to
the corresponding acids (75), apparently more readily than the chlorides (D. R. Smith, 2 ) , the trimer in each case being more resistant to attack.
The infrared absorption spectra of the tetraphosphonitrilates (19)
show that hydrogen bonding in the solid is extensive, so that the disPO(0H)- and -N =
tinction between the tautomeric forms -NH
P(0H)- is not as great as these formulas would suggest. Nevertheless, the alkali-metal salts of tri- and tetraphosphonitrilic acid react
with sodium hypochlorite to give N-chloro derivatives (93).
b. Esterification. The chlorides react smoothly with alcohols and alkoxides, though side reactions giving hydrogen chloride and alkyl chlorides also occur. These may in part be avoided by working a t a low
temperature. The hexamethoxy derivative of the trimer (bp 127'/0.1
360
mm) was prepared by Dishon ( $ I ) , though most of the product was

converted into a water-soluble polymer. The hexaethoxy compound
(NP(OEt)2)3 was prepared by Rats and Hess (61), by the action of
sodium ethoxide on the chloride, as a clear viscous oil, changing to a
jelly on standing. On heating this ester to 125", diethyl ether was eliminated irreversibly.
Eta\
/OEt
P
\P
PEt\
4
OEt
P//O\
P//
+ EtoO
/ NN
N
I
(4)
II
I
II
I
I1
I I1
Goldschmidt and Dishon (36') were able to replace all the chlorine in
the rubberlike high polymer, though the product was deficient in alkoxy
groups, probably on account of a similar condensation process.
By the action of ethyl iodide on the tri- and tetrasilver salts of the
appropriate phosphonitrilic acids, Ratz and Hess (61) have prepared
the ethyl esters of the N-ethyl phosphonitrilates (EtN PO (OEt) ) 3,4.
The trimeric ester is a crystalline compound which forms a dihydrochloride.
Trimeric phosphonitrilic azide has been prepared (37) from the
trimeric chloride and sodium azide in solution in acetone.
Esterification, hydrolysis, and other reactions of the phosphonitrilic
series may often be catalyzed by pyridine and other donor molecules;
this effect will be considered in more detail in Section 111, C, 2, d.
c. Ammonia and Primary and Secondary Amines. The chlorides also
react with ammonia, and with primary and secondary amines, though
complete elimination of the chlorine is difficult. Tri (phosphonitrile)
diamidotetrachloride P3N3(NH2)&la has been .repeatedly investigated
(29, 64, 86, 86) ; it melts at 162' with decomposition, and is much more
soluble in ether and dioxane than in nonpolar solvents (29). The completely ammonated (PN(NH2)J can be obtained by prolonged action
of liquid ammonia on the trimer (see p. 126 in ref. 3). On heating, it
loses ammonia to give, eventually, a product with the composition of
phospham (PN:NH).. A less pure form of phospham has also been
obtained by passing ammonia gas over the molten trimeric chloride (20).
Ammonolysis of the tetrameric chloride yields the diamidohexachloride and the tetraamidotetrachloride ($66).Again, deammonation of the
completely substituted tetramer gives phospham (see p. 127 in ref. 8 ) .
The reactions of primary and secondary amines with the phosphonitrilic chlorides have been studied by several investigators. Complete
aminolysis is difficult to attain (see p. 127 in ref. 3), though fully substituted derivatives of the trimer and tetramer have been reported.
N
/ NN
N/
NN
N
"
N
PHOSPHONITRILIC DERIVATIVE%
361
The reactions of aliphatic and aromatic amines with the trimeric

chloride have been studied by Schenk (7'0)and by Bode, et al. (14), who
give details of the preparation of the dichlorotetraanilide, the hexap-toluidide, the hexapiperidide, the tetrachloro-bis-dimethylamide and
the corresponding compound with ethylenediamine. The tetrachloroTABLE 111
AMINE
DPRIVATIVEB
OF PIIOBYHOKITRILIC CHLOHIoE TILIMEII
AND TETRAMER
Derivative
Fully substituted
derivatives of trimer
hexaanilide
P3Ns(NH.CeH&
hexa-o-toluidide
PLNa(lVH .CsH,. CH8)s
hexa-p-toluidide
PaNs(NH .CJ34*CHs)s
hexapiperidide
I'zh'n(NCbHl,)r
Melting points and references

264' ( I I ) , 267" (70), 268' (40)
241-242" (20)
242" (Id),243" (/to)
266" (14),slnter point

231" (YO), decomposes below
mp
Iicx~Lpheny lhy drazide

1%?-200 (YO), 200" (20)
P3Ns(NH
.CsHs)s
diphenyltetraanilide
198-199" (14)
PsNdC6Is)dNH CoH&
diphenyldianikdediami$e
218' (14)
P~N~(C~H&(NH*CJIS)~(NH?)I
tetraar,ilideditlmide
PaNs(NH .CsHs?r(N I&
256-257' (14)
+
Tetra substituted
derivatives of trimer
tetraanilide
PsNa(NH.C S H ~ ) ~ C ~ ~
tetra-ptoluidide
I'sNs(NH. CSH, .CHs)rCI?
diphenyldiamide
PsNI(CIHS)~(NHZ)~C~~
diphenyldianilide
1'3Na(CaH&(NH .CaHd?Ch
Fully substituted
octaanilide
derivativeof tetramer PdN,(NH. CSH&
162163' (14)
193' (14)
244" (TO)
o-phenylene diamide may possess a spiran structure. A number of amido

derivatives of diphenyltriphosphonitrilic chloride were also prepared,
such as the diphenyldiamide and diphenyldianilide and the diphenyl
tetraanilide. Melting points of the amine derivatives are givea in Table
111. Many of these compounds are basic to perchloric and hydrochloric
acids. Although attention has been concentrated on the trimer, there
362
S . 1,.
PADDOCK AND H. T. SEARLE
seems little doubt that similarly detailed work could be carried out with
the tetrameric chloride.
Bode and Clausen i l l ) have repeated earlier work on the anilide
(WP(NHPh) 2) 3 (40). On heating, this compound loses aniline reversibly.
(Compare the behavior of the ethyl esters, Section 111, C, 2, b.) The
products are soluble in nitromethane and nitrobenzene, and possibly
contain the grouping 2 P = N Ph. On further heating, insoluble phenylphosphams (NP:NPh), are formed without further loss in weight; for
tlicse the structure shown in formula I1 is suggested.
Ph
i
1'11
Indirect support is given to this formulation by an infrared study

(82) of phospham (NP:NH)., which retains the strong absorption near
1300 em-' characteristic of the phosphonitrilic P-N bond, and shows

other features which also occur in the spectra of the imidophosphates.
d. Tertiary Ainines and Donor Solvents. Tertiary amines such as
pyridine, quinoline, and tribenaylamine also react with the phosphonitrilic chlorides (H. Schaperkijtter, quoted in ref. 5) ; weak complexes are
probably formed. One important consequence of the effect is that hydrolysis and esterification are accelerated in the presence of pyridine.
The products of hydrolysis are pyridinium salts of polymeric acids,
from which the pyridine can be removed by keeping in vacuo over
sulfuric acid (71).Disruption of the ring need not occur, however. For
instance, the butyl ester of the trimer [P3N3(OBu)s1bp 170-171*/0.3
mm] was prepared by Dishon (31) by the interaction of the chloride
and butanol in pyridine at 0".
Other donor molecules react with phosphonitrilic compounds. Dioxane
is said to facilitate the depolymerization of the high polymer (55) and
of the phenyl derivatives (13). Acetone forms complexes with the amino
acid derivatiwa of the tetramer (see p. 128 in ref. 3). It is possible that
ether acts as Icatalyst for the hydrolysis of the trinieric chloride. The
effect of these solvents is subtle, since the solubilities of the trimeric
chloride in ether, in pyridine, and in dioxane show no sign of any specific
chemical interaction.
The elucidation of thc effect of donor solvents, and its extension to
363
the preparation of new types of derivative, may be expected to become

increasingly important in phosphonitrilic chemistry.
3. Alkyl and Aryl Derivatives
No phosphonitrilic alkyls have been obtained. Stokes (85) continued

the work of Couldridge (20) on the reaction of zinc ethyl with the trimeric chloride; a t high temperatures a violent reaction took place, but
no identifiable products could be isolated. It is probable that alkyl phosphines were formed. W. Philpott ( 2 ) found the trimeric chloride in ethereal
solution to be unreactive to methyl Grignard reagents and to aluminum
alkyls. Unsuccessful attempts have been made by H. T. Searle (2) to
prepare the methyl derivatives by reaction of Me2PC13with ammonium
chloride, either dry or in tetrachloroethane. Most of the expected hydrogen chloride is evolved, but ring closure does not take place.
In the aromatic series, Bode and Bach (10) were able to isolate only
linear products from the reaction of PhPC14 with NH4C1. Treatment
of the trimeric chloride in ethereal solution with bromobenzene and
sodium gave no isolable product. Partially and fully phenylated compounds have, however, been obtained in other ways.
Thus, Rosset (66) prepared the hexaphenyl derivative of the trimer
(P3N3Phs) by the action of phenylmagnesium bromide on the trimeric
chloride in boiling toluene. Bode and Bach (10) were unable to repeat
this work, but obtained a small yield of the hexaphenyl compound (mp
185-187C) by using anisole as a solvent. Its constitution was proved
by the quantitative yield of diphenylphosphinic acid obtained on hydrolysis. The main product contained halogen, which reacted with silver
perchlorate in alcohol to give the linear compound Ph(PPhZN),H *
HC104. The trimeric chloride also undergoes the Friedel and Crafts
reaction (10) to give the diphenyl derivative P3N3C14Ph2,mp 92.5"C.
The tetrameric chloride also reacts with phenylmagnesium bromide
to give partially and completely phenylated products (13).These include two octaphenyl derivatives (which may be boat and chair forms)
and two tetraphenyl derivatives. Some of these compounds when hydrolyzed in the presence of dioxane gave PsNSPho, suggesting that the formation of monomer is facilitated by this solvent. Two (probably linear)
phenylated cleavage products containing halogen were also isolated.
Ultraviolet light has no effect on the phosphonitrilic chlorides themselves. Dishon and Hirshberg (32) found that solutions of the trimeric
and tetrameric chlorides in hydrocarbon solvents became brown on irradiation with light from a mercury arc, and were able to isolate from
the product distillable ghen 1 and decahydronaphthyl derivatives, the
P
formation of whivq suggests radical attack.
364
4. Reactions with Acceptor Molecules

In many of the reactions referred to above, the phosphonitrilic compounds have behaved as acceptors; hydrolysis, esterification,' aminolysis,
and fluorination can all be understood in terms of nucleophilic attack a t
the phosphorus atom. Phosphonitrilic compounds also possess marked
donor properties, though these have not been studied systematically
until recently.
Besson and Rosset (7) found that nitrogen dioxide formed an unstable complex with the trimeric chloride, one molecule of which took
up between two and three molecules of the oxide. Sulfur trioxide reacts
with the trimeric chloride a t 40" to give an addition compound (PNCb
S03)3, a hygroscopic solid which yields sulfuric and sulfamic acids on
hydrolysis ( 3 4 ) . Bode et al. (14) prepared (PNC12)3 * HC104 and
(PNCl,), * 2HC104 by the ,action of perchloric acid on solutions of the
chlorides in acetic acid; Stokes noted the solubility of the trimeric
chloride in sulfuric acid ( 8 5 ) .
By the use of cryoscopic measurements in 100% sulfuric acid, D. R.
Smith ( 2 ) has found that the trimeric chloride takes up one proton, and the
tetramer, pentamer and hexamer two each, thus confirming and extending the preparative work on perchlorates. Although the base strengths
of ail the polymers are of the same order as that of anthraquinone, they
may be shown in various ways to differ significantly among themselves.
The principal way in which the acid-base equilibrium has been
studied is the partition of the chlorides between sulfuric acid and
n-hexane. A plot of the partition coefficient against acid concentration
is shown in Fig. 2. It will be seen that the trimer is an appreciably
stronger base than the others; as might be expected, the bromides are
stronger and the fluorides weaker bases. A similar method has been used
in the study of the azulenes (59).
The calculation of the numerical values of the base constants from
the partition results is difficult for two reasons. First, no method has
been found for the determination of the concentration of neutral base
in the acid layer, and second, the partition coefficient is strongly dependent on polymer Concentration even in dilute solutions. Nevertheless, by the-use of simplifying assumptions the value p K , = -8.2 may
be obtained for the strength of the conjugate acid of the trimer (D. R.
Smith, 2 ) . This is closely confirmed by indicator studies, and is consistent
with the shift of the H-C1 stretching frequency at 3.46 p observed in solutions of hydrogen chloride in the trimer (A. C. Chapman, 2 ) .
The difference in base strengths, and consequent difference in partition coefficients between sulfuric acid of a chosen concentration and an
365
/I
-2 .O
80
09 90
I
91
92
93
I
94
I
95
Concentration of acid (wt
I
96
I
97
I
98
I
99
H,SO,)
FIG.2. The partition coefficients of the phosphonitrilic chlorides between sulfuric

acid and n-hexane. The initial concentration of each polymer in the n-hexane was
3.5 x lod m.
inert solvent, forms the basis of a useful method of separating the phosphonitrilic chlorides ( 1 ) . In a comparable way, McCauley et al. (50)
have separated the isomeric xylenes by utilizing their different basicities
to an HF/BF3 mixture. Although the phosphonitrilic chlorides are, like
hydrocarbons, stable toward most cold concentrated acids, they are decomposed by hydrogen fluoride to ammonium hexafluorophosphate ( 1 2 ) .
IV. Structure
A. GENERAL
The foregoing chemical reactions have shown the retention of the
ring structure of the phosphonitrilic compounds through a variety of
react.ions in which the chlorine atoms are replaced by other electronegative substituents. They have chiefly concerned the trimeric and tetrameric chlorides, because these are more abundant than the higher polymers. It is known, however, that the pentameric chloride, for instance,
is both thermally stable and resistant to hydrolysis. The higher polymers
have not yet been separated but the mixture of approximate composition' (PNC12)8-9 is also comparatively inert. The individual members of
366
which it is composed may therefore be expected to exhibit similar chemical behavior to the trimer and tetramer. As a class, the phosphonitrilic
chlorides react much less readily than acid chlorides normally do.
There are two general points of interest: first, the interpretation of
the stability of the polymers in terms of their known structural characteristics, and second, whether or not the cyclic structures proved for
the trimeric and tetrameric chlorides are found in the higher members.
The structural evidence relating to these two points. will be detailed
in Section IV, B.
B. EXPERIMENTAL
1. Trimeric Compounds
The structure of the trimeric chloride was investigated by Brockway

and Bright by electron diffraction (16). The structure they proposed, in
which each phosphorus atom carries two chlorine atoms in a plane
pcrpendicular to the P3Ns ring, has been confirmed by the crystallographic work of A. Wilson (6).The results are given in Table IV, in which
TABLE IV
THE MOLECULAR
DIMENSIONS
OF
THE TRIMERIC
AND TETRAMERIC
PHOSPHONITRILIC
CHLORIDES
Compound
and
method
1'-S bond
(PXC1I)I E.U:
(PNCl,), x-ray
(PNC1z)d X-ray
1.65 k 0.03
1.61 f 0.04
1.67 f 0.04
length
P-C1 bond
lenDgth
(A)
(-4)
1.97 & 0.03
1.99 +_ 0.03
1.99 f 0.03
Cl-P-C1
angle
(degrees)
N-P-N
angle
(degrees)
107-110
101 rt. 3
105.5
1'21 +_ 3
117
P-N-P
angle
(degrees)
assumed to he 120
119 f 3
123
ED. denotes electron diffraction.
those of Ketelaar and de Vries (45) for the tetrameric chloride are also
included for convenience. Some apparent inequalities in P-C1 bond
lengths in the trimer may disappear after three-dimensional refinement
of the crystal structure. The main point here is the shortness of the
P-N bond lengths compared with 1.78 A, the value obtained for the
length of the single P-N bond in sodium phosphoramidate (39).
The infrared absorption spectrum of the trimeric chloride has been
determined by Daasch and others (24, 25, 4 6 ) , and is shown, for the
sodium chloride region. in Fig. 3. It is dominated by the absorption band
at 1218 cm-l, which is due principally to the P-N stretching mode and
occurs at slightly different positions in the spectra of the other polymers.
367
Frequency (cm-')
FIG.3. The infrared absorption spectrum of the trimeric chloride, together with
portions of the spectra of higher polymers (Section V, 6.) The numbers denote n
in (PNCL),.
If the P-N bond were single, absorption would be expected in the region
700-800 cm-l (41). Such absorption is absent from the spectrum of the
trimeric chloride.
The detailed structure of the infrared absorption and Raman spectra
(24, SO) has been shown by Daasch to be in complete agreement with
the planar structure of Dsh symmetry. From the Raman frequencies
Iribarne and de Kowalewski (43) calculated the P-N stretching frequency of trimer to be 4.0 millidynes/A. This value depends on the
assignment of a line a t 100 cm-l to C1-P-Cl angular deformation. Both
the assignment and the numerical value of the stretching constant' seem
reasonable, but since the line was not observed by Daasch the conclusion should be accepted cautiously.
A partial analysis of the crystal structure of the trimeric bromide
has been carried out ( 9 ) , and is consistent with a planar molecule. The
mixed chlorobromides P3N3C15Br,P3N3C14Br2,P3N3C1&r4 have the same
or nearly related crystallographic symmetry and cell dimensions, and presumably similar structures (Section 111, B) . Their infrared absorption
spectra (64) also show that the order of the P-N bond exceeds unity.
There is some thermochemical evidence of bond-strengthening.
S. B. Hartley (8)has found the heat of formation of crystalline triphosphonitrilic chloride to be AH, = -196.3 3 kcal/mole. By making some
reasonable assumptions, the P-N bond energy is found to be 75-80 kcal/
mole, appreciably greater than that expected for a single bond.
2. Tetrameric Compounds
In the tetrameric chloride crystal, the ring is puckered, with Sa symmetry, though with equal bond lengths round the ring (4.5). This may
be contrasted with the most recent data for cyclooctatetraene, which has
368
N. L. PADDOCK AND H. T. SgAaLD
been proved (6) to have the Dza configuration in the vapor phase, with
the unequal bond lengths C=C = 1.334 f 0.001 A, C-C = 1.462 C
0.001A.
The spectroscopic data on the tetrameric chloride (24) conflict with
the crystal structure determination. A larger number of fundamentals
in both Raman and infrared, and of coincidences between them, would
be expected than are in fact observed. The molecule must therefore have
a higher symmetry in solution than in the crystal, and the data are most
nearly in accord with a planar model of symmetry C4,,, though DZdsymmetry is not excluded.
The crystal structure of the tetrameric fluoride is a t present being
determined (F. R. Tromans, 2 ) . No detail is a t present available, but the
space-group determination shows that the molecule is centrosymmetrical.
,.
..
FIG.4. The structure of ( a ) tetraphosphonitrilic chloride and (b) tetraphosphonitriiic fluoride. The phosphorus atoms are shaded and the halogen atoms dotted.
The phosphorus atoms must therefore lie in a plane, and the nitrogen
atoms in another intersecting it, the configuration being similar to the
eight-membered ring in octamethyl tetrasiloxane (83).It is unlikely that
these planes coincide. The resulting ring, of Czh symmetry, is, however,
more nearly planar than that in the chloride. The two rings, which may
be regarded as boat and chair forms, are illustrated in Fig. 4. The ends
of the boat or chair are formed from two atoms each, rather than one
as in the cyclohexane series.
3. Xedium Polgmers
a. General. There is no certain knowledge of the structure of the

higher polymers, until the rubberlike high polymer is reached. Cyclic
structures for the pentamer upwards have been accepted only reluctantly,
369
involving as they do rings of ten members or more. It is, however, difficult to write a linear structure for (PNC12), except as a di-radical or a
highly polar entity.
There is no sign of behavior as a radical; the thermal stability of the
carefully purified pentamer and hexamer is of the same order as that of
the trimer and tetramer. The second possibility can be ruled out, since
the dipole moments are all small, being of the order of 0.5 D (46)(Section 111, B ) , and show no large differences as the series is ascended.
The mixture of compositjon (PNClz)s.9 has a comparably low polarity
(H. T. Searle, 9).
The evidence is strengthened by the preparation from the cyclic
:rimer or tetramer of the related linear compounds (PNCl,). * PCl,
(Section 11, C), which have quite different properties. The stable existence of such large rings has in any case seemed less unreasonable since
the isolation of the separate members of the cyclic dimethylsiloxane series
(Me2SiO). up to the nonamer, which has a 18-membered ring (see p. 80
in ref. 65).
While the balance of evidence is therefore in favor of a cyclic formulation for all compounds (PNClZ). except the elastomer, a direct
crystal structure determination of the pentamer or higher polymer would
be of great value.
b . Spectroscopic Data. The ultraviolet absorption spectra of several
members of the series have been published (46). Figure 5 includes the
spectra of the trimeric to hexameric chlorides obtained by D. R. Smith ( 2 )
(cf. Krause, @), and of a phosphonitrilic oil, approximately (PNCL)8.
They differ characteristically from those of the end-stopped polymers
(Fig. 1).
Since the spectra show little change in position or intensity as the
series is ascended, W-W* transitions are excluded. Figure 6 shows that
the spectra are comparatively slightly affected by solution of the chlorides in sulfuric acid; absorption cannot, therefore, be due to an n-T*
transition from the nitrogen atom. It may be a result of excitation of
the unshared elec'trons on the halogen atoms, consistent with the effect
of bromine substitution on the position and intensity of the absorption
band. The ultraviolet spectra therefore give no direct information on
the structure of the ring.
The infrared spectra of trimer to heptamer have been published (46)
though so far complete interpretation is restricted to the two lowest
members (24).There is a common feature, however, in the very strong
absorption in the 1200-1350 cm-1 region, principally contributed by the
P-N stretching vibration. Daasch and Smith (25) examined the infrared
absorption spectra of the phosphonitrilic esters, and showed that the
370
N. L. PADDOCK AND H. T. SEABLE
3.c
000
2.c
00
\
u)
2
[r
-i
1.C
0.c
I
I
10 210
220
230
240
250 260
27C
.O
Wavelength (mp)
FIG.5. The ultraviolet absorption spectra of the phosphonitrilic chlorides. Molar
extinction coefficient & calculated on a PNCL monomer basis. KEY:--- trimer ;
.-. .- tetramer;
pentamer; - hexamer;
soluble oil
(approximately octamer) ; numbers denote n in (PNCI,!,. The molar extinction
coefficient is calculated on a PNCL monomer basis.
-.
- - - + -
371
P HOSP HONITRILIC DERIVATIVES
+ 4.c
0,o00
+3.c
000
+2.c
00
u)
2
0
+ 1.c
0.0
I
-1.0
200 220
240
260
280
300
320 340
).I
Wavelength (rnp)
FIQ.6. The effect of solvents on the ultraviolet absorption spectra of (PNCl&
and (PNBr&. The molar extinction coefficient is calculated on a PNCL or PNBa
monomer basis.
P-N stretching frequency was only slightly affected by the mass of the
substituent over the range CH30 to CsH170. This frequency is therefore
characteristic of the ring itself, and, provided due account is taken of
other contributing factors, may be used as a measure of its stability.
372
N. L. PADDOCK AND H. T. SBABLE
It is principally dependent on the electronegativity of the atom attached

to the ring, frequency increasing steadily with electronegativity, as
shown in Table V.
TABLE V
THEP-N STRETCHING
FREQUENCIES
OF THE PHOSPHONITRILIC
HALIDES
P-N stretching
Compound
frequency
(cm-1)
Reference*
1170
1218
1242
1287
1322
I305
1325
1324
1291
1240
1365-1380
* Reference f identifies work carried out by A. C. Chapman.
4. Nigh Polymer
Meyer et al. (53) examined the X-ray fiber pattern of the stretched
high polymer, and with some assumptions based on a consideration of
other phosphorus compounds, calculated the form of the chain. It has
the helical configuration common in polymers, the set of chlorine atoms
forming a screw of opposite hand to that of the phosphonitrilic skeleton.
The principles of construction seem to be the maintenance of P-N-P
and N-P-N angles close to those found in the lower polymers, and a
dihedral angle which minimizes the repulsion of nonbonded atoms.
c. STRUCTC-RAL
THEDRY
Thc evidence from X-ray, spectroscopic, and thermocheinical investigations considered in the previous sections shows clearly that the lower
phosphonitrilic chlorides possess cyclic structures which exhibit a high
degree of stability, suggestive of aromatic character. Such a view of
their structure is compatible with their low reactivity compared with
other acid chlorides, and with the maintenance of the ring structure in
a variety of derivatives. It is easy to write Kekulk-type structures for
the trimeric chloride (111); such structures have also been suggested
PHOSPHONITRILK DERIVATIVES
c1
373
CI
\P'
C1 Ny
\A
\N
II /
C1
/ XN/ \cl
C1
for the tetrameric chloride ( 4 5 ) . The way in which aromatic character

can arise in these compounds will be considered in the present section;
it will be seen that there are important differences from aromatic series
which involve only boron, carbon, or nitrogen.
In benzene the situation is clear and well-known. Three of the four
valency electrons of each carbon atom may be regarded as occupying
sp" hybrid orbitals, mutually spaced a t angIes of 120" in a plane; overlap of the orbitals of adjacent carbon atoms provide for a hexagonal
ring firmly held together by o-bonds at their natural angle. A hydrogen
atom is attached to each carbon atom in the plane of the hexagon, and
in the direction of the third 'sp2 orbital. The fourth electron occupies a
pz orbital perpendicular to the plane; overlap between these orbitals
round the ring gives rise to the characteristic delocalization and its consequences of short and strong bonds.
The lowest state has full circular symmetry and is nondegenerate; it
can therefore accommodate two electrons. The next higher state is doubly
degenerate, the two orbitals having each a nodal. plane perpendicular
to the ring and to each other, the combination having full circular syminetry and therefore exhibiting the characteristics of a closed shell. Six
electrons complete the tn-o lowest shells, accounting for the special stability of benzene. The next highest level is also doubly degenerate, so
that a set of eight electrons (e.g., in cyclooctatetraene) corresponds to
a half-filled uppermost shell and no special stability; aromaticity would
be expected to occur next in the ten-electron system of cyclodecapentaene ( 4 2 ) .
There are, however, significant differences when silicon, phosphorus
or sulfur is inclbded in the ring, since the possibility arises of using
d-orbitals for *-bonding ( $ 1 ) . These differences show up particularly
clearly in the phosphonitrilic halides. In these compounds the electrons
of the nitrogen atom may, to a first approximation, be regarded as occupying sp2 hybrid orbitals, one in each bond to a pposphorus atom,
two in a lone pair. The fifth electron can use a p , orbitd for the formation of a *-bond, as in pyridine. The phosphorus atom has five electrons
for four bonds, arranged tetrahedrally. Again to a first approximation,
374
N . L. PADDOCK AND H. T. SEAaLE
we may think of these bonds as formed from sp3 hybrid orbitals, leaving a d-orbital for t-bonding. The important difference is that a d-orbital
has g-symmetry with respect to inversion at its center, whereas the
adjacent p-orbitals have u-symmetry, and for maximum overlap must
alternate in sign, as shown in Fig. 7. The repeating unit therefore con-
FIG.7 . The system of overlapping orbitals responsible for dn-pu bonding.
sists of two P N units, and there is inevitably a mismatch of sign in the

trimer, which has three such units. The trimer may therefore be expected to have a lower energy per *-electron than the tetramer, in which
the signs can be matched all round the ring.
A simplest possible molecular orbital treatment of the t-electrons
(23) shows that the symmetry properties have a deeper significance, in
that they require the highest filled orbital to be nondegenerate, so that
any even number form a closed shell. By contrast, if the aromatic system is formed solely from p-orbitals, the lowest filled orbital only is
nondegenerate, leading to Hiickels rule, that closed shells are formed
only from 4% 2 electrons. Benzene is aromatic, cyclooctatetraene is
not, corresponding to the calculated difference of t-electron energies and
the alternating bond-lengths in cyclooctatetraene.
On the other hand, all the phosphonitrilic chlorides display comparable aromaticity. The assumption of a particular difference in electronegativity between phosphorus and nitrogen allows the calculation
of w-electron energies in terms of the appropriate resonance integrals.
Table VI compares the aromatic hydrocarbons and the phosphonitrilic
chlorides. The calculated energies per electron relative to the Coulomb
parameters ac: and ap are expressed in terms of the appropriate resonance
TABLE VI
ELECTRON,
I N TERMS OF
RESONANCE
INTEGRALS
CALCTL.4TED ENERGIES PER
No. of electrons
Aromatic hydrocarbonsac
1.0B
(NPCL),
1.78s
1.338
1.876
QP+
8
1.208
1.896
THE
APPROPRIATE
10
12
1.306
1.906
1.248
1.906
PHOGPHONITRILIC DERIVATIVES
375
integrals /3 and 6. It must be emphasised that the steady increase in velectron energy in the phosphonitrilic series is a result of the symmetry,
not of any particular arithmetical assumptions. Any even number of electrons forms a closed shell, and therefore has aromatic properties associated with it.
V.
Discussion
A. FURTHDR
ASPECTSOF THE THEORY
There are two further aspects of the theory which are important in
its application to the phosphonitrilic series. The first is that a 3d orbital
is normally too diffuse to overlap effectively with a 2p orbital (22); it
becomes sufficiently compact to do so in the present instance only if a
partial positive charge is induced on the phosphorus atom by electronegative substituents attached to it.
The second aspect concerns the geometry of the system. It is assumed
in the theory that the ring is planar, though since the d-orbitals have
considerable lateral extension, this requirement is of smaller importance
than in the series of aromatic hydrocarbons. The phosphonitrilic halides
may therefore be expected to exhibit some flexibility without appreciable loss of aromatic character. At some &age, however, a limit will
be reached. As the ring continues to increase in size, the steady increase
in r-electron energies will be increasingly offset by folding of the ring
and consequent partial uncoupling of the r-electrons, so that the observed energies should pass through a maximum.
The existence and main properties of the series of phosphonitrilic
halides can be accounted for in these terms, even though only a small
proportion of the valeney electrons is considered by the theory, and in
spite of the neglect of U-W interaction, which must be more important
here than in the hydrocarbon series. Other factors which influence stability will become apparent in Section V, C.
B. CHARGEDISTF~IBUTION
EFFECTS
The shortness and equality of the P-N bonds in the trimeric and
in the tetrameric chloride have been demonstrated in Section IV,B. The
thermochemical evidence for a bond order greater than unity in the
trimeric chloride is confirmed by the high value of the observed P-N
stretching frequency, which is maintained or even increased as the series
is ascended. The effect of mass on ring vibration frequencies is not large
(94, 26), so that the higher and lower infra-red absorption frequencies
characteristic of the phosphonitrilic fluorides and bromides, respectively,
are indications of their relative bond strengths, and are consistent with
the aromatic theory.
376
N. L. PADDOCK AND H. T. SEABLE
The phosphonitrilic halides are weaker bases than are some related
nonaromatic compounds, just as pyridine is a weaker base than piperidine. The bond type in the phosphonitrilic series is probably the same
as in the phosphine-imines R3P:NR (80, 8 1 ) ; the stability of these
compounds, too, increases with the electronegativity of the substituents
on the phosphorus. Tetraphenylphosphine-imineis one of the most stable
of these compounds, and forms a hydrochloride. Its base strength is
therefore greater than those of the phosphonitrilic halides, presumably
because the opportunities for electronic delocalization are less. The
N-ethyl ester of the trimeric phosphonitrilic acid, which has a cyclic
structure, but in which resonance of the same type as in the phosphonitrilic halides cannot occur, also forms a hydrochloride (61).It is, therefore, a comparatively strong base, in spite of the inductive effect of the
two oxygen atoms on the phosphorus. The extreme weakness of the phosphonitrilic halides as bases cannot therefore be wholly due to the inductive effect of the halogens.
The connection between aromatic character and base strength is confirmed by comparisons between members of the phosphonitrilic chloride
series. The calculated difference in *-electron energy between the trimer
and tetramer (Table VI) is greater than between successive higher members. The P-N bond stretching frequency (Table V) in the trimer is
appreciably less than in the tetramer ; correspondingly, the base strength
of the trimer (Fig. 2) is the greater. As the series is ascended, successive
differences between frequencies and between base constants become
smaller, as expected, though it would be unwise to draw too detailed
conclusions from their numerical values.
The trimeric bromide is a stronger base than the chloride; the
fluoride is weaker again, since it is by far the least soluble in sulfuric
acid. Qualitative observations show that ease of attack by nucleophilic
reagents decreases with increasing base strength. I n the absence of aromaticity, different effects occur. Skrowaczewska and Mastalerz (75')
investigated the action of hydrogen chloride on compounds of the type
R2N * PO(OR)s, where the substituents R may be the same or different.
The P-N bond is presumably single. The products of the reaction are
the amine hydrochloride and the chloro-ester. Substituents which inmeasc the base strength increase susceptibility to attack, the proposed
mechanism involving the intermediate formation of a salt
l&N.PO(OR)Z
+ HC1 e [&hH.PO(OR)21C1-
(5)
the positive charge facilitating attack on the phosphorus by tlie chlorine

atom of another hydrogen chloride molecule.
This possibility is not open to the phosphonitrilic halides, which are,
377
by reason of their aromatic character, too weakly basic to form hydrochlorides. The partial positive charge on the phosphorus atom and its
reactivity to nucleophilic reagents therefore increases with the electronegativity of the attached groups. The basicity of the nitrogen is thereby
diminished as a secondary effect.
The charge separation which has been mentioned in Section V, A. as
being one of the characteristics of d w - p aromaticity can cause stabilization in its own right, by electrostatic interaction in the sense P+N-.
One would certainly expect structures of this type to be very flexible,
and the larger rings to be stable. A similar polar bond Si+O- has been
suggested to account for the high internal mobility observed in the
siloxane series (67).This effect cannot, however, be a major factor in
the phosphonitrilic series, because it would require the strongest base
to exhibit the highest P-N stretching frequency, contrary to observation.
C. GEOMET~ICAL
EFFECTS
More detailed consideration of the tetrameric halides brings out some
of the other factors which affect stability. On the simple theory, the
tetramer should be more aromatic than the trimer, and should be planar.
The evidence on both points is conflicting. The relative magnitudes of
the P-N stretching frequency and the relative base strengths of the
trimeric and tetrameric chlorides suggest the greater aromaticity of the
tetramer; the relative bond lengths and the puckered structure of
the tetrameric ring suggest the contrary.
As noted above, coplanarity is not likely to be such an important
requirement for dr-pr as for p r - p n bonding. Hybridization of the spZ
type at the nitrogen would require an N-P-N angle of 150" for a flat
ring. Steric requirements for strong U- and &bonding may well, therefore, conflict to some extent with resultant puckering to a boat form
and lengthening of the P-N bond from 1.61 A in the trimeric chloride
to 1.67 A in the tetrameric chloride. This figure is still well below the
1.78 A determined for a single P-N bond (39).
Inter- and intramolecular forces between the large nonbonded chlorine atoms are also likely to be important, as shown by the structural
change which takes place on dissolution and which may sccount in part
for the discrepancies noted above. In the tetrameric fluoride, however,
not only is the aromaticity accentuated by the electronegative substituent, but, since the halogen atoms are smaller, they interact lesF strongljThe more nearly planar chair f o m i3 therefore adopted.
There is no detailed structural knowledge of the medium polymers,
though the infrared spectra do allow a comparison of the P-N stretching
frequencies of the individual members. As an indication of aromatic
378
N. L. PADDOCK AND H. T. SpuaLE
character, however, the observed frequency must be used cautiously.

This is because molecular vibrations are not independent of one another,
and even if it were possible to isolate a pure bond-stretching constant,
it would be a measure of the total bond strength, not merely of the
contribution of the r-electrons. Nevertheless, it is consistent with the
present theory that the P-N frequency (Table V) should increase to a
maximum with increasing ring size, decrease with increasing folding of
the ring, and increase again to its highest value in the rubber-like high
polymer, in which steric inhibition of mesomerism should be least.
Tentative confirmation of this view is provided by another general
feature of the infra-red spectra. As the chloride series is ascended, absorption near 750 cm-I, which is absent in the trimer, increases in intensity, as shown in Fig. 3. Although an aromatic phosphonitrilic ring
should be fairly flexible, a high degree of folding, which must occur in
the pentamer upwards, would be expected to lead to uncoupling of the
n-electrons and the formation of single bonds, with a characteristic
absorption in the region observed. The assignment of the frequency must
be regarded a t present, however, only as a working hypothesis.
A further feature of the theory ($3) is that, because the highest
filled orbital is nondegenerate, a phosphonitrilic dimer would be expected to be less unstable relative to the trimer than is cyclobutadiene
to benzene. The dissociation of P4N4F2Clsand P4N4F4C14 (Section 111,
C, 1) , probably to dimeric forms, is suggestive but not conclusive; PsNaFa
and P4N4F8 are reported to be stable to 300" (75) and P4N4C18 to
4 4 5 O (73).
D. A GENERALELIMINATION
REACTIOS
The formation of phosphonitrilic chlorides by ammonolysis of phosphorus pentachloride is considered (Section 11, D) to proceed by formation of ammonium hexachlorophosphate, followed by elimination of hydrogen chloride and a series of related condensation processes. Similar
reactions are quite common in inorganic chemistry, as will be seen below.
Their common feature is that the transfer of charge characteristic of
intermediate salt or complex formation between donor and acceptor
molecules facilitates simultaneously the release of positive and negative
groups from adjacent atoms. In ammonium hexachlorophosphate, for
example, the formation of the salt reduces electrostatic interaction, making easier the removal of the hydrogen atoms as protons. A chloride ion
would be easily detached from a hexachlorophosphate ion.
There are many examples involving ammonium ions; for instance,
the decomposition of ammonium nitrate to nitrous oxide. Ammonium
379
tetraphenylborate on heating yields benzene, ammonia, and triphenyl

boron (94).
NHBPhr + NHa
+ CsHs + PhaB
(6)
From tetramethylammonium chloride and lithium borohydride the

thermally stable tetramethylammonium borohydride is produced (4),
supporting the hypothesis that a comparable salt is formed intermediately
in the reactions of borohydrides with ammonium chloride to give boraaole (69),
NH4Cl+ LIB&
-+
[NHIBH~]-+ f(BH*NH)s 3H2
(7)
or with methylammonium chloride to give N-trimethyl borazole (68).

MesNHaCl
+ LiBH4-t [NH8Me.BHJ+ B(BH.NMe)s + 3H2
(8)
The reaction of boron trichloride with ammonium chloride is even more

nearly analogous to that of phosphorus pentachloride ( 4 8 ) ,
NH4Cl
+ BCla
--f
[NH4BC14]--t +(NH.BCl)s
+ 3HC1
(9)
The. corresponding ammonium tetrafluoroborate, like ammonium hexafluorophosphate, is stable to heat, and does not give the B-trifluoroborazole.
The reaction is not restricted to phosphorus, silicon, and boron compounds; the tetrameric (NSEt)4 is, for instance, formed from ethylamine
hydrochloride and sulfur dichloride (49). Nor is it restricted to nitrogen
compounds; the very stable cyclic phosphinoborines are formed by elimination of hydrogen from the dimethyl-phosphine borine complex ( 1 7 ) .
The initial charge separation makes the reaction possible; its completion clearly requires stability of the products, which may be attained
in different ways. In the borazoles and the phosphonitrilics, aromatic
character is postulated, and if conditions do not allow its development
cyclization is discouraged. I n the phosphonitrilic series, the strength of
the r-system increases with the electronegativity of the substituents on
the phosphorus, so that it is perhaps not surprising that derivatives of
phosphorus pentachloride carrying groups of low electronegativity like
methyl do not easily give cyclic polymers on treatment with ammonium
chloride.
VI. Conclusion
The phosphonitrilic chlorides have normally occupied a somewhat

uncomfortable place in textbooks of inorganic chemistry, in that although
they contain no carbon, in their properties they resemble organic rather
380
than inorganic compounds. This review will have succeeded in its purpose if it has demonstrated their systematic relation to other groups of
compounds, both organic and inorganic, through the mode of their formation and their properties. Their theoretical treatment as r-electron
systems extends the conventional modes of thought, applicable particularly to womatic hydrocarbons, and can act as a guide to further experimental work.
The range of experimental methods employed has not so far been
very wide; further work on the spectroscopic, thermochemical, and magnetic properties and on the kinetics of the formation and substitution
reactions can be expected to be illuminating. Most work has been done
on the trimeric chloride, the lowest stable member of the series, and the
most abundant. As in the boron hydride series, significant advances are
likely to result from investigations of the higher polymers.
LIST OF SYMBOLS
R-U*
U-d
Unit cell dimension (Angstrom units,&

Angles between triclinic cell axes; in monoclinic systems @=angle
between a and c axes (degrees)
Or: Refractive indices of biaxial crystal in increasing order of magnitude (numbers)
Extraordinary, ordinary, refractive indices of uniaxial crystal
Optic axial angle (degrees)
Absolute temperature
Melting point, boiling point
Vapor pressure ; following letter denotes the condensed phase
Density; measured except where otherwise stated. The subscript denotes the temperature in degrees Centigrade
Crystal, liquid
exp. Coefficient of cubical expansion
Standard heat of formation
Latent heat of sublimation
Latent heat of fusion
Latent heat of vaporization
Dipole moment (Debye units, D)
Or: lo- cm; m p = lo- cm
Electronic transition of nonbonded electrons to a higher antibonding
level
Electronic transition from a r-ground state to a higher antibonding
level
Z
Pnam
Molar extinction coefficient

Extinction coefficient, path length 1 cm, concentration 1% (/v)
Number of molecules per unit cell.
This and other space-group symbols are explained in International
hhles for S-ray Crystallography, Kynoch Press, 1952.
381
REFERENCES
*
1. Albright & Wilson (Mfg.) Ltd., Patent applied for.
2. Albright & Wilson (Mfg.) Ltd., Research Department, unpublished work.
3. Audrieth, L. F., Steinman, R., and Toy, A. D. F., Chem. Rev. 32, 109 (1943).
4. Banus, M.D., Bragdon, R. W., and Gibb, T. R. P., Jr., J. Am. Chem. SOC.74,
2346 (1953).
6.Bastiansen, O., Hedberg, L., and Hedberg, K., J. Chem. Phys. 27, 1311 (1957).
6. Besson, A., Compt. rend. 114,1479 (1892).
7 . Besson, A., and Roaset, G., Compt. rend. 143,37 (1906).
8.Bode, H.,
2.anorg. Chem. 252, 113 (1943).
9. Bode, H., Structure Repts. 12,227 (1949).
10. Bode, H., and Bach, H., Ber. 75B, 215 (1942).
11. Bode, H., and Clausen, H., Z.anorg. Chem. 258,99 (1949).
12. Bode, H., and Clawen, H., 2.anorg. Chem. 265,229 (1951).
13. Bode, H.,
and Thamer, R., Ber. 76B, 121 (1943).
14. Bode, H., Butow, K., and Lienau, G., Ber. 81,547 (1948).
16, Bondi, A.,J . Phys. Chem. 55,1355 (1951).
16. Brockway, L. O., and Bright, W. M., J . Am. Chem. SOC.65, 1551 (1943).
17. Burg, A.B., and Wagner, R. I., J. Am. Chem. SOC.75,3872 (1953).
18. Compagnie Francaise des Matieres Colorantes, British Patent 774,694 (May 15.
1957); U.S.Patent 2,782,133(Feb. 19,1957).
19. Corbridge, D. E. C., Actu C w s t . 6,104 (1953).
20. Couldridge, W.,
J . Chem. SOC.53,398 (1888).
21. Craig, D.P., Chem. & Znd. p. 3 (1958).
22. Craig, D.P., and Magnuason, E. A., J . Chem. SOC.p. 4895 (1956).
23. Craig, D. P., and Paddock, N. L., Nature 181,1052 (1958).
24. Daasch, L. W., J. Am. Chem.Soc. 76,3403 (1954).
26. Daasch, L. W., and Smith, D. C., Anal. Chem. 23,853 (1951).
26. de Ficquelmont, A. M., Compt. rend. 200,1045 (1935).
27. de Ficquelmont, A. M., Compt. rend. 204,689 (1937).
28. de Ficquelmont, A. M., Compt. rend. 204, 867 (1937).
29. de Ficquelmont, A. M., Ann. chim. 1111 12, 169 (1939).
30. de Ficquelmont, A. M., Magat, M., and Ochs, L., Compt. rend. 208, 1900 (1939).
31. Dishon, B., J. Am. Chem. SOC.71,2251 (1949).
32. Dishon, B. R., and Hirshberg, Y., J. Polymer Sci. 4,75 (1949).
33. Fialkov, Ya. A., Kuzmenko, A. A., and Kostromina, N. A., Ukrain. Khim. Zhur.
21,556 (1955).
34. Goehring, M., Hohenschuts, H., and Appel, R., 2.Nuturforsch. 9b, 678 (1954).
36. Gilpin, J. E., Am. Chem. J . 19, 352 (1897).
36. Goldschmidt, F., and Dishon, B., J. Polymer Sci. 3, 481 (1948).
37. Grundmann, C., and Rats, R., 2.Naturforsch.lob, 116 (1955).
38. Hildebrand, J. M., and Scott, R. L., in The Solubility of Non-electrolytes, 3rd
ed., p. 46. Reinhold, New York, 1950.
39. Hobbs, E., Corbridge, D. E. C., and Raistrick, B., Acta Cryst. 6,621 (1953).
*Other references to early work may be found in Audrieth et al. (3). Ref. 2
identifies work done by members of the Researcli Department of Albright & Wilson
(Mfg.) Ltd. Full publication may be expected during 1958-59, particular contributions being traceable through the name quoted in the text.
382
40. Hofmann, A. W., Ber. 17, 1909 (18%).

41. Holmstedt, B., and Larsson, L., Acta Chem. Scand. 5,1179 (1951).
&. Hiickel, E., Z . Physik 70,204 (1931).

43. Iribarne, J. V.,and de Kowalewski, D. G., 1.Chem. Phys. 20,346 (1952).
4. Jaeger, F. M., and Beintema, J., Proc. Acad. Sci. Amsterdam 35,756 (1932).
46. Ketelaar, J. A. A., and de Vries, T. A., Rec. tmv. chim. 58, 1081 (1939).
46. Krause, H. J., 2.Elektrochem. 59, 1004 (1955).
47. Lange, W.,and von Krueger, G.,Ber. 65,1253 (1932).
48. Laubengayer, A. W., and Brown, C . A,, J. Am. Chem. SOC.
77, 3699 (1955).
49. Lengfeld, F., and Stieglitz, J., Be7.28,2742 (1895).
60. McCauley, D. A., Shoemaker, B. H., and Lien, A. P., Znd. Eng. Chem. 42, 2103
( 1950).
61. Meyer, K. H., Trans. Faraday SOC.32,148 (1936).
69. Meyer, K. H., and van der Wyk, A. J. A., J . Polymer Sci. 1.49 (1946).
63. Meyer, K. H., Lotmar, W., and Pankow, G . W., Helv. Chim. Acta 19, 930 (1936).
54. Moureu, H., and de Ficquelmont, A. M., Compt. rend. 198, 1417 (1934).
55. Moureu, H., and de Ficquelmont, A. M., Compt. rend. 213,306 (1941).
56. Narath, A,, Lohman, F. H., and Quimby, 0. T., J. Am. Chem. SOC.78, 4493 (1956).
57. Patat, F., and Frombling, K., Monatsh. 86,718 (1955).
58. Patat, F., and Kollinsky, F., Makromol. Chem. 6,292 (1951).
69. Plattner, PI. A., Heilbronner, E., and Weber, S., Helv. Chim. Acta 32, 574 (1949).
60. Proctor, J. E., unpublished results.
61. Ratz, R., and Hem, M., Ber. 84,889 (1951).
62, Renaud, P., Compt. rend. 194,2054 (1932).
63. Renaud, P., Ann. chim. [ill 3,443 (1935).
64. Rice, R. G., Daasch, L. W., Holden, J. R., and Kohn, E. J., J . Znorg. & Nuclear
Chem. 5, 190 (1958).
60. Rochow, E. G., in An Introduction to the Chemistry of the Silicones, 2nd ed.
Wiley, New York, Chapman & Hall, London, 1951.
66. Rosset, H., Compt. rend. 180,750 (1925).
G Y . Roth, W. L., J. Am. Chem.Soc. 69,474 (1947).
68. Schaeffer,G. W., and Anderson, E. R., J. Am. Chem. SOC.71,2143 (1949).
69. SchaefTer, G. W., Schaeffer,R., and Schlesinger, H. I., J. Am. Chem. SOC.73, 1612
(1951).
70. Schenk, R., Ber. 60B,160 (1927).
71. Schenk, R., and Romer, G., Ber. 57B,1343 (1924).
72. Schmitz-Dumont, O., Z . Elektrochem. 45,651 (1939).
73. Schmitz-Dumont, 0.)
and Braschos, A,, Z . anorg. Chem.243, 113 (1939).
7 4 . Schmitz-Dumont, O., and Kulkens, H., Z.anorg. Chem. 238,189 (1938).
Y5. Seel, F., and Langer, J., Angew. Chem. 68,461 (1956).
76. Sharpe, A. G., unpublished observation.
77. Skrowaczewska,Z., and Mastalerz, P., Roczniki Chem. 29,415 (1955).
78. Specker, H., Z . anorg. Chem. 263, 133 (1950).
79. Specker, H., Angew. Chem. 65,299 (1953).
80. Staudinger, H., and Braunholtz, W., Helv.Chim. Acta 4,897 (1921).
81. Staudinger, H., and Hauser, E., Helv. Chim. Acta 4,861 (1921).
82. Steger, E., Angew. Chem. 69,145 (1957).
83. Steiniink, H., Post, B., and Fankuchen, I.,Acta Cryst. 8,420 (1955).
84. Steinman, R., Schirmer, F. B., Jr., and Audrieth, L. F., J. Am. Chem. SOC. 64,
2377 (1942).
383
86. Stokes, H. N., Am. Chem. 3.17,275 (1895).

86. Stokes, H. N., Ber. 28,437 (1895).
87. Stokes, H. N., Am. Chem. J . 18,629 (1896).
91. Stokes, H. N., 2.anorg. Chem. 19,36 (1899).
92. Tassin, W.,2.K ~ k t31,
. 304 (1899).
9% Taylor, M. C., U. S. Patents 2,796,321 and 2,796,322 (June 18, 1957) ; British Patent
767,979 (Feb. 13, 1957).
94. Ting Li Chu, 3. Am. Chem. SOC.75,1730 (1953).
THE SULFURIC ACID SOLVENT SYSTEM
R J . Gillespiel and E. A . Robinson'

University College. London; Englrrnd
I . Introduction . .
. . . . . . . .
A. Some Physical Properties of Sulfuric Acid
B . The Self-Dissociation Reactions of Sulfiiric
C . Experimental Methods . . . . . .
1. Cryoscopic Measurements . .
. .
2 . Conductimetric Measurements . . .
3. Spectroscopic Measurements . . . .
I1. Solutions in Sulfuric Acid
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Acid
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
386
387
387
390
390
391
393
394
.
.
.
.
.
.
.
.
.
.
.
.
A . Nonelectrolytes . . . . . . . . . . . . . . . . .
B. Electrolytes . . . . . . . . . . . . . . . . . .
1. Strong Bases . . . . . . . . . . . . . . . . .
a . Metal Hydrogen Sulfates . . . . . . . . . . . .
b . Simple Conjugate Acid Formation . . . . . . . . .
c . Conversion of Oxy and Hydroxy Compounds into Their Sulfato
Derivatives
. . . . . . . . . . . . . . . .
d . Anhydride Formation . . . . . . . . . . . . .
2 . Weak Bases . . . . . . . . . . . . . . . . .
3.Acids . . . . . . . . . . . . . . . . . . . .
4. Acid-Base Reactions . . . . . . . . . . . . . .
. 394
. 395
. 395
. 395
. 397
398
399
400
400
. 403
.
.
111. The Behavior of Some Inorganic Compounds in Sulfuric Acid . . . . 405

A . Sulfato and Hydrogensulfato Compounds . . . . . . . . . 405
B. Group 111: Boron . . . . . . . . . . . . . . . . . 407
1. Boric Acid and Boric Oxide . . . . . . . . . . . . . 407
2 . Tetra(hydrogensu1fato)boric Acid . . . . . . . . . . . 408
C . Group IV . . . . . . . . . . . . . . . . . . . .
410
1. Silicon . . . . . . . . . . . . . . . . . . . .
410
a. Triakylsilanols and Their Derivatives . . . . . . . . . 41.0
b . Dialkybilanols and Their Derivatives . . . . . . . . . 410
c . Monoakylsilanols and Their Derivatives . . . . . . . . 411
d . Silicic Acid Derivatives . . . . . . . . . . . . . . 412
2.Tin . . . . . . . . . . . . . . . . . . . . .
413
a . Hexa(hydrogensu1fato)stannic Acid . . . . . . . . . . 413
b Tetramethyltin . . . . . . . . . . . . . . . . 413
. . . . . . . 414
c . Trimethyltin Sulfate . . . . . . .
d Di-a-butyltin Diacetate . . . . . . . . . . . . . 414
e Methykitannonic Acid . . . . . . . . . . . . . . 414
3. LeadlHexa(hydrogensu1fato)plumbic Acid1 . . . . . . . . 414
416
D.Group V . . . . . . . . . . . . . . . . . . . .
1. Nitrogen
. . . . . . . . . . . . . . . . . . . 416
' Present Address: Hamilton College, McMaster TJniversity, Hamilton, Ontario,
Canada .
385
.
.
.
386
R. J . CILLESPIE AND E. A. ROBINSON
2. Phosphorus .
. . . . . . . . . . . . . . . . .
3. Arsenic . , . .
4 . Antimony , .
E. Group VI . . . .
1. Sclcnium Dioxide
2. Tellurium Dioxide
3. Sulfur Dioxide .
References
.417
. . . . . . . . . . . . . . . .417
.
. 418
. . . . . . . . . . . . . . . . . 419
. . . . . . . . . . . . . . . .
419
. . . . . . . . . . . . . . . . 420
. . . . . . . . . . . . . . . .
420
. . . . . . . . . . . . . . . . 421
I. Introduction
Interest in sulfuric acid-water mixtures and anhydrous sulfuric acid

as solvents has considerably intensified in recent years following the
pioneer work of Hantzsch (53, 54),mainly in the period 1907-1911, and
the work of Hammett (55, 56, 57, 84) in the 1930s. It is perhaps surprising that the chemical and physical properties of anhydrous sulfuric
acid have been comparatively little investigated until recent years considering that sulfuric acid is probably the most widely used of inorganic
chemicals and is available in an excellent state of purity.
This review is concerned in particular with recent developments in
the chemistry of solutions in anhydrous sulfuric acid. As a solvent it is
of interest from % number of points of view. It has long been used as a
medium for certain reactions such as, for example, aromatic nitration,
although its role in this particular reaction has only recently been understood ( 3 6 ) . It is a good solvent for a wide variety of organic compounds, w r y many of which give stable solutions from which they may
be recovered unchanged simply by dilution with water. It is a highly
acidic medium and as such has been used for the study of the basicity
of very weak bases such as ketones and nitro-compounds. Although it is
an extremely weakly basic solvent a few substances have nevertheless
been found which show acidic behavior. I n general our knowledge of
acids and bases has been considerably widened as a result of the study
of sulfuric acid solutions. Our understanding of the behavior and relationships of a wide variety of sulfates, sulfato complexes, and mixed
oxy- and hydroxysulfates can be considerably increased by considering
them in terms of the sulfuric acid solvent system just as Franklin did
so successfully for nitrogen compounds in the liquid ammonia solvent
system (27). Sulfuric acid resembles water as a solvent in many ways
despite its much higher acidity; and a comparison of such properties as
the conductivities, viscosities, densities, and activities of electrolyte solutions in sulfuric acid with those of aqueous electrolyte solutions is proving
valuable in aiding our understanding of these solutions and of electrolyte
solutions in general.
387
A. SOMEPHYSICAL
PROPERTIES
OF SULFURIC
ACID
Sulfuric acid of 100% composition may be prepared (37) by adding
aqueous acid to oleum until it has the maximum freeaing point of 10.371OC.
It is a very viscous liquid with a conveniently wide liquid range. It has
a dielectric constant higher than that of water and a conductivity greater
than that of most other pure liquids. A minimum in the conductivity of
the Hz0-S03 system occurs very nearly, but not exactly, at the composition H2SO4;in fact at a molal concentration of 0.002 excess water at
25" ( 4 9 ) . Its high viscosity and boiling point show that it is a highly
associated liquid in which there are presumably strong hydrogen bonds
between the molecules. It has been shown that the solid has a layer-type
structure in which each sulfuric acid molecule is hydrogen-bonded to
four others (78). The structure of the liquid may well resemble that of
the solid, just as that of water is in some important ways similar to that
of ice. The existence of hydrogen bonds, of length 2.85 A, between sulfuric acid molecules has been deduced from X-ray diffraction data on
97% aqueous sulfuric acid (19,aO). Values of some of the physical constants of sulfuric acid are given in Table I.
TABLE I
SOMEPHYSICAL
CONSTANTS
OF SULFURIC
ACID
Temperature
Property
Freezing point
Boiling point
Viscosity
Density, d2S
Dielectric coastant
Specific conductance
Heat capacity
Heat of fusion
Value
10.371"
290-317"
24.54 centipoise
1.8269
100
120
1.0439 X lo-* ohm-lcm-l
0.338 cal/deg/gm
2560 cal/mole
("C)
References
69
87
25
25
25
10
25
25
10.37
4-4
4.4
8,349 46
8,349 46
49
68
67
B. THESELF-DISSOCIATION
REACTIONS
OF SULFURIC
ACID
Despite its extremely high acidity sulfuric acid belongs to the large
class of solvents known as amphiprotic or amphoteric solvents which
have both acidic and basic properties. In all solvents of this type proton
transfer between solvent molecules, or autoprotolysis, occurs in which
one molecule acts as an acid and the other as a base. Thus in water we
388
R. J. GILLESPIE AND E. A. ROBINSON
have the familiar reaction,

in liquid ammonia the reaction
and in sulfuric acid
2HZSO, = H&3OI+
+ HSOd-.
(3)
The extent of the autoprotolysis is a measure of both the acidic and

basic strengths of the solvcnt and is given by the autoprotolysis constant
or ionic product; for example, for water K,, = [H80+] [OH-] = 10-14
(25") and for sulfuric acid K,, = [H3S04+] [HS04-] = 1.7 x
i l O o ) . The autoprotolysis constant of sulfuric acid is greater than that
for any other solvent that has been studied. Such a large value implies
that, in spite of its very high acidity, sulfuric acid must also he appreviably basic.
There are, however, some additional self-dissociation processes in
sulfuric acid which niust also be considered. The primary self-dissociation reaction mnv be considered t o be
2H,SOI = HzO
+ HPS~OF.
(4)
Both of these products are further ionized ( 28); water extensively according t o the equation
H20
+ H2SO4 = H30+ + HSO4-
(ti)
and disulfuric acid partially, according to the equation

HzSzOi
+ HzSOc = H&O4+ + HS207-.
(6)
Since the ions H3S04+ and HSOa- are in equilibrium as a consequence

of the autoprotolysis reaction [Eq. ( 3 ) ] it follows that the ions H:@+
and HS207- must also be in equilibriuni
2H2SO4 = HaO+
+ HSZ0;-.
(7)
This has been called the ionic self-dehydration reaction ( 2 9 ) .Only four of
the five equilibria, Eqs. (3) to (7) are independent and it has been
found convenient to discuss the self-dissociation in t e r m of Eqs. (31,
( 5 ) , (61, and ( 7 ) . Values for the corresponding equilibrium constants
are given in Table 11. The values a t 10" were obtained from a detailed
study (4, 6 ) of the freezing points of solutions of metal hydrogen
sulfates, water, and disulfuric acid, each of which represses the selfdissociation equilibria in a different way. Table I11 gives the concentration of each of the products of the self-dissociation. The total molal
concentration of 0.0424 a t 10" corresponds to a freezing point of 10.625"
389
TABLE I1
CONSTANTSFOR THE SELF-DISBOCIATION
EQUILIBRIUM
REACTIONS
OF SULFURIC
ACID
Reaction
10"
25"
for the hypothetical undissociated acid. The values of the self-dissociation constants a t 25" were obtained (266) from the values a t 10" and
the electrical conductivities at 10" and 25" of solutions of metal hydrogen sulfates, water, and disulfuric acid. The values given in Tables I1 and
I11 are more accurate than those that have been given previously (88,89,
TABLE I11
(m)OF THE SELF-DISSOCIATION
SPECIES
CONCENTRATIONS
IN SULFURIC
ACID
Species
10"
25"
~-
HSOdHaSOI+
H,O+
HS207H&OT
H*O
~~
0.0150
0.0113
0,0080
0.0044
0.0036
o.Ooo1
Total
-~
0.0178
0.0135
0.0088
0.0045
0.0043
o.Ooo1
0.0424
0.0490
62) as certain simplifying assumptions which are not entirely justified
were made in earlier work.

The ions that are formed by the autoprotolysis of any solvent are
of particular importance in the chemistry of solutions in that solvent.
The anion is formed by the ionization of any base and is the strongest
possible base that can exist in that solvent. The cation is formed by any
acid and is the strongest acid in the solvent. Thus in sulfuric acid the
strongest possible base is the hydrogen sulfate ion and all intrinsically
stronger bases are fully ionized to give the hydrogen sulfate ion
B
+ H&04
Acids give rise to the HsSOa+ ion
BH+
+ HSOd-.
(8)
390
C. EXPERIMENTAL
METHODS
The two main methods by which sulfuric acid solutions have been
investigated are cryoscopic and electrical conductivity measurements.
Some use has also been made of Raman, infrared, and ultraviolet absorption spectra.
1. Cryoscopic Measurements
The cryoscopic method was used in Hantzsch's original work (53, 54)
and was later developed by Hammett and Deyrup (56) and by Gillespie
et al. (48). Sulfuric acid freezes a t 10.371"C and has a relatively large
molal freezing-point depression or cryoscopic constant k = 6.12 (3.2).
It is thus a convenient solvent for cryoscopic measurements provided that
adequate precautions are taken to prevent absorption of water from the
atmosphere. From freezing-point measurements the number, V, of moles
of particles (ions and molecules) produced by one mole of any solute may
be obtained. Some examples are given in Table IV.
TABLE IV
EXANPLES
OF VALUESOF v
AND 7 FOR
VARIOUBSOLUTES
System
+
+
+
+
+
+
+
+
+
+ +
+
+ +
+
+
+
+
+
+
+
+ +
CH3C02H HzS04 = CH~COIHZ+ HSOICzHbOH H2S04 = CzHaSO4H H30+ HSO4CaHa(NH2)2 2H2S04 = C&(NHs)z++
2HSO4(CHs)sSiOSi(CHa)3 3HzS04= 2(CH&Si . SO4H HIO+ HSOaH"0s
2H2S04 = NOz+ HjO+ 2HSO4Bz03 9H2SO4 = 3Hs0+ 2B(HSO4)4- HSOI6HZSOI = 3HaO+ B(HSO4)(- 2HSO4K204 3HzSO4 = NO+ WOz'
H30+ 3HSO4-
2
3
3
1
1
4
4
6
6
2
1
2
1
2
A good deal of confusion and much controversy between early workers

(54, 55, 7 7 ) resulted from a lack of understanding of the nature and
extent of the self-dissociation of sulfuric acid, which is repressed by

most solutes and thus affects the freezing-point depressions that they
produce. Following Hammett and Deyrup (56) it became common practice to carry out cryoscopic measurements in sulfuric acid containing
sufficient water to depress the freezing point to approximately 10.0" in
order to repress largely the solvent self-dissociation. This is not, however, an entirely satisfactory procedure. The self-dissociation is not completely repressed ( 5 ) , and allowance for it should still in principle
be made. Moreover water has been found, in the case of nonelectrolytes and weak electrolytes at least, to have an effect on the depression
391
produced by the solute (10, 39, 42). This has been attributed by Brayford and Wyatt (10) to salting-out and salting-in effects of the
hydroxonium hydrogen sulfate produced by the water. These effects have
led, in at least two cases, to erroneous or inaccurate interpretations of
freezing-point measurements, carried out in the presence of small amounts
of water (7, 10, 39, 42). Since the repression of the self-dissociation
equilibria by the added electrolyte can be allowed for by making use
of the constants given in Table I1 (6),it would seem preferable, except
perhaps when only very approximate results are required, to use 100%
sulfuric acid, rather than slightly aqueous acid, as a solvent for cryoscopic measurements, as has in fact been done in recent work (see e.g.
94, 39).
For freezing-point measurements made in 100% sulphuric acid v is
given by the expression
v
=O(1
+ 0.0028)/6.12+m- md/m
where 6 is the freezing-point depression calculated from the freezing-point

(10.625) of hypothetical undissociated sulphuric acid, is the osmotic
coefficient, m is the stoichiometric concentration of the solute allowing for
any solvent used up in its ionization and ma is the total concentration of
the self-dissociation species (6, 34). Since + is in general not known it
must be assumed to have the value of unity and an approximate value of
v, denoted here by 4,may then be calculated from the expression
=e(l
+ 0.0028)/6.12m - md/m.
Some workers who have used slightly aqueous sulphuric acid as solvent
have expressed their results in terms of the van%Hoff i-factor
i = O/6.12m
Thus i is a further approximation to
V.
2. Conductimetric Measurements
A few measurements of the conductivities of solutions in sulfuric acid

were made by the early workers (6,63,61,63)but there were considerable
differences between their results and conclusions, again partly because
of a lack of understanding of the solvent self-dissociation and the contribution that it made to the conductivities of the solutions studied. A
comprehensive investigation of the conductivities of solutions in sulfuric
acid was first made by Gillespie and Wasif (46) and this showed that
such mea&ements were of considerable value in the study of the
ioniaations of solutes in sulfuric acid.
For a pure liquid, sulfuric acid has a high electrical conductivity
(@)-0.01044 ohm-lcm-l at 25. This is due primarily to its extensive
392
R. J . GILLESPIE AND E. A. ROBINSON
autoprotolysis and also to the fact that both the ions thus formed, H3S04+
and HS04-, have relatively high mobilities. The mobilities of these ions
are in fact much greater than those of any other ions in sulfuric acid.
This is shown, for example, by the values given in Table V for the cation
TABLE V
C'ATIOS TRANSPORT
NVMBERS
FOR SOMEELECTROLYTES
(49)
Electrolyte
AgHSOi
KHSO,
SaHS04
LiHSO,
Ba(HSO&
Sr(HSO&
t.
0.249
0.624
0.792
0.556
0.174
0.211
0.026
0.030
0.021
0.013
0.009
0.007
transport numbers, t + , of various metal hydrogen sulfates (43). The

mobilities of these metal ions are extremely small in sulfuric acid because of its very high viscosity-24.5 centipoise a t 25O-which is approximately 25 times that of water. The hydrogen sulfate ion and the
hydrogen ion (H3S04+)are however unaffected by this high viscosity
because they conduct almost entirely by a proton-transfer mechanism
(43) quite analogous to that generally accepted to explain the abnormally high mobilities of H 3 0 + and of OH- in water ( 8 0 ) . The mechanism is illustrated very diagrammatically for H3SOe+ in Fig. 1. Successive transfer of protons along hydrogen bonds result in an effective
movement of H3S04+ ions through the solution without the need for
the actual diffusion of individual ions. Values of the mobilities of H3S04+,
393
TABLE VI
IN 100% SULFURIC
ACIDAND WATER
IONICMOBILITIES
AT INFINITE
DILUTION
(25')
Ion
h0(HnS04)
Xo(HnO)
-
349.8
198.6
50.1
73.5
63.6
(80)
HS04-, and other ions in 100% HzS04 are given in Table VI together
with the values for some ions in water a t infinite dilution. The mobilities
of the H3S04+ and HS04- ions in sulphuric acid are quite comparable
with those of the H30+ and OH- ions in water. However, the mobilities
of other ions in sulphuric acid are much smaller than the mobilities of
the corresponding ions in water because of the very high viscosity of
sulphuric acid.
The conductivities of solutions of acids and bases in sulfuric acid are
thus determined almost entirely by the concentration of H3S04+ or
HS04- respectively. All simple bases ionizing according to Eq. (8), that
is, mono (hydrogensulfates) , have very similar molar conductances and
all bases which give rise to two hydrogen sulfate ions, that is, di(hydrogensulfates) such as nitric acid
HNO8
+ 2H~S04= NO,+ + &Of + 2HSO4-
(10)
have molar conductances approximately twice those of a mono (hydrogensulfate) a t the same hydrogen sulfate ion concentration (45). Molar
conductances of a number of typical electrolytes are given in Table VII.
Electrical conductivity measurements thus provide a means of determining the number, y , of moles of hydrogen sulfate ions or hydrogen ions
produced by one mole of any solute. Some typical examples are given
in Table IV. The value of y from conductivity measurements taken
together with that of Y from freezing-point measurements often enables
the mode of ionization of a solute t o be determined unambiguously.
3. Spectroscopic Measurements
Differences in the visible and ultraviolet absorption spectra of bases

and their conjugate acids have been utilized in determining their degrees of ionization in sulfuric acid-water mixtures (61,68, 55) and, in
394
TABLE VII
MOLARCOWDUCTANCES
OF SOME
ELECTROLYTES
(45)
Mono(hydrogensu1fates)
31
KHIHSOc
(CHa)pCO
Hz0
CZHsOH
CHSCOaH
0.1
0.2
0.3
0.4
0.6
0.8
1 .o
157
118
104
94.4
82.2
73.7
66.3
I53
116
104
95.0
80.1
72.4
63.9
150
114
100
90.5
78.3
69.6
63.2
156
117
103
93.1
79.8
70.1
62.2
152
116
99.3
89.5
77.1
67.9
59.8
__
Di(hydrogensulf ates)
HNOs
o-C&(NHAr
Ba(HS04h
0.05
0.10
0.15
0.20
0.30
0.40
304
228
201
177
155
138
306
236
20 I
187
153
130
304
221
179
150
132
110
the case of very weak bases such as nitro compounds, in oleums ( 7 ) .

Raman spectra have been used to identify and to measure approximately the concentration of various species in sulfuric acid solutions.
Thus the nitronium ion NO2+ was identified (17, 58, 76) and its linear
structure proved by the observation of its single sharp and intense Raman
line a t 1400 em-', and the hydrogen sulfate ion may be recognized and its
concentration estimated by a characteristic strong Raman line a t 1050
cm-l (58,75).
II. Solutions in Sulfuric Acid
A. NONELECTROLYTES
Sulfuric acid is a poor solvent for nonelectrolytes. The hydrogenbonding interactions between the sulfuric acid molecules are so strong
that unless 8 solute is strongly solvated, either because it is ionic
or because it can strongly hydrogen bond with sulfuric acid, it is unlikely to be able to disrupt sufficiently the structure of the sulfuric acid
to enable it t o dissolve. Yet if a molecule is sufficiently basic to form
strong hydrogen bonds with sulfuric acid it is likely that some proton
transfer along the hydrogen bonds will occur, resulting in a t least a small
degree of ionization. Thus in order to behave as a nonelectrolyte a
395
solute must have sufficient basic characteristics to form strong hydrogen

bonds with the solvent but must be insufficiently basic to protonate. The
only nonelectrolytes that are known at present are sulfuryl chloride ( 4 8 ) ,
picric acid ( l o ) , and a few other polynitroaromatic compounds (7, 10,
4 2 ) . Trichloracetic acid and chlorosulfonic acid may also be nonelectrolytes although the evidence is not conclusive (45, 48). Dimethylsulfate and phosphorus oxychloride, which early workers believed to be
nonelectrolytes, have been shown to react with the solvent ( 4 8 ) .
It is interesting to note that the conductivities of solutions of nonelectrolytes in sulfuric acid are lower than that of the solvent itself,
presumably because the nonelectrolyte decreases the mobilities and
perhaps also the concentrations of the highly conducting H3S04+ and
HS04- ions (42, 4 5 ) . The effect of nonelectrolytes on the mobilities of
these ions has been attributed to the strong hydrogen-bonding interactions between the solute and the solvent molecules. This decreases the
effectiveness of the latter in the proton-transfer conduction process (45).
B. ELECTROLYTES
Sulfuric acid is a good solvent for electrolytes, as is consistent with
its high dielectric constant and the polar nature of its molecules. Many
of the soluble electrolytes are rather unfamiliar and cannot be obtained
a t all in aqueous solution, whereas many of the electrolytes that are
encountered in aqueous solution are not stable in sulfuric acid.
1. Strong Bases
a. Metal Hgdrogen Sulfates. The alkali metal and some other metal
hydrogen sulfates dissolve in sulfuric acid as fully ionized binary electrolytes and are strong bases analogous to the hydroxides in water
The corresponding normal sulfates are completely converted to the

hydrogen sulfates and formally they are analogous to the metal oxides
in water (37).
The solubilities of a number of metal sulfates are given in Table

VIII. The sulfates of Al, Fe(III), Ni(II), Be, and Cd(I1) have also
been investigated and found to have solubilities less than 0.1 mole %
( 2 3 ) . It is interesting to note that the solubilities of the hydrogen sulfates in sulfuric acid closely parallel those of the hydroxides in water and
396
TABLE VIII
SOLUBILITIES
OF METALSULFATES
IN SULFURIC
ACID AT 25"
Sulfat,e
Solubility
(mole %)
Solid phase
Reference
60
60
59
-59
60
59
14.28
9.28
9.11
8.85
5.28
5.16
Soluble > 2
Soluble > 2
Soluble > 2
Soluble > 2
Soluble > 2
23
25
$3
93
85
1.2
0.78
0.18
0.17
0.19
0.08
0.02
cuso4
HgSO4
65
59
-5.9
59
59
59
59
the acetates in anhydrous aretic acid. As is indicated in Table VIII the

solid phase in equilibrium with the saturated solution is in the case of
the more soluble sulfates a solvate of the hydrogen sulfate, while, for
the less soluble it is the unsolvated hydrogen sulfate or the normal
sulfate. The formation of solvates is some indication that these electrolytes are solvated in solution. Additional evidence of this solvation
has been obtained from freezing-point (37), density (44) viscosity (33,
44) conductivity (45) and transport-number (43) measurements on
their solutions. The extent of solvation decreases from Li+ to Cs+ for
the alkali metals and the divalent B a t + and Sr++ are more highly
solvated than any of the alkali metal cations.
Salts of other inorganic acids are almost without exception either
insoluble, for example, AgCl, CuBrP, A1Cl3, and A1POt7 or undergo complete solvolysis as is illustrated by the following examples:
+ HtS04 = Na+ + HSO4- + HCI

+ HzSO4 = NH4+ + HSO4- + HCIO4
SaaPOI+ 3HzS04= 3Na+ + 3HS04- + HaPOI
KNO1 + H2S04 I(+ + HS0.- + HNOa.
SaCl
SH4ClOP
its
(15)
(16)
(1 7)
(18)
Sometimes the free acid that is formed undergoes further reaction

is the case with nitric and phosphoric acids (see b and c ) . All such
397
soluble salts thus give strongly basic solutions as a result of solvolysis.

Solvolysis occurs partly because of the rather high concentration of
hydrogen sulfate ions in sulfuric acid but mainly because the acids from
which these salts are derived are exceedingly weak acids or do not behave as acids a t all in solution in sulfuric acid. Of the acids mentioned
rtbove HCl is un-ionized and sparingly soluble, and probably reacts to
some extent to give chlorosulfonic acid; perchloric acid is exceedingly
weak, and both phosphoric and nitric acids behave as bases (see Section 3).
b. Simple Conjugate Acid Formation. Most organic compounds, with
the exception of aliphatic hydrocarbons, some aromatic hydrocarbons,
and their halogen derivatives, are soluble in sulfuric acid. This is because
they contain atoms such as 0, N, S, or P with one or more unshared
pairs of electrons which almost always exhibit basic properties in sulfuric acid. Thus a large number of organic compounds behave as strong
bases, being completely converted into their conjugate acids. Sulfuric
acid may be said to exert a considerable leveling effect on the strengths
of bases. Examples include ketones,
(19)
+ HSOI-
(20)
+ I*zso4 = CH&OOHa+ + HSOA-
(21)
(22)
CIIaCO COTTO IIzSO4
CI&COII+ C ~ N S HSO4-
amines,
CsH6NH2
+ Has04 = CsHs
NH3+
carboxylic acids,
CHsCOOH
esters,
C~HLCOOC~HS
= CsH&OOH+
- CIHh + HSO4-
and amides
CsHsCONHy
+ HzS04
CsHsCONHs+
+ HSO4-.
(23)
Many of these solutions are quite stable and the organic compound
can be recovered unchanged simply by diluting the sulfuric acid solution with water. In other cases subsequent reactions ensue, such as
hydrolysis, in the case of some esters (35, 70) and amides ( 1 4 ) , for
example,
+
+
+
+
+
+
CHIOC~HICOOCHI 2HzSO4 = CHsOCd&COOH2+ CHSOiH HSO4WrCONHz HzSO4 = CeHsCOzH+ NH4+ HSOaH&O,-.
(24)
(25)
A detailed review of the behavior of organic compounds in sulfuric acid

has been given by Gillespie and Leisten (36).
Phosphoric acid provides an example of an inorganic substance that
behaves as a simple base [Eq. (@)I.
398
c. Conversion of Oxy and Hydrozy Compounds into Their Sulfato

Derivatives. A number of other types of basic behavior, in addition to
simple protonation, are also known. Thus many oxy and hydroxy compounds are converted to sulfato derivatives with the elimination of
water, which ionizes as a base according to Eq. ( 5 ) . The sulfato
compound that is formed may be a nonelectrolyte, or it may itself exhibit basic properties either by protonating or by ionizing off a hydrogen
sulfate ion, or it may even behave as an acid. The simplest examples
of this type of behavior are given by a number of substances of the
general formula XOH which are converted to the hydrogen sulfate XS04H
XOH
+ 2Hzso4
m04H
+ HsO+ + HSO4-
(26)
and thus behave as bases. In addition the hydrogen sulfate itself sometimes acts as a base either forming its conjugate acid
xso4H + HSOc = XSO~HI HSOI(27)

or ionizing into X+ and S04H- just like a metal hydrogen sulfate. Ethyl
alcohol is a base of this type being converted to ethylhydrogen sulfate
(30,48).
CIHsOH
CZHsSOIH
+ H30+ + HSOI-.
(28)
There is some evidence (SO), although it is not very conclusive that

ethyl hydrogen sulfate is itself a rather weak base ionizing slightly according to the equation
C,H&O*H
+ H2SO4 = CzHaSOiHi+ + HSO4-.
(29)
Xitric acid (St?), triphenylcarbinol (%), and mesitoic acid (35) all
provide examples of the formation of a hydrogen sulfate XS04H that
is fully ionized to X+ and HS04- so that the over-all ionizations of
these compounds are as follows:
HSOj
+ 2HZSO4 = NO%++ H,O+ + 2HSO4-
+ 2HzSO4 = PhaC+ + HjO+ + 2HSO4Me3CeH2C02H+ 2HS0, = MesCsH2. CO+ + HIO+ + 2HSO4-.

PhaCOH
(30)
(31)
(32)
Such positive ions formed by the ionization of hydrogen sulfates in

sulfuric acid are all very strongly electrophilic and can only exist in a
very weakly basic solvent such as sulfuric acid or in a suitable nonprotonic solvent. Their formation in sulfuric acid is due not only to its
high acidity but also to the very low activity of water when in dilute
Bases of the type exemplified by nitric acid and triphenylcarbinol have been
called secondary bases but as there are a number of other similar, but nevertheless
dxfferent, and equally important types of basic behavior it does not seem necessary
to single out this particular one for a special name.
399
solution in sulfuric acid. This low water activity results because

the water is largely converted to HsO+ and the remaining unionized
water is strongly hydrogen bonded to the solvent. One advantage of
sulfuric acid as a solvent for such ions is that it is a good solvent for
electrolytes whereas nonprotonic solvents are often poor solvents for
electrolytes. Thus in sulfuric acid relatively concentrated solutions of
these reactive ions can be obtained. This is of considerable importance,
particularly in those cases where such species have been postulated as
unstable reaction intermediates in other solvents. The nitronium ion is,
of course, the reactive intermediate in aromatic nitration under many
conditions ( 3 6 ) .
Anhydrides X20 can similarly be converted to a hydrogen sulfate
and thus act as bases
+ 3HzS04 = 2XSO4H + HaO+ + HSOd-.
XzO
(33)
Hexamethyldisiloxane behaves in this way (Section 111, C, 1, a)
(CHa)sSiOSi(CHs)x 3HzS04 = 2(CH&SiS04H
+ HsO++ HS04-
(34)
and the trimethylsilicon hydrogen sulfate that is formed appears to be

a nonelectrolyte. Dinitrogen pentoxide forms the fully ionized nitronium
hydrogen sulfate (Section 111, D, 1 )
Nz06
+ 3HzSO4 = 2N02+ + HaO+ + 3HS04-.
(35)
Other more complicated examples of the formation of sulfato-derivatives are given in Section 111. Only one more example of particular interest will be mentioned here, namely, that in which the sulfate formed
is an acid. Thus boric acid is converted t o boron tri(hydrogensu1fate)
(Section 111, B),
B(0H)a
+ 6HzS04 = SB(SO4H)a + 3HaO+ + 3HS04-.
(36)
But boron tri(hydrogensu1fate) is an acid and reacts with the hydrogen sulfate ion,
B(SO4H)a
+ HSOC
B(SOIH),-.
(37)
However, only one-third of the hydrogen sulfate formed is neutralized

so that the over-all reaction of boric acid is still that of a base,
B(OH),
+ 6HzS04= 3H80++ B(HS04)4- + 2HS04-.
(38)
d. Anhydride Formation. Still another type of basic behavior is provided by a number of hydroxy compounds that are converted to their
anhydrides as, for example, is the case with a number of weak inorganic
acids,
400
2. Weak Bases
Some substance6 with very weakly basic properties are only partially
ionized in sulfuric acid and it is possible to measure their degree of
ionization and hence obtain their basicity constants by means of cryoscopic, conductimetric, and spectroscopic measurements. A number of
nitro-compounds have been carefully studied by several different methods. It may be seen in Table IX that the results obtained by the differTABLE IS
DISSOCIATION
CONBTANTS
OF %ME
ConductiCyrometric
ecopic
(42, 71) (31,39, 4.2)
Base
pterl-Butylnitrobenzene
pNitrotoluene
+Nitrotoluene
m-Nitrotoluene
Nitrobenzene
p-Chloronitrobenzcirr
Nitromethane
Acetonitrile
Benzoditrile
pToluenemlfonic acid
Benzenesulfonic arid
Dipheny lsuifone
0.095
0.067
0.023
0.010
0.004
0.0025
0.16
0.07
-
0.10
0.094
0.062
0.020
0.011
0.003
0.004*
-
0.026*
0.011 *
0.016*
W E A K BASE8
Spectroscopic
Tit,rrtiont (41)
H&Or
(7)
0.10
0.077
0.024
0.013
0.024
0.009
0.0046
-
HB(HSOA
0.097
0.069
0.01i
0.009
* These values are probably rather high as the measurements were made in slightly
aqueous rather than l 0 0 ~ sulfuric
,
acid (see ref. 39).
t Calculated from the position of minimum conductivity in the titmtions of these
acids (Section 11, B, 4).
ent methods are in excellent agreement. The conductivities of solutions
of a number of nitro compounds are shown in Fig. 2 together with those
for a fully ionized binary electrolyte (C6H5C02H2+ HSOa-) for comparison. Other weak bases include sulfones, sulfonic acids, and nitriles.
Their basic dissociation constants are also given in Table IX. Some
nitriles have also been found ( 7 1 ) to undergo hydrolysis according to
the equation
CHaCN
+ 2HsSO4 = CHsCONHa++ HhOi-.
(41)
3. Acids
The majority of substances that are normally regarded as acids, that

is from the standpoint of aqueous solution chemistry, no longer behave
401
Concn (mole per kg of soh 1

FIG.2. Conductivities of solutions of some nitro compounds. KEY:A, C&COnH;
B, p-CsH,Me NO%;C , o-CaH4Me* NO*; D, m-CsH4Me NO?; E , PhNO?; F, pCaH'
CI * NO,; G, MeNOz. Reproduced from ref. 4%
as acids a t all in solution in sulfuric acid, but as bases of various types

as is shown by the following examples:
Hydrofluoric acid reacts to form fluorsulfonic acid

Hydrogen chloride is not very soluble in sulfuric acid but probably
reacts to some extent to give chlorsdfonic acid. Hydrogen bromide and
iodide are oxidised to their respective elements. Even perchloric acid,
which is often regarded as the strongest known acid, shows little or no
402
acid properties in solution in sulfuric acid, as is implied, for example,

by the virtually complete solvolysis of metal perchlorates (Section 11,
B, 1, a ) . Perchloric acid must be a very weak acid whose ionization
according to the equation
HClO4
+ HZSO, = HoSOi+ + Clod-
(48)
is very slight. It appears that fluorsulfonic acid is slightly stronger than

perchloric acid. It causes a small but definite increase in the conductivity
when dissolved in sulfuric acid, and it can be "titrated" with suitable
bases (3) (Section 11, B, 4 ) . From such measurements the equilibrium
constant for its dissociation according to the equation
HSOaF
+ HZSO, = H3SOi' + SOsF-
(49)
has been estimated and is given in Table X. The conductivity of its

TABLE S
DISSOCIATION
CONST.4NTS
Acid
OF SOME
ACIDS
Reference
I;,
1.7 x
(10")
Very weak
Very weak
1.8 x 10-8 (10")
3x
(25")
1.1 x lo-* (loo)
1.4 X lo-* (loo)
Moderately strong
Strong
* Ka(HZS0,)
= Kep/10.2.
solutions is shown together with that of other acids and the nonelectrolyte
sulfuryl chloride in Fig. 3.
The first acids of the sulfuric acid system t o be recognized (29)
were disulfuric acid, H2S207,
and the higher polysulfuric acids, I-I~S~OIO
etc., which are present in oleum. Dilute oleum contains mainly disulfuric
acid, H2S207,and from cryoscopic measurements on such solutions the
first dissociation constant of disulfuric acid has been found to be 1.4 X
( 5 ) . It is approximately 30% ionized in a 0.1 m solution;
H?&O7
+ HZSO, = H,SOp+ + HS107-.
(50)
The coiiductivities of its solutions, which are shown in Fig. 3, are also
consistent with this value for its acid-dissociation constant. The highcis
polysulfuric acids are present in increasing amount in more concentrated
oleums and they appear to be somewhat stronger acids than disulfuric
403
2.0
--' 5
'E
c
- 1.5
0
1.0
0
0.10
0.05
0.15
Concentrotion (mole per kg,,,,)
FIG.3. Conductivities of solutions of some acids.
acid but it has not been possible to measure their dissociation constants
because several of them are always present together in any oleum.
Recently evidence has been obtained for the existence of several complex hydrogensulfato acids, namely tetra (hydrogensulfato) boric acid
HB (HS0.J 4 which is a strong acid, hexa (hydrogensulfato) plumbic acid
HzPb (HS04)6 whose first dissociation constant is comparable with that
of disulfuric acid, tetra (hydrogensulfato) arsenious acid HAS(HS04)
which is very weak, and hexa (hydrogensulfato) stannic acid H2Sn (HS04)6
whose strength has not yet been accurately measured although it is
probably somewhat stronger than the corresponding plumbic acid. These
acids are fully discussed in Section 111.
4. Acid-Base Reactions
Acid-base neutralization reactions can be carried out in sulfuric acid
as in any other amphoteric solvent. In sulfuric acid the neutralization
reaction is essentially
&SO,+
+ HSO4- = 2H804.
(51)
Compare
&O+ OH- = 2H2O

(52)
in water. Since the ions H3S04+ and HSOr- have much higher conduc-
404
R. J . GILLESPIE A N D E. A. ROBINSON
tivities than any other ions in sulfuric acid such neutralization reactions
are very conveniently followed by measuring the electrical conductivity
of the solution which decreases as base is added to acid or vice versa,
passes through a minimum and then increases again (Fig. 4 ) . The position of this minimum conductivity depends on the strength of the acid
and the base. If the autoprotolysis were the only self-dissociation process
of the solvent and if the conductivity of HaS04+ were equal to that of
HS0,- then the minimum conductivity in the reaction of a strong base
"b/"'HZSZO,
'bb/"'H [E(HS0414]
FIG.
4. Acid-base conductimetric titrations.
( 4 A ) : A . Nitrobenzene:. B,. m-Nitrotoluene; C and D, Beneoic acid;
(4B); A , m-Sitrotoluene; B, Potassium hydrogensulfate; C, o-Nitrotoluene.
with a strong acid would come exactly a t the ratio of added base to
moles of initial acid nb/nal = 1.0, that is, a t the true end-point of the
reaction. If allowance is made for the difference in the mobilities of
H3S04+and HS04- and also for the other self-dissociation processes of
sulfuric acid then it may be shown (26) that the minimum conductivity
would he expected a t the mole ratio nb/?i,,'= 0.98. It is evident that
when a weak acid is titrated with a strong base the iiiiniiiiuiii conductivity will occur a t na/n," < 1.0 and when a strong acid is titrated with
405
a weak base it will occur at ne/n$ > 1.0. It may be shown (26) that in
general for an acid with a dissociation constant K, and a base with a
dissociation constant K b the minimum conductivity at 25" occurs at
nb/nai = 0.98(1
+ 0.017/&)/(1 + 0.014/Ka).*
Thus if the position of minimum conductivity is determined experimentally a value of either K , or K b may be obtained if the other is
known. Some examples of titrations involving the strong acid HB (HSO4)4
and the weak acid H&07 and several strong and weak bases are shown
in Fig. 4. Values of the basic dissociation constants of several nitro
compounds calculated from the position of the minimum conductivity
in titrations of both H2S207
and HB(HS04)4 are in good agreement
with values obtained by other methods (Table IX).
111. The Behavior of Some Inorganic Compounds in Sulfuric Acid
A. SULFATO
AND HYDROGENSULFATO
COMPOUNDS
The inorganic chemistry of the sulfuric acid solvent system has not
yet been as extensively studied as that of some other solvents such as
liquid ammonia, but a sufficient variety of certain classes of compounds
has now been investigated for it to be possible to give a reasonably
systematic account of their behavior.
Because only a very few anions are known a t present that do not
undergo extensive, if not complete, solvolysis in sulfuric acid, investigations have been mainly confined to the behavior of sulfates and hydrogen sulfates, which are the analogues of the oxides and hydroxides in
water. It must not be concluded, however, that the chemistry of solutions in sulfuric acid is therefore without any great interest, because
a great variety of different cations are known, many of which cannot
exist in aqueous solutions. When an oxide or hydroxide is dissolved in
sulfuric acid there is in general a tendency for it to be converted to a
sulfate or hydrogen sulfate with the elimination of water (Section 11,
B, 1, c), although the extent to which this occurs varies widely from one
element to another. Thus, for example, the oxides and hydroxides of the
more electropositive metals are completely converted into the corresponding hydrogen sulfates, while phosphoric acid, for example, is merely protonated and undergoes no conversion into a sulfato derivative. I n general a wide variety of compounds intermediate between the oxides and
hydroxides of the aqueous system and the sulfates and hydrogen sulfates
of the sulfiuic acid system may be formed. Some of the possible compounds and the relations between them are illustrated in Fig. 5 for the
*Except for HZO and Hi307 for which slightly different equations are needed.
406
FIG.5. Oxy-, hydroxy-, sulfato-, and hydrogen sulfato-derivatives of a trivalent

element X and a tetravalent element Y .
cases of a trivalent atom or group S and a tetravalent atom or group Y .

Moving from left to right across the diagram corresponds to the addition of water, from top to bottom to the addition of sulfur trioxide, and
diagonally from upper left to lower right to the addition of sulfuric acid.
The compounds in the top row belong entirely t o the water solvent
system while those in the bottom diagonal belong entirely to the sulfuric acid solvent system. It appears that one of the important factors
determining the extent of sulfation is the electronegativity of the group
S (or Y ) , t h e extent of sulfation increasing with decreasing electronegativity. The formulae given in Fig. 5 are empirical only, and are
not intended to represent the actual molecular constitution of the compounds. Thus they may in fact be polymeric, for example, Sbz03 and
RSiO HSOa (Section 111, C, 1, c ) , or they may be ionized either
completely or partially as bases, for example,
NO . HSO'
KO+
+ HS0,-
CHI Sn(HS04)r = CHI . Sn(HSO,)z+

f
+ HSOI-
(53)
(54)
or in a few cases as acids, for example,

B(HSO4)s ---t HB(HSOI), -+ HaSOI+
+ B(HS01)d+ Pb(HSO4)s"
Pb(HSO4)r + HZPb(HSO4)s -+ 2HaSO4'
(55)
(56)
in exactly the same way as oxides and hydroxides may behave as either
acids or bases, or both.
We are now in a position to discuss the behavior of the compounds
of some of the elements of Groups 111, IV, V, and VI, that have so far
been studied, Apart from the fact that compounds of the alkali and
alkaline earth metals are generally converted to ionic hydrogen sulfates
407
and that compounds of a number of other metals generally give insoluble

sulfates (Section 11, B, 1, a ) practically nothing is known of the behavior of the compounds of elements other than those discussed below.
B. GROUP111: BORON
Boric acid and boric oxide and probably many other boron compounds are converted in sulfuric acid t o boron tri (hydrogensulfate),
B (HS04)3, which behaves as a strong acid, HB (HSOa)4. The evidence
for the formation of this acid and its salts provides an interesting example of the application of the cryoscopic and conductimetric methods
for determining modes of ionization ($4).
1. Boric Acid and Boric Oxide
It has been known for a long time that boric acid is very soluble in
sulfuric acid, and various compounds of boric acid and boric oxide with
sulfuric acid have been reported. One of these compounds has the composition H3B03 3s03, and may be regarded as boron tri(hydrogensulfate), B (HSO,) (12).Hantzsch (65) suggested that this compound
is formed when boric oxide is dissolved in sulfuric acid. Freezing-point
measurements (g4) showed that Y = 6 for both H3B03 and Bz03. The
formation of B(HSO4)3 would give Y = 7 and Y = 8 for boric acid and
boric oxide respectively
HsBOs
+ 6HpSO4 = B(HSO4)j + 3HsO+ + 3HSO4+
(57)
&Os
9H2so4 = 2B(HSOi)s'+ 3H,O+ + 3HSO4(58)
which is not in agreement with the experimental observations. Conductivity measurements, Fig. 6, showed that for boric acid = 2 and for
Concentration ( M I
FIG.6. Conductivities of solutions of boric acid and boric oxide. Reproduced from
ref. 94.
408
boric oxide y = 1, which is again not in agreement with Eqs. (57) and
(58), which require y = 3 in both cases. If, however, we assume that
boron tri (hydrogensulfate) combines with hydrogen sulfate ion to form
the boron tetra(hydrogensu1fate) ion
B(HSO4)s
+ HSO4-
B(HSO4)r-
(59)
by analogy n-ith
BF3
+ F- = BF4-
and (15)
B(0H)s
+ OH- = B(OH)(-
Eqs. (57) and (58) can be rewritten as follows
+ 6HZSO4 = B(HSO4)a- + 3Hs0+ + 2HS04B20j + 9HZS0, = 2B(HSO4)(- +:3HsOf + HSOa-.
HaBOa
(60)
(61)
It may be seen that now v = 6 and y = 2 for boric acid and v = 6 and
= 1 for boric oxide in agreement with the experimental results.
2. Tetra(hydrogensu1fato)
boric Acid
The conclusion that the boron tetra (hydrogensulfate) ion is formed
in these solutions is of great interest since, as it apparently does not
undergo solvolysis, the corresponding acid HB (HSO4) must be a strong
acid in the sulfuric acid solvent system, and must be extensively ionized
according to Eq. ( 5 5 ) above. The solutions formed by dissolving boric
oxide or boric acid in sulfuric acid contain a mixture of hydroxonium
boron tetra (hydrogensulfate) and hydroxonium hydrogensulfate. Solutions of the free acid can be prepared by removing the hydroxonium ion
by means of the reaction
HsO+
+ SO8 = Hasod+
(62)
by dissolving the boric acid or oxide in oleum instead of sulfuric acid.

Equations (60) and (61) then become
Bb03
+ 3HpSs07 = HsSO,+ + B(HSO&- + H2S04

+ 3H&01 + 4R,SOd = 2HsS01++ 2B(HSOi)r-.
(63)
(64)
Cryoscopic and conductimetric measurements (94) confirm these reactions

and therefore that the acid HB (HS04)4 is very extensively, if not completely, ionized. Such a solution of HB(HSO4)l can be titrated with a
strong base such as KHS04 and the titration can be followed conductimetrically (Fig. 4):
KRSOi
base
+ HB(HS0i)r = KB(HS04)r + H804.

acid
salt
(65)
solvent
The conductivity falls rapidly as KHSO, is added to a solution of HB(HS04)a
409
and the highly conducting H&h+ ion is replaced by the poorly conducting
K+ ion. It passes through a minimum a t the mole ratio nKHS04/niHB(HS04)4 =
0.98, where 7 2 ~ ~ is~ the
0 , number of moles of KHSO4 added and niHB(HS04)r
is t.he initial number of moles of HB(HS04),,and then rises again because of
the HS0,- arising from the excess KHSO,. At the minimum the solution
contains the fully ionized but relatively poorly conducting salt potassium
boron tetra(hydrogensulfate), K+ -B(HS04)4-.In agreement with this interpretation of the conductivity results is the fact that up to the composition
nKHSO./n*HB(HSO,), = 1.0 the freezing point hardly changes (Fig. 7). This
2 .o
1.0
0
K HS
O4Pi
H [B ( H SO,)
41
FIQ.7. Freezing points of HIB(HSOdrl-KHS04 solutions. KEY: u,

experimental curve ; - - -, calculated ideal curve.
is because HaS04+ is being replaced by K+, and B(HS04)4- remains unchanged in the solution; hence the total number of solute particles remains
unchanged. After this composition the freezing point decreases as a consequence of the extra ions due to the excess KHS04 (24).
Attempts to obtain simple salts, for example, KB(HSOr)4,from such
solutions have not so far been successful (41). In order to get a solid to
crystallize out it was found necessary to prepare considerably more
concentrated solutions than those used in the physical measurements,
and under these conditions the solid sodium, potassium, ammonium, and
410
strontium salts that were obtained had a ratio of sulfate to boron of

less than four. They appear t o contain polymeric anions formed from
the boron tetra (hydrogensulfate) anion by the elimination of sulfuric
acid. One of the simplest of such anions that can be envisaged is the
dimeric ion containing bridging sulfate groups
+ LH,SO,.
HSOd
SO,H
HSO/
(66)
SO,H
\.SO,/
There are many niore complex possibilities for anions derived from
compounds intermediate in composition between B (HSOI)3 and Bz(SO4)3.
This is analogous to the behavior of the borate ion, which, in dilute aqueous solution has the formula B(OH)4- (15)although many solid borates
contain complex anions such as B*07=and B,O1I4-.
C. GROUPIV
1. Silicon
A number of organosilicon compounds have been investigated by

several groups of workers (26, 76, 79, 8s).The results of some recent
conductivity measurements (26) are shown in Fig. 8.
a. Tnalkylsilanols, RsSiOH, and Their Derivatives. The experimental values (26, 76, 79) of v = 4.1 and y = 1.0 for hexamethyldisiloxane
indicate that it ionizes according to the equation
[(CHa)sSi]zO 3HzS04 = 2(CH3)&3iSO,H
+ HaO++ HSO4-.
(67)
Price (79j suggested that trimethylsilicon hydrogen sulfate ionizes furt,her to give the trimethylsiliconium ion,
(CHa)&iS04H= (CH&3iC
+ HS04-
(68)
but there is no evidence from the cryoscopic or conductimetric measurements that this occurs to any appreciable extent (26). The results of
measurements on solutions of trimethylethoxysilane (26), triethylethoxysilane (76), and triethylsilanol (76) show that they also react to give
the corresponding trialkylsilicon hydrogensulfates
RaSiOEt
RsSiOH
+ 3HaS04= R3SiSOIH+ EtS0,H + HsO+ + HS04+ 2HzSO4 = RaSiS0,H + HaO+ + HSOt-.
(69)
(70)
b. Dialkylsilanols, R&( O H )I, and Their Derivatives. The experimental values of v = 7.2 and y = 2.0 obtained for dimethyldiethoxysilane
(26) are in agreement with the formation of the corresponding di(hydrogensulf ate)
jCH8)4Si(OEt)24- 6HZSO4= (CHs)2Si(S04H)s 2EtS04H
+ 2H80++ 2HS04-.
(71)
411

/
.I
2 H20J/
2.r
H20,
2.r
,/
r
0
I.!
1.0
0.01
0.02 0.03 0.04 0.05 0.06 0.07 0.00 0.09 0. b

Concentration (mole per kg roln)
FIG.8. Conductivities of solutions of some silicon compounds. KEY:A , 0 Hexamethyldisiloxane,

Trimethylethoxysilane; B, Dimethyldiethoxysilane; C, Methyltriethoxysilane.
The value of i = 12.6 obtained by Price (79) for the compound [ (CH,)2
Si0l4 is also consistent with the formation of dimethylsilicon di(hydrogensulfate)
[(CH8)&0]4
+ 12H2S04= 4(CH8)2Si(S04H)2+ 4HaO++ 4HS04-.
(72)
The rather high values of v and i given by these compounds can be

attributed mainly to nonideality of the solutions as has been discussed
in detail by Flowers et al. (26).
c. Monoalkylsilanols, R S i ( 0 H )J , and Their Derivatives. If methyltriethoxysilane formed methylsilicon tri (hydrogensulfate) values of
= 10 and y = 3 would be expected:
CHaSi(0Et)s
+ 9Hk3O. = CHSi(SO4H)a + 3EtHSOh 4-3&0+ 4-3HSO,-.
The actual experimental values obtained (26) were v = 8.0-7.7 and

= 2.6-2.2 over the concentration range 0.015-0.055 m. It appears therefore that in this case conversion to the corresponding sulfato derivatives
is not complete. Thus if the oxide (S04H)2CH&3iOSiCH3(S04H)2 were
412
formed v = 8.5 and y = 2.5 would be expected

2CH3Si(OEt)s
+ 15HzSO4 = (SO4H)zCHa . SiOSi CHI(SO~H)~

+ 6EtSO4H -+ 5H,O+ + 5HS0,-.
*
(73)
The smaller values of v and y actually obtained indicate that higher

polymers are in fact formed. In general one can write
+ (ti + 3/n)H2S0,
CAsSi(OEt)3
1/n HSO, . CH3

~ i 4 - - - @ ~ > ) - - ~ ~ S O ~ H ]
SO,H
SOIH
n--2
SO4H
+ 3EtHSOI + (2 + l/n)HtO+ + (2 + l/n)HSO,-.

Dimer
Trimer
Tetramer
n
2
3
8.5
8.0
7.75
(74)
3.5
2.3
2.25
The formation of a mixture of these polymers would account satisfactorily for the experimental observations. The formation of cyclic polymers as suggested by Price (79) would give v = 7.5-7.0 and y = 2.0
CH3Si(OEt)3-+ l/n[CHaSi(SO,H)O],
+ 3EtSOdH + 2H30+ + 2HSO4-
(75)
and is not therefore in good agreement with the experimental results.

The values of i between 8.5 and 8.0 obtained (79) for the compound
[ (CHs) sSiO]SSiCH3 are consistent with the formation of t,he dimers and
trimers described above
3(CH&3iSOIH
v=7+3/n
+ (2 + l/n)HtO+ + (2 + l/n)HS04-
(76)
d. Silicic dczd Derivatives. The compounds Si (OEt) 47 PhaSiOH7

PhsSi (OH)?, Si (OAc)1, and (naphthyl)$3iOH appear to be insoluble
in sulfuric acid (26). They probably react to form insoluble polymers.
Some cryoscopic and conductimetric measurements have been made on
solutions of SiPli4 which dissolves to some extent with the formation of
only a relatively small amount of insoluble polymer. The interpretation
of the results is not very certain but they appear to indicate that benzene
sulfonic acid and a three-dimensional oxy (hydrogensulfato) polymer are
formed ( 2 6 ) .
SiPh, -+ 4PhSOaH
+ 1/2n[Si,Ol(HSO.)r],,1 + 2.5Ha0+ + 2.5HSOa-.
(77)
413
THE SULFUBIC ACID SOLVENT SYSTEM
The compound Si[OSi(CH3)sl4 appears to give a similar polymer (79).

Si[OSi(CHa)s],
~(CHS)SS~SO~H
1/2n[Si2O~(HSO4)2I,m
+ 2.5H,0f + 2.5HS01-.
(78)
It may be noted that in solution in sulfuric acid no more than two

hydrogen sulfate groups become attached to a silicon atom. In cases
where a tri- or tetra(hydrogensu1fate) could in principle be formed oxysulfato compounds are in fact obtained.
6. Tin
a. Hexa (hydrogensulfato)stannic Acid. Stannic acid and stannic
sulfate appear to be almost insoluble in sulfuric acid, but a number of
salts of sulfato-stannic acid, HzSn(SO4)3, have been prepared by evaporating mixtures of stannic oxide, a metal sulfate, and sulfuric acid (13,
85, 86), for example, RbzSn(S04)3, Kd3n(S04)3, Ag,Sn(S04) 3 * 3Hz0,
CaSn(S04)3 3Hz0, PbSn(S04)3 3Hz0. All the hydrated salts contain three molecules of water, and they may perhaps be formulated as
salts of tri (hydrogensulfato) stannic acid, HzSn(OH)3(HS04)3, for example, AgzSn(OH)3 (HS04)3. Evidence for the existence of hexa (hydrogensulfato) stannic acid, HzSn(HS04)8, in solution in sulfuric acid has
been obtained from the study of the freezing points and conductivities
of solutions of tetraphenyltin, triphenyl tin hydroxide, and tin tetraacetate in sulfuric acid (41). The results of these measurements are
consistent with the following modes of ionization for these compounds,
and are not in agreement with other modes of ionization that may be
formulated.
+ 14&% = &Sn(HSO& + 4PhSOaH + 4H,O+ + 4HS04PhsSnOH + 13HsSO4 = Hkb(HsO4)~+ 3PhSOsH + 4HaO+ + 4Hso4S ~ ( O A C+) ~lOH&O, = H2Sn(HS04)s+ 4AcOHp+ + 4Hso4P4Sn
(79)
(80)
(81)
The values of Y and y decrease with increasing concentration in each

case and this can be attributed to the equilibrium between the free acid,
its anions, and hydrogen sulfate ion,
H2Sn(HSO&
+ 2HS04- = HSn(HS04)e- 4-HSO4- + &So4

= Sn(HSO&-
+ 2HnSO4.
(82)
In the case of the phenyl compounds, the phenyl groups are rapidly and
quantitatively removed to form benzene sulfonic acid and water, as is
also the case for phenyl boron and silicon compounds.
b. Tetramethyltin. When tetramethyltin is dissolved in sulfuric acid
at room temperature there is a rapid evolution of methane as a result
of the reaction
(CHa)Sn
+ &so4
(CH&Sn SOdH
+ C&.
(83)
414
A similar reaction is known t o occur with the halogen hydracids. When

this reaction was allowed to proceed to completion, cryoscopic and conductimetric measurements (41) on the resulting solution showed that
v = 2 and y = 1, indicating that trimethyltin hydrogen sulfate ionizes as
a strong base according to the equation
(CH&Sn
SOdH = (CHs),Sn+
+ HSOI-.
(84
This reaction can also be written in the form

(CHASn SO4H
1
+ HSOh = (CH&Sn . SO4Hz+ +
(85)
and it is not possible a t present to distinguish between these two possibilities. In any similar cases we shall for simplicity write the reaction
in the first of the forms given above.
c. Trimethyltin Sulfate. This behavior of trimethyltin hydrogensulfate is confirmed by that of trimethyltin sulfate, which was found (41)
to give v = 4 and y = 2, in accord with the ionization
+ 2HS04-.
[(CH,)&I]~SO, = 2(CHa)*Sn+
(86)
Thus trimethyltin hydrogensulfate is a strong base in sulfuric acid. The

corresponding hydroxide is, on the other hand, only a weak base in
water (11).
d. Di-n-Bzityltin Diacetate. Ciyoscopic and conductimetric measurements ( 4 1 ) on di-n-butyltin diacetate indicate that it forms the di(hydrogensulfate) which behaves as a strong base
(n-Bu)zSn(OAc)s 4H2SO, = (n-Bu)zSnSO,H+ 2AcOH*+ 3HS04-. (87)

e. Methylstannonic Acid. Methylstannonic acid dissolves in sul-
furic acid only very slowly even a t 60". However, solutions could be
obtained sufficiently concentrated to enable cryoscopic and conductimetric measurements to be made (41). The results show that methyltin
tri (hydrogensulfate) is formed
CHaSnO(OH)
+ 5HzSO4 = CH$n(HSO& + 2HSO+ + 3HSOI-
(88)
which then ionizes as a weak base.

In contrast to the behavior of silicon all the tin compounds investigated are completely converted t o the corresponding sulfato comfiounds,
which is consistent with the smaller electronegativity of tin. The basicity
of the tin hydrogen sulfates decreases as the number of hydrogen sulfate
groups on the tin atom increases, thus R3SnS04H and R2Sn(HSOO2 are
strong bases, while RSn(HS04)3 is a weak base and S n ( H S 0 4 ) 4is an
acid.
3. Lead [ H e m( hydrogensulfato)plumbic Acid]
Plumbous salts are in general converted to plumbous di (hydrogensulfate) which is a fully ionized strong base (Section 11, B, 1, a ) . Lead
415
dioxide is insoluble in sulfuric acid and the only soluble compound of

Pb(1V) that has been investigated is lead tetraacetate, which, like tin
tetraacetate, is fully converted to the tetra (hydrogensulfate). Since this
behaves as an acid, it may be regarded as hexa(hydrogensu1fato) plumbic
acid, H2Pb(HSO4)6 (40)
Pb(0Ac)r
+ 10HzSO4 = HzPb(HS0a)s + 4AcOHZ+ + msoa-.
(89)
From the variation of the values of v and y with concentration it was

possible to deduce values for the two dissociation constants of this acid,
K1 = 1.1 x lo-* and Kz = 1.8 X
H*Pb(HSOl)s-+ HaSO4+
+ HPb(HSOd)s-+
2H&O4+
+ Pb(HSOi)r*.
(90)
Solutions of hexa (hydrogensulfato) plumbic acid on warming deposit

bright yellow plumbic sulfate P b (SO4)2 which like Sn (S04)2 is only
slightly soluble in sulfuric acid. It seems reasonable to suppose that the
insolubility of these sulfates results from their being polymeric. Their
structures have not been determined. One possibility is illustrated in
Formula (I). Salts of trisulfatoplumbic acid have been prepared ( 1 6 ) ,
(1)
for example, KzPb(S04)3 and these are also only slightly soluble in
sulfuric acid, as are the corresponding sulfatostannates. The structures
of these salts are not known; they could be imagined as containing discrete anions containing three chelated sulfate groups as in Formula (11),
but their insolubility is more consistent with a structure containing a

three-dimensional polymeric anion, such as that represented rather diagrammatically in Formula (111). Salts of the corresponding hydroxy acids
are well-known, for example, K2Sn(OH)6 and NazPb(OH)6. It is note-
416
(111)
worthy that just as sulfatoplumbic acid readily loses sulfuric acid on

warming to deposit insoluble plumbic sulfate, so hydroxyplumbic acid
readily loses water and deposits insoluble PbOz on warming, while both
hydroxy- and sulfatostannic acids appear to be rather more stable.
D. GROUPT1. Nitrogen
Nitric acid, metallic nitrates, and dinitrogen pentoxide are converted

to nitronium hydrogen sulfate, which is a fully ionized strong base (36,
45, 47, 58, 7'5):
HI'iOa
KNOa
NzOs
+ 2HzS04 = NOz+ + Ha0 + 2HS04+ 3HzS01 = K+ + NOz+ + Ha0 + 3HSOd+ 3HzSO4 = 2 x 0 +~ HaO+
~
+ ~HSOI-.
+
(91)
(92)
(93)
Dinitrogen trioxide and nitrites react to give fully ionized nitrosonium

hydrogen sulfate (47, 7'5)
KzOa
P;aNOZ
+ 3HzS04 = 2NO+ + HaO+ + 3HSOp+ 3H2S01 = Na+ + NO+ + HsO+ + 3HS04-
(94)
(95)
and dinitrogen tetroxide gives a mixture of nitronium and nitrosoniuni

hydrogen sulfates (47, 75)
WzO4
+ 3HzSO4 = N02+ + NO+ + HsO' + 3HSOa-.
(96)
There is no doubt that nitronium and nitrosonium hydrogen sulfates are

ionized as bases according to the equations
and not, for example, according to the equation

NO,
HSO.
+ Heso4 = NO2
SOIH~+ HSO4-
(99)
as both the nitronium and nitrosonium ions have been positively identified by means of their Raman spectra (36,58, 75).
417
Ammonia and all the amines that have been investigated behave as
strong bases in sulfuric acid. Even triphenylamine, which is generally
regarded as an exceedingly weak base, incapable of forming salts, has
been shown (50) by means of cryoscopic and conductimetric measurements to be a strong base in sulfuric acid, ionizing completely according
to the equation
PhaN
+ HZSO1 = PhaNH+ + HSO4-.
(100)
2. Phosphorus
Phosphoric acid behaves as a strong base, as is shown, for example,

by cryoscopic and conductimetric measurements (41) on KHzP04. They
show that v = 4 and y = 2, in agreement with ionization according t o
the equation
KHzPO4
+ 2HsSO4 = Kf + HIPOI++ 2HSO4-.
(101)
Like triphenylamine, triphenylphosphine is a strong base (50) being

fully protonated according to the equation
PhrP
+ His04 = P W H + + HSOd-.
(102)
In addition it undergoes either rather slow sulfonation or oxidation, or

both.
3. Arsenic
The pentoxide is insoluble in sulfuric acid a t room temperature, and

dissolves only slowly even at temperatures near the boiling point of the
acid. Arsenious oxide, on the other hand, is moderately, if rather siowly,
soluble in sulfuric acid. Cryoscopic and conductimetric measurements
(41) have shown that it ionizes in an analogous manner to N203,
forming As0 HSOa,
A&Os
+ 3HzS01
2As0 * HSO,
+ HaO+ + HSO4-.
(103)
However, whereas the ionization of NO HSO4 is complete, that of

I s 0 HS04 is only approximately 50% complete in a 0.05 M solution
of A~203,
As0 * HSO. = ABO+ HSO1-.
(104)
Moreover the results indicate that a t higher concentrations some polymeric cation is formed in increasing amounts. It has not so far been
possible to determine the nature of this ion. The simplest polymeric ion
that can be formulated would appear to be As304+
AsO+
+ A&Oi
&d&+.
(105)
Whereas the antimony1 ion SbO+ and the bismuthyl ion BiO+ are well-
418
known, this appears to be the first definite evidence for the arsonyl
ion AsO+.
By dissolving arsenious oxide in oleum, it can be further sulfated and
arsenic tri (hydrogensulfate) is obtained. This appears to behave as a
very weak acid, and therefore it can probably be formulated as tetra
(hydrogensulfato) arsenious acid, HAS(HSO,) 4 ,
If further Asz03 is added to a solution of AsCHSOI), freezing-point and

conductivity measurements indicate that a polymeric species, which is
probably (S04H)2AsOAs(S04H) is formed,
.4%0a
+ 4As(HSOd)S = ~(SOIH)~A~OAS(SO~H)Z.
(108)
Addition of still more Asz03 finally leads to the formation of As0 * HSO,,
A*Og
+ As(HSO4)a = 3As0 . HSOI.
(109)
A number of compounds of AszO3 with sulfur trioxide have been

described, which may be formulated as oxysulfato derivatives with
formulae such as have been given in Fig. 5. Thus, from solutions of
As203 in aqueous sulfuric acid, the'compound As203 * SOB or (As0)&304
has been crystallized ( 1 ) . The same compound has also been obtained
from the flues of pyrites burners (76). From 98% HzS04the compound
As203 2503 has been obtained ( I ) : this may be formulated as
AszO(S04)2. From the reaction of As203 and sulfur trioxide the forma-.
tion of the compounds As203 * 3S03, As203 * 4SO3, Asz03 6S03, and
As203 8S03 has been claimed (I, 72,81).The first may be formulated as
As2(S04)s, and the others may be polysulfates. Several complex sulfates
of As(II1) have been described (66),for example, PbO . Asa03 * 2s03
and CaO -h203
. 3S03. These may all be formulated as derivatives of
oxysulfato acids intermediate between HAsOz and HAs(S04)2. Thus the
first may be written as Pb[AsO(S04)]2, and the second as Ca[As20
(SO,) 3]. The relationships between the various compounds of arsenic
(111) that appear to exist in sulfuric acid solutions is shown diagrammatically below (also see Fig. 5 ) .
&SOI'
4. Axt imon y
+ h(HSO4)4-
HAs(HSO4)d c As(HSO4)g
HAOT
HIO
(HSO&AsOAs(HSQ)r
H&OT
HzO
As0 * HSOI -+ AsO+
Jr
Jr
+ HSOI-
The trioxide is insoluble in sulfuric acid at room temperature, but it

dissolves on strong heating, and on cooling crystals of Sb2(S04)3are
419
deposited (2).From more aqueous sulfuric acid both SbO * S04H and
(SbO)zS04 have been obtained (2). A number of salts of the acid
HSb(S04)2 have been prepared (52, 6 5 ) . They are insoluble in sulfuric
acid a t room temperature. It has been suggested that the salts contain
the Sb(S04)z- anion in which the two sulfate groups are chelated to the
antimony (82).However, their insolubility in sulfuric acid makes it
much more probable that they contain large polymeric anions in which
the sulfate groups bridge from one antimony atom to another.
E. G ~ U VI
P
1. Selenium Dioxide
It has been known for many years that selenium dioxide is soluble
in sulfuric acid to give a bright yellow solution (73). It has been shown
recently (25), by cryoscopic and conductimetric measurements, that in
dilute solutions it behaves as a weak base ( K b = 4.4 X
SeOz
+ HZSO, = HSe02+ + HSOa-
(110)
forming the SeO OH+ ion. It is probable that the un-ionized portion of the selenium dioxide is present as selenious hydrogensulfate,
SeO (OH) (SOrH) . The reaction of selenium dioxide with sulfuric acid
would then be more correctly written
SeOz
+ H&04 = SeO(OH)(HS04) = SeO . OH+ + HS04-.
(111)
Selenious hydrogensulfate (Formula IV) could also be described as selenious-sulfuric acid, by analogy with disulfuric acid (Formula V) and
selenic-sulfuric acid (Formula VI) , although in sulfuric acid it behaves
as a base rather than as an acid.
Small amounts of the ion HSeZO4+ are also present in dilute solutions of SeOz in sulfuric acid, together with the un-ionized diselenious
hydrogensulfate, SezOj (OH) (HS04) (Formula VII) . Relatively greater
\/
XS/
Se
HO
/ \
0
0
!e
/je
NS/
/ \0/ \0/ / \
HO
OH
420
R. J . GILLESPIE AND
E. A. ROBINSON
amounts of this dimeric ion and probably higher polymeric ions are
formed with increasing concentration of selenium dioxide.
When selenium dioxide is added to an oleum the conductivity decreases and passes through a minimum when one mole of SeOz has
been added for each mole of disulfuric acid originally present in the
oleum. The simplest interpretation of this result is that a largely
un-ionized complex Se02 H2S20, is formed. This may be formulated
either as SeO ( H S O r ) or SeO (OH) (HS207). Confirmation of such an
un-ionized species is obtained from freezing-point measurements which
show that a t the composition of minimum conductivity Y = 1.3. There
is also some evidence for the existence of the corresponding derivatives of diselenious acid, that is, Se203(HSO,) 2 or Se203(OH) (HSz0.r).
It is not possible on the basis of available experimental data to distinguish between these alternative formulations, although the first is perhaps the most probable in each case and is also consistent with the
general scheme of oxysulfato derivatives given in Fig. 5. The relationships between the sulfato compounds of selenium(1V) for which evidence has been obtained and the corresponding hydroxides and oxides
are shown below.
Se02
SeoOdOHh
SeO(OH)*
Se20dOH)(SOJU
SeO(0H) (SOaH)
SezOdSOJVp
SeO(S04H),
:?. Tellurium Dioxide
It is, perhaps, a t first sight surprising that tellurium dioxide is insoluble in sulfuric acid, particularly because a compound 2TeO2 So3
(9) has been described, and as Te02 is soluble in other strong acids, for
example, HN03, HC104, and HS03F, from which compounds such as
2Te02 H N 0 3 (64) and 2Te02 * HC104 (18) have been isolated. All
these compounds can be regarded as derivatives of ditellurous acid, for
example, Te203(OH) (NO3), Tes03(OH) (ClO,) , and Tez03(Sod). It
seems possible that Te02 does form a sulfate when treated with sulfuric
acid, but that being highly polymeric it is insoluble in sulfuric acid.
3. Sulfiir Dioxide
Sulfur dioxide is moderately soluble in sulfuric acid (51, 74) and

gives a solution that has a n electrical conductivity which is very slightly
greater than that of the solvent ( 4 1 ) . This is probably caused by a rather
slight ionization according t o the equation
SO1
+ HSOI = HSOz+ + HSOd-.
(112)
The solubility of sulfur dioxide is greater in oleum and increases rapidly
421
with increasing oleum concentration, that is, with increasing acidity of

the solvent (74). This suggests that the basicity of sulfur dioxide is
essentially responsible for its solubility in sulfuric acid and oleum,
presumably because of the formation of a hydrogen-bonded complex
which precedes the actual proton transfer accompanying ionization;
Ha04
+ SO,
HSOaOH
* *
.0
4 4
HSO4-
+ HSOt+.
Evidence for the basicity of the S=O group is also given by the fact
that a number of substituted diphenyl sulfoxides, R2S=0, have been
shown to behave as strong bases and the corresponding sulfones, RzSOz,
as weak bases in solution in sulfuric acid (38);
+ Ha01 RS=OH+ + HSO4&SO, + Ha04 = &SOnH+ + HSOd-.
R&l--O
(113)
(114)
1. Adie, R.H.,J. Chem.Soc.55,157 (1889).

9. Adie, R. H., 1.Chem. Soc. 57; 540 (1890).
3. Barr, J., and Gillespie, R. J., unpublished.
4. Bass, S. J., and Gillespie, R. J., J . Chem. SOC.to be published.
6. Bass, S. J., Gillespie, R. J., and Robinson, E. A., J. Chem. SOC.to be published.
6 . Bergius, F., 2.physik. Chem. 72,338 (1910).
7 . Brand, J. C. D., Horning, ITT. C., and Thornley, M. B., J. Chem. Soc. p. 1374
( 1952).
S. Brand, J. C. D., James, J. C., and Rutherford, A., J . Chem. SOC.p. 2447 (1953).
9. Brauner, B., J . Chem. SOC.55,382 (1889).
10. Brayford, J. R., and Wyatt, P. A. H., J . Chem. SOC.p. 3453 (1955).
11. Bredig, G., 2.physilc. Chem. 13,303 (1894).
12. dArcy, F. R., J. Chem. SOC.55, 155 (1889).
13. Druce, J. G.F., Chem. News 128,33 (1924).
14. Duffy, J. A., and Leisten, J. A., Nature 178, 242 (1956).
15. Edwards, J. O.,Morrison, G. C., Roes, V. F., and Schultz, J. W., J . Am. Chem.
Soc. 77,266 (1955).
16. Elbs, K., and Fischer, F., Z . Hektrochem. 7,243 (1901).
17. Fheant, S., and Chedin, J., Mdm. sew. chim. dtat (Park) 40,292 (1955).
18. Fichter, F., and Schmid, M., 2.anorg. Chem. 98,141 (1916).
19. Finback, C., and Viervoll, H., Tidsskr. Kjemi, Bergvesen Met. 3.36 (1943).
20. Finback, C., Rplnning, O., and Vien-011, H., Tssskr. Kjemi, Bergvesen Met. 4,
26 (1944).
21. Flexser, L. A., and Hammett, L. P., 1. Am. Chem. SOC.60, 885 (1938).
2%. Flexser, L. A., Hammett, L. P., and Dingwall, A. J. Am. Chem. SOC.57, 2103
(1935).
25. Flowers, R. H., and Gillespie, R. J., unpublished.
24. Flowers, R. H., Gillespie, R. J., and Oubridge, J. V., J. Chem. SOC.p. 1925 (1956).
g6. Flowers, R. H., Gillespie, R. J., and Robinson, E. A., J. N U C ~ U&TInorg. Chem.
In preas. (1959).
42!2
1.
Flowers, R. H., Gillespie, R. J., and Robinson, E. A., J. Chem. SOC.,to be pubwed.
8.Franklin, E. C., The Nitrogen System of Compounds, Rheinhold, New York,

1936.
$8. Gillespie, R. J. J . Chem. SOC.p. 2493 (1950).
B.Gillqie, R. J., J. Chem:Soc. p. 2516 (1950).
90. Gillespie, R. J., J. Chem. SOC.p. 2537 (1950).
31. Gilleupie, R. J., J. Chem. SOC.p. 2542 (1950).
38. Gilleapie, R. J., 1. Chem. Soc. p. 1951 (1954).
33. Gillespie, R. J., Discussions Faraday Soc. 24,230 (1957).
34. Gilleapic, R. J., and Cole, R. H., Trans. Faraduy Soc. 52,1325 (1956).
36.Gillespie, R. J., and Leisten, J. A., Quart. Revs. 8,40 (1954).
36. Gillespie, R. J., and Millen, D. J., Quart. Revs. 2, 277 (1948).
S7. Gillespie, R. J., and Oubridge, J. V., J. Chem. Soc. p. 80 (1956).
38. Gillespie, R. J., and Pssserini, R. C., J. Chem. Soc. p. 3850 (1956).
39. Gillespie, R. J., and Robinson, E. A., J . Chem. Soc. p. 4233 (1957).
40. Gillespie, R. J., and Robinson, E. A., Proc. Chem. SOC.145 (1957).
41. Gillespie, R . J., and Robinson, E. A., unpublished.
43. Gillespie, R. J., and Solomons, C., J. Chem. Soc. p. 1796 (1957).
4. Gillespie, R. J., and Wasif, S., J. Chem. Soc. p. 209 (1953).
4. Gillqie, R. J., and W a i f , S., J . Chem. SOC.p. 215 (1953).
46. Gillespie, R. J., and Wasif, S., J . Chem. Soc. p. 221 (1963).
46. Gillespie, R. J., and \Vhite, R. F. M., Tram. Faraday Soc, 54, 1846 (1958).
47. Gillespie, R. J., Graham, J., Hughes, E. D., Ingold, C. K., and Peeling, E. R. A.,
1. Chem. SOC.p. 2504 (1950).
48. Gillespie, R. J., Hughes, E. D., and Ingold, C. K., J. Chem. Soc. p. 2504 (1950).
49. Gilfespie, R. J., Oubridge, J. V., and Solomons, C.,J . Chem. Soc. p. 1804 (1957).
60. Gillespie, R. J., Oubridge, J. V.,and Robinson, E. A., unpublished.
61. Gold, V . , and Hawes, B . W. V., J . Chem. Soc. p. 2102 (1951).
68. Gutmann, A., Arch. Pharm. 236, 477 (1898).
63. Hantzsch, A., 2.physik. Chem. 61,257 (1907).
64. Rantzsch, A,, 2. pkysik. Chem. 62, 626 (1908); 65, 41 (1908); 68, 204 (1909);
Guzz. chim. ital. 39(i), 120 (1908) ; 39(ii), 512 (1908); 41(i), 645 (1911).
66. Hammett, L. P., and Deyrup, A. J., J. Am. Chem. SOC.54,2721 (1932).
66. Hammett, L. P., and Deyrup, A. J., J. Am. Chem. Soc. 56,827 (1933).
67. Hammett, L. P., and Paul, M. A,, J. Am. Chem. Soc. 56,827 (1934).
68. Ingold, C. K., Millen, D. J., and Poole, H. G., J . Chem. Soc, p. 2576 (1950).
69. Kendall, J., and Davidson, A. W., J. Am. Chem. Soc. 43, 979 (1921).
60. Kendall, J., and Landon, M. L., J. Am. Chem. SOC.42,2131 (1920).
61. Kendall, J., Adler, H., and Davidson, A. W., J. Am. Chem. Soc. 43. 1846 (1921).
68. Kirkbride, B. J., and Wyatt, P. A. H., ?inns. Fnrnday SOC.54,483 (1958).
63. Kneitach, R., Ber. 34,4089 (1901).
(7.6. Kothner, D., Ann. 399.1 (1901).
66.Kiihl, H., 2.anorg. Chem. 54,256 (19O7),
66. Kiihl, H., Arch. Pharm. 245,377 (1907).
67. Kunzler, J. E., and Giauque, W. F., J . Am. Chem. SOC.74, 804 (1952).
68. Kumler, J. E., and Giauque, W. F., J. Am. Chem. SOC.74, 3472 (1952).
69. Ihmler, J. E., and Giauque, W. F., J. Am. Chem. Soc. 74, 5271 (1952).
70. Leisten, J. A., J. Chem. SOC.p. 1572 (1956).
71. Liler, M., and Kosanovic, D., 1.Chem. Soc. p. 1084 (1958).
423
72. Mellor, J. W., A Comprehensive Treatise on Inorganic and Theoretical Chemistry, p. 332.Longmans Green, New York, 1929.
73. Meyer, J., and Langner, M., Ber. 60,285 (1927).
74. Miles, F.D., and Carson, T., J. Chem. SOC.1946,786.
76. Millen, D.J., J. Chem. SOC.1950,2600.
76. Newman, M.S., Craig, R. A., and Garrett, 8 .B., J. Am. Chem. SOC.
71,869 (1949).
77. Oddo, G., and Scandola, E., 2. physik. Chem. 62, 243 (1908); 66, 138 (1908);
Guzz. chim. ital. 38(i), 603 (1908); 39(i), 569 (1909); 39(ii), 1 (1908); 40(ii),
163 (1910).
78. Pascard, R., Compt. rend. 240, 2162 (1955).
79. Price, F.,J.Am. Chem. SOC.70,871 (1948).
80. Robinson, R.A., and Stokes, R. H., Electrolyte Solutions, p. 116. Butterworths,
London, 1955.
81. Schultz-Sellac, C., Ber. 4, 109 (1871).
88. Sidgwick, N. V., The Chemical Elements and Their Compounds, p. 798. Oxford
University Press, London and New York, 1950.
83. Szmant, H. H., Devlin, 0. M., and Brost, G . A., J. Am. Chem. SOC.73, 3059
(1951).
84. Treffers, H.P., and Hammett, L. P., J . Am. Chem. SOC.59, 1708 (1937).
86. Weinland, R. F., and Kuhl, H.,
Ber. 39, 2951 (1906).
86. Weinland, R.F., and Kuhl, H., Z. anorg. Chem. 54, 244 (1907).
87. Yost, D.M., and Russell, H., Systematic Inorganic Chemistry, Prentice-Hall,
New York, 1944.
AUTHOR INDEX
Numbers in parentheses are reference numbers and are included to assist in locating references in which the authors names are not mentioned in the text. Numbers
in italics indicate the page on which the reference is listed.
Baldwin, H. W., 6 (101,60
Bagshawe, B., 329 (4), 343
Balfour, A. E., 229 (21,864
Abegg, R., 23 (11, 60
Ball, D.L., 23 (10, 24 (111, 60
Abitz, 245 (44) 266
Ballhausen, C.J., 6 (91),68,95 (1191,106
Adamson, A. W., 36 (2,3),60
(1191,118
Adie, R. H.,418 (11, 419 (2), 421
Bancie-Grillot, M., 337 (51, 34.9
Adler, H.,391 (611,4.98
Banus, M.D., 379 (41,381
Ahrens, L. H., 207 (1001, ,921
Barbaud, J., 325 (28,291, 3.43
Albert, P.,341 (l), 348
Barber, W. A., 59 (4), 65 (4), 109
Altshuller, A. P., 196 (41,197 (31,203(21,
Barker, F.B., 46 (12),60
818
Barr, J., 402 (31,4.91
Amaldi, E., 268 (11, 3i2, 315 (231, 343
Barrett, J., 40 (13), 60
Amis, E.S., 29 (41,40 (291,60
Baruseh, M.R., 59 (121), f l d
Amphlett, C. B., 17 (51,60
Bade, R.,341 (61, 34.9
Anderson, A., 33 (61,34 (61, 35 (61, 60 Baskin, Y., 224 (461,866
Anderson, E. R., 379 (681, 388
Bass, S. J., 388 (4,51, 390 (51, 391 (5),
Anderson, H. L., 321 (2),34.9
402 (51, 4.91
Anderson, J. S.,25 (71,60
Baatiansen, O., 368 (51,381
Anderson, R. C., 279 (85,86,87), 3i4
Bate, G. L.,339 (71,343
Anderson, R. E., 321 (571,344
Bauer, S. H., 132 (4,6,661, 148 (5), 164,
Andrew, E. R., 309 (21,Sf8
166
Appel, R.,364 (341,381
Bauer, W. H., 163
Arimoto, F.S., 83 (31,s (3), 86 (3), 109 Baxendale, J. H., 40 (13, 14, 151, 60
Armstrong, W.D., 38 (81, 60
Beach, J. Y., 132 (61,164
Asher, R. C., 242 (11, 864
Beamer, W.H., 336 (31,343
Asprey, L. B., 39 (1211,153
Bearcroft, D.J., 21 (161, 60
Atchison, G. J., 336 (31,$43
Becker, W.E., 46 (171,60
Aten, A. H. W., 272 (5,61,273 (5,61, 276
Beckert, O.,90 (851, 91 (851,111
(3), 281 (4, 51, 301 (51, 318
Beer, M., 83 (223), 116
Atoji, M., 146 (1, 2,3), 148 (291,164,166
Beinoravichute, Z. A., 83 (1391,f i S
Audrieth, L. F., 348 (31, 349 (841, 354
Beintema, J., 356 (44),38.9
(841, 357 (841, 359 (31, 360 (31,362
Bemski, G., 303 (7), 318
(3), 38f, 388
Benkeser, R. A., 84 (5), 109
Auerbach, F., 23 (11,60
Benson, G. C., I66 (61, 167 (5, 61, 182
Augustyniak, W. M., 303 (7), 312
(71,819
B e m n , R. E., 83 (6),85 (61, 92 (61, 109
B
Bentley, W. C., 327 (971,346
Bergiua, F.,391 (6), 4.9i
Bernal, J. D., 224 (31,864
Bach, H.,352 (101, 363 (101, 381
Berndt, A., 91 (7,81, 109, 110
Baumel, A., 76 (1821,114
Bertaut, E. J., 164 (8,9). 207 (8). 819
Baker, F.B., 39 (91,60
425
426
AUTHOR INDEX
Resson, A , , 350 16), 355 (6), 359 (7),361
(7), 381
Betts, R. H.,
32 (19),39 (lS),60
Bichowsky, F.R.,189 (101,219
Biltz, W.,246 (41,284
Birmingham, J. M.,
60 (217),61 (10,2111,
66 (9, 2121, 68 (211, 2171, 69 (211,
2171, 71 (213), 72 (2171, 74 (9, 211,
216, 217), 75 (216), 76 (211, 2161,
80 (2161, 81 (9, 211, 212, 216), 82
(213), 110, 114, 116
Bleick, W. E.,197 (111, 204 (11), ,919
Bloom, A. L.,134 (7), 155
Bode, H.,350 (81, 352 (101, 355 (81,356
(81,361 (11, 14), 362 (11, 131, 363
(10,.13),364 (14), 365 (121,367 (91,
381
Boehm, H.P., 224 (341,226 (81, 227 (8),

229 (81,230 (81, 966
Boke, K., 78 (11, 12, 13), 110
Bottcher, R.,89 (43, 101-(441,104 (441,
110
Bollnow, 0. F., 167 (121,219
Bondi, A., 348 (151,381
Bonner, N.A., 14 (211,33 (6), 34 (6),35
(61,36 (20a1,43 (201,60
Boone, J. L., 151 (8), 165
Bormann, F.,166 (131,219
Bowen, H.J. M., 335 (8), 337 (91, 3.ds
Born, M.,158 (14, 17, 19, 201, 160 (16),
163 (161,164 (171, 166 (191,171 (201,
172 (21),174 (21),192 (181,210 (15),
219
Boyd, G. E., 273 (211, 281 (9, 211, 298

(9,211,300 (211,312
Brack, A., 61 (1091, 65 (1091, 112
Brady, I. G.W., 6 (a),
60
Bragdon, R. W., 379 (41,381
Bragg, J. K.,148 (9), 149 (91, 166
Brand, J. C. D.,82 (141, 110, 387 (81,
391 (71, 394 (71, 395 (71, 400 (71,
421
Braschos, A.,350 (731,354 (731,355 (731,
358 (731,378 (731,382
Brauer, P., 166 (22), 168 (221, d19
Brauner, B., 420 (91,491
Braunholtz, W.,376 (M),
38.9
Brayford, J. R., 391 (101, 395 (10).
Bredig, G.,414 (111, 491
Breslow, D.S., 97 (15), 110
Bretschneider. 0..
231
266
(n),
Brewer, F. M.,338 (61),344

Brewer, L.,211 (491,220
Bright, W. M., 366 (161,35'1
Briscoc, H.V. A., 25 (71,60
Broadhead, G. D., 84 (171,85 (16, 18),
110
Brockway, L.O.,366 (161,381
Broda, E.,275 (67),281 (671,313
Brodersen, I<., 233 (59),266
Brodie, B. C., 226 (61,248 (51,265
Bronsted, J. N.,28 (231,60
Brooksbank, W.A., 335 (lo), 343
Brost, G.A., 410 (831, 493
Brown, C. A., 379 (481,382
Brown, C. I., 43 (241,60
Brown, F. W., 294 (631,295 (631, 31.9
Brown, H.,
338 (11, 27), 343
Brown, H.C., 146 (101,166
Brown, R.H.,
43 (941,62
Brown, W.L.,284 (101,288 (10,271, 292
(10,271, 293 (27), 302 (10,271, 305
(271,312, 313
Brubaker, C. H., 43 (251,60
Biichner, E.H., 187 (23,24), 919
Biitow, K.,361 (141, 364 (141, 381
Burell, E.J., 280 (84), 314
Burg, A. B.,150 (111, 151 (S), 165, 379
(171,381
Burgus, W. H.,276 (111, 918
C
Cabell, M. J., 319 (12),330 (12),337
343
Cabera, N., 286 (12), 287 (12), 312
Cadenbach, G.,236 (191,241 (191,266
Cain, C.E.,85 (102),112
Cali, P.J., 333 (50), 336 (38,50), 343,344
Calvin, M.,105 (361,110
Campbell, N. L., 256 (871,206
Carlson, T.A.,274 (131,277 (131,312
Carpenter, L. G.,42 (261,43 (26),60
Carroll, L.,119 (12),266
Carson, T.,420 (741, 421 (74), 423
Carter, A. H.,229 (71,264
Casler, R.,288 (141,312
Chalmers, T.A.,268 (78), 314
Chambers, W.J., 132 (551, 133 (551, 166
Chapman, A. C., 364, 372
Chapman, S.,201 (119),221
AUTHOR INDEX
427
Crawford, B. L.,Jr., 119 (181,123 (181,

Chaudron, G.,341 (I), 342
124 (18),125 (631,126 (181, 128 (IS),
Chedin, J., 394 (171,421
129 (181,130 (18), 142 (181,145 (18),
Chiang, R. S. H., 278 (30), 279, (301,
147 (181,151 (181,153 (18), 166, 166
313
Croatto, U., 272 (221,277 (231, 31.9
Chien, J. C. W., 281 (IS),81.2
Croft, R. C., 254 (lla), 255 (10,11, Ila),
Christ,man,D., 279 (161,31.9
258 (llb), 260 (121, 261 ( 1 3 , 266
Clancy, E.F., 318 (961, 366
Clarke, F. P.,294 (171,297 (171,302 (17, Csalan, E., 229 (381, 965
Curby, R. J., 83 (168,1701,113
19), 305 (17,18, 191, 319
Curie, I., 315 (19), 328 (181,343
Clarke, R. P.,148 (131,149 (131,166
Clausen, H., 361 (111,362 (111, 365 (121, Curphey, T.J., 82 (1651, 83 (1651,113
Curran, S. C., 332 (601,344
381
Clauss, A., 226 (81, 227 (81, 229 (81, 230
D
(8), 2G6
Cleary, R.E.,273 (201,276 (20),281 (201,
Daasch, L. W., 350 (641, 354 (641,355
298 (201,319
(641, 356 (641, 366 (24, 251, 367 (24,
Cloason, R. D., 102 (19), 103 (19), 110
641, 368 (241,369 (24, 251, 372 (24,
Cobble, J., 273 (211, 281 (9,211, 298 (9,
25), 375 (24,251,381, 383
21),300 (211,319
Dabagian, H. J., 334 (391,3.43
Coffield, T. H.,102 (191, 103 (191,110
DAgostino, O., 268 (11, 312, 315 (231,
Cohen, C. A., 64 (2051, 114
343
Cohen, D.,39 (56, 121~1,40 (27,28, 29,
DAns, J., 66 (27),68 (271,110
1221, 60, 61, 63
dArcy, F. R.,407 (121, 491
Cole, R.H., 387 (341,4%
Davidson, A. W., 391 (611,396 (591,4.22
Connally, R.E., 325 (141,343
Davidson, N.,10 (331, 43 (24), 60
Connick, R.E.,6 (301,60
Davies, T.H.,283 (go), 314
Cook, G.B., 324 (151,8.43
de Boer, F., 163 (251, 166, (26,271, 170
Coon, J+.H., 338 (161,343
(26, 271, 919
Cooper, H.G.,297 (551,313
Corbridge, D. E. C., 356 (21, 359 (191, de Boer, J. H., 189 (1251, 190 (291, 192
(29, 1231, 195 (1231, 203 (28), 204
366 (391,377 (391,381
(1221,91.9, $21, 290 ( 8 ) , 312
Coryell, C. D., 17 (751, 33 (751,49 (751,
Deck, C. D., 16 (134),63
61, 322 (171,3.43
de Ficquelmont, A. M., 354 (55),355 (29.
Cosgrove, J: F.,326 (631,3.44
55), 357 (27,291, 358 (281,360 (26.
Cotton, F. A., 58 (211,60 (2171,61 (159,
29, 54), 367 (301,381, 38.9
214), 64 (214), 68 (2171,69 (2171,70
(22,2141, 72 (25, 2171, 74 (22, 216, de Kowalewski, D. G., 367 (43),382
217, 2181, 75 (2161,76 (216,2181, 77 Delbecq, C. J., 337 (201,3.43
(22, 25), 80 (2161, 81 (22, 25, 216, de Maine, M. M.,272 (511,275 (501,281
(50,53,541,285 (501,287(50,53,W,
218), 87 (221, 89 (22,1591, 90 (1591,
293 (541, 302 (50, 63, W), 303 (53,
91 (1591,93 (201,95 (231,109 (1171,
541, 304 (50, 541, 305 (50, 53, 541,
ii0,112,ii3,114,1~6
306 (53, 541, 308 (53, 541, 313
Couldridge, W., 360 (20), 361 (20), 363
De Moulds, V., 229 (71,266
(20), 381
Dent, B. M., 166 (811,202 (801, 220
Coulson, C. A., 125 (141, 166
De Puy, C. H., 68 (281, 110
Cowley, J. M., 257 (91,258 (91,966
Craig, D.P.,95 (26), 110, 373 (211, 374 de ROW,A. M., 272 (51,273 (51,281(51,
301 (51,313
(231, 375 (221,378 (231,381
Devlin, 0. M., 410 (83), 493
Craig, R. A., 410 (761,423
de Vries, T. A., 356 (451,366 (45), 367
Craig, R.P., 43 (24),60
(451,372 (45), 382
Cramer, W.H.,
3 (1181,62
428
AUTHOR INDEX
Deyrup, A. J., 386 (55,561, 390 (55, 561,
Emersleben, O., 164 (30, 31, 321, 166 (30,

31, 32), 819
393 (551, 422
Emery, A. R., 164

Endell, J., 231 (711,166
120 (16, 501, 121 (23, 501, 133 (231,
135 (161, 135 (23, 501, 136 (231, 137 Engelmann, F., 74 (341, 110
(501, 139 (151, 142 (151, 143 (15, 16, Epstein, P., 164 (33,341,219
Erber, J., 275 (67), 281 (671, 313
231, 144 (231,145 (231, 150 (231, 166,
Eriks, K., 119 (231, 121 (231, 133 (23),
166
135 (231, 136 (231, 143 (231, 144
Dienes, C. J., 281 (251,294 (631, 295 (63),
Dickerson, R. E., 119 (16, 17, 23, 44, 501,
298 (241, 312, 313

Dingwall, A., 393 (221, 421
Dishon, B., 360 (31, 361, 362 (311, 363
(321, 381
Dodgen, H. W., 16 (311, 50
Dodson, R. W., 10 (331,32 (32, 621, 33
(117), 38 (471, 42 (26, 501, 43 (26,
1011, 44 (591, 60, 61, 62, 268 (571,
269 (581, 278 (771, 279 (771, 281 (56,
771, 308 (771, 313, 314
Doering, W. von E., 68 (281, 110
Doumerc, J., 328 (221, 343
Doyle, J. R., 73 (291, 110
Drever, R. W. P., 332 (60), 344
Druce, J. G. F., 413 (131, 421
Duffy, J. A., 397 (141, @1
Duke, F. R., 9 (341,38 (35),60
Dullenkopf, W., 248 (911, 166
Duncan, J. F., 324 (151, 325 (211, 343
Dunite,,J. D., 58 (30, 311, 77 (30, 31, 321,
78 (31, 321, 80 (321, 95 (321, 110
Dwyer, F. P., 13 (361, 16 (361,61
Dymond, J. A., 337 (91, 343
Dzurus, M., 245 (14), 247 (141, 256 (151,
258 (151, 266
E
Eberhardt, W. H., 119 (181, 123 (181, 124
(181, 126 (181, 128 (181, 129 (181,
130 (181, 142 (181, 145 (181, 147 (181,
151 (181, 153 (18), 166
Edwards, J. O., 408 (151, 410 (151, 9 1
Edwards, L. J., 132 (251, 133 (261, 134
(261, 144 (251, 146 (251, 164, 166
Eguchi, M., 341 (761,361
Eider, E., 16 (371, 61
Eiland, P. F., 58 (331, 77 (331, 110
Eiafeld, K., 107 (1041, 11%
Elbs, K., 415 (161, 421
Ellis, P., 17 (381, 18 (381, 61
Ekhner, B., 105 (351, 110
(231, 145 (231, 150 (231,166
Erschow, A., 193 (135), 221

Evans, J. B., 278 (26), 31.2
Evans, M. G.,7 (391, 61, 197 (35), 198
(351, El9
Evjen, H. M., 163 (361, 166 (361, 919

Ewald, P. P., 158 (371, 164 (371, 167 (371,
219
Ewing, H., 28 (1441, 53

Eyring, H., 8 (841, 10 (841, 40 (841, 61
F
Fankuchen, I., 368 (831,382

Faraggi, H., 328 (18, 221, 34.3
Feder, H., 6 (40),61
Feltham, R. D., 105 (361, 110
Fhneant, S., 394 (171, 421
Fensham, P. F., 253 (16), 265
Fermi, E., 268 (11, 319, 315 (231, 343
Fialkov, Y. A., 352 (331,381
Fichter, F., 420 (181, 4.22
Finback, C.,387 (19, 201, 421
Finch, G. J., 224 (171, 266
Fischer, A. K., 68 (371, 74 (37), 110
Fischer, E. O., 57 (421, 58 (701, 59 (55),
60 (51, 571, 61 (48, 49, 50, 58, 591,
62 (57, 701, 64 (381, 65 (55, 621, 66
(56, 811, 67 (481, 68 (49),69 (51, 73).
70 (50,60,631, 71 (5'7,591, 72 (39, 47,
581, 73 (76, 77, 871, 74 (39, 51, 56,
58, 59, 66, 70, 71, 73, 74, 76, 871, 76
(66, 181), 77 (40, 70, 152, 198, 199,
ZOO), 81 (58, 76, 77, 871, 82 (42, 74,
1321,86 (47,721,87 (51,52,80,86),88
(52, 621, 89 (43, 60, 82, 861, 90 (57,
60, 61, 69, 851, 91 (52, 60, 69, 85),
92 (391, 93 (52, 82, 861, 94 (70, 177),
95 (41), 96 (411, 97 (41, 42, 701,
98 (53, 541, 99 (53, 64, 75), 100 (6%
79, 901, 101 (44, 67, 68, 83, 841, 102
(631, 103 (651, 104 (44, 53, 64, 74,
AUTHOR INDEX
429
Geissler, P. R., 270 (481, 272 (481, 278

79, 83, 91, 2011, 105 (451, 106 (53,
(481, 279 (481, 280 (481, 313
2011, 107 (781, 108 (451, 109 (461,
George, P., 16. (441, 61
110,111, 112, 113, 114
Giacomello, G., 272 (221, 312
Fischer, F., 415 (161, 4.91
Giannini, U., 97 (133, 1341, 112
Fischer, K. W., 107 (1051, 112
Fletcher, R. C.,284 (101, 288 (10, 271, Giaque, W. F., 387 (67, 68,691, 422
Gibb, T. R. P., Jr., 379 (41, 381
292 (10, 271, 293 (271, 302 (10, 271,
Gillespie, R. J., 386 (361,387 (34,37, 44,
305 (271, 312, ,913
46, 491, 388 (4, 5, 291, 389 (26, 28,
Flexser, L. A., 393 (21, 221, 4.91
291, 390 (5, 32, 481, 391 (5, 24, 32,
Flowers, R. H., 389 (261, 391 (241, 393
39, 42, 45, 491, 392 (431, 393 (26, 45),
(261, 395 (231, 396 (231, 402 (241,
394 (451, 395 (23, 37, 42, 45, 48),
404 (261, 405 (261, 407 (241, 408
396 (23, 33, 37, 43, 44, 45), 397
(241, 410 (261, 411 (261, 412 (261,
(351, 398 (30, 35, 36, 481, 399 (361,
419 (251, 4.91, 4.99
400 (31, 39, 41, 421, 401 (421, 402
Ford, M. D., 133 (261, 134 (261, 155
(3, 5, 24, 29, 30, 40, 411, 404 (261.
Ford, P . T., 132 (191, 166
405
(261, 407 (241, 408 (241, 409
Fowler, R. H., 192 (381,219
(411, 410 (261, 411 (261, 412 (261,
Fox, M., 270 (281, 278 (281, 313
413 (411, 414 (411, 415 (401, 416 (36,
Franck, J., 8 (1041, 10 (1041, 62
45, 471, 417 (41, 501, 419 (251, 420
Frank, F. C., 163 (391, 219
(411,
@g
Franklin, E. C.,386 (271, 388 (271, @g
Gilman, H.! 67 (93), 111
Franklin, R. E., 230 (181, 266
Fridenhagen, K., 236 (19, 201, 241 (19, Gilman, H. S. A., 32 (191, 60
Gilpin, J. E., 353 (351, 381
201, 266
Girardot, P. R., 154
Freiser, H., 83 (1661, 84 (1661, 113
Frenzel, A., 229 (381, 248 (221, 251 (211, Gist, L. A., Jr., 67 (931, 111
Glendenin, L. E., 337 (201, 343
255 (211, 266
Godhring, M., 364 (341,381
Friedlander, G., 324 (241, 343
Friedman, L., 80 (88), 111, 150 (341,166, Goggin, D., 84 (51, 109
Gold, V., 420 (511, 4.92
269 (291, 270 (291,913
Goldberg, E., 338 (11, 271, $43
Frisch, K. C., 67 (891, 94 (891, 111
Fritz, H. P., 100 (901, 105 (45, 1271, 108 Goldhaber, S., 278 (301, 279 (301, 313
Goldschmidt, F., 360 (361, 381
(451, 109 (461, 110, 111, 112
Goldschmidt, V. M., 172 (401, 219
Frombling, K., 358 (571, 382
Froitaheim-Kiihlhorn, H., 65 (2311, 116 Goldsmith, M., 252 (231, 266
Golovnya, R. V., 71 (1441, 83 (140, 143,
Fuoss, R. M., 7 (411, 61
1451, 84 (141, 142, 1431, 85 (140, 1421,
Furlani, C., 104 (911, 111
119
Furman, S. C., 38 (421,61
Gordon, B., 279 (861, 314
Gordon, B. M., 37 (451, 61
G
Goubeau, J., 195 (41, 421, 219, 248 (911,
266
Galachow, G., 193 (1351, 291

Gallagher, J. D., 326 (251, 343
Galster, H., 107 (921, 111
Garner, C. S., 17 (1381, 36 (701, 37 (851,
38 (421,45 (46, 1401, 48 (851, 61, 62,
63
Garrett, A. B., 410 (761, 4.93
Gates, H. S., 7 (431, 24 (691, 61
Gaudin, A. M., 327 (261, 343
Gould, E. S., 83 (2041, 114

Graham, J., 416 (471, 422
Graham, P. J., 83 (94, 951, 85 (94, 951,
111,112
Gray, P., 174 (431, 177 (431, 188 (431, 198

(431, 219
Green, J. H., 271 (311, 281 (31, 321, 302

(311, 313
Griffith, J. S., 208 (441, 219
430
AUTHOB INDEX
Hartmann, H., 6 (51, 65), 51

Hartshorne, N. H., 291 (381, 315'
Harrison, G. E., 335 (331,843
Harvey, J. A., 319 (40), 343
Hassel, O.,224 (%), 266'
Hasted, J. B., 3 (521, 61
Hauffe, K., 253 (26), 265
Hauser, C. R.,83 (99, loo), 84 ( l w ) , 85
(101, 102,120), 86 (loo), 118
Hauser, E., 376 (811, 382
Haven, A. C., 83 (31, 85 (31, 86 (31, 109
Havens, W. W.,Jr., 320 (34, 101), 343,
345
Hames, B. W. V., 420 (51), 4.22
Hawthorne, M. F., 136 (241, 155
H
Heal, H. G.,39 (M), 51
Hedberg, K., 368 (5), 3Sl
Hedberg, L., 36s (5), 3S1
Haber, F., 160 (48), 219
Heidt, L. J., 38 (541, 61
Hadzi, D., 228 (24), 265
Heilbronner, E., 364 (591, 3S2
Hafner, K., 65 (2311, 115
Hafner, W., 60 (51). 61 ( 5 0 ) , 69 (511, Hein, F., 99 (1071, 107 (103, 104, 105, 106,
1071, 118
70 (50), 74 (51), 87 (51, 52, 861, 88
(52), 89 (86), 90 (85), 91 (52, 85), Heinlein, L., 252 (41), 265
93 (52, 86), 98 (53, 541, 99 (53), Helm, D., 66 (1671, 67 (167), 81 (lG"7,
92 (1671, 113
104 (531, 106 (531, 111
Helmholtz, L., 174 (93), 189 (93), 190
Haigh, C. P., 327 (31), 3.43
(93), 220
Hall, R. M. S., 272 (331,279 (331,513
Hallam, B. F., 86 (97), 89 (96,981, 91 Henderson, J. W., 297 (55), 313
Hennig, G. R., 245 (141, 247 (14), 248
(98),93 (97), 112
(29), 250 (27),251 (281, 252 (28, 301,
Halperin, J., 41 (491, 42 (491, 51
253 (28), 254 (%), 256 (151,258 (15),
Hnmaguchi, H., 339 (321, 343
259 (291, 262 (27, 27a, 281, 265
Hamill, W. H., 273 (20),276 (20),280
242
(34, 841, 281 (20), 298 (20), 318, HBrold, A., 236 (31, 321, 237
(31), 251 (311,254 (311,266
513, 31;
Hamilton, W. C., 125 (21), 134 (211, 144 Herr, W. P., 273 (39), 281 (39), 315
Herwig, W., 99 (227, 228), 100 ( n 7 ) , 107
(21), 165
(227, 2281, 116
Hammett, L. P., 386 (55, 56, 57, &390
I),
(55,56),393 (21,22,55), 4.21,428,423 Herzfeld, K. F., 214 (461,219
Herzog, L. F.,330 (351, 543
Hammond, J. A., 152, 153
ffntzsch, A., 386 (53, 541, 390 (53: 541, Herzog, S., 105 (35), 110
Hess, M., 360 (611,382
391 (531,407 (531, 422
Harbottle, G., 42 (50), 61, 270 (37), 271 Hildebrand, J. M., 355 (38),S S l
(311, 272 (66),273 (35, 36, 41), 275 Hillman, M., 151 (22), 165
(35,371, 276 (35,66), 278 (371, 281 Hilsdorf, G. J., 270 (481, 272 (48), 278
(481, 279 (481, 280 (481, 31s
(31,35,41,661, 298 (35,66),300 (351,
Hindman, J. C., 38 (551, 39 (56,
302 (31,66), 307 (41),315
40 (27,28,29,122), 60, 61,53
Hardy, H. R., 40 (14, 16),60
Hirshberg, Y., 363 (32), 3S1
Hare, P. E., 211 (491,920
Hirshfeld, F. L.,119 (23, 44), 121 (23),
Harker, D., 136 (301,166
133 (231, 135 (231, 136 (231, 143 (231,
Harnwell, C. P., 2 (1201, 62
144 (231, 145 (23), 150 (23), 166, 166
Hartley, S. B., 367
Griffon, H., 335 (28,291, 343

Grillot, E., 337 (5, 301, 343
Grimm, H. G., 204 (451, 214 (461, $19
Gross, B., 166 (47), 819
Grossman, J. J., 45 (461, 61
Grubert, H., 61 (48, 49),67 (481, 68 (491,
72 (47), 86 (47), 110
Grundmann, C., 360 (37), 3SI
Gryder, J. W., 38 (47), 61, 268 (57), 513
Gurnee, E. F., 3 (481, 4 (48), 61
Guter, G. A., 136 (20), 145 (20), 156
Gutmmn, A, 419 (521, 428
Gyarfas, E. C., 13 (36), 16 (36),51
431
AUTHOR: INDEX
Hobbs, E., 366 (391,377 (39),3Sl

Hodes, I., 198 (501, 920
Hofmann, A. W., 361 (401,362 (40), 389
Hofmann, H. P., 59 (551, 65 (551, 66
(561, 74 (561,111
Hofmann, U., 224 (341, 226 (8,33, 371,
227 (8, 33, 35, 361, 229 (8, 33, 381,
230 (81, 231 (711,248 (22,601, 250
(60), 966, 966
Hogness, T. R., 2 (1201,69
Hohenschutz, H., 364 (341,3S1
Hojendahl, I<., 163 (521, 166 (521, 169
(521,201 (521,202 (521,990
Holden, J. R., 350 (641, 354 (641, 355
(641, 356 (641,367 (641, 382
Holmes, A.,68 (1901,114
Holmes, 0.G., 6 (911,66, 208 (971,621,
273 (461,313
Holmes, W. S., 353, 354
Holmstedt, B.,367 (41),3S2
Holst, R.,227 (351, 966
Holstein, T.,
2 (571,3 (571, 61
Hoppe, R.,166 (511, 168 (51), 170 (511,
220
Hornbeck, J. A.,2 (581,61
Horne, R. A.,44 (59), 61
Hornig, E. O., 279 (401,313
Hornig, H. C.,14 (611, 38 (601,61
Homing, w. C.,391 (71,394 (71,395 (71,
400 (71,421
Hontman, E. L., 330 (371,343
Hough, W. V., 132 (251, 133 (261, 134
(26), 144 (251, 146 (25), 166
Howell, P.A., 119 (16,171, 120 (161, 135
(161,166
Hrostowski, H. J., 119 (271, 129 (281,166
Huang, K.,158 (17), 164 (171,919
Hudgens, J. E., 334 (391, 336 (381, 3.43
Hudis, J., 32 (621,33 (631,61
Hiickel, E., 373 (421,889
Hiickel, W.,60 (1081, 119
Huggins, M. L., 166 (541,172 (53,941,174
(541,189 (531, 190 (531, 192 (541,990
Hughes, D. J., 319 (401, 343
Hughes, E. D., 390 (481, 395 (481, 398
(481,416 (471,499
Hughes, E. W., 164
Huizenga, J. R., 39 @1), 69,339 (71,343
Hummel, R. W., 330 (421,338 (411, 3@
Hund, F.,164 (551,166 (55,561,168(551,
220
Hunt, J. P., 5 (641,36 (20a), 60, 61

HurB, J., 341 (61,3.43
Hush, N.S., 208 (571,213 (571,220
Hutchinson, G. W.,326 (431, 344
Hylleraas, E. A., 166 (681, 167 (581, 181
(591,189 (591,990
I
Ibers, J. A.,257 (9), 258 (91, 966
Ilse, F. E.,6 (651,61
Ingham, A. E., 174 (821, 175 (821, $90
Ingold, C. K., 390 (481, 394 (581, 395
(481,398 (481,416 (47, 581, &$
Iredale, P., 326 (44),3&
Iribame, J. V., 367 (431,389
Irsa, A. P., 80 (881, 111, 150 (341,166
Irvine, D.H.,16 (441,61
Irvine, J., 17 (751, 33 (751, 49 (751, 61
Issleib, K., 61 (1091,85 (1091,119
J
Jach, J., 273 (411,281 (411,307 (411,313
Jacobs, P..W. M., 169 (601,980
Jacobson, R., 148 (291,166
Jaeger, F.M.,358 (441, 3%
Jaffe, H. H., 95 (1101,Il2
Jakovlev, J. V., 336 (451,344
James, J. A., 333 (471,336 (46,47), 344
James, J. C., 387 (81, &1
Jenkins, E. N., 339 (481,342 (49),344
Jervis, R.E., 337 (531,344
Jira, R.,60 (571,61 (58, 591, 62 (571,70
(60, 1111, 71 (57, 591, 72 (58), 73
(871,74 (58, 59, 871, 81 (58, 871, 89
(601,90 (57,60,61),91 (601,111,112
Jijnck, P., 253 (42,431,965
Johnson, R. E.,46 (171,60
Joliot, F.,315 (191, s@
Jonassen, H. B., 73 (291, 110
Jones, R.,135 (611,136 (611,137 (61,62),
166
Jones, R. E., 165 (611,$90

Jones, W.H., 45 (1401, 63
Juza, R.,195 (621,196 (62, 631,$20, 252
(41), 253 (39, 40, 42, 43), 254 (391,
666
K
Kahn, M., 46 (12,861,60
Kant, A., 333 (501, 336 (50)) 344
432
AUTHOB INDEX
(641, &,9
Kaplan, L., 58 (1121,80 (1121,81 (1121, Kothner, D.,
Kohn, A.,328 (221,343
119
Kapustinskii, A. F., 158 (671, 172 (661, Kohn, E.J., 350 (64),354 (641,366(641,
356 (041,367(641,%,9
175 (641, 178 (64, 651, 179 (681,
189 (64,65, 66), 190 (661, 192 (66, Kojima, M.,185 (1321,W1
Kollinsky, F.,358 (581, S8,9
69), 195 (65,66),2C3 (881, 980
Kornfeld, H., 192 (181,919
Kasper, J. S., 136 (30),146 (31),166
Kate, J. J., 58 (112),80 (1121,81 (112), Kosanovic, D.,400 (711,@8
Koski, W.S., 138 (351,148 (481,160(33,
11.9
341, 166, 166, !274 (13),277 (13,421,
Kauer, E., 78 (1131,119
818,313
Kaufman, J. J., 138 (351,150 (341,166
Kmarnovskaja, L. I., 1Sg (701, 169 (701, Kostromb, N. A., 352 (331,381
Kraua, C. A., 7 (711,61
188 (70,1241,920,221
Kazarnovskii, I. A., 198 (1241, 990, Hf Krause, H. J., 350 (461, 3S4 (401, 365
(461, 366 (461, 369 (43,
,%#
Kazarnovskii, J., 189 (721,990
Krause, J. T., 0 (221,60
Kealy, T.J., 57 (1141,58 (1141,119
Krishnan, K. S., 163 (751,820
Keenan, T.R., 37 (66).61
KritAaya, I. I., 84 (1361,112
Keilin, B., 148 (65),166
Kiihl, H.,
413 (85,86),418 (661,419 (851,
Kelly, J., Jr., 134 (32),165
4@, 4f3
Kendall, J., 163 (73), 166 (73).220, 391
Iiiilkens, H , 350 (74),352 (74),355 (741,
(611,396 (59,60),&??:
358 (741,389
Kennedy, J. W., 276 (11). 312. 324 (241,
Kunzler, J. E., 387 (67,68,691,
4%
343
Kurras, E.,107 (106),118
Kern, D. M. H., 39 (67).51
Kursanov, D.N., 84 (1151,112
Kerrigan, J., 148 (29), 155
Kester, W. I,., 5S (112),80 (112), 81 Kureen, F.,146 (701,166
Kuzrnenko, A. A., 352 (331,331
(112),I18
Ketelaar, J. A. A., 356 (45),366 (45), 367
(451,372(45), 382
L
Khuranov, S. S., 84 (146),113
Kierstead, H.A., 32 (681, 61
Ladd, M. F. C., 173 (76),189 (76), 190
King, C. B., 38 (116),68
(761,207 (761,990
King. E.L., 7 (43),23 (11, lB),
24 (11,
69,128). 27 (1281,35 (1071,36 (1071, Laidler, D.S.,224 (451,966
Lande, A.,158 (19,201, 166 (19,
171
50, 61, 62, 53
(201,219, B O
King, W. R.,Jr., 36 (70),61
Landel, A., 256 (611,258 (611,866
Kirkbridge, B. J., 389 (622,4.22
Landon, M. L., 396 (801, 429
Kleinberg, J., 322 (51,521, 344
I\llernrn, W.,195 (42), ,919, 245 (44),2G6 Landshoff, R., 181 (78,791, 990
Lang, H.J., 326 (251,348
Klixbull-Jorgensen, C.,
6 (91),62
Lange, W.,350 (47),3 9
Iirnetko, E.A., 262 (44a),966
Langer, J., 350 (75), 354 (75),355 (75),
Iineitsch, R.,391 (631, 429
359 (751,378 (751,382
Kobayashi, S., 182 (741,990
Koch, G. K., 272 (51,273 (51,281 (51, Langner, M.,419 (731,49.9
Larsson, L., 367 (411,388
301 (5),319
Latimer, W. M.,5 (731, 9 (721,61
Kochetkova, N.S., 83 (1350, lla!
Laubengayer, A. W.,379 (481,$82
Kodarna, G., 164
Iiogler, H. P., 65 (62),70 (63),88 (621, Laudenklos, H.,146 (691,166
99 (W, 100 (641,102 (03), 103 (65) Laurence, G.S., 33 (74), 61
Lauterbur, P.C., 138 (351,166
104 (a),
111
Laves, F., 224 (461, 866
Konig, E., 227 (361,S 6
n),
AUTHOR INDEX
Lavine, L., 135 (361, 166

Leary, J. K., 356
Leboeuf, M. B., 325 (141,343
Leddicote, G. W., 335 (101, 343
Lednicer, D., 85 (1021, 112
Lee, W. H.. 173 (761, 189 (761, 190 (761,
207 (761, 220
Legnfeld, F., 379 (491, 382
Leigh, K., 32 (191, 60
Leipfinger, H., 74 (66, 1161, 75 (1161, 76
(66, 1161, 111, 112
Leisten, J. A., 397 (14, 35, 701, 398 (351,
@I, 4%
Lemal, D., 90 (1601, 113
LennardJones, J., 166 (811, 174 (821,
175 (821,202 (801, 220
Leto, J. R., 109 (1171, 112
LBvBque, P., 341 (61,343
Levey, S., 270 (431, 278 (431, 279 (431,
313
Levi, H., 316 (361,343
Levy, R. B., 207 (951, 221
Levy, S., 69 (1611, 113
Lewis, J. E., 318 (721,344
Lewis, W. B., 17 (751, 33 (751, 49 (751,
61
433
41), 125 (411, 126 (18, 411, 128 (18,

391, 129 (18, 39, 411, 130 (181, 131
(391, r33 (231, 135 (16, 23, 36, 37,
501, 136 (23, 40, 411, 137 (40, 41, 501,
139 (151, 142 (15, 181, 143 (15, 16,
231, 144 (23, 42, 431, 145 (18, 231,
146 (1, 2, 3, 43, 711, 147 (W, 148
(291, 150 (231, 151 (181, 153 (181,
164, 166, 166
Lipson, H., 224 (47), 266
Livingstone, R., 28 (231, 60
Lloyd, D., 68 (1251, 112
Lohman, F. H., 359 (561, 3%
London, F., 174 (831,990
Longuet-Higgins, H. C., 97 (1261, 112,
124 (45), 145 (471, 147 (461, 152 (461,
166
Lotmar, W., 358 (531,362 (531, 372 (531,
382
Loveridge, B. A., 319 (851,344
Lowdin, P. O., 181 (84, 85, 861, 220
Lucht, C. M., 136 (301,166
Lubbe, H., 252 (411, 266
Luttke, W., 105 (1271, 112
Lumry, R. W., 10 (112), 32 (1121, 62
Lund, L. G., 355
Lundquist, S. O., 182 (871, 189 (87), 220
Lunn, E. G., 2 (1201, 62
Libby, W. F., 8 (761, 10 (761, 14 (61, 771,

23 (761, 38 (601, 61, 268 (451, 269
(291, 270 (28, 29, 681, 272 (44, 70).
278 (281, 279 (681, 313, 314
M
Liehr, A. D., 92 (1181, 93 (1181, 95 (119),
106 (1191,110,112
McCallum, K. J., 273 (46, 471, 276 (471,
Lien, A. P., 365 (501, 383
281 (741, 301 (74), 313, 314
Lienau, G., 361 (141,364 (141, 381
McCarty, L. V., 146 (311, 148 (9), 149
Liler, M., 400 (711, 422
(91,165
Lindsay, J. K., 83 (99,1001, 84 (loo), 85
McCauley, C. E., 270 (48, 721, 272 (481,
(95, 101, 102, 1201, 86 (1001, 11.9
278 (48, 721, 279 (481, 280 (481, 31.9,
Lindsey, R. V., Jr., 83 (6, 94, 951, 85 (6,
31 4
941, 92 (61, 109, 111, 11.2
McCauley, D. A., 365 (501, 382
Lindstrom, E. G., 59 (1211, 112
McClelland, J. D., 252 (301, 266
Linhard, M., 7 (781, 61
McClure, D. S., 6 (91), 62, 208 (971, 221
Linnenbom, V. J., 32 (791, 61
McConneU, H. M., 13 (801, 37 (801, 62
Linnett, J. W., 97 (1221, 112
Lippert, E. L., Jr., 119 (231, 121 (231, 133 McCoy, R. E., 148 (51, 164
(231, 135 (231, 136 (231, 143 (231, McDonnell, F. R. M., 234 (481, 245 (481,
247 (481, 248 (481, 252 (481, 266
144 (231, 145 (231, 150 (231, 164, 166
Lippincott, E. R.,58 (123, 1241, 80 (123, McGeachin, H. M. D., 355
Machlup, S., 284 (101, 288 (101, 292 (101,
1241, 81 (123, 1241, 112
302 (101, 318
Lipscomb, W. N., 119 (16, 17, 18, 23, 37,
44, 501, 120 (16, 501, 121 (23, 37, 38, McKibben, J. L., 326 (251, 343
SO), 122 (411, 123 (18, 411, 124 (18, Mackintosh, W. D., 337 (531, 344
434
AUTHOR INDEX
Macoll, A., 95 (261, 110

Maddock, A. G.,271 (311, 272 (22, 51,
52), 273 (36, 471, 275 (501, 276 (471,
277 (231, 278 (521, 279 (521, 281 (31,
32, 50, 52, 54), 285 (49, 501, 287 (50,
54), 293 (541, 302 (31, 50, 52, 541,
303 (541, 304 (49, 50, 541, 305 (50,
54), 306 (54), 308 (M),
312, 313
Aladelung, E., 158 (88), 163 (881, 166 (881,
220
Magat, M., 367 (301, 381
Magee, J. I,., 3 (481, 4 (481, 51
Magnusson, E. A., 375 (221,3861
Magnusson, L. B., 39 ( 8 0 , 62
Mahlman, H. A., 335 (lo), 343
Maltbie, MI.
McC., 192 (961, 221
Mangold, D. J., 151 (22), 156
Mantica, E., 97 (1341, 112
Mapper, D., 326 (86, 89), 329 (861, 332
(86), 336 (871, 337 (86,88), 345
Marcus, R. A., 9 (821, 10 (82), 19 (821,
40 (831, 62
Marcus, R J., 8
10 (841, 28 (1441,
40 (84), 52> 53
Margenau, H., 175 (89,901, 220
Mark, H., 224 (25), 265
Mark Wood., Sister hl. John, 21 (691,
51
Marsh, R. E., 91 (71, 109

Marr, J. Ti-., 297 (55), 313
Mastalerz, P., 376 (771,382
Maybury, P. C., 148 (481, 166
Mayer, J. E., 172 (21, 94), 174 (21, 93),
175 (911, 189 (931, 190 (93), 192 (961,
207 (92, 95), 819, 220,291
Mayo, D. TV., 84 (1281, 112
hlayr, G , 328 (541, 344
Mlarzanti, G., 97 (133, 134), 112
Mrier, D. J., 37 (851, 48 (85),59
Memke, IT. W., 321 (57), 322 (55, 561,
344
Meister, H., 61 (1291, 66 (1291, 112

Mellish, C E., 331 (591, 332 (581, 344
Mellor, J. TY.,418 (721, 423
Meinick, A, 46 (871, 62
Meyer, E. G., 46 (86, 871, 52
Meyer, J., 419 (73), 4-23
Meyer, I<. H., 358 (51, 52, 531, 362 (531,
372 (53). 552
Mic*hacl,J. P., 43 (251, 60
Milburn, R., 27 (881, 35 (891, 52
Miles, F. D., 420 (741, 421 (741, 423

Miilen, D. J., 386 (36),394 (58, 751, 398
(361, 399 (36), 416 (36, 58, 75), 42%
4933
Miller, H. C., 132 (551, 133 (55), 166
Miller, J. J., 136 (241, 166
Miller, J. M., 268 (571,269 (561,281 (561,
313
Miller, S. A., 57 (130), 112

Mills, 0.S., 89 (981, 91 (981, 112
Milman, M., 280 (581, 313
Mitchell, J. H., 2 (901, 62
Moffit, W., 6 (911, 52, 95 (131), 112, 128
(49), 166
Moljk, A., 332 (60), 344
Moore, E. B., Jr., 119 (501, 120 (50), 121
(501, 135 (50), 137 (50), 156
Morris, D. F. C., 192 (98), 207 (1001, 208
(981, 211 (981, 221, 338 (GI), 344
Morrison, G. C., 408 (15), 410 (15), 421
Morrison, G. H., 326 (63),336 (621, 344
Mott, N. F., 286 (12, 59), 287 (12, 59),
312, 313
Motz, I<. L., 83 (169, 171), 113, 114
Moureu, H., 354 (551, 355 (55), 360 (541,
382
Mrozowski, S., 250 (49), 262 (501, 2G6
Mijller, H., 99 (1071, 107 (1071, 112
Rluetterties, E. I,., 132 (55), 133 (55),
166
Mulay, L. N., 82 (1321, 112

Rfurmann, R. I<., 21 (92), 27 (921, 28
(921, 33 (921, 34 (92), 62
Muschlitz, E. E., Jr., 3 (931, 52
Myers, H., 20 (130), 21 (1291, 23 (1291,
24 (1291, 31 (1291, 34 (129), 53
Myers, R. J., 119 (27), 129 (281, 155
N
Sakahara, A,, 58 (2251, 115
Sancollas, G. H., 7 (391, 51
Sarath, A,, 359 (561, 33.2
Sath, A., 281 (601,313
S a t t a , G., 97 (133, 134), 112
Kelson, R. D., 58 (123,1241, 80 (123,124),
81 (123, 1241, 118
Nesmeynnov, A. N., 71 (1441, 83 (135,
137, 138, 139, 140, 143, 145, 147, 148),
84 (136, 141, 142, 143, 146), 85 (140,
142), 86 (1471, 112, 115
AUTHC)R INDEX
Nesmeyanov, 0. A., 83 (143, 1471, 84

(1431, 86 (1471, 113
Neumann, H. M., 43 (941, 62
Newburg, N. R., 97 (151, 110
Newkirk, A. E., 146 (311, 166
Newman, M. S., 410 (761, 423
Newton, T. W.,
38 (951, 39 (961, 62
Nicholas, L., 71 (194), 84 (1941, 114
Nikitina, T. V.,71 (1441, 113
Nilsson, G., 278 (611,313
Nordman, C. E., 134 (511, 166
Norman, J. H.,151 (221, 166
Norment, H. G.,133 (521, 166
Norton, F. J., 148 (91, 149 (91, 166
Novak, A.,228 (241, 266
Nyholm, R. S., 95 (261,110
0
Ochs, L., 367 (301, 381
Oddo, G., 390 (771, 423
Odeblad, E., 339 (651, 340 (64, 661, 844
Odeblad, S.,340 (661,344
dfele, K., 101 (67, 681,111
Ogard, A. E., 22 (981, 24 (981, 28 (981, 31
(971, 48 (971, 49 (981, 60, 62
R. A., Jr., 125 (531, 132 (53, 54),
134 (32, 531, 166, 166
Ohlerich, G.,226 (371, 2
6
'
Onsager, L.,107 (2291, 116
Opp, K., 196 (631,220
Orgel, L. E., 6 (911, 28 (991, 62, 58 (30,
311, 77 (30, 31, 321, 78 (31, 321, 80
(321, 95 (26, 32), 97 (1261, 110, 112,
208 (44, 1011, 219, 221
Orleman, E. F., 39 (671, 61
Osgerby, J. M., 85 (181, 110
Osmond, R. G.,336 (671, 844
Otlet, R. L.,331 (591, 344
Oubridge, J. V., 387 (37,491, 391 (24,491,
395 (371, 396 (371,402 (241, 407 (241,
408 (241, 417 (50), 4.91,&2
Overhauser, A. W., 297 (621,298 (621, 303
(621, 305 (621, 313
Owen, J., 6 (911, 18 (1001, 62
Owen, R. B., 325 (68),344
Ogg,
P
Paddock, N. L., 374 (231, 378 (231, 381
Page, F. M., 195 (1021, 221
Page, J. A., 58 (1491, 76 (1491, 113
435
Paine, D. H., 350, 355

Palm, C., 90 (691, 91 (691, 111
Pankow, G. W.,358 (531, 362 (531, 372
(531,582
Pannell, J. H., 327 (a),

3.43
Parchen, F. R., 38 (351, 60
Parkins, W. E., 294 (631, 295 (631, 3f3
Parnes, Z. N., 84 (115), If9
Parry, R. W.,
164
Parshall, G.W.,83 (94, 951, 85 (94, 951,
111,112
Parsons, T. D., 148 (291,166
Pertington, J. R., 158 (1031, 9.91
Pascard, R., 387 (781, @3
Passerini, R. C., 421 (381,4.99
Patat, F., 358 (57, 581, 38.9
Pate, B. D., 333 (911, 335 (90,911, 336
(911, 346
Paul, M. A., 386 (571,422

Pauling, L.,78 (179), 114, 171 (1051, 175
(1041, 187 (107),.9.91
Pauson, P. L., 57 (114, 15Oa1, 58 (1141, 60

(2171, 68 (2171, 69 (2171, 72 (2171,
73 (1511, 74 (151, 217, 2181, 76 (151,
217, 2181, 81 (151, 2181, 84 (17, 1501,
85 (181, 86 (971, 89 (96, 981, 91 (981,
93 (971, 110, 119, 113, 116
Payne, J. A., 331 (59), 332 (581, 344
Pease, R. N., 148 (13, 741, 149 (13, 741,
166, 166
Peeling, E. R. A., 416 (471, &2
Penna-Franca, E., 43 (101),69
Penneman, R. A., 39 (1211, 63
Penney, W. G., 6 (1021, 6.9
Pepinsky, R., 58 (331, 77 (331, 110
Peraldo, M., 97 (1341,118
Perevnlova, E. G., 71 (1441, 83 (139, 140,
143, 145, 1471, 84 (141, 142, 143, 1461,
85 (140, 1421, 86 (1471, 113
Peterson, M. L., 83 (94, 951, 85 (94, 951,
111, 112
Pfab, W., 58 (701, 62 (701, 74 (70),77 (70,
1521, 94 (701, 97 (701, 111, 113
Phillips, W .D:, 132 (551, 133 (551, 166
Philpott, W., 363

Picciotto, E., 328 (69), 3 4
Pierson, D. H., 325 (70, 711, $44
Piesbergen, U., 74 (711, 211
Pimentel, G. C., 129 (281, 166
Pink, R.C.,234 (481,245 (481, 247 (481,
248 (481, 252 (481, 966
436
AUTHOB INDEX
Raymond, W.H.A., 335 (331, 3 4

Redvanly, C. S., 279 @7), 314
Reed, G.W.,339 (32,731, 3 4 , 3-44
Reid, A. F., 340 (741,SG
Reiffel, L.,334 (981,346
Renaud, P., 356 (631, 358 (62, 631, 382
Reynolds, L. T., 74 (1641, 81 (1641, 83
(164),93 (201,110,113
Reynolds, W. L., 10 (1121,32 11121, 6.8
Rice, C. N., 32 (133,63
Rice, R. G.,350 (641, 354 (641, 355 (841,
356 (641,367 (641, 382
Rich, A,, 58 (311,77 (311,78 (311, 110
Rich, R.L., 20 (1301,37 (1101,62, 63
Richards, D. H., 333 (471,336 (46, 471,
344
Richards, J. H.,
82 (1651,83 (1651,113
Richards, R. E., 132 (19),165
Richardson, J., 18 (1111,69
(921, 52, 53
Richmond, 11. H.,
83 11661, 84 (1661, 113
Post, B., 368 (831, 38.9
Ridler, K.E.W.,2 (901,52
Postmiis, C., 3 (107), 36 (1071,62
Potratz, H.A., 14 (211,50, 339 (71, 343 Rieder, W.,275 (671,281 (671, 319
Riemschneider, R., 66 (167),67 (1671,81
Prsstwood, R. J., 42 (1081,5.9
(1671,92 (1671,113
Price, F.,410 (791, 411 (791, 413 (791,
Riesebos, P., 272 (61,273 (61,318
@3
Riley, H. L., 226 (52,53), 229 (2,71,264,
Price, W. C., 131 (561, 132 (56), 166
266, 866
Primak, W., 237 (511,Z66, 294 (641, 305
Rinehart,
K. L.,83 (168, 169, 170, 1711,
1651,313
113, 114
Pringsheim, P., 288 (14),3 . 9
Pritchard, H.,158 (1081, 189 (1081, ,921 Roberts, J. K.,82 (1851,114
Roberts, M. de V., 145 (47), 166
Proctor, J. E ,353 (601,382
Pryce, M. N. L.,208 (57),213 (571,$90 Robinson, E. A., 388 (51,389 (261,390
(51, 391 (5, 391, 393 (26),400 (39,
411, 402 (5, 40, 411, 404 (261, 405
Q
(261,409 (411,410 (261,411 (26),412
(261, 413 (411,414(411,415 (401,417
Quartermann, L., 237 (511,666
(41,501, 419 (251,420 (411,@I, 4.@'
Quimby, 0.T., 359 (561,38.9
Robinson, R. A., 392 (801, 393 (801, @3
Robinson, R. M., 229 (2),264
R
Rochow, E. G.,82 (1321, 112, 348 (a),
353 (651,369 (651,38g
Rabideau, S. W., 59 (109), 68
Romer, G.,348 (71), 362 (711,382
Rita, R., 360 (37,61), 381,38.9
Rplnning, O.,387 (201,421
Raistrick, B.,366 (391,377 (391,381
Rona, E.,44 (1131,68,339 (751,344
&mires, F.,69 (1611,113
Rosenberg, H.,83 (163, 1951, 84 (1621,
Rasetti, F.,268 (11, 312
113, 114
Rausch, M.,83 (163,1951,84 (128, 1621,
Rosenblum, M., 58 (219), 62 (2191, 74
86 (72),111,1lt,113, 114
(2191,81 (2191, 82 ( 2 W , 83 (2241,
Rauscher, H., 272 (661,276(661,281 (661,
85 (1721,94 (2181, 114, 116
298 (661, 302 (661,313
Ross, V. F.. 408 (15). 410 (151,4.91
Ray,J., 134 (321,166
Rosset, G.,359 (71,364 (7), 381
Ray, J. D.,132 (541,166
Pino, P.,97 (133,134), 11.9

Pinson, W.H.,
330 (351,343
Piper, T. S., 61 (1591,64 (2151,65 (2151,
68 (I57),81 (2151,88 (154,155, 1571,
89 (154,155, 156, 1591, 90 (153,156,
159, 1601, 91 (154,156, 158, 1591,83
(153),94 (215),113,116
Pitzer, K., 5 (731,61
Plane, R. A.,13 (1031, 35 (103), 62
Plaas, R., 226 (81, 227 (81, 229 (81, 230
(81, 266
Plattner, P. A., 364 (591,382
Platzman, R., 8 (1041,10 (104), 61
Plumb, R C.,318 (721,344
Pontecorvo, B.,268 (11, SIC
Poole, H.G.,394 (581, 416 (581, 4@
Posey, F. A.,5 (731,7 (1311,21 (92),27
(921,28 (921,29 (1061, 33 (921,34
AUTHOR INDEX
437
Schirmer, F. B., Jr., 349 (841, 354 (.&I),

357 (841,389
Schlafer, H. F., 6 (511,61
Schlapp, R., 6 (1021,69
Schleede, A.,236 (791,237 (791,966
Schlesinger, H.I., 146 (731,148 (29),150
(111,166, 166, 379 (691,389
Schlogl, K.,86 (1781,114
Schmaeling, V.,166 (1121,168 (1121,.??.91
Schmeckenbecher, A., 253 (39,42, 431,
254 (391,966
Schmid, M.,420 (181,491
Schmidt, P.,253 (40,421,966
Schmitz-Dumont, O.,350 (73, 741, 352
(741,354 (731,355 (73,741, 358 (72,
73,741,378 (731,389
Schomaker, V.,78 (1791,91 (81,110, 114
Schoolery, J. N.,129 (641, 134 (7,641,135
(61,641, 136 (61,641, 137 (61,621,
166, 166
Schreiner, S.,69 (731,74 (73,741,82 (741,
104 (741,111
Schrodt, A. G.,272 (701,314
SchrolI, G.,84 (51,109
Schuhl, C.,341 (61,34.9
Schuler, R. H., 270 (48,721,271 (711,272
(481, 278 (48,721,279 (481,280 (481,
813,314
Schultz, D. R., 164
S
Schults, J. W.,
408 (151, 410 (IS), 4.91
SchultzSellac, C., 418 (811,49.9
Sagane, R., 341 (761,344
Sakamoto, Y., 164 (1111, 166 (54, 109, Schuh, H.,255 (651, 259 (651,866
1111, 168 (1111, 174 (541, 192 (541, Schulze, E.,236 (66, 671, 237 (671,239
(671,240 (671,242(801, 243 (66,721,
220,281
245 (671,966
Salmon, L., 327 (781,330 (921, 333 (n),
334 (921,336 (871,337 (921,3.44, 3415 Schwab, G. M., 76 (180, 181, 182), 114
Schwarz, H.A., 280 (34,841, 313, 314
Sandel, V., 102 (191,103 (191,110
Schwen, R., 60 (1oS1,lf9
Scandola, E.,390 (771,493
Scott, R. L., 355 (381,381
Scarrot, G.G.,326 (431,344
Schaeffer, G.W., 136 (201,145 (20), 166, Searle, H.T., 351, 369
Sebera, D.,21 (161,60
379 (68,691,388
Schaeffer, R., 119 (171,135 (58, 61), 136 Sebera, D. K., 26 (114a),69
(58,59,61),137 (58,61,621,149 (601, Seeholzer, 99 (751,111
Seel, F.,350 (751,354 (751,355 (751,359
150 (801, 166, 166, 379 (691,388
(751,378 (751,38.9
Schsfheutl,248 (781,966
Segrd, E.,268 (11,319
SchiiperkGtter, H.,362
Schenk, R.,348 (711,361 (701,362 (711, Seibold, E. A., 78 (1831,92 (1831,114
Seitz, F.,158 (1131,189 (1131,190 (1131,
388
991, 289 (731,314
Scherer, J. R., 125 (631, 166
114
Schilovtseva, L. S., 84 (1421,85 (1421, Semenow, D. A., 82 (la),
&us, D.,73 (76, 77, 871, 74 (76,871,75
113
Rosset, H., 363 (661, 389

Rossini, F. D., 189 (101,919
Roth, W.L., 377 (671,388
Rowland, F.S.,270 (681,279 (68,69,88),
313, 314
Roy, S. K., 163 (751,990
Rubisch, O.,243 (54,721, 244 (511, 248
(541,966
Ruch, E., 76 (1751,94 (173,174, 175, 176,
177), 109 (1741,114
RiidorlT, G.,231 (62,63,711,232 (621,233
(641,235 (631, 966
Riidorff, W., 226 (571,231 (62,63,711,232
(621,233 (59,641,234 (581, 235 (631,
236 (66,671, 237 (671,239 (671,240
(57,671, 243 (66,759, 245 (671,246
(571,248 (58, 60, 681, 249 (681, 250
(55,601,251 (561, 254 (731,255 (651,
256 (61,701, 258 (611,259 (651, 261
(691,966
Ruess, G.,226 (751,227 (741,228 (741,
229 (741,231 (711,252 (751,966
Ruff, O.,231 (76,771,866
Rundle, R. E., 147 (571,166
Russell, H.,
387 (871,4.83
Rutenberg, A. C., 5 (1141,62
Rutherford, A.,387 (81,&l
438
AUTHOB INDEX
(1851,81 (76, 77, 871, 96 (1851, 107

(781,I l l , 114
Seyfang, A P ,328 (79,801,334 (80).3&!
Seyfertlt, D., 61 (lo),110
Shankar, J., 281 (60),313
Shapiro, I., 148 (651, 1.56
Sharman, L. J., 281 (741, 301 (741,314
Sharpe, A. G., 350 (761,3S2
Shaw, P.D.,
84 (1281,112
Shaw, P.F. D., 280 (58),313
Shepp, A.,148 (5), 154
Sheppard, J. C., 14 (115),16 (1151,52
Sherman, J., 158 (1141,166 (114,1151,168
(114),178 (115). 189 (114),190 (1141,
192 (114), 197 (ll4), 205 (114),208
(114),207 (ll4),208 (1141,211 (1141,
213 (114), 221
Sherrill, M. S., 38 (1161,52
Shigeta, J., 341 (761,344
Shoemaker, B.H., 365 (501,3S.2
Shoemaker, D.P , 91 (220,221. n2), 93
(2211,115
Shore, S. G., 154

Sidgwick, N. V., 118 (65a),166, 419 (821,
@3
Siecke, W. F., 248 (68),249 (681, 266

Silbiger, G., 132 (66),156
Sib, V.,254 (73)' 266
Silverman, J.. 33 (1171,52
Simonetta, M.,
95 (1861, 114
Simons, J. H., 3 (93,1181, 52
Simukovn, N. A , 71 (144),83 (145),113
Singer, L.,38 (8), 50
Skrowaczewska, Z., 376 (771,3SS
Slansky, C.S.,5 (731,51
Slater, J. C., 3 ( 119), 52
Smales, A. A., 318 (82),319 (12,%), 324
(81), 326 (84,86, 891, 328 (801, 329
(861,330(12.42,92,93,941,332 (861,
333 (91), 334 (80.921, 335 (90,911,
336 (67,83, 87, 911, 337 (12,86, 88,
921, 338 (811,342 (491,343, 366, 346
Smith, D. C.,366 (251,369 (251,372 (251,
375 (25),3Sl
Smith, D. R.,359, 364, 369
Smith, M.E.,38 (541,51
Smyth, H.D., 2 (1201,52
Sneeden, W.,
82 (14), 110
Sneeaum, J. S.,68 (1251,112
sogo, P.,105 (361, 110
Sokol, P.E.,83 (1701,I13
Soldate, A. &I.,
131 (671,156
Solomons, C.,387 (49), 391 (42,49), 395
(a),
400 (421,401 <42), 42.2
Sowden, E. M., 335 (951,345
Specker, H.,
357 (78,791, 3S2
Spooner, R.C.,38 (1161,62
Spoor, N. L.,25 (7),60
Stahl, H.O.,87 (86),89 (861,90 (851,91
(851,93 (Se),lOO (79,1871.101 (187),
104 (79),iii,114
Starovskii, 0. V., 83 (1451, 113
Staudenmaier, L.,226 (81),266
Staudinger, H.,376 (80, 811, 382
Steger, E.,362 (821, 382
Stehle, P. F., 146 (lo), 155
Stehn, J. R., 318 (961, 3-46
Steinfink, H., 368 (B),382
Stcinman, R., 348 (3), 349 ( W ,354 ( 8 4 ,
357 (841, 359 (31, 360 (3), 362 (31,
3S1, %2
Stephanou, S. E., 39 (1211,53

Stewart, D.C., 327 (971,346
Stieglitz, J., 379 (491,362
Stitch, S.R.,335 (951,345
Stone, C. A., 334 (981,345
Stock, A., 118 (68),146 (68, 69, 701, 158
Stokes, A. R., 224 (471,266
Stokes, H.N.,348 (85,891,349 (891,354
(W),
355 (891,356(891,357 (891,359
(85,86,87, 88,90, 91), 360 (85, 861,
363 (a),
364 (85),393
Stokes, R. H., 392 (80),393 (801, 493
Stone, F. S., 284 (751,289 (751,314
Stumpp, E.,261 (69),2GG
Suck, H.,
230 (201,241 (20),2G5
Sue, P.,273 (761,314, 341 (1, 991, 34%
345
Sugarman, N., 322 (17),S i 3

Sullivan, J. C., 39 (56,121a), 40 (27,28,
29, 1221, 60, 51, 53
Summers, L.,68 (188, 189, 190),114
Sung Moon, 83 (171),114
Sutin, N.,270 (37),272 (33,521,275 (371,
278 (37, 52, 771,279 (33,52, 77), 281
(52,771,302 (521,308 (771,31% 314
SutlitTe, L. H., 38 (123), 40 (151,60, 53
Sutterlin, W., 146 (701,156
Sutton, I. E.,78 (1831, 92 (183),114
Sutton, L.E.R.,95 (26),110
Sworykin, A., 242 (821,266'
Symons, M. C . R., 15 (124),63
439
AUTHOR INDEX
Ssekely, G., 336 (1001,346

Szilard, L.,268 (781,314
Szmant, H.H.,
410 (831,423
Szymanski, H.,305 (651,313
T
Tammann, G.,242 (821,266
Tassin, W., 356 (921,383
Taube, H.,5 (64,73, 1141, 6 (10,401, 7
(1311,13 (1031,16 (311,20 (1301,21
(16, 92, 125, 1291, 22 (98, 1431, 23
(128,1291,24 (98,128,129),26 (114a,
1251,27 (88,92,125,1281,28 (92,98,
1271, 29 (1061,31 (97, 126, 1291, 33
(921,34 (92,1291,35 (1031,37 (1101,
40 (1271,41 (491,42 (491,48 (971,49
(98,1431, 60, 61, 62, 63
TaiigbGl, K.,281 (541,287 (541,293 (541,
302 (54),303 (54),304 (541,305(541,
306 (541,308 (541,319
Taylor, A., 224 (451,2G6
Taylor, M. C., 359 (931,383
Taylor, R.C., 164
Taylor, T.I.,320 (1011,346
Tebboth, J. A.,57 (1301,112
Temple, R.B., 233 (861, ,966
Templeton, D.H.,165 (611, 166 (1181,
168 (1181, 170 (1181, 179 (1161, 181
(1181, 2.20, 9.21
Thamer, R., 362 (131,363 (131,381
Thiele, H.,227 (851,248 (83), 255 (841,
Tremaine, J. F.,57 (1301,11.2

Trifan, D.S.,71 (1941,84 (1941,114
Tromans, F. R.,354,355,356,368
Tmchida, R.,68 (225),116
Tmtsui, M.,107 (nS1,116
Turkevich, A., 338 (1021, 339 (32, 731,
843,844,
Turton, C. N.,279 (69,88),314
U
Uloth, R. H., 68 (189,190),114
Ubbelohde, A. R.,234 (481,246 (481, 247
(481, 248 (481, 252 (481, 966
Unsold, A., 171 (1201,221
Unzhakow, G. M., 195 (121),
Uri, N.,197 (35),198 (351,919
Urry, G.,146 (731,148 (291, 166, 166
v
Van Alten, L.,32 (1321,63
Van Arkel, A. E., 192 (1231, 195 (1231,
204 (1231,991
van Berkum, J. B. M., 281 (41, 318
Vand, V., 294 (791,814
van der Straaten, H., 272 (61,273 (61, $19
van der Wyk, A. J. A., 358 (521,889
Van Liempt, J. A. M., 203 (281,919
Van Santen, J. H.,166 (26, 271, 170 (26,
271, 919
van Styvendael, M., 328 (691, 344

Van Vleck, J. H., 0 (1331,63
Van Zeggeran, F., 166 (61, 167 (61, dl9
266
Thiele, J., 60 (1911, 61 (1921, 64 (191, Vedeneev, A. V., 198 (1241, 891
Venkateswarlu, K. S., 281 (60),313
1921,65 (1921,114
Verwey, E. J. W., 166 (26,271,189(1251,
Thomas, F. G.,325 (211,343
190 (291, 192 (291, 919, 9.21
Thomas, J. C.,88 (1931,114
Vickery, R. C., 256 (871,96G
Thomas, J. G.N.,39 (531,61
Thompson, H.D.,333 (501,336 (50),344 Viervoll, H., 387 (19,20), 421
Vigoureux, S.,87 (801,111
Thompson, H.W.,233 (861, 966
Thornley, M. B.,391 (71,394 (71,395 (71, Vilchevskaya, V. D.,83 (1481, 113
Vineyard, G.H.,
281 (25),318
400 (71,4.21
Vogel, M.,83 (163, 1951, 84 (1621, 113,
Tierney, P. A., 146 (101, 166
Ting Li Chu, 379 (941, 383
114
Vogt, F.,226 (751,252 (751,966
Tomkins, A., 338 (1021,346
Voitlander, J., 76 (180, 181, 1821, 11.4
Topping, J., 201 (1191,$21
Toy, A. D.F., 348 (31,359 (3),360 (31, Volkenav, N. A., 83 (137, 138, 1481, 113
von Brauchitsch, M.,66 (271,68(271,110
362 (31,381
von Hevesy, G.,316 (361,343
Treffers, H. P., 386 (841,&3
von Kriieger, G., 350 (471,38.2
Trefonns, I,., 119 (441,146 (711,166
440
AUTHOR INDEX
Vorres, K.S., 36 (31 60

Voeburgh, W.C., 35 (891,51
W
Waddington, T. C., 166 (126, 127, 1281,
169 (126,lZ,128), 174 (43,1271, 176
(126), 177 (43, 1271, 188 (43, 126),
198 (431, 199 (1271, 200 (I%),
201
(1261,119, 221
Wagner, R. I., 379 (171,381
Rahl, A. C., 14 (1151,16 (37,115, 1341,
32 (791, 33 (63),37 (45), 42 (108),
Whipple, R. O.,72 (251,77 (251,81 (25),

110
White, R.F. M., 387 (461,422

Whiting, M. C., 58 (219), 62 (2191, 74
(2191,81 (2191,82 (2191, 83 (2241,
94 (2191,115
Whitman, G. M., 83 (94,95),8.5 (94,951,
111,11,9
Wiberley, S. E., 151

Wiese, H. K., 64 (205), 114
Rfilkins, R.G., 17 (381, 18 (381,51
Wilkinson, D.H., 326 (104),346
Wilkinson, G.,58 (21, 149,219), 60 (207,
217), 61 (10, 159, 208, 211, 214), 62
(2191,64 (214,2151, 65 (215),66 (9,
51, 52, 53
212), 68 (37,157, 211, 2171, 69 (209,
Wallace, P. R.,250 (88),866
211, 2171, 70 (22, 2141, 71 (207,
Wang, J. H., 275 (80),314
209, 2131,72 (25,206, 2171, 73 (1511,
Ward, R. L., 13 (135),45 (1351,49 (1351,
74 (9, 22, 37, 151, 164, 207, 211,
63
216, 217, 218, 2191, 75 (2161,76 (149,
Rartik, T., 146 (72,73), 156
151,211,216,218),77 (22,25),80(88,
Wasif, S.,387 (441,391 (451,392 (431,
2161, 81 (9, 22, 25, 151, 164,2% 207,
393 (49,394 (451,395 (451,396 (43,
211, 212, 215, 216, 218, 219), 82 (213,
44,45),416 (451,422
219), 83 (164), 87 (22), 88 (154,155,
Wadey, A. T., 148 (741, 149 (74), 156
157, 208), 89 (22,154, 155, 156, 159),
Rattenburg, A., 321 (1031,346
90 (153, 156, 159, 1601, 91 (154,156,
Waugh, 3.S.,109 (117),111
158, 159), 93 (1531,94 (215,2191, 95
Weaver, H.E., 13 (80),37 (801,52
(23), 109 (2101, 110, 111, 113, 114,
Weaver, J. R., 154
116
Weber, J. R., 38 (123),53
Willard, J. E., 270 (43,821, 278 (26,30,
Weber, S.,364 (591,382
43), 279 (30, 40, 43), 280 (82). 281
Webster, R.K., 330 (93,941,346
(151,313, 313, 314
\Veinland, R. F., 413 (85,861, 493
Weinmayr, V., 83 (2021, 84 (202, 2031, Williams, M. J. G.,17 (381, 18 (38),51
Williams, R. R., 273 (20), 276 (201,280
114
(34,841, 281 (20, 83), 298 (20, 83)
Weisendanger, H. U.D., 45 (14O),53
31.2, 313, 314
Weiss, E.,77 (198,199,200),82
104
Wilman, H., 224 (171,165
(196,197,201), 106 (2011,114
Weiss, J., 8 (1371,10 (1371,32 (1361,53 Wilson, A.,355,356,366
93 (221).
Weissman, S. I., 13 (135), 17 (138), 45 Wilson, F.C.,91 (220,221,222),
116
(135),49 (1351,53
Wilson, S. A., 242 (11,264
Welch, A. J. E., 309 (81),$14
Wilson, U.,351
Weliky, X., 83 (204), 114
Wellmann, M.,236 (79), 237 (79), 166 Wirzmiiller, A,, 66 (81),89 (821,93 (82),
101 (831, 104 (83),111
Wettz, J. E., 6 (1391,53
Wesselink, G. A., 272 (5), 273 (51, 281 Rolf, A. P., 279 (85,86,87), 314
Wolf, F.,3 (1411,53
(5).301 (51,31.2
West, C. D., 193 (129), 195 (129), 991 Wolf, L.,83 (2231, 115
Wolfgang. R., 279 (691,279 (88),314
Wexler, S.,281 (91,298 (91,$13
Wheatley, P. J., 119 (16, 171, 120 (16), Wood, A. J., 326 (86,891, 329 (86),332
(86),336 (87),337 (86,88),345
135 (16),146 (3),164, 155
AUTHOR INDEX
Woodward, R. B., 58 (2191,62 (2191,74 Yuster, P., 288 (141,Jld

(2191, 81 (2191, 82 (2191,83 (2241, Yuster, P. H., 337 (20), 343
94 (2191,116
Wright, K. A., 288 (271, 292 (271, 293
(271,302 (271, 305 (271, 313
Wyatt, P. A. H., 389 (621,391 (lo), 395
(101,4-21, g-2
Wylie, G., 182 (71,-219
YY
Yamada, S., 58 (2251,116
Yamashita, J., 182 (130,131), 185 (1321,
g-21
Yamaeaki, M., 125 (751,166

Yankwich, P. E., 270 (891,314
Yatsimirskii, K.B., 179 (68,133,1341,189
(133, 1341, 192 (691, 195 (1331, 196
(1331,197 (1331,199 (1341,2(10 (1331,
201 (1331,203 (681,B O , 2-21
Yosim, S., 283 (901, 314
Yost, D.M.,387 (871,4.98
441
SUBJECT INDEX
A
Ammonia, phosphonitrilic halides and,
Absolute enthalpies, lattice energies and,

206
Acceptor molecule reactions, phosphonitrilic halides and, 364-385
Acetylides, lattice energies of, 198
Acid-base reactions, sulfuric acid solutions and, 403-405
Acid formation, simple conjugate, sulfuric acid solutions and, 397
Acids, sulfuric acid solutions and, 400403
Activated complex, bridged, redox reaction mechanisms and, 19-32
outer-sphere, redox reaction mechan i s m ~a d , 12-19
Activation analysis, charged particles
and, 334341
confiicting nuclear processes and, 331334
experimental,
activating collision measurements
and, 327-328
activity measurements and, 324-326
gamma spectrometry and, 326-327
precision and accuracy and, 328-331
radiochemical purification and, 322-
Ammoniates, boron hydrides, 154

Anhydride formation sulfuric acid solutions and, 399
Annealing,
radiation,
interpretation of, 298-309
thermal,
interpretation of, 283-298
Annealing functions, mixed kinetics and,
360362
323
sample preparation and, 321
source preparation and, 323-324
irradiation and, 316-319
neutron,
applications of, 334339
radiation sources and, 319-321
Activation energy,
defect number,
post-recoil annealing effects and, 297298
processes distributed,
postrecoil annealing effects and, 294297
293-294
Aryl derivatives, phosphonitrilic halides

and, 383
Azides, lattice energies of, 198-199
B
B.HbO, compounds, topology of, 152-153
Balance equations, boron hydride and,
138
Bifluorides, lattice energies of, 199

Billiard-ball model, see Elastic-collision
model.
Bimolecular recombination, diffusion-controlled, annealing and, 292
Bonding,
nonionic contribution, lattice-energies
and, 206-214
Born-Haber cycle, account of, 158-169
Borohydrides, lattice energies of, 197
Borofluorides, lattice energies of, 203
Boron chloride BSCL, structure of, 151152
Boron halides, 146-148

Boron hydride, ammoniates of, 154
balance equations and, 138
nuclear resonance studies,
AI(BHi)s and, 132
BH'- md, 131-132
BtHe and, 134
BsH, addition compounds and, 133-
Activity measurements, activation analysis and, 324-326

Alkyl derivatives, phosphonitrilic halides
and, 303
Amides, lattice energies of, 196
442
134
BJL- and, 132-133

Barn and, 134
Bdas and, 134-135
SUBJECT INDEX
B;H, and, 135

BsHlo and, 135-136
BmHm- and, 136
BloH14and, 136
BzHaBr and, 138
&oH&r and, 137
B J L I and, 138
BioHuL and, 136-137
BloHmI(mp, 116') and, 137
b10Hd (mp, 72") and, 137
poasible ions of, 144-146
reactions among, 148-151
structures of, 118-121
topological theory of, 138-144
valence theory,
diborane and, 124-126
electron deficiency and, 121-122
higher hydrides and, 126-128
molecular orbitals and, 128-131
three-center bond and, 122-124
Boron nitride compounds, graphite and,
261-262
Brush-heap model, see Random-fragmentation model
C
Caesium graphite, 236-248
Carbonates, lattice energies of, 202-203
Carbon monofluoride, homopolar bonding and, 230-233
Cations, hydration of, 5
Chalcogenides, lattice energies of, 192,
193
Charge distribution,
effects, phosphonitrilic halides and, 375377
Charged particles, activation analysis

and, 339-341
Chemical properties, phosphonitrilic
halides and, 357-365
Collision measurements, activation analysis and, 327-328
Conductimetric measurements, sulfuric
acid and, 391-393
Correlated pairs, post-recoil annealing
effects and, 283-290
Cryoscopic meaaurements, sulfuric acid
and, 390-391
Cyanates, lattice energies of, 199-200
443
Cyclic polymers, phosphonitrilic halide

preparation and, 349-351
Cyclopentadiene,
alkali metal salts,
interaction with metal halides and,
60-61
interaction with metal carbonyls and,

61
interaction with metal halides and,

61
interaction with metals and, 59

magnesium bromide,
interaction with metal salts and, 5960
thallium,
preparative methods and, 61-62
Cyclopentadienyl metal complexes,
cyclopentadienyl metal carbonyls and,
86-90
cyclopentadienyl metal nitrosyls and,

90-91
physicochemical investigations of, 91

technical applications of, 97
Cyclopentadienyl metal compounds,
bonding of, 93-97
chemical properties,
indenyl compounds and, 73,75
presently known compounds and, 6373
molecular structure of, 92-93
physicochemical, crystallographic proyertia,
magnetic and, 74, 75-76
mass spectrographic and, 79-80
polarographic and, 76-77
spectroscopic and, 80,81
thermodynamic and, 80-82
X-ray and, 77-79
preparative methods,
cyclopentadiene alkali metal saltsmetal halides interaction and, 6061
cyclopentadiene-metal
interaction
and, 69
cyclopentadiene-metal carbonyls interaction and, 61
cyclopentadiene-metal halides interaction and, 61
cyclopentadienyl magnesium bromide-metal salts interaction and,
69-80
444
SUBJECT INDEX
cyclopentadienyl thallium and, 6162

ferrocene and, 61
Elimination reaction, phosphonitrilic

Epithermal-reaction
model, SzilardChalmers reaction and, -269-270
Esterification, phosphonitrilic halides
and, 359-360
Dialkylsiianols, sulfuric acid and, 410-411

Diazocyclopentadiene, 68
Dicyclopentadienyl iron, see also FerroF
cene
five-membered ring systems, complex
Fast neutrons, radiation annealing and,
formation by, 57-58
300301
Dicyclopentadienyl metal compounds,
Ferrocene, 71, 82-83
see Cyclopentadienyl metal comderivatives of, 84-86
pounds
preparative methods for, 61
Dispersion energy (London) term UL,
substitution reactions and, 83-84
lattice energies and, 174-175
Donor solvents, phosphonitrilic halides
G
and, 362-383
Drift velocity, redox reaction mechaGamma rays, radiation annealing and,
nisms and. 2-3
299-300
E
Elastic-collision model, Sailard-Chalmers
reaction and, 268-2439
Elastic-inelastic collision model, SzilardChalmers reaction and, 269
Electrolytes,
sulfuric acid solutions,
acids and, 400-403
acid-base reactions and, 403-405
anhydride formation and, 399
metal hydrogen sulfates and, 395-397
simple conjugate acid formation and,
397
sulfato derivative conversion and,

398-399
weak bases and, 400
Electron affinity, determination, lattice
energies and, 203-204
Electron transfer,
correlated pairs,
post-recoil annealing effects and, 287288
phase,
redox reaction mechanisms and, 2-4
solution,
redox reaction mechanisms and, 8-12
Electrostatic interactions,
correlated pairs, post-recoil annealing
effects and, 285-287
gas
Gamma spectrometry, activation analysis

and, 326-327
Geometrical effects, phosphonitrilic halides and, 377-378
Graphite,
crystal structure of, 224-226
homopolar bonding,
carbon monofluoride and, 230-233
graphite oxide and, 226-230
tetracarbon monofluoride and, 233236
intercalation compounds,
comparative survey of, 263-264
mode of reaction of, 224-226
polar bonding,
boron nitride compounds and, 261262
caesium graphite and, 236-248

graphite salts and, 248-251
halogen-graphite compounds and.
251-254
metal chloride intercalation and, 254259
metal oxide intercalation and, 260262
metal sulfide intercalation and. 260262
potassium graphite and, 236-248

rubidium graphite and, 236-248
residue compounds of, 262-263
445
SUBJECT INDEX
Graphite oxide, homopolar bonding and,

226-230
Graphite salts, 248-251
H
Halides, lattice energies of, 189, 190
Halogen-graphite compounds, 251-254
High polymer, phosphonitrilic halides
and, 372
Homopolar bonding,
graphite,
carbon monofluoride and, 230-233
graphite oxide and, 226-230
tetracarbon monofluoride and, 233236
Hot zone,
Szilard-Chalmers reaction, recrystallization and, 290-292
Hot-zone model, Szilard-Chalmers reaction and, 270-271
Hydration enthalpies, lattice energies
and, 18&188
Hydrides, lattice energies of, 189, 191, 192
Hydrogensulfato compounds, sulfuric
acid and, 405407
Hydrolysis, phosphonitrilic halides and,
359
Hydrosulfides, lattice energies of, 195-196

Hydroxides, lattice energies of, 192, 194,
195
Hypothetical compound stability,

determination, lattice energies and,
214-217
I
Imides, lattice energies of, 196
Indenyl compounds, eighth-group elements and, 73,75
Intercalation,
metal chlorides,
graphite and, 254-259
metal oxides,
graphite and, 260.262
metal sulfides,
graphite and, 260-262
Intercalation compounds,
graphite, comparative survey of, 263204
Ionic crystals,
lattice energies,
Kapustinskiis formula and, 177-179
Templetons calculations and, 179-181
nuclear recoil and, 272-278
Irradiation,
activation analysis, 318319, sources of,
319-321
K
Kapustinskiis formula, lattice energies
and, 177-179
Kinetics,
mixed, annealing functions and, 293294
1
Lattice energies,
alkali metal salt values,
amides and, 196
azides and, 198-199
bifluorides and, 199
borofluorides and, 203
borohydrides and, 197
chalcogenides and, 192, 193
cyanates and, 199-200
cyanides and, 196-197
halides and, 189, 190
hydrides and, 189, 191, 192
hydrosulfides and, 195-196
hydroxides and, 192, 194, 195
nitrates and, 201
superoxides and, 197-198
thiocyanates and, 200
alkaline earth salt values,
acetylides and, 198
carbonates and, 202-203
chalcogenides and, 192, 193
imides and, 196
peroxides and, 198
calculation uses,
absolute enthalpies and, 206
electron affinity determination and,
203-204
fluoride, oxide effect on oxidation

state and, 218
hypothetical compound stability determination and, 214-217
446
SUBJECT INDEX
nonionic contribution to bonding

and, 206-214
proton affinity determination and,
204-206
extended classical calculation,
dispersion energy (London) term UL
and, 174-175
Madelung term UY and, 162-171
permanent electrical multipole term
UQand, 176-177
repulsive energy term -UB and, 171174
zero-point energy term UZ and, 176
hydration enthalpiea,
calculation from, 186-188
ionic crystals,
Kapustinskiis formula and, 177-179
Templetons calculations and, li9181
quantum mechanical prediction of, 181186
Linear polymers, phosphmitrilic halide
preparation and, 351-352
Nonelectrolytes, sulfuric acid solutions

and, 394-395
Nuclear processes,
conflicting, activation analysis and,
331-334
Nuclear recoil,
chemical eflects,
Szilard-Chalmen reaction and, 271281
ionic crystals and, 272-278
molecular crystals and, 278-281
Nuclear resonance, boron hydrides and,
131-138
0
Oxidation state,
fluoride d e c t ,
lattice energies and, 218
oxide effect,
lattice energies and, 218
M
Madelung term UY,lattice energies and.
162-171
Magnetic moments, cyclopentadienyl
metal compounds and, 74, 75-76
Mass spectrograph, cyclopentadienyl
metal compounds and, 79-80
Metal chlorides, intercalation in graphite
and, 254-259
Metal hydrogen sulfates, sulfuric acid
solutions and, 395-397
Metal oxides, intercalation in graphite
and, 260-262
Metal sulfides, intercalation in graphite
and, 260-262
Molecular crystals, nuclear recoil and,
278-281
Monoalkylsilnnols,sulfuric acid and, 411412
N
Keutron,
activation analysis, applications of,
334-339
Nitrntes, lattice energies of, 201
Permanent electrical multipole term Uo,

lattice energies and, 176-177
Peroxides, lattice energies of, 198
Phosphonitrilic halides,
charge distribution effects and, 375-377
general elimination reaction and, 378379
geometrical effects and, 377-378
preparation, 348-349
cyclic polymers and, 349-351
h e a r polymers and, 351-352
reaction course and, 352-353
properties, 353-354
acceptor molecule reactions and, 364365
alkyl derivatives and, 363
ammonia and, 360-362
aryl derivatives and, 363
donor solvents and, 362-363
esterification and, 359-360
hydrolysis and, 359
physical and, 354-357
polymerization and, 357-358
primary amines and, 360-362
secondary amines and, 360-362
tertiary amines and, 362-363
447
SUBJECT INDEX
structural theory,
further aspects of, 375
structure, 365-366
high polymer and, 372
medium polymers and, 368-372
spectroscopic data and, 369-372
structural theory and, 372-375
tetrameric compounds and, 367-368
trimeric compounds and, 366-367
Physical properties,
phosphonitrilic halides and, 354-357
sulfuric acid and, 387
Polar bonding,
graphite,
boron nitride compounds and, 261262
caesium graphite and, 236-248
graphite salts and, 248-251
halogen-graphite compounds and,
251-254
metal chloride intercalation and, 254259
metal oxide intercalation and, 260262
metal sulfide intercalation and, 260262
potmium graphite and, 236-248
rubidium graphite and, 236-248
Polarograph, cyclopentadienyl metal
compounds and, 76-77
Polymerization, phosphonitrilic halides
and, 357-358
Post-recoil,
annealing effects, Szilard-Chalmers reaction and, 281-311
Potassium graphite, 236-248
Primary amines, phosphonitrilic halides
and, 360-362
Proton &ty,
determination, lattice energies and,
204-206
Protons,
inelastic scattering, redox reaction
mechanisms and, 3
Pyridine cyclopentadienylide, 68
Q
Quantum mechanical prediction, lattice
energies and, 181-186
R
Radiation, see Irradiation
Radiochemical purification, activation
analysis and, 322323
Random-fragmentation model, SzilardChalmers reaction and, 270
Random-walk process,
correlated pair recombination, post-recoil annealing effects and, 288-290
Rate law for exchange,
radioactive recoil, derivation of, 310311
Reactant ion,
environment interaction, redox reaction mechanisms and, 4-7
Recrystallization,
hot zone,
Ssilard-Cfialmers reaction and, 290292
spherical zones,
retention increase rate and, 309-310
Redox reactions,
drift velocity and, 2-3
mechanisms,
bridged activated complex and, 19-32
gaa phase electron transfer and, 2-4
net two-electron change and, 41-44
nonaqueous solvent reactions and,
44-46
outersphere activated complex and,
12-19
reactant ion-environment interaction
and; 4-7
solution electron transfer and, 8-12
uncertain classification and, 32-41
Repulsive energy term - Un, lattice energies and, 171-174
Ring systems,
five-membered,
complex formation by, 57-97
seven-membered,
existence of complexes of, 107-109
six-membered,
complex formation by, 97-107
Rubidium graphite, 236-248
S
Sample preparation, activation analysis
and, 321
448
SUBJECT INDEX
Secondary amines, phosphonitrilic halides

and, 360-362
Self-dissociation reactions, sulfuric acid
and, 387-389
Silicic acid derivatives, sulfuric acid and,
412-413
Source preparation, activation analysis

and, 323324
Spectroscope,
cyclopentadienyl metal compounds
and, 80,81
phosphonitrilic halides and, 39-372
sulfuric acid and, 393-394
Spherical zones,
recrystallization of, retention increase
rate and, 309-310
Structural theory, phosphonitrilic halides
and, 372-375
Sulfato compounds, sulfuric acid and,
405-407
Sulfato derivative conversion, sulfuric

acid solutions and, 398-399
Sulfuric acid,
antimony and, 418-419
arsenic and, 417-418
boric acid and, 407-408
boric oxide and, 407-408
dialkylsilanols and, 410-411
di-n-butyltin diacetate and, 414
experimental,
conductimetric measurements and.
391-393
cryoscopic measurements and, 390391
spectroscopic measurements and,

393-394
hexa(hydrogensu1fato)stannic acid and,
sulfur dioxide and, 420-421

tellurium dioxide and, 420
tetra(hydrogensu1fato)boric acid and,
408-410
tetramethyltin and, 413-414

trialkylsilanols and, 410
trimethyltin sulfate and, 414
Sulfuric acid solutions,
electrolytes,
acids and, 400-403
acid-base reactions and, 403-405
anhydride formation and, 399
metal hydrogen sulfates and, 395-397
simple conjugate acid formation and.
397
sulfato derivative conversion and,

398-399
weak bases and, 400

nonelectrolytes and, 394-395
Superoxides, lattice energies of, 197-198
Szilard-Chalmers reaction,
solids,
elastic-collision model and, 268-269
elastic-inelastic collision model and.
269
epithermal-reaction model and, 269270
hot-zone model and, 270-271

nuclear recoil chemical effects and.
271-281
postrecoil annealing effects and, 281311
random-fragmentation
model and.
270
thermal exchange model and, 293
413
hydrogensulfato compounds and, 405407
lead[hexafhydrogensulfato)plumbic
acid1 and, 414416
methylstannonic acid and, 414
monoalkylsiIanols and. 411-412
nitrogen and, 416-417
phosphorus and. 417
physical properties of, 387
selenium dioxide and, 419420
self-dissociation reactions of, 387389
silicic acid derivatives and, 412-413
sulfato compounds and, 405-407
Templetons calculations, lattice energies

and, 179-181
Tertiary amines, phosphonitrilic halides
and, 362363
Tetracarbon monofluoride, homopolar
bonding and, !233-236
Tetrameric compounds, phosphonitrilic
Thermal exchange model, Sdard-Chalmers reaction and, 293
Thermodynamic properties, cyclbpentadienyl metaf compounds and, 8082
449
SUBJECT INDEX
Thiocyanates, lattice energies of, 200

Topological theory, boron hydrides and,
138-144
Trialkylsilanols, sulfuric acid and, 410

Tricyclopentadiehyl metal compounds,
see Cyclopentadienyl metal compounds
Trimeric compounds, phosphonitrilic
Triphenylphosphonium cyclopentadienylide, 68-69
V
Valence theory,
boron hydride,
diborane and, 124-128
electron deficiency and, 121-122
higher hydridea and, 128128
molecular orbitals and, 128-131
three-center bond and, 132-124
W
Weak bases, sulfuric acid solutions and,
400
X
Dispersion energy (London)
term UG
Uy,see Madelung term UY
Ua, 8ee Permanent electrical multipole
term UQ
Un, see Repulsive energy term --'UB
UZ,see Zero-point energy term UZ
UL,
see
X-irradiation, cyclopentadienyl metal

compounds and, 77-79
z
Zero-point energy term
giee and, 176
UZ,lattice ener-

Advances in Inorganic Chemistry and Radiochemistry 1 (1959)

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

Advances in Inorganic Chemistry and Radiochemistry 1 (1959)

Încărcat de

Drepturi de autor:

Formate disponibile

Advances in

This Page Intentionally Left Blank

ACADEMIC PRESS INC.

ALL RIGHTS RESERVED

ACADBMIC PRESS INC.

United Kingdom Edition

Library of Congress Catalog Card Number: 59-7692

First Printing, 1959

PRINTED IN THE UNITED STATES OF AMERICA

* Present address : Hamilton College, McMaster University, Hamilton, Ontario,

This Page Intentionally Left Blank

This Page Intentionally Left Blank

Mechanisms of Redox Reactions of Simple Chemistry

Compounds of Aromatic Ring Systems and Metals

Recent Studies of the Boron Hydrides

Lattice Energies and Their Significance in Inorganic Chemistry

Graphite Intercalation Compounds

The Srilard-Chalmers Reaction in Solids

The Phosphonitrilic Halides and Their Derivatives

The Sulfuric Acid Solvent System

This Page Intentionally Left Blank

MECHANISMS OF REDOX REACTIONS OF SIMPLE CHEMISTRY

The principal concern in this chapter will be the mechanism of reaction

may be taken as typical, are part of the material of instruction in general

MECHANISMS OF REDOX REACTIONS

is large enough so that electron transfer over nuclear separations as large

MECHANISMS OF REDOX REACTIONS

MECHANISMS OF REDOX REACTIONS

IV. General Consideration of the Electron Transfer Process in Solution

Let us consider first eome general features of electron transfer between

Fe++ aq. 2 Fc+++aq.

A mechanism of this kind, consisting of discrete steps, can be rejected on

MECHANISMS OF REDOX REACTIONS

The electrostatic interactions are allowed for in the various treatments

MECHANISMS OF REDOX REACTIONS

for the reactions of the second type a species of geometry: LsM+"+lX

sphere necessary to complete the reaction will be dependent on the same

MECHANISMS OF REDOX REACTIONS

+ Z-MCa+++ +d-MC;+++ + l-MCs++.

- Mn04=, I r C l p - IrCls=, Fe(CN)64- - Fe(CN),F, M O ( C N ) ~ 'Mo(CN)s=, F e ( ~ h e n ) ~ +-+Fe(phen)g+++, Fe(dip),++ - Fe(dip):i

abbreviation cpn represents the cyclopentadiene radical, GH;.

MECHANISMS OF REDOX REACTIONS

MECHANISMS OF REDOX REACTIONS

No specific effect of C1-

rearrangements as are necessary are relegated to outer spheres where the

type. Although Co(phen)3++

has a magnetic moment of 5.2 Bohr magnetons (%I),

MECHANISMS OF REDOX REACTIONS

systems. The role of the con-

In summary of the work discussed in this section it can be said that

The term bridged will be used to imply a configuration in which an

MECHANISMS OF REDOX REACTIONS

merits special discussion since it is a representative of the important class

to, Of equal significance are the experiments on the isotope fractionation

nation (1.0035for XI4compared to Pi)is also reported for the nitrogen in

MECHANISMS OF REDOX REACTIONS

For a large number of reactions of Cr(II1) complexes with Cr++, a

+ Cr*++ = Cr++ + 2C1- + Cr*Cl++.

It is difficult to see an alternative formulation that accommodates the

+ Cr*++-+ Cr++ + C1- + Cr*Cl++.

+ Cr*+++ Cr++ + Cr*Cl++

The next stage :

+ 5H+ = 5NH4+ + CrCl++.

+ Cr++ = Cr++ + 2C1- + CrCl++.

+ Cr++-+ Cr++ + C1- + CrCl++.

+ Cr+++ Cr++ + CrCl++