Documente Academic
Documente Profesional
Documente Cultură
I NORGANIC CHEMISTRY
AND
RADIOCHEMISTRY
Volume 7
Advunces in
INORGANIC CHEMISTRY
AND
RADIOCHEMISTRY
EDITORS
H. J. EMELEUS
A. G. SHARPE
University Chemical laborafory
Cambridge, England
VOLUME 7
7959
NEW YORK
COPYRIGHT(O
1959
BY
ACADEMIC
PRESS
INC.
Published by
ACADEMIC PRESS INC. (LONDON) LTD.
BERKELEY
SQUARE
HOUSE,LONDONW. 1
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
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
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
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
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
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
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
. 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
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
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
I. Introduction
. . . . . . . . . . . . . . . . .
11. Electron Transfer in the Gae Phase . . . . . . . . . .
111. 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. 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
+ Ce(1V)
Fe+++ Ce+++
(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
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
+ e- aq.
+ e- aq. - products.
/,
Oxidizing agent
--+
(3)
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
11
12
H. TAUBE
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++
(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
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
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
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-
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.
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)
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
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
-
23
(7)
(8)
(9)
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++
(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
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
25
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
27
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++ +;
(12)
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)
(15)
Cr(OH&,++4-(NH8)6CoOH2+++ products
(16)
30
H. TAUBE
RoX++
07)
1-1
[RoXCrj++-+ products.
1. Y
(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
A)
33
TABLE VII
(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+
34
H. TAUBE
35
complex. But in this case it is not unlikely that an activated complex such
as
b+44+
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
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:
+ 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
39
Np4+
(23)
U4+
(24)
(25)
+
(U4+)[Ce(IVI)[k(H+)-z + k(H+)-a]
(26)
(27)
(28)
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++
(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
41
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*)-.
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
+ k,(TlBrz-)(TlBr'-).
[.
]-
Br Br Br
T1 T1
Rr Br Br
= -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
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
45
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'
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
(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)
(35)
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
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"
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)
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).
29. Cohen, D., Sullivan, J. C., Amis, E. S., and Hindman, J. C., J . Am. Chem. SOC.
78,1543 (1956).
51
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,
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.
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
E. 0. Fischer
. .
and H P Fritz
I . Introduction .
. .
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
. .
82
. . . . . . . . .
83
84
86
. . . . .
. . . . .
.
.
.
.
.
.
.
.
80
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
89
. 90
. . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . . .
Metal Complexes .
.
.
90
. 91
. 91
.
91
.
92
. 92
. 93
.
97
References
87
87
89
97
97
98
99
99
99
101
101
101
101
102
103
106
106
107
109
1. Introduction
57
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
58
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.
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
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).
+ 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
+ 2CsHs
Me(CJl&
+ zC0 + Hs.
(7)
6. Special Methods
+ C&+
TIC& 4- HtO.
62
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
Metal
'
l 1
IH
COI
Cyclopentadienyl Compounds
I P I
PIII
Hn
lo'
I*
64
+ 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.
65
(10)
66
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
68
69
70
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
72
73
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
D. PHYSICOCHEMICAL
AND CRYSTALLOGRAPHIC
PROPERTIES
OF PURE
METALCYCLOPENTADIENYLS
1 . Magnetic Investigations
100
200
300
400 500
TOK-
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
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
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
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
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
~
~-
80
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
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
84
85
Ferrocenylamine
Aniline
10-9
7.2 X 10-"
1.55
PK
8.81
10.14
86
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
89
90
(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.
91
92
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
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
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
97
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
(11)
+ S&- + 40H-
-+
2Cr(CgH&
+ 2H20 + 2S03.
(12)
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
B. PREPARATIVE
METHODS
The synthesis from metal chloride, aluminum chloride, the aromatic
compound, and aluminum powder (in those cases where reduction was
99
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
100
high vacuum at 150, and deposits as brown-black crystals of mp 284285'. It is only moderately soluble in organic solvents such as benzene
+ SOH-+
5Mo(CeH~.)s MOO,-
+ 4Hz0 + 2CeHe.
(13)
101
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
+ 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
103
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
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
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^
BONDING
107
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.
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
+ 3CO.
109
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
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).
113
(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
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).
William N Lipscomb
School of Chemistry. Univenity of Minnesota. Minneapolis 14. Minnesota
I . Introduction
. . . . . . . . . . . . . . . . . . . 118
. . . . . . . . . . . . . 118
. . . . . . . . . . . . . . 121
. . . . . . . . . . . . . . 121
. . . . . . . . . . . . . . 122
. . . . . . . . . . . . . . 124
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
126
128
C.&Es-.
. . . . . . . . . . . . . . . . . . .
.
.
.
.
.
132
. . 133
.
134
.
134
.
134
. .
. .
.
.
.
.
.
.
135
135
.
136
.
136
.
136
. . 137
. . 137
. . 138
References
. 144
. . . . . . . . . . . . . . . . . 146
Acknowledgments
. 148
. 151
. 154
. .
117
154
118
WILLIAM N. LIPSCOMB
1. Introduction
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.
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.
+=
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
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
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
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
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
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).
--
131
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
132
WILLIAM N. LIPSCOMB
(a 1
133
(b)
FIQ.16. Possible BaHa structures. Preference for structure (a) is suggested by the
valence theory.
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
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.
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]=
137
138
0. B5HsBr
WILLIAM N. LIPSCOMB
(MP,
32-34')
AND
B6HJ
(MP,
53')
+ +
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
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
-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
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)
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
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
146
WILLIAM N. LIPSCOMB
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
147
+ q + 2t + 2y = 3 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
2BHS,
+ B2H6--+
(higher products)
(2)
- k(BH,)(B,Ha)
-
k6(B2H6)3/2
149
exchange is
2BHa
+ De + BHlD + HD (surface)
BHD
+ BzHs * B A D + BHa,
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
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
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.
153
4p = p
- n + 28 + x + 3t + 2y + z
+ 6n = p - n + s + s + 2t + 2y + 22
3 ( p - n)
s+x=q
+ + +
+ +
154
WILLIAM N. LIPSCOMB
3. AMMONIATESOF BORONHYDRIDES
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
. .
T C Waddington
The Uoiversity Chemical laboratories. Cambridge. England
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
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
I 58
7. C. WADDINGTON
206
206
207
208
.
.
.
.
.
211
211
213
213
214
218
218
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 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
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
161
LATTICE ENEBGIES
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
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
+ mZ3)
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
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
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
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
118,61
118
'6,27
167
LATTICE ENERGIES
--czi2-kl
/ -
2rvp
exp ( - r h 2 / k Z )
h
L
-22
ij
3
erfc (d;krij).
r,j
(20)
- 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
__
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).
(23)
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
- 24.3(a - 1).
(24)
(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
+ (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
$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
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
~~
11.33725
-UR
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)
172
T. C. WADDINGTON
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
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
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
Uz = NA-hv,,
4
(36)
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
177
LATTICE ENERGIES
(39)
B. THEKAPUSTINSKII
AND TEMPLDWN
CALCULATIONS
OF THE
LATTICE
ENERGIES
OF IONICCRYSTALS
Y,
the number in a
178
T. C. WADDINGTON
(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
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)
(9)
+ 2RT, is given by
(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)
- 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
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
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
32
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
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
(n)
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;
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
+_
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
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
OF THE
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
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
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
-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
-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
204
T. C. WADDINGTON
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.
+ 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
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
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
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
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
!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
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
(51)
CM~~X.
+ 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
(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
+ MC12.
216
T. C. WADDINGTON
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
(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
+ 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
+ +
U
(72
1)RT = AHfoMn+(g) nAH,OX-(g) - AHfoMX,(s)
or rewriting the expression
(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
(55)
> 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
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).
W Rudorff
University of Tiibingen.
Tuebingsn. Germany
. . . . . . . 224
. . . . . . . 226
. . . . . . . 226
.
. 226
. . . . . . . . 228
. . . . . . . . 230
. . . . . . . . 230
. . . . . . . . 230
. . . . . . . . 231
. . . . . . . . 233
. . . . . . . . 233
. . . . . . . . 234
(CF) . . . . . . 235
v . Comparative
References
Survey
. 262
. . . . . . . . . . . . . . . . . 263
. . . . . . . . . . . . . . . . . . . .
223
264
224
B ~ R F F
-i
2
I
'
7.41A
2.45A
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
227
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
229
C-Atoms
O = 0-Atoms
6,OH-Groups
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
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).
233
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.
235
practically unaltered
a difference from the
in identical positions
for preparations with
5.5A
0 F- A toms
=
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
-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
24C
+ Kliq + C2,K;
AH = -29.2 kcal.
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
239
l+a=4.90KX-
0 K-Atoms
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
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.
243
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
244
w.
RUDORFF
245
w. RUDORFF
246
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
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
(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
253
w.
254
RUDORFF
D. INTERCALATION
OF METALCHLORIDES
IN GRAPHITE
1. Preparation and Properties
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.
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;
SiCL
GeCh
TiCll
SnCla
SnCt
PbClz
PbCL
ThCL
ThIi
PCla
PC16
ASClr
SbCls
BiCls
VCL
SOlCl,
SOCl,
SeOCIz
TeCL
MnCl?
COClZ
NiCll
256
w. RUDOEFF
257
258
w. RUDORFF
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
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
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
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?.?
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
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
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
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.
Swelling of
of oxidation.
graphite oxide in water is reyersible and does not change the degree
265
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.
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
.
.
.
.
.
.
.
. . 310
. ,311
. . 311
. .
312
1. Introduction
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
-
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
271
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
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
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
275
Diluent
K&d4
K&d4
KzCd,
K&d4
KzCrO4
KMn04
KMnOl
KMnOl
KzSO4
KzSOi
K2BeF4
K2BeF4
None
KCIO4
Kclo4
None
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
277
278
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
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
280
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
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
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
284
= 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
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
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.
X(THEcr04+- o= DISTANCE)
- 0: distances in
xw
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
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
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 -
= D,,e-E/kT
(12)
289
D = 12vj
=
$2v*-EfkT
(13)
290
29 1
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
292
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 the symbols have their usual meanings and the retention is measured from Ro.
293
'
+ RX G X + RX*
(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).
294
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
pt =
PO.
One obtains
p u [I - (1 - y) Bt exp (-h'/kT)]*'1-7
PO
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
296
untouched. Thus
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,
any y
(37)
297
becomes
Pt
=jE:POW)dE
(38)
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
Eo
- an
(40)
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
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
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
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
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)
305
306
600
500
400
4% 3oo
-1%
200
I00
0.5
1.0
E, IN ELECTRON VOLTS
1.5
307
AR
30
5
P
I-
e
g
2
20
z
W
I-
z
w
"Ea
10
ANNEALING TEMPERATURE
OK
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
309
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
and since T
= r0
- ut.we have
To3
Toa
Inteeratine
+
+
+
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
n(X*) =
i.e.
cn
(c + 1)
n ( X * ) - n(X*),
c+l
c - n ( X * ) , exp
oo
311
R,
--- Rt
1 - R,
- c exp [
vt].
ACKNOWLEDGMENTS
The authors acknowledge with thanks helpful conversations with
312
R
Ro,Rt,R
AR.
r,ro
T
T.
T,
t
U
J7
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,
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
316
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:
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:
(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
320
D. H. F. ATKINS A N D A. A. SMALES
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
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.
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
ACTIVITY~ ~ E A S U R E M E N T S
It has already been mentioned that difficulties are to be expected in
ACTIVATION ANALYSIS
325
326
D. H. F. ATKINS A N D A. A. SMALES
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
D. H. F. ATKINS A N D A. A. SMALES
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 ($7)
Method
G1
Wl
214
219
216
205
204
21.3
21.4
28.5
19
25
rubidium content of G1 and W1. Table V taken from Smales and Webster
(9.9)contains some published results for the rubidium content.
The accuracy of the activation analysis method when applied to sea
water has been checked by comparison with other methods in the case of
rubidium and strontium. The rubidium content (Table VI) of a sample of
TABLE VI
RUBIDIUMIN SEAWATER
Anal?-st
Method
Rb content
(pglliter)
121.4 f 5
120 f 10
sea water has been investigated by Smales and Salmon (92) using neutron
activation and by Smales and Webster (94) with an isotope-dilution
method.
Hummel and Smsles (42) determined the strontium content of sea
water by neutron activation, isotope dilution using Sr8*, and a tracer
technique with SrBS(see Table VII) .
TABLE VII
STROWTIUM IN S E A W A T E H
Method
Strontium content
(mg/liter)
8.1 f 0.1
8.1 f 0.3
8.0 f 0.1
ACTIVATION ANALYSIS
331
From the published results for the accuracy of activation analysis applied to three widely different matrix materials it will be seen that the
accuracy compares very favorably with other modern techniques. The case
of cobalt is particularly interesting, inasmuch as the results obtained by
activation analysis indicated the unreliability of conventionaI analytical
methods when applied to this estimation at low levels.
V. Conflicting Nuclear Processes
Limitations may be imposed on activation analysis by conflicting nuclear processes. These have been mentioned above as becoming of increasing importance in neutron activation when the particle energy increases
above that necessary to bring about the required interaction.
I n the nuclear reactor the normal reaction used in activation analysis
is the ( n , y ) induced by thermal neutrons but there are several routes by
which a nuclide A of atomic number 2 and mass M may be produced:
1. A F - l
2. AF+l
AF
AF
5. AF$.
(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
(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
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
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:
ACTIVATION ANALYSIS
337
338
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
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
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
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
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)
N . L. Paddock
. .
and H T Searle
I . Introduction
. . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
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
1. Introduction
A. GENERAL
+ NH,C1+
PNCli
+ 4HC1
(1)
PHOSPHONITRILIC DERIVATIVES
349
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
C1
P
''
(1)
+ 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
PHOSPHONITRILIC DERIVATIVES
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 )
352
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)
PHOSPHONITRILIC DERIVATIVES
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)*
PHOSPHONITRILIC DERIVATIVES
355
356
TABLE I
CRYSTALLOORAPHIC DATAFOR TEE
PHOSPHONFPRILIC
HALIDES
(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'
(a = 112O)
10.79
5.93 2 Tetragonal
P42/n
44
63
A. Wilson, 2
64
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
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
PHOSPHONITRILIC DERIVATIVES
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
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
PHOSPHONITRILIC DERIVATIVES
359
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
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
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
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)
362
S . 1,.
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
363
PHOSPHONITRILIC DERIVATIVES
364
365
PHOSPHONITRILIC DERIVATIVES
/I
-2 .O
80
09 90
I
91
92
93
I
94
I
95
I
96
I
97
I
98
I
99
H,SO,)
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 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
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.
PHOSPHONITRILIC DERIVATIVES
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
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
PHOSPHONITRILIC DERIVATIVES
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
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
+ 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
P-N stretching
Compound
frequency
(cm-1)
Reference*
1170
1218
1242
1287
1322
I305
1325
1324
1291
1240
1365-1380
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
374
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-
TABLE VI
ELECTRON,
I N TERMS OF
RESONANCE
INTEGRALS
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
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)
PHOSPHONITRILIC DERIVATIVES
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
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
PHOSPHONITRILIC DERIVATIVES
+ CsHs + PhaB
(6)
-+
(7)
(8)
+ 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
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
Z
Pnam
PHOSPHONITRILIC DERIVATIVES
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
PHOSPHONITRILIC DERIVATIVES
383
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 . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
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
.
.
.
.
.
386
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
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
+ HSOd-.
(3)
+ HPS~OF.
(4)
Both of these products are further ionized ( 28); water extensively according t o the equation
H20
(ti)
(6)
+ 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
+ H&04
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
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
=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
392
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
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
(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
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
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
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
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
SH4ClOP
its
(15)
(16)
(1 7)
(18)
397
(19)
+ HSOI-
(20)
(21)
(22)
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,
+
+
+
+
+
+
(24)
(25)
398
+ 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
CZHsSOIH
+ H30+ + HSOI-.
(28)
(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
(30)
(31)
(32)
399
XzO
(33)
+ HsO++ HS04-
(34)
(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
(36)
But boron tri(hydrogensu1fate) is an acid and reacts with the hydrogen sulfate ion,
B(SO4H)a
+ HSOC
B(SOIH),-.
(37)
(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
(41)
3. Acids
401
402
(48)
(49)
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
(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
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
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
KO+
+ HS0,-
+ HSOI-
(53)
(54)
+ B(HS01)d+ Pb(HSO4)s"
(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
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
(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)(-
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)
(63)
(64)
salt
(65)
solvent
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
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
+ 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
+ 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
(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)
+ 2H80++ 2HS04-.
(71)
411
.I
2 H20J/
2.r
H20,
2.r
,/
r
0
I.!
1.0
0.01
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
(72)
412
(73)
+ (ti + 3/n)H2S0,
CAsSi(OEt)3
SO,H
SOIH
n--2
SO4H
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],
(75)
3(CH&3iSOIH
v=7+3/n
+ (2 + l/n)HtO+ + (2 + l/n)HS04-
(76)
(77)
413
~(CHS)SS~SO~H
1/2n[Si2O~(HSO4)2I,m
+ 2.5H,0f + 2.5HS01-.
(78)
+ 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)
+ 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
SOdH = (CHs),Sn+
+ HSOI-.
(84
(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)
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)
(88)
415
(89)
+ HPb(HSOd)s-+
2H&O4+
+ Pb(HSOi)r*.
(90)
(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),
416
(111)
D. GROUPT1. Nitrogen
+ 2HzS04 = NOz+ + Ha0 + 2HS04+ 3HzS01 = K+ + NOz+ + Ha0 + 3HSOd+ 3HzSO4 = 2 x 0 +~ HaO+
~
+ ~HSOI-.
+
(91)
(92)
(93)
(94)
(95)
(96)
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
(100)
2. Phosphorus
(101)
+ His04 = P W H + + HSOd-.
(102)
A&Os
+ 3HzS01
2As0 * HSO,
+ HaO+ + HSO4-.
(103)
(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 ,
+ 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
(109)
&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-
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
(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
(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),
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
(112)
421
+ 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);
R&l--O
(113)
(114)
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.
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
(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
(n),
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
428
AUTHOR INDEX
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
AUTHOR INDEX
429
430
AUTHOB INDEX
431
AUTHOR: INDEX
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
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
433
434
AUTHOR INDEX
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
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
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
436
AUTHOB INDEX
AUTHOR INDEX
437
438
AUTHOB INDEX
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
439
AUTHOR INDEX
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,
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
440
AUTHOR INDEX
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),
AUTHOR INDEX
YY
Yamada, S., 58 (2251,116
Yamashita, J., 182 (130,131), 185 (1321,
g-21
441
SUBJECT INDEX
A
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
B
B.HbO, compounds, topology of, 152-153
Balance equations, boron hydride and,
138
134
SUBJECT INDEX
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
443
thallium,
preparative methods and, 61-62
Cyclopentadienyl metal complexes,
cyclopentadienyl metal carbonyls and,
86-90
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
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
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
445
SUBJECT INDEX
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
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
446
SUBJECT INDEX
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
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
random-fragmentation
model and.
270
413
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
449
SUBJECT INDEX
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
z
Zero-point energy term
giee and, 176
UZ,lattice ener-