MECHANISMS OF REDOX REACTIONS OF SIMPLE CHEMISTRY

H. Taube
Department of Chemistry, University of Chicago, Chicago, Illinois

I. Introduction . . . . . . . . . . . . . . . . . . . 1
11. Electron Transfer in the Gae Phase . . . . . . . . . . . . 2
111. Interaction of Reactant Ions with Environment . . . . . . . . 4
IV. General Consideration of the Electron Transfer Process in Solution . . 8
V. The Outer-Sphere Activated Complex . . . . . . . . . . . 12
VI. The Bridged Activated Complex . . . . . . . . . . . . . 19
VII. Reactions Proceeding by Mechanisms of Uncertain Classification . . . 32
VIII. System in Which a Net Two-Electron Change Is Involved . . . . 41
IX. Reactions in Nonaqueous Solvents . . . . . . . . . . . . 44
X. Conclusion . . . . . . . . . . . . . . . . . . . . 46
References . . . . . . . . . . . . . . . . . . . . 50

1. Introduction

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

of metal ions in which a change of oxidation state takes place on reaction.
These reactions, of which
Fe++ + Ce(1V) 3 +
Fe+++ Ce+++ (1)
may be taken as typical, are part of the material of instruction in general
and analytical chemistry and as such are familiar to anyone with training
in chemistry. Nevertheless, fundamental questions as to the means by
which the change in oxidation state takes place were until recently not
formulated, let alone answered.
The general subject includes as a proper part problems of specific chem-
istry such as are encounted in the reaction, for example, of Fe++ with
Mn04-: What are the steps by which the system proceeds to the final
products, and what are the properties of the intermediate oxidation states
of Mn (or of Fe) which must be involved for such a complex over-all re-
action? Important for inorganic chemistry as such questions are, for the
most part they have been set aside, and attention is directed rather to the
description of the individual steps. Given a process of simple order, we
shall consider questions such as these: What is the closest distance of ap-
proach of oxidant and reductant? What is the arrangement of other groups
besides the reactant metal ions in the activated complex? What motions
of these groups are necessary to consummate the reaction? How are the
1
2 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-trans-

fer 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 experi-
mental 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, involv-
ing 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. Simi-
lar 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
OF CROSSSECTION
COMPARISON FOR ELECTRON
TRANSFER
Ti
WITH GASKINETICCROSSSECTION
2‘.

T, x (cm*) T. X 1016 (cmz)

He 54 15
Ke 65 21
A 134 42

presents the data taken from Hornbeck’s paper comparing the cross sec-
tion for charge transfer with the “normal” cross section for ion-molecule
interaction including atom polarization. Ticonsiderably exceeds T,, and
 MECHANISMS OF REDOX REACTIONS 3
is large enough so that electron transfer over nuclear separations as large
as 5-8 Bohr radii must be invoked. For N2+ in N2 the effect on drift veloc-
ity is smaller than is observed for the inert gases, and the mobility is only
7.5% less than the normal gas kinetic mobility. The difference between an
inert gas on the one hand and a more complex molecule on the other can
be attributed to the circumstance that the equilibrium internuclear sepa-
rations for N2+ and N2 are different, as are the separations of the vibra-
tional levels. Since the electron transfer process is adiabatic, the energies
at the two sites must match closely to make transfer possible.
During the Notre Dame Symposium, Muschlitz and Simons (93) sum-
marized work on the cross sections for inelastic scattering of protons by
various gases. The cross sections for 100-volt protons on He, Ne, Kr, and
A are in the ratios 0, 1.5, 48, and 28 (52); for a complex molecule such as
C2Ha,a value of 112 is observed; there is a decrease in the cross section
for inelastic scattering as the kinetic energies of the ions increase (118).
The interesting qualitative features are: the maximum in the cross section
for the inert gas as atomic number increases and the large cross section
for C2Hs (and other complex molecules). Presumably, the large cross sec-
tion for complex molecules undergoing electron transfer to H+ comes
about because, by using vibrational states, these molecules can provide
many ways of bringing about the energy matching.
Since shielding by other electrons at large distances for a test electron
is complete, and since the cross sections for electron transfer when simple
atoms are involved are large compared to atomic dimensions, hydrogen-
like wave functions can be used to describe the electron distribution a t
the large radii in question. Thus calculations of the probability of electron
transfer, at least in simple cases, meet with a fair degree of success. Such
calculations have been made by Holstein (57) for the symmetrical inert
gas systems, using Hartree-Fock wave functions for the outermost shells.
Gurnee and Magee (48), dealing with the same systems, have used the
Slater wave functions (119) and have chosen in each case a value for the
parameter Q (which in the hydrogen-like case would be (21) where 1is
the ionizing potential) such that the one-electron wave function is approx-
imately satisfied for the range of distances in question. Satisfactory agree-
+
ment of calculated and experimental cross sections for Ne Ne+ and
He 3- He+ is obtained. The treatment of Gurnee and Magee also includes
double charge transfer reactions; the theory and observations made for
such reactions (141) are germane to the issue of 1 versus 2e changes in
+
redox reactions. Experimentally, the cross section for Ne++ Ne =.Ne
+ Ne++ is about 1/4 of that for the l e process in the same system, while for
A, the ratio is about 1/2. Theory (48) suggests a ratio of about 4/2 for both
sets of reactants. For the discussion of charge transfer in diatomic mole-
4 H. TAUBE

cules, Gurnee and Magee used Heitler-London functions for the electronic
part of the wave functions and harmonic oscillator and rigid rotator func-
tions 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+, HD-
HD+, 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 prob-
ability 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 al-
ways 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 chemist’s 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 consid-
erably greater than for solvent having only solvent as neighbor. The forces
that give individuality to the different ions, whether we consider the in-
fluence 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 coordi-
nation spheres completed, have with the surrounding medium can be un-
derstood largely as differences that arise from the slightly different radii
of the two central ions. The distinction between groups in the first coordi-
nation 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
 MECHANISMS OF REDOX REACTIONS 5
field than do those in the second sphere of coordination, so that the distinc-
tion 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 elec-
trolyte .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 coor-
dination 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 rea-
sons. 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 differ-
ent 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 coordina-
tion 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 coor-
dination numbers.
Most of the classical physicochemical methods for the study of hydra-
tion 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 coordi-
nation 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
6 H. TAUBE
these species (tn of the order of 20 to 40 hr at 25'). I n a further devel-
opment (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 hydra-
tion 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 mag-
nitude 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 applica-
tion arising from the incomplete resolution of the electronic bands and
from unresolved questions of intensities. The work on the electronic spec-
tra 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 labili-
ties of hydrated paramagnetic cations is observation of the line broaden-
ing 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 tech-
nique has been applied by Werte (139) to C1- substitution and by Con-
nick (30) to H 2 0 substitution in hydrated cations. There has been a re-
vival 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
 MECHANISMS OF REDOX REACTIONS 7
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 substi-
tution 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 in-
dependent, 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 as-
sociation 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 exam-
ple, no distinction is made between structures such as AI+++(OH& C1- and
AI+++(OHB)~
(OH,), C1-1.
8 H. TAUBE

IV. General Consideration of the Electron Transfer Process in Solution

Let us consider first eome general features of electron transfer between

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

+ e- aq.
1 1

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

I;-,

Oxidizing agent + e- aq. - products.

/,
--+ (3)
A mechanism of this kind, consisting of discrete steps, can be rejected on
the basis of kinetic evidence for all systems which have thus far been
studied in detail (although it would be dangerous to reject it as a possibil-
ity for very powerful reducing agents in water or liquid ammonia). For if
reaction (2) forward is slow compared to (3), the rate of reaction will be
independent of the concentration of oxidizing agent, while if (2) is main-
tained as an equilibrium with (3) rate determining, the rate of reaction
will be inverse in the concentration of Fe+++. Such rate laws have not
heen observed for the systems of simple chemistry which we are discussing.
IIowxer: the feature that the rates can be inverse in the concentration of
the product formed from the reducing agent has frequently been observed
for systems of complex chemistry. In every such case, this kinetic feature
can reasonably he attributed to the formation of an intermediate oxidation
state.
If the model is modified by postulating that the electron is strongly
affected by the presence of the oxidizing agent, the conditions for barrier
penetration are also met, and it is difficult to see what experimental cri-
teria would distinguish the two cases, whether the electron surmounts a
 MECHANISMS OF REDOX REACTIONS 9
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 suf-
ficiently 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 agents-
+ +
for 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 substi-
tutions in the first coordination sphere are not readily accepted; in the
other there are important changes in the first coordination sphere accom-
panying the electron transfer, and further, at least one of the species read-
ily 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 combina-
tions 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 elec-
tronic 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 at-
tempting 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 em-
phasized different ones of these factors, and serious differences of opinion
are recorded.
30 H. TAWE

The electrostatic interactions are allowed for in the various treatments

of the problem (82?,84,137),but the quantitative validity of the treatments
covering close distances of approach has been correctly questioned (104).
In discussing the frequency of electronic transition, Libby (7'6) uses the
radial dependence of hydrogen-like wave functions, with due regard for
the charges of the ions, and discusses electron exchange between transition
metal ions over distances of several Angstrom units; Zener (142), in dis-
cussing the overlap of d electrons of nearest neighbor transition metal
cations (nearest neighbor Mn cations are separated by O=) in a solid
such as LaMn03, states, ". . . Mn+3 ions are sufficiently far apart as to
have no appreciable overlapping." Marcus (82) considers the ions, includ-
ing the first coordination shell, as conducting spheres, takes the distance
of approach for electron transfer to be the sum of the radii of the spheres
and concludes that factors other than the frequency of transition of the
electron determine the rate of Teaction. Marcus et al. (84) leave the prob-
ability of the electron transfer, depending as it does on distance, as a run-
ning parameter, balancing the increasing electrostatic repulsion on close
approach against the greater probability for barrier penetration under
these circumstances. The assumption of the complex ion as a conducting
sphere, necessary because knowledge of the electron distribution in such
ions is so little understood, does, however, gloss over the extraordinary
differences in rate observed for different ligands, and no theory can be
considered satisfactory that does not make allowance for the chemical
individualities of the systems.
For reactions in which there are no changes in tlie first coordination
sphere, the effects arising from the requirement for nonequilibrium polari-
zation of the surroundings are relegated to regions beyond the first sphere
of coordination, but they are nevertheless important. Marcus (82) has
given a mature discussion of the contribution made to the free energy of
activation of these processes by the requirement for solvent reorganiza-
tion. Electrolyte effects also exist, for the transfer of an electron from one
site to another requires the eventual redistribution of ions; as in the case
of dipole interactions, the changes must occur prior to the electron transfer
as part of the requirement that no energy change accompany electron
transfer. There may in fact be an interplay between the solvent dipole and
ion atmosphere contributions, and it is conceivable that the transfer of a
single ion from one site to another can make up a large part of the total
energy requirement of the activation of the environment. A specific mech-
anism (53, 112) of this type which has been suggested is a proton shift
from the coordination shell of the reducing agent t o that of the oxidizing
agent on electron transfer (or described as H-atom transfer).
To make the discussion concrete, let us take as the activated complex
 MECHANISMS OF REDOX REACTIONS 11
for the reactions of the second type a species of geometry: LsM+"+lX
M+nW5,where L , X , and W are ligand groups which may all be alike,
and the M's are metal ions. In arriving a t a configuration having X as a
bridging group, either M+"+l or M+" (or both) has undergone substitu-
tion in the first coordination sphere. We will also assume that the electronic
transfer or redistribution of the electron itself is extremely rapid. This is
an independent assumption which does not follow from the assumed geom-
etry, but it seems to be required to explain why certain systems adopt this
kind of activated complex in undergoing electron transfer reactions. Im-
portant differences as compared to the previous class are realized for these
systems. The electron is effectively distributed over the two sites, and the
binding resulting therefrom lowers the energy needed for the formation
of the activated complex. Whereas in the previous case the electron waits
for an appropriate fluctuation without being able to influence the changes
required in the site to which it will transfer, in the case under present dis-
cussion it affects the energy required to produce the fluctuation there. Pre-
sumably the reaction is consummated by some fluctuation which causes
the separation of Mn+l and Mn, and may require other changes in the GO-
ordination spheres as well. If, for example, the fluctuation in question is
the transfer of X from an equilibrium position close to the oxidizing cation
to one close to the reducing cation, the redistribution of the electron affects
the energy that is necessary to stretch the Mn+l-X bond and to compress
the Mn-X bond.
A feature of the bridged activated complex is that the bridging group
may move from the oxidizing agent to the reducing agent, that is, in the
direction opposite to that of the electron. When the bridging group is nega-
tively charged, the process is electrically equivalent to the transfer of an
atom from the oxidizing agent to the reducing agent, and thus the demand
for reorientation of solvent is less than in the previous mechanism. This
feature also operates to encourage traffic by a path which in certain other
respects appears to be highly unfavorable to reaction.
For reactions of this class, even less than for those discussed earlier,
is it possible to account quantitatively for the rates. In fact, i t can be
considered an ambitious goal even to predict whether in a particular in-
stance substitution to form the binuclear complex is rate determining or
whether, as was assumed in the foregoing discussion, equilibrium to form
the binuclear complex is established rapidly, and the slow step is disrup-
tion of the binuclear complex. For a quantitative discussion, the energy
of interadion in such a binuclear complex must be assessed. This energy
will depend greatly on the electronic structures of the metal ions, on the
properties of the bridging group, and on the properties of the other ligands.
Similarly, the energies required to produce the changes in coordination
12 H. TAUBE

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

facturs. At the present time, far from being able to treat these factors
quantitatively, we still lack a full description of the activated com-
plexes for systems of the widely different electronic structures which must
be considered.
In solvents of low dielectric constant the differences between the two
types of activated complex become. less pronounced. Thus, considering for
example the system involving electron exchange between CIOz and C102-,
when the solvent is water the mechanism would fairly clearly be of the
first type; however, in a solvent of low dielectric constant, free C102-
does not exist, but it will always be in association with the counter cation.
When the dipole moment of the solvent is low, the only mechanism for
bringing about the matching of energy of the electron at two sites is the
motion of the cation, and we thus have the parsimony of atomic readjust-
ments characteristic of the second class. The electron will follow the mo-
tion of the cation from C102- to C102, and the activated complex can be
considered to be of the bridged type [C102NaC102] in the case under pres-
ent discussion.
The two types of activated complex that have been discussed present
t,he gross geometrical features of two types which have been experimen-
tally delineated; however, the types that are observed do not necessarily
present the other features which were referred to in the cases chosen for
the qualitative theoretical discussion. In the next two sections the experi-
mental evidence on the nature of “outer sphere” and “bridged” activated
complexes is discussed (the terms have no connotations other than the
geometry of the activated complexes implied by them), and the general
observations relating to the kinetic behavior of the corresponding systems
will be outlined.

V. The Outer-Sphere Activated Complex

The discussion of experimental methods which opens this section ap-

plies equally well to the next two sections. The development of new meth-
ods 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 prin-
cipal importance is the fact that the new methods attracted many capablc
 MECHANISMS OF REDOX REACTIONS 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 in-
volve, 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 success-
fully 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 ex-
change 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 re-
action partner, resulting in line broadening (136).Another new approach
depends on the broadening of the magnetic resonance absorption of a nu-
cleus, caused by the presence of the exchange partner. (80)
A very ingenious method which in essence satisfies the conditions of no
net change, and which can be applied to ions that exist in d-1 forms, was
introduced by Dwyer and Gyarfas (36).This method exploits the change
in the rotatory power of a system in which a net change of the following
kind is taking place:
d-MCa++ + Z-MCa+++ +d-MC;+++ + l-MCs++. (4)

It is regrettable that with all the activity that has developed in meas-
uring 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 in-
volve 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: MnOr-
The abbreviation phen represents 1,lO-plienanthroline, and dip represents
2,2’-bipyridine.
14 H. TAUBE

- Mn04=, I r C l p - IrCls=, Fe(CN)64- - Fe(CN),F, M O ( C N ) ~ '-

Mo(CN)s=, F e ( ~ h e n ) ~ +-+Fe(phen)g+++, Fe(dip),++ - Fe(dip):i
+ + + , 0s (dip) 3+ + - 0 s (dip) 3+ + + , Co (dip) 3+ + - Co (dip) 3 + + +. The

list is not intended to be exhaustive but only to provide an introduction to

reagents of this class. Oxyions, CN- and halide complexes are represented,
various central metal ions, cations and anions, and species of Coordination
number 4,6, and 8.
Both members of each couple are substitution inert, and for all le-
redox reactions which are possible for mixtures within this group, whether
involving virtual or net chemical changes, electron transfer takes place
much more rapidly than does substitution in the first coordination sphere.
Thus, even in the absence of direct proof that no labilization of the coordi-
nation sphere occurs in the activated complex, it is aimost certain that no
interpenetration of the coordination spheres occurs. The outer sphere ac-
tivated complexes are not necessarily restricted to reagents of this kind,
but it is only for them that it is possible to assert with some confidence
that this type of activated complex operates. It is also characteristic of
the molecules in this group that the change in oxidation state takes place
with little change in the dimensions of the molecule; thus these systems
approach fairly closely the conditions which were chosen for the first class
which was given theoretical discussion in the previous section.
The group considered here has been restricted to molecules which can
reasonably be regarded as coordinatively saturated. There are numerous
molecules, some of which are particularly important, for which certain
elements only of the structure of the first coordination sphere are known
to be preserved on electron transfer. These include3: C102- - C102,U02+
- U02++,etc., Fe(cpn)* - F e ( ~ p n ) ~ +etc., , FeII(porphyrin) - Fe"'
{porphyrin) ,etc. Here the ligands specified in the formulas do not provide
for coordinative saturation of the molecule. In UOz+ + , for example, the
oxygens are colinear with the U atom, and groups held around the girdle
of this molecule are readily replaced; furthermore, it is not known whether
changes in the girdle coordination accompany electron transfer. On the
basis of structures of the reactants alone, no conclusions as to the geome-
tries of the activated complexes can be reached for this class of reagents.
The system among those of present interest which has been most thor-
oughly studied is exchange between Mn04- and Mn04=. A number of
workers (21,61,77,115) have done experiments with this reaction, but the
most successful and complete study is that of Sheppard and Wahl (115).
The rate of the reaction has been proved to be first order in each of the
reactants, as was tacitly assumed in the earlier discussion of reactions of
this class. The specific rate a t 0" is reported as 710 M-I sec-l, E as
a The abbreviation cpn represents the cyclopentadiene radical, GH;.
 MECHANISMS OF REDOX REACTIONS 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= - Mn04-
reaction there is no interpenetration of coordination spheres. Symons (124)
has shown that when Mn04- is reduced to Mn04= by labelled water con-
taining OH-, the isotopic composition of Mn04- is unaltered. This reac-
tion cannot occur without accompanying electron transfer between MnO4-
and 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, 0.16 730

NaOH, CsOH, each 0.08 1730
CsOH, 0.16 2470
NaOH, 0.16, Co(NH&Cla, 0.001 1860
NaOH, 0.16, NaaFe(CN)a, 0.001 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 un-
favorable to the electron transfer so that a readjustment of these atmos-
pheres 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 equi-
librium 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 read-
justment 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 indi-
cated 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
 MECHANISMS OF REDOX REACTIONS 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 outer-
sphere type. These systems differ in that one reaction partner is substitu-
tion 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 FOR SOME COBALT
EXCHANGE COMPLEXES

Temper-
ature k E
IJ ("C) (M-l sec-l) (kcal) Reference

Co(phen)3++- Co(phen)$+++ - -0 3.5 - 38

Co(en)a++ - Co(en)s+++a 0.98 25 6 X lo-' -14 75
CO(NH~)G,++- Co(NH&+++ 0.98 45 <7 x 10-6 - 75
Co(en),++ - Co(NH&+++ 0.98 25 2 x 10-4 - 75

No specific effect of C1- rtt 1-M level.

Not enough systematic work has been done to evaluate the various
factors that affect the rates for the class of reactions which has been con-
sidered, but a few general observations may be in order. Contributing t o
the high rates which are observed for most of them is the circumstance
that because of the electronic structures, very little change in geometry
takes place in the first coordination sphere on electron transfer, and such
'The abbreviation en represents ethylenediamine.
18 H. TAUBE

rearrangements as are necessary are relegated to outer spheres where the

intensities of the interactions are much reduced. The factors that con-
tribute to the remarkable result that for many of these ions an electron
can be accepted without much increase in internuclear distance are funda-
mental to understanding the results on electron transfer. For most mem-
bers of the class the electron population is such that the electrons can
avoid the ligands (the exceptions are in the Co+ + complexes), but this
is not a sufficient condition. Thus, while in the couple V(OHa)a++ -
V ( O H Z ) ~ + + the
+ electrons can also avoid the six ligands around each
ion, a considerable expansion in the first coordination sphere results when
V + + + absorbs an electron. Strong x interactions of the dc electrons with
ligand orbitals must be invoked, but to explain the observations it must
further be supposed that the de electrons of the metal are bonding rather
than antibonding (100, 111).One is led to conclude that bonding of this
type for dr electrons is particularly effective when the ligands have multi-
ple bonds or conjugated multiple bonds, but it may not be limited to
such cases. Thus in terms of the geometrical features which have been
considered, I r C l p - IrCla= probably does not differ from Fe(CN)$- -
F e ( C N ) p , for example, but when a parallel case for the first row transi-
tion series is considered, for example VClz v e r m VC13, a large change
in dimension again results on change in oxidation number.
Besides the influence that the ligand has in determining the geometrical
and electronic structure of the complex, there may also be the direct (but
related) influence of affecting the distribution of the d electrons in the
complex and thus the probability of barrier penetration. Unfortunately,
there are no observations which expose separately the geometrical and
“conductivity” factors, but the second factor will readily be discernible
in some of the systems discussed in the next section.
The slower rates for the Co systems, summarized in Table 111 com-
pared to those for systems described earlier, can be attributed to the
circumstance that large readjustments of the bonds must be made to pro-
vide sites equally favorable to occupancy by the electron. This circum-
stance is itself a result of the electronic structure. The seven d electrons
of Co++ can not all be accommodated in the dc orbitals and so prevent
the close approach of ligands to the metal; when there are only six d
electrons, as in Co+ + +, these can be accommodated in the dr set of or-
bitals and thus do not interfere as markedly with the close approach of
the ligands to the central ion. The difference in dimensions of Co(II1) and
Co(I1) complexes is probably much less marked for the Co(phen) com-
/
plexes than it is for those of the Co-N- type. Although Co(phen)3++
\
has a magnetic moment of 5.2 Bohr magnetons (%I),
the complex prob-
 MECHANISMS OF REDOX REACTIONS 19
ably has a configuration close to that which would make the electronic
state with one unpaired electron the ground state. Thus the readjustment
required in the Co (phen) system preceding electron transfer is probably
/
a great deal less than it is for the Co-N- systems. The role of the con-
\
jugated system of double bonds in phenanthroline in bringing the d elec-
trons to the surface of the complex ion may also be important in causing
a greater rate of electron transfer for the Co(phen) as compared to
/
Co-N- case.
\
In summary of the work discussed in this section it can be said that
the composition of the activated complex is known at least with respect to
the exchanging species, and it is further known, at least for some systems,
that reaction takes place without interpenetration of the coordination
spheres. Important work which remains to be done is, an extension of the
study of the salt effects, and the influence of ligand properties. Important
fundamental questions which remain to be answered are: What is the opti-
mum distance of approach for electron transfer? Precisely how is the
energy matching brought about? How does the nature of the ligands and
the solvent influence the probability of barrier penetration? Rough ideas
as to the optimum distance of approach can perhaps be obtained from
studies of the influence of ionic strength. If for a given salt concentration,
the dimension of the ionic atmosphere is large compared to the separation
of the ions undergoing reaction, a normal salt effect can be expected, but
for the reverse case anomalies can be expected. If the distance of approach
required for electron transfer is large, these anomalies will appear in dilute
solution. It should be stressed that if the electron transfers over a large
distance salt effects will be observed even for a reaction involving a neutral
species, for example, C102 - ClOz-, and it is precisely for such a system
that there is the greatest hope of learning something about the geometry
of the activated complex from study of the salt effects.

VI. The Bridged Activated Complex

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

atom or group of atoms is part of the coordination sphere of both cation
partners in the redox reaction. The characteristic feature is that the co-
ordination sphere of a t least one of the partners has been entered, so that
new bonds are established in making the activated complex. It is by no
means necessary that this geometry correspond to the “strong overlap”
(8.2) case, although, as will be adduced from the evidence for the cases
20 H. TAUBE
which have been studied, strong electronic interaction between the cations
must be invoked.
It is precisely the feature that bonds are broken and new bonds are
formed in generating the activated complex that provides experimental
accesa to some features of mechanism for these systems. In properly
chosen systems, it can be expected that the new bonds formed in the proc-
ess 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 qual-
ify 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 con-
taining (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 observa-
tion 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, sub-
stantially 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).
 MECHANISMS OF REDOX REACTIONS 21
Complex ions of the series (NH~)&o"~Lcan be formed in great vari-
ety, 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 trans-
fer, 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 indi-
cated; 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++ j (5)
merits special discussion since it is a representative of the important class
of reactions in which both members are aquo ions. The system has the
useful and distinctive feature that a variety of testa of mechanism can be
brought to bear, and definite conclusions can be derived from the results.
In common with other systems of this class (see below), the rate law has
two terms (92) [the residue C O ( N H ~(or ) ~Cr(NH3)a) will frequently be
represented by Ro (or Rr)] :

The kinetic data are summarized in Table IV. The specific rate kz which is
the coefficient for the term (RoOH++) (Cr++) is obviously equal to kz'/
K where K is the dissociation constant of RoOH2+ + +. The value of K
has beendetermined (16) as 1.2 x at 25" and p = 1.00; the associated
value of AH is 10 kcal. The value of k, is a revision of that reported
earlier (99), but the new value of k i agrees well with the earlier one.
The tracer work on oxygen atom transfer has already been referred
22 H. TAUBE

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

effects for atoms bound to the oxidizing agent. There is a strong discrimi-
nation (1.035)in favor of the reaction of 016Hcompared to 018H in the
bridging position. This result indicates that the bridging does not merely
function to bond the metal ion centers but that stretching of the ColI1-
OH- bond is, in fact, part of the activation process. A significant discrimi-

TABLE IV
KINETICDATA(143) FOR THE REACTION
RoOH2+++ Cr++ + (p = 1.00;AT 20°C)

Jt1 k?' k? E, E2 A&$ A&$

.\I-' ser-' sec-1 M-1 sec-1 kcal hcal e.u. e.u.
~~

In HzO 0.50 1.58 1.5 X loG 3.3 4.8 -48 - 15

In D20 .I3 0.40 - - - - -

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

the bound ammonias so that stretching of Co1"-NH3 bonds in the activa-
tion process in indicated. However, complete interpretation of this result
is not possible, because it is not known whether the observed effect is
distributed over the fiye nitrogen positions or m-hether it is localized in a
single bond.
The data on the effect of changing the solvcnt froiii H20 to D20 are
incomplete, because the change in K with solvent in not known. It is un-
likely, however, that the change in K will account completeIy for the de-
crease in kz' by a factor of 3.8 which is observed when DzO is used in
place of HzO [it should be noted that exchange of the aquo protons is
rapid, so that in DzO the reactant species is (NH31sCoODz+++],and a
residual Hz0-D20 effect is undoubtedly left. The origin of this effect may
lie in this, that stretching of the 0-H bond takes place in the activation
process. This conclusion is also strongly indicated by the marked change
in kl which occurs when the isotopic change in the solvent is made. How-
ever, the conclusion cannot be insisted on for the kl reaction, because the
case for oxygen transfer in the reaction is not as strong as for the k2' path.
In the context of these results it should be mentioned that even for a sys-
tem which does not involve a hydrogen-containing bridging group (and
which certainly does not react by H-atom transfer), namely the reaction
of (NH3)&rCl++ with C r + + , a decrease in rate by approximately 30%
is observed (98) in changing the solvent from H20 to D20. Thus there is
a noticeable solvent isotope effect even short of stretching H-0 bonds in
the bridging group or transferring H from the coordination sphere of re-
ducing agent to oxidizing agent.
 MECHANISMS OF REDOX REACTIONS 23
For a large number of reactions of Cr(II1) complexes with Cr++, a
bridged activated complex is obviously also involved. Among these is a
reaction of almost classical interest: the catalysis by Cr+ + of the dissolu-
tion of anhydrous CrC13 (I).The product of the reaction has been shown
to be CrC1+ + [rather than Cr (OH2)e+ + + as would be expected for ordi-
nary dilute solutions if complete equilibrium were rapidly established] ,
and the C1- retained has been proved not to have passed through the solu-
tion (129). The reaction can be formulated as
CrCla + Cr*++ = Cr++ + 2C1- + Cr*Cl++. (7)
It is difficult to see an alternative formulation that accommodates the
observations which have been cited. The rapid dissociation of CrC12+ un-
der the influence of Cr++ (I29)-this also proceeds rapidly only to the
CrCl+ + stage-can be understood in a similar manner:
CrC12+ + Cr*++-+ Cr++ + C1- + Cr*Cl++. (8)
The next stage :
CrCl++ + Cr*+++ Cr++ + Cr*Cl++ (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++ + 5H+ = 5NH4+ + CrCl++. (10)
The systems take this course only when Cr++ is present; in the ordinary
+
aquotization process the products (NH3)&rOH2+ + + X - are formed.
The rates of reaction are strictly first order in the Cr(II1) complex and
in the catalyst Cr++, and are independent of acidity over a wide range
of concentration. The mechanisms are obviously similar to those discussed
in the previous paragraph. Kinetic data on the reactions of the Cr(II1)
complexes are summarized in Table V.
The data shown in Table V make possible comparison of the efficiency
of various groups as electron mediators.6 As shown by the quantitative
data for Cr (111) complexes, and indicated also by qualitative observa-
tions with Co(II1) complexes, the iodo complexes yield to reduction much
more rapidly than the fluoro. This result indicates that the electron trans-
fer does ngt occur by direct overlap (76) of the d orbitals of the metal
ions but rather through the agency of the bridging groups (148).The order
‘Use of this term was suggested t o us by Prof. John R. Platt, Department of
Physics, University of Chicago.
24 H. TAUBE
TABLE V
THERATESOF REACTION
OF Cr++ WITH V.uuons Cr(II1) COMPLEXES

Temper
k ature
Oxidant (M-1 Sec-1) (“C) A m AS$ Reference
kcal e.u.

CrF++ 2.6 x lo-* 27 13.7 -20 11

CrCl+ + 8.3 f 2, 9.1 & 1 0 - - 128,ll
CrBr++ >60 0 11
CrNCS++ 1.8 x 10-4 27 - - 11
CrN3++ >l.2 0 - - 11
(NH&CrF++ 2.7 x 10-4 25 13.4 -30 98
(NH&CrCl++ 5.1 X 10-2 25 11.1 -23 98
(NH&CrBr++ 3.2 X 10-1 25 8.5 -33 98
(NHa)5CrI+ + 5.5 f 1.5 25 - - 98
frane-CrC12+(69) -23 x 102 189

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 circum-
stance that d electrons or orbitals are not used in the same way for bind-
ing 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 quali-
tatively 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 octahe-
dral 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 am-
moniated 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 mecha-
nism by which the hexammino ion is reduced is not known; since the
bridged activated complex has been made difficult of access, electron
transfer may in fact take place through the coordination spheres of the
 MECHANISMS OF BEDOX BEACTIONS 25
two reactant ions. To arrive at a bridged activated complex in the casc
of coordinated NHs would require the dissociation of protons. Even with
OH- as the attacking base, this requires about 14 kcal (7) in the way of
activation energy; when only water is available, the activation energy
would be considerably greater.
Experiments with oxidizing agent8 of the pentamminecobalt (111)
class and with complex organic molecules occupying the sixth coordina-
tion position have led to some new and interesting observations. In all of
the systems referred to, only a single (NHs)&o is attached to each ligand;
in every system, transfer of the organic ligand to chromium is observed on
electron transfer, so that direct attack of Cr++ on the organic ligand can
with confidence be accepted as a feature of the reaction mechanism. The
data on these systems are summarized in Table VI.
In the reaction of the acetato and butyrato complexes, attack can be
only at the carbonyl group, and following the argument of Ball and King,
it is likely that it takes place on the oxygen which does not have the
Co(II1) residue. In view of the similarity of the rate of reaction of the
acid succinate complex to that of the acetato or butyrato, it is reasonable
to conclude that for the acid succinate the attack is also at the carbonyl
adjacent to the Co(II1) center; this conclusion is supported by the fact
that the methylsuccinate complex also reacts at about the same rate. The
acceleration noted in proceeding to the succinate ion can be attributed to
the influence of negative charge and/or of chelation of Cr++ in the acti-
vated complex.
Fumarate and succinate appear to afford a clear cut intercomparison ;
the more rapid rate for fumarate, in spite of the circumstance that there
is no possibility of benefit from chelation in the activated complex, sug-
gestn that, for this ion, attack is not restricted to the carbonyl adjacent
to the Co(II1) but takes place at the remote end. The conjugated bond
system provides a mechanism for electron transport, and there is made
available to the system the benefit of avoiding close approach of the posi-
tive charges of the cations. The comparisons including the phthalate com-
plexes support the conclusion reached. For the meta and ortho complexes,
the conducting systems of bonds are lacking, in the first case because of
the relative positions on the benzene ring, and in the second because the
close approach of the carboxyls prevents them from assuming a coplanar
configuration. In the p-phthalate complex, the conjugated bond system
can be achieved, and the rate of reaction is correspondingly more rapid.
An important new effect of acid does not fit the form adopted for re-
porting the data in Table VI. Examples of the usual behavior, that the
reaction with a bridging anion is more rapid than with the acid in the
bridging position, are shown there. With fumarate as the bridging group,
26 H. T A W E

TABLE VI
ACIDSAS ELECTRON
CARBOXYLIC MEDIATORS,
AL = 1.00

Temper-
ature k E
Ligand ("C) (M-*sec-l) (kcd) G S Reference
e.u.

Acetate 25 0.15 126

Butyrate 25 0.08 186
Crotonate 25 0.18 126
Succinic acid 25 0.19 114
Succinate ion 25 -1.0 114a
Me-succinate ester 25 0.22 114a
Oxalic acid 5 >20 126
Maleic acid 5 >20 126
Fumaric acid 5 0.5 7.5 33 114a

-
Fumarate ion 5 -2 114a
o-Phthalic acid 25 0.057 5.1 47 114~
o-Phthalate ion 25 10 114a
m-Phthalic acid 25 0.10 2.6 56 114a
p-Phthalic acid 25 -40 114a

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 ac-
celerated path. This effect, which is marked also for the p-phthalate com-
plexes, presumably arises from this: In a complex constituted as follows,
(?r:H3)&0"' 0 H 0 Cr"
\ I //
c-c=c-C (11)
0 A 'OH
conjugation between the metal ion centers is incomplete. Placing a proton
on the carbonyl oxygen adjacent to the Co(II1) causes a redistribution of
the carbonyl electrons and improves the conjugation through the molecule.
As is required by this interpretation, there is no hint of acceleration by
acid when bifunctional ligands which are unconjugated (succinate, 0- and
m-phthalates) are the bridging groups.
The very rapid rates of reaction observed for the oxalate and maleate
complexes are not accommodated by any of the factors which have been
discussed thus far. The experiments with the succinate complexes show
the chelation effect to be small. Chelation implies attack by Cr++ at the
carbonyl adjacent to the Co(II1), and if this is the case the benefit from
having a conjugated structure is not required; in fact, if there is chelation
with maleate, the coplanar arrangement of the carboxyls required for effec-
 MECHANISMS OF REDOX REACTIONS 27
tive conjugation is impossible. In explanation of the difference between
maleate and fumarate, it is suggested that two different mechanisms of
electron transfer by the bridging groups must be discussed. In one of these
mechanisms, exemplified by fumarate, as electron density is removed a t
one end of the conjugated 7r system, it is replaced at the other; in the sec-
ond, there is net transfer of an electron from the reducing agent to the
bridging group. Such a mechanism does not require the same configura-
tion of atoms as does the one discussed for the fumarate ion and will come
into question for systems which have unoccupied low-lying orbitals. The
fact that, of all the bridging groups discussed, oxalate (88) and maleate
are the most easily reduced, fits in with the suggestion made about the
mechanism of electron transport for these groups as electron mediator.
The suggestion also explains the difference between maleate and o-phthal-
ate, which have similar geometry about the carboxyls, but differ in re-
ducibility. It should be mentioned that with neither fumarate nor maleate
is there a change in configuration of the bridging group on electron trans-
fer. Isomerisation of the maleate is not required by the mechanism pro-
posed because the Cr+ +, in attacking maleate, may impose restrictions on
its geometry.
Thus far the only ligand effects which have been discussed are those
concerned with their role as electron mediators in bridging positions. This
takes account of only one of the eleven coordination positions which need
to be considered for a bridged activated complex between two reactants of
normal coordination number six. It is to be expected that changing groups
in nonbridging positions will also influence rates, and such effects can
easily and unambiguously be demonstrated. Taking (NHJ &oOHz+ + +
as oxidizing agent ana Cr++ as reducing agent, it is observed (125)that
there is a marked acceleration of the reaction by SO4=, and particularly
by pyrophosphate, and both of these ions are incorporated into the product
Cr(II1) complex. That a group other than these ligands is involved as a
bridging group is demonstrated by using (NH,) &oCl+ + as oxidizing
agent; when pyrophosphate is present, both C1- and pyrophosphate are
incorporated in the product Cr (111) complex. Groups differ enormously in
their capacity to accelerate the rate of electron transfer by simple ligand
intervention. Thus, in the system now under discussion, pyrophosphate
is very effective, SO,= less so, and an effect of C1- at 0.1 M level is not
discernible (92).It seems likely that there will be a parallelism between
the ability of the ligand to stabilize Cr(II1) over Cr(I1) and its ability
to accelerate the rate of oxidation by simple attachment to Cr++ in the
activated complex (the relationship would presumably hold in the reverse
way also). The ligand effect of C1-, though slight, must exist to explain
cataIysis by Cr++ of the CrCI---C1- exchange (128) ; or, looking at it
28 H. TAUBE

in the reverse direction, to explain why CrC12+ reacts so much more rapidly
with Cr++ than does CrCl++ (see Table V).
Cis and trans positions can be distinguished in the bridged activated
complexes under present discussion, and in considering the influence of
nonbridging ligands, it is of interest to differentiate the effects at the
two positions. Orgel (99) has suggested that for Co(II1) and Cr(II1)
complexes the incoming electron is accepted in the d,2 orbital, the energy
of this orbital being lowered to the necessary extent by moving groups
trans to each other from the metal ion center. The stretching of the
Co(II1)-OH- bond, when OH is the bridging group, has been demon-
strated (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 inter-
action 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 ex-
tent 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 activa-
tion (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++ + Cr++ +; A S = -33 at p = 1.0 (98) (12)
and
+
(NR3MbBr++ Hg+++; aS = -16 at, p = 0 (S) (13)
is instructive. In both systems the activated complexes presumably have
the 8811163 general geometry and are of the same charge type, so that the
entropy effects caused by concentration of charge in the dielectric should
 MECHANISMS OF REWX REACTIONS 29
be much the same. The disparity in the values of A S is very great, how-
ever, 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 attribu-
table to this, that much more in the way of simultaneous bond readjust-
ment 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, simul-
taneous 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 en-
countered thus far, the concentration of such intermediates appears to be
so small that direct detection is difficult, if not impossible, yet there are
powerful arguments which support such a formulation of the mechanism.
When an activation energy as small as 4 kcal is in question for the
bridged activated complex, we face the difficulty that the activation energy
for substitution on the Cr++ probably exceeds this value. A mechanism
in which the reaction occurs in a single step,
+
(NH&.CoOHr+++ Cr+++, (14)
would require activation energy at least as great as the activation energy
for substitution in the more labile partner (here Cr++), and considerable
additional contributions from other sources can be expected as well. The
situation would not be materially improved by adopting the formulation
(we assume Cr+ +aq. is a hexa-aquo ion) :
Cr(OH2),++P Cr(OH2)6++ 4- HzO (15)

Cr(OH&,++4-(NH8)6CoOH2+++ products (16)

AH# over-all would necessarily exceed AH for process (15), and 4

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

equilibrium. In this situation, the reaction can be represented as:

+ Cr(OH?),++C [RoXCr(H,0)6]++ + H,O

I:,
RoX++ 07)
1-1

[RoXCrj++-+ products.
1. Y

Then the rate of reaction will be given by

k,k,
(L + k2) (Cr+*) (RoX++),
If k v 1 > > k2, AH$ over-all = AH1 + AH2$. The value of AH1 can be
considerably less than AH$ for substitution, because energy is paid back
in the new Cr++-X bond formed, and there may also be a gain from en-
ergy 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 correspond-
ing activation energy is probably larger. The compensating factor is neces-
sarily in the entropies of activation. The entropies of activation for sub-
stitution 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 acti-
vation 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 ten-
ters. 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 ap-
proach 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 com-
pare various groups as electron niediators would be to refer to the coeffi-
cients for the rates expressed in terms of the concentration of the binuclear
intermediates, but so little is known of the relative stabilities of such spe-
cies that an analysis of this kind must be put aside for the present.
The electronic interpretations of the observations for the particular sys-
tems 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
 MECHANISMS OF REDOX REACTIONS 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 dissoci-
ates, it separates into Cr”(1V) and CrX++. Attempts (97) to demon-
strate this reaction have thus far failed, not necessarily because the chem-
istry 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 com-
plex 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 acti-
vated 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 ap-
plies 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 mecha-
nism are more di5cult to derive.

VII. Reactions Proceeding by Mechanisms of Uncertain Classification

The characteristics which have made possible the conclusions about

mechanism for the systems considered thus far are (a) both oxidizing
agent and reducing agent undergo substitution much more slowly than
electron transfer or (b) the oxidizing agent and the oxidized product com-
plex undergo substitution much less rapidly than the electron transfer
takes place, but the reducing agent is substitution labile. The majority of
the systems do not meet these sets of conditions, and for them the conch-
:ions about mechanism are more limited in scope. The reactions are, never-
theless, as interesting and important, and it is a challenge to devise new
experiments in the hope of differentiating between the possible mecha-
nisms. The models for the activated complexes which have been proposed
to explain the electron transfer in a typical case, such as F e + + - F e + + + ,
tire the two which have been discussed thus far, together with a modifica-
tion of the outer-sphere activated complex in which a specific role is as-
signed to protons in carrying the electrons from the reducing agent to the
oxidizing agent. Because the greatest effort has been expended on the
Fe+ + - Fe+ + + reaction, with respect to both experimentation and the-
orizing, this system will be considered first, and such conclusions about
mechanism as seem justified drawn. The problems of mechanism for other
systems are similar, but usually some new element is featured, so that
most of the systems for which rates have been measured, and which have
come to the attention of the author, are referred to.
The early lore on the rate of Fe+ + - Fe+ + + electron exchange is con-
fused by paradoxes (19,68,79,152,136). Partly because of this, the prob-
lem challenged a large number of skillful experimenters] and a general
growth of activity in the entire field ensued. The first definitive experi-
ments were done by Dodson (32); his results are consistent with those of
Linnenbom and Wahl (79)and with those of Betts et al. (19),but not
with those of van Alten and Rice (I,%?), or those of Kierstead (68).
I n Table VII is a summary of the data which have been obtained on
the Fe++ - Fe+f+ reaction. The observation that the values of E are so
nearly the same for a variety of paths has inclined some authors (112)
to the view that a common process, assumed to be H-atom transfer be-
tween the hydration spheres, is taking place for all the systems. This view
has been strengthened by the observation made by Hudis and Dodson
(62) that the specific rate coefficients for the terms
(Fe++)(Fe+++)and (Fe++)(F'e++ A)

(H+)
 MECHANISMS OF BEDOX REACTIONS 33
TABLE VII
SUMMARY OF KINETICDATAON Fe++ - Fe+++EXCHANGE
(TEMPERATURE
25", p = 0.5)

ko
(M-1 Sec-1) E MS Reference

Fe+++ 0.87 9.9 -25 117

FeOH++ 1010. 7.4 - 20 ii7
FeNCS++ 41.5 7.5 -28 74
FeF++ 9.7 9.1 -21 63
-24
-
FeC1++ 9.7 8.8 117
FeF2+ 2.5 9.5 -22 63
FeCIz+ 15 I17
FeF8 -0.5 63

each decrease by a factor of approximately 2 in changing to DzO as the sol-

vent. 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 trans-
fer of hydrogen atoms does not take place-although substantial weaken-
ing 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 formula-
tions will satisfy equally the kinetic requirements:
(a) Fe+++aq. X- Fe++aq. (no X--Fe bonds)
(b) (XFe++)aq. Fe++aq.
(4 Fe+++aq. FeX+ aq.
(4 [FeXFe]4+

Observations made by Lewis et aE. (75), Anderson and Bonner ( 6 ) , and

others (98),for systems of the charge type +2, +3, show that when sub-
stitution in the first sphere of coordination of the oxidizing agent is ex-
cluded, C1- does not exert its usual catalytic effect. Thus it seems likely
that, in the Fe++ - Fe+++ systems, activated complexes (a) and (c) can
34 H. TAUBE
be excluded. It is admitted that this conclusion is based on reasoning by
analogy, but there appears to be no basis for concluding that such reason-
ing will fail here. The argument is strengthened when it is noted that,
when C1- is placed in the first coordination sphere of Co (111) and Cr(II1) ,
the usual large acceleration of rate by C1- is observed. The fact that the
reaction of Cr+ +, in catalyzing the dissociation of CrC1+ + [for this pur-
pose an activated complex such as either (b) or (e) would serve], is very
slow compared to electron transfer accompanied by atom transfer [ acti-
vated complex (d) 1, taken with the previous arguments, suggests that in
the present system the activated complex (d) also serves as the reaction
path.
The comparison (12.9) of the relative rates at which a series of Fe(II1)
complexes react with Cr++, with the relative rates a t which the same
complexes react with Fe+ + , suggests that similar mechanisms operate in
the two series. The argument that the Cr++ - Fe(II1) reactions, at least
with catalytic groups other than OH- in the activated complex, proceed
by a bridged activated complex, is this. When Cr++ reacts with Fe+++ in
the presence of X - , catalysis by X - is observed and CrX++ is formed
quantitatively by this path (129).When Cr (OH,) 6 + + + is oxidizing agent,
vatalysis by C1- has not been observed (6, 92), and CrCl++ is not
formed. The formation of CrCl++ in the reaction of Fe(II1) with Cr++,
in the presence of C1-, proves that a Cr-CI bond is joined in the activated
eomplex. The comparison of the sensitivity to C1- catalysis of Fe(III),
which is substitution labile, to that of Cr(II1) or Co(III), which are not,
shows that substitution in the first coordination sphere is a condition for
marked catalysis by C1-. Hence, the conclusion that C1- makes a bond
not only to Cr+ + but also to Fe+ + + in the activated complex is strongly
indicated if not, in fact, proven. This line of argument, therefore, also
supports the bridged activated complex for the reaction of Fe++ with
Fe+++.
At best, the kind of argument used here can indicate that a major part
of the reaction takes place by a particular reaction path. Based as it is
on gross comparisons of rate, it does not make possible an assessment of
the contribution to the total reaction of rival, but minor, paths. And yet,
only such a complete description can be regarded as satisfactory. To take
n particular example, we are perhaps prepared to accept the conclusion
that with X = F a part of the reaction takes place by the bridged activated

*Plane and Taube report marked catalysis by C1- of the Cr++- Cr+++reaction,
but the data of Anderson and Bonner allow at most a slight effect. It is quite likely
that the conclusion of the former authors is wrong. They observed erratic and
unexplained catalysis in several of their experiments and were probably misled by
such accidental catalysis appearing in the experiment on the effect of C1-.
 MECHANISMS OF REDOX REACTIONS 35
complex. But in this case it is not unlikely that an activated complex such

b+44+
as

also accommodates a good fraction of the reaction. It obviously would be

of some interest to discover how the reaction divides between the two
routes for the series of the halides.
Very little that is definite can be said about paths involving more than
one nonsolvent ligand in the activated complex. When a single anionic
ligand is present, it is reasonable, in view of the rather general behavior
that anions are better electron mediators than neutral molecules, that this
ligand be used in the bridging position. The most reasonable disposition
of a second anionic ligand is much more difficult to choose; further work
on effects for nonbridging ligands using substitution-inert oxidizing agents
is required to illuminate this subject. As before, it is likely that only con-
figurations in which the ligands are directly attached to either Fe++ or
Fe+++ would appear to come into consideration. But even granting this,
the second X - may be in association with the oxidant, with the reductant,
or with both simultaneously as in a system in which a double bridge is in-
volved. The trans effect noted in the reactions of Co(II1) complexes with
Cr++ does not necessarly hold for oxidizing agents of differing electron
structure. Thus when Fe+ + + is being reduced the dislocations required
to cause this center to accept the incoming electrons are probably more
symmetrical than those sufficient for Cr(II1) or Co(II1) as the oxidizing
agent.
A number of other electron exchange reactions of the +2, $3 charge
type have been studied, those of Cr, Co, V, and Eu. Anderson and Bonner
(6) observed for the Cr++ - Cr+++ exchange the rate law

Even in strongly acidic solution the acid inverse path is the dominant one,
and a good value of k, was not obtained. They report 18 x M-l
sec-’ as an upper limit for kl at 25°C and p = 1.0 and k2‘ under the same
conditions as 1.0 X sec-l. From the data of Plane and Taube
(103) on the catalysis by Cr++ of the Cr(OH2)6+++- H20 exchange,
and using Anderson and Bonner’s 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 assuni-
ing 6; if a bridged activated complex is involved, then only 5 will be ex-
changed, 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 cir-
cumstance 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
match at the two sites. The exchange between and C0Y-I (Y rep-
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 known-
electron transfer may be through the intact coordination shells or may re-
quire some dislocation on the substitution-labile Co (11) complex. Adam-
son 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 pro-
vide an interesting comparison with the systems already described in this
section because, for the pair, only electrons in the dr levels come into ques-
tion. 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
 MECHANISMS OF REDOX BEACTIONS 37
as to the rate of electron transfer can be drawn. The issue of relation of
rate and mechanism to electronic structure is also raised in comparing the
kinetic behavior of Eu++aq. - Eu+++aq. (86) with the systems already
discussed. The rate of exchange is very slow, and no contribution by an
aquo or even hydroxo path could be detected (the reduction of water by
Eu+ + limits the time during which exchange can be measured). Exchange
in the presence of C1- was observed, and this reaction path is described by :
Rate (Msec-1) = 6.5 X 10~le-20~800/RT(E~++)(E~+++)(C1-),
at p = 2.0.

This rate law can be cast into the same form as those for the Fe+++ -
Fe++ exchange, but the data on the stability of EuCl++ are lacking, and
a meaningful and dependable comparison cannot be made. It seems likely,
however, that even when expressed in the form (Eu++) (EuCl++), the
corresponding activation energy will be as large as 15 kcal.
The rate of electron exchange between Cu(1) and Cu(I1) in 12 F HC1
has been measured (80) by a nuclear resonance technique. The specific
rate is reported as 5 x 107 M-l sec-1. In the context of this result, the
observation by Gordon and Wahl (45) that the bimolecular reaction be-
tween 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 trans-
fer 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 re-
action of Ag(I1) does not take place between ions of charge +2, but exten-
sive loss of protops takes place in forming the activated complex ( 4 5 ) .
Considerable work on electron exchange reactions of cations of oxida-
tion state +3, +4 has been done. In only one of the systems of thia class
studied are the ions involved (reactants and products) actually of charge
+3 and +4. Keenan (66) reports the rate of electron exchange between
Pu+++ and Pu4+ to be given by:
Rate (Msec-1) = 3 X ~O~(PU+++)(PU~+)~-'*'C"/~~
+ 2.2 X 106(Pu+++)(PuOH+++)e-2*800/RT,at p = 2.0. (21)
38 H. TAUBE

We encounter again an extremely low activation energy, as in the Cr+ + -

(NH3)&oOH++ reaction, and perhaps again the formation of a binuclear
complex in the system is indicated.
The kinetic interpretation of reactions in which C e ( N ) is a reactant
is complicated because Ce (IV) is apparently considerably hydrolyzed
even in acidic solution (116) and furthermore may condense to polynu-
clear 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 es-
sential 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 composi-
tion ( 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 exten-
sive changes in degree of hydrolysis for one or both metal ion centers ac-
company reaction. Thus Furman and Garner (48) report the rate law for
the V (111)-V (IV) exchange as

V(IV) is presumably present as VO++, thus loss of H+ from the V(II1)

center must take place to convert it to V(IV). Inverse acid dependence
is reported for the reaction of Go+++ with Cef3; the conclusion that the
activated complex for this reaction also contains C104- is somewhat ques-
tionable (1231.
The redox reactions of the ions of U, Np, P u , and Am provide a fertile
field for exploring the influence of ( H + ) on the rates of reactions which
involve a change in the state of hydrolysis of the metal ion center. A sin-
gular feature of the chemistry is that, whereas species of oxidation state
+3 and f 4 can exist as ions of the same respective charge, those of + 5
and + 6 in water exist as ions of the type Tu02+ and TuOz++. With the
exception of Am4+, ions corresponding to each of the oxidation states + 3
to +6 exist for each element, and a large number of reactions between
members of this class, as well as with other reagents, are possible.
A complete review of the reactions of these interesting ions is not at-
tempted, and the reader is directed to papers by Newton (95) and Hindmsn
(55) for a more complete discussion. A few of the reactions are referred to
in order to develop some of the features of the chemistry. Among the reac-
 MECHANISMS OF REDOX REACTIONS 39
tions in which the extent of hydrolysis is increased are:
Np4+ + Fe+++ + 2H20 = NpOz+ + Fe++ + 4H+ (23)
U4++ 2Fe++++ 2H20 = UOZ+++ 2Fe++ + 4H+ (24)
U4+ + 2Ce(IV) + 2H20 = UOZ++ Ce+++ + 4H+. (25)
The rate laws are, respectively,
k(Np4+)(Fe+++)(H+)-a (81) (26)
+
(U4+)(Fe+++)[k(HH+)-1 k’(H+)-*)] (18) (27)
(U4+)[Ce(IVI)[k(H+)-z + k’(H+)-a] (9) (28)
In the reactions of U4+,the primary products of the bimolecular steps are
presumably formed by l e - transfer. In the three reactions the loss of pro-
ton is needed, not only to provide a proper bridging group (if indeed the
mechanism is of this type), but also as a pure ligand effect. Direct loss of
an electron from an ion such as U*+ or Np4+ may require an oxidizing
agent of exceptional electron affinity. Prior proton loss, so that the state of
hydrolysis of the reducing agent approaches that of the product, does give
an opportunity for energy matching with oxidizing agents of ordinary po-
tential. It is not at all clear, however, why the differences in the composi-
tions of the activated complexes appear; indeed, it would be expected that
Ce (IV) in reacting with U (IV) could get by with a weaker inverse depend-
ence on (H+) than is the case for Fe+++, contrary to the case actually
observed.
In another group of reactions, the extent of hydrolysis of some metal
ion centers is diminished on electron transfer. It will suffice to cite the rate
laws for the four disproportionation reactions:
+
2 T ~ 0 2 + 4H+ = T%Oz++ + Tu“ + 2Hz0. (29)
They are
k(UOz+)’(H+) (63,67), (NpOz+)’[k(H+) + k’(H+)’] (66‘1, ( 1 2 1 ~ ) ~
k(PuOt+)a(H+) (1091, k ( A m ~ o ~ + ) ( H +(111).
)~
The necessity for the increase in rate with acidity can qualitatively be
understood by following the arguments given earlier. It is difficult to un-
derstand, however, why such apparently similar substances show such
variety in the rate laws for similar reactions. Particularly striking is the
extreme behavior of AmOz+.
An important result which is significant for all the reactions which have
been considered was obtained by Newton (96) in a study of the reaction
2PuO2++ + U4++ 2Hz0 = UOz++ + 2PuOz+ + 4H+. (30)
40 H. TAUBE
The form of the rate law suggests that U4+and Pu02++ react to form an
intermediate containing both U and Pu; all three reactions, formation of
intermediate, intermediate to reactants, and intermediate to products, are
rate determining. It is d s c u l t to formulate an intermediate which behaves
in this manner without invoking a species with definite linkages between
U and Pu, and thus the observation indicates a bridged activated complex
for this reaction as well.
The exchange reaction Np02+ - Np02++ has been studied thoroughly
by Coheo et al. (873. The interesting feature of this system is that part
of the coordination sphere remains intact on electron transfer. However,
since substitution in girdle positions can take place readily, i t is by no
means certain that the activated complex is of the outer-sphere type, and
in fact the authors incline to the view that a bridged activated complex is
involved in some paths. The reaction is first order in each of the redox
species and is independent of (H+) over a considerable range; k a t O'C,
p = 1, is 29 M-I sec-', E = 8.3 kcal M-I and ASS = -24 e.u. C1- in-
creases 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 demon-
stration 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 understand-
ing 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 ques-
tion ivhether the intimate contact obtaining in a bridged activated com-
plex 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 be-
tween reductant and oxidant. It is highly unlikely that in the reaction of
Fef + + with hydroquinone ( 1 4 1 5 ) reaction takes place by electron trans-
fer through the intact coordination shell of Fe+ + + (as suggested by Mar-
cus, 851, when substitution on Fe+ + + can so easily occur. In the reaction
of Cr+ + with quinone, chromium-oxygen bonds are established (127) ;
by analogy, therefore, a similar mechanism can be expected in the Fe+++
reaction. Not many reactions of this kind, of simple chemistry, have been
studied. An interesting reaction recently studied (IS) is that of Fe++
with CO(C~O~)~=. The reaction is quite rapid: k at 20.3"C, p = 0, is 11 x
 MECHANISMS OF REDOX RBACTIONS 41
lo3 M - l sec-l, E = 12.6 kcal/mole-l and A S N 0. Because the reaction
can be studied at low concentration, it proves useful in testing theories of
the influence of inert ions on the rates of ionic reactions.
No claim can be made that a complete review has been made of the
reactions which qualify for discussion in this section. An attempt was
made, however, to present enough of the observations so that the different
aspects of behavior which seem significant are outlined. Much interesting
work readily suggests itself; it is much more difficult, however, to devise
experiments for these labile systems which will lead t o definite answers to
the important question: how are the various groups known to be in the
activated complex arranged with respect to each other?

VIII. Systems in Which a Net Two Electron Change Is Involved

The systems which are treated here are those in which a net 2e- change
takes place for both partners on reaction. In some instances, the reaction
may proceed by a series of two le- steps; in others, it is almost certain
that the reaction proceeds directly by a 2e- change.
A feature which distinguishes the 2e- processes as a class from the le-
class is this, that with few exceptions, serious dislocations in the coordina-
tion sphere accompany reaction. Thus in the changes Cl(VI1) + Cl(V)
+ Cl(II1) + C1(I),there is a progressive reduction in coordination num-
ber 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 coordina-
tion 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 accompa-
nied by atom transfer has been obtained. Thus when SOS" brings about
the changes c103-+ CIOz-, CIOz- + C10-, BrOo- + BrO-, essentially
complete transfer of an atom of oxygen for each 2e- stage of oxidation
occurs (49). The reaction can be formulated (using C103- as an example
of an oxidizing agent), as involving as an intermediate
(OnSOC10*)-.
The reaction is completed by decomposition into SO3 and ClOz- (or, if a
+ +
base such as H20 is required, into SO4- 2H+ C102-). Since C1(V)
undergoes substitution less readily than S(1V) ,the bridging oxygen is de-
rived from c103-rather than SO3=, and net transfer is therefore observed.
This kind of mechanism apparently operates also when MnOz is the oxi-
42 H. TAUBE

dizing agent; when SO4= is formed, one atom of oxygen is derived from
the oxidizing agent (49)
The kind of mechanism in which atom transfer occurs can perhaps be
reasonably expected for these systems in which a state of changed electron
population but unchanged coordination would presumably be a t a very
high energy. A mechanism in which changes in the coordination sphere of
sach reactant accompanies electron transfer over a distance, even if fa-
vorable 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 mole-
cules about T1 is undoubtedly strongly disturbed by the pair of s electrons
that constitute the electronic difference between the two states of oxida-
tion. 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 es-
perimental 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 sep-
arate paths: (Tlf) (T1+++), (T1+) (TIOH++) , and (T1+) (TINOS++)
(108).For the path (Tl+) (TIOH++) , k a t p = 6.0 and a t 25°C 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 dimin-
ish 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 in-
crease. The similar system with Br- added has been thoroughly worked
aut (26). It shows the interesting behavior that the exchange rate first de-
creases, 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
 MECHANISMS OF REDOX REACTIONS 43
quantitatively accounted for by the rate law:
+ +
k~(Tl+)(Tl+++) k2(T1Brz+) ks(TIBra) + k,(TlBrz-)(TlBr'-).
The second and third terms presumably correspond to the establishment of
the redox equilibrium with TI+ and Br2, the first and fourth to direct ex-
change. The coefficients at 25", p = 0.5 M H+ are 1.2 X lov4 M-'
sec-l, 6.2 x 10-7 sec-', 2.7 x lo-' sec-l, and 7.4 x 10-1 M-l sec-l.
With CN-, a diminution in rate at low CN- is observed, and the in-
crease sets in after approximately enough CN- has been added to form
T1(CN)4- (101). Brubaker and Michael (66) require, to explain their
observations on the effect of SO4=, complexes which in addition to TI+
and Tl+++ contain respectively, zero, one, and three sulfate ions.
It is impossible at the present level of knowledge of this subject to find
a unique explanation of the interesting ligand effects which have been
observed. Carpenter and Dodson (26')propose for the activated complex
of composition (TlBr2-) (TlBr4-) the structure

[.
Br Br Br
T1 T1
Rr Br Br
]-
and this is certainly reasonable. It is not clear, however, why a double
bridge should be required in these systems, nor why the activated com-
plex (Tlf) (TI+++) (Br-)2 is not also an effective pathway to products.
Further, it is difficult to see why (TlCl++) (Tl+) is actually less effective
than (TI+) (Tl+++), or why SO4= is more effective than Br- or C1- in
promoting the electron transfer.
The exchange between Sn(I1) and Sn(IV) in strong HC1 has been
studied (64). The solutions show strong interaction absorption, and it is
of interest to inquire into the relation between the species causing the ab-
sorption and the activated complexes. That they are not precisely the
same is true by definition, but it is possible that a small dislocation of
binuclear species causing the interaction absorption suffices to form the
activated complex. The observations on the Sn(I1) - Sn(IV) exchange
are summarized by the equation
Rate (M sec-l) = (Sn(I1)) (Sn(1V)) 7.5 X 10se-10s500/RT; Ai3$ = -50e.u. (31)
Unfortunately, the formulae of the species Sn(I1) and Sn(IV) are not
known, nor is the composition of the activated complex defined with re-
spect 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 re-
actions was made by Neumann and Brown (94). Realizing that SbC1,-
44 H. TAUBE

reaches dissociation equilibrium only slowly, they studied the exchange

between Sb(II1) and SbCls- in HCl solution. There is a diminution in rate
at high C1-, suggesting the conclusion that the optimum composition for
-
the activated complex is Sb+ + + * 3C1- SbC16-. A reasonable structure
for an activated complex of this formula is two condensed octahedra shar-
ing one face.
Thus, evidence for multiple halide bridges in bringing about exchanges
comes up also in the Sb(II1) - Sb(V) case, and in fact finds support also
from work in nonaqueous media.
For the cases considered thus far, it suffices to construct a relatively
simple bridged activated complex, and no evidence for complicated chem-
istry involving intermediate oxidation states is obtained. Not all systems
of the present class are this simple. Thus Rona (113) finds for the exchange
U4+ - U02++ the rate law

and this and other similar observations add a new dimension to the field:
Why in some cases, but not in others, can the transfer of oxidation states
be brought about simply by transfer of one or more groups?

IX. Reactions in Nonaqueous Solvents

There are certain experiments done in hydrogen-labile solvents which

could economically have been discussed in relation to the corresponding
experiments in water. However, work in solvents as little different from
HzO as CH30H promises such new and striking results, that it seemed
preferable to emphasize the importance of the work by collecting the
relevant data in one section.
Experiments on the Fe+ + - Fe+ + + exchange in various alcohols have
been done by Horne and Dodson (59).It is remarkable that the rate of
exchange decreases to very low values at zero content of water, so low
that within the precision of the experiments no exchange is detected. As
HzO is added, the specific rate first increases linearly, then a t a diminish-
ing rate. At comparable concentrations of water, say 1 M , the rate in iso-
propanol is much less (perhaps by a factor of than it is in ethanol.
If a bridged activated complex is indeed involved, the results indicate that
an alcohol is much less effective than is water, a result that seems quali-
tatively reasonable on the basis of such a formulation of mechanism; for
if substantial separation of the groups in the bridge must be achieved in
the activated complex, this is more difficult for alcohol than for H2O. The
large difference between ethanol and water does not seem explicable on
 MECHANISMS OF REWX REACTIONS 45
the basis of a simple barrier penetration mechanism. But left unexplained
still is the tremendous difference in rate between isopropanol and ethanol
as solvents. Is there perhaps a change in coordination number for Fe (111)
between ethanol and isopropanol as solvents? The large bulk of isopropa-
no1 would favor a lower coordination number. To complete the hypothesis,
it would be necessary to assume that the coordination number of Fe++
remains at 6, From this brief essay into the changes produced by the sub-
stitution of alcohols for H20, it is evident that extension of this work is
very worthwhile. An added attractive feature is that the CHsOH remains
liquid down to -98". At this temperature other ions may behave like Cr++
does in water at room temperature and definite evidence about mechanism
may be obtainable simply from product analysis.
Experiments in NHs would also appear to have special significance in
view of the fact that the bridged activated complex, at least in acid solu-
tion, is not readily accessible for an ammoniated ion. A beginning in this
field has been made by Grossman and Garner (46) who studied the Co
(11)-Co(III) exchange. The exchange rates were observed to be much the
same as in water; the authors attach significance to the circumstance that
the rates are comparable to those for the Co (NH;O8+ + + - NHs exchange
( I @ ) , and the comparison does in fact suggest that more than electron
penetration of the coordination spheres is involved.
The experiments of Ward and Weissman (135) have already been re-
ferred to. They represent the only work on redox reactions of highly polar
substances in solvents of low dielectric constant and are to be regarded
as the beginning of a field of work which can be expected to develop con-
siderably. An interesting feature of the results is not only the magnitude
of the specific rates they were able to measure but also the dependence
of rate on the nature of the cation. Table VIII contains a summary of
their data.
TABLE VIII
THE EXCJUNQE
OF ALKALI
NAPETEALENIDES
WITH NAPETHALENJZ

Cation Solvent k (M-1 Sec-1)

K*
Na+
Tetrahydrofurane
Tetrahydrofurane
5.7 *-1107x 10'
Li+
IS+
Na+
Tetrahydrofurane
l,%Dimethoxyethane
1,Z-Dimethoxyethane
*-10'
4.6 f 3 X 108
7.6 3 x 107

~~~ ~

Interesting differences in the kinetics of the Sn(I1) - Sn(1V) and

Sb(II1) - Sb(IV) reactions are produced by changing from water to
media in which the species exist as discrete molecules. For the conditions
46 K. TAUBE

under which the reactions are studied in water, the system can make up
from the supply of C1-, or deliver to the surrounding medium, whatever
is needed to meet the demands of the activated complex. But for the con-
ditions under which the following experiments were run, the systems must
make do with solvent molecules or with reactant molecules. When CClp
is solvent, little help can be expected from it for any influences in the
first coordination sphere of each reactant which might be required.
The exchange reaction of SnClz with SnCll in CzH50H (86) takes
place by the rate law
Rate ( M sec-1) = 1.4 X 1018 e--23,700/RT (SnCI2)(SnCL); (32)
ASS is calculated as -0.4 e.u. The activation energy for the reaction is
much higher than that in concentrated HCl; presumably, some use is made
in the latter case of the large fund of additional C1- available. When
CH30H is the solvent (87) the rate is given by
5.5 x 1010 (SnCls)(SnCL).
e-20*m/RT (33)
The rate law for the exchange reaction of SbCls with SbCl5 in CCll is:
Rate ( M sec-1) = 106 e - l Q f " / R T (SbClb) + 4 X 106ee-15s000/RT(SbC&)(SbCls)'. (34)
The first term presumably corresponds to operation of the equilibrium to
form CI2 and SbC13; the second demands a remarkable activated complex
which Barker and Kahn (12) have succeeded in formulating in a plausible
way. For the analogous exchange of P C 4 with Pels (17) only the decom-
position term is observed. For this path,
Rate (Msec-1) = 1.2 x 10' e-15933/RT (PCb). (35)
These kinds of reactions, extended to include systems in which there
is a net change, and emphasizing solvents such as CCla, deserve much
more attention. There appears to be a better opportunity to define the
activated complex than is the case when water is a solvent, in part because
the specific influence of this solvent is probably much reduced, and in part
because the substitution lability of the molecules has been much reduced.
Thus, in the SbC& - SbC13 case, it may be possible to study C1 exchange
between the two forms, and a comparison of these data with those on Sb
exchange may illuminate the mechanism further.

X. Conclusion

Much of the work reported has taken place in the last few years. Dur-
ing 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
 MECHANISMS OF REWX REACTIONS 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 out-
lined in the individual sections, and there too some of the problems per-
taining to each area have been acknowledged. In this section, a few addi-
tional points will be raised which bring the various sections into closer
context.
An aspect of this subject which merits systematic and intensive ex-
ploration 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+ + + sys-
tem 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 dif-
ference 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 elec-
tron 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 deter-
mining, 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 substi-
tution-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 in-
creasing acidity of the corresponding aquo ions and the correlation sug-
gests that Fe++ +, because of its greater capacity for polarizing the ligand,
benefits less from the substitution of OH- for HzO. The decreased sensi-
tivity of the Fe+++aq. reaction compared to that of (NH3)&oOHz+++
to the substitution of DzO for HzO fits in with this suggestion-far less in
the way of stretching the OH bond is necessary to make the electrons
48 H. TAUBE

TABLE IS
COMPARISON OF LIGANDEFFEms FOR SYSTEM8 OF COMMON CHARQE TYPE
(COMPARISONS +
FOR M ( I 1 I ) L M++)

Relative k for

Fe+++- Fe++ 1 1160 11

F e + + +- Cr++ 1 2800 18
(H?O)&r+++ - Cr++ 1 > 8 X lo' >2 x 10'
(Ir;H&Co+++- Cr++ 1 3 x 10' >50
Eu++" - Eu++ 1 > loo"
8 Calculated assumiiig for EuCi++ is < 1 (U),
and this is probably
a generous upper limit.

available in the case of F e + + + , as compared to Co(II1). A remarkable

result exposed in Table IX is the enormous change in the relative efficien-
cies of C1- and OH- even for cases of the same charge type. This again
may be a consequence of the differing capacities of the ions to cause dis-
sociation of the proton in the activatcd complex. A comparison which
avoids this complicating factor, as for example of the halide ions, would
be very worthwhile.
The relative efficiencies of ligands in promoting electron transfer appear
also to be a strong function of charge type. Thus C1- is almost ineffective
in promoting electron transfer between Ce+ + + - Ce(1V), although it is
quite effective in systems of the charge type +2, +3 in which the inner-
sphere activated complex is accessible and even more effective for Cu ( I ) -
Cu(111.The evidence (97) for the reaction of CrCl++ with M n + + + is
that M n + + + chooses a path involving OH- as bridging group in prefer-
ence to CI-, even when the acidity is large. It may be fairly general that
when ions of large positive charge are involved the hydroxy path pre-
vails. As has already been pointed out, the analysis of these effects is diffi-
cult and complex. The efficiency of the bridging group in making the
binuclear coniplex must be considered, the rate of decomposition of the
binuclear complex to products, and for the labile systems there is also the
question of whether the catalytic group is involved in a bridging or simple
ligand position, or both.
The study of binuciear intermediates is strongly called for. Thus far
such species have not been detected in a direct measurement, but it seems
likely that with properly chosen cations, ligands, and conditions, ap-
preciable concentrations can be built up. This subject has been a matter
of concern in investigations of "interaction absorption," but in most of the
 MECHANISMS OF REDOX REACTIONS 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 con-
centration of Co (en) 3+ + + can be interpreted qualitatively on the assump-
tion that the process
Co(ex&++ (quartet state) + Co(en)a++(doublet state)

ia to some extent rate determining.

A recent observation which may lead to an advance in understanding
the operation of the bridged versus outer-sphere activated complex is this:
C r ( d i ~ ) ~ +(143)
+ has been shown to react very rapidly with Co(II1)
complexes, including Co (NH3)S+ + + ; V (dip)Q + + reacts much less rap-
idly with the same Co(II1) complexes. In these reactions we are almost
certainly concerned with outer-sphere activated complexes. It will thus be
possible to compare rates for the two types of mechanisms for a common
group of oxidizing agents which can be formed in great variety.
An area which has been little touched on is the influence of cations in
promoting redox reactions between anions. There is of course the obvious
effect of inducing reactions by generating intermediates in a redox process,
and such effects have been the subject of considerable study. But there
should also be effects which arise from including both reactants in a com-
plex with the metal ions. In such a case the intermediates would not neces-
sarily be formed as discrete entities. The question at issue is this: Do
cations act as electron mediators for reactions of anions in the same way
that anions Berve for cations? There is no reason why this should not be
the case and, in fact, the work of Ward and Weissman (235) on cation
influence on the rate of electron transfer between naphthalene and naph-
thalenide salts is a beginning in this field.
Much can be learned from studies in nonaqueous solvents. Some of the
special features to be expected from work in NH3 and CH30H have been
pointed out. In addition, NH3 offers the interesting possibility of studying
reactions of solvated electrons, and some new kinetic features can be ex-
50 H. TAUBE

pected when reducing agents are used which approach the alkali metals
in reducing potential but perhaps do not yield an appreciable equilibrium
concentration of solvated electrons.
A vigorous development in these areas, in others which have not been
touched on, and in some not anticipated by the author, can be expected.

REFERENCES
1. cf. Abegg, R., and Auerbach, F., “Handbuch der Anorganischen Chemie,” Vol.
4, p. 76. Hirzel, Leipzig, 1921.
2. Adamson, A. W . , J .Phys. Chem. 55,293 (1951).
3. Adamson, A. W., and Vorres, K. S., J. Inorg. 14 Nuclear Chem. 3, 206 (1956).
4. Amis, E. S., “Kinetics of Chemical Change in Solution,” p. 11. Macmillan, New
York, 1949.
5. Amphlett, C. B., Quart. Revs. 8,219 (1954).
6. Anderson, A., and Bonner, N. A., J. Am. Chem. SOC.76,3826 (1954).
7. Anderson, J. S., Briscoe, H. V. A., and Spoor, N. L., J. Chem. SOC.1943, 361.
8. Armstrong, W. D., and Singer, L., Anal. Chem. 26, 1047 (1954).
9. Baker, F. B., unpublished; referred to in ref. 98.
10. Baldwin, H. W., and Taube, H., to be published.
If. Ball, D. L., and King, E. L., J. Am. Chem. SOC,80, 1091 (1958).
f2. Barker, F. B., and Kahn, M., J. Chem. Phys. 78,1317 (1956).
13. Barrett, J., and Baxendale, J. H., Trans. Faraday SOC.52,210 (1956).
14. Baxendale, J. H., and Hardy, H. R., Trans. Faraday Soc. 50, 808 (1954)
16. Baxendale, J. H., Hardy, H. R., and Sutcliffe, L. H., Trans. Faraday SOC.47,
936 (1951).
16. Bearcroft, D. J., Sebera, D., Zwickel, A,, and Taube, H., to be published.
17. Becker, W. E., and Johnson, R. E., J. Chem. Phys. 79,5157 (1957).
18. Betts, R. H., Can. J . Chem. 33, 1780 (1955).
19. Betts, R. H., Gilman, H. S. A., and Leigh, K., J. Am. Chem. SOC.72, 4978 (1950).
20. Bonner, N. A , J. Am. Chem. SOC.71, 3909 (1949).
2Ou. Bonner, N. A., and Hunt, J. P., J. Am. Chem. SOC. 74, 1866 (1952).
21. Bonner, N. A., and Potratz, H. A., J . Am. Chem. SOC.73, 1845 (1951).
B. Brady, I. G. W., and Krause, J. T., J. Chem. Phys. 27, 304 (1957).
23. Bromted, J. N., and Livingstone, R., J. Am. Chem. SOC.49, 435 (1927).
24. Brown, C. I., Craig, R. P., and Davidson, N., J. Am. Chem. SOC.73, 1946 (1951).
25.Brubaker, C. H., and Michael, J. P., J. Znorg. & Nuclear Chem. 4, 55 (1957).
$6. Carpenter, L. G., and Dodson, R. W., private communication.
fl. Cohen, D., Sullivan, J. C., and Hindman, J. C., J. Am. Chem. SOC.
76, 352 (1954).
28. Cohen, D., Sullivan, J. C., and Hindman, J. C., J. Am. Chem. SOC. 77, 4964
( 1955).
29. Cohen, D., Sullivan, J. C., Amis, E. S., and Hindman, J. C., J . Am. Chem. SOC.
78,1543 (1956).
30. Connick, R. E., private communication.
31. Dodgen, H. W., and Taube, H., J. A m . C h e m SOC.
71,3330 (1949).
32. Dodson, R. W., J . Am. Chem. SOC.72,3315 (1950).
33. Dodaon, R. W.,and Davidson, N., J. Phys. Chem. 56,866 (1952).
34. Duke, F. R.,I. Am. Chern. Soe. 70,3975 (1948).
35. Duke, F. R., and Parchen, F. R., J. Am. Chem. Soc. 78, 1540 (1956).
 MECHANISMS OF REDOX REACTIONS 51
36. Dwyer, F. P., and Gyarfas, E. C., Nature 166,481 (1950).
67. Eichler, E., and Wahl, A. C., J . Am. Chem. SOC.80, 4145 (1958).
38. Ellis, P., Wilkina, R. G., and Williams, M. J. G., J . Chem. SOC.1957,4456.
39. Evans, M. G.,and Nancollas, G. H., Trans. Faraday SOC. 48,363 (1953).
40. Feder, H., resulk9 described by Taube, H., J. Phys. Chem. 58,523 (1964).
41. FUOSS, R. M., J.Am. Chem. SOC.56,1027,1031 (1934).
42. Furman, S. C., and Garner, C. S., J.Am. Chem. SOC.74,2333 (1952).
43. Gates, H. S., and King, E. L., J . Am. Chem. Soc. 80, 5017 (1958).
44. George, P., and Irvine, D. H., J . Chem. SOC. 1954,587.
46. Gordon, B. M., and Wahl, A. C., J . Am. Chem. SOC.80,273 (1958).
46. Grossman, J. J., and Garner, C. S., J. Chem. Phys. 28,268 (1958).
47. Gryder, J. W., and Dodson, R. W., J . Am. Chem. SOC.73, 2890 (1951).
48. Gurnee, E. F., and Magee, J. L., J . Chem. Phys. 26,1237 (1957).
49. Halperin, J., and Taube, H., J . Am. Chem. SOC.74,35,380 (1952).
60. Harbottle, G., and Dodson, R. W., J . Am. Chem. SOC.70, 880 (1948); 73, 2442
(1951).
61. Hartmann, €and I., Schlafer, H. F., 2. phyaik. Chem. (Leipzig) BlW, 116
(1951) ; Z. Naturforsch. 6a, 754 (1951).
62. Hasted, J. B., Proc. Roy. SOC.A205,421 (1951).
66. Heal, H. G., and Thomas, J. G..N., Trans. Faraday SOC. 45,ll (1919).
64. Heidt, L. J., and Smith, M. E., J. Am. Chem. SOC. 70,2476 (1948).
66. Hindman, J. C., in preparation.
66. Hindman, J. C., Sullivan, J. C., and Cohen, D., J . Am. Chem. SOC.76, 3278
( 1954); 79,4029 ( 1957).
67. Holstein, T., J . Phys. Chem. 56,832 (1952).
68. Hornbeck, J. A., J . Phys. Chem. 56,829 (1952).
69. Horne, R. A., and Dodson, R. W., private communication..
60. Hornig, H. C., and Libby, W. F., J . Phys. Chem. 56,869 (1952).
61. Hornig, H. C., Zimmerman, G. L., and Libby, W. F., J. Am. Chem. SOC.72,
3808 (1950).
62. Hudis, J., and Dodson, R. W., J . Am. Chem. SOC.78,911 (1956).
68. Hudis, J., and Wahl, A. C., J . Am. Chem. SOC.75,4153 (1953).
64. Hunt, J. P., and Taube, H., J . Chem. Phys. 19,602 (1951).
66. Ilse, F.‘E., and Hartmann, H., Z . physik. Chem. (Leipzig) B197, 239 (1951);
2.Naturforsch. ba, 751 (1951).
66. Keenan, T. R., J. Phys. Chem. 78,2339 (1956).
67. Kern, D. M. H., and Orleman, E. F., J . Am. Chem. SOC.71,2102 (1949).
68. Kierstead, H. A., J . C h m . Phys. 18,756 (1950).
69. King, E. L., Sister M. John Mark Woods, and Gates, H. S., J . Am. Chem. SOC.
80, 5015 (1958).
70. King, W. R., Jr., and Garner, C. S., J. Am. Chem. SOC.74, 3709 (1952).
71. Kraus, C. A., J . Chem. Educ. 12,567 (1935).
78. Latimer, W. M., “Oxidation Potentials,” p. 50. Prentice-Hall, New York, 1952.
76. Latimer, W. M., Pitzer, K., and Slansky, C. S., J . Chem. Phys. 7 , 208 (1939).
74. Laurence, G. S., Trans Faraday SOC.53, 1316 (1957).
76. Lewis, W. B., Coryell, C. D., and Irvine, J., J. Chem. SOC.Suppl. 2, 5386 (1949).
76. Libby, W. F., J. Phys. Chem. 56,863 (1952).
77. Libby, W. F., J. Am. Chem. SOC. 62,1930 (1940).
78. Linhard, M., Z . Elektrochem. SO, 224 (1944).
79. Linnenbom, V. J., and Wahl, A. C., J . Am. Chem. SOC.71,2589 (1949).
52 H. TAUBE
80. McConnell, H. M., and Weaver, H. E., J. Chem. phys. 25,307 (1956).
81. Magnuason, L. B., and Huieenga, J. R., J . Am. Chem. SOC.73,3202 (1951).
88. Marcus, R. A., J. Chem. Phys. 24, W (19.56).
8.9. Marcus, R. A., J. Chem. Phys. 26,872 (1997).
84. Marcus, R. J., Zwolinski, B. J., and Eyring, H., J. Phys. Chem. 58, 432 (1954).
86. Meier, D. J., and Garner, C. S., J. phys. Chem. 56, 853 (1952).
86. Meyer, E. G., and Kahn,M., J . Am. Chem. Soc. 73,4950 (1951).
87. Meyer, E. G., and Melnick, A., J. Chem. Phys. 61,367 (1957).
88. Milburn, R., and Taube, H., to be published.
89. Milburn, R., and Vosburgh, W. C., J . Am. Chem. SOC.77,1352 (1955).
90. Mitchell, J. H., and Ridler, K. E. W., Roc. Roy. Soe. A l e , 911 (1934).
91. See Moffitt, W., and Ballhausen, C. J., Ann. Rev. Phys. Chem. 7, 107 (19561,
for references to the work of L. Orgel, C. J. Ballhausen, C. Klixbull-Jorgensen,
and J. Owen. See also Holmes, 0. G., and McClure, D. S., J . Chem. Phys. 26,
1886 (1957).
92. Murmann, R. I<., Taube, H., and Posey, F. A., J. Am. Chena. SOC.79, 262
(1957).
93. Muschlitz, E. E., Jr., and Simons, J. A., J. Phys. Chem. 56, 837 (1952).
94. Neumann, H. M., and Brown,R. H., J. Am. Chem. SOC. 78,1843 (1956).
96. Newton, T . W.,in preparation.
96. Newton, T. W.,J. Phys. Chem. to be published.
97. Ogard, A. E., and Taube, H., J. Phys. Chem. 62,357 (1958).
98. Ogard, A. E., and Taube, H., J. Am. Chem. Soc. 80,1084 (1958).
99. Orgel, L. E., Rept. X" Conseil inst. intern. Chim. Solvay p. 289 (1956).
100. Owen, J., D ~ ~ C Z LFaradny
S ~ ~ OSOC.~ S19, 127 (1955).
101. Penna-Franca, E., and Dodson, R. W., J. Am. Chem. Soc. 77, 2651 (1955).
1 W . Penney, W. G., and Schlapp, R., Phys. Rev. 41, 194 (1932).
10.9. Plane, R. A., and Taube, H., J . Phys. Chem. 56.33 (1952).
104. Platzman, R., and Franck, J., 2.Physik 138,411 (1954).
106. Posey, F. A., and Taube, H., J. Am. Chem. Soc. 75,4099 (1953).
106. Posey, F .A., and Taube, H., J. Am. Chem. SOC. 79, 265 (1957).
207. Postmus, C., and King, E. L., J. Phys. Chem. 59,1208 (1955).
108. Prestwood, R. J., and Wahl, A. C., J. Am. Chem. Soc. 70, 880 (1948); 71, 3137
(1949).
109. Rabideau, S. W., J. Am. Chem.Soc. 79,6350 (1957).
110. Rich, R. L., and Taube, H., J. Phys. Chem. 58, 6 (1954).
111. Richardson, J., private communication.
112. Reynolds, W. L., and Lumry, R. W., J. Chem. Phys. 23,3460 (1955).
119. Rona, E., J. Am. Chem. SOC.72,4339 (1950).
114. Rutenberg, A. C., and Taube, H., J. Chem. Phys. 20, 825 (19s).
1 1 4 ~Sebera,
. D. K., and Taube, H., to be published.
116. Sheppard, J. C., and Wahl, A. C., J . A m . Chem.Soc. 79,1020 (1957).
116. Sherrill, M. S., King, C. B., and Spooner, R. C., J. Am. Chem. SOC.65, 170
(1943).
117. Silverman, J., and Dodson, R. W., J. Phys. Chem. 56,846 (1952).
118. Simons, J. H., and Cramer, W. H., J. Chem. Phys. 26,1272 (1952).
119. Slatcr, J. C., Phys. Rev. 36,57 (1938).
1gO. Smyth, H. D., Harnwell, C. P., Hogness, T. R., and Lunn, E. G., Nature 119, 85
(1927).
 MECHANISMS OF REDOX REACTIONS 53
121. Stephanou, S.E.,Asprey, L. B., and Penneman, R. A., U. S. Atomic Encrgy
Commission Rept . AECU 925 (1950).
12la. Sullivan, J . C.,Cohen, D., and Hindman, J. C., J . Am. Chem. SOC.76, 4275
(1954).
12% Sullivan, J. C., Cohen, D., and Hindman, J. C., J . Am. Chem. SOC.79, 3672
(1957).
1.93.Sutcliffe, L. H., and Weber, J. R., Trans Faraday SOC.52, 1225 (1956).
1.84. Symons, M. C.R., J . Chem. SOC.1954,3676.
126. Taube, H., J . Am. Chem.Soc. 77,4481 (1955).
126. Taube, H.,Chem. Rev. 5469 (1952).
1U.Taube, H., unpublished observations.
12%. Taube, H., and King, E. L., J . Am. Chem. SOC.76,4053 (1954).
129. Taube, H., and Myers, H., J . Am. Chem. SOC.76,2103 (1954).
130. Taube, H., Myers, H., and Rich, R. L., J . Am. Chem. SOC.75,4118 (1953).
131. Taube, H., and Posey, F. A,, J . Am. Chem. SOC.78,15 (1956).
13g. Van Alten, L., and Rice, C. N., J . Am. Chem. SOC.70,883 (1948).
133. Van Vleck, J. H., Phys. Rev. 41,208 (1932).
134. Wahl, A. C.,and Deck, C. D., J. Am. Chcm. SOC.76,4054 (1954).
136. Ward, R. L.,and Weissman, S. I.,J. Am. Chcm. SOC.79,2086 (1957).
136. Weiss, J., J . Chem. Phys. 19, 1066 (1951).
18'. Weiss, J.,Proc. Roy. SOC.A222,128 (1954).
158. Weissman, S.I., and Garner, C. S., J . Am. Chent. SOC: 78, 1072 (1956).
139. Wertz, J. E., J. Chem. Phys. 24,484 (1956).
140. Weisendanger, H. U. D., Jones, W. H., and Garner, C. S., J. Chem. Phys. 27,
668 (1957).
141. Wolf, F.,Ann. Physik 34,341 (1938).
1.42. Zener, C., Phys. Rev. 82,403 (1951).
I@. Zwickel, A.,and Taube, H., to be published.
144. Zwolinski, B. J., Marcus, R. J., and Eyting, H.. Chem. Rev. 55, 157 (1955).