Sunteți pe pagina 1din 35

Copyright 1971.

All rights reserved


SOLUTION THERMODYNAMICS IN METALLIC AND 8506
CERAMIC SOLID SYSTEMS
C. B. ALCOCK
Department of Metallurgy and Materials Science
University of Toronto, Toronto, Canada
INTRODUCTION
The thermodynamic data relating to the solid systems which are used in
high-temperature devices, or which are processed at high temperatures to
produce devices that operate near room temperature, are essential to the
chemical aspects of the work of the materials scientist. Such information
makes i t possible to calculate accurately the position of equilibrium in high
temperature service, or at some stage during a chemical process, by a pro
cedure which does not demand a fundamental understanding of the atomic
nature of the materials being used or processed. Quantitative experimental
studies must be made of clearly defned thermodynamic properties and the
results of such experiments can then be brought together in tabular or graph
ical compilations. It would appear that this aspect of the work of the high
temperature chemist is universally applicable and could eventually lead to
a complete quantitative description to suit the needs of the materials scien
tist under nearly all foreseeable circumstances.
The shortcomings of this emphasis toward the collection of thermody
namic data have two i mportant aspects which seem to stress the need
for a more mechanistic approach than a mere collection of data. First,
too frequently, the materials scientist must make additi9ns to simple systems
which have been experimentally documented in order to produce a new de
vice, and second, an atomic model suggests ideas for explaining or predicting
transport properties of materials. In seeking to modify the behavior of a
device material, one frequently needs guidance on the ways in which alloy
ing additions can be i ncorporated and on the limits of solubility and the
chemical potential as a function of composition. No bald collection of thermo
dynamic properties can produce signposts. A model of a system, such as
Sievert's law for example, can be tested experimentally and the results then
help develop the atomic model to a higher level of usefulness.
Clearly the modeling aspect of the work of the chemist in this feld of
materials research must be supported by a highly developed array of experi
mental skills, and it is frequently the case that the more sensitive the experi
mental probes, the more i lluminating are the results
.
In this review, we shall
219
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
Quick links to online content
Further
ANNUAL
REVIEWS
220 ALCOCK
collect some examples of what has been achieved in the feld of thermody
namic information and what can be done to test some models with these
data. Clearly the materials available for such an essay are extremely numer
ous, and some selection, not only of systems but even of the results which
have been published for those systems, must be applied to make a sensible
limit to the discussion. The author has chosen to consider only metallic and
ceramic systems since there are abundant examples to be found in this feld,
and this choice presents a reasonably coherent selection of topics. No apology
will be made for the arbitrarily chosen limits of the discussion since the aim
is to present only a few facets of a very broad topic. It is hoped that an efect
of the treatment will be to sound, certainly not for the frst time, a note of
caution to those who feel that further efort is likely to be unfruitful in an
area of research which is considered passe. Indeed, the power and scope of
modern experimental methods present an impressive collection of skills.
The homemade and frequently crude high-temperature devices of the im
mediate postwar period have been replaced with complex instruments and
new techniques. To select only a few, high-temperature calorimetry, Knud
sen efusion studies with the mass spectrometer, and electrochemistry with
solid electrolytes are examples of new weapons which can be applied to the
study of high-temperature chemistry. There have been a number of reviews
(1-3) of the applications of these and many other devices.
It is worthwhile pointing out at this juncture that, although the experi
mental tools are reasonably precise, the reproducibility of results for a given
system between a number of groups of workers usually leaves much to be
desired. As an example, the results collected by Rizzo, Bidwell & Frank (4)
for the standard free energy of formation of W02(s) from their own emf mea
surements and other sources have an uncertainty limit of 300 cal g
'
atom
-
1
oxygen. They further showed that within their own group of experiments on
this system, the values for materials of diferent impurity content (at the
100 ppm level) could vary by as much as the 300 cal which covered all of
the studies. Needless to say, for any given confguration of electrodes and
electrolytes, it is commonly found that emfs are reproducible to 1 mY,
which corresponds to 50 cal g
.
atom-1 oxygen.
In a similar vein, Cater (5) reports that data from eight laboratories for
the vapor pressure of gold yielded mean values for the heat of sublimation
at 298K of 86,520 to 88,570 cal gatom-
1
by third-law calculation. This
range of heats of sublimation corresponds to a range of about a factor of 2
in the measured vapor pressures at 1500C. It is certainly true that any one
set of results would show a better reproducibility than this, but the mech
anics of constructing and operating vapor pressure measuring systems at
this temperature is still sufciently ill defned that the comparison of the
results of one group of workers with those of another can leave such uncer
tainty concerning the thermodynamic properties.
The proposition that calorimetric data used in conjunction with free
energy measurements lead to a more accurate evaluation of a system appears
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 221
in principle to be a sounder approach to the determination of entropies of
formation than so-called second-law techniques. Those of us who remember
the cause celebre of the standard heat of formation of a quartz, which
changed from -209.9 kcal'mole
-
1 to -217.S0.5 kcal mole
-
1 as recently
as 1962, might be excused some scepticism. The advent of accurate high
temperature calorimetry certainly promises new power to high-temperature
chemistry, and a comparison between a number of groups of workers should
outline the attainable limits of reproducibility in this feld as in the area of
free-energy measurements.
I t is against this background of the limitations of accessibility of the facts
that our success with model building should be judged, and it is hoped that
this aspect will emerge from the subsequent parts of this article.
DILUTE SOLUTIONS OF NONMETALS I METALS
The thermodynamics of solution of nonmetals in pure metals and alloys is
of considerable practical importance both in the incorporation of desirable
dilute solutes and in the elimination of unwanted impurities. One of the
major areas of interest has been the range of composition of validity of
Henry's law for dilute solution, and the interaction coefcients between dilute
solutes.
The rate of accumulation of experimental information has depended on
the accuracy of the techniques available and in this feld there is a wide
spectrum. Probably the most precise measurements have involved the solu
tions of nitrogen and hydrogen in metals and alloys.
The information which can be obtained by careful application of nitrogen
solubility measurements at moderate temperatures is best exemplifed by
the studies of Wriedt & Darken (6) of the efects of cold work on the solubil
ity of nitrogen in pure a iron between 300 and 450C. The nitrogen chemical
potential i n the gas phase was fxed by means of HdNH3 mixtures
.
These
studies showed that the purely dilute solution resulting from the pressure of
nitrogen in the gas phase is augmented by an additional solubility due to the
attractive interactions between dissolved nitrogen atoms and dislocations
which can be i ntroduced by cold working. I t was found necessary to assume,
in i nterpreting the results, that there are two important kinds of sites: screw
dislocations, to which the transfer of a nitrogen atom from the normal lattice
position involved a heat of transfer of -2560 cal mole-1, and edge disloca
tions (and possibly microcracks) , to which the heat of transfer was -20,530
cal mole -1. Only a small amount of this "excess" nitrogen above the true
volume solubility was ascribed to the latter sinks, the major efect being due
to the screw dislocation interaction. These conclusions, which could be
reached by an analysis of the chemical data, were supported by micrographic
studies of the iron samples which were employed. Efects of line defect traps
such as these are only signifcant at low temperatures because the large
number of dislocations which are introduced by cold working is substantially
reduced at higher temperatures when the samples can anneal.
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
222 ALCOCK
Simple solubility studies in which a gas mixture which fxes the chemical
potential of a given element i s equilibrated with a condensed phase can be
carried out over a range of pressures from several atmospheres down to
about 10-
4
Torr. I t would seem probable however that extreme care must
be taken at the low-pressure limit, especially as the temperature increases,
to avoid errors which can arise from the presence of foreign gases (e. g. , water
vapor), which are the products of outgassing and leaks in the system. Mass
spectrometric analysis of the gas phase would seem unavoidable to establish
the validity of experimental results under low-pressure conditions. The
reproducibility of solubility measurements has been reviewed by Lange &
Schenck (7) , who show that even in the case of hydrogen solubilities errors
of 30% can readily accumulate for small solubilities.
The studies of carbon solubilities are usually made either by direct equi
libration with the element or with CH
4
/H2 and CO/C02 mixtures. Because
of the relative instability of methane, the higher-temperature studies (above
1000C) have usually been made with CO/C02 mixtures. The latter system
also defnes an oxygen potential and so cannot be used for studies of carbon
solubilities in metals having a high oxygen afnity. That a number of these
metals also form stable carbides, and therefore that studies of the dilute
solutions of carbon i n the metals demand the imposition of very low carbon
potentials via an equilibrating gas phase, probably accounts for the almost
complete lack of quantitative data. It has recently been shown feasible to
determine the free energies of formation of carbides by the use of galvanic
cells having CaF2 as the solid electrolyte (8). This promises that the measure
ment of the properties of the dil ute solutions of carbon in metals and alloys
which are in equilibrium with solid carbide phases will soon be a fruitful feld
of study.
The experimental position with regard to oxygen solubilities has been
vastly improved by the development of solid oxide electrolyte techniques,
and recent studies by Fromm (9) of the dilute solutions of oxygen in solid
niobium and tantalum at temperatures around 1000C show the potential
of this method. The work of Kubaschewski et al (10
-
12) and later of Kom
arek & Silver (13) has provided another very signifcant extension of our
experimental capabilities. If a sample of a metal which forms a very stable
oxide and has a reasonably high vapor pressure, such as calcium at 1000C,
is equilibrated with a dilute solution of oxygen in a metal and a separate
phase of CaO is formed, then the oxygen potential of the resulting dilute
solution of oxygen i n the metal will equal that of Ca/CaO at 1000C. Results
for higher oxygen potentials, which produce more concentrated solutions of
oxygen in the metal, can be obtained by maintaining the source of calcium
at a lower temperature than the metal sample, since the calcium activity
over the metal-oxygen alloy is thereby reduced. This technique was applied
to elucidate the dilute solution properties of oxygen in the metals titanium,
zirconium, hafnium, and vanadium.
The i nterpretation of the results of dilute solution studies range from
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 223
the quantum mechanical analyses, which have been successful in describing
the energies of solution of hydrogen in metals, to the formal statistical analy
ses of this and other solutes in metals and alloys. Although there is no funda
mental confict between these two approaches, at present the former is very
restricted in applicability when compared with the latter, and the major
technique is still couched in terms of "billiard-ball" atomic models rather
than of eigenfunctions. This has the added advantage that the pictorial na
ture of the approach suggests ways of calculating the confgurational entropies
of dilute solution.
DILUTE SOLUTIONS OF HYDROGEN I THE TRANSITION METALS
The idea has long been current ( 14) , ( 15) that the thermodynamic prop
erties of very dilute solutions of hydrogen in solid metals and alloys may be
accounted for by assuming that the electron is donated to the Fermi surface
of the conduction band and that the proton exists independently. The quanti
tative aspects of this concept can be considered in terms of the simple
Thomas-Fermi calculation of Mott ( 16) , who showed that the conduction
electrons will tend to cluster around the proton, the increased density dn(r)
at a distance r from the proton being given by
where A is a screening constant. According to Herzfeld & Goeppert- Mayer ( 17)
where q is the charge on the proton and l/(ope/on) No is the density of
states of the electrons at the Fermi surface. The electron chemical potential
Je is here equated to the energy of the electrons at the Fermi surface. The
screening of the proton by the conduction electrons results in a decrease in
energy and hence in the main exothermic contribution to the heat of solution
of hydrogen in a meta\
.
Subsequent i mprovements of the simple model (18)
have shown that the electron density undergoes oscillations near the proton
which are not described by the simple Thomas-Fermi approximation of
Mott, but the substantial aspect of this model, which produces a dependence
of the energy of solution of hydrogen on the density of states at the Fermi
surface, remains. The details of attempts to calculate heats of solution by
this technique are set forth in a comprehensive review by Ebisuzaki &
O'Keefe ( 19) and will not be repeated here, but it is worthy of mention that
the experimental results show that the expected relationship to the density
of states is found. There is an al most linear decrease of the heat of solution
of hydrogen in the transition metals with increasing values of the screening
constant. The latter term may be obtained empirically from measurements
of the temperature coefcient of the electronic contribution to the specifc
heat, ", of the solvent metal since
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
224 ALCOCK
k is Boltzmann' s constant and V is the atomic volume.
McQuillan and co-workers (20) have shown that, consistent with this
model, the heats of solution of hydrogen in the body-centered cubic alloys
between Ti-Nb and Nb-Mo increase from -14 kcal in / Ti to zero in the
molybdenum-rich Nb-Mo alloys and to + 10 kcal in pure Mo. In the same
range of composition the density of states at the Fermi surface decreases
from about 0.3 to 0.05 cal
-
1
cm -3. In a parallel fashion the heat of sol ution
of hydrogen in Pd-Ag alloys becomes increasingly exothermic with increasing
silver content, whilst in the composition range from pure Pd to the equia
tomic composition the electronic specifc heat coefcient l drops from 2.4 to
about 0. 2 X 103 cal deg.
-2
mole
-
1 (21), showing a drop in the density of
states at the Fermi surface.
Although this "screened proton" model for dilute solutions of hydrogen
in metals has adequate support from experimental studies of pure solvent
metals and of alloys, one problem remains in the further j ustifcation of the
model. This arises from the theory's prediction of a repulsion between protons
in solution. This interaction plays a signifcant part in the thermodynamics
of more concentrated solutions, but from experimental data it is possible to
conclude that the interaction between the species in solution should be at
tractive rather than repulsive (22).
According to the standard statistical mechanical analyses, the maximum
term in the grand partition function for hydrogen dissolved in a metal can
be obtained by diferentiation (23). The resulting expression shows that
dependence of the activity of hydrogen in solution in the metal as a function
of concentration includes two interaction energy terms. The frst of these is
the extremely dilute solution term which has been discussed above and which
represents the interaction between the dissolved atom and the metal matrix;
the second term accounts for the interactions between the dissolved species.
The experimental data for the Ta-H system have been analyzed in this way
(24) and it has been shown that the results are consistent with a small attrac
tive energy between the dissolved hydrogen species. The statistical mechani
cal expression that is obtained is
The second term on the right-hand side of this equation is the hydrogen
hydrogen interaction term and this is found to be negative according to the
measurements of Mallett & Koehl (25).
The resolution of this problem may come by considering that hydrogen
solutions in palladium increase the lattice parameter of the metal, indicating
a dilatation due to solution. Brodowsky (26) suggests that in dilute solutions
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 225
the strain energy is reduced by pairing the dilatational centers, and hence
the efect on the energy of solution is of an apparent attraction between pairs
of dissolved atoms. In a further study of dilute solutions of boron in palla
dium (27) support was found for the suggestion that the elastic interactions
between solutes, which may be described in terms of interaction energies and
hence in quasi chemical terms, are the source of the apparent attraction be
tween very dilute solutes.
Gallagher & Oates (28) have demonstrated from a collection of experi
mental data that the excess partial entropy of solution of hydrogen in metals
is a linear function of the partial heat of solution. The results for the hcp and
fcc metals fall on one straight line, and those for bcc metals on another. The
linear relationship is what might be expected if the entropy term were vibra
tional in origin and were thus related to the strength of the metal-hydrogen
bond.
Gallagher and Oates make the reasonable assumption that hydrogen
enters the tetrahedral sites in the bcc and hcp metals and the octahedral
sites in the face-centered metals. In the equation which relates the activity
coefcient to the partial thermodynamic quantities
the values of { are 6 for bcc, 2 for hcp, and 1 for fcc. The extensive collection
of data assembled in this paper is represented by the equations
AHXS 1O-4.HH + 11.8;
- lO-4.HH + 15.7;
for bcc and hcp metals
for fcc metals
where ABn is in cal g. atom-1 and /" in entropy units. The authors' observa
tion that the diference between these two lines is roughly R In 6 might be
attributed mainly to the diference in the choice of { for the diferent struc
tures.
Ricca & Giorgi (29) showed in a study of the dilute solutions of Hz and
D2 in Q Zr (hcp) that the solutions obey Sievert's law and that the partial
molar heats and excess entropies of solution of hydrogen and deuterium are
independent of composition (XH 10
-
4
-10-3) . Ricca concluded however
(30) that the simple model in which hydrogen atoms are considered to be
located in tetrahedral sites, and the excess partial entropy arises solely from
the vibrational contribution, cannot explain the experimental results. Be
cause the dissolved particle is the nucleus only, in this circumstance there
might be a unique translational contribution to the excess partial entropy.
The excess partial entropy was obtained by subtracting the simple con
fgurational term
.
Sconfig R In
2
Nzr - NH
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
226 ALCOCK
where NH and NZr are the number of hydrogen and zirconium atoms re
spectively in these calculations.
The vibrational frequencies which were calculated in these studies can
not yet be compared with measured results such as those obtained from
neutron difraction. The only data extant at present seem to be for the
hydrides (31).
TH SOLUTON OF CARBON I ' IRON
The dilute solutions of carbon in metals and alloys have not been so
extensively studied with respect to the dependence on the solvent metal as
has hydrogen. The major efort so far has been with the solutions of interest
to the steel industry. It is found that at quite low concentrations of carbon
in solution in ' iron, the solute shows positive deviations from Henry's law.
The earliest statistical mechanical calculations to account for this behavior
were made by Darken & Smith (32), who assumed that carbon atoms in
neighboring sites exerted a small repulsive force on one another. The equa
tion for the dependence of the activity of carbon on concentration and
temperature which was then obtained was
Nc Nc
log a.- [13 - 120] -+ 1
NFe NFe
The confgurational partial entropy is -R In Xc in this model. Here, the
constant I depends upon the choice of standard state for carbon in solution
in iron, and 8 is given by 8 exp ( -e/ RT), where e is the energy of repulsion
between neighboring carbon atoms. Ban-ya, Elliott & Chipman (33) fnd in
a more complete study of these solutions that E is temperature dependent, and
they suggest that this quantity be separated into a heat and entropy term
thus
8 " exp - [ - J
RT R
where h has the value 3140 cal mole
-
l and S the value of 0.86 cal deg-l
mole-l. These authors also show the treatment of their results according to
models which were developed for these solutions subsequent to Darken and
Smith's work.
The main alternative springs from developments of Spretnak & Speiser's
model (34), in which a carbon atom excludes a number of neighboring sites
from occupancy by other carbon atoms. This is tantamount to making h a
very large positive quantity in Darken and Smith's treatment. This "block
ing" of sites by interstitial solutes has been investigated mainly by McLellan
and co-workers (35) in recent years. The model has been elaborated to allow
for the overlap of the excluded volumes which occurs increasingly with
increasing carbon concentration. Ban-ya, Elliott & Chipman (33) show that
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 227
their experimental results can be accommodated to this model, which is
essentially athermal, by the equation
yc
.
ac
1 7 Yc + lSyc2
'
I t can thus be concluded that the experimental information obtainable by
means of existing techniques is not accurate or informative enough to be
the basis of a decision between the two extremes of the interaction repulsion
energy with a normal confgurational entropy on one hand and the athermal
"blocking" model on the other. The basic diference between the two ap
proaches is that the partial heat of solution of carbon should be a measurable
function of composition according to the Darken-Smith model, but not ac
cording to the Speiser-Spretnak model. The solute interaction term appears
to be too small to be easily obtained from second-law treatment of free
energies of solution, and hence accurate calorimetry or the evolution of a
new experimental technique seems to be mandatory.
The small interaction energies which are to be expected between atoms in
dilute solution are not amenable to measurement in high-temperature sys
tems because of the inherent experimental errors. A successful method for
producing these quantities at low temperatures seems to be the measurement
of internal friction. Two examples of these studies are dilute solutions of oxy
gen in tantalum (36), in which the attractive energy between oxygen atoms
was found to be -2500 cal mole-I, and of oxygen in niobium (37), in which
the results were -1600 cal, -4100 cal, and -6200 calmole-1 for the
formation of oxygen pairs, triplets, and quadruplets respectively. The
experimental values were obtained at low temperatures ("'300C) and
hence the concentrations of these entities are much higher than would be
found at temperatures around lOOOC where the conventional measurements
have been made.
To sum up, the behavior of dilute nonmetallic solutes can be adequately
described by Henry's law as long as solute-solute interactions are negligibly
small. Since these appear to be of the order of 1000 cal mole-I, the con
centration of solute usually exceeds 1 atomic percent before a measurable
departure from the ideal law appears. It is impossible at present to estimate
the composition range over which the ideal law will continue to apply be
cause we have no reliable model of solute-solute interactions in metallic
solutions.
BINARY METALLIC SYSTEMS
Concentrated solid solutions.-The simplest atomic model of binary
alloys, the strictly regular solution, requires that the pairwise bond energies
be independent of composition, the crystal structure remain unaltered across
the whole concentration range, the atoms be randomly distributed, and the
Kopp-Neumann law apply to the heat capacities. Amongst the many liquid
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
228 ALCOCK
alloy systems which have been measured (38) , there have been very few
systems for which the experimental results conform to this behavior, and it
would seem unlikely that the regular solution model anticipates anything
but the gross properties of mixing in the solid state. Indications from the
liquid studies are that the model should ft most closely the behavior of alloy
systems which are formed from elements of the same group of the Periodic
Table. Certainly the results for Ag-Au (39) and Mg-Cd (40) solid solutions
conform reasonably well at temperatures close to the liquidus. The testing
of this model in the solid state is severely limited by the fact that very few
systems show a wide range of isostructural solid solutions. The Hume
Rothery alloys, Cu and Ag with the IIb, IIIb, IVb elements, represent the
typical behavior of many binary systems, having a number of solid solution
ranges of difering crystal structures across the composition range.
The most trenchant contribution of free-electron theory to the calcula
tion of the stabilities of metallic alloys has arisen from subsequent improve
ments of the Mott calculation, which was used in connection with hydrogen
solubilities. Blandin & Deplante (41) have summarized these calculations to
show that the potential suggested by the Thomas-Fermi calculation should
be replaced by an oscillatory potential and the resulting energy of interaction
between two solute species can therefore be either attractive or repulsive
depending on the position of nearest neighbors. The form of the interaction
energy is
1ZlZ2
cos 2kpr
E(r)
kp(hkp + 1)
r3
Here Zl and Z2 are the charges of interacting ions and kp is the wave vector of
electrons at the Fermi surface. The details of this model may not be quantita
tively successful in a number of real alloys because the ion core interactions
other than those which occur via the conduction electrons are not considered.
In fact the earlier calculations of the cohension of copper, for example, con
tained a sizable ion core interaction term (42) and it is still not clear, at least
to the author, whether or not this suggestion has been consigned to limbo as a
result of recent developmen ts.
The oscillatory potential calculations do appear to be consistent with a
reasonable amount of experimental data, and that the energy of interaction
between two solutes depends upon the Fermi energy in this model suggests
that in any alloy system where the metals have diferent valencies, at least
the subregular model (43) should be applied. Theoretical speculations based
on the free-electron theory also suggest that the boundaries between adjacent
solid-solution ranges are fxed only by the requirement that the conduction
band electrons fnd the lowest translational energy levels. Unfortunately,
there do not appear to be sufcient results for any system to make it possible
to test this idea quantitatively. The only system on which a signifcant
number of studies of heats of mixing over a wide composition range by a
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
-

"
l
o
-1000
<-2000
-3000
0
,
'
1
,
,
,
,
\
I
'
SOLUTION THERMODYNAMICS
\1"
,

\
" .
.
I
02
.
,
\
I
I
04
,



Xx
1 1
I I
06
XZ
229
o
x
I
oe 10
FIGURE 1. The experimental results for the heats of formation of copper-zinc alloys
as a function of composition (44-48).
number of workers have been made is the Cu-Zn system, and the experi
mental results do not lead to an unambiguous conclusion (Figure 1). I ndeed
the combination of heat-of-mixing data with results for free energy of mixing
i n this system indicates that the most signifcant diference between the a
and { solid solutions in the Cu-Zn system is the much higher nonconfgura
tional entropy i n the { solutions than the a solutions. The Ag-Zn and Au-Zn
systems are only partially worked out, there being much less information
concerning the heats of mixing i n the zinc-rich phases. The experimental data
to support or refute the free-electron theory of the Hume-Rothery rule for
phase boundary compositions seem at present quite inadequate, and the
narrow ranges of composition and temperature over which some of these
solutions exist clearly demonstrate the need of much painstaking work to
el ucidate the theoretical model.
The most promising start in the comparison of theoretical predictions
with experimental results i n copper and silver dilute alloys seems to be the
measurements by Kleppa (49), which have largely been confrmed in subse
quent work (50) on the heats of formation of the a solid solutions. These are
the copper- or silver-rich alloys with elements from Groups II
-
IVb. The
theoretical predictions (51) do not relate to the magnitudes of the partial
heats of solution of the dil ute constituents, but suggest that the curvature
of the LHM-composition relationship should i ncrease with i ncreasing valency
of the solute element. This prediction is borne out by experiment and indi-
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
230 ALCOCK
cates that a change in the electron/atom ratio must be considered as a cause of
departure of the thermodynamic properties of binary metallic alloys from
the strictly regular solution model.
Another signifcant departure from this idealized behavior is to be found
in the fnite values of nonconfgurational entropies of mixing. Two important
sources of deviation are in the nonadditivity of the heat capacities which is
required by the Kopp-Neumann rule, and in magnetic contributions arising
from the mixing of atoms with unflled d shells.
On solid alloys direct calorimetric evidence concerning heat capacities
is even more sparse than heats of mixing, but Freedman & Nowick (52) used
the phase diagrams for a number of systems displaying a very limited solid
solubility to calculate the excess partial entropies of the dilute solutes. It
was found from such calculations that the excess entropy is usually larger,
the larger the endothermic heat of mixing. The heats of mixing must neces
sarily be endothermic in order to bring about the narrow range of solid solu
tions. In more concentrated solutions, including those with an exothermic
heat of mixing, Kubaschewski (53) has demonstrated that the excess entropy
of mixing, at its maximum value, is related to the maximum value of the heat
of mixing by a linear equation
where TlB is the boiling point of component 1, etc. Furthermore, it appears
that these maxima in heat and entropy of mixing occur at approximately the
same composition.
The solid solutions which contain transition metals as one component
will have an electron-spin contribution to the entropy of mixing which has
the term
Bpin @
R L Xi In (ii + 1)
where Xi is the atom fraction of the species bearing an atomic moment
ii
Bohr magnetrons per atom. This "magnetic" entropy will only be present in
paramagnetic alloys and disappears in the ferromagnetic state. It therefore
follows that there must always be a fnite contribution to the excess entropy
from this source in paramagnetic alloys. In most of those systems which have
been reasonably accurately studied, this factor appears to make a signifcant
contribution (54, 55). The precise determination of the magnetic entropy
requires knowledge of the atomic moments for each species, and these may
vary with composition (56). It would seem that the information available at
present is insufcient to allow a precise a priori calculation of this term and
that, in those few cases in which sufcient data are at hand, the magnetic
entropy only accounts for part of the excess entropy.
One feature of palladium alloys which has appeared is a change in sign of
the heat of mixing across the composition range. This efect obviously can-
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 231
not be accounted for through the use of a constant pairwise bonding model,
but is predicted by free-electron theory for the mixing of two elements of
diferent valency (57). The efect is found clearly in Ag-Sn alloys, for example
(58). Of course it is difcult to see exactly what is meant by the "valency"
of a transition metal species, especially since the d band population will
change as a function of composition on the formation of alloys (59). Never
theless, the change of sign of the heat of mixing is quite well established for
the alloys between Co, Ni, Fe+ Pd (54, 55). The alloys between Co, Ni,
Fe+Pt (60, 61), and Cu+Pd, Pt (62, 63) all have larger exothermic heats of
mixing than these palladium alloys and no range of composition in which .
the heat of mixing is endothermic. This suggests that the change of sign is
related to the weakness of the interactions in the Co, Ni, Fe+Pd alloys,
which allows the efects of the electron/atom ratio in raising or lowering the
Fermi surface to become signifcant.
IntermetaUic compounds.-A number of compounds have now been in
vestigated either by means of liquid-tin calorimetry (64) for heats of forma
tion or by vapor pressure measurements for free energies of formation. There
are two theoretical approaches to the prediction of the stabilities of inter
metallic compounds; the frst to be discussed here is due to Kubaschewski
(65). A study of the crystal structures and interatomic distances in inter
metallic compounds reveals that there is usually an increased density of
packing of the atoms in many of the compounds when compared with the
component elements. Laves (66) has brought together a considerable amount
of structural evidence in support of this observation. Kubaschewski began
the calculation of the energy of formation of a compound by assuming that
the like-atom bond strengths, BAA, BAB, can be directly obtained from the
heats of vaporization and the coordination numbers in the pure elements.
Unlike-atom bond strengths were then obtained by use of the geometric mean
postulate of Pauling (67) EAB (EAA EBB)112. In order to make allowance for
the variation of bond strength with bond length and for the contributions
from atoms beyond the frst coordination shell, an inverse dependence of the
bond strength on the internuclear distance was assumed.
Kubaschewski thus arrives at an "efective coordination number" which
for example in an AB alloy would be, for the metal A,
2rA is the internuclear distance between neighboring atoms in pure A, dA
is the distance between A atoms in the compound, and dAB is the correspond
ing distance between A and B nuclei.
The values for energies of formation which are obtained by using this
theory are in many cases quite close to the measured values; the latter are
seldom better established than 500 cal mole-I. A particular series of com-
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
232 ALCOCK
pounds which would seem amenable to this treatment are the Laves phases of
general formula AB2 These show a considerable increase in density on
formation from the elements. A comparison of the calculated heats of forma
tion with those obtained by liquid-tin solution calorimetry by King &
Kleppa (68) and by Kubaschewski (65) using data from the literature shows
that the agreement is usually within 2000 cal g' atom-1.
Support for the general idea has also been obtained in Knudsen efusion
studies of the isomorphous compounds of general formula UX2 (X Ga, In,
Si, Ge, Sn, Pb) in which it was found that the heats of formation of these
compounds, which were obtaine
d by second-law calculation, were in the
expected sequence (69, 70) .
Another chemical bond approach has been proposed by Brewer (71) and
this is based on Engel's (72) original postulate. The crystal structures of
metallic systems are related to the number of sand P bonding electrons per
atom. The suggested sequence is
ela
Structure
1-1.75
bcc
1 .8-2.2
hcp
2.5-3.0
fcc
In Engel's original publications the values were only 1.0, 2.0, and 3.0, but
this integral assignment has now been abandoned.
The crystal structure therefore provides a clue to the electronic states of
atoms in pure metals or alloys and intermetallic compounds. According to
this approach, the number of sand p electrons which are used in bonding a
particular metallic system frequently difers from the number of these elec
trons in the separated gaseous atoms. A Group II element has no bonding
electrons in the atomic ground state, and therefore the one of the s electrons
must be promoted to the higher-energy p level before the bonding confgura
tion is achieved. Thus, the Group I I S
2
confguration is changed to sp by
promotion of one electron, and there are now two bonding electrons avail
able. Brewer shows that the promotional process can involve d electrons in
the transition metal systems and lead to the formation of unpaired d elec
trons, which can also be used for bond formation. According to this theory,
the d electrons in no way contribute to the criterion for crystal structure but
may add to the cohesion. Using the heats of sublimation of the elements,
Brewer was able to obtain values for s, p, and d bonding energies for a wide
range of confgurations and to predict from these calculations the existence
of intermetallic compounds such as ZrPtg having a very high stability (73) .
I t certainly appears that a self-consistent model of bonding in inter metallic
compounds could be generated from this concept, but because the of un
certainty of the values of the promotional energies and because the energy of
formation of a compound is obtained as the small diference between large
numbers, it seems that this theory's useful regime is only in the strongly
interacting systems. A description of the weakly bonded systems, such as
those treated by Kubaschewski, would not seem convincing at the moment.
Furthermore, in order to accommodate a number of metallic systems to the
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 233
theory, it has become necessary to have mixtures of bonding states, thus
Sp O
.
6 corresponds to a mixture of the sand sp confgurations. These two
states have energies which do not difer widely for a number of elements, but
the inclusion of the possibility of mixing states in the post hoc description of
a system introduces a degree of fexibility which detracts from the predictive
capability of the approach. Finally, as Massalski (74) pointed out, a given
binary alloy can assume a number of diferent crystal structures as the tem
perature is changed. The example considered by Massalski is the Cu-Ga
system in which, at 20 atomic percent gallium, the hcp phase appears at
350C, the fcc phase at 600C, and the bcc structure at 860C
.
Here, indeed,
the balance of stability is probably too subtle to be demonstrated convinc
ingly through the Engel-Brewer approach.
These two theories deal with bonding in intermetallic compounds in
terms of a model of bonding in the metallic state which could be called
"covalent" and appear able to account semiquantitatively for compounds
having a small heat of formation or those of the transition metals having ab
normally large heats of formation. There remains an important group of
compounds which ft neither description and which is sometimes described
as "heteropolar" or "partially ionic." The treatment of these by the chemi
cally inclined workers in the feld involves the use of the electronegativity
concept. It would seem inappropriate to attempt to calculate the stabilities
of systems having partially flled conduction bands by means of this basically
molecular approach. A review by Robinson and Bever of the stabilities of
compounds having the NiAs structure clearly shows that the connection is
only, at best, semiquantitative. Thus, FeSe has twice the heat of formation
of NiSe but the same diference in electronegativity, and MnSb has twice
the stability of MnBi (75).
The screening by the conduction band, which has been shown to occur
when an impurity of diferent charge from the host lattice ions is immersed in
dilute metallic solution, should remove the large Madelung contribution to
the stability of a metallic system in which electron transfer takes place. The
ionic contribution to bonding should therefore be insignifcant in metallic
systems. Mott (76) showed that the residual "charge" on each Wigner
Seitz cell in a binary alloy could be responsible for the energies of order
disorder transformation, since the periodic structure of the ordered phase
would include only the Madelung contribution. The Thomas-Fermi pro
cedure was again employed, and the result for the equiatomic CuZn com
pound showed that screened charges, + 1/2e for Zn and -1/2e for Cu, were
reduced by the conduction band to 0.075e. The contribution to the co
hesion of the compound from these screened charges would therefore amount
to approximately 1 kcal g' atom-I. Further calculations based on the oscil
latory potential by Harrison & Paskin (77) lead to a result for the CuZn
alloy which is close to that of Mott, and so we may assume that this approach
represents the free-electron view of charge transfer reasonably well. If this
is so, the"ionic" contribution in the Madelung sense to bonding in com
pounds which are substantially metallic and in which screening can occur
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
234 ALCOCK
seems very small and should not be expected to account for the larger heats
of formation, but may be responsible for order-disorder energies.
The contribution to bonding which results from electron transfer may
come about simply because the transferred or relocalized electron is at a lower
energy level. This need not supply a Madelung contribution and so should
not be thought of as an ionic bond, but the energy of transfer from one atom
to another might be related to the diference between the atomic electro
negativities. This seems especially reasonable following Mulliken's sugges
tion (78) that the electronegativity is related to the electron afnities and the
ionization energies of atoms. The efect on the heat of mixing of two elements
might thus still be related in a semiquantitative way to the diference be
tween the respective electronegativities.
The model proposed by Varley (79) , in which an exothermic contribution
to the heat of mixing occurs as a result of electron transfer from a conduc
tion band derived from one component to that derived from the other com
ponent (a "two-band" model of alloys), contains the essence of this sugges
tion, and the theory produced substantially correct predictions of the exo
thermic component of heats of mixing.
In connection with the disordering energies of metallic compounds, it can
be seen from the previous discussion that the use of the constant pairwise
bonding model in the conventional Bragg-Williams calculation cannot be
expected to be quantitatively satisfactory. Hillert (80) has found the use of a
concentration-dependent interaction energy to be satisfactory if the expres
sion for the heat of mixing in the strictly regular equation fHM =aXAXB (a
is a constant of the system) is replaced by a more general unsymmetrical
expression AH
M
= fXA 2XB.
This is Hardy's (43) subregular model if the entropy of mixing is main
tained in the random form
/M = R I Xi In Xi
To summarize, the general conclusions which can be drawn from the
experimental evidence concerning binary alloys would seem to lean heavily
towards an abandonment of the constant pairwise bonding model, except as a
special case when elements from the same group of the Periodic Table are
alloyed. The subregular model would seem to be the next most credible basis
for discussing binary alloy systems, except where d-d band transitions are
likely to be involved. Even here though it is not safe to assume, except as a
rough approximation, that the confgurational entropy term is the only part
of the temperature-dependent component of the free energy of mixing.
INTERSTITIAL CMPOUNDS
The thermodynamic properties of borides, carbides, silicides, and nitrides
have received a good deal of attention in recent years. The data for heats of
formation are quite substantial; some are the result of direct calorimetric
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 235
studies, but the majority are obtained by free-energy-of-formation measure
ments. The problems of achieving high accuracy for the thermodynamic
properties can be minimized when data are available over a wide range of
temperatures, and no single technique can sufce. Results can be obtained by
vapor pressure studies only at temperatures around 1750C for the stable
compounds, and electromotive force and gas-solid equilibration studies are
usually made for these systems at around lOOOC. There is therefore a large
temperature interval between the results for the same compound, and sig
nifcant changes of state and ranges of non stoichiometry occur in the tem
perature interval. Interrelation of the results for any given system is there
fore difcult, and use of the second-law procedure very l imited. The purely
experimental problem of gathering a signifcant amount of information to
establish one system accurately is very much hampered by the prevalence of
solid-solution formation and by the ease with which impurities can be built
into an originally pure substance by chemical reactions which occur during
the attempted measurement. It is of interest, then, to begin our discussion of
the stabilities of interstitial compounds by referring to the few measurements
which relate to solution formation.
Two studies have been made of nitride-carbide solid-solution formation.
That of Grieveson was on the system TiC-TiN (81) and there have been a
number of studies of the UC-UN system (82-84). In both systems the solid
solutions showed Raoultian behavior of the pseudobinary metal carbide
metal nitride mixture. It follows that the residual nitrogen pressure in any
system which is to contain a refractory carbide of a metal that also forms a
stable nitride at high temperature must be controlled, or at least monitored,
at a very low level.
The same consideration applies to oxygen in the gas phase since solid
solutions of oxide in carbide can also behave according to Raoult's law. The
measurements of oxygen solubility i n UC (85) were made i n the temperature
range 1200-1375C by means of the technique evolved by Komarek ( 13).
In the frst experiments, the carbide samples and liquid uranium were con
tained in CaO crucibles and equilibrated with controlled calcium pressures
in order to establish the thermodynamics of dissolution of oxygen at the
U-UC"Ol_" boundary. The composition of the oxycarbide was diferent when
U02 and U02-UC2 mixtures were brought to equilibrium with carbide samples
in which the uranium activity was less than unity. The oxygen pressures used
throughout this study were between 10-35 and 10-
2
5 atm.
Potter (86) has also studied the Pu C"Ol_" phase in the temperature range
1300-1500C and fnds that the li miting composition is approximately
PUCO
.
2700.67 in the presence of plutonium metal and PuCO.UOO.37 in the
presence of PU
2
C3 These two sets of results make it quite clear that the
properties of the ternary system metal/carbon/nonmetal not only depend
on the chemical potentials of oxygen and the nonmetal in the coexisting gas
phase, but are also directly related to the metal activity.
As Grieveson found with the Ti-C-N system, a further complication in the
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
236 ALCOCK
behavior of many of the refractory carbides and nitrides is the wide range of
compositions representing departure from stoichiometry, TiC"N1_", over
which a single phase can exist. Thus, at 1000C, this phase can contain any
where between 30 and 50 atom percent carbon plus nitrogen. This means that
results for the pseudobinary stoichiometric mixture can be obtained only
when the metal , carbon, and nitrogen activities are all known or when the
activity of one component is known over all the phase feld and a ternary
Gibbs-Duhem is carried out to obtain the values of activities at the pseudo
binary.
In the light of these studies, i t can be readily appreciated that work with
solids whose chemical composition is well defned can be a very arduous task,
and the paucity of measurements at low temperatures probably refects
some of these problems. At high temperatures, where vaporization is sig
nifcant, the measurement of activities of both components has aided the
study of refractory carbides. With the precision achieved with mass spectro
metric studies, it is now feasible to gather valuable information at tempera
tures approaching 2000e. The problems associated with the measurements
of single activities with the spectrometer-such as an estimation of the
ionization cross section of each species measured-can be obviated by the
use of Belton and Fruehan' s adaptation of the Gibbs-Duhem expression (87).
The existence of a stable gaseous dicarbide which is the analog of the gaseous
monoxide is also a valuable aid in the determination of free energies of
formation of carbides. De Maria and co-workers (88
-
90) and Cuthbert,
Faircloth, Flowers & Pummery (91) obtained thermodynamic i nformation
for alkaline and rare-earth dicarbides from vapor pressure studies. The
alkaline-earth compounds vaporized principally to give the metal vapor and
solid carbon, whilst the rare-earth carbides showed a partial pressure of
MCz(g) which varied from system to system.
Storms (92) has obtained the activities of carbon in the U-C system by
comparing the VCz/V partial pressure ratio at a given VIC ratio with that
in equilibrium with the UC2 +C mixture at the same temperature.
Because of the relative i nstability of methane, the studies of CHz/Hz
metal carbide equilibria which could be used to establish carbide thermo
dynamics are few and restricted to temperatures below 1000e. Recent
studies by Alekseev et al (93) on the Ti-C system are i n quite good agreement
with those of Grieveson (81), thus showing the reliability of the method. The
shortcoming of this procedure is, of course, that the carbon activity can only
be measured down to about lO-z with any reliability at 1000C, since
pH
j
PH
2
2
in equilibrium with carbon is 10
-
2 At lower temperatures, when
this ratio increases, the achievement of equilibrium takes a few days, and
mi nor gas leaks in the gas-handling system can be an important source of
error.
The use of a mixture of CO and CO2 to defne a carbon potential in the
gas phase has been of limited use in the study of carbides, and Kleykamp (94)
has concluded from recent emf studies that the coexisting oxygen potential
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 237
defned by a COC02 mixture can lead to incorrect results even when the
metal has a relatively low oxygen afnity. The afnity of chromium for
oxygen is less than those of the Group IV and V elements, which form re
fractory carbides, but the earlier measurements of the stabilities of CraC2 and
Cr7Ca by means of CO/C02 equilibria produced thermodynamic values for
these carbides which were signifcantly different from those of Kleykamp.
He used the cells
Pt/Cr, CrFdCaF2/CrF2, Cr3C2, C/Pt
and
Pt/Cr, CrFdCaFdCrFz, CrCg, CrgC2/Pt
i n order to measure the chromium activities i n the CraC2, C and Cr7Ca,
CraC2 mixtures. His results for the free energies of formation of these phases
IGO - 7200 - 8. 0T cal mole-1 (880-1 1 100K)
AGO 23, 750 - 8. 5 T eal mole
-1
(920-10800K)
can be compared with Richardson's assessment (95) of the earlier data from
which the following equations can be derived:
IGO - 20, 100 - 2. 7ST cal mole-1
IGo - 41 ,600 - 6. 15T cal mole-1
The diferences in the stabilities of these technologically very important
carbides, which are obtained by gas equilibration and by solid-state electro
chemistry, amounts to 16 kcal mole
-
1 at 10000K for Cr7Ca and to about the
same error for CraC2
The experimental scatter i n the emf measurements appears to be about
10 m V, and this accounts for an uncertainty of 700 cal mole
-
1 i n the mea
surement of the free energy of formation of CrgC2 This is because the elec
trode CrF2, CraC
2
, C has the carbide and carbon at unit activity and hence
IGo RT In aCr
3
and from the cell measurements 2FE RT In aC
r'
However, Worrell & Chipman (96) studied the CO pressures over
V20a-C- (VO, VC) and obtained from their results a two-term equation for
the formation of stoichiometric VC
V + C -VC; IGO 22, 200 + 1 . S T eal mole-1
The solid solution (VC, VO) was assumed to conform to Raoult's law, and
from Mah & Kelley's (97) data for the V20a of vanadium oxide, the vanadium
activity in the three solid-phase mixture can be obtained and, since carbon
is present as a pure phase, the free energy of formation of the monocarbide
can be calculated. The temperature-independent term of this equation is in
very good agreement with the standard heat of formation of the monocarbide
-24,400 2000 cal mole-I, which was obtained by Mah (98) calorime
trically.
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
238 ALCOCK
The experiments with niobium and tantal um oxides reacting with carbon
to form NbC and TaC were much simpler to interpret since the solubilities
of oxygen in the carbide phases under these conditions were negligibly small
Nb02 + 3 C -- NbC + 2CO
Ta205 + 7 C -- 2TaC + SCO
I n the latter cases, the heats of formation, which were calculated by the
second-law procedure, were in good agreement, after correction to room
temperature, with calorimetric heats of formation.
I n the realm of calorimetric studies, the work on the heats of formation of
carbides and nitrides, which have a wide range of non stoichiometry, has
brought the interesting suggestion that the heat of formation per g' mole is a
linear function of composition. For example,
Ti + x/2N
2
-- TiN,(0.S93 x 0.98)
- AHo 1 6. 5 + 62.0x kcal mole-l
These results (Morozova & Khernburg 99) were obtained by oxygen
combustion calorimetry. The samples were analyzed for their oxygen con
tents, and the two extreme samples gave the results TiNo
.
9s0o
.
o7 and
TiNo.6930o.oo9. The efects of these impurity concentrations should not
therefore be too marked. Other similar studies in which the analytical in
formation is not so complete are those of Kornilov and co-workers ( 100) for
the systems NbC" and TaC". The results are again given in the simple two
term equation form
-AHO(NbC") 18. 19 + 14.00x kcal mole
-1
(0. 739 x 0. 913)
-A(TaCx) 14. 13 + 19. 73x kcal ' mole
-
1
(0.733 x 0. 961)
Huber et al (101, 102) report compositional dependence for their studies of
NbC" and TaC" by the three-term equations
AHO(NbC,) 6.60 - 70. 95x + 30. 75x
2
kcal mole-1
(0.686 x 0.984)
and
A(TaC") 22. 81 - 103. 78x + 46.88x2 kcal mole-1
(0. 72 x 0.998)
as the result of computer analyses, but state that the simpler expression
AHO(TaC.) - 1 1 . 92 - 22. 67x kcal mole
-
1
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
40
3
8
-
36
'
34 f

32 1
C

::
i 30 -

0;
1 28 r
26 f
24
22

04
SOLUTION THERMODYNAMICS
05

!.
TaCx-
I i I
I
0.6 07 0-8 0,9 1,0
C/TQ atom ratio
239
FIGURE 2, The results for the heats of formation of TaCx as a function of composition
(92). The dotted line represents the functional relationship given in the text.
also fts their results for tantalum monocarbide within the experimental error
(Figure 2) .
The connotation of the simple two-term expression for the heat of forma
tion per mole of a nonstoichiometric carbide is that since the heat of mixing
per g
'
atom AHM is given by
(
Xc is the atom fraction of carbon)
and from the standard thermodynamic equation
it follows that if
AHO a + bx then AHc b
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
240 ALCOCK
These results therefore i ndicate that the working model for a nonstoi
choimetric carbide or nitride phase over a wide range of composition should
be that the partial molar heats of solution of the components should be
constant. I t has been previously been mentioned that analyses of the thermo
dynamics of phases with a wide range of composition can be made success
fully by means of the grand partition function
.
The variation of the activity
coeficient of each component includes a partial heat of solution which may
be a function of composition. This functional dependence i s related to va
cancy-vacancy i nteraction energies when a defective compound is con
sidered. Thus, Libowitz ( 103) ascribes the dependence of the thermodynamic
properties of hydrides on composition to vacancy i nteraction energies having
the values 4.3 kcal mole
-
1 for DH3
-
o, 1. 5 kcal mole
-
1 for ZrH2
-
o, and 0. 35
kcal ' mole-1 for PdHx. Hoch ( 104) has concluded that the free energy results
for TaCx could be accounted for without such a vacancy interaction term
and thus that the partial heats of solution are constant, which i s consistent
with the calorimetric studies. The values of a obtained from these simpl e
linear expressions for the calorimetric results have no obvious structural
signifcance, and i t wil l be valuable to see if further studies show any relation
ship between IHo and a.
The studies of refractory silicides and borides at high temperatures have
been almost exclusively carried out employing vapor pressure techniques.
This is the obvious consequence of the lack of suitable gas mixtures which can
defne boron or silicon potentials. The silicides of technical importance are
those of high melting point, and the information on the compounds of
Groups IV, V, and VIE is fairly complete. For most of these silicides, Searcy
( 105) lists heats of formation which have been obtained either by second-law
calculations from vapor pressure measurements or by direct calorimetric
measurement ( 106) . Only recently has an electromotive force technique been
applied to the study of silicides, and i n this study ( 107) the thermodynamic
properties of tantal um silicides were obtained by the use of a solid oxide
electrolytes. Levine and Kolodney studied the oxygen concentration cells
wi th electrodes made by suitable combination of the silicide phases (e. g. ,
Pt/Fe, FeO/ThOr Y20
3
/Si02, TaSi2, Ta5Si
3
/Pt) in the temperature range
900-1 100C.
The difculty associated with this study is that the oxide which is in
equilibrium with the mixed silicide phase changes across the Ta-Si composi
tion range and could also change with temperature. Si02 was shown to be
the oxide i n equilibrium with TasSi3-Ta2Si, and Ta20s with Ta2Si-Ta9Si
1
2
Clearly when the oxides of the two components of the silicide have roughly
the same stability and the heat of formation of the silicide is great, the
changing activities of the components in a given mixture could cause the
stable coexisting oxide phase to change from one element, tantal um, to the
other, silicon, across a particular temperature range. The relatively slow
rates of the solid-state reactions associated with this oxygen transfer could
lead to slowly drifting emf.
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 241
The use of the fuoride electrolyte CaF2 would add extra problems due to
the volatilities of BFa and SiF 9. I t is not surprising then that the only success
ful electrochemical study with borides or silicides has involved a metal which
forms a very stable fuoride. Aronson & Auskern ( l08) measured the free
energies of formation of thori um borides in the temperature range 800-950C
by means of the ceIIs
Th, ThF4/CaF2/ThF4, ThB6, B
/CaF2/ThF4, ThB6, ThB4
/CaFdThF4, ThB, Th(a)
OXIE SYSTEMS
The study of metaIIic oxide single phases, solid solutions, and interoxide
compounds has been a major involvement of high-temperature chemists in
the past decade. The volume of publications and the resulting wealth of
information is much lal-ger than that i n the earlier sections of this article, but
it cannot be said that the state of our knowledge in this area is far in advance
as a result .
. The two most exhaustively studied of the single-oxide systems are un
doubtedly FeOl+X and U02x. These two oxides have major technological
signifcance, and the interests of the steel and nuclear reactor industries have
been weII served in the oxide feld.
The studies of FeOl+x were originated by the classical work of Darken &
Gurry (109) , who used gas-solid equilibration, and one of the most recent is
by Fender & Riley, using solid-state electrochemistry ( 1 10) . The results of a
number of studies between these two temporal and experimental limits
( 1 1 1
-
1 13) appeared to be interpretable in terms of Fe
2
+ vacancies and Fe
3
+
positive holes, or possibly by clusters of interstitial Fea+ ions ( 1 14) . The
ordering of defects however was not contemplated, since the thermodynamic
properties I:H02 and I:S02 appeared to behave uniformly across the whole
phase feld. Fender and Riley' s latest work, however, lends support to the
suggestions of Raccah & Vallet ( 1 15) that there are three regions within the
FeOl+x phase feld which are separated by second-order transitions. These
transitions probably involve the long-range ordering of defects in a char
acteristic manner in each region.
A typical result is that shown for the composition Fe01.116 in which
I:H1/202 takes the values -61 , 020 460, -64,090 5 10, and -66,090 420
cal g atom
-
1 in the three regions. Koch (1 16) made some high-temperature
X-ray studies with single crystals of FeOI+X which gave results consistent
with the formation of a cubic superlattice having ordered vacancies in octa
hedral sites with clusters of 4Fe
3
+ ions on tetrahedral sites. There is some
evidence, then, to support the idea of defect clustering in this system at
relatively high temperatures, up to 1 350C. It appears that though the
change in partial molar heat of solution of oxygen is small going across these
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
242 ALCOCK
ordered structures, the random model of point defects cannot be applied to
FeOl+x
'
The amount of work which must be applied to any metal oxide system to
determine the defect structure appears to have been increased as more de
tailed studies have opened up new problems. The electrochemical studies of
Nb20s-x by Blumenthal, Moser & Whitmore ( 1 17) indicated that the phase
was homogeneous over the concentration range Nb02

4
6
o
to Nb02.496 at
temperatures around 900e. Schafer, Bergner & Gruehn ( l 1 8), however,
concluded from gas-solid equilibrium studies at 1300C that six phases occur
in the composition range Nb02.4
1
7 to Nb02
.
5oo. The annealing times required
in the preparation of these samples were from 15 to 25 hr at 1300C. It thus
appears that the formation of the ordered defect structures could require
treatment at elevated temperatures over a long period of time. Schafer and
co-workers could not analyze their results to give separate partial heats and
entropies of solution since the studies were made at only one temperature,
but they observe that the change in free energy when one of the intermediate
phases was formed from the appropriate mixture of Nb02
.
4
1
7 and Nb02
.
6
QO
had the maximum value of 188 cal mole-
1
of the intermediate phase.
The Ce02-X ( 1 19) phase exists over the range of values of x from 0 to
0.3 at temperatures above 1 100e. At lower temperatures the ordered struc
tures of a series of compounds of general formula Ce,,02n_2 are formed. When
the basic structure is of the fuorite type, in which ordered systems of ox
ygen vacancies are formed, a new system appears for each value of n. The
partial heat of solution of oxygen per g
.
atom IH1(20z
decreases from about
- 1 15 kcal to a minimum at Ce01.6
1
of about -93 kcal and then increases
to -100 kcal at Ce01 . 76 before decreasing again. In the isostructural Pu02
-
X
phase over the same range of composition and in the temperature range 700-
1 140C, the partial heat of solution of oxygen decreases steadily from -120
kcal g atom-
1
at x =0. 02 to -89 kcal at x =0. 35 ( l20) . This range of solid
solutions does not yield a series of compounds analogous to Ce,02n
-
2 on
cooling.
Both partial molar free energies and partial molar heats of solution of
oxygen in the U02 phase have been determined separately between U02
. 000
and U02. 020 ( 121) . The results for the partial free energies have been obtained
by gas
-
solid and emf measurements with reasonable agreement; the results
for the heat (from Gerdanian & Dode 122) were obtained by use of the
Calvet microcalorimeter at 1 100e. The results suggest that between
U02.000 and U02006 the partial molar heat of solution of oxygen changes from
about - 100 to about -30 kcal g. The partial molar entropy of solution of
oxygen changes very rapidly in this region, corresponding to the behavior of
a component in a highly ordered nonstoichiometric system according to the
classical Wagner-Schottky model . The partial heat should remain constant
however i n the dilute solution range, and the calorimetric results are some
what disturbing for this reason.
As a fnal example, Zador and the author (123) found that the dioxide
phase Mo01. 999 to Mo02. 22 showed the behavior expected of a random defect
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 243
model with no appearance of intermediate phases after lengthy anneals i n
the temperature range 800 to 1050C. The partial molar heat of solution of
oxygen changed from a fairly constant value of -65 kcal g atom-1 in the
metal-rich range to -53 kcal g atom
-
1 in the oxygen-rich range.
The reduction of V02, Ti02, MoOa, and WOa leads not to the formation of
any appreciable range of nonstoichiometry, but to the appearance of a num
ber of intermediate compounds ( 124, 125). These phases, now called Magneli
phases by most authors, have compositions which ft into general formulae
such as (V, Ti)n02n-lJ (Mo, W)nOa
n
-
lJ (Mo, W)nOan
-
2. There have been very
few measurements of the two phase equilibria involving these oxides. but
Ackermann & Rauh (126) have studied some of the tungsten oxide phases,
and Katsura & Hasegawa ( 127) have studied equilibria involving V20a,
Va05, V
4
07, V509, V
6
011, and V02. Both studies were above 1000C, and
hence the establishment of the ordering of the defects should have been com
plete.
Katsura and Hasegawa measured the oxygen potentials of the two-phase
systems at 16000K by means of solid oxide electrolytes and obtained a value
of the free-energy change for the reaction which spans the phase felds of the
Magneli phases, Vn02n-1, i.e. ,
.
G
O 1
600 - 18,000 cal
Ackermann and Rauh represented their results, which were obtained by
vapor pressure measurement, for the standard free energies of formation of
the tungsten oxides by simple two-term equations, and these can be given as
W + 1 .4502 - W02 . 90
(W 20068/20)
W + 1 .3602 -t W02. 72
(W
1
8049/1 8)
.GO - 190,000 + 54. 0T cal mole-1
.GO - 182,200 + 52. 3T cal mole-1
At the mean temperature of the study, 1200C, the diference in stability
between these oxides is therefore 3600 cal mole-1 of oxygen.
DEFECT EQUILIBRIUM CONSTANTS
Within the single-phase nonstoichiometric systems, much efort has been
applied to the elucidation of defect interactions by means of the mass-action
law. What is most surprizing is that in a number of cases one value of the
constant covers a wide range of composition for one system and the use of
the term association prevails. It needs nO stressing here that the situation
is in many respects similar to that in dilute aqueous solutions of simple elec
trolytes, in which it is quite unreasonable to use the mass-action law except
under extreme dil ution. The measurable departures from stoichiometry i n
the high-temperature oxide systems constitute much higher concentrations
of defects than those which should be handled in this way. However, it is of
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
244 ALCOCK
i nterest to bring together the successful representations of the experimental
data to see if there is a commonality of behavior.
The transition metal oxides MnO, FeO, CoO, and NiO have all been
studied with respect to nonstoichiometry. The phases all have the NaCI
crystal structure and only show oxygen-excess nonstoichiometry. The ease
of formation of higher-valency cations Ma+ decreases with increasing atomic
number i n this series, and the range of values of x in M01+" is limited by the
formation of the phase M30
4
, except for nickel, in which pressures greater
than one atmosphere of oxygen are required to form the higher oxides.
Manganous oxide was studied by Davies & Richardson ( 128) , who estab
lished the thermodynamics of nonstoichiometry by equilibration of the phase
with CO-C02 mixtures. At temperatures between 1500 and 1650C, the value
of x could be raised to 0.045 at the highest oxygen pressures. The mass
action constant which described the small deviations could be accounted for
by assuming a large intrinsic concentration of positive holes ( 129) and the
higher departure from stoichiometry could be accounted for by means of the
now-standard equation 1/202+2M2+2M3
+
+VM2
+
+0-. The oxygen
pressure dependence of x is thus one sixth power above MnOo
.
olo and one
quarter power below this value.
\\Tustite has been shown to be a complex phase, and the mass-action con
stant data provide very little information which helps to resolve the prob
lem. Swaroop & Wagner ( 130) fnd a one-sixth power dependence of the
vacancy concentration on the oxygen pressure at low defect concentrations;
this changes to one quarter power at higher defect concentrations. It might
be concluded, then, that the defects, the FeH ions, and the vacancies are
fully dissociated at the metal-rich boundary. Brynestad & Flood ( 131) point
out, however, that i n MgO-FeO solid solutions, the one-sixth power law
does not apply, and since the defect concentration is then lower than in
metal-rich FeOI+x, the fully dissociated model should be applicable a for
tiori.
Their alternative suggestion is a "blocking" model based on a fully asso
ciated complex. It is assumed that two Fe3+ ions are always fully associated
with each ferrous ion vacancy, and that each vacancy blocks the nearest
neighbor cation sites from occupation by another vacancy. The mass-action
constant then takes the approximate form
( nFe3+ )3
1
-
_
2nFe2
where ni is the number of ions of the ith kind. This much more complex
expression accounts for the experimental results over the whole composition
range, both i n pure wustite and i n the MgO-FeO solid solutions.
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 245
The exact structure of the complex has not yet been resolved, but the
alternative suggestion of Roth (132), based on a neutron difraction study,
has most experimental support. The equation for the addition of oxygen to
wustite is
Here Fei3+ is an interstitial ferric iron.
Nickel oxide ( 133) and cobaltous oxide ( 134) both show the simple
p
11
6
0
2 dependence of the departure from nonstoichiometry on oxygen pres
sure which would appear to support the simple point-defect model, but
Brynstad and Flood's model could also be valid here. The deviations from
stoichiometry which have been studied are very small, being limited by the
attainment of one atmosphere oxygen dissociation pressure.
I t thus appears that the ease of formation of the higher-valency ion Mn+
and the insertion of this ion into interstitial sites, which correspond to the
structure of the next-higher oxide Ma04, determines the manner in which the
deviation from stoichiometry in the M0
1
+
x phase depends on the oxygen
pressure. I n this connection it is interesting that the ratio of the oxygen pres
sures for the two-phase equilibria MO-Ma0
4
/M-MO have the values 10
+
2
7,
10+2, and 10+11 at 1000C for manganese, iron, and cobalt, respectively. The
phase Nia04 has not yet been studied. Although the ratio is very much larger
for manganese than for iron and cobalt, the oxidation of Mn' to Mn
a
+
only
occurs substantially over 9 orders of magnitude from MnOl. O05 to Mn01. 05
at 1650C, according to Davies and Richardson.
The oxides Ce02, PU02, and U02 all have the fuorite structure, the frst
two representing the highest valency and U4+ the lowest valency of these
cations in the stable oxide systems. The defects associated with the non
stoichiometric ranges of these oxides appear to occur mainly through the
anions, there being anion vacancies in Ce02
-
X and PuOz--: and interstitial
oxygens in U02
+
x (135) . The heat changes for the reactions
2PU203 + O2 - 4PU02
and
are -197 and - 170 kcal respectively, and the two non stoichiometric phases
have very similar properties with respect to range of existence and partial
heats of solution of oxygen, as was shown earlier.
The defect equilibrium constant for PuOz--x depends upon the minus one
sixth power, according to the results of Markin & Rand (120) . Atlas &
Schlehman (136) , who studied this system at a higher temperature, found
the dependence to be of the minus one fourth power in the same range of
nonstoichiometry, The defect formation reaction could be written
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
246
ALCOCK
or
depending on the formation of interstitial metal ions or oxygen vacancies.
The oxygen pressure dependence can then be related to the concentration of
defects and the power will be "determined" in the case of the interstitial
species by the extent to which the electrons are trapped to form Pua+ ions.
The results of Bevan & Kordis (1 19) for Ce02z and of Tetenbaum &
Hunt (137) for UOH, show behavior very similar to Pu02
-
z, the power de
pendence of the defect concentration on the oxygen pressure being about one
ffth. Over the wide range which was covered by Bevan and Kordis, the
change of slope of the log p02/log x plot can be clearly seen.
There is no analogous region for cerium and plutonium oxides correspond
ing to U02
+
" and it must sufce here to remark that Aukrust, Forland &
Hagemark (138) interpreted the results for this oxide by a defect equilibrium
which associates two US ions in a complex with an oxygen interstitial ion
2 UH + 1/2 O2 -2 UH Oi
Willis (135) showed by means of neutron difraction that the structure
was more complex than this, as far as the oxygen interstitial ion is concerned.
I t appears that these results must be interpreted as showing that there are
two types of oxygen ion, 0' at (0. 5, 0. 38, 0. 38) and 0" at (0. 41, 0. 41, 0. 41)
rather than at the (0. 5, 0. 5, 0. 5) position which would normally be antici
pated. These ions occur in equal numbers to one another and an equal num
ber of vacant sites are on the oxygen lattice. Such an arrangement of oxygen
ions is found in the U
4
09 lattice. This evidence, together with the defect
equilibrium constant found by Hagemark, suggests that the 2U. Oi "com
plex" is the progenitor of U
4
09, which is the phase which is next formed on
oxidation of U02.
The oxides V02, Ti02, Nb02, and Mo02 are all of the rutile structure with
slight modifcation in the latter two systems. It was mentioned earlier that
no metal-rich region of nonstoichiometry in V02-
x has yet been observed,
but the other three oxides have been successfully prepared in the metal-rich
composition range. The defect equilibrium constants for the reduction may
be written either as
with the formation of oxygen vacancies, or as
In the latter case, where an interstitial ion is postulated, the charge of the
interstitial ion may be reduced to Ma+ by the trapping of one of the electrons,
or MS
+
ions can be formed on normal lattice sites. In the Magneli phases
which are formed in both V02-: and Ti02z, the shear planes contain MS+
ions with the twelvefold oxygen coordination typical of the hexagonal close
packing of the M203 phases. If it were assumed that the defect structure in
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS
-7o r-
-90
'
.
t
-10
0
-1
1
0
-
1
20 !. g g g g g

g
CPITION X IN M
2-X
247
FIGURE 3. The relationship between chemical potential and atom fraction of oxygen
in Ti02_z, according to a number of separate studies (148) ,
the homogeneous range Ti01
.
99c Ti02
.
000 is the Ti
a
+
ion which takes up a posi
tion corresponding to the oxygen coordination of the shear planes in the series
Tin02n
-
l , the defect equilibrium constant would involve the oxygen partial
pressure to the minus one fourth power. The value obtained from a number
of studies is closer to the minus one ffth power which corresponds to the un
trapped electron and Tii
4
+
interstitial ion.
In the other oxides Nb02-x and Mo02-x in which a signifcant range of non
stoichiometry has been found, the defect equilibrium constant only involves
the minus one half power of the oxygen pressure. It is therefore concluded
that the number of electrons which are released by reduction at this early
stage are small compared with the intrinsic concentration.
The difculties that are inherent in the study of mass-action constants
are well represented by the collection of results shown in Figure 3, which
summarizes the studies made on substoichiometric Ti02
SOLID SOLUTIONS OF OXIDES
The Hume-Rothery rule that solid solution can take place to a great
extent between species whose radii do not difer by more than 15 percent is
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
248 ALCOCK
well exemplifed in the literature. In relation to oxide solid solutions, this
rule holds weJl when the oxides have the composition close to stoichiometry,
but the introduction of nonstoichiometry adds a new chemical dimension
which afects the working of the rule.
The near-stoichiometric solid solutions show a smaJl departure from
Raoult's law and measurements of FeO-MgO in equilibrium with iron have
been made, with excellent agreement, by a number of workers (139-141).
The results of Hahn & Muan ( 142) for the NiO-MgO and NiO-MnO solid
solutions in equilibrium with nickel were made with H2-C02 gas mixtures in
gas-solid equilibration. These results have been confrmed by Seetharaman
& Abraham ( 143) , ( 144) , who used solid oxide galvanic cells. The thermo
dynamic data for these systems do not provide very much information except
when the results suggest phase separation at low temperatures.
The study of the oxidation of solid solutions can provide very interesting
information in connection with solution models, since it is possible to dilute
one species in a proposed complex simply by the addition of a chemically
stable species. Thus, in FeO-MgO solid solutions, it is only the ferrous ion
which can be oxidized, and it is of considerable interest to discover whether
the heat of oxidation of the ferrous iron is the same in the solid solution as in
pure wustite. According to Iyengar & the author (145) the partial heat of
solution of oxygen in these solutions is independent of the Fe/Mg ratio
from 3. 2 to 20 mole percent wustite, and the value is the same, within experi
mental error, as that in pure wustite -67.25 kcal g atom
-
1 to within
5 kcal.
Aronson & Clayton ( 146) showed that the partial heat of solution of oxy
gen in (VI Th
1
-!)02- solid solutions remained constant ( 2.5 kcal g
. atom-I) from y =1 to y=0. 29, and Roberts & Markin ( 147) report that the
partial heat of solution of oxygen in (UIPUl-)02
+
" solutions remains sensibly
constant until the average Pu valency is reduced to 3. 1 .
All of these results suggest that the defect mechanism which operates in
the pure oxide, is also responsible for the oxidation behavior of the solid
solutions.
These studies of nonstoichiometry in oxide systems show yet again that
the sol ution problems of the materials scientist in metallic and ceramic
systems remain largely unresolved. In many cases, it is not the gross thermo
dynamic properties which are now of major signifcance, and if we are to
pursue the business of correlating thermodynamic properties with atomic
models, then the materials which are investigated must be defned with re
spect to chemical composition in the most rigorous way, both before and after
the measurement. The techniques which we use at present, and which give us
thermodynamic results to 200 cal on the average, are still not sufciently
precise since all too frequently the balance of stability hangs on a few calories.
The choice of a preferred atomic model out of a number of suggested models
is becoming increasingly difcult as the art of fashioning new models pro
gresses. When a sufcient number of parameters is available. a number of
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 249
models can be ftted to the same imprecise data, but at least the confict
should lead towards some more defnitive experiment. Recent years have seen
an increasing use of more than one technique in investigating a given system.
Thus, the use of free-energy and calorimetric techniques in tandem seems to
be more readily accepted as the desirable modus operandi. It seems clear,
however, that structural information such as high-temperature X-ray and
neutron difraction data will be required to an increasing extent as model
building is pursued. Because of the radical changes which might be brought
about by the presence of relatively minor impurity levels, it would seem de
sirable that thermochemists working in this feld seek an alliance with their
structural colleagues so that their measurements may be made on equally
well-defned substances in the physical as well as the chemical sense.
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
250 ALCOCK
LITERATURE CITED
1. Kubaschewski, 0., Evans, E. LJ., A
cock, C. B. 1967. Metallurgical
Thermochemistry. London: Perga
mon 4th ed.
2. 1967. Proc. Brit. Ceram. Soc. No. 8
3. Rapp, R. A. Ed. 1970. Physicochemi
cal Measurements in Metals Re
search, Vo!' 4. New York; Inter
science
4. Rizzo, F. E., Bidwell, L. R., Frank,
D. F. 1967. Trans. Met. Soc. AIME
239; 1901
5. Cater, E. D., See Ref. 4, Part I, 21
6. Wriedt, H. A., Darken, L. S. 1965.
Trans. AIME 233; 1 1 1 , 122
7. Lange, K. W., Schenck, H. See Ref. 4,
Part I, 267
8. Markin, T. L., Bones, R. J., Wheeler,
V. J. See Ref. 2, 51
9. Fromm, E., 1970. J. Less-Common
Metals 22 : 139
10. Allen, N. P., Kubaschewski, 0., von
Goldbeck, O. 1951. J. Electrochem.
Soc. 98; 417
11. Kubasehewsk, 0. , Deneh, W. A.
1953. J. Inst. Metals 82; 87
12. Ibid 1955. 84;440
1 3. Komarek, K. L., Silver, M. , 1962.
Thermodynamics of Nuclear Ma
terials, 774. Vienna: IAEA
14. Isenberg, I. , Phys. Rev. 79: 736 1950
15. Wagner, C., 1944. Z. Phys. Cher.
A 193:386
16. Mott, N. F., 1936. Proc. Cambridge
Phil. Soc. 32: 281
1 7. Herzfeld, K. F., Goeppert-Mayer, M.
1934. Z. Phys. Chem. B 26:203
18. Friedel, J. 1952. Phil. Mag. 43: 153
19. Ebisuzaki, Y., O'Keefe, M. 1967.
Progr. Solid-State Chem. 4: 187
20. McQuillan, A. D. 1967. Phase Stabil
ity in Metals and Alloys, 375.
New York: McGraw-Hill
21. Hoare, F. E. 1963. Electronic Struc
ture and Alloy Chemistry of the
Transition Elements, 29. New York:
Interscience
22. Lacher, J. R. 1937. Proc. Roy. Soc.
A 161 ;525
23. Fowler, R., Guggenheim, E. H. 1949.
Statistical Thermodynamics. Cam
bridge Univ. Press
24. Hoch, M. 1964. Trans. AIME 230: 138
25. Mallett, N. W., Koehl, B. G. 1962.
J. Electrochem. Soc. 109:61 1
26. Brodowsky, H. A. 1965. Z. Phys.
Chem. N. F. 44: 129
27. Brodowsky, H. A., Schaller, H. J.
1969. Trans. Met. Soc. AIME 245 :
1015
28. Gallagher, P. T. , Oates, W. A. 1969.
Trans. Met. Soc. AIME 245 : 179
29. Ricca, F., Giorgi, T. A. 1967. J. Phys.
Chem. 71 :3627
. .
30. Ricca, F. See Ref. 29, 3632
31. Pan, S. S. , Webb, F .. J. 1965. Nuc!.
Sci. Eng. 23: 188
32. Darken, L. S., Smith, R. P. 1946.
J. Am. Chem. Soc. 68: 1 1 72
33. Ban-ya, S., Elliott, J. F., Chipman,
J. 1969. Trans. Met. Soc. AIME
245 : 1 199
34. Speiser, R., Spretnak, J. W. 1955.
Trans. A m. Soc. Metals 47 :493
35. McLellan, R. B., Gerrard, T. L. ,
Horowitz, S. J., Sprague, J. A.
1964. Trans. Met. Soc. AIME
239: 528
36. Powers, R. W., Doyle, M. V. 1959.
Trans. Met. Soc. AIME 21 5:655
37. Gibala, R. , Wert, C. A. 1966. Acta
Met. 14: 1095
38. Hultgren, R., Orr, R. L., Anderson,
P. D. , Kelley, K. K. 1963. Selected
Values of Thermodynamic Proper
ties of Metals and Alloys. New
York; Wiley
39. McCabe, C. L., Schadel, H. M., Jr.,
Birchena1l, C. E. 1953. J. Metals
5 : 709
40. Trumbore, F. A., Wallace, W. E. ,
Craig, R. S. 1952. J. Am. Chem.
Soc. 74: 132
41. Blandin, A. , Deplante, J. L. 1963.
Metallic Solid Solutions. New York;
Benjamin
42. Fuchs, K. 1935. Proc. Roy. Soc. 151 :
585
43. Hardy, H. K. 1953. Acta Met. 1 :202
44. Blair, G. R., Downie, D. B. 1970.
Metal Sci. J. 4: 1
45. Orr, R. L. , Argent, B. B. 1965. Trans.
Faraday Soc. 61 : 2126
46. Weibke, F. 1937. Z. Anorg. Chem.
232 :289
47. Samson-Himmelstjerna, H. O. 1936.
Z. Metallk. 28: 197
48. Korber, F. , Oelsen, W. , Mitt, K. W.
1937. Inst. Eisenforsch. 19:207
49. Kleppa, O. J. 1956. J. Phys. Chem.
60:842
50. Kleppa, O. J., King, R. C. 1962. Acta
Met. 10: 1183
51. Friedel, J. 1954. Advan. Phys. 3;446
52. Freedman, J. F., Nowick, A. S. 1958.
Acta Met. 6: 1 76
53. Kubaschewski, O. 1968. Thermody
namics of Nuclear Materials, 685.
Vienna: IAEA
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
SOLUTION THERMODYNAMICS 251
54. Alcock, C. B. , Kubik, A. 1969. Acta
Met. 1 7:437
55. Bidwell, L. R., Rizzo, F. E., Smith,
J. V. 1970. Acta Met. 18: 101 3
56. Goodenough, J. B. 1966. Magnetism
and the Chemical Bond. New York:
Interscience
57. See Ref. 51
58. Kleppa, O. J. 1955. Acta Met. 3:255
59. Crangle, J. 1963. Electronic Structure
and Alloy Chemistry of the Transi
tion Elements, 51. New York: Inter
science
60. Taylor, R. W., Muan, A. 1962. Trans.
Met. Soc. AIME 224:500
61. Schwerdtfeger, K., Muan, A. 1965.
Trans. Met. Soc. AIME 233: 1904
62. Myles, K. M., Darby, ]. B. 1968.
Acta Met. 16:485
63. Landolt, C., Muan, A. 1969. Trans.
Met. Soc. AIME 245 : 791
64. Leach, ]. S. Ll. See Ref. 3, 199
65. Kubaschewski, O. 1959. Nat. Phys.
Lab. Symp. No. 9, Pap. 3C. H. M.
Stationery Ofce
66. Laves, F. 1956. Theory of Alloy Phases,
1 24. Cleveland, Ohio: ASM
67. Pauling, L. 1960. Nature of the Chem
ical Bond. Ithaca, New York:
Cornell Univ. Press. 3rd ed.
68. King, R. C., Kleppa, O. J. 1964. Acta
Met. 12: 87
69. Alcock, C. B. , Grieveson, P. 1961-
1962. J. Inst. Metals 90:304
70. Alcock, C. B., Cornish, ]. B. , Grieve
son, P. 1966. Thermodynamics, 1 :
211. Vienna: IAEA
71. Brewer, L., 1967. Phase Stability in
Metals and Alloys, ed. P. S. Rud
man, J. Stringer, R. I. Jafee, 39.
New York: McGraw-Hill
72. Engel, N. 1949. Kem. Maanedsbl.
30:53, 75
73. Brewer, L. 1967. Acta Met. 15 : 553
74. See Ref. 71, 566-67
75. Robinson, P. M. , Bever, M. B. 1967.
Intermetallic Compounds, ed. J. H.
Westbrook, 38. New York: Wiley
76. Mott, N. F. 1937. Proc. Phys. Soc.
London 49:258
77. Harrison, R. ]. , Paskin, A. 1963.
Metallic Solid Solutions, Paper V,
ed. Friedel, Guinier. New York:
Benjamin
78. Mullikan, R. S. 1934. J. Chem. Phys.
2 : 782
79. Varley, J. H. O. 1954. Phil. Mag. 45 :
887
80. Hillert, M. See Ref. 77, Paper XLVII
81. Grieveson, P. 1967. Proc. Brit. Ceram.
Soc. No. 8, 137.
82. Naoumidis, A., Stocker, H. J. 1968.
Thermodynamics of Nuclear Ma
terials, 287. Vienna: IAEA
83. Sano, T., Katsura, M. , Kai, H. See
Ref. 82, 301
84. Leitnaker, J. M. See Ref. 82, 317
85. Steele, B. C. H. , Javed, N. A. , Alcock,
C. B. 1970. J. Nucl. Mater. 35: 1
86. Potter, P. E. See Ref. 82, 337
87. Belton, G. R. , Fruehan, R. J. 1967.
J. Phys. Chem. 71 :1403
88. Balducci, G., Capalbi, A., DeMaria,
G., Guido, M. 1969. J. Chem.
Phys. 51 : 2871
89. Balducci, G., DeMaria, G. , Guido, M.
See Ref. 88, 2876
90. DeMaria, G. , Balducci, G., Capalbi,
A., Guido, M. 1967. Proc. Brit.
Ceram. Soc. No. 8, 127
91. Cuthbert, ]., Faircloth, R. L. , Flowers,
R. H. , Pummery, F. C. W. See Ref.
90, 155
92. Storms, E. K. 1967. The Refractory
Carbides. New York: Academic
93. Alekseev, V. I. , Panov, A. S. , Fiveiskii,
E. V., Shvartsman, L. A. See Ref.
82, 446
94. Kleykamp, H. 1969. Ber. Bunsenges.
Phys. Chem. 73:354
95. Richardson, F. D. 1953. J. Iron Steel
Inst. 1 75:53
96. Worrell, W. L., Chipman, ]. 1964.
Trans. Met. Soc. AIME 230: 1682
97. Mah, A. D. , Kelley, K. K. 1961. US
Bur. Mines, Rep. No. 5858
98. Mah, A. D. 1963. US Bur. Mines
R.I. 6177
99. Morozova, M. P., Khernburg, M. M.
1966. Zh. Fiz. Khim. 40: 1125
100. Kornilov, A. N. , Zaikin, I. D. ,
Skuratov, S. M. , Shveikin, G. P.
1966. Zh. Fiz. Khim. 40: 1070
101. Huber, E. J., Jr., Head, E. L. , Holley,
C. E., Jr., Storms, E. K., Krikorian,
N. H. 1961. J. Phys. Chem. 65 :
1846
102. Huber, E. J+ , Jr., Head, E. L. , Holley,
C. E., Jr., Bowman, A. L. 1963. J.
Phys. Chem. 67: 793
103. Libowitz, G. G. 1963. Non-Stoichio
metric Compounds, Advan. Chem.
Ser. No. 39 74
104. Hoch, M. , Juenke, E. F., Sjodahl,
L. H. 1968. Thermodynamics of
Nuclear Materials, 497. Vienna:
IAEA
105. Searcy, A. W. 1970. Chemical and
Mechanical Behaviour of Inorganic
Materials, ed. A. W. Searcy, D. V.
Ragone, U. Colombo, 51 . New
York: Interscience
106. Robins, D. A., Jenkins, I. 1955. Acta
Met. 3:598
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
252 ALCOCK
107. Levine, S. R., Kolodney, M. 1969. J.
Electrochem. Soc. 1 16: 1420
108. Aronson, S., Auskern, A. 1966. Ther
modynamics, 165. Vienna: IAEA
109. Darken, L. S., Gurry, R. W. 1945. J.
Am. Chem. Soc. 67: 1398
1 10. Fender, B. E. F., Riley, F. D. 1969.
J. Phys. Chem. Solids 30:793
1 1 1 . Haufe, K., Pfeifer, H. 1953. Z.
Metallk. 44:27
1 12. Rizzo, F., Smith, J. V. 1968. J. Phys.
Chem. 72 :485
1 1 3. Kleman, M. 1965. Mem. Sci. Rev.
Met. 62: 457
1 14. Richardson, F. D. 1948. Discuss.
Faraday Soc. 4: 256
1 15. Raccah, P., Vallet, P. 1961. C. R.
Acad. Sci. 253 :2682
1 16. Koch, F. 1967. PhD thesis. North
western Univ.
1 1 7. Blumenthal, R. N., Moser, J. B.,
Whitmore, D. H. 1965. J. Am.
Ceram. Soc. 48:61 7
1 18. Schafer, H., Bergner, D. , Gruehn, R.
1969. Z. Anorg. Chem. 365 : 31
1 19. Bevan, D. J. M., Kordis, J. 1964. J.
Inorg. Nue!. Chem. 26: 1509
120. Markin, T. L., Rand, M. H. 1966.
Thermodynamics, 1 :145. Vienna:
JAEA
121. IAEA 1965. Thermodynamics and
Transport Properties of UO. Tech.
Rep. Ser. No. 39. Vienna
122. Gerdanian, P., Dode, M. 1968. Ther
modynamics of Nuclear Materials,
41. Vienna: IAEA
123. Zador, S., Alcock, C. B. 1970. J. Chem.
Thermodynam. 2 :9
124. Andersson, S., Jahnberg, L. 1963.
Ark. Kemi 21 :413
125. Hurien, T. 1959. Acta Chem. Scand.
13:365
126. Ackermann, R. J. , Rauh, E. G. 1962.
J. Phys. Chem. 67:2596
127. Katsura, T., Hasegawa, M. 1967.
Bull. Chem. Soc. Jap. 40: 561
128. Davies, M. W., Richardson, F. D.
1959. Trans. Faraday Soc. 55 :604
129. Smyth, D. M. , 1961. J. Phys. Cher.
Solids 19: 167
130. Swaroop, B. , Wagner, J. B. 1967.
Trans. Met. Soc. AIME 239: 1215
131. Brynestad, J. , Flood, H. 1958. Z.
Elektrochem. 62 :953
132. Roth, W. L. 1960. Acta Crystallogr.
13:140
133. Tret'yakov, Vu. D., Rapv, R. A. 1969.
Trans. Met. Soc. AIME 245 : 1235
134. Gleitzer, C. 1962. Bull. Soc. Chim.
Fr. Vol. 75
135. Willis, B. T. M. 1963. Nature 197: 755
136. Atlas, L. M. , Schlehman, G. J. 1966.
Thermodynamics, 2:407. Vienna:
IAEA
137. Tetenbaum, M., Hunt, P. D. , 1968.
J. Chem. Phys. 49:4739
138. Aukrust, E., Fj!rland, T., Hagemark,
K. 1962. Thermodynamics of Nu
clear Materia/s, 713. Vienna: IAEA
139. Hahn, W. C., Muan, A. 1962. Trans.
Met. Soc. AIME 224: 416
140. Schmahl, L. G., Frisch, B., Stock, G.
1961.Arch. Eisenhuettenuw. 32 :297
141. Shashkina, A. V., Gerasimov, Va. 1.
1953. Ser. Fiz. Khim. 27 :399
142. Hahn, W. C., Muan, A. 1961. J. Phys.
Chem. Solids 19:338
143. Seetharaman, S. , Abraham, K. P.
1968. Indian J. Technol. 6: 123
144. Seetharaman, S., Abraham, K. P.
1968. Trans. IMM C 209
145. Alcock, C. B., Iyengar, G. K. 1967.
Proc. Brit. Ceram. Soc. No. 8 219
146. Aronson, S. , Clayton, J. C. 1960. J.
Chem. Phys. 32: 749
147. Roberts, L. E. J., Markin, T. L. 1967.
Proc. Brit. Ceram. Soc. No. 8 201
148. Alcock, C. B. , Zador, S. , Steele,
B. C. H. 1967. Proe. Brit. Ceram.
Soc. No. 8 231
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
CONTENTS
STRUCTURAL CHARACTERIZATION OF MATERIALS BY USE OF ELECTRON
MICROSCOPY AND SPECTROSCOPY, Victor A. Phillips and Eric
Lifshin 1
DEVELOPMENT AND ApPLICATION OF THEORETICAL TECHNIQUES TO
PROBLEMS IN MATERIALS SCIENCE, J. C. Phillips 93
ANOMALOUS PROPERTIES OF THE VANADIUM OXIDES, John B. Good-
enough
101
THE STRUCTURE OF DEFECTS IN SOLIDS, J. H. Crawford, Jr., and L.
M. Slifkin 139
MODELS OF THE GEOMETRICAL, ELECTRONIC, AND VIBRONIC STRUC-
TURES OF IDEAL CLEAN SOLID SURFACES, C. B. Duke 165
SOLID-STATE PHASE TRANSFORMATIONS, C. M. Wayman 185
SOLUTION THERMODYNAMICS IN METALLIC AND CERAMIC SOLID SyS-
TEMS, C. B. Alcock . 219
CRYSTAL GROWTH, R. A. Laudise, J. R. Carruthers, andK. A. Jackson. 253
SOLID THIN FILMS, E. Kay 289
PHYSICS OF STRENGTHENING MECHANISMS IN CRYSTALLINE SOLIDS,
William D. Nix and Ricardo A. Menezes. 313
ELECTRONIC AND OPTICAL PHENOMENA IN SEMICONDUCTORS, M. E.
Jones and R. T. Bate 347
CORROSION SCIENCE, 1970, David A. Vermilyea 373
AUTHOR INDEX
399
SUBJECT INDEX
411
vii
A
n
n
u
.

R
e
v
.

M
a
t
e
r
.

S
c
i
.

1
9
7
1
.
1
:
2
1
9
-
2
5
2
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
a
n
n
u
a
l
r
e
v
i
e
w
s
.
o
r
g
b
y

M
i
d
d
l
e

E
a
s
t

T
e
c
h
n
i
c
a
l

U
n
i
v
e
r
s
i
t
y

o
n

0
2
/
1
3
/
1
2
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
Quick links to online content
Further
ANNUAL
REVIEWS

S-ar putea să vă placă și