Solid Solutions of the Alkali Halides

Arthur V. Tobolsky

Citation: The Journal of Chemical Physics 10, 187 (1942); doi: 10.1063/1.1723683
View online: http://dx.doi.org/10.1063/1.1723683
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/10/3?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Temperature Dependence of the Elastic Constants of the Alkali Halides. II. The Solid Solution KCl–KBr
J. Appl. Phys. 38, 446 (1967); 10.1063/1.1709357

Theory of Alkali Halide Solid Solutions

J. Appl. Phys. 33, 2815 (1962); 10.1063/1.1702554

Solid Solutions of the Alkali Halides. II. The Theoretical Calculation of Lattice Constants, Heats of Mixing,
and Distributions between Solid and Aqueous Phases
J. Chem. Phys. 19, 149 (1951); 10.1063/1.1748149

Calculation of the Heats of Formation of Alkali Halide Solid Solutions from Hildebrand's Equation
J. Chem. Phys. 18, 771 (1950); 10.1063/1.1747770

The BornMayer Model for Ionic Solids and the Heats of Formation and Lattice Spacings of Alkali Halide
Solid Solutions
J. Chem. Phys. 17, 1095 (1949); 10.1063/1.1747119

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 MARCH. 1942 JOURNAL OF CHEMICAL PHYSICS VOLUME 10

Solid Solutions of the Alkali Halides

ARTHUR V. TOBOLSKY
Albert Plaut Fellow in Chemistry, Princeton University, Princeton, New Jersey
(Received December 18, 1941)

X-ray studies of solid solutions of alkali halide pairs with a common ion at room temperature
and at 550°C show that these pairs can be divided into three groups depending on the percent
deviation 0 of their lattice parameters. For values of 0 less than 6 percent miscibility is com-
plete at room temperature. For values of 0 between 6 and 13 percent miscibility is complete at
550°C. For values of 0 larger than about 13 percent miscibility is not complete at 550°C. In
terms of the Born theory of ionic lattices, the free energy of mixing, and Vegard's law, an
approximate theory for the phase curve of these solid solutions is derived. For a typical alkali
halide with an electrostatic energy of 180 kcal. at the absolute zero of temperature the equation
of the phase curve is
902 /T(1-2x)+ln x-In (l-x) =0,
where x is the mole fraction of one component dissolved in the other. The temperature above
which miscibility is complete is given by T=4.50 2• The heat of mixing is approximately
-902 cal. The predictions of this theory are in good agreement with the above-mentioned
experimental observations. .

SOLID SOLUTIONS OF THE ALKALI HALIDES
I Tmixtures
has long been known that when certain
of alkali halides with a common ion
are brought into contact solid solution formation
takes place. 1 Only recently,however, has the rate
of attainment of equilibrium for this solid solu-
tion formation been carefully studied. 2 It has
been found that at room temperature if the salt
mixtures are kept dry, it will take a matter of
several weeks for equilibrium to be reached. The
presence of moisture will cut the time to one day.
At high temperatures if the salt mixtures are
fused, quickly quenched, then ground in an agate
mortar and annealed in a Victor Meyer furnace
at 550°C, it will take only a few hours for equi-
librium to be reached.
Continuing the work of Matsen and Beach in
this laboratory, a study was made to determine
what are the stable phases when equimolal
mixtures of alkali halide pairs with a common
ion are brought together at room temperature
and at 550°C. In general, each component is
somewhat soluble in the other and two solid
solution phases occur. The extent of solution is
greater if the lattice parameters are more nearly
equal, and the amount of solubility increases
with the temperature. For certain mixtures for
1 For a general review see Bruni, Chem. Rev. 1, 345-375
(1924).
2 Matsen and Beach, J. Am. Chem. Soc. 63, 3470-3473
(1941) .
which the lattice parameters are nearly equal
miscibility is complete at room temperature and
a single phase will result when equilibrium is
reached. For others, for which the percent dif-
ference between the lattice constants is somewhat
larger two phases result at room temperature but
miscibility is complete at 550°C. For a third class
of these mixtures for which the lattice parameters
are widely different the solubility will be small
and two phases will result for all temperatures
below the melting temperature.
The method used in this study to determine
the equilibrium phases is an x-ray powder
method fully described in the paper by Matsen
and Beach. No attempt was made to determine
quantitatively the composition of the phases,
but merely to determine whether or not mis-
cibility was complete. In most cases where it
was found that two phases were formed, these
phases were found to be nearly pure phases on
the criterion that their lattice constants were not
much different from the lattice parameters of the
pure phases. A similar study had previously been
made by the x-ray method for some of these
mixtures at room temperature. 3 Attempts to
determine the width of the miscibility gap of the
phase diagram had also been made by other
methods. l The various results have been collected
3 Havighurst, Mack, and Blake, J. Am. Chem. Soc. 47,
pp. 29~43.

187

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 188 .~ RT H UR V. TO B 0 L SKY

TABLE I.

Alkali halide'
pairs Equilibrium conditions: this study Remarks from other studies

NaF-KF 15.4 15°C Two nearly pure phases*

550°C Two nearly pure phases
NaCI-KCI 11.5 15°C Two nearly pure phases H.M.B.:t Solution of a few percent KCI in NaC!
550°C Complete miscibility Bruni:t Mixed crystals formed from melt destroyed
at room temperature
NaBr-KBr 10.8 15°C Two phases Bruni: Mixed crystals formed from melt destroyed
550°C Complete miscibility at room temperature
NaI-KI 9.1 15°C Two phases H.M.B.: Very low solubility at room temperature
5500C Completely miscible Bruni: Mixed crystals formed from melt destroyed
at room temperature
NaF-NaI 39.8 15°C Two nearly pure phases
550°C Two nearly pure phases
XaCI-NaBr 15°C Complete miscibility Studied by many investigators
550°C Complete miscibility
NaBr-NaI 8.8 15°C The x-ray lines could not
be interpreted
550°C Complete miscibility
KF-KCI 17.8 15°C Two nearly pure phases
550°C Two nearly pure phases
KF-KBr 23.5 15°C Two nearly pure phases
550°C Two nearly pure phases
KF-KI 32.3 550°C Believed to be two nearly
pure phases. KF lines do
not show very well
KCI-KBr 4.8 15°C Completely miscible
550°C Completely miscible
KCI-KI 12.3 15°C Slight dissolving of KCI Bruni: At room temperature the miscibility gap
in KI extends from 7-90%. Near the melting tempera-
550°C More extensive dissolving ture the miscibility gap is between 49 and 91 %
ing of KCI in KI but
not completely miscible
KBr-KI 7.1 15°C Two phases H.M.B.: KBr soluble to the extent of 15 or 20% in
550°C Completely miscible KI at room temperature
Bruni: Miscibility gap between 34 and 93%

* By this phrase we mean that the x-ray picture shows two phases that differ very slightly from the pure components.
t Reference 3. Havighurst, Mack, and Blake.
tReference 1. Within the approximations of the theory to be developed lOO(a,-al)/al =100(a'-at)/a, =J()()(a,-al)/!(al+a,)/2] =.1, so
that it does not matter which particular definition we adopt for purposes of tabulation.

in Table I. The quantity 0 that appears in this
table is the percent deviation between the lattice
constants al and a2 of the component salts.
It will be seen, as had been noted previously,3
that for salt mixtures for which the value of 0
is less than 5 or 6 percent miscibility is complete
at room temperature. Working under the super-
vision of Dr. Beach in this laboratory I have
made further x-ray studies at high temperatures
and noted that for values of 0 between approx-
imately 6 percent and 13 percent miscibility is
partial at room temperature and complete at
550°C (i.e. only one phase is observed in the
x-ray powder photograph). For larger values of 0
two phases are always observed and especially
at room temperature the phases are very nearly
the original pure components.
In order to explain these phenomena, I have
developed a theory for the phase diagram of
these salt mixtures in terms of the Born theory
for ionic lattices, Vegard's law, and the free
energy of mixing. It is commonly accepted that
in a solid solution such as KCi-KBr the Ci- ions
substitute for Br- ions at random in the KBr
lattice. Superstructure lines which characterize
ordered lattices such as occur in many alloys
have never been observed for mixed crystals of
alkali halides. One must therefore assume that
if there is any ordering phenomena the Curie
point for order-disorder occurs far below room
temperature. In this theory it has been assumed
that the substitution is perfectly random at all
temperatures, i.e. whatever mixed crystals occur
are perfect mixed crystals.
A general expression for the la ttice parameter of
a solid solution in terms of the parameters of the
components has been suggested to be of the form
+
an =al n (az n -aln)x,

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 SOLID SOLUTIONS OF THE ALKA.LI HALIDES 189

where x is the mole fraction of component 2 in
the solid solution and n is a constant to be em-
pirically determined.
The special case where n = 1 is known as
Vegard's law. This law has been found to hold
true for a large number of solid solutions. Values
of n = 3 and n = 8 have also been proposed. The
la tter value of n = 8 has been proposed as a resul t
of a theoretical study by Grimm and Herzfeld
but has not been verified experimentally. The
previously mentioned studyof Havighurst, Mack,
and Blake has shown that for values of 0 less
than 6 percent solid solutions of the alkali
halides at room temperature obey Vegard's law
rather closely. It is, however, hard to distinguish
experimen tall y between the val ues n = 3 and
n= 1 for these particular solid solutions. For
simplicity in the theoretical derivation of the
phase diagram, I have assumed the validity of
Vegard's law.
By far the largest portion of the energy of an
alkali halide is given by the Born evaluation of
the electrostatic energy of the crystal lattice
which is
is in general endothermic, i.e., their formation
absorbs heat. This is in accordance with the fact
that solubility in the solid state increases with
temperature. Also it is experimentally observed
that the heat changes are greatest in the pairs
KCI-NaCl, KBr-NaBr, KI-Nal whose mixed
crystals are formed at high temperatures but are
unstable at room temperature. Then comes KCI
KI which has a large miscibility gap at room
temperature, but at high temperature has a
small miscibility gap. The heat of formation of
the solid solution of KBr-KI is somewhat less;
the miscibility which is complete just below
melting shows at 25°C a gap between 34 and 93
percent. Finally, the smallest heat of formation
within the series is shown by KCI-KBr where
solubility in the solid state is complete at room
temperature. 6
If we assume the validity of Vegard's law, we
can make an approximate calculation of these
heats of formation. The energy U oo of an equi-
TABLE ILt
-Uo - U calculated -U experimental
(Sherman) (Mayer) (Sherman)

LiF 240.1 242.0

where N=Avogadro's law number, e=e1ectronic Li Cl 193.3 200.2 198.1
charge, z=charge on ion, aD = distance from any LiBr 183.1 189.7 189.3
Li I 170.7 176.2 181.1
ion to nearest neighbor, A = Madelung constant.
The term in -l/m represents the repulsive Na F 215.0 214.5
Na Cl 180.4 184.4 182.8
energy of the lattice, where m is approximately 8. Na Br 171.7 176.5 173.3
If we collect all the terms in the numerator of Na I 160.8 166.1 166.4
the above expression and call them 11;[, we have KF 190.4 191.8
Uoo = -M/ao. KCl 164.4 167.5 164.4
K Br 157.8 161.7 156.2
The U oo represents by far the largest contri- KI 149.0 153.1 151.5
bution to the energy of the ionic crystal lattice. RbF 181.8 184.2
Various other contributions such as van der Rb Cl 158.9 162.7 160.5
Rb Br 152.5 156.7 153.3
Waal's energy, polarization energy, etc., range Rb I 144.2 148.4 149.0
from 1 to 5 percent as is seen from the values of
Cs F 172.8 175.8
U o and U for the alkali halides listed in Table II. Cs Cl 148.9 155.6 155.1
U represents the total energy of the lattice 5 (not Cs Br 143.5 150.0 148.6
Cs I 136.1 142.7 145.3
including thermal energy which is given by a
Debye term). t See reference 5.
U 0 is calculated from the Born theory described above using values
The heat of formation of these solid solutions of m that are determined by use of experimental values of the com-
pressibility and generally lie in the vicinity of 9. The lattice parameter
used is that measured at room temperature and therefore U 0 is slightly
4 U oo is the electrostatic energy of the lattice at the different from Uoo because of thermal expansion. Mayer's more accurate
calculations involve the use of an exponential term for the repulsive
absolute zero of temperature. energy and also includes van der \Vaals energy.
"See Seitz, Modern Theory of Solids, Chapter 2. The
more penetrating analysis of Mayer in which an exponential
law for repulsion between the ions is used does not alter the 6 The foregoing remarks on heats of formation of these
results sufficiently to warrant its use in the approximations solid solutions are adapted from reference 1 as are the
here used. experimental data cited in Table III.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 190 ARTHUR V. TOBOLSKY

molal solid solution is given by have compiled the experimental and the calcu-
lated values for the heats of formation of some
Uoo= -M/[(al+a2)/2J.7
of these solid solutions. It is only in the. cases
Let us take for example the reaction KCI+KBr where Vegard's law is obeyed and the miscibility
= (solid solution KCI, KBr). (al and a2 are here is complete, i.e., cases where /) is smaller than
the lattice parameters o(the pure 6 percent that we can expect heats of formation
measured at room temperature to check the
~H=-
2M
+ (M M)
-+- calculated values.
(al +a2) /2 al a2 In order to derive the phase curve for these
solutions we consider the question: Suppose we
components which form the solid solution.)
mix in equimolal proportions two alkali halides.
-4ala2+a2(al+a2)+al(al+a2») of lattice parameters a2 and at, with a2 ~al' If
~H=M we allow the mixture to come to equilibrium the
( (al +a2) (ala2) .
general result will be two phases, a solid solution
of component 1 in component 2, and a solid
------'" solution of component 2 in component 1. It is a
(al+a2) (ala2) 2a 10000
consequence of the assumptions and approxima-
approximately ~H\/>9/)2 since M/a'" 180,000 cal. tions of this theory that the amount of com-
ponent 2 in component 1 is equal to the amount
(a is a kind of average between al and a2).
of component 1 dissolved in component 2. This
\Ve have therefore an approximate evaluation does not alter the general shape of the phase
of the heats of formation of these solid solutions diagram and for simplicity we shall first derive
m terms of the deviation the phase curve on this basis. \Ve shall later
discuss this question more fully from a theoretical
standpoint and compare with experimental ob-
servations.
The total Gibbs free energy for one mole of
In the above derivation we have used the fact alkali halide crystal is U+Ad+pV=F. U is the
that M/a= - Uo is on the average 180,000 electrostatic energy of the crystal. 8 If we neglect
calories for the alkali halides. In Table III we expansion of the crystal, U= Uoo. Ad represents
the Debye expression for the Helmholtz free
TABLE III.
energy. p V is negligibly small below the melting
4R (calculated) 4R (observed) point.
System in calories in calories
Since a large portion of the free energy is com-
KCl-RbCl 4.53 166 203 prised by the term Uoo , and since the Debye
KBr-KCl 4.8 187 220 temperatures and the extra electrostatic energies
NaCl-NaBr 5.5 236
KBr-KI 7.1 389 (390) of expansion of the alkali halides are all of the
NaI-KI 9.1 646 (1250)
NaBr-KBr 10.8 960 (1400) same order of magnitude, we shall assume that
NaCi-KCl 11.5 1188 (2150) insofar as it has any effect on the process of
mixing F= U oo .
The numbers enclosed in parenthesis are unreliable because the
equimolal solid solutions of the corresponding systems are unstable Now let us suppose that we mixed two alkali
at room temperature. Since aH is obtained by measuring heats of
solution in water this necessitates using quenched solid solutions which halide crystals (one mole of each). The free
are of course thermodynamically unstable. energy of the two solid solution phases formed is
7 Actual1y we should here use
F=_(M + M)+2RT{XlnX+ C1-X)ln(1-X»)
1I= l~~/m[~-~(~)m]+~RHRD(~)+PV a' a"
for the heat content of one mole of alkali halide crystal.
But this is not very different from Uoo and we are using where x is the mole fraction of component 1 in
the latter as a good approximation. The heat of reaction the one solid solution phase or the mole fraction
is of course -tl.H. (9R8/8+RD(8/T) represents Debye
energy.) ao represents the lattice parameter of absolute 8 For a complete discussion see Slater, Introduction to
zero, a the lattice parameter at temperature in question. Chemical Physics, Chaps. 23 and 13.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 SOLID SOLUTIONS OF THE ALKALI HALIDES 191

·s

·If
t
y==7'" ·3
W

.J.

-/

.0
.0 .Z- .3 'I .S (" ·7 .1' /0

;t:: -+
FIG. 1.

of component 2 in the other solid solution phase

and where a' is the new lattice parameter of the
first solid solution phase and a" is the lattice
parameter of the second solid solution phase. By
Vegard's rule
a'=al+(aZ- al)x,
+2RT[ln x-In (1-x)]=0. (2)
a" =az+ (al-aZ)x.
Let us put
Using these values a2-al aZ-al
0)=--, 02=--,
al a2
F=
al+(a2-al)x aZ+(al- aZ)x M 01 M 02
0=- ------
+2RTlxlnx+(1-x) In (1-x)}. al (1+01X)Z az (1-02X)2

In order to determine x we wish the free energy +2RT[ln x-In (1-x)]. (3)
to be a minimum. After carrying through a little
Since 01 "'02"'0.05; x:-:::; 1 we can expand by
algebra we can find x as a function of T and 0
binomial theorem.
the percent deviation of the two lattice param-
eters.
aF M(aZ-al) Jv!(al-aZ)
-=0= +----- 2M(a2- a l)2
ax [al+(aZ-al)x]Z [aZ+(al-aZ)x]Z
- x+2RT[ln x-In (1-x)], (4)
+2RT(ln x-In (1-x» =0. (1) a2 3

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16
 192 ARTHUR V. TOBOLSKY

M(a2- a l)2(a2+al) _ 2M(a2-~)2X TABLE V.

° a1 2a2 2 al at Critical mixing

temperature
Critical mixing
temperature

(5) 1 0
4.5 Abs. 10 450.00 Abs.
2 18.0 11 544.5
40.5 12 648.0
2M(a2-al)2 ja 3(1- 2x) "
4
5
72.0
112.5
13
14
760.5
882.0
6 162.0 15 1012.5
+2RT[ln x-In (l-x)]=O. (6) 7 220.5 16 1152.0
8 288.0 17 1300.5
9 364.5 18 1458.0
(a is a kind of average between al and a2.) For 19 1714.5
the alkali halides we have seen that the average
value of Uo o is 180,000 cal. For the purpose of
temperatures the theory IS 111 accord with the
obtaining a general equation suitable for all
experimental data obtained so far. The initial
alkali halides we set
steepness of the curve explains why many mix-
M ja'" 180,000 caL, R= 2 cal./degree, tures completely miscible at high temperatures
give at room temperature an x-ray photograph
0= 100(a2-al)/a. in which the two phases appear nearly pure.
Hence 9152/T(1-2x)+ln x-In (1-x) =0. We let It is a con~equence of the approximations of
this theory that the mole fraction of component
T
Y=---,
90 2
y=1-2x /[In (l~X) l 1 dissolved in component 2 is equal to the whole
fraction of component 2 dissolved in component
1. This can be seen by assuming that a certain
Ymaxirnum = !- amount x of component 1 is in solid solution 1
and a certain amount y of component 2 is in
We can make a table of y against x. See Table IV.
solid solution 1. The composition of the other
The graph (Fig. 1) shows a plot of the mole
solid solution phase (solid solution 2) is then
fraction dissolved against T /90 2• The charac-
determined by the original amounts of the pure
teristic feature of the curve is that it rises very
components. By equating the partial potentials
steeply near the axis of zero concentration and
of each component in the two phases we find
then rounds off to 'reach a maximum beyond
that if we start out with equal amounts of the
which miscibility is complete. The critical mixing
two pure components, x will be equal to y. The
temperature for various values of a is given in
partial free energies are of course obtained by
Table V. It is seen that this theory predicts
setting up the expressions for the total free
complete miscibility at room temperature for
energy of each phase according to the same
values of a less than 8 percent as compared with
approximations used above and differentiating
the value of 5 or 6 percent that would be ex-
with respect to the amount of component in the
pected on the basis of experiment. At higher
phase. The consequence that x=y is not borne
TABLE IV. out by experiment, but the general shape of the
x 0,01 0.02 0.05 0.10 0.20 0.30 0.40 0.50
phase curve should not be radically wrong
Y 0.21750.24730.30040.36400.43290.47230.49380.5000 because of this error.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.240.225.121 On: Mon, 24 Nov 2014 15:33:16