26 Kenneth S.

Pitzer

ground state is minimum. Thus the optical excitation of
the solvated electron from the ground state to the excited
state i s followed by the rearrangement of the solvent mol-
ecules. In the semicontinuum model12 the optical excita-
tion is followed by the expansion of the cavity. On the
other hand, in bhe configuration coordinate model, the op-
tical excitation is followed by the change of the orienta-
tional polarization from Po to PI (see Figure 4). The
change of the orientational polarization from Po to PI
implies the decrease of the orientational polarization
around the electron (see Figure 2). The decrease of the
orientational polarization implies the deorientation of the
solvent molecules. Thus in the configuration coordinate
model the optical excitation is followed by the deorienta-
tion of the solvent molecules.
In the present calculation the effect of the electronic
polarization is neglected. This effect will be taken into
consideration in our later paper.
Acknowledgments. The authors wish to thank Professor
T. Watanabe and Dr. M. Natori for helpful discussions.

Thermodytriarnics of Electrolytes. I . Theoretical Basis and General Equation

Kenneth S. Pitter
inorganic Materials Research Division of the Lawrence Berkeley Laboratory and Department of Chemistry, University of Caiifornia,
Berkeiey, California 94720 (Received May 72, 7972)
Publication cosfs assisted by The U.S. Atomic Energy Commission

A system of equations for the thermodynamic properties of electrolytes is developed OR the basis of theo-
retical insights from improved analysis of the Debye-Huckel model as well as recently published numeri-
cal calculations for more realistic models. The most important result is the recognition of an ionic
strength dependence of the effect of short-range forces in binary interactions. By modifying the usual
second virial coefficients to include this feature, one obtains a system of equations which are only slight-
ly more complex than those of Guggenheim but yield agreement within experimental error to concentra-
tions of several molal instead of 0.1 M . If one compares instead with the recently proposed equations of
Scatchard, Rush, and Johnson, the present equations are very much simpler for mixed electrolytes (and
somewhat simpler for single electrolytes) yet appear to yield comparable agreement with experimental
results for both single electrolytes and mixtures.

The thermodynamic properties of aqueous electrolytes
have been extensively investigated both experimentally
and theoretically. The monographs of Harned and Owen1
and of Robinson and Stokes2 provide excellent summaries.
While the detailed nature of these solutions is so complex
that an ab inibio quantum-statistical theory is not feasi-
ble, the data appear to relate to few enough independent
parameters 'to make relatively exact semiempirical repre-
sentation possible. It is the present objective to develop
equations which reproduce the measured properties sub-
stantially within experimental accuracy, which are com-
pact and convenient in that only a very few parameters
need be tabulated for each substance and the mathemati-
cal calculations are simple, which have appropriate form
for mixed electrolytes as well as for solutions of a single
solute, and whose parameters have physical meaning as
far as possible.
In 1960 Brewer and the writer3 selected as the best
available system one proposed and applied to dilute solu-
tions by Guggenkeid with modifications suggested by
Scatchard5 for concentrated solutions. While this system
was useful in providing a simple and compact summary of
experimental data, it did not fully satisfy the other de-
sired qualities. Recent theoretical advances of Friedman
and collaborators"? and of Card and Valleaus provide im-
portant insights and support for the greatly improved
semiempirical treatment proposed below. Indeed it is in-
teresting that a key idea can be obtained by introducing
the Debye-Huckel model and distribution function into
modern equations relating such functions to the osmotic
pressure. But first we review the Guggenheim equations.
(1) H. S. Harned and 8. 8.Owen, "The Physical Chemistry of Electro-
lytic Solutions," Reinhold, New York, N. Y., 1950.
(2) R. A. Robinson and R. H, Stokes, "Electrolytic Solutions," Butter-
worths, London, 1959.
(3) K. S. Pitzer and L. Brewer, revised edition of "Thermodynamics" by
G. N. Lewis and M. Randall, McGraw-Hili, New York, N. Y., 1961.
(4) (a) E. A. ,Guggenheim, Phil. Mag., [7] 18, 588 (1035); (b) E. A , .
Guggenheim and J. C. Turgeon, Trans. Faraday Soc., 51, 747
(1955); (c) E. A. Guggenheim, "Applications of Statistical Mechan-
ics, Clarendon Press, Oxford, 1966, pp 166-170; here Guggen-
helm surprisingly abandons the limitation to 0.1 M for validity of his
equations for the difference in ~pfrom a reference salt. A5 is ob-
vious from Figure 1 or the figures and tables of ref 4, this is a
crude approximation.
(5) G. Scatchard. Chem. Rev., 19, 909 (1939); also in W. J. Hamer,
Ed., "The Structure of Electrolytic Solutions." Wilev. New York. N.
Y., 1959, p 9.
(6) J. C. Rasaiah and H. L. Friedman, J. Chem. Phys., 48, 7242 (1968);
50, 3965 (1969).
(7) P. S . Ramanathan and H. L. Friedman, J Chem. Phys.. 54, 1086
(1971).
(8) (a) D: N. Card and J. P. Valleau, J. Chem. Plrys., 52, 6232 (1970);
(b) J. C. Rasaiah, D. N. Card, and J. P. Valleau, J. Chem. Phys.,
56, 248 (1972).

The Journal of Physical Chemistry, Voi. 77, No. 2, 1973

Thermodynamics of Electrolytes
~ ~ ~ ~ e n h ~ ihard ~ -Equations
" S ~ ~ i , c
Since the Guggenheim4 equations have had consider-
able success, it is desirable to recall them at this point
and to discuss the aspects that need improvement. The
equations for osmotic and activity coefficients are
Here the s u m over M and X cover all positive and nega-
tive ions, respt:ctively; A, is the usual Debye-Huckel coef-
ficient with N , Avogadro's number and d, the density of
the solvent; and I ie; a useful parameter which expresses
the distance a t which the electrostatic energy for singly
charged ions in the dielectric just equals thermal energy.
The quantities are constants (at given T and D )
analogous to second virial coefficients which represent the
net effect of various short-range forces between the M and
X ions. It should be noted that Guggenheim followed
Br~nsted'sg princEple of specific interaction and excluded
terms related to short-range forces between ions of like
sign.
We shall not distinguish between direct interactions of
solute species at short distances and changes in solvation
with concentration since bolh phenomena influence the
effective interionic potential of average force and thereby
the second and higher virial coefficients. Likewise the dis-
tinction between molality and concentration will be ig-
nored since tho statistical calculations based upon molec-
ular models are used only to suggest appropriate forms for
empirical use and to provide qualitative understanding.
'The advant,age of the molality in practical calculations,
especially if the temperature varies, is overwhelming.
It is also useful to consider the total excess Gibbs ener-
gy (for a solution containing n, kg of solvent) which is ob-
tained by approprrate integration of the activity or osmot-
ic coefficient.
r(n.) = (3/X3)[in( I -I- - c (x2/2)1
For a single 1-1electrolyte eq 1and 2 reduce to
In y := --tA,ml/2/(1 + m1/2)]+ 2 P ~ x m
(8)
cp _- 1 E= -(A,m1/2/3) u(rnlf2)+ P M x m (9)
Guggenheim and Turgeondb showed that eq 8 and 9 fit-
ted, essentiallly within experimental error, the data for 1-1
electrolytes in water at 0" and at room temperature at
concentrations up to 0.1 M . They acknowledge, however,
and others have showin that substantial discrepancies arise
at higher concentrations. In seeking equations that can be
used at higher concentration one may examine separately
269
0.4
-m
o 0.3
f
B
I
x
2s
a_
-1s 0.2
0 I___
0 0.5 1.0 15
ml/2
Figure 1. Difference in osmotic coefficient of several electro-
lytes from that of RbN03.
the first and second terms on the right side. The first
term in eq 6 must remain a general function of ionic
strength if these equations are to maintain their simplici-
ty and utility for mixed electrolyte solutions. We shall re-
turn later to the question of an improved mathematical
form of this first term. Next we note that, by taking dif-
ferences between the properties of different electrolytes of
the same type and at the same concentration, the first
term cancels and we have (for 1-1 electrolytes)
In Y M ~ -Y In Y M'X' = 2 m ( 0 w y - i ( 3 h f y ) (10)
PWX"- 'FM'X' = m ( @ M y- pMfxf) (11)
These equations (10 and 11) still do not fit the experimen-
tal properties above 0.1 M with constant P's.
Following a suggestion of Scatchard,5 Brewer and the
writer3 considered @ to vary slowly with concentration and
produced thereby a very compact tabulation of the experi-
mental data for pure electrolyte solutions. If the P's are no
longer constants, however, the relationships between eq 1,
2 , and 6 and between 8 and 9 are no longer valid. We shall
return to the derivation of correct relationships corre-
sponding to the variable @'s in a later section but here
point out in Figure 1 the qualitative nature of the depen-
dence of fl on concentration. The notable (and initially
surprising) feature is the substantial change in p at low
concentration in contrast to its relative constancy at high-
er concentration. While the change in @ below 0.1 M is
not large, there is no reason to doubt that the curves
maintain their slope in that region. Thus an improved
theory which adequately accounts for these effects at
higher concentration will presumably also represent an
improvement below 0.1 M .
Recently Scatchardlo and coworkers'j have extended
and elaborated the Guggenheim equations in several
ways. First the Debye-Huckel term in eq E is subdivided
into a series of terms with different coefficients of 1112 in T
corresponding to different distances of closest approach
for the solute components. A p p r o ~ ~ derivatives
ia~~ then
(9) J N Brqnsted, Kgl Dan Vidensk Selsk Mat Fys Medd 4, (4)
(1921), J Amer Chem ~ O C44,877
, (1922), 45,2898 (1923)
(10) G Scatchard, J Amer CkLem Soc , 83,2636 (1961)
(11) G Scatchard, R M R u b , and J S Johnson J Phys C h e m , 74,
3786 (1970)
The Journal of Physical Ghernisrry, Vol. 77, No. 2, 1973
270
yield correct but very complex formulas for the osmotic
and activity coefficients. Secondly, the Br$nsted principle
of specific interaction is abandoned and terms are intro-
duced for the short-range interaction of ions of like sign.
Finally, arrays of third and fourth virial coefficients are
added. On this basis, Lietzke and Stoughton12 were able
to represent accurately the osmotic coefficients of 20 pure
electrolytes; also several systems of mixed electrolytes
have been treated.l[)$J~ This system allows enough terms
to be included to represent experimental data accurately;
but the equakions are very complicated and there seems
little promise OC simple physical interpretation of the pa-
rameters. Consequently, it seems worthwhile to seek sim-
pler equations with fewer and more meaningful parame-
ters.
Also noteworthy is the work of Reilly, Wood, and Rob-
inson13 which ih directed primarily to the process of mix-
ing of pure electrolytes. Their general pattern of equations
is similar in many aspects to that presented hereafter, but
these authors did not apply their equations to the numeri-
cal expression of the properties of single electrolytes and
thus did not deal with one of the principal subjects of this
paper.
Hard-Core Effects in Debye-HCekel Theory
Traditional Debye-Huckel theory of electrolyte solu-
tions1.2 recognizes the distance of closest approach a in
the calculativn 3f the electrostatic energy of the distribu-
tion of ions hut ignores the kinetic effect of the hard core
on the osmotic pressure or other properties. As Kirk-
wood,14 among others noted, this hard core effect cannot
be treated rigorously by the traditional charging process
methods of cdculating free energy. Van Rysselberghe and
Eisenberg15 did add approximate hard-core terms to the
traditional Debye-Hhckel formulas, but the feature of
particular interest was lost in that approximation.
Recently it has been shown6e16 that there are several
equations which relate the intermolecular potential and
the radial distribution function (also called the pair corre-
lation function) to thermodynamic properties. Each equa-
tion will give the same set of thermodynamic results if the
distribution function is exact, but different results arise
from approximate distribution functions. The most conve-
nient equation for a hard-core potential is the so-called
“pressure” equation which yields the pressure of a pure
fluid or the osinotic pressure of a solution.
Where n is the osmotic pressure, c, and c I , are concentra-
tions of species i arid .I, c is the total solute concentration
2 c L ,urJis the interniolecular potential, and gzJ is the radi-
al distribution function. The sums cover all solute species.
We introduce ihe pot entia1
u2/= w, rCa
uLJ= x,zJe2/Dr, r 2 a
and obtain
n - chT = (e2/BD,l
I J
1-
~ c , c l z , z I g1,(r)4nrdr
Where the second term, involving g(a), the radial distri-
bution function at the outer surface of contact, constitutes
The Journal of Physfcal Chemistry. Yo/ 77, No. 2, 7973
Kenneth S. Pitzer
the kinetic effect of the hard core. Note that we have as-
sumed the same core size a for all species of ions but dif-
ferent charges z, and z J in accordance with Debye-Huckel
theory. The usual Debye-Hiickel distance 1 / is~defined
K~ = (4ae2/DkT)~ I Z ~ ~ C , (15)
and the function
qJr) = [z,zJe2/DkT(1+ ~ a ) ] ( e - x ( r - @ ’ / r ) (16)
In the Debye-Huckel derivation the radial distribution
of ions around a given ion is obtained from Boltzmann’s
law with the interaction energy given by the product of
the charge on one icn and the average electrical potential
surrounding the other ion. Summation over the various
ions yields a charge density which, after approximation by
its linear term, is substituted into the Poisson equation.
The solution of the Poisson equation yields an explicit re-
sult for the electrical potential. The quantity qLI(r)de-
fined in eq 16 is the resulting exponent in the Boltzmann
factor for the radial distribution function, which is g L I ( r )
= exp[-q,,(r)I.
The charge density around a given ion t now becomes
In the expanded expression the first term vanishes be-
cause the solution is electrically neutral. Since qcl con-
tains the factor z J , the second or linear term is a sum over
cJz,2 and does not vanish; indeed, it yields the charge dis-
tribution in Debye-Huckel theory. The third term in-
volves a sum over cJz13 which vanishes for symmetrical
electrolytes and is relatively small in other cases. Thus for
symmetrical electrolytes there is no inconsistency with the
linearized Poisson-Boltzmann solution in retaining the
third term, but it has no effect on the charge density and
therefore no effect on the conventional Debye-Huckel re-
sults for thermodynamic properties.
Now, however, we wish to include the kinetic effect of
the hard core by the use of eq 14 which involves the sum
over c,c,g,,(a). In discussion of this sum we again consider
the expanded expression for the radial distribution func-
tion
gJr) = 1 - q L I ( r+
) Y2qZj2(r) ~ I . (17)
for r L a; g = 0 for r < a, of course. In this case the sum
for the first term is finite while that for the second term is
zero. But for the hard-core effect the third term again
yields a finite contribution. Thus for symmetrical electro-
lytes this third term can be included without any incon-
sistency with the Debye-Huckel derivation, and its inclu-
sion should yield an improved approximation for the
hard-core effect. Even for unsymmetrical electrolytes the
effect of the third term 0x1 the charge density is small;
hence the “inconsistency” of its retention is small and it
(134 may be desirable to include its contribution t o the hard
(13b) core-effect.
+
(12) M. H. Lietzke and R . W. Stoughton, J. Phys. Chem., fie, 508
(1962).
(13) (a) P. J. Reilly and R. H. Wood, J. Phys. Chem., 73, 4292 (1969);
(b) P. J. Reiliy, R. H. Wood, and R. A. Robinson, ibid., 75, 1305
(1971).
(14) J. c. Kirkwood, Chem. Rev., 19, 275 (1936); J, C , Kirkwood and J.
C. Poirier, J. Phys. Chem., 58, 591 (1954).
(15) P. Van Rysselberghe and S, Eisenberg, J . Amer. Chem. SOC.,61,
3030 (1939); 62, 451 (1940).
(16) T. L. Hill, “Statistical Mechanics,” McGraw-Hill, New York, N. Y . ,
1956.
Thermodynamics of Electrolytes
Since the tneory is only approximate, however, there is
no reason why. the exponential expressions for g,, should
not be used, if desired, and Card and Valleausa have
found good agreement between such distribution functions
and their Monte Carlo calculations for 1-1 electrolytes up
to about 1Ad.
For our puiposes i t is preferable to use the three-term
expansion of cq 17 since this allows simpler mathematical
expression of the results while still providing a good ap-
proximation whencmr the general Debye-Huckel solution
i s valid. The osmotic coefficient is found, on this basis, to
be
la -K J
p - 1 ~ 3-
ckT I == 24nc(l C Ka>
)] (18)
--’-)
48n ( K4a
I all ions have charges & z this reduces to
Now K = (4nl)ll%zcl’2 and I = e2/DkT as defined earlier.
In eq 18 and 19 the first term on the right arises from the
electrostatic energy and the second term from the hard
core; each has interesting implications.
This result for the electrostatic effects differs from the
traditional Debye-Huchel formula although it reduces to
the same limiting law when ~a can be neglected as com-
pared to unit). Such a difference is not unexpected since
the distribution functiou is approximate except in the
limit of small K. In Figure 2 we compare the alternate
therrnodynaimrc expressions based on the Debye-Huckel
distribution f mctaon with the theoretically exact but
numerically ~ ~ ~ ~ ~ Monte ~ r s Carlo
o m statistical
e treat-
ments of the Game molecular model ( a = 4.25 A, aqueous
solution at ‘25”) The new expression, eq 19, agrees, within
computational. ~ ~ c e r t a ~with n ~ ythe
, Monte Carlo results
o ~t~o I = 0.5 M . Substitution of the
for c o n c e n t i ~ a t ~up
conventional term for the electrostatic effects causes a se-
rious discrepancy. It is also interesting that this theoreti-
cal curve of eq 19 (or the Monte Carlo results) fits the ex-
perimental data for HBr very well.
In view of the approximations of the molecular model,
hOWe~er,we shall regard both expressions for the thermo-
dynamic effect of the charge distribution as candidates for
adoption on the basis of success in fitting experimental
data.
O f greater interest is the hard-core term in eq 18 and
19. The first term within the brackets, 2xa3/3, is indepen-
dent of concentration and was considered by Van Ryssel-
berghe and Eisexzberg.15 The second term within the
brackets, howevei, does depend, through K , on the ionic
strength of thc solution. Furthermore, this expression in
brackets varies with concentration in just the same man-
ner as was shown in Figure 1 with an initially rapid de-
crease from the value at zero concentration to a smaller
arid relatively consta~ntvalue at high concentration.
This decrease in the second virial coefficient with ionic
strength arises from a similar decrease in the average ra-
dial distribution function at contact. Thus it is important
to note that the inore rigorous statistical treatments by
the Monte C‘ario method yields a similar behavior of the
radial ~ ~ s t r i ~ 9lunction
~ t i ~ n at contact. Furthermore, the
aPrisnnthaa and Friedman? on a model
with a soft repulsive potential show a similar decrease
271
-0. I u
0.o 0.5
-p2
I
Figure 2. Osmotic coefficient for the Debye-Huckel model
(a = 4.25 A, aqueous solution at 25’): the solid circles are calcu-
lated values by the Monte Carlo method,6athe solid curve is from
the present theory with eq 19, the dashed curve differs from the
solid curve by substituting the conventional D-H expression for
the electrostatic energy for the corresponding term in eq 19,
and the triangles are experimental values for HBr.
with increase in ionic strength in the average radial distri-
bution function at a comparable radius.
The Debye-Huckel distribution function was also in-
serted in the so-called “compressibility” equation of sta-
tistical mechanics.6 The result is much more complex,
mathematically, although similar in qualitative behavior.
Hence it does not add anything useful for our present pur-
poses.
While the Debye-Huckel treatment does not allow the
consideration of different ionic radii within the same solu-
tion, we can compare the properties of different solutions,
each of a single solute with a particular radius sum. We
consider the difference in osmotic coefficient for symmet-
rical electrolytes with radius sums at and a2 and ion
charge z.
“Fz - cp1 - 2rr 4-
c
Here the first term inside the braces arises from the elec-
trostatic effects and the second and third terms from the
hard-core effects ( i e . , the same order as in eq 19). The
quantity ((p2 - cpl)/c corresponds to the quantity ((p2 -
cpl)/m which was shown in Figure 1 and would be a con-
stant if the Guggenheim equations were valid. It is appar-
ent from eq 20 that it will yield the same qualitative be-
havior as was shown in Figure l; specifically, the second
term is constant, but both the first and third terms de-
crease from a finite value a t zero concentration to very
small values as K increases.
The principal consequence of this section is to suggest
that the properties of electrolyte solutions can be ex-
pressed by an “electrostatic” term plus a virial coefficient
series in which the virial coefficients may be functions of
the ionic strength of the solution. More particularly it is
suggested that the second virial coefficient will vary with
ionic strength in the pattern indicated by Figure 1 or eq
18 and 19. An alternate and probably improved form is
also proposed for the “electrostatic” term.
The suggestion of a second virial coefficient which is a
The Journal of Physical Chemistry, Vo;. 77, No. 2, 1973
272 Kenneth S. Pitzer

function of the ionic $strength is, of course, implicit in the reduce to

much more complex theory of Mayer.17 The preceding
argument leads more simply to this same type of expres-
sion and thereby should help to clarify the reasons for this
ionic strength dependence as well as suggesting simple
mathematical forms. We turn now to the choice of equa-
tions wbich wi 11 provide improved empirical representa-
tion of the properties of electrolytes. While the discussion
up to this point should help make the following choices
plausible, the final adoption of any particular equation
will be based on its empirical effectiveness.

Let 11s now s~t u p a system of equations for the thermo-

dynamic properties of pure or mixed electrolytes in rather
general form We will later choose the exact form for cer- where v = UM 4- ux. The terms for triple interactions with
tain functions ky comparison with experimental data. The all ions of the same sign (pMMMand pxxx) have been
total excess Gihbs energy for a solution containing n, kg dropped in eq 25 and 26 and hereafter since they seem
of solvent and s,,n,, . . . moles of' solute species i,j, . . . is certain to be exceedingly small.
taken to be From eq 25 and 26 it is apparent that the properties of
single electrolytes are determined by the combinations
(2UMUXxMX UM'XMM 4- ux2Axx) and ( U M ~ M M X
uXpMxx). A similar but much more cumbersome examina-
(21) tion of the equations for mixed electrolytes indicates addi-
Mere f(1)is a function of ionic strength (also temperature tional observable combinations of the X and p functions.
and solvent properties) expressing the effect of the long- The following definitions of the more directly observable
range electrostatic forces; f ( I ) may have the form of the quantities appear to be most convenient. Quantities ap-
first term of eq 6 or the similar form which can be ob- pearing only for mixed electrolytes are distinguished by
tained by integration Qf eq 18; XEl(Z) is a function of ionic the use of greek instead of latin letters.
strength with the qualitative behavior indicated by Figure
1 on the second term 011 the right in eq 18 or 19. The effect
of short-range forces between species i and j is, of course,
the basis for A,](l). We also include a term for triple ion
interaction which may be significant at high concentra-
tion but igncre any dependence of p l l R on ionic strength.
We assume that the X and p matrices are symmetric, i e . ,
A,, = Allr etc.
The equation<; for the activity and osmotic coefficients
follow from the eppropriate derivatives of G e X
CMX* = [3/(uMuX)f/2][uMpMMX-b YXfiMXXI (30)
CMXY = 3hCMX" (31)
@MN = h M N - (ZN/2ZM)XMM (ZM,f2ZNIXNN (32)
~ M N '= d 8 ~ ~ / d ~l M= N -
' ( Z N / 2 Z M ) X M M ' ~-
mi (ZM/2ZN)XNNf (33)
$MNX 6pMNX (3ZN/ZM)PMMX (3ZM/ZN)PNNX (34)
Also we define
f " = Y2V - ( f / I ) l (35)
f' = 1hf (36)
where f = dfldl, h?,' L= dX,,/dI, and m, = n,/n,, etc. In these terms the properties of the single electrolyte
combination of terms for the neutral electrolyte M ( U M ) take the simple form
X ( U , ) yields for its activity coefficient
Cp - 1 = IZMZXlf"-b m(2UMUX/U)BMXP+
m2[2(uMUX)3~2/u]CMxP(37)
In y = I z ~ z x l f y+ m ( 2 u ~ u x / u ) B ~-tx ~
T?Z2[[2(VMUX)3'2/U]CMXr (38)
In the case of symmetrical electrolytes the coefficients of
mB and m2C reduce to unity; m is the usual stoichome-
tric molality.
(17) J. E. Mayer, J Chem Phys, 18, 1426 (1959), arid subsequent
In a solution of the single electrolyte MX these equations work includinq E. Meeron, J Chem. Phys., 26, 804 (1957)

The Journal of Physical Chemistry, Vol. 77, No. 2, 1973

ihermodynamic,s 04 Electrolytes 273

The expressions for mixed electrolytes become some- y ) throughout eq 44. Although the formuIas for this case
what cumbersome and we will restrict ourselves a t this are more complex, no new quantities are introduced as
time to a few examples. Consider first a mixture of two compared to those for the set of solutions with common
ions of one sign, M[and N, with a common ion X of oppo- ions MX-NX, MY-NY, MX-MY, and NX-NU. Hence
site sign but with unrestricted magnitudes of charge ZM, the measurement of the set of solutions with common ions
zN, and zX. Electrical neutrality requires that provides all the parameters needed to calculate the prop-
erties of the mixture without a common ion. Reilly and
' n X =z I L M / z X I m M + IzN/ZXlmN Wood13a discuss in greater detail the process of mixing
electrolytes with equations whose terms can be related to
The osmotic coefficient for this solution is those presented here. We turn now to the use of these
cp - 1 -- [ 2 / ( r n ~-E.: m~ 4- equations for the numerical representation of experimen-
tal data.
mX)]{If'+ m M m X B M X ' +
+
NX' -t a M n N ( B M N f IBMN') Single Electrolytes
12'X/ZMf1'2"mX2CMX'' + A comprehensive program is now underway for the se-
lZX/ZNI " 2 T b N m X 2 c N X v f mMmNmX$MNX) (39) lection of optimum functions for f and a,, and the evalua-
tion of constants in B,j as well as Cr, For various electro-
and the ionic s ~ , r e rin
~ this
~ t ~case
~ is lytes. At this time a few initial results will be reported
showing the accuracy of agreement which has already
+ /ZV/zX/)mM + /zNzX/(l +
i %[/Z~Zxj(1 I ZN/ZX/)mNI been obtained. The comprehensive tables of osmotic coef-
If the two electrdyles are symmetric this expression can ficients presented by Robinson and Stokes2 provide the
be readily expressed in terms of the molality m of the experimental data for this initial phase of evaluation.
common ion and the solute mole fraction y of NX. Thus Three functions were considered for f , the electrostatic
for MX-NX mixtures one obtains for the two activity term
coefficients as well as the osmotic coefficient ~DHOQ -A, [P/2/(1 +b P ) ] (45)
(0 - 1 = z2f' + Ti'Z{(L --.Y ) B M $ + yBNX' + fDHc' = -& P'2a(bP'2) (46)
yi?. - Y)(BMN + ~ B M N ' ) ]+
m2{( i - ~ ) C M X+' YCNX' + y(1 - Y)$MNX] (40) f~' =-AP(P2/4){1 + [3/(1 + b11'2Y]\ for b1"' < 2.0
In y M X = Z2p -I- (47a)
nz{h'~x'+ ~ ( B N x-
' BMX' + BMN) + = 1/3A,(21/b)'l3 for bI'" > 2.0 (47b)
r(1 -I :b'fi8MN'I -b m2(@MX' -k y ( C N X ' - CMX' +
YZ$MNX) + Yzu(1 - Y)$MNXI (41) A, = Y3Ay = Y3(2nNodw/1000)1'213'2 (48)

In Y N X -I z2fr + A,, is the usual Debye-Huckel constant for the osmotic

iTL{BNxY"b (1 - Y ) ( B M X ' -BNX' + coefficient and all three forms reduce to the usual limiting
law a t small I. The first form, DMQ for "Debye-
&N) f y(l - y)lBMN'] -$. m2{C N X Y + (1 - y ) ( CMX' - Huckel-osmotic," is that derived above, eq 18 and 19, by
CnxP -$. YZ$MNX) + Y 2 ~ ( 1- Y)$MNX) (42) use of the Debye-Huckel radial distribution function in
the "pressure" equation of statistical mechanics which
This formulation shows clearly the quantities B and $ yields the osmotic pressure directly. The second form,
which mise foi mixtures in contrast to the other terms DHC for "Debye-Huckel-charging," is the familiar result
which can be determined by the properties of the pure obtained by initially calculating the activity coefficient
NIX and NX solutions. It is interesting that 8 and 0' show from a charging process and subsequent conversion by the
different coniposition dependence in eq 41 and 42 but the Gibbs-Duhem equation to the osmotic coefficient. The
same dependence in eq 40. third form is one recommended by Glueckaufl8 which
The mixture MX-NY where there is no common ion goes smoothly from the limiting law into a cube-root de-
yields more complex equations. pendence on I at high concentration. The parameter b is
to be optimized separately for each form.
Two forms were tested for the second virial coefficient
Br' = p ( 0 ) + [P'"/(l + ,P'Z)Z] (49)
Brr*= pi01 - Pi11 exp( - - & / 2 ) (50)
The first form is suggested directly by eq 18 and 19 while
the second is considered because it is an especially simple
form having the same general properties, namely; (a) fi-
nite value a t zero ionic strength, (b) rapid change linear
in 11'2 at low ionic strength, and (c) smosth approach to a
constant value at high ionic strength. Although in princi-
ple all three parameters a , p i * ) , and may be adjusted
for each solute, it was hoped that a: might remain the
same for broad classes of electrolytes and that there would

(18) E. Glueckauf, Proc. Roy. SOC.,Sei. A, 310,444 (1969)

The Journal of Physical Chemistry, Vol. 77, No. 2, 1973

274 Kenneth S. Pitzer

TABLE Parameters tor Thermodynamic Functions for Dilute

Aqueous Electrolytes at
_I---

Electrolyte pc01 @!I I 103

HCI 0.1802 0.2753 0.6

LiCl 0.1575 0.2811 0.8
NaCl 0.0781 0.2659 0.7
KCI 0.0460 0.2186 0.4
CSCl 0.0320 0.0273 1.5
NaN03 0.0059 0.1714 0.7
NH4N03 ---0.0143 0.1045 1.5
RbN03 -0.0663 -0.0623 1.9
Ca(C10.4z 0.5789 2.5883 1.1
MgClz 0.4869 2.1062 3.3
CaC12 0.4162 2.2324 1.6
Na2CrO4 0.1 186 1.8765 2.1
Na2S04 0.0428 1.3491 1.6

ab = 1.20, 08 = 2.00, C'O = 0, range 0-2 M

TABLE I I: Par<amdersfor Thermodynamic Functions for Aqueous
Electrolytes ai 25" from I) to 6 Ma
Electrolyte p") e,+
HCl 0.25503 -0.00059
NaCl 0.26495 4-0.00122
KCl 0.20887 -0.00082
CSCl 0.01336 -0.00049
NaN(33 8.17964 -0.00067
KN03 0.04939 +0.00660
RbN03 -0.01 736 4-0.00528
a b = 1.20 and N = 2.00 throughout.
be a systematic relationship between p ( 0 ) and p(1).In spe-
cial cases, however, the form of I39 can be modified or
terms added without affecting the simple form for most
solutes. For this initial test the third virial coefficient C@
was omitted.
The six possible combinations off' and B,+were tested
systematically with the osmotic coefficient values up to 2
M for the 13 electrolytes listed in Table I, by varying b
and CY independently but holding each the same for all
substances The best general agreement was obtained for
fDHO and B,," with b = 1.2 and CY = 2.0; in this case the
over-all standard deviation was 0.0015. The other combi-
nations of f and * gave standard deviations about
twice as large and not significantliy differing one from an-
other.
Table I shows the values of p ( 0 ) and p(1) for each sub-
stance with f D H o ' o and BII* equation and the standard
deviations for each solute. Since osmotic coefficients are
ordinarily reported to only 0.001, it is apparent that the
agreement it3 .well within experimental uncertainty in
most, if not all, cases.
With this choice of form o f f " and B,+and these values
of b and CY the third virial coefficient C,+ was restored to
eq 37 and data for several electrolytes were fitted up to 6
M. The experimental data are from Robinson and Stokes2
with recent revisions for NaN03 and KN03 from Bezbo-
ruah, Covington, and Robinson.19 The results, given in
Table 11, show excellent agreement within experimental
error. It is also significant that p ( 0 ) and p(1) have not
changed much with the addition of the third virial coeffi-
cient and the extension to more concentrated solutions.
Thus the second virial coefficient expression appears to be
yl/Z
Figure 3. Radial distribution function at hard-core contact as a
function of ionic strength from Monte Carlo
a x 103
1.7
well established even though its physical interpretation in
0.7 terms of molecular quantities is not simple.
0.4 Lietzke and Stoughton12 fitted the osmotic coefficient
1.5 data for several electrolytes to the four constant equation
1.4 comprising the conventional Debye-Wiickel term fDHC "
12 with adjustable distance of closest approach and second,
1.0 third, and fourth virial coefficients, each taken as a con-
stant. While the respective calculations of Lietzke and
Stoughton and the present research are not precisely com-
parable, the new results of Table I with two adjustable
parameters for each substance fit the data up to 2 M as
well or better than the earlier work with four adjustable
constants fitted to data ranging up to concentrations from
3 to 6 M . Likewise the results in Table HI with three ad
justable constants compare favorably with those of the
earlier equation using four adjustable constants.
More definitive results for 1-1 and 1-2 electrolytes as
well as calculations for higher valence types will be pre-
sented in a subsequent paper.
Comparison of the @(Ol and p(1) values in Table I shows
a generally parallel trend but substantial individual varia-
tion. In order to understand the difference between p ( 0 )
and p(1) it is desirable to consider the relative values of
the radial distribution functions at contact for hard cores.
Figure 3 shows g + - ( a ) and g++(a) = g - - ( a ) as a function
of concentration from the Monte Carlo calculations of
Card and Valleauga which were for a 1-1 aqueous electro-
lyte with a = 4.25 A. The results of R a ~ ~ ~ and a ~ h a ~
Friedman7 for soft-core models show the same general
trend but are not as directly interpretable. The curves on
Figure 3 show clearly that the short-range interaction of
unlike charged ions becomes relatively much more impor-
tant at low ionic strength. At high ionic strength the
short-range interactions are weighted approximately 3.5 to
1 for unlike and like charged ions, respectively. This ratio
increases to more than 10 to 1 at low ionic strength. Since
it is the sum of short-range interaction effects in eq 27
and 28 which constitute the B,+ and %I? functions, it is
clear that the principal contribution to @ ( I ) will come
from the short-range interaction of unlike charged ions
(19) C P Bezboruah, A. K Covington, and R A. Robinson, J Cham
Thermodyn , 2,431 (1970).

The Journal of P t i y s i x i i Chemistry, Vol. 77, No. 2, 1973

Thermodynamics of Eiectrolytes
while the nnteractions of like and unlike charged ions will
both contribute to @o) in a proportion given by the mag-
nitudes of g+ - ( u ) , g++(a), and g - - ( a ) at high concentra-
tion. With multiply charged ions the short-range interac-
tion of like-charged ions will be less important, of course,
and may be negligible even for p ( 0 ) .
It is apparent f ~ o mTable I that the ratio of p(1) to p ( 0 )
for 2-1 electrolytes is much greater than for 1-1 solutes.
This is in general accord with eq 18 where the two terms
in the brackets correspond to the two terms in B+‘. In the
limit of zero conceutration the ratio of the second term to.
the first is (L2/2a21 for 1-1 electrolytes and (2P/a2) for 2-1
solutes. Since 1 = 7.1 A and a is frequently near 4 A, ratios
near 1.5 and 6 would be plausible for p ( 1 ) / p ( O )for 1-1 and
2-1 solutes, respectively. While the values in Table I show
too much variability to give much meaning to this com-
parison, nevertheless, it seems likely that this increase in
as compared tlo /3(O) arises from the increased ionic
charge as it affects the last term in eq 18.
Since the electrostatic interaction function f was not
changed for individual solutes, the Bp function also in-
cludes a correction LO f for the effective distance of closest
approach. Eqlation 20 and the accompanying discussion
describe this correcl ion which will contribute to both p ( 0 )
and p(1).If the effective distance of closest approach for a
particular solute is less than that implied in f , this contri-
bution to /Yo) and kY1) will be negative. Hence the small
negative values in Table I are understandable.
In conclusion of this section the chosen forms of‘f and B
are summarized for excess Gibbs energy and activity coef-
ficient as well as osmotic coefficient. The subscripts are
omitted since the alternate forms will not be used further.
f“ -A“ [P”/(l + bP’2)]
p = --A+[ P a/(l -P bW2) + (2/b) In (1C b1112)]
pG” = -.A’ ( 4 1 / b )111(1 + bP’2)
5”c pco, + p e - a P ’ 2
gr = 2p(O) $. (ap(l)/*zn x
[I - exp( - a F ) ( I + ,PI2 - Y2a21)]
B”x -- p(O! i-
(Z@‘l!,/’dI)!r)lb .- exp(-aP2)(1 + aP”)]
The superscript G x indicates the form for the excess
Gibbs energy; the value b = 1.2 has been selected for use
with all substances sit 25” aqueous solution; also the value
a = 2.0 waa found to be satisfactory for the 1-1 and 2-1
electrolytes listed 11 Table I but a may be adjusted for
each substance if desired.
Mixed E ~ ~ ~ ~ ” r ~ ~ y ~ @ ~
The therinodynamx treatment of isopiestic data for
mixed electrolytes lias been simplified considerably by the
use of equations which clearly distinguish the effects of
mixing from those of the pure components. This is further
facilitated by analytical equations representing accurately
the properties of the pure components. Recent papers of
Rush, Scatchard, and others11920 offer excellent illustra-
tions. For mixing solutions of equal ionic strength with a
common ion, e 3 40-42 show clearly that any nonlinear ef-
fect arises only from interactions of ions of the same sign
either as pairs in OMN or in combination with one ion of
the opposite sign in $MNX. The linear terms for the activi-
ty coefficient (but not the osmotic coefficient) also include
BMN and $hmjt as well iaS functions for the pure compo-
nents.
275
0.03
-
u.
I.
2 0
EIN
+
=.
0” -0.03
-0.06
0 I 2 3
rn
Figure 4. Parameter for the difference in short-range interactions
of like charged ions as a function of molality. Data from Harned
and Owen’ for mixtures of HCI with the sait indicated; the sizes
of circles show their estimates of experimental error.
It is desirable to estimate theoretically the expected de-
pendence of OMN on ionic strength. Equation 32 shows
that this term is a difference in second virial coefficients
for ions of the same sign. Figure 3 shows that g + + rises
from a very small value at zero concentration to a value
somewhat larger but still much less than unity at high
ionic strength. Thus we expect BMN to increase from a
smaller value at zero concentration to a relatively con-
stant value at high concentration.
The best test of this picture is the extensive array of
galvanic cell measurements which yield the activity coef-
ficient of HC1 in mixtures with LiCI, NaC1, KC1, and
CsC1. Harned and Owen1 summarize this work and ex-
press the results in terms of coefficients a12 and a21 which
are given in our terminology by the equations (where y is
the solute fraction MC1)
(51)
(52)
In YHCl = In YHCIO - 2.303tu12my (574
In YMCI = In ~ M C I O - 2,303azlm(l - y ) (57b)
(53)
Here -y~clOand YMClo are the activity coefficients of the
(54) pure electrolytes at the molality m.
Comparison of eq 57 with 38, 41, and 42 indicates con-
(55) sistency provided the terms my(1 - ~ ) I @ M andNMm2y(1
’
- Y)$MNX may be neglected. On that basis one obtains
(56)
OH,M + (m/2)$H,M,Cl = -(2.303/2)(0(1~-I- a211 (58)
- BMCI“+ ~ ( C H C -~ CMCL“)
BHC~’ ’
(2.303/2)(a12 - ( ~ 2 1 ) (59)
Figure 4 shows the values of BH,M + (n’b/2)$H,M,Cl ob-
tained from Harned and Owen with the size of the circles
indicating their estimates of experimental error. Since the
second term will be negligible in dilute solutions, we may
interpret the curves below 1 M with respect to BH,M. It is
clear that the general nature of these curves corresponds
to the argument given above; the decrease in 0 at low con-
centration corresponds to the decrease in radial distribu-
tion function at hard-core contact for pairs of ions of like
sign.
In the more concentrated solutions the slope of the
curve for CsCl-HCl indicates a significant value of
$H,cs,CI. For the other curves the corresponding term is
very small.
One can also compare, in accordance with eq 59, our
B” values with the differences (0112 - ,211 from Harned
and Owen. The agreement is satisfactory for the constants
120) Y. C. Wu, R. M. Rush, and G. Scatchard, J Phys. Chem., 72, 4048
(1968); 73, 2047,4433,4434 (1969).
The Journal ofPhysical Chemistry, Vol. 77, No. 2, 7973
276
A NoCl- No NO3
O.O!j 0 KCI -KN03
ONaCI-KCI
0 1 ~----' I i
2 4 6 mixing processes with a common ion. This last equality is
m just the cross-square rule of Young, Wu, and Krawetz,23
Figure 5. Difference in osmotic Coefficient of mixed electro-
lytes From pure components at the same ionic strength. The
curves for NaCI-Kr1O3 and KCI-NaN03 are calculated by eq 61
with parameters from pure electrolytes and mixtures with a com-
mon ion
of Table I (through 2 M ) or those of Table 11.
Since we find BH,IM to vary with ionic strength, 0H,M'
cannot be zero. The resulting nonlinear term in eq 41 is,
however, very small and within experimental error at 0.1
M . This effect might have been detected at higher con-
centrations, especially in HC1-CsC1 solutions, but the ap-
propriate measwements were not made. Hence there is no
conflict hetween our interpretation and available data.
We consider next the change in osmotic coefficient for
mixtures MX-NX with a common ion. From eq 37 and 40
one obtains
A p m = y( I -Y)171(6~N + l0MN' + m$'MNX) (60)
It is of interest to consider a series of systems with the
same pair of ion4 of like sign, M and N, but with different
common ions X, X', etc. In this situation the O M N term
must be the same throLghout but #MNX will differ with X.
We considered the series NaCl-KCI, NaBr-KBr, and
NaNCQ3-KN08 for which osmotic coefficients have been
measured by several investigator~.~S~~1~Zz It is found, how-
ever, that the hromi es differ so slightly from the chlo-
rides that little is gained by their inclusion. The data for
the chlorides and nitrates are shown in the middle of Fig-
ure 5 . Actually there are many points for the system
NaC1-KCl; the two shown define the straight line which
represents all the points quite well.
It is apparent from Figure 5 that the data for systems
with a common ion cain be represented by straight lines.
But there is so iietPe information at concentration below 1
M that one cannot determine whether there is curvature
in that regiop. I n view of the theory and the data on the
HCI-MCI systerns given above one would expect curva-
ture toward zero in p m in the dilute region. Since the
values of [A(p,/my(Z -- y ) ] for mixtures with a common
1011 are small, this unct?rtainty from possible curvature in
the dilute regioc is also small and we can omit the terms
in 9' from osmotic coefficient treatments until such time
as improved ex~ier~merital accuracy justifies their inclu-
sion. On this basis we find BN%,K = -0.012, B(Cl,N03) =
, $N~,IK,CI -0.0018, +(Na,K,NOs) = -0.012,
$4Na,Cl,NIO3) =: $(K C3,NOz) = -0.0060.
The Journal 01 Physical Chemtsrry, Vol. 77, NO. 2, 7973
Kenneth S. Pitaer
These values determined from the mixtures with a com-
mon ion suffice, when combined with values from Table I1
for the pure electrolytes, to calculate A y m for the
mixtures without common ion. The equation for mixing
MX with NY, which follows from (43) and (33) and with
the omission of O', is
[Ap,/my(l - y)] = [ O h m + OXY +
+
( ~ / ~ ) ( + M N x $MNY + $MXY -+ lk~x~Pl+
+
[BMYV B N X V - BMX@- B N Y ' +
m(CMY' c N X @-GMX' CNY')] (61)
It is notable that all terms in the first brackets remain
unchanged if M and N (or X and U) are interchanged
whereas all terms in the second brackets change sign.
Thus the sum of [Apm/my(l - y ) ] for mixing MX--NY
and for NX-MY is just twice the first brackets and is ex-
actly the sum of the four corresponding quantities for the
although it is more commonly applied to heats or Gibbs
energies of mixing.
Figure 5 shows the experimental values of Apm for mix-
ing NaCl-KN03 and KCl-NaN03 together with the
curves calculated from eq 61. The fit for the NaCI-KNO3
system is almost perfect. For the KCl-NaN83 mixtures
the agreement is less perfect but appears to be within ex-
perimental uncertainty. If the data for the mixtures with-
out a common ion had been included in an adjustment of
all parameters, a better fit would doubtless be possible,
but it seems preferable in most cases, and much easier, to
evaluate parameters from the simplest systems to which
they are related.
The agreement shown on Figure 5 suggests that the
cross-square rule is satisfied for this system whereas Bez-
boruah, Covington, and Robinsonls concluded that the
difference probably exceeded experimental error. These
authors did not compare the A p of mixing quantities di-
rectly but rather calculated AG of mixing instead. This
calculation of AG involves the integral of (Aym/m)dm
from zero to the molality of interest. It is evident from
Figure 5 the curves we predict in the very dilute region
are quite different from the lines one might draw through
the points without reference to other information. Appli-
cation of the cross-square condition directly to A(pm av-
oids this difficulty and appears to me to yield agreement
within the uncertainty of experiments. It should be em-
phasized that our present, approach and eq 61 predict
much more than the cross-square rule; they predict the
two individual curves for the systems without a common
ion instead of predicting only their sum.
Although the point24 has been made previously that an
ion pair interaction quantity such as # M r J must be the
same regardless of the ion of opposite sign, most experi-
mental data have been interpreted without considering
this limitation. An exception is the very recent work of
Scatchard, Rush, and Johnson11 where it is shown that
this limitation can be imposed without difficulty for the
various mixed electrolytes formed from NaC1, NazSOa,
MgC12, and MgS04. This point was also recognized by
(21) A. K. Covington, T. ti. Lilley, and R. A. Robinsan, J. Phys. Chem.,
72,2759 (1968).
(22) R. A. Robinson, J. Phys. Chem., 65, 662 (1961);R. M. Rush and
R. A. Robinson, J. Tenn. Acad. Sci., 42. 22 (1968).
(23) T. F. Young, Y. C. Wu, and A. A. Krawetz, Discuss. Faraday Soc.,
24, 37, 77 (1957).
(24) ti. L. Friedman. J. Chem. Phys., 32, 1134 (:960):"ionic Solution
Theory," Interscience, New York. N. Y:. 1962.
Ab lnifio Calculations for Thioformaldehyde 277

Wood and Smith26 in making their very convincing case
that tbe heat of mixing of 1-1 electrolytes with a common
ion arises primarily from differences in the interaction of
pairs of ions of the same sign and consequently disproves
Brfinsted's principle of specific ion interaction.
Summary
Et is shown from an improved statistical treatment
based upon the Debye-Huckel ion distribution function
that the effect of short-range forces between pairs of ions,
i.e., the second virial coefficient, should depend also on
the ionic strength. Numerical calculations on more realis-
tic models confirm this general effect in which second vir-
ial coefficient9 for pure electrolytes decrease rapidly with
increase in ionic strength in the very dilute region yet be-
come nearly constant a t higher concentrations. This pat-
tern of behavior was known experimentally and seemed
puzzling heretofore.
By including the appropriate ionic strength dependence
in the second vjrial. coefficients a relatiely simple system
of equations for the thermodynamic properties of electro-
lytes is developed which appears to represent experimen-
tal behavior to much higher concentration than any previ-
ous system of comparable complexity. Numerical parame-
ters and comparisons with experimental data are present-
ed for the osmotic coefficients of several pure electrolytes,
for the activity coefficients of a series of mixtures HC1-
MCl, and for the osmotic coefficient for all possible
mixtures of sodium and potassium chlorides and nitrates.
A much more extensive treatment of many particular sys-
tems is in progress, and it is expected that a very compact
summary can be given of the extensive array of experi-
mental information on aqueous electrolytes.
Acknowledgment. I wish to thank Mr. Guillermo
Mayorga for carrying out some of the numerical calcula-
tions. This research was sponsored by the U. S. Atomic
Energy Commission.
(25) R. H. Wood and R. W. Smith, d . Phys. Chem., 69, 2974 (1965)

b Initio CallctilPations for the Ground and Low-Lying Triplet States of T h ~ ~ f ~ r ~ a ~ d ~ ~ y d

N. Colin b i r d * and Jerrald R. Swenson

Photochemistry Unit, Department of Chemistry, University of Western Ontario, London. Ontario, Canada (Received June 5, 7972)
Pub!icafion costs assisted by the National Research Council of Canada

The bonding, energetics, and geometry for the ground, 3(n,7r*), and 3(?r,~*) states of thioformaldehyde,
H&S, have been investigated by ab initio molecular orbital calculations. The calculated C-S and C-H
bond lengths of 1.631 and 1.090 A, respectively, and the calculated HCH angle of 114.6" are all in good
agreement with the microwave structural results. In contrast to formaldehyde itself, the cr and 7r bonds
are predicted to be polarized in opposite directions in H&S. Carbon-sulfur bond distances and excita-
tion energies are predicted for the 3(n,7r*) and 3(a,a*) states of H2CS by calculations with and without a
3da orbital on sulfur.

Introduction simulated by a linear combination of three Gaussian-type

Th~oform~ldehyde is the smallest thio ketone, a func-
tional group whose photochemistry has recently attracted
considerable iaterest.2 Since the relative stabilities and
the equilibrium geometries for the excited states of a mol-
ecule influence its photochemical behavior, the energies
and C--S bond feaigths for the excited states of thio ke-
tones are of interest. Although semiempirical calculations
on thio ketones have been ~ e p o r t e dno
, ~ study of the ge-
ometry changes that occur upon excitation has been
made. We report here ab initio molecular orbital studies
of the geometry, bonding, and energetics in the ground,
Z(n,r*)?and 3(s,a*) states of thioformaldehyde.
etails of the Calcdations
All wave functions have been determined by the closed-
shell SCF-LCAO-MO method of Roothaan.4 A minimal
basis set of Slater-type orbitals (STO's) was used to ex-
pand the molecular orbitals. Each s and p-type STO was
orbitals according to the STO-3G expansions of Pople and
coworkers.5 All integrals required were evaluated using
subroutines from the POLYATOM system.6 The energies
of the triplet states were evaluated from the ground-state
wave function using Pople's formula7
AEij == C j - ci - J3j
where EL is the eigenvalue of the occupied molecular orbi-
(1) Publicatiorl No. 57 of the Photochemistry Unit.
(2) D. R. Kemp, A. H. Lawrence, C. C. Liao, A. 0 . Loutfy, P. de Mayo,
A. A. Nicholson, and S. Paszyc, Pure Appl. Chem., 367 (197'1).
(3) J. Fabian and A. Mehihorn, Z. Chem., 7 192 (1967); {bid.. 9 , 271
(1969).
(4) C. C. J. Roothaan, Rev. Mod. Phys., 23, 69 (1951).
(5) (a) W. J. Hehre, R. F. Stewart, and J. A. Pople, .I. Chem. Phys..
51, 2657 (1969); ( b ) W. J. Hehre, R. Ditchfieid, A. F. Stewart, and
J. A. Pople, ibid., 52, 2769 (1970).
(6) Available from the Quantum Chemistry Program Exchange. Indiana
University, Bloomington, Indiana, as QCPE 47A.
(7) J. A . Pople, Proc. Phys. SOC..London, Sect. A , 68, 81 (1955).

The Journal of Physical Chemistry, Vol. 77. No. 2, 1979