Fkid Phm IUqullibrh, 13 (1983) 269-296 269

Elsevier Science Publtiers B.V., Amsterdam - Printed in The Netherlands

ELECTROLYTE SOLUTIONS

G. MAURER
BASE AG, D-6700 LUDWIGSHAFEN a. Rh. (Fed. Rep. Germany)

ABSTRACT

The correlation and prediction of thermodynamic properties of

liquid mixtures containing electrolyte components needed for phase
equilibrium calculations are discussed, Models to describe aqueous
and nonaqueous solutions with one or more electrolytes dissolved are
reviewed.

INTRODUCTION
The influence ef electrolyte components on phase and chemical
equilibria has $6 be taken into account in the design and the opera-
tion of many s@paratfan processes in chanical engineering. For
example the solubility of a nonelectpelyto gas in an aqueous solu-
tion can be enhanced (salting in) or reduced (salting out) by the
presence of one or more salts. glmilarly there is often a change in
the volatility of a liquid mixture caused by electrolyte compounds.
Therefore salts can be used as extractive agents in areotropic
distillation. An example of the combination of phase equilibria with
chemical reactions is the solubi1it.y of carbon dioxide and ammonia
in aqueous solutions that occurs in sour water systems. The vapor
pressure of a ternary mixture of these components is considerably
lower than would be expected from the solubility of the single solu-
te systems. Liquid-liquid phase equilibria in electrolyte systems
are cammon in ion exchange extraction. Solid-liquid equilibria must
be known for salt purification processes such as solid-liquid ex-
traction and crystallization. Furthermore many chemical reactions in
aqueous phase depend upon the pH of the solution.
The thermodynamics of phase equilibria in electrolyte systems is,
concerned mainly with the properties of liquid phases.It is well
known that we still lack a theerstically sound and easy-to-apply mo-

0378-3812/83/$03.00 0 1983 Elrevier Science Publishers B.V.

de1 for describing the properties of non-electrolyte liquid mixtu-
res. Thermodynamic properties in electrolyte solutions are influen-
ced not only by short-range interactions (mainly between the nearest
neighbors) but also to a considerable extent by long-range interio-
nit forces. Therefore it is not surprising that a complete theoreti-
cal account of the thermodynamic properties of electrolyte solutions
still poses more problems than for nonelectrolyte systems. The field
of electrolyte solutions has received much attention over the last
few decades and much progress has been achieved. Some excellent re-
views of the current state of the fundamental theory of electrolyte
solutions are available in the literature (cf. 1 - 31. They show
that the currently avai 1 able statistical thermodynamic models are
either too simple or too complex for use in chemical engineering.
For representing phase equilibria in electrolyte systems it is
therefore common practice to combine the old electrostatic theory of
Debye and Hiickel or extensions of it with methods approved for non-
electrolyte mixtures.
This contribution relates to the methods now availab?e to the
chemical engineer who has to handle electrolyte solutions.
After a short summary of the basic relationships (standard state,
activity coefficient etc.1 two examples of recent developments in
the theory of electrolyte solutions are discussed. The next chapter
deals with aqueous solutions of strong nonvolatile electrolytes,
then methods to describe the influence of salts on the volatility of
mixed solvents are reviewed. The two remaining chapters deal with
salt effects on the solubil ity of nonelectrolyte and electrolyte ga-
ses in water.

FUNDAMENTAL THERMODYNAMICS
In calculating phase equilibria the chemical potentials ul of all
components present in coexisting phases have to be known. The chemi-
cal potential p of a component i in a liquid solution is generally
expressed as the sum of the chemical potential of i in a reference
state, a contribution from ideal mixing and a correction for real
mixing effects.

A number of reference states and concentration scales are common

for electrolyte solutions. For a none1 ectrolyte sol vent component
the usual reference state is chosen according to Raoult’s law; For
 271

supercritical compounds and ionic species it is usually chosen

according to Henry's law. Concentrations are expressed either on a
mole fraction scale: Xi = lli/ xtlj, on a molal scale: mi=ni/(lOOO g
solvent) or on a molar scale (ci=nf/(lOOO cm3 solution). The activi-
ty coefficient is related to the excess Gibbs energy GE of the
mixture by:

RT In yi = ( aG E/anI ) T ,
p
,
“j*i
(1)

When a strong electrolyte consisting of va anions of valenczy z, and

vc cations of valency zc is dissolved in an aqueous solution, the
procedure implied by equation (1) cannot be performed, because
electrically equivalent amounts of anions and cations are dissolved
simultaneously. This is the reason for defining a mean ionic activi-
ty coefficient for the electrolyte AvaCvc (cf.( 4)):

l/(VlJ’VC)
yt =
[
Y,“q ‘Vc””1 (2)

Another common method of expressing the influence of ionic par-

ticles on the behavior of a solution is the osmotic coefficient. On
a molal concentration scale the osmotic coefficient 0 is defined by

-1000 Inos
Q:_.-
Xvim, Ms (31

where the sum'is over all molecular electrolytes and MS and a, are
the molecular weight and the activity of the solvent S, respective-
ly.

THEORY OF ELECTROLYTE SOLUTIONS

Debye-HDckel model
The modern theory of electrolyte solutions was established by
Debye and Hiickel around 1920 (5). Their model has been extensively
described in many textbooks (4, 6), so only the essential features
are repeated here. Deviations from the properties of the mixture at
infinite dilution are assumed to result only from long-range inter-
ionic forces. The Debye-Hiickel model incorporates three steps to
derive a model for an ionic solution: First an interionic potential
is selected (a hard core potential with an electrostatic contrfbu-
tion superimposed), then the electric energy around an arbitrarily
chosen central ion in the solution is calculated in terms of concen-
trations and valencies of the ions and the properties of the solvent
272

by assuming a spherically symmetric radial distribution function of

the Boltzmann type. In the third step the Gibbs free energy of the
ionic solution is calculated from the electrostatic potential assu-
ming a kind of charging process. The activity coefficient of an
ionic component i on a molar concentation scale then becomes:

Am
Iny :- zi2. (41
ld.6 . d-m

where Zi, d and I, stand for the charge number of ion i, the hard
sphere diameter of the ions and the ionic strength of the solution
(Xc = 0,5XVici) and A and B designate solvent properties which are
mainly influenced by the dielectric behaviour and the temperature.
In very dilute solutions the term B.d.rc in equation (41 becomes
negligible compared to unity and the limiting form of the Debye-
Hiickel ionic activity coefficient on a molar concentration scale is:

(5)

Results for the mean ionic molar activity coefficient Y, from the
Debye-HDckel limiting law agree well with experimental results only
at ionic strength up to about 0.002 molar. At less than about 0.05
ma1 ar, the mean ionic activity coefficient can be described by the
Debye-HBckel law when a number is chosen appropriately for the
hard sphere diameter d. For still higher values of ionic strength,
as is shown in Figure 1, the Oebye-Hiickel law fails, as contrary to
the experimental results, always negative deviations from ideality
are predicted. A great deal of work has been done and is still going
on to improve the theory of electrolyte solutions. Improvements have
been achieved in each of the three steps mentioned.

Extension of the Debye-HDckel model by introducing results from

modern statistical thermodynamics
Rasaiah and Friedman (7) applied recent progress in statistical
thermodynamics to derive more rigorous relationships between the in-
termolecular potential, the radial distribution function and the os-
motic coefficient of an electrolyte solution:

7 r (y ) gij (r) 4lt r2 dr

(6)

0
 273

I I I I

Debye - Htickel - Pitzer /

0
d = 3.9 A

-0.5

-1 .o

-0.5

exp. results (4 >

-2.0 0 LiCI
l AgN03
\
I

.l .2 .5 1 2 3 4 5 6
molal ionic strength

Fig. 1. Comparison between measured and calculated mean activity

coefficientsyk of LiCL and AgN03 in water at 25 'C and
high ionic strength
274

where ci+ is the number density of component i, c+ = 1 Ci+ and i and

j designate ionic species. Pitzer used the hard core electrostatic
energy interionic potential uij of Debye and Hiickel and basically
the same Boltzmann expression for the radial distribution function
CJij (3). He al so expanded the exponentials in much the same way as
it was done in the original theory but retained not only the second
but also the third term in that expansion. This three term form in-
troduced in equation (6) yields the following results for the osmo-
tic coefficient of a 1:l single electrolyte solution:

2Tcd3 n4
Q =‘-~K::(,*Hdl c+ l
3 l 48~ c-2
d
(l+xd)* (7)

where K is the Debye-Hlckel inverse length, which depends on the

ionic strength of the solutfon and the dielectric properties of the
sol vent. In the above model the hard core diameter (distance of
closest approach) is expected to be the sum of contributions from
ion i and ion j: d = di + dj. (As shown in Figure 1, this equation
offers a considerable improvem’ent over the original Debye-Htickel
theory for single solute systems as mean ionic activity coefficients
larger than unity can be predicted. Experimental results can be re-
produced nearly within experimental uncertainty for ionic strength
numbers up to about 5 molar. In spite of the impressive success of
the model for 1 : 1 electrolytes and in spite of the simplicity of
the basic equations, the model is only of limited use because its
extension to multicomponent solutions is neither straightforward nor
approved.

Non-primitive model by Planche and Renon

When the forces between the particles are described using a
“hard-core” and an “electrostatic” contribution, in electrolyte
theory such a model is usually called “primitive”. The most promi-
sing “non-primitive” model seems to be that of Planche and Renon
(8). Planche and Renon take into account interactions between all
kinds of species present in the liquid mixture, i.e. both ionic and
mol ecu1 ar ones. Besides a hard-core and an electrostatic contribu-
tion to the potential energy, short-range interactions are incl u-
ded. The dielectric constant of the solvent is corrected for the
presence of ions. With a -statistical thermodynamic approximation the
radial distribution function is derived and the internal energy as
we1 1 as the Helmhol tz free energy are calculated. The model requires
 215

seven parameters (1 size -, 3 energy - and 3 solvation parameters)

to describe the properties of the pure solvent, one size parameter
for each ionic compound and one energy-parameter for each binary
interaction. Detailed results of this method have been published for
only nine halides showing for the osmotic pressure a very good
agreement with experimental data up to an ionic strength of about 6
molal. For further binary and also ternary systems only a summarized
comparison is available (91, stating an agreement with experimental
results similar to the two-parameter Pitzer equation mentioned
later.

SEMIEMPIRICAL METHODS FOR THE DESCRIPTION OF PHASE EQUILIBRIA IN

AQUEOUS SYSTEMS OF STRONG ELECTROLYTES
As mentioned earlier, theoretically based models are either too
complex or have not been tested as necessary for application in en-
gineering calculations. Therefore semiempirical methods are usually
preferred when phase equilibria in aqueous systems of electrolytes
are considered. Much as for the description of nonelectrolyte sy-
stems, both chemical and physical models are in common use. Conse-
quently both types of model will be discussed in the following chap-
ters.

Solvations models
Chemical models assume ions to be solvated and calculate the ac-
tivity coefficients taking into account the decrease in the number
of particles resulting from that solvation. Many models have been
described in the literature. Only the Robinson-Stokes model (10)
will shortly be discussed here. Solvation reduces the total number
of particles existing independently in the solution. Therefore a
correction to the usual Debye-HDckel expression must be introduced.
In principle, such an equation implies two parameters for a single
electrolyte solute system: the number of solvent molecules in the
solvation shell and the distance of closest approach of the solva-
tes. If both parameters are taken as constants and are determined
from experimental results for the osmotic coefficient, the solvation
model is able to give a good representation of the experimental re-
sults at ionic strength numbers up to about 4 molal. As the solva-
tion number and consequently also the diameter of the solvate de-
crease at high ionic concentrations - as the number of free solvent
molecules decreases - this method fails at high ionic strength.
Furthermore a simple extension to multisolute electrolyte systems
276

would require that a constant solvation number has to be assigned to

a certain ionic compound. Unfortunately, the solvation number of an
ion turns out to depend also on the nature of the companion ions in
the sol uti on. Similar deficiencies are observed in other solvation
models. Recent development aim to combine group contribution methods
with solvation models to describe solvent activities in single as
well as multielectrolyte solutions.

Semiempirical extensions of the Debye-HUckel law

Many attempts to extend the Debye-Hlckel law by adding semiempi-
rical correction terms have been published. The most elaborate and
most successful extension is that of Pitzer (11 - 17). Although
there have been some further developments (18 - 201, a detailed com-
parison (21) recently revealed that Pitzer’s method is still the
most successful for engineering calculations. Therefore mainly
Pitzers’s work and a more recent development by Chen et al. (72) are
discussed here. Additionally, simpler procedures by Funk and Vera
(23 - 25) and Hala (26) are shortly reviewed. Furthermore a predic-
tive procedure by Meissner et al. (27) which is of use in estimating
the properties of electrolyte solutions when no or only scare expe-
rimental results are available, is described.

Pitzer’s model
Pitzer proposed an equation for the excess Gibbs free energy of
an aqueous solution of electrolytes which is a modification of
earl ler proposals by Guggenheim and Scatchard:

1000 GE
- - =L(l) f f mjmk
MW nwRT
l hij mi mj * f f: cl;k mi
(8)

The first term is a modified Debye-Hlckel contribution; The second

term is a correction for short-range binary interactions between
ions; it is a function of ionic strength. The third term corrects
for ternary interactions between ionic species and is assumed to be
independent of the ionic strength.

The virial coefficients ?X and k* are symmetric and include short-

range interactions between ions with charges of both the same and
opposite sign. These coefficients cannot be measured individually
but only in certain combinations. In single electrolyte systems usu-
ally Brdnstedt’s principle for interactions between ions is assumed

( hii = uifi* = 0). This principle may be applied also to multi-so-

lute electrolyte systems when interactions between equally charged
ions are considered. Representation of experimental data can be
further improved by taking differences in binary interactions bet-
ween ions of the same sign into account. For an excellent represen-
tation of experimental data of multicomponent solutions, ternary pa-
rameters which can only be determinded from experimental data on
ternary or multisolute electrolyte systems are needed. For most
engineering applications at ionic strength numbers up to about 6 mo-
lal, third virial coefficients and interactions between ions of the
same sign of charge can however be neglected. Virial coefficients
have been tabulated by Pitzer and coworkers for many electrolytes at
25 'C; extensions to higher temperatures including weak electrolytes
have been published (12 - 17, 28 - 34). In Figure 2 the osmotic
coefficient of an aqueous NaCl-MgS04 solution as calculated with
Pitzer's model is compared by way of experimental results at ionic
strength numbers up to 9 molal, demonstrating the excellent agree-
ment which is generally achieved with Pitzer's method. A very im-
pressive demonstration of the capability of that method has been
published by Harvie and Weare (35) who compared successfully predic-
ted and measured mineral solubilities in water.

Model by Chen, Britt, Boston and Evans

A more recent model which might be able to become as valuable as
Pitzers's correlation has been developped by Chen,. Britt, Boston and
Evans (72). The excess Gibbs energy is expressed as a sum of two
contributions. Long range interionic contributions are taken into
account by means of Pitzer's modification of the Debye-Hlckel law,
whereas short range interactions are incorporated by means of the
local composition concept adapted from the NRTL equation for non-
electrolyte liquid mixtures. The model requires two binary parame-
ters for a singly electrolyte - single solvent system. Parameters for

more than one hundred aqueous systems are published. Extensions for
multicomponent systems, mixed solvents and partially dissociated
electrolytes are being developped.

Funk-Vera Model
When highly concentrated electrolyte solutions are considered, it
may be advisable to, change to models with a different reference sta-
te for the solute. Vapor-liquid equilibria representation at tempe-
ratures where the pure salt is fused indicate that the type of equa-
tions which is well known for nonelectrolyte mixtures is able to
278

1.10

1.05i-

8 1.00
+
.-E
v
‘S
g .95 t-

u
‘E .9c I-

.8E i-

.8CI-
m&,C[ = 0.485

-75i-
0 1 2 3 4 5 6 7 8
molal ionic strength

Fig. 2. Comparison between measured and calculated osmotic coef-

ficients of an aqueous NaCl-MgS04 solution at 25 ‘C
(0 exp. results by Wu et al. (47); -Pitrer’s model 1
 279

describe the excess Gibbs free energy adequately also in binary wa-
ter-salt systems (22). Only in the very dilute region will neglec-
ting the Debye-Htickel term result in unacceptable discrepancies bet-
ween calculated and experimental properties of the salt. Obviously,
over a large concentration range the structure of an ionic solution
is very similar to that of the fused salt and only at very low ionic
concentrations does that structure break down. For many calculations
the pure fused salt or the saturated binary solution might therefore
be a more appropriate reference state for an electrolyte component
than the infinitely diluted solution. A description wherein the sa-
turated solution of the strong electrolyte in the solvent is taken
as a reference state was proposed by Funk (23) and extended by Funk,
Vera and coworkers (24, 25, 36 - 38).

Estimation method of Meissner et al.

As experimental data over a wide concentration range are often
only available at temperatures around 25 'C, estimation methods are
mainly of interest where higher temperatures are involved. Meissner
and coworkers developed a procedure for predicting mean ionic acti-
vity coefficients at 25 'C which was later extended to other tempe-
ratures (27, 39 - 46). Empirically it was found that for single
electrolyte systems there exists a curve family for the term

as a function of the total ionic strength of the solution. Each cur-

ve of that family corresponds to a certain value of q which depends
on the nature of the dissolved electrolyte. From one experimental
data point - at a preferably high ionic strength - the constant q
can be determined and the mean activity coefficient of the electro-
lyte in question at other ionic strength numbers ranging from about
0.1 to 20 molal results from the appropriate curve. While originally
this method was only known as a graphical approach for single elec-
trolyte - water systems at 25 OC, it was later extended to estimate
mean activity coefficients and the activity of water in mixed sy-
stems at high temperatures. Furthermore analytfcal equations for the
curve family were given. By way of example, starting from the osmo-
tic coefficient of a 6 molal aqueous NaCl solution at 25'C (0 = 1.271
(47)) the activity of water in a 10 molal aqueous NaCl solution at
3OO'C is estimated to be 0.69; the experimental results is 0.76 (49).
280

Hala’s method
In phase equilibria including electrolyte components the electro-
lyte is often completely dissociated in one phase, whereas it exists
as’a neutral molecule in the other phase. An example is a vapor-li-
quid equfl i brium in a binary strong electrolyte-water system such as
HCl -H20. In that system the vapor-liquid equilibrium at a temperatu-
re T and a low pressure p can be expressed by
v,,+ + VCI- V,,’ + “,-I-
.prf K, ( T.P)
YHCI %CI’ -Y
(10)

yw.P = f,(T,p) xw Yw (11)

yi and fi are the true mole fraction in the gas phase and the liquid
phase reference fugacity of component i. xw is the true mole frac-
tion of water in the liquid phase while xHClf is a pseudo mole
fraction of HCl defined by
lI(VH+ l VCIT)
/ vH* VCI- \
(12)
;K+e =Jbf,’ y- j
i stand for the true mole fraction in the liquid phase
and the stochiometric number of ionic component i. The acitivfty
coefficient of water,yw, and the mean rational activity coefficient
of HCl Yxl:(defined according to eq (2)) are calculated from the ex-
cess Gibbs energy GE:

( ““+ +vc,-) RT In YHC,‘- = ( aGE/ .hHCl ) T,p,nW

RT in VW = (aG E I an,) T,P,nHC,

For the excess Gibbs free energy Hala introduced two terms

GE/W I (Q,hab) “,
(15)

nt being the total number of moles of (undfssociated) solute and

solvent. (?a incorporates the influence of long-range coulombic for-
ces, whereas Qb results from short-range interactions. Hala proposed
a modified Debye-Hiickel term for Qa and a Porter-type term for Qb.
In a binary system such as HCl-H20 there are four parameters which
all depend on temperature. By way of example, Figure 3 demonstrates
that Hala’s procedure is suitable for correlating vapor -liquid
equilibria in aqueous solutions of single strong electrolytes.
Hala’s method can easily be extended to multielectrolyte aqueous
systems.
 1.0 I ’ A’
P
mean deviation in P
0
( Yca[c -Yexp )HC[ : ’ 0.0°2
0.8
0 exp. Haase et al. (48) /
-cal. Hala ( 26) 0

06
I
0
0
=4”
0.4

0.2

0.15 X,.,C[ 0.20 0.25

Fig. 3. Vapor-liquid equilibrium for HCl/H20 at 25 "C -

experimental results compared with Hala's correlation
282

NONAQUEOUS ELECTROLYTE SYSTEMS

General remarks
Electrolytes dissociate not only in water but also in other polar
solvents, such as ammonia or methanol. In principle, the methods
discussed above are al so sui tab1 e for these el ectrolyte-nonaqueous
solvent systems, as long as the parameters are properly fitted to
the experimental results. In practice, however, very little is known
about the application of these methods to nonaqueous solvents. Still
less is known where the solvent is a multicomponent mixture. In mul-
ticomponent solvents the dielectric properties of the solvent as
well as binary and ternary interaction parameters depend on the com-
position of the solvent. From a theoretical point of view the
Planche-Renon model might be a good choice, but only a few results
are available at the moment (81. Among the other methods discussed
above, only results for Hala’s procedure have been pub1 ished. As
salts may be applied as extractive agents in distillation, there is
extensive literature on salt effects in distillation which was re-
viewed a couple of years ago by Furter et al. (50, 511. Furthermore
there were two symposia of the American Chemical Society on that
subject (80 - 81). Therefore only a short review will be given here.

Hal a’s method for mu1 ticomponent solvents

The influence of LiCl on the vapor-liquid equilibrium in the
binary water-methanol system will be considered by way of example.
Starting from equation (151, Hal a used the same principles as dis-
cussed above. He only substituted Porter’s expression by a three-
suffix Margules equation. By applying a similar procedure as dis-
cussed above only one ternary parameter had to be fitted to experi-
mental results. In Figure 4 a comparison between calculated and mea-
sured gas phase compositfons is shown, proving that Hala’s method is
a fine carrel ation instrument.

Other correlation techniques for describing salt effects in vapor-

liquid equilibria for destillation processes
Many techniques for correlating salt effects in vapor-liquid
equilibria have been pub1 ished (501, but none of them has been suc-
cessfully tested as to the prediction of phase equilibria in multi-
component system from binary data alone. Of the mu1 titude of models
based either on physical fnteraction or chemical reaction (salvation
 101 mm Hg

Pexp

4 ia3
HgO - .l .2 .3 .4 .5 .6 .8
CH3 OH moleI mole

O---_-O exp. results Brou 1 et al. (26 1

__ * talc. results
0-

Procedure: T, Liquid composition given

Gas-phase composition calculated
Parameters: 2 binary, 1 ternary
Deviations: Mean deviation of predicted CH30H concentration
f 0,012 mole/mole (25 data POifttS)

Fig. 4. Vapor-liquid equilibrium of H20-CHjOH-LiCl at 6O'C - Comparison between

experimental results and calculated data from Hala's correlation
models) only the work of Furter and coworkers, Rousseau and cowor-
kers and a recently pub1 ished proposal by Schmitt and Vogelpohl will
be discussed. Furter and coworkers (52, 53) use a Setchenow-type
equation to express the influence of a dissolved nonvolatile elec-
trolyte on the relative volatility

a = Yl x2*/(x1* Y2) (16)

of a binary mixture of components 1 and 2:

Toglo ( a s/ a 0) = k3x3 (17)

Subscripts s and o designate salt-containing and salt-free binary

mixtures, yf is the mole fraction of i in the vapor phase, xi* the
mole fraction of i in the liquid phase on a salt-free basis and x3
is the mole fraction of the (undissociated) nonvolatile electrolyte
in the liquid phase. Originally it was hoped that the Setchenow pa-
rameter k3 would not depend strongly on the composition of the bina-
ry solvent mixture, but this has proved to be true only in a few ca-
ses. For rough estimates eq. (17) may nevertheless be used.
Rousseau and coworkers (54 - 56) applied methods well accepted
for the description of liquid phases of nonelectrolyte compounds in
thef r “pseudo-binary approach”. To a binary mixture of components 1
and 2 a salt 3 is added to enhance the volatility of component 1.
Components 2 and 3 are considered to behave 1 ike a pseudo pure com-
ponent 2*. Activity coefficients are correlated with good success by
applying commonly used equations such as van Laar, Wilson, NRTL and
Uniquac. An extension of that method was applied to quaternary mix-
tures, which were reduced to “pseudo-ternary” mixtures (56).
Recently Schmitt and Vogelpohl (57) al so used a “pseudo-binary”
approach for describing vapor-liquid equilibria in a salt-containing
binary sol vent mixture, but additionally they used the .hypothesis
that a salt added to a mixture of solvents complexes with these sol-
vent components. The degree of formatfon of complexes was assumed to
be proportional to the maximum solubilities of the salt in the pure
solvents at the temperature fn question. This method only needs bi-
nary data and yields a good representation of ternary data. The pro-
cedure can be extended to multicomponent solvent mixtures, but no
experience for such an extension is available.
 285

SALT EFFECTS ON THE SOLUBILITY OF SPARSELY SOLUBLE NONELECTROLYTE

COMPONENTS IN WATER
Salting out and salting in
Organic liquid and solid components as well as nonelectrolyte ga-
ses are usually only sparsely soluble in water. These low solubili-
ties are often influenced considerably by the presence of electroly-
te components. For example, if one mole of sodium chloride is added
to one kilogram of water, the solubility of benzene is reduced by
about 30 % and that of nitrogen by about 25 %. As the solubility is
reduced, these salt effects are called "salting out". Especially
when the electrolyte forms large ions as for example the tetraalkyl-
ammonium salts, the solubility may also be increased. This phenome-
non is called "salting in". For example the solubility of methane in
water at 25 "C increases by about 17 % when one mole of (n-CqHg)NBr
is added to one kilogram of water.
Salting-in and salting-out effects are usually correlated by
means of the Setchenow equation
(1B)

where So and S, are the solubilities of component i in the salt-free

and salt-containing solutions on any molecular concentration scale,
cs is the molarity of the salt S and kfls is the salting coefficient
or salting constant of i caused by the electrolyte S.
The salting coefficient depends on the natures of the dissolved
electrolyte and the solute, as well as on temperature and salt con-
centration. Many methods have been proposed to predict the salting
coefficient itself as well as its dependence on salt concentration
and temperature. Some are reviewed in the following chapters.

Methods to predict salting coefficients

When for the solute i the reference state is chosen according to
Henry's law for the solubility of i in the salt-free solvent, it
can be shown that the following relation holds: (S,/S,)i =.y~,~
where ~4,~ is the activity coefficient of solute i in the salt-con-
taining liquid phase. Therefore the salting constant can be calcula-
ted from an appropriate model for the excess Gibbs free energy of
the electrolyte solution. The first procedure was published by Debye
and McAulay in 1925 (SS).Assuming that salt effects on nonelectroly-
te solutes are caused by a change in the dielectric constant of the
solvent which again results from the presence of nonelectrolyte
286

species, and following similar ideas to those put forward in the

fundamental work by Debye and Hiickel (51, the following expression
for the activity coefficients yi,s was derived:

In yp q a? nj*zj’l
Wie* a,,,on,c( dj 1 / (8lt E ED kT)
species
(19)

The influence of the ionic species is expressed by nj*which is the

number of moles of ionic component j in one cm3 of solution, charge
number zj and the size of the ions (dj = radius of ion j 1. The in-
fluence of the solute on the salting coefficient is only expressed
by coefficient Wi which describes the change in the dielectric con-
stant of the solution caused by the solute molecules (c =ew.
(I-wini*)), where cw stands for the dielectric constant of pure wa-
ter and ni* for the number of moles of solute f in one cm3 of solu-
tion, It follows from the electrostatic theory of Debye and McAulay
that the solute/solvent combination should determine the sign of the
salting coefficient, i.e. whether salting-in or sal tfng-out occurs.
This is in contrast to the observed behavior which shows that there
is a non-negligible influence of the nature of the electrolyte. For
example methane is salted out in aqueous ammonium bromide solutions
but salted in in aqueous tetrabutyl ammonium bromide. Nevertheless,
if in eq. (191 wi and dj are properly chosen, salting-out effects
can be predicted with reasonable accurancy. For example, McDevit and
Long (59) argue that, when the ionic radii were taken from crystal-
lographic data and the decrement of the dielectric constant of water
was calculated as the difference between those of benzene and water,
salting coefficients for benzene in water estimated from the Debye-
McAulay theory often do not deviate by more than 20 per cent from
experimental results (cf. Table 11.
Much work has been done to allow for the shortcomings of the
Debye-McAul ay theory, the main efforts being aimed at explaining
salting-in. Literature published until around 1950 was reviewed by
Bockris et al. (60) and Conway et al. (61). In 1952 McDevft and Long
(591 developped an expression which relates the salting constant to
volumetric properties of the dissolved salt and the dissolved non-
electrolyte sol ute. Their main idea was that a dissolved electrolyte
may compress or expand the solvent, resulting in salting-out or sal-
ting-in, respectively. In accord with experimental data in the theo-
ry by McDevft and Long the properties of the salts are able to
 281

change the sign of the salting coefficient:

+,*: .,O ( vg - v,o) / ( Pw RT ‘n (‘0) 1 (21)

Vi0 and vso are the partial molar volumes of solute i and salt S in
infinite dilution in water, vs is the molar volume of the pure salt
as a liquid and Dw is the compressibility of water. The partial mo-
lar volumes in water can be determined from volumetric data, whereas
the molar volume of the liquid salt has to be estimated, as the salt
is usually a solid at the temperature of interest. Salting coeffi-
cients predicted by the McDevft-Long equation are consistently too
high by a factor of 2 to 3 when liquid solutes such as benzene are
considered, but they are of a similar quality to those from the
electrostatic theory of Debye and McAulay, when smaller molecules
are dealt with. However it is worth mentioning that for most nonpo-
lar solutes the theory by McDevit and Long predicts the right order
of salting constants which is often a great help in estimating salt
effects. The observed discrepances between calculated and measured
salting coefficients were attributed by McDevft and Long to diffe-
rences in the size of the ions and solvent and solute molecules.

TABLE 1
Comparison between observed and calculated salting coefficients
in aqueous benzene solutions at 25 'C

T
salting coefficient ks benzene/m01 dm-3
salt obsd. calculated
a b C d e

NaCl .195 -130 .42 .I55 .166 ,105

LfCl .141 .127 .31 .155 .172 .045
NaJ .095 .122 .27 -149 -157 .036
CsCl .088 .121 .26 .145 .149 -.037

a) Debye-McAulay; b) McDevit-Long; c) Conway et al. parameters

from density data; d) Conway et al. parameters from mobility data;
e) Scaled Particle Theory.

Conway, Desnoyers and Smith combined a solvatfon model with the

Debye-McAulay procedure (61). In applyfng that model one faces the
problem that the number of water molecules in the hydration shell
and the diameter of that shell have to be estimated from other phy-
sical properties. Furthermore both numbers are not unique, but de-
pend on the method and physical property used for estimation. Gene-
rally the model by Conway et al. can predict salting-out as well as
salting-in, but provides no improvement over the electrostatic theo-
ry by Debye and McAulay when only salting-out coefficients are con-
sidered.
More modern methods to predict salting coefficients are based on
the Scaled-Particle Theory by Pierotti (621. Similar to the orginal
Debye-HiIckel theory, Pierotti calculated the change in the Gibbs
free energy when a solute is dissolved from two terms: One is for
creating a cavity in the sol vent structure in which a solute molecu-
le is brought. In this first step the solute molecules are conside-
red to behave like hard spheres. The second term results from “char-
gi ng” the hard spheres with intermolecular forces between solute and
sol vent species. Shoor and Gubbins extended the Scaled-Particle
Theory to electrolyte solutions (63). Later on Masterton and Lee
(641 used the Shoor-Gubbins equation to calculate salting coeffi-
cients as the sum of three terms: ki.s = ka + kb + kc. kc is a volu-
metric contribution which needs the partial molar volume of the non-
electrolyte solute in the solvent; kb results from the soft part of
the intermolecular potential; For the nonpolar part of that poten-
tial a Lennard-Jones 6-12 potential with arithmetic mixing rules for
distances and geometric mixing rule for energies is used. Polar con-
tri butions resul ti ng from interactions between the permanent dipole
of a solvent species and a permanent or induced dipole of the solute
1 are also considered. ka resul ts from introducing the hard sphere
solute molecules into the electrolyte solution. It depends on the
concentration of the salt in the solution, the apparent molar volume
of the salt in water at infinite dilution and the diameters of all
species present. When these diameters are taken as the Lennard-Jones
distances, besides pure substance parameters, the only information
needed on mixtures are volumetric properties. In all cases ka is
positive and kb is negative, whereas kc may be posivtive or negati-
ve, but its contribution to the salting coefficient remains small.
Sfmflar to the method of Conway et al. the results of the Scaled
Particle Theory strongly depend on the choice of the numbers for the
parameters. As shown in Table 1, salting-out coefficfents for ben-
zene calculated from the Scaled Particle Theory generally do not
 289

agree much better with experimenta data than those calculated by

other methods.
Onda et al. (65) used a proposa by van Krevelen and Hoftijxer to
predict salt effects. Van Krevelen and Hoftijzer modified the
Setchenow equation by introducing the ionic strength instead of the
molarity of the salt to describe the influence of the concentration
of the electrolyte components:
(21)
lOq~o(So1S,)i = k, s* I

They assume that three effects contribute to the salting coefficient

(k*i,s = ksolute+kcation+kanion) which describe the influence of the
nature of solute, cations and anions. From experimental results the
different contributions were evaluated. The influence of temperature
on the contributions of cations and anions can be neglected; whereas
it must be considered for ksolute. For many solute and ionic species
Onda et al. report the different contributions; their tables may be
used to estimate salting coefficients. By way of example, predicted
and measured solubilities of some gases in water are compared in
Table 2, confirming a fair to good agreement..For multielectrolyte
aqueous systems Onda et. al (66) proposed to extend eq. (21):

log)0 (Sg'Ss)i = X k"j, lj

all (221
electrolytes j

where kT,j is the salting coefficient of solute i in an aqueous so-

lution of single electrolyte and Ij is the ionic strength of that
electrolyte j in the multicomponent solution.
TABLE 2
Salting-out constants for the solubility of nonpolar gases in
aqueous electrolyte systems at 25 'C - Predictions by Onda's
correlation compared with experimental results

k i,s/mole.dm-3 source of
solute obsd. predicted exp. data

Ar .102 -113 Clever, Holland (67)

Ar .143 -084 Clever, Holland (67)

02 .184 .162 Shoor, Gubbins (63)

C2H4 .127 -105 Morrison, Bfllet.(68)
CO2 -017 .022 Onda et al. (66)
C2H4 .271 .298 Onda et al. (66)
290

Influence of temperature and salt concentration on the salting

coefficient
As the salting coefficient is related to the activity coefficient
of the solute the influence of temperature on the salting coeffi-
cient can be described by the temperature-dependence of the activity
coefficient. It is well known from thermodynamics of liquid mixtures
that no adequate model to describe the influence of temperature is
known, even for nonelectrolyte systems. The salting coefficient usu-
ally decreases with temperature, runs through a minimum and increa-
ses again. It was shown by Clever and Ho11 and that the theory of
McDevit and Long is able to describe that curve qualitatively (67).
Clever and Ho1 1 and argue that the influence of temperature on the
salting coefficient mainly results from that on the partial molar
volume of the dissolved salt. Their method has been applied recently
by Mason and Kao (82) and Pawlikowski et al. (731 to estimate the
temperature dependence of salting-out constants.

SOLUBILITY OF WEAK ELECTROLYTE GASES IN AQUEOUS SOLUTIONS

The removal of acid gases such as carbon dioxide and hydrogen
sulfide from effluent gas streams by absorption is a typical example
of a separation process where the solubil ity of weak electrolyte ga-
ses in aqueous solutions has to be considered. Of special interest
are the solubilities of ammonia, carbon dioxide, hydrogen sulfide
and sulfur dioxide in pure and salt-containing water and in aqueous
alkanolamine solutions. The vapor-liquid equilibrium is strongly in-
fluenced by chemical reactions in the liquid phase. Applying the
concept of Henry’s constant Hi for the solubility of NH3 and CO2 and
Raoult’s law for the solvent (water) the following three equations
hold for the vapor-liquid equilibrium:

(H . m . y )i = P (y .rP)i; i = NH3, CO2 (23)

(fo x .Y Iw= rJ (Y.q )w (24)

where mi, yi. Yi and Vi designate molality. gas-phase mole frac-

tion, activity coefficient and gas-phase fugacity coefficient of
mol ecu1 ar component 5. P is the pressure and f,o is the standard
state fugacity of water. The solubility of the gases is influenced
 291

considerably by chemical reactions: the first dissociations of am-

monia and carbon dioxide (NH3 + H20 eNHe+ + OH-, CO2 + H20*HC03-
+ H+), the second dissociation of carbon dioxide (HC03- +CO3'
+ H+), the carbamate reactfon (NH3 + HCO3-+ #H2COO- + H20), and the

dissociation of water (H20 ‘FH+ + OH-). The three equations for va-
por-liquid equilibrium must therefore be supplemented by 5 equations
for chemical equilibria. As an example here only the equations for
the dissociation of ammonia and the carbamate reaction are given:

h)(T) = (m-V)mH; (m.V)OH- (m-y)~H~ (x-y),

I (25)
K4(T) = (m.j' )MHZCoo- ( X’Y)W / (m’y)NH3 (m YhCOj

Additionally the mass balance in the liquid phase as well as in the

gaseous phase and the condition of electroneutrality have to be ta-
ken into account. Therefore when the temperature and the overall
concentrations in the liquid phase are given there remains a system

of 12 nonlinear equations to be solved, provided Henry constants,

equilibrium constants, fugacity coefficients and activity coeffi-
cients are available.
Van Krevelen et al. applied very restictive assumptions to solve

that system (69). Their model is able to give a good representation

of vapor-liquid equilibria in ammonia-rich mixtures at temperatures
between 20 and 60 'C in a comparatively small concentration range
even when hydrogen sulfide is present in the mixture in addition to
carbon dioxide and ammonia. As this model is described and compared
with experimental results in detail elsewhere (70) it will not be
discussed here. In view of the limiting restrictions applied by van
Krevelen et al. this model was not extended to multicomponents

systems. More modern methods aim to solve the nonlinear equations by

adapting recent results for the activity coefficients in aqueous so-
lutions of strong electrolytes, especially the method of Pitter and

coworkers. Edwards et al. (71) use Pitzer's expression for the ex-
cess Gibbs energy in an aqueous solution of strong electrolytes to
calculate activity coefficients of all solute species (molecular and
ionic) but they neglect contributions from ternary interactions. As
Pitzer's equation is used to describe the influence of interactions
between ionic and molecular species as well as between molecular
species a very large number of binary parameters must be known. For
example in the system NH3 - CO2 - H2S - SO2 - H20 considered by
Edwards et al. besides four molecular solute species there are 8
ionic species e.g. NH4+, H+. OH-, HC03-, Cog’, HS-, S’ and NH2COO-.
Even if it is assumed that all interaction parameters are symmetric
there remain 156 binary parameters, which in principle all depend on
temperature. It is obvious that some approximations have to be used
to reduce the number of parameters considerably. A complete set of
revised parameters was pub1 i shed recent1 y (70 1. Furthermore
Pawlikowski et al. extended the temperature range for the NH3 - CO2
- H20 system to I50 ‘C by fitting some of the parameters to new
experimental results for the vapor-liquid equilibrium in that system
(73). A comparison between measured and calculated partial pressures
of ammonia and carbon dioxide using interaction parameters by
Pawlikowski et al. shows good agreement (cf. Figure 5). Beutier and
Renon (74. 751* split the excess Gibbs free energy GE of a multi-
component mixture of ionic and molecular compounds into three terms
describing the influence of interactions between ionic species -
GijE -, between ionic and molecular solutes - GAGE - and bet-
ween molecular solutes - GaaE. For the ionic contribution GijE
Pitzer’s equation is used, but unl ike the method of Edwards et al.
ternary interactions were al so taken into account. The ion-molecule
contribution Gi a E is described by an expression based upon the
Debye-McAulay electrostatic theory discussed above. Molecule-molecu-
le contribution GaaE is assumed to result only from interactions
between like solute species. A Margules type of expression is used
for G,, E which contains only one parameter per molecular solute.
Generally the procedure by Beutier and Renon provides similar agree-
ment with experimental results as the method of Edwards et al.. Chen
et
-* al (76, 20) al so used Pitzer’s equation for the excess Gibbs
free energy to describe activity coefficients in electrolyte solu-
tions. Similar to the method of Edwards et al. that expression was
extended to describe also the influence of interactions including
molecular solutes. But in contrast to the assumptions made by
Edwards et al. ternary interactions and interactions between ions
having the same sign of charge are not neglected a priori. In apply-
ing that method one again faces the problem of having to determine a
larger number of parameters. Chen et al. selected the important pa-
rameters by a preliminary order of magnitude analysis. The parame-

*) There are some misprints in the original paper by Beutier and

Renon which were recently corrected. These misprints are the
reason for the discrepancies mentioned in (701.
meters found to be significant were then fitted to experimenta VLE
data. By way of example for the hydrochlorid ac id-water system at
25 "C 6 parameters were found to be significant and consequent Y
fitted to experimental results.
Similar procedures as discussed here for aqueous systems were
applied by Mather and coworkers to correlate solubility data of CO2
and H2S in some aqueous amine solutions (77 - 79) and by Pawlikowski
et al. to describe the influence of low levels of phenol on the so-
lubility of ammonia in water (83).

1.6
- colt
Fuwiikowski et al. (73 I
1.2 . 0,u exp
0
\

0.8 mw3=9.6

0
0.4

0
* k__ mNH3 -3.1
k,

20 I I I 1 I

16 I
mNH3=3.1

12
’ / %:

0 mNH3 = 9.6
8
I -
I
4 0
,d, -
01 $ I
0123456
mco2
Fig. 5 Phase equilibria for NH3/C02/H20 system at 1OO'C.
294

REFERENCES

1 H.L. Friedman, Ann. Rev. Phys. Chem., 32 (1981) 179 - 204.

2 H.L. Friedman, in S.A. Newman (Ed.), Thermodynamics of Aqueous
Systems with Industrial Applications, ACS Symp. Ser. No. 133,
Washington DC, 1980, pp. 547-560.
K.S. Pitzer, Accounts of Chem. Res., 10 (19771 371 - 377.
R.A. Robinson and R.H. Stokes, Electrolyte Solutions, Second
Edition Butterworths, London (1970).
Debye and E. Hiickel, Phys.Zeitschrift, 24 (1923) 185 - 206.
J:S. Newman, Electrochemical Systems, Pentice-Hall Inc., Engle-
wood Cliffs N.J. (1973).
J.C. Rasaiah and H.L. Friedman, J. Chem. Phys., 48 (1968) 2742-
2752.
H. Planche and H. Renon, J. Phys. Chem., 85 (1981) 3924 - 3929
see also H. Planche Ph.D-Thesis, Ecole National Superieure des
Mines de Paris (1982).
9 H. Renon, H. Planche, W. Filrst and F.X. Ball, paper presented
at the 2nd World Congress Chem. Eng. Montreal 1981.
10 R. H. Stokes and R.A. Robinson, J. Am. Chem. Sot., -70 (1948)
1870-1878.
K. S. Pitzer, J. Phys. Chem., 77 (1973) 268 - 277.
:: K. S. Pitzer and G. Mayorga. J. Phys. Chem., 77 (1973) 2300 -
2308 with corrections ibid. 78 (1974) 2698.
13 S. Pitzer and 6. Mayorga, J. Solution Chem., 3 (1974) 539 -
5K46
14 ; Pftzer and J. J. Kim, J. Am. Chem. Sot., -96 (1974) 5701 -
sKio7:
15 K. S. Pitzer, J. Solution Chem., 4 (1975) 249 - 265.
16 K. S. Pitzer and L. F. Silvester. ibid. 5 (1976) 269 - 278.
17 K. S. Pitzer, R. N. Roy and L. F: Silvesfer, J. Am. Chem. Sot.,
99 (1977) 4930 - 4936.
18 K L. Cruz and H. Renon. AIChE-J., 24 (1978) 817-830.
19 C. C. Chen, Computer Simulation of Chemical Processes in the
Electrolytes, PhD-Thesis, MIT (1980).
20 C. C. Chen, H. I. Britt, J. F. Boston and L. B. Evans, in S.A.
Newmann (Ed. 1, Thermodynamics of Aqueous Systems with Indu-
strial Applications, S. A. Newman ed. ACS Symp. Ser.No. 133,
Washington DC, 1980, pp. 61 - 89.
21 F. X. Ball, W. Fiirst and H. Renon, private communication of a
paper to be published
K. S. Pitzer, Ber. Bunsenges. Phys. Chem., 85 (1981) 952- 959.
f: Funk, Ind. Eng. Chem. Process Des. Dev., 13 (1974) 362 -
j,,W-
24 R. Vega and E.W. Funk, Desalination, 15 (1974) 225 - 242.
J.H. Vera, Can. J. Chem. Eng., 55 (1977) 484-486.
f: E. Hala, Proc. Intern. Symp. Distillation 1969,
Inst. of Chem. Engrs. Symp. Series 32, 5 - 13.
27 H. P. Meissner, in S. A. Newman (Ed.) Thermodynamics of Aqueous
Systems with Industrial Applications, ACS Symp. Ser. No. 133,
Washington DC, 1980, pp. 495 - 511.
28 Silvester and K. S. Pitzer, J. Phys. Chem., 81 (1977)
:82:- - 1828
29 K. S. Pitze; R Peterson and L. F. Silvester, J. Solution
Chem.. 7 (19;8?45.- 56.
30 L. F. -Silvester and K. S. Pitzer, ibid., 327 - 337.
31 K. S. Pitzer and L. F. Silvester. J. Phys. Chem., 82 (1978)
1239 - 1242.
32 D. J. Bradley and K. S. Pitzer, ibid., 83 (1979) 1599 - 1603.
K. S. Pitzer and J. C. Peiper, ibid. 84 (1980) 2396 - 2398.
:: P. s. z. Rogers and K. S. Pitzer, ibid., 85 (1981) 2886 - 2895.
 296

35 C.E. Harvie and J.H. Weare, Geochim. Cosmochim. Acta, 44 (1980)

981 - 997.
36 H. A. Correa and J. H. Vera, Can. J. Chem. Eng., 53 (1975)
204 - 210.
37 R. Vega and J. H. Vera, ibid., 54 (19761 245 - 248.
38 E. W. Funk, in S. A. Newman (Ed.), Thermodynamics of Aqueous
Systems with Industrial Applications, ACS Symp. Ser.No. 133,
Washington DC, 1980, pp. 717 - 739.
39 H. P. Meissner and J. W. Tester, Ind. Eng. Chem. Process Des.
Dev., 11 (1972) 128 - 133.
40 H. P. Meissner and C. L. Kusik, AIChE-J., 18 (19721 294 - 298.
41 H. P. Meissner, C. L. Kusik and J. W. Tester, ibid., 661 - 662.
42 C. L. Kusik and H. P. Meissner, Ind. Eng. Chem. Process Des.
Dev. 12 (19731 112 - 115.
H. P. Meissner and C. L. Kusik, ibid., 205 - 208.
4: C. L. Kusik and H. P. Meissner, AIChE Symp. Ser. 173 (19781
14 - 20.
45 H. P. Meissner and C. L. Kusik, Ind. Eng. Chem. Process Des.
Dev., 18 (1979) 391 - 394.
46 H. P. Meissner and M. P. Manning, Proceedings of the 2nd World
Congress Chem. Eng. Montreal, (1981) Vol. 5, 55 - 59.
47 Y. C. Wu, R. M. Rush and G. Scatchard, J. Phys. Chem., 73
(1969) 2047 - 2053.
48 R. Haase, H. Naas and H. Thumm, Colln. Czech. them. Commun.
Engl. Edn., 25 (19601 579.
49 C. Liu and W. T. Lindsay, J. Solution Chem., 1 (19721 45 - 69.
50 W. F. Furter and R.A. Cook, Int. J. Heat Mass Transfer, 10
(19671 23 - 36.
51 W.F. Furter, Can. J. Chem. Eng., 55 (1977) 229 - 239.
52 A. I. Johnson and W.F. Furter, ibid., 38 (1960) 78 - 87.
D. Meranda and W.F. Furter, AIChE-J., 17 (1971) 38 - 42.
:: R. W. Rousseau, D.L. Ashcraft and E. M. Schoenborn, AIChE-J. 18
(1972) 825 - 829.
55 J.E. Boone, R.W. Rousseau and E. M. Schoenborn, Adv. Chem.
Ser. 155 (1976) 36 - 52.
56 R. W. Rousseau and J.E. Boone, AIChE-J., 24 (19781 718 - 725.
57 Schmitt and A. Vogelpohl, Fluid Phase Equilibria, 2 (1982)
Y67 - 176.
58 P. Debye and J. McAulay, Phys. Z., 26 (1925) 22 - 29.
59 W. F. McDevit and F.A. Long, J. Am. Chem. Sot., 74 (1952)
1773 - 1777.
60 J. 0. M. Bockris, J. Bowler-Reed and J. A. Kitchener, Trans.
Faraday Sot., 47 (1951) 184 - 192.
61 B. E. Conway, J. E. Desnoyers and A. C. Smith, Phil. Trans.
Roy. Sot. London, A 256 (1964) 389 - 437.
R. A. Pierotti, J. Phys. Chem.. 69 (19651 281 - 288.
S. K. Shoor and K. E. Gubbins, J. Phys. Chem., 73 (19691 498
- 505.
64 W. L. Masterton and T. P. Lee, J. Phys. Chem., 74 (1970) 1776 -
1782.
65 K. Dnda, E. Sada, T. Kobayashi, S. Kito and K. Ito, J. Chem.
Eng. Japan, 3 (19701 18 - 24.
66 ibid. 137 - 142.
67 H. L. Clever and C. J. Holland, J. Chem. Engng. Data, 13 (19681
411 - 414.
68 T. J. Morrison and F. Billet, J. Chem. Sot.. (19521 3819-3822.
69 D. W. van Krevelen, P. J. Hoftijzer and F. F. Huntjens, Rec.
Trav. Chim. Pay-bas, 68 (1949) 191 - 216.
70 G. Maurer, in S. A. Newman (Ed.), Thermodynamics of Aqueous
Systems with Industrial Applications, ACS Symp. Ser.No. 133,
Washington DC, 1980. pp. 139 - 172.
71 T. J. Edwards, G. Maurer, J. Newman and J. M. Prausnitz,
AIChE-J., 24 (1978) 966 - 976.
72 C.C.Chen, H.1 .Britt, J.F. Boston, L.B. Evans, AICHE-J., 28
(1982) 588 - 596.
73 E. M. Pawlikowski, J. Newman and J. M. Prausnitz, Ind. Eng.
Chem. Process Des. Dev., 21 (1982) 764 - 770.
74 D. Beutier and H. Renon, ibid., 17 (1978) 220 - 230 and 19
(1980) 722.
75 H. Renon, in S. A. Newman (Ed.), Thermodynamics of Aqueous
Systems with Industrial Applications, ACS Symp. Ser.Uo. 133,
Washington DC, 1980, pp. 173 - 186.
76 C. C. Chen, H. I. Britt. J. F. Boston and L. B. Evans,
AIChE-J., 25 (1979) 820 - 831.
77 A. E. Mather and R. D. Deshmukh, in S. A. Newman (Ed.),
Thermodynamics of Aqueous Systems with Industrial Applications,
ACS Symp. Ser.Uo. 133, Washington DC, 1980, pp. 49 - 60.
78 R. D. Deshmukh and A. E. Mather, Chem. Eng. Sci., 36 (1981)
355 - 362.
79 F. Y. Jou, A. E. Mather and F. D. Otto, Ind. Eng. Chem. Process
Des. Dev., 21 (1982) 539 - 544.
80 W. F. Furter, (Ed.) Thermodynamic Behaviour of Electrolytes in
mixed Solvents
Part I , Adv. Chem. Series 155, 1976
81 Part II , Adv. Chem. Series 177, 1979
American-Chemical Society, Washington DC
82 0. M. Mason and R. Kao, in S. A. Newman (Ed.), Thermodynamics
of Aqueous Systems with Industrial Applications, ACS Symp.
Ser. No. 133, Washington DC, 1980, pp. 107 - 138.
83 E. M. Pawlikowski, J. Newman and J. M. Prausnitz, AIChE-J. 29
(1983) in press.