Você está na página 1de 18

Journal of Solution Chemistry, Vol. 28, No.

4, 1999

Thermodynamics of Electrolytes. 13. Ionic Strength


Dependence of Higher-Order Terms; Equations
for CaCl2 and MgCl2
Kenneth S. Pitzer,1,4 Peiming Wang,1 Joseph A. Rard,2,*
and Simon L. Clegg3

Received December 16, 1998

While the original ion-interaction (Pitzer) equations of 1973 were adequate for
many electrolytes to the limit of solubility, additional terms are needed for some
systems of large solubility. A simple pattern of ionic-strength dependence is
proposed for third, fourth, and higher virial coefficients. It is found to be very
effective in representing the complex behavior of CaCl2(aq) as well as that of
MgCl2(aq) at 25°C. Equations without ion association are presented for each
system that are valid for the full range to 11.0 and 5.9 mol-kg -1 , respectively,
as well as simpler equations for limited molality ranges.

KEY WORDS: Electrolyte solutions; osmotic coefficient; activity coefficient;


thermodynamics; calcium chloride; magnesium chloride.

1. INTRODUCTION
The primary advance in papers 1 and 2 of this series(1,2) arose from the
recognition and accurate representation of the ionic-strength dependence of

1
Department of Chemistry and Lawrence Berkeley National Laboratory, University of Califor-
nia, Berkeley, California 94720-1460. Current address for Peiming Wang: OLI Systems, Inc.,
108 American Road, Morris Plains, New Jersey 07950.
2
Geosciences and Environmental Technologies, Earth and Environmental Sciences Directorate,
Lawrence Livermore National Laboratory, University of California, Livermore, California 94550.
3
School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, U.K.
4
Deceased December 26, 1997.
f
We dedicate this paper to the memory of Kenneth S. Pitzer in recognition of his many
invaluable contributions to solution chemistry.
265
0095-9782/99/0400-0265$16.00/0 C 1999 Plenum Publishing Corporation
266 Pitzer, Wang, Rard, and Clegg

the second virial coefficient in the low molality range. Basic theory for
electrolyte solutions(3) provides for ionic-strength dependence of the various
virial coefficients. However, for most systems and with accurate second
virials, the third virial coefficient is so small that any ionic-strength depen-
dence is negligible. Thus, the ion-interaction (Pitzer) equations were presented
with third virial coefficients independent of ionic strength.(1,2,4,5) They are,
of course, dependent on pressure and temperature. This was found satisfactory
for mixed as well as pure electrolytes(4) and even extending to solid saturation
molalities, in most cases, in the extensive investigations of Harvie and Weare(6)
and of Filippov and associates.(7) Subsequently, fourth and higher virial coeffi-
cients were added for electrolytes extending to very high molality. Examples
are CdCl2, ZnCl2, LiCl, and CaCl2. For the CdCl2, ZnCl2, and LiCl, these
treatments were quite satisfactory,(8,9) but for the important case of CaCl2,
there are now very accurate data(10) and the 1985 fit of Ananthaswamy and
Atkinson (11) with virial coefficients through the sixth order, although good, can
now be improved. This paper presents and tests for CaC!2(aq) and MgCI2(aq) a
simple formulation of ionic-strength dependent terms extending to any order.
Recently, Archer(12) introduced an ionic-strength term at the third virial
level for NaBr, which is soluble to 9 mol-kg - 1 at 25°C. He also found that
this addition improved the fit to the very accurate data for NaCl(aq).(13)
Archer's extension with an ionic-strength dependent third virial term was
generalized by Clegg, Rard, and Pitzer(14) to self-associated electrolytes and to
electrolyte mixtures. They applied these equations to solutions of H2SO4(aq),
which contain equilibrium mixtures of H+, SO2-, and HSO4-. Their model provides
a very accurate representation of the available osmotic and activity coefficients,
enthalpies of dilution, and heat capacities to 6.1 mol-kg -1 . The degrees of dissocia-
tion of the HSO4 ion calculated with this model are in good agreement with
experimental values from spectroscopic measurements.
Rard and Clegg(10) used similar self-association models for CaCl2(aq), which
include CaCl+ ion pairs. There is ample evidence for ion pair formation above
about 5 mol-kg-1 in these solutions,(15) but neutron diffraction data(16) show no
significant association at 4.5 mol-kg -1 . Rard and Clegg(10) obtained an accurate
model for CaCl2(aq) valid to 8.0 mol-kg -1 , and, by including additional virial
terms, a second model valid over the full molality range to 11.0 mol-kg -1 .
However, the association constant K°(CaCl+) is poorly known at 25°C, in contrast
to the more precisely known value of K(HSO4). This uncertainty for K°(CaCl+)
gives rise to large uncertainties in the predicted degree of dissociation of CaCl+
and the corresponding ionic activity coefficients for CaCl2(aq) solutions, which
then depend strongly on the assumed value of K°(CaCl+) and on whether higher-
order virial terms are included in the model.
Both models presented by Rard and Clegg yield very accurate values
for the stoichiometric quantities c and G±, and thus their equations are
Ionic Strength Dependence 267

satisfactory when CaCl2(aq) is used as an isopiestic reference standard or to


calculate the emf of a reversible cell. However, use of distorted values of
ionic activities can cause difficulties when the thermodynamic properties of
mixed electrolytes containing CaCl2 are being analyzed, since any anomalous
predictions of the extent of formation of CaCl+ must be compensated for the
values of the ionic activity coefficients of the other electrolytes. For such
mixture calculations, there are advantages in having an accurate thermody-
namic model in which CaCl2(aq) is treated as a fully dissociated electrolyte.
There are alternate forms of expression for an ionic strength dependence
of the third virial coefficient—as there were for the second virial term.(l) We
have considered alternate forms not only for the third virial but also for the
fourth and higher terms. A form is selected that retains the relatively simple
expressions of the second (virial) coefficient level for all higher coefficients.
At the third virial level, its contribution is very similar to that of Archer's
expression. With reoptimization of all parameters, the fit for NaBr and NaCl
should remain excellent, but there is no reason to change Archer's equations
for these cases. It is, for cases using fourth or higher coefficients, that the
advantages of our new form become important.
These equations are used here to represent the osmotic coefficients of
CaCl2(aq) and MgCl2(aq) at 25°C, and good quality fits were obtained for
both systems. These same equations should be applicable to even more highly
soluble electrolytes. However, for miscible or nearly miscible systems, e.g.
H2SO4(aq) and HNO3(aq), where the molality can approach or become infi-
nite, equations based on the mole fraction scale are more suitable.(5)

2. EQUATIONS FOR PURE ELECTROLYTES


We first recall the expression for the second virial coefficient for the
osmotic coefficient of a pure electrolyte

with possible additional terms in B(3), a3, etc.; in addition, B(2) is often zero.
If we define

Eq. (1) becomes


268 Pitzer, Wang, Rard, and Clegg

Then the second virial expression for the excess Gibbs energy is

For the natural logarithm of the mean activity coefficient, one has

Note that if a = 0, x = 0, exp(-x) = 1, and limx->0 g(x) = 1; thus, one can


write the ionic-strength dependence of each term in the same form. Then the
B(0) term may be viewed as having the form B(0) exp(-AoI1/2), but with
a0 = 0.
For higher virial coefficients, we generalize Eqs. (4) and (5) for any
one term in the nth virial for the excess Gibbs energy to

Then appropriate differentiation yields, for the osmotic coefficient

and for the natural logarithm of the mean activity coefficient

Thus, for the fourth virial, if there are three terms, one has for the excess
Gibbs energy and with xDO = 0

with

Then the fourth virial for the osmotic coefficient is


Ionic Strength Dependence 269

and for the natural logarithm of the mean activity coefficient is

A complete expression through the sixth virial for the osmotic coefficient
is then

with |zMzx|fP the Debye-Huckel term. If there is no ionic-strength depen-


dency for the third and higher virials, this reduces to the expression adopted
for CaCl2(aq) by Ananthaswamy and Atkinson.(10) We now consider
CMX, DMX, ... to have possible ionic strength dependences with

For the mean activity coefficient, the corresponding equations are


270 Pitzer, Wang, Rard, and Clegg

At the third virial level, the only difference from the Archer term for
the osmotic coefficient is the use of the first power of I instead of I1/2 in xCL.
However, for the Gibbs energy and the activity coefficient, the expressions
corresponding to Eqs. (5) and (17b) are more complex for the Archer term
and progressively much more complex for the fourth or higher virials.
At each level, the ionic-strength-independent term, B(0), C(0), D(0), etc.,
is obtained with a = 0, x = 0. It is only at the second virial level that there
is useful guidance from basic theory for the A's. The other nonzero a values
are chosen for optimum fit to experimental data. While purely empirical, the
higher order terms with nonzero a are very useful in fitting experimental
data since their contribution is relatively localized in a limited molality range.

3. EQUATIONS FOR MIXED ELECTROLYTES


While no new, detailed, treatments for particular mixed system are
presented here, a few comments are appropriate. Above the third virial level,
a complete expression with all possible mixing terms becomes very complex.
It seems best to follow the method introduced by Filippov etal.(8) for mixtures,
including CdCl2, and used by Anstiss and Pitzer(9) for ZnCl2-NaCl-H2O.
For these 2-1 electrolytes, the fourth and fifth virial terms of the M 2+ , Cl-
electrolyte should be dominated by (M 2+ )(Cl - ) 3 and (M 2+ )(Cl - ) 4 interactions
(or complexes). It is then assumed that the excess Gibbs energy from the
fourth and fifth terms is

Comparison with Eqs. (7), (8), and (10) then yields, for the D and E terms

For the contribution of the (M 2+ )(Cl - ) 3 and (M 2+ )(Cl - ) 4 components


to the higher virial terms of the osmotic and activity coefficients, one finds

Then, for mixtures, terms of high order such as O(Na, Cd, 3Cl) are added if
Ionic Strength Dependence 271

needed and the resulting equations successfully represent mixed electrolyte


properties.(8,9)
At the third virial level, the situation for 1-1 electrolytes is straightfor-
ward. The equations in general use(4-6) remain satisfactory if the CMX, CMX,
and CMX quantities become ionic-strength dependent. In particular,

For cases involving multiply charged ions, the established definition(4,5)


of CMX is

It is the factor |zxzx|1/2 that is inconsistent. It arose from rather complex


considerations of the mixing terms PMNX, P MXY and their definitions for ions
of differing charge.(1,4,5)
Clegg et al.(14) treat sulfuric acid as a mixed electrolyte with ionic-
strength-dependent third virial terms. Their Appendix I includes appropriate
general equations for Archer-type terms. With amendments, these equations
would serve for the present ionic-strength function. We prefer to delay a full
treatment of mixed systems until one or more examples are considered,
including fourth virial and possibly higher terms. At this time, we suggest
only that an ionic strength-dependent CMX obtained from Eq. (23) or from
Eqs. (15b), (17b), and (24) be divided by |zMzx|1/2 for use in the familiar
equations for mixed electrolytes.(1,4,5)

4. EXPERIMENTAL DATA AND RESULTS FOR CaCI2


Rard and Clegg(10) presented, for CaCl2(aq) at 25°C, a very extensive
and carefully reviewed data base. Included were new isopiestic measurements
in the range 4-10 mol-kg -1 referenced to H2SO4(aq). That data base was
used for the present calculations. Since a full description has been pub-
lished,(10) only a few comments are needed here. In general, the change from
the 1977 review of Rard et al.(17) or from the older review of Robinson and
Stokes(18) is very small. Rard and Clegg(10) give full consideration to various
electrochemical cell measurements, but find only a few to merit inclusion
for the general equation. It is only in the very dilute region, below 0.2 mol-
kg - 1 , that their effect is important. Since their inclusion would complicate
the present calculations, the cell data were omitted. In their place the P values
at or below 0.2 mol-kg -1 , from the final table of recommended values of
Rard and Clegg, were added.
272 Pitzer, Wang, Rard, and Clegg

Since the basic objective was equations that are not only accurate, but
also convenient, for use for either pure CaCl2 or for mixed solutions, several
equations of increasing complexity are presented for increasing ranges of
molality. We consider first the complex equation, but still without ion associa-
tion, for the full range from 0 to 11 mol-kg -1 (i.e., 3.5 mol-kg -1 above
saturation). It extends to the fifth virial coefficient with nine terms in all.
The parameters are given in Table I, along with those for the simpler equations.
The Debye-Huckel parameter AP = 0.391476 kg l / 2 -mol - l / 2 is based on the
dielectric constant equation of Archer and Wang.(l9) There are five ionic
strength-dependent terms with a values selected after various trial calcula-
tions. The B, C, D, and E values are obtained by the least-squares method.
For mixtures the molality of CaCl2 seldom exceeds 7 mol-kg -1 and the
ionic strength is less than 21 mol-kg -1 . In addition, the interpretation of the
fifth virial is ambiguous with respect to CaCl4 and Ca2Cl3 interactions.
Hence, an equation limited to the fourth virial is desirable; we present one
valid to 7 mol-kg -1 with seven terms. The fourth virial is certainly dominated
by CaCl3 interactions and can be treated as a w(Ca, 3Cl) term in the Filippov
system described above. This equation, valid to 7 mol-kg - 1 , also suffices for

Table I. Parameter Values for CaCl(aq) in Eqs. (14-17) with 3, 5, 7, and 9 Ion-
Interaction Parameters

Parameter Units np = 9a np = 7a np = 5a np = 3b
B(0) kg-mol-1 0.569081 0.464899 0.446867 0.31588
B(1)
(2)
kg-mol -1 1.59385 1.55840 1.57819 1.6140
B kg-mol -1 0 0 0 0
ABI kg1/2-mol1/2 3.0 2.5 2.45 2.0
C(0) kg 2 -mol -2 -0.0556802 0.00459539 -0.00663855 -0.00017103
C(2)(1) kg2- mol - 2 0.215755 -0.00941176 -0.0151049
C kg 2 -mol -2 -0.317714 -0.0729924 -0.0421769
ACI kg-mol -1 0.15 0.15 0.15
AC2 kg-mol-1 0.25 0.25 0.25
D(()) kg-mol-3 0.00282839 -0.000745381
D(1) kg3-mol-3 0.0170624 -0.00241580
D(2) kg 3 -mol -3 -0.0244910
ADI kg3/2-mol-3/2 0.030 0.030
AD2 kg3/2-mol-3/2 0.035
(0)
E kg4-mol-4 -0.0000509632
mmax mol-kg -1 11.0 7.0 5.2 2.5
Sf 0.0033 0.0026 0.0022
Number of points 430 355 293
a
This study.
b
Pitzer and Mayorga (Ref. 2).
c
Standard deviation of the fit.
Ionic Strength Dependence 273

the interpretation of many isopiestic measurements where CaCl2 is the refer-


ence electrolyte; those for MgCl2, as discussed below, are an example. For
solutions of still lower ionic-strength, it is convenient to have parameters
that can be used in the programs in common use that extend only to the third
virial coefficient. Parameters are given in Table I for a treatment with C(1)
and C(2) terms for the range to 5.2 mol-kg -1 .
We also considered the appropriate treatment for a still more limited
range and with only B(0), B(1), and C(0) terms. The parameters originally given
in 1973(2) for the range to 2.5 mol-kg-1 yield such good results and are now
so widely used that we recommend their retention without change. These
parameters are included in Table I.
Each of these functions is compared with the recommended values of
Rard and Clegg(10) in Fig. 1 and in Tables II and III. The maximum deviations

Fig. 1. Comparison of (a) osmotic coefficients and (b) activity coefficients of CaCI2(aq) from
Rard and Clegg (Ref. 10) with those calculated using equations from this study (Eqs. 14 and
16; np = 5, 7, 9) and from Pitzer and Mayorga (Ref. 2; np = 3). np = 9; np =
7; np = 5; np = 3.
274 Pitzer, Wang, Rard, and Clegg

Table II. Comparison of the Osmotic Coefficients of CaCl2(aq) from Table 11 of Rard
and Clegg with Those Calculated Using Eq. (14) and Ion-Interaction Parameters in
Table I
ma Rard and Clegg np = 9b np = 7b np = 5b np = 3c
0.001 0.9623 0.9623 0.9622 0.9622 0.9621
0.002 0.9493 0.9494 0.9492 0.9492 0.9490
0.005 0.9275 0.9275 0.9271 0.9272 0.9269
0.01 0.9078 0.9079 0.9074 0.9074 0.9072
0.02 0.8871 0.8871 0.8866 0.8867 0.8866
0.05 0.8628 0.8627 0.8627 0.8629 0.8636
0.10 0.8527 0.8527 0.8533 0.8535 0.8553
0.20 0.8588 0.8589 0.8593 0.8594 0.8613
0.30 0.8748 0.8753 0.8751 0.8750 0.8758
0.40 0.8944 0.8957 0.8950 0.8948 0.8941
0.50 0.9163 0.9181 0.9172 0.9170 0.9150
0.60 0.9401 0.9419 0.9411 0.9410 0.9382
0.70 0.9655 0.9668 0.9664 0.9664 0.9633
0.80 0.9924 0.9928 0.9929 0.9929 0.9900
0.90 1.0205 1.0200 1.0204 1.0205 1.0180
1.00 1.0495 1.0482 1.0489 1.0490 1.0474
1.20 1.1100 1.1078 1.1087 1.1088 1.1092
1.40 1.1733 1.1715 1.1721 1.1721 1.1745
1.60 1.2393 1.2387 1.2387 1.2386 1.2426
1.80 1.3079 1.3089 1.3084 1.3082 1.3129
2.00 1.3793 1.3816 1.3807 1.3805 1.3850
2.25 1.4725 1.4754 1.4744 1.4743 1.4772
2.50 1.5699 1.5717 1.5712 1.5712 1.5711
2.75 1.6705 1.6702 1.6703 1.6706
3.00 1.7731 1.7705 1.7713 1.7718
3.25 1.8759 1.8723 1.8736 1.8740
3.50 1.9778 1.9751 1.9765 1.9767
3.75 2.0788 2.0785 2.0795 2.0793
4.00 2.1796 2.1818 2.1820 2.1812
4.25 2.2805 2.2841 2.2834 2.2819
4.50 2.3808 2.3845 2.3828 2.3811
4.75 2.4792 2.4820 2.4795 2.4784
5.00 2.5743 2.5753 2.5727 2.5735
5.25 2.6643 2.6635 2.6615 2.6661
5.50 2.7479 2.7457 2.7450
5.75 2.8241 2.8209 2.8220
6.00 2.8920 2.8887 2.8918
6.25 2.9514 2.9486 2.9531
6.50 3.0022 3.0005 3.0049
6.75 3.0447 3.0443 3.0463
7.00 3.0796 3.0804 3.0762
7.25 3.1074 3.1092
7.50 3.1291 3.1314
7.75 3.1453 3.1476
Ionic Strength Dependence 275

Table II. Continued


a
m Rard and Clegg np = 9b np = 7b np = 5b np = 3c

8.00 3.1570 3.1588


8.25 3.1649 3.1658
8.50 3.1697 3.1694
8.75 3.1721 3.1707
9.00 3.1727 3.1702
9.25 3.1720 3.1688
9.50 3.1704 3.1671
9.75 3.1683 3.1654
10.00 3.1659 3.1641
10.25 3.1635 3.1633
10.50 3.1610 3.1629
10.75 3.1586 3.1628
11.00 3.1563 3.1625
a
Units, mol-kg-1.
b
This study: np is the number of ion-interaction parameters used in the fit.
c
Pitzer and Mayorga (Ref. 2).

remain below the range of deviations of the more accurate measurements


(±0.3% of P) as shown in the figures of Rard and Clegg.(10)
A comparison of the present np = 5 fit for m < 5.2 mol-kg-1 to
the corresponding Archer equation(12,13) fit to a comparable molality range
indicates that the present model gives a better representation of the osmotic
coefficients of CaCl2(aq) by up to 0.004 in P. However, the present model
has one more adjustable parameter.

5. EXPERIMENTAL DATA AND RESULTS FOR MgCl2


Goldberg and Nuttall (20) in 1978 and Rard and Miller (21) in 1981 reviewed
the available data for MgCl2(aq) at 25°C. Rard and Miller also reported
important new isopiestic measurements at higher molalities. Above 2.5 mol-
kg~', the new measurements give significantly higher osmotic coefficients
than the earlier data that were primarily from Stokes.(22) The difference, which
exceeds 0.01 above 3.0 mol-kg -1 , is surprising and probably arose from
contamination with alkali chlorides in some of the earlier studies.(21) The
Rard and Miller values were adopted for the isopiestic ratios. These and
other P values from isopiestic measurements were recalculated for the more
recent equations for the reference solutes: Archer(13) for NaCl(aq), Clegg et
al.(14) for H2SO4(aq), and our new np = 7 equation for CaCl2(aq).
Below 2.5 mol-kg-1 there are several sets of isopiestic measurements(22-29)
The Rard-Miller(21) and Goldberg-Nuttall(20)evaluations of osmotic coeffi-
276 Pitzer, Wang, Rard, and Clegg

Table III. Comparison of the Activity Coefficients of CaCl2(aq) from Table 11 of Rard
and Clegg with Those Calculated Using Eq. (16) and Ion-Interaction Parameters in
Table I

ma Rard and Clegg np = 9b np = 7b np = 5b np = 3c

0.001 0.8886 0.8886 0.8884 0.8884 0.8882


0.002 0.8508 0.8509 0.8505 0.8505 0.8503
0.005 0.7871 0.7872 0.7865 0.7865 0.7860
0.01 0.7290 0.7292 0.7282 0.7283 0.7277
0.02 0.6651 0.6652 0.6640 0.6642 0.6636
0.05 0.5786 0.5786 0.5777 0.5779 0.5781
0.10 0.5187 0.5187 0.5183 0.5186 0.5197
0.20 0.4716 0.4716 0.4714 0.4717 0.4735
0.30 0.4538 0.4541 0.4536 0.4538 0.4552
0.40 0.4476 0.4484 0.4476 0.4477 0.4485
0.50 0.4479 0.4490 0.4481 0.4482 0.4483
0.60 0.4527 0.4540 0.4530 0.4532 0.4526
0.70 0.4609 0.4621 0.4613 0.4615 0.4606
0.80 0.4721 0.4730 0.4724 0.4726 0.4715
0.90 0.4859 0.4864 0.4859 0.4861 0.4852
1.00 0.5021 0.5021 0.5017 0.5020 0.5013
1.20 0.541 1 0.5405 0.5403 0.5406 0.5408
1.40 0.5891 0.5884 0.5881 0.5884 0.5900
1.60 0.6467 0.6467 0.6460 0.6463 0.6493
1.80 0.7153 0.7164 0.7152 0.7154 0.7197
2.00 0.7965 0.7989 0.7973 0.7974 0.8024
2.25 0.9191 0.9227 0.9207 0.9209 0.9256
2.50 1.0702 1 .0736 1.0715 1.0720 1.0744
2.75 1.2554 1 .2568 1.2552 1.2561
3.00 1.4811 1.4791 1.4784 1.4797
3.25 1.7534 1.7487 1.7488 1 .7505
3.50 2.0795 2.0753 2.0759 2.0774
3.75 2.4695 2.4703 2.4702 2.4708
4.00 2.9377 2.9463 2.9440 2.9427
4.25 3.5010 3.5170 3.5106 3.5068
4.50 4.1762 4.1965 4.1845 4.1787
4.75 4.9786 4.9985 4.9803 4.9758
5.00 5.9208 5.9351 5.9120 5.9176
5.25 7.0112 7.0155 6.9914 7.0257
5.50 8.2526 8.2448 8.2265
5.75 9.6418 9.6228 9.6190
6.00 11.169 11.144 11.162
6.25 12.820 12.796 12.837
6.50 14.575 14.562 14.612
6.75 16.416 16.422 16.439
7.00 18.322 18.352 18.257
7.25 20.276 20.329
7.50 22.263 22.334
7.75 24.270 24.350
Ionic Strength Dependence 277

Table III. Continued

ma Rard and Clegg np = 9b np = 7b np = 5b np = 3c

8.00 26.291 26.366


8.25 28.321 28.378
8.50 30.360 30.386
8.75 32.411 32.399
9.00 34.477 34.427
9.25 36.566 36.485
9.50 38.685 38.589
9.75 40.841 40.754
10.00 43.042 42.993
10.25 45.292 45.315
10.50 47.598 47.722
10.75 49.960 50.207
11.00 52.377 52.752
a,b,c
See footnotes, Table II.

cients from these data are essentially the same and need not be discussed in
detail here. Only the values given large weights in these reviews were included
in the present calculations. A few more recent measurements of Kuschel and
Seidel(30) were included.
For the very dilute range, the values from the freezing point measure-
ments of Gibbard and Gossmann(31) can be converted using thermal data to
yield accurate osmotic coefficients at 25°C; these were included up to 1.02
mol-kg -1 . In addition, in the extremely dilute range below 0.01 mol-kg-1,
activity coefficient values from diffusion data of Harned(32) were included.
For an accurate representation over the full range to 5.9 mol-kg -1 , an
equation extending to the fourth virial with six terms was required. Since
there is no ambiguity of interpretation of the fourth virial as an MgCl3 term,
the equation is fully satisfactory for use in mixed electrolytes. Nevertheless,
it is convenient to have a simpler equation if it is valid over a substantial
range. Thus, we also present an equation with only B(0), B(1), C(0), C(1) terms
that is valid to 4.5 mol-kg -1 . This is essentially the same pattern as the
Archer equations(12,13) for NaBr(aq) and NaCl(aq), the only difference being
in the simpler ionic strength dependence of the C(1) term in the present
equation. The parameters for these equations are in Table IV.
It is interesting to compare the equation first presented in 1973(2) and evaluate
its validity for further use. As expected, the MgCl2 np = 3 equation deviates in
the range above 2.0 mol-kg-1 where the Rard-Miller measurements(21) differ
significantly from the earlier values. Below 2.0 mol-kg -1 , the agreement shown
in Fig. 2 is acceptable for many purposes. However, the new np = 4 equation
278 Pitzer, Wang, Rard, and Clegg

Table IV. Parameter Values for MgCl2(aq) in Eqs. (14-17) with 3, 4, and 6 Ion-
Interaction Parameters

Parameter Units np = 6a np = 4a np = 3b
B(0) kg-mol -1 0.563113 0.512290 0.35235
B(1) kg-mol -1 1 .95628 1 .97367 1.6815
kg-mol -1
(2)
B 0 0 0
ABI kg1/2-mol-1/2 3.0 2.8 2.0
C(0) kg 2 -mol -2 -0.0279311 -0.00676278 0.0025959
C(1) kg 2 -mol -2 0.0468272 -0.0665035
C(2) kg 2 -mol -2 -0.129805
ACI kg-mol -1 0.15 0.23
AC2 kg-mol -1 0.25
D(0) kg3-mol-3 0.00131506
mmax mol-kg -1 5.9 4.5 2
of 0.0028 0.0030
Number of weighted points 176 150
a
This study.
b
Pitzer and Mayorga (Ref. 2).
c
Standard deviation of the fit.

Fig. 2. Differences of the experimental osmotic coefficients (symbols) of MgCI2(aq) from the
model equation, Eq. (14), and parameters given in Table IV, fit to 5.9 mol-kg -1 (np = 6): +
Ref. 21; • Ref. 27; A Ref. 29; X Ref. 28; * Ref. 22; * Ref. 24; - Ref. 25; - Ref. 26; a Ref.
30; 0 Ref. 31; o Ref. 23. Differences are also shown for the model fit to 4.5 mol-kg-1 (np
= 4, dotted curve) and the model by Pitzer and Mayorga (Ref. 2; np = 3, dashed curve) from
those fit to 5.9 mol-kg-1 (np = 6).
Ionic Strength Dependence 279

is no more complex than the Archer equations(12,13) that are coming into general
use; hence, the new np = 4 equation should be used for new research.
Tables V and VI give osmotic and activity coefficient values for these
equations; Fig. 2 shows comparisons with the array of measurements for the
appropriate range of molality.

6. DISCUSSION
If there is significant ion association at very low concentration, it is
unambiguous and can be evaluated from thermodynamic, conductance, or
structural measurements. Aqueous sulfuric acid is an example. In addition,
for cases where there is partial association at low concentration with redissoci-
ation at higher molality, as in MgSO4(aq) and other 2-2 electrolytes, the
method with B(2) = -K/2 is satisfactory(5,33) At higher concentrations, the
population of positive and negative ion near neighbors is substantial on a
Debye-Huckel basis and is subject to large variation for differences in short-
range forces. Thus, any assignment to ion association is ambiguous and
alternate descriptions may be both simpler and preferable for various reasons.
The method described in this paper is shown to give an accurate descrip-
tion of the CaCl2(aq) system at 25°C where any significant ion association
occurs only at high molality. This system has important advantages over a
representation that includes an assumed ion-association constant. One is
convenience of calculation, even for the pure electrolyte. The calculation of
the fractional association at each point and the corresponding ionic strength is
avoided. For mixed electrolytes, it is more important to avoid the unnecessary
complexity introduced by the assumption of additional species of ambiguous
molality. Thus, we strongly recommend the present system when higher-
order terms are needed, in addition to the long-established formulation (1,2,4,5)
with B(0), B(1), B(2), and C terms.

ACKNOWLEDGMENTS
The work at Berkeley was supported by the Director, Office of Energy
Research, Office of Basic Energy Sciences, Division of Chemical Sciences
of the U. S. Department of Energy under Contract No. DE-AC03-76SF00098.
The contribution of J. A. R. was performed under the auspices of the Office
of Basic Energy Sciences (Geosciences) of the U. S. Department of Energy
by the Lawrence Livermore National Laboratory under contract No. W-7405-
ENG-48. That of S.L.C. was supported by an Advanced Research Fellowship
(GT5/93/AAPS/2) from the Natural Environment Research Council of the
U.K.
280 Pitzer, Wang, Rard, and Clegg

Table V. Calculated Osmotic Coefficients of MgCl2(aq) Using Eq. (14) and Ion-
Interaction Parameters in Table IV

ma np = 6b np = 4b np = 3c

0.0001 0.9870 0.9870 0.9870


0.0005 0.9726 0.9725 0.9723
0.00 1 0.9627 0.9627 0.9622
0.005 0.9292 0.9291 0.9275
0.01 0.9107 0.9107 0.9083
0.05 0.8703 0.8709 0.8682
0.10 0.8629 0.8637 0.8633
0.20 0.8724 0.8727 0.8753
0.30 0.8929 0.8923 0.8953
0.40 0.9182 0.9171 0.9192
0.50 0.9463 0.9451 0.9459
0.60 0.9765 0.9754 0.9749
0.70 1.0082 1.0074 1.0059
0.80 1.0413 1.0410 1.0388
0.90 1.0757 1.0758 1.0733
1.00 1.1113 1.1118 1.1093
1.20 1.1863 1.1874 1.1850
1.40 1.2661 1.2672 1.2651
1.60 1.3505 1.3512 1.3489
1.80 1.4391 1.4390 1.4358
2.00 1.5313 1.5304 1.5253
2.20 1.6265 1.6250
2.40 1.7242 1.7222
2.60 1.8237 1.8218
2.80 1.9246 1.9232
3.00 2.0265 2.0260
3.20 2.1290 2.1296
3.40 2.2320 2.2337
3.60 2.3352 2.3378
3.80 2.4388 2.4417
4.00 2.5427 2.5449
4.25 2.6734 2.6725
4.50 2.8055 2.7982
4.75 2.9396
5.00 3.0765
5.25 3.2171
5.50 3.3624
5.75 3.5135
5.90 3.6074

a,b,c See footnotes, Table II.


Ionic Strength Dependence 281

Table VI. Calculated Activity Coefficients of MgCl2 (aq) Using Eq. (16) and Ion-
Interaction Parameters in Table IV

ma np = 6b np = 4b np = 3c

0.0001 0.9613 0.9613 0.9612


0.0005 0.9184 0.9184 0.9179
0.001 0.8894 0.8893 0.8885
0.005 0.7900 0.7899 0.7870
0.01 0.7337 0.7337 0.7294
0.05 0.5897 0.5902 0.5837
0.10 0.5332 0.5342 0.5288
0.20 0.4905 0.4913 0.4883
0.30 0.4771 0.4775 0.4753
0.40 0.4761 0.4761 0.4739
0.50 0.4823 0.4821 0.4793
0.60 0.4935 0.4932 0.4898
0.70 0.5087 0.5086 0.5045
0.80 0.5276 0.5276 0.5228
0.90 0.5498 0.5500 0.5448
1.00 0.5753 0.5759 0.5701
1.20 0.6370 0.6380 0.6315
1.40 0.7142 0.7156 0.7082
1.60 0.8097 0.8109 0.8022
1.80 0.9266 0.9274 0.9163
2.00 1.0693 1.0693 1.0541
2.20 1.2427 1.2417
2.40 1.4530 1.4511
2.60 1.7076 1.7051
2.80 2.0152 2.0130
3.00 2.3866 2.3858
3.20 2.8346 2.8369
3.40 3.3749 3.3817
3.60 4.0268 4.0390
3.80 4.8138 4.8306
4.00 5.7653 5.7820
4.25 7.2429 7.2420
4.50 9.1292 9.0679
4.75 11.551
5.00 14.682
5.25 18.763
5.50 24.136
5.75 31.284
5.90 36.705
a,b,c
See footnotes. Table II.
282 Pitzer, Wang, Rard, and Clegg

REFERENCES
1. K. S. Pitzer, J. Phys. Chem. 77, 268 (1973).
2. K. S. Pitzer and G. Mayorga, 7. Phys. Chem. 77, 2300 (1973).
3. H. L. Friedman, Ionic Solution Theory, Chap. 3 (Wiley-Interscience, New York, 1962).
4. K. S. Pitzer and J. J. Kim, J. Am. Chem. Soc. 96, 5701 (1974).
5. K. S. Pitzer, in Activity Coefficients in Electrolyte Solutions, 2nd edn., Chap. 3, K. S.
Pitzer, ed. (CRC Press, Boca Raton, FL, 1991).
6. C. E. Harvie and J. H. Weare, Geochim. Cosmochim. Acta 44, 981 (1980).
7. V. K. Filippov and V. I. Nokhrin, Russ. J. Inorg. Chem. 30, 282, 289 (1985).
8. V. K. Filippov, N. A. Charykov, and V. Rumyantsev, Dokl. Akad. Nauk, SSSR, Fiz. Khim.
296, 665 (1987) (Engl. Transl. p. 936).
9. R. G. Anstiss and K. S. Pitzer, J. Solution Chem. 20, 849 (1991).
10. J. A. Rard and S. L. Clegg, J. Chem. Eng. Data 42, 819 (1997).
11. J. Ananthaswamy and G. Atkinson, 7. Chem. Eng. Data 30, 120 (1985).
12. D. G. Archer, J. Phys. Chem. Ref. Data 20, 509 (1991).
13. D. G. Archer, 7. Phys. Chem. Ref. Data 21, 793 (1992).
14. S. L. Clegg, J. A. Rard, and K. S. Pitzer, J. Chem. Soc. Faraday Trans. 90, 1875 (1994).
15. R. C. Phutela and K. S. Pitzer, J. Solution Chem. 12, 201 (1983).
16. N. A. Hewish, G. W. Neilson, and J. E. Enderby, Nature 297, 138 (1982).
17. J. A. Rard, A. Habenschuss, and F. H. Spedding, J. Chem. Eng. Data 22, 180 (1977).
18. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, 2nd edn. revised, (Butterworths,
London, 1965).
19. D. G. Archer and P. Wang, J. Phys. Chem. Ref. Data 19, 371 (1990).
20. R. N. Goldberg and R. L. Nuttall, J. Phys. Chem. Ref. Data 7, 263 (1978).
21. J. A. Rard and D. G. Miller, J. Chem. Eng. Data 26, 33, 38 (1981).
22. R. H. Stokes, Trans. Faraday Soc. 41, 642 (1945).
23. R. A. Robinson and R. H. Stokes, Trans. Faraday Soc. 36, 733 (1940).
24. R. A. Robinson and V. E. Bower, 7. Res. Natl. Bur. Stand. U.S. A70, 305 (1966).
25. R. F. Platford, 7. Phys. Chem. 72, 4053 (1968).
26. Y. C. Wu, R. M. Rush, and G. Scatchard, J. Phys. Chem. 72, 4048 (1968); 73, 2047 (1969).
27. Yu. G. Frolov, V. P. Nikolaev, M. Kh. Karapet'yants, and K. K. Vlasenko, Russ. J. Phys.
Chem. 45, 1054 (1971).
28. J. Padova and D. Saad, 7. Solution Chem. 6, 57 (1977).
29. D. Saad, J. Padova, and Y. Marcus, 7. Solution Chem. 4, 983 (1975).
30. F. Kuschel and J. Seidel, 7. Chem. Eng. Data 30, 440 (1985).
31. H. F. Gibbard, Jr., and A. F. Gossmann, 7. Solution Chem. 3, 385 (1974).
32. H. S. Harned, in The Structure of Electrolytic Solutions, Chap. 10, W. J. Hamer, ed., (Wiley,
New York, 1959).
33. K. S. Pitzer and G. Mayorga, 7. Solution Chem. 3, 539 (1974).

Você também pode gostar