You are on page 1of 18

Gaxhimica et Cosmochimica Acfo Vol. 54, pp. 3265-3282 0016-7037/90/$3.00 + .

w
Copyright 0 1990 Pergamon F’ress pk. Printed in U.S.A.

The influence of pressure on the activity coefficients of the solutes and on the solubility
of minerals in the system Na-Ca-Cl-S04-Hz0 to 200°C
and 1 kbar, and to high NaCl concentration
CHRISTOPHE MONNIN *
CNRS, UniversitC Paul Sabatier Laboratoire de Gtochimie, 38, rue des Trente-Six Ponts, 3 1400 Toulouse, France

(Received March 6, 1990; accepfed in revisedform September 27, 1990)

Abstract-A model is presented which is used to calculate the effect of pressure on activity coefficients
of aqueous solutes in the system Na-Ca-Cl-S04-HZ0 to 200°C. Literature data for the density and com-
pressibility of aqueous binary solutions of Na2S04 and CaC12to 200°C are used to calculate the first and
second pressure derivatives of Pitzer’s ion interaction model parameters, as well as the standard molal
compressibility and volume of these two salts. Empirical correlations between the apparent molal volume
and compressibility of the aqueous electrolytes are used to guide the choice of the temperature dependent
expressions used for the numerical representation of the derivatives of Pitzer’s parameters with respect
to pressure. For sodium sulfate solutions, such correlations are used to extrapolate compressibilities to
200°C. The change in the thermodynamic properties of thcCaSO2 ion pair with pressure is taken into
account by the variation of its dissociation constant. The volumetric properties (partial molal volumes
and compressibilities) of multicomponent solutions in the Na-Ca-Cl-S04-Hz0 system can be predicted
from the information generated here and the volumetric equations of ROGERS and PITZER ( 1982)
for NaCl.
This model is then combined with the high temperature model of MOLLER( 1988) of the same system
in order to calculate activity coefficients at high pressures to 200°C. The resulting model is validated by
comparing calculated and measured solubilities of anhydrite and gypsum in pure water and in NaCl
solutions up to 6 M. The agreement between the calculated and measured solubilities of the calcium
sulfates is typically better than 10% up to 200°C and 1 kbar. The relevance of temperature and pressure
corrections to the activity coefficients of aqueous solutes is discussed in regard to the assumed accuracy
with which geochemical models are able to calculate mineral solubilities.

I. INTRODUCI’ION experimental data, especially the solubility of calcium sulfate


salts in common ion ternary solutions.
A NEW GENERATIONOF aqueous solution models based on
The modelling of the influence of pressure on the ther-
Pitzer’s ion interaction approach has recently been developed
modynamic properties of such a system then requires the
and applied to a variety of low temperature geochemical
evaluation of pressure effects on the activity coefficients of
problems ( see WEARE, 1987, and references therein ) . These
typical strong electrolytes like Na2S04 and CaClz, and on
models are able to accurately predict mineral solubilities in
that of weak electrolytes like CaS04. MONNIN ( 1989) has
complex systems over a wide range of composition to high
shown that Pitzer’s approach can be used to calculate the
concentrations. The success of this approach has triggered
densities of the solution and the partial molal volumes of
continuing efforts to extend such models to higher temper-
strong electrolytes in the Na-K-Ca-Mg-Cl-SO, -CO9 -HC03 -
atures ( PABALANand PITZER, 1987; MOLLER, 1988; GREEN-
Hz0 at 25°C. In this paper, this approach is extended to
BERGand MOLLER, 1989) and to minor and trace elements
higher temperatures for the Na-Ca-Cl-SO4 -Hz0 system.
( MONNIN and GALLINIER, 1988). The purpose of this paper
Measured densities of binary solutions of Na2 SO, and CaC12
is to incorporate the effect of pressure into the model for the
taken from the literature are used to build temperature de-
“simplified seawater” system Na-Ca-Cl-S04-Hz0 at high
pendent equations for the standard molal volumes and for
temperature built by MOLLER( 1988 ) from free energy data.
the first pressure derivatives of Pitzer’s interaction parameters
At 25°C HARVIE et al. (1984) have been able to model
for the activity coefficients. Consideration of the compres-
this system within the strict ion interaction approach: all
sibility of the binary solutions allows the calculation of the
electrolytes are considered as strong and no ion pairs need
standard molal compressibility of the salts as well as the sec-
be included. But at higher temperature, the decrease of the
ond pressure derivatives of the interaction parameters. From
dielectric constant of the solvent enhances ion pairing.
the knowledge of the variation of the stability constant of the
MOLLER( 1988) has found that the CaSO$( aq) ion pair was
CaSOi(aq) ion pair with pressure, one can calculate the dis-
clearly required to accurately reproduce all of the considered
tribution of species at high pressure in the Na-Ca-Cl-SO.+-
Hz0 system. With this information, the influence of pressure
* Present address: Department of Geology and Geophysics, Yale on the activity coefficients of the solutes in multicomponent
University, New Haven, CT 065 1 l-8 130, USA. solutions can be evaluated. It is shown how the model can
3265
3266 C. Monnin

be used to predict the solubility of calcium sulfate salts in For the standard conditions, one has
concentrated NaCl solutions. Finally, the paper discusses the
relevance of pressure corrections for the activity coefficients FTC_ aVO
of aqueous solutes in regard of the general accuracy with i ap i T’
which geochemical models can predict mineral solubilities.
The apparent molal compressibility is a measure of the
difference between the total compressibility of the solution
II. MODEL EQUATIONS
and that of the solvent:
The variation with pressure of yhlx, the activity coefficient
KT - 1OOOK,q = wr- 1~OPwLxv
of a given solute MX in a complex solution, can be calculated +Kr = (9)
m m
by integrating the following relation:
It is thus possible to express the compressibility of a solution
(1) of any complexity as a function of the solute concentrations
simply by taking the derivative of the volume equation
in which vMx is the stoichiometric number of ions of the ( MONNIN, 1989) with respect to pressure. For a binary so-
electrolyte MX, vMx is the partial molal volume of the solute lution, Pitzer’s ion interaction model leads to
MX in the solution and pMx its standard molal volume, which ~~=iP---f~~- 2vMvxRTm(Bgx + 2(v~zMImCfix)
depends only on temperature and pressure. R is the perfect
gas constant and T the absolute temperature. (10)

The pressure dependence of the partial molal volume of with


MX can be obtained from the knowledge of its partial molal
compressibility K,, :

a%,,
c-1
ap T
= -&fx (2)
and

All the necessary equations and definitions for the specific


volume of the solution and the partial molal volume of the
(12)
solutes in Pitzer’s formalism are given in an earlier paper
( MONNIN, 1989) and will not be repeated here. The various AK is the Debye-Hiickel slope for the compressibility. The
expressions for the compressibility are totally symmetric with second virial coefficient is
those for the volumes but nevertheless require some specific
definitions.
The coefficient of isothermal compressibility & can be 7

expressed as a function of V, the total volume of the system,


as = P gi” + &$“g( a,@+&i”g( a2fi) ( 13a)

&_f
($ 1. T
(3) g(x) = f ( 1 - ( 1 + x) exp( -x)). (13b)

Kr is the isothermal compressibility of the system: The superscript K indicates that the labeled quantity is the
first derivative with respect to pressure of the corresponding
= v&. (4) quantity for the apparent molal volume and thus the second
derivative with respect to pressure of the corresponding
quantity for the activity coefficient:
The specific volume of an aqueous solution is
lOOOv, + me,
(5)
XK=(tg),=($qT (14)
‘= lOOO+mM,

in which v, is the specific volume of water, 9, the apparent x being &:, &,$, a:&, or CMx. The other parameters (b,
molal volume of the solution, m the molality of the solute, (Y, , az) have the usual significance (see PITZER, 1987); that
and MS its molar mass. Expressions (4) and ( 5 ) lead to is, b = 1.2 for all electrolytes, (Y, = 1.4 and (Ye= 12 for a
2-2 salt and LY,= 2.0 and a2 = 0 for salts of lower valence
(6) type.
The expressions of the excess compressibility of a multi-
where &r is the apparent molal isothermal compressibility. component solution and of the partial molal compressibilities
It is proportional to the derivative of the apparent molal vol- of the solute can be easily derived from the above definitions
ume of the solution with respect to pressure: and the corresponding expressions for the volume of the so-
lution and the partial molal volumes of the solutes ( MONNIN,
a&= -
($ 1. T
1989). The conventional partial molal compressibility of an
ion can be obtained by differentiation of the expression of
Effect of pressure on activity coefficients 3261

its partial molal volume with respect to pressure. For a cation Such expressions have been used to describe the variation
M in a multicomponent solution, one gets of the interaction parameters with T as well as that of the
various standard molal properties of the aqueous species or
the standard chemical potentials of the solids (see MOLLER,
1988). The large number of adjustable coefficients allows
- 2RT 2 m.(B&, + ZCK,,) Eqn. ( 19) to represent, as a function of T, the behavior of
a all the parameters investigated in this work. Earlier studies
have shown that Pitzer’s interaction parameters (see, e.g.,
- RT C C mcma(z2,Bz + Iz~lCf + d&-a) ROGERS and PITZER, 1982) and the standard properties of
c a
aqueous species ( TANGERand HELGESON,1988) vary rapidly
- 2RT 2 m,t& - z&RT cc mcmcf EOzl(I) at high temperature when the critical point of water is ap-
c C>C proached and, and at low Tin the vicinity of an anomaly of
the thermodynamic properties of supercooled water. In Eqn.
- RT CC m.m.~(z~EB~r(Z) + J/f&t). (15) (19), the termsas/(680 - T)anda,/( T- 227)areincluded
a>a‘
for their rapid variations in these temperature ranges, without
In a similar way, the partial molal compressibility of the any further theoretical justification. There exist semi-theo-
anion X is retical models of the thermodynamic properties of aqueous
vi
Kx=px-Zz,A,Kp+?ln(l + be)
electrolytes at infinite dilution, the most extensive being those
derived by Helgeson and his coworkers (HELGESON and
4 l+bti b KIRKHAM, 1976; TANGERand HELGESON,1988) which have
- 2RT 2 m,(B& + ZC$) fewer fitting parameters than Eqn. ( 19). The theoretical basis
E of such models provides some guidance for the eventual ex-
trapolation of the standard properties of aqueous electrolytes
- RT C 2 m,m,(ziB~ + Iz~lCf + d&d to high temperatures, which is beyond the scope of this work.
c a
In the temperature range considered here, the purely empirical
- 2RT 2 m,f3Ex - z$RT C 2 mam.tEO”,:(Z) representation of the data with Eqn. ( 19) has been found
a a,*’ sufficient. Some indications on the variations of the inter-
action parameters with temperature and pressure have been
- RT 22 m,m,~(z$EOf$(Z) + I/;~x). (16)
c>c’ empirically found (see section V) and are used to guide the
form of the mathematical representation of the parameters.
In Eqns. (15) and (16), Z = c, m,z,- Bg is the derivative
The values of all the XK parameters (see Eqn. 14) and the
of the second virial coefficient (Eqns. 13a and 13b) with re-
standard molal compressibilities for fixed temperatures can
spect to the ionic strength.
thus be obtained with Eqns. (9) and ( 10) by numerical
Bs = j3$*Kg’(a,fi)/Z + /3$Z’*Kg’(a2fi)/Z (17) regression of experimental data on the compressibility of bi-
with nary solutions as a function of molality. In a similar way,
values of the X” parameters and of the standard molal vol-
umes can be obtained by regression of density data ( MONNIN,
g’(x)=--$[l-(l+x+G)exp(-x)].
1989). These quantities can then be represented as a function
of temperature with the aid of Eqn. ( 19). The coefficients
“f$‘( I) is the derivative with respect to the ionic strength of
resulting from the analysis ofthe available experimental data
the electrostatic function E@(Z). These electrostatic terms
are given in Table 1 for the volume and in Table 2 for the
were included in a previous study on densities and partial
compressibility. However, before studying the properties of
molal volumes ( MONNIN, 1989), and they will be ignored
solutions, one must know the thermodynamic properties of
here. The partial molal compressibilities of the mean salts
the solvent, for use in the expressions of the apparent molal
can then be calculated by additivity of the conventional partial
volumes and compressibilities.
molal volumes of the ions. If v,,, and vqcaare the stoichio-
metric coefficients of the ions c and a in the salt ca, then one
has III. VOLUMETRIC PROPERTIES OF WATER

K- = v,,& i- v,,Ka. (18) The equation of state derived by HAAR et al. ( 1986) is
now a commonly accepted standard for the calculation of all
In Pitzer’s formalism, there is no indication as to how the the thermodynamic properties of water. However, its for-
parameters of the model vary with temperature or pressure. mulation (Helmoltz free energy in temperature-density co-
In papers published to date (see, e.g., ROGERS and PITZER, ordinates) makes it inconvenient for routine calculations.
1982; PABALANand PITZER, 1987; MOLLER, 1988), the pa- KELL ( 1975 ) provides rational expressions of the properties
rameters obtained from fits at fixed temperature or pressure of water as a function of the common variables temperature
have been represented by empirical expressions of the fol- and pressure from which the density and compressibility can
lowing type: be easily obtained. Unfortunately, they are limited to tem-
Parameter(T) = a, + azT + aST2 + aT3 + a51n T peratures below 150°C. It has thus been necessary to generate
such convenient expressions of the volumetric properties of
+ &/T + a,/( T - 227.0) + as/(680.0 - T). (19) water for higher temperatures. This has been done by fitting
3268 C. Monnin

values of the density or the compressibility obtained from Between 150 and 300°C the isothermal compressibility
the HGK equation to simple empirical polynomials. of water can be calculated by Eqn. (24)) obtained in a way
similar to that for the density (Eqn. 2 1):
A. Density
pr = -4.514512 10e4 + 1.498765 lo-?
KELLY( 1975 ) gives the following expression of the density
of water pwin kg/m 3 as a function of temperature ( t in degree - 1.778673 10m7r2+ 1.043988 10-9r3
C) up to 150°C: - 2.999736 10-‘Zr4 + 3.453965 IO-r5r5 (24)
pw = (999.83952 + 16.945176~ - 7.9870401 10-3t2 with u = 5.1 lo-’ bar’, i.e., roughly 1%.
- 46.170461 10-6t3 + 105.56302 10-9t4
C. Adiabatic Compressibility
- 280.54253 10-‘2tS)/( 1 + 16.87985 10-3t). (20)
It is also necessary to analyse sound speed data (see the
From 150 to 300°C density values along the liquid-vapor appendix). The adiabatic compressibility /3~of a medium of
equilibrium curve generated by the Haar, Gallager, and Kell density p is related to the speed of sound U by
(HGK) equation have been fit to
p,,, = 1.0210864 - 6.915445 10-4t + 2.828443 1O-6~2 Ps=--$. (25)

- 3.131825 lo-Rt3 + 1.015801 IO-Iat4 The speed of sound lJo (in m/s) in pure water has been
- 1.373334 10-Y (21) measured between 0 and 100°C by DEL GROSSOand MADER
( 1972), who give the following expression:
with a standard deviation of 1.6 10e5 g/cm3.
U. = 1402.338 + 5.0371 lr - 5.80852 lo-*r2
B. Coefficient of Isothermal Compressibility + 3.342 10-4r3 - 1.478 10-6r4 + 3.1464 10-9r5. (26)

KELL( 1975) gives two expressions of & versus temper- In this work, it has not been necessary to evaluate U. above
ature. The first is recommended for temperatures below 100°C.
100°C:
IV. DEBYE-HtkKEL SLOPES FOR THE APPARENT
& = (50.88496 + 0.61638t + 1.459187 10-3t2 MOLAL VOLUME AND COMPRF2&IBILITY

+ 20.08438 10-6t3 - 58.47727 10-9t4 The solvent properties appear in the expression of the ap-
parent molal quantities and in the Debye-Htickel limiting
+ 410.411 10-‘2t5)/( 1 + 19.67348 l0-5t). (22)
law slopes. The Debye-Hiickel constants depend directly on
Between 100 and 150°C Eqn. (23) should be used: temperature and pressure, and on the thermodynamic and
dielectric properties ofthe solvent, but also on their derivatives
Pr = (50.884917 + 0.62590623r + 1.3848668 10-3rZ ,with respect to T and P. While there is a general agreement
about the choice of the HGK equation as an international
+ 21.603427 10e6r3 - 72.087667 10m9r4
standard for the thermodynamic properties of water, the
+ 465.45054 lo-i2r5)/( 1 + 19.859983 10s5r). (23) consensus is not so wide for its dielectric properties. Several

Table 1. Fitting coefficients (eq. 19) for the standard nolal volume and the interactioncoefficients
for the al went molal volume (efxx stands for 10fXx).

aI a* =n a6 a7

8.525003e-2 -3.581619 7.234513e-3 .5.839699e-6 0. -7.515469et4-3.007338et2 0.


5.369952e-5-2.653801e-' 3_625554e-10 0. 0. 0. 2.202016e-3 -2.68293e-2
1.145144e-5-4.527545e-1 4.34633e-11 0. 0. 0. -2.165713e-4 .595239e-4

1.281259et3 -3.292342 4_267199e-3 0. 0. -1.231424e+5-l.O67946e+3 9.132116etl

5.3088e-5 4.33707e-6 -3.262413e-9 0. 2.70953e-4 1.42266e-1 4.51986e-3 0.

3.200188e-3-l.O92875e-I -7.101661e-9 0. 0. -8_373935e-1 4.901041e-2 0.

/ 0. 0. 0. 0.

I I
1.752e-6 -8.55e-2 0. 0.

4.175OOOe+l 4.121094e-I -2.841949e-4 0. 0. 0. -5.545000et2 4.352001e+l


r - 0.
1: / ,“Il_lY;:::::::Y
9.949027e-5 4.549863e-' -3.008955e-1C 1.201352e-2

2.34138e-5 -1.405386e-' 1.971700e-10 ,2.146539e-1: 0.

(a) from ROGERS and PITZER(1982). (b) coefficientsof equation 34 (with c,= ai).
Effect of pressure on activity coefficients 3269

Table 2. Fitting coefficients (eq. 19) for the standard nolal compressibility end the

interaction coefficients for the apparent molal compressibility (efxx stands for 10f”‘).

NaCl (a) i 1.266280e-1 -3.941704e-4 9.678052e-7

pc0,*r -l.l17349e-7 3.49265e-10 -8.357192e-13

Na, SO, K* (b) -2.33e-2 7.2e-4 0. 0. 0.

!3’O’*‘L (cl 2.4e-8 -l.l4e-3 0. 0. 0.

CaCl, K” -1.627103e-1 9.5793e-4 -1.466741e-6 -l.l8024e-1 0.

13~~‘*r (d) -3.456839e-8 1.536791e-10 -1.821653e-13 0. I 0.


(a) from ROGERS and PITZIIR (1982). (b) coefficients of eq. ~2 (with k,= a, 1.
(c) coefficients c i eq. 33 (with b,= aI). (d) valid above 8O.C (see text).

sets of values for the Debye-Hiickel slopes have been calcu- The analysis of the data of GATES and WOOD ( 1989) for
lated from various equations for the dielectric constant of CaC12 requires a knowledge of A, at high pressures for a few
water (mainly those of BRADLEY and PITZER, 1979, and of temperatures. The needed slopes have been calculated from
UEMATSU and FRANCK, 1980). ANANTHASWAMY and AT- the values reported by BRADLEY and PITZER (1979) at
KINSON ( 1984) and BEVER and STAPLES ( 1986) have recently rounded pressures.
tried to normalize the values of these parameters. Both these In a similar way, A,, the Debye-Huckel slope for the ap-
authors have used the HGK equation for the thermodynamic parent molal isothermal compressibility, can be calculated
properties of water, but retain different expressions for the by the following expressions:
dielectric constant: the Bradley-Pitzer formulation for An-
anthaswamy and Atkinson, and the Uematsu-Franck equa- l between 0 and lOO”C, and from the values of Ananthas-
tion for Beyer and Staples. These two equations reproduce wamy and Atkinson (u = 1.6 10e6),
the data for the dielectric constant with comparable accu-
racies, but lead to different values for the derivatives of e AKr = 2.160213 1O-2 - 2.151289 10-4T
versus T and P, which lead to differences in the values of the
Debye-Htickel slopes. These differences are, at most, 25% for + 7.258216 IO-‘T2 - 8.437141 IO-“T3; (29)
A+, the slope for the activity and osmotic coefficients, but
they can be as large as 50% for AK, the slope for the com- l from 100 to 300°C with the Bradley-Pitzer values (u = 2.8
pressibility. In the absence of a general agreement on that 10-4),
topic, the present work retains the values of the slopes gen-
erated from the Bradley and Pitzer equation for the dielectric & = -2.94314 + 4.932726 lo-‘T
constant of water.
- 2.760752 10-4T2 + 7.908295 lo-‘T3
BRADLEY and FITZER ( 1979) give tables of the various
slopes up to 350°C and 1 kbar. These have been recalculated - 1.128366 10-13T“ + 6.623415 10-‘3T5
with more significant figures up to 100°C by ANANTHAS
WAMY and ATKINSON ( 1984). These values have thus been - 3.156183 lo2 (30)
fit here to the polynomials given below: 680.0 - T .

l from 0 to lOO”C, the values of& the slope for the apparent Sound speed measurements in electrolyte solutions require
molal volume given by Ananthaswamy and Atkinson, can knowledge of the Debye-Hiickel slope for the apparent molal
be represented with a standard deviation of 4.1 1O-4 by adiabatic compressibility. The relationship to its isothermal
counterpart is ( MILLERO, 1979)
A, = 8.106377 - 1.256008 IO-‘T+ 7.760273 10-4T2

- 2.098163 10-6T3 + 2.25777 lo-‘T4; (27)


&r=&+(Br,w-&S,w) , (31)
l from 100 to 3DO”C, the Bradley-Pitzer values for A, are fit
to
an expression in which AE and AC- are the Debye-Hiickel
A, = 3.849971 1O+2- 6.982754T + 3.877068 10-2T2 slopes for the coefficient of thermal expansion and the heat
capacity. MILLERO ( 1979) indicates that the difference A,
- 1.11381 lo-‘T3 + 1.589736 lo-‘T4 - AKs is equal to 0.6 10m4 cm3 - kg’j2. mol-3/2 - bar-’ for
6.469241 lo4 25°C. So, for this temperature, one has (ANANTHASWAMY
- 9.395266 lo-‘IT5 + (28) and ATKINSON, 1984)
680.0 - T ’
with (r = 4.07 10m2( T is the temperature in Kelvin). A, = -3.78 10e4 cm3 - kg’j2 - mol-3/2 -bar-‘,
3270 C. Monnin

thus leading to
A, = -4.37 10s4 cm3-kgif2 .molW3’* *bar-‘.

V. EMPIRICAL CORRELATIONS BETWEEN THE


APPARENT MOLAL VOLUMES AND
COMPRESSIBILITIES OF BINARY SOLUTIONS

HELGESON and KIRKHAM (1976,p. 182)and MILLERO


et al. (1987) have shown that the standard partial molal
compre~ibility of an aqueous electrolyte is correlated to its
standard partial molal volume. MKLERO et al. ( 1987) have
shown that the following relation holds for sodium and mag-
nesium sulfates and chlorides:

Moreover, MILLERO ( 1979) has shown that the excess


partial molal volumes V - 8’ for electrolytes in seawater are
proportional to their excess partial molal compressibilities R
- p. These two remarks suggest that the apparent molal
volume and compressibility are then correlated. In Fig. 1,
the apparent molal adiabatic compre~ibility of Na2S04 ob-
tained from the data of MILLERO et al. ( 1987) is plotted
versus its apparent molal volume for three temperatures. The
same graph for CaClz (Fig. 2) uses the data of MILLEROet FIG. 2. Correlation between the apparent molal volume (in cm’/
al. ( 1977a) and of PERMANand URRY ( 1929) for the com- mol) and the apparent molal compressibility (in 104cm3/mol/bar)
pressibility. The apparent molal volumes are calculated by for CaC12.
the equations given in later sections. Figures I and 2 show
that for these two solutes there is a linear reIa~onship between
the one expressed by relation (32) are approximate (see Fig.
(Py and (PK,the slope of which is temperature dependent. In
13 for CaClr and Fig. 2 in MILLEROet al., 1987) and only
Pitzer’s model, such relationships must correspond to a cor-
indicate tendencies. Lastly, a previous study ( MONNIN, 1987)
relation between the interaction parameters for the apparent
has shown that the variations of i3(‘)*”and C” with temper-
molal volume (i.e., @co)~”and C”) and their counterpart for ature are linked by the following relation:
the compressibility (i.e., @(o)*x and CK). This has been found
to be the case for Na2S0,, for which one can write c‘” = c, i- C*/3’0’~“_ 134)
p(OLK= b, + ~*/+W. (33) Here it has been convenient to use this relation for Na2 S04,
but not for CaC12, for which accurate high temperature data
Such relationships can be used to estimate the compressibility exist (see below).
in a temperature range where only the density is known. It
should be remembered, however, that correlations such as VI. VOLUME~C PROPERTIES OF AQUEOUS
~L~ONS OF NaCI

The volumetric properties of NaCl(aq) between 0 and


300°C and to 1000 bars have been studied by ROGERSand
PITZER ( 1982 ), who, however, did not give an expression to
directly calculate vga,-, but only tables of values. Between 0
and 2OO”, their values have been fitted to Eqn. ( 19). The
expressions for /3$6 and Cga;,, reported in Table 1 come
directly from their paper.
In the same way, their values for the standard molal com-
pressibility between 0 and 200°C have been adjusted with
Eqn. 19 (Table 2). The required expressions for ,L?$f and
C&c, have been obtained by taking the pressure derivative
of their expressions of @$$ and C&c,.

VII. VOLUMETRIC PROPERTIES OF BINARY


SOLUTIONS OF NaZSO,
10 15 al 2s 0”
A. Density
FIG. 1.Correlation between the apparent molal volume (in cm 3/
mol) and the apparent molal ~rn~~~ibility (in 104cm3/mol/~r) Density data at 25°C have been evaluated previously
for Na2 SO,. (MONNIN, 1989). Between 0 and 100°C. Miller0 and his
Effect of pressure on activity coefficients 3271

coworkers ( MILLERO and KNOX, 1973; CHEN et al., 1977,


1980; Lo SURDO et al., 1982; CONNAUGHTONet al., 1986)
have provided highly accurate data, better than any other
data in this temperature range (see Fig. 3 ) . These data were
evaluated within Pitzer’s ion interaction formalism by CON-
NAUGHTONet al. ( 1986), who gave expressions for the tem-
perature dependence of the interaction parameters valid to
1OO’C. The model has been parameterised again in order to
include the few data available above 100°C: those of ELLIS
( 1968 ) and of PEpINov et al. ( I 985 ) . The data of I&AIBULLIN
and NOVIKOV ( 1973 ) is aberrant (Fig. 4) and was not in-
cluded. The data used in the regression are represented by
an asterisk in the captions to Figs. 3 and 4. Only the data for
which the influence of pressure can be neglected (P < 20
bars) have been considered in the fit. Preliminary calculations
based on compressibility estimates (see below) have shown
that the variation of the standard partial molal volume at
. ,
100°C for a pressure of 20 bars is lower than 0.5 cm3/mol
and that the variation of the interaction parameters is neg-
0 1”
I

m
I

ligible. It is, however, important to calculate the density of FIG. 4. Apparent molal volume (in cm3/mol) of Na2S04at 100°C
water at 20 bars to obtain the apparent molal volumes. High versus the square root of the molality (filled circles: International
Critical Tables. 1928; inverted triangles:KHAIBULL~N and NOVIKOV,
pressure density data have been used as a check of the model
1973; open circles: EuIs, 1968,* for 20 bars, squares: FEPINOVet
(see below). al., 1985; triangles: FABUSSet al., 1966; black triangles: KOROSIand
Without the && parameter, data below 100°C cannot FABUSS,1968 ). The asterisk indicates data selected for the calcula-
be fit adequately. On the contrary, it is not needed to repro- tions.
duce the data of ELLIS ( 1968 ) and PBPINOVet al. ( 1985 ) up
to 200°C. The coefficients reported in Table 1 have been
calculated by first fitting to Eqn. ( 19) values of vkasso, ob-
175’C, and 510.10m6 at 200°C. The data of FEPINOVet al.
tained for each temperature between 0 and 200°C. Then, the
( 1988) is less accurate (u N 3.0 10m4).
standard molal volume is forced to the values given by this
new expression and the interaction parameters are calculated.
The selected data below 100°C are reproduced with a stan- B. Compressibility
dard deviation ofabout lo-’ cm3/g, while above lOO”C, the MILLEROet al. ( 1987) used Pitzer’s model to analyse their
data of ELLIS( 1968 ) are calculated with standard deviations sound speed data, which they converted to isothermal com-
of 75.10e6 cm3/g at 125°C 120.10-6 at 15O”C, 260.10m6 at pressibilities. Their equations have here been used to generate
values of the compressibility up to IOO’C, which have then
been fitted to the expressions below in order to extrapolate
the compressibility to 200°C using the correlations described
in section V. The relationship between k and r” given by
these authors (Eqn. 32, Table 2) has been retained. /?g&
has been found to be related to @$&, by Eqn. (33), the
coefficients of which are given in Table 2. The fit to the data
below 100°C is better than 1%. Correlations like those ex-
pressed by Eqns. (32) and (33) allow one to estimate the
compressibility of Na* SO4 solutions to 200°C because of the
availability of F&o, and &&,, at these temperatures.

C. High Pressure Density


The specific volume at high pressure can be obtained by
integration of the compressibility. If the compressibility is
independent of pressure, then
P
v(P) - o(P0) = - K,dP = -K,(P - PO). (35)
0 a5 s PO
FIG. 3. Apparent molal volume (in cm3/mol) of Na2S04 at 50°C Comparison between high pressure specific volumes cal-
versus the square root of the molality (f&d circles: International culated by Eqn. (35 ) and measured values allows a test of
Critical Tables, 1928; open triangles: MILLERO and KNOX, 1973*;
filled triangles: ELLIS,1968, for 20 bars, inverted triangles: CHENet the method of calculation of the compressibility and an es-
al., 1977*; black squares: MIKHAILOV et al., 1957). The asterisk timate of the variation of KT with pressure. CHEN et al. ( 1977 )
indicates data selected for the calculations. have measured the apparent volume of Naz SO, solutions at
3272 C. Monnin

low concentration (molalities between 0.009 and 0.33 M)


for pressures up to 1000 bars at temperatures of 0, 25, and
50°C. In Figs. 5 and 6 the variation of the apparent molal
volume versus pressure are shown for solutions of molality
0.148 M (Fig. 5) and 0.335 M (Fig. 6), for the three cited
temperatures. The straight lines are calculated by Eqn. (35 ).
At 0°C the predicted apparent volume is in good agreement
with the measurements up to about 500 bars. The difference
at higher pressures could be consistent with a decrease in KI
when the pressure increases. But, at 25 and 50°C the data

.;1/1_
of Chen et al. ( I977 ) roughly plot on a straight line, the slope
of which (which is the apparent molal isothermal compres-
sibility) is in accordance with the calculation for the 0.335
M solution (Fig. 6) but not with the 0.148 M solution (Fig.
5). These results as well as those obtained for the other so- 0 500 1
P bar
lutions studied by these authors (not shown here) do not
allow any firm conclusions to be drawn about the variation FIG. 6. Variation of the apparent molal volume of a Na2S0, so-
of the compressibility with pressure in the considered pressure lution (molality = 0.335 M) as a function of pressure (experimental
data Of CHEN et al., 1977, at O”C,circles; 25”C, squares; and 5O”C,
range. It nevertheless seems reasonable to take KT indepen- triangles). The lines are calculated independently of the data.
dent of P.

VIII. VOLUMETRIC PROPERTIES OF BINARY


AQUEOUS SOLUTIONS OF G&l2 data supplement the older measurements made by ELLIS
( 1967) for low concentrations to 200°C.
A. Densities at 1 Bar All of these data (converted to apparent molal volumes)
are plotted in Figs. 7, 8, and 9 versus the square root of the
In an earlier paper (MONNIN, 1987), literature density data
ionic strength for 50, 125, and 175°C. Figure 7 shows that,
of CaQ was compiled and evaluated with Pitzer’s ion in-
for 50°C Kumar’s data is not consistent with the general
teraction approach. Because of a lack of data, this study was
trend at high concentration. In Fig. 9 one can also see the
limited below 1OO’C. Since this paper, new data have been
published ( KUMAR, 1986; GATES and WOOD, 1989) which discrepancy with the data of GATES and WOOD ( 1989) at
allow the extension of the composition, temperature, and 35.5 bars, which cannot be due to the different pressures of
pressure ranges of this earlier study. The most complete set both sets of data (35.5 and 20 bars). Attempts to fit all the
of measurements is that of GATES and WOOD ( 1989) who data sets together were unsuccessful. So Gates and Woods
give densities of CaClz solutions up to 327°C 400 bars, and
6.4 M. The data of KUMAR (1986) extend to 200°C for the

1
1-
same concentration range and a pressure of 20 bars. These

25-
0

51 FIG. 7. Apparent molal volume of CaClzat 50°C as a function of


0 500 P bar 1000 the square root of molality (filled hexagons: PERMAN and URRY,
1929;filled circles:BOCATYKHand EVNOVICH,1972; black triangles:
FIG. 5. Variation of the apparent molal volume of a Na2S04 so- InternationalCritical Tables, 1928; black squares: ELLIS, 1967, for
lution ( molality = 0.148 M ) as a function of pressure (experimental 20 bars;triangles: ASKMANOVICand KREY, 1970; diamonds: KUMAR,
data OfCHEN et al., 1977,at 0°C circles; 25”C, squares; and 50”C, 1986, for 20 bars; open circles: GATES and WOOD, 1989*). The
triangles. The lines are calculated independently of the data. asterisk indicates data selected for the calculations.
Effect of pressure on activity coefficients 3213

Extrapolation of the straight lines of Figs. 10 and 11 for tem-


peratures above 100°C yields values for the standard molal
volume as well as the interaction parameters for a I bar hy-
pathetical pressure, which have been fit with the values at 1
bar below 100°C to Eqn. 19.

B. Compressibility

In Figs. 10 and 11, one can see that at 50,76, and 127”C,
the, standard molal volume of CaC12 and the parameter
P &yZ are linear functions of pressure up to 400 bars. The
slopes of the lines of Fig. 10 are proportional to the standard
molal compressibility, which is then independent of pressure
at least up to 127’C and 400 bars. At 177”C, there are only
two points in Fig. 10 from which no conclusion about the
variation of KT(CaCIZ) with P can be drawn. The value of
j%Q obtained at this temperature is in good agreement with
the trend deduced from the data at lower T (Fig. 12). At
25“C, the values of the standard molal volumes at high pres-
sure obtained from the fits of the density data of GATES and
WOOD ( 1985 ) are compared to those predicted from the val-
ues of the standard isothermal molal compressibility (see the
appendix) and of the standard molal volume at 1 bar pre-
viously determined ( MONNIN, 1989). The agreement is very
good.
In a similar way, the values of &$$ obtained from the
I I I _.-J
1 2 densities at high pressure are plotted in Fi 1la. Let us con-
Jiii ?
sider first the temperatures above 25°C. /3:& varies linearly
FIG. 8. Apparent molal volume of CaCl, at 125°C as a function with pressure in the considered pressure range. The value of
of the square root of the molality at 20 bars (squares: KUMAR,1986; /3:&c (which is the derivative of ag!& with respect to pres-
open circles:ELLIS,1967;black triangles: GATES and WOOD, I989* ) . sure) can be calculated from the slopes of these lines. For
The asterisk indicates data selected for the calculations.
these temperatures, the parameter C&J, can be set equal to
zero without any degradation of the fit. On the contrary,
CLa, is clearly needed to fit the data at 25’C. It nevertheless
data have been preferred over any other because of their con-
sistency and their extensive P-T-m ranges. A main difference
in the database between Na2 SO4 and CaCIZ is the availability
of extensive high pressure density data for calcium chloride.
In effect, Gates and Wood give densities up to 400 bars at
25°C (GATES and WOOD, 1985), later supplemented by
measurements at higher temperatures by the same authors
(GATES and WOOD, 1989). But, because of the limited data
available for sodium sulfate, only the data below 2OO’C has
been considered.
Isothermal-isobaric fits of the data of GATES and WOOD
(1985, 1989) have been performed to obtain values of the
standard molal volume of calcium chloride and Pitzer’s in-
teraction parameters. These values are reported versus pres-
sure in Figs. 10 and 11 for several temperatures. The standard
deviations of these fits lie between 1O-4 and 10m3 g/cm3,
which is consistent with the standard deviation of the fit (0.33
kg/m3) of the same data using spline functions cited by
GATES and WOOD ( 1989). An earlier study (MONNIN, 1987)
has shown that the parameter j3&& is not required to fit the
data for this system. It was necessary to include the parameter
C&a, in the regression only at low temperature (below 50°C)
and at low pressures (below 300 bars). Although the param-
eters j3(‘)‘”and C” are strongly correlated ( MONNIN, 1987),
FIG. 9. Apparent molal volume of CaCIZat 175°C and 20 bars as
it was more convenient, in the present case, to give an expres- a function of the square root of molality (squares: KUMAR,1986;
sion of C” as a function of T, instead of j3(‘)*‘, as done in circles: ELLIS,1967; black triangles: GATESand WOOD, 1989*, for
MONNIN ( 1987 ) or for sodium sulfate in the present work. 35.5 bars). The asterisk indicates data selected for the calculations.
3274 C. Monnin

“at best” through the points has a slope comparable to that


determined by MILLERO et aI. ( 1987) for NaCl, Na2S0.,,
MgCl, , and MgSOd. Such a linear correlation does not allow
an accurate representation of the data and can be used only
to guess values of c, as done above for Na2S04. As such,
values of the standard isothermal compressibility of aqueous
calcium chloride between 25 and 177°C have been fit to
expression ( 19), but the sole values of /3:2&t above 50°C
have been used in the fit (see Table 2). For the calculation
of the compressibility at 25°C the parameters given in the
appendix are to be used.
As a final check of the above equations, the compressibit
ities of concentrated caicium chloride solutions determined
by PERMANand URRY ( 1929) from compression data have
been compared to the calculated values. The experimental

FIG. IO. Variation of the standard molaf votume of CaCl, with


respect to pressure at various tern~~tu~ (in degrees C). See text
for discussion.

converges to zero with increasing pressure (Fig. 11b). The


dashed lines in Figs. 10 and 11 are predicted from the values
of the parameters at 25°C ( MONNIN, 1989) and that of their
derivatives with respect to pressure obtained from compres-
sibility data (see the appendix). The agreement with the val-
ues obtained from the high pressure densities is reasonable.
Figure 1la shows that the slope of the lines above 25°C
are largely different from that for 25°C. There is a steep de-
crease in the value of @g&t with respect to temperature, which
reaches an afmost constant value above 50°C. KUMAR and
ATKINSON( 1983) report density and sound speed measure-
ments at 5, 15, and 35°C. Their density data have been shown
to be inconsistent with other data (Fig. 3 in MONNIN, 1987 ).
Their adiabatic apparent molal compressibilities can be cor-
rectly fitted with Pitzer’s equations, but without the parameter
@g&2which clearly overtits the data and leads to erratic values
of the standard molaI compressibility, as can be seen in KU-
MARand ATKINSON(1983). The values of J? at 5, 15, and
35°C obtained from these data are not consistent with that
of 25°C although they may be taken as consistent with the
values above 50°C (Fig. 12a). On the contrary, the values
of &&: at these tern~mt~~ smoothly join with those at
high T (Fig. 12b). In this case, the point at 25°C appears
aberrant. It should be kept in mind that this value has been
determined from independent high quality data (sound speeds
of MILLER0et al., 1977a, and high pressure densities ofGATEs
and WOOD, 1985) and cannot be discarded. An experimental
check of Kumar and Atkinson’s dam is first required. These
data have not been included in the final fits. I 1
The standard molaI compressibility of aqueous calcium 4w Pbars
chloride is reported in Fig. 13 versus its standard molal vol- FIG. I I. Variation with pressureof the parameter Sg& at various
ume. One can see that there is a rough correlation between temperatures (in degrees C), and of CLc,, at 25°C. See text for dis-
the two quantities. The straight line (drawn by hand) passing cussion.
Effect of pressure on activity coefficients 3275

where ip stands for ion pair. Aqueous calcium sulfate (com-


posed of the free ions Ca’+( aq) and SO:-( aq)) will be noted
CaS04, while the neutral calcium sulfate ion pair is
CaSOt . Note that in Moller’s model, the activity coefficient
of the neutral species is equal to 1. So, in order to calculate
the thermodynamic properties of the electrolyte CaSO4 at
high pressure, one must know the activity coefficients of the
-200 - free aqueous species Ca2+ and SO:-, as well as the stability
constant of the ion pair at the considered pressure.
The variation of this stability constant with pressure can
be calculated from the standard volume and compressibility
of reaction (36). In their study, FISHER and FOX ( 1979)
-300 - \, compared the values of r$ and of wp determined from
their conductivity measurements at high pressure (to 2000
-I#’ l bars) to the existing literature values, which showed a great
scatter as well as large disagreement concerning the variations
of Arg and flp with T and P. All references nevertheless
indicate an increase of the dissociation constant of CaSOi
with pressure. FISHER and Fox ( 1979) give - 11.7 cm’/mol
for ATi and 30.0 10m4cm3/mol/bar-’ for ti$ at 25°C.
Among others, they rejected the values of MILLERO et al.
( 1977b) based on volumetric measurements. They did not
find a temperature dependence of the standard volume and
compressibility of reaction (36) in the narrow range that they
considered ( IO-30°C). Their values at 25°C have been ten-
tatively retained for the whole temperature range considered
in this work (up to 200°C).
The partial molal volume and compressibility of the free
calcium and sulfate ions can be calculated by the equations
FIG. 12. Variation with temperature of the standard molal com- given in section II and the interaction parameters determined
pressibility of CaClz and of the parameter fig)& These quantities
have been obtained from the high pressure densities of GATESand in sections VI, VII and VIII. Moller’s model for the activity
Wool ( 1989; squares) and from the sound speed data of KUMAR coefficient of CaS04 includes non-zero values of /3(O),/3(l),
and ATKINSON ( 1983; triangles). The filled circles are values at 25°C and PC*) for the interactions between Ca’+(aq) and
(see the appendix). The curves are fit to the data (see text). SOi-( One must then estimate the variation of these

data can be reproduced within about 5% for molalities below


5 M. The deviation for the three data points above this con-
centration may increase to 25%. It should be noted that this
older data compare very well with the most recent measure-
ments of GATES and WOOD ( 1989). For example, at 50°C
Perman and Urry give (au/&‘) = -1.61 10m5 cm3/g/bar
for m = 6.4 M, while the value of (Au/ AP) calculated from
Gates and Wood’s data is -1.67 10e5. The present model
leads to - 1.80 lo-‘.

IX. VOLUMETRIC PROPERTIES OF THE WEAK


-200 -
ELECTROLYTE CaSO, (aq )
As stated in the introduction, MOLLER( 1988) has shown
that accurate modelling of the thermodynamic properties of
the Na-Ca-Cl-SO.,-Hz0 system required the introduction of
the ion pair CaSO:( aq), whose dissociation reaction can be
written as
CaSO:( aq) = Ca’+(aq) + SO:-(aq). (36)
The mass action law for this reaction is
FIG. 13. Standard molal compressibility of aqueous calcium chlo-
md+ - mot- - Y&SO, ride versus its standard molal volume (the temperature reported for
Kip =
mip each point is in degrees C).
3216 C. Monnin

parameters with pressure. This could, in principle, be done timated values of all the interaction parameters needed by
from volumetric data for aqueous solutions of calcium sulfate Pitzer’s equations, we can predict partial molal volumes and
as a function of molality. Such data exist ( MILLERO et al., compressibilities of solutes in multicomponent solutions, and
1977b), but the values of the stability constant of CaSOt from that obtain activity coefficients at high pressure. If the
derived from it have been rejected by FISHERand Fox ( 1979 ) , solubility products of the considered minerals are known,
as discussed above. In this study, /3$&o.,,&Lo,, and @g&o, then the solubility can be calculated by simultaneous solution
have been tentatively kept independent of pressure, and their of the mass action law for the reactions of dissolution of the
values at the considered temperature are those given by mineral and of dissociation of the ion pair.
MOLLER ( 1988). As a result, the partial molal volume and
compressibility of CaSO,(aq) in a binary solution reduces
to the Debye-Htickel term in Eqns. ( 16) and ( 17) for I? and A. Solubility Products of Anhydrite and Gypsum
in the corresponding expressions for v. The standard molal at High Pressure
volume and compressibility of CaSO, can be calculated by
additivity: The variation of the solubility product of a mineral with
pressure can be calculated by integrating Eqns. (40) and (4 1):

and
8 ln f&S T, P)
ap
=_AV%T, RT
0
(40)
7

K$&& = G&l, + G&so1 - 2G&CI (39)


and
from the expressions of the standard molal volumes and
compressibilities of NaCl, Naz S04, and CaC12 as a function
of temperature given in the previous sections. = -ti(T, P), (41)

X. SOLUBILITY OF CALCIUM SULFATES IN where A VP and AK: stand for the standard volume and com-
WATER AND IN N&l SOLUTIONS AT pressibility of the reaction of dissolution of the considered
HIGH PRESSURES TO 200°C solid. For anhydrite, one has
One of the most important advantages of Pitzer’s ion in- A!‘:( T, P) = a&++( T, P)
teraction model lies in the fact that it provides an accurate
means for predicting the thermodynamic properties of com- + vPoim(T, P) - V!&(T, P) (42)
plex solutions once those of the simple systems contributing
to the mixture are known. This is of great relevance for geo- with a corresponding expression for AK?. SKINNER( 1963 )
chemical modelling of aqueous systems where one has to indicates that the variation of the molar volumes of sulfates
deal with large ranges ofconcentration and composition, and in the range 20-1OO’C is 0.1-0.3%, and of 0.2-1.0% be-
a large variety of species. Before the model reaches the stage tween 20 and 200°C. From the data of BIRCH ( 1963), one
of practical calculations, a large number of parameters have can calculate that the molar volumes of sulfates decrease by
to be determined from experimental data, as done here for about 0.2% for a pressure increase of 1000 bars. It is then
volumetric properties. A common feature ofthe main papers totally justifiable to consider the molar volumes of minerals
devoted to the parameterisation of Pitzer’s model for geo- independent of T and P, and to take them as equal to their
chemical purposes (see WEARE, 1987, for a review) is that values at 25’C and 1 bar, which are 45.95 cm3/mol for an-
the authors are generally careful in validating the predictions hydrite and 74.94 cm3/mol for gypsum (ROBIE et al., 1979).
of the model against measured properties for complex sys- The integration of Eqns. (40) and (4 1) leads to the classical
tems. For example, in MONNIN( 1989), the calculated partial expression below:
molal volumes of salts in seawater have been compared to
Ap:t$; PO) (p _ po)
the measured values at 25°C. For this temperature, the in- Ln Ksp( T, P) = Ln K,( T, PO) -
clusion of mixing terms (the 0’ and $” parameters) has not
been necessary. A recent study ( CONNAUGHTONet al., 1989) + G!(T, PO)
for the system Na-Mg-Cl-S04-Hz0 up to 95°C has shown 2RT (P- POF. (43)
that these parameters increase with temperature, but roughly
stay within the limit (k5.0 10-6) previously determined and Ln Ksp( T, PO) is given by MOLLER( 1988). In their study of
discussed ( MONNIN, 1989). So the same conclusion regarding the solubility of calcite, fluorite, and celestite at high pressure,
the mixing parameters pertains to higher temperature: they MACDONALDand NORTH ( 1974) found that In Kspis a linear
can be safely ignored in regard to the uncertainties of the function of pressure up to 1000 bars. MILLERO ( 1983) in-
other parameters of the model. dicated that the slopes of the straight lines obtained by MAC-
In his thorough study, MILLERO ( 1979) noted that the DONALD and NORTH ( 1974) compare well with those cal-
only direct measurements of pressure effects on activity coef- culated from volumetric data only if the compressibility of
ficients of electrolytes in mixed solutions as seawater come the reaction of dissolution is taken into account. MILLERO
from measurements involving ionic equilibria such as the ( 1983) also showed that the values of the equilibrium constant
solubility of minerals. Such data may serve as tests to validate for an ionic equilibrium can be overestimated by about 40%
the present model. Now that we dispose of calculated or es- at 1000 bars if A@ is neglected.
Effect of pressure on activity coefficients 3217

a D-
1 -L
‘bar
FIG. 14. Calculated (solid lines) anhydrite solubility in pure water versus pressure at temperatures up to 2o0°C.
The filled circles represent data of DICKSON et al. ( 1963) and BLOUNTand DICKSON ( 1973).

B. Solubility of Anhydrite in Pure Water binary solutions used in the parameterization of the model
and in NaCl Solutions are limited to lower T-Pconditions, and that a certain number
Anydrite solubility in pure water is plotted versus pressure of approximations and correlations have been used, leading
in Fig. 14. One can see the good agreement between the pre- to uncertainties at these high temperatures and pressures.
dicted values and the experimental data of DICKSON et al. The curves for 200°C represent an extrapolation outside the
(1963) and of BLOUNTand DICKSON(1973). T-P-m domain of parameterization of the model. The ac-
Fig. 15 displays the experimental data for anhydrite sol- curacy with which the present model calculates anhydrite
ubility at 1 bar and 1OO’C as a function of NaCl molality. solubility up to 200°C and 1 kbar is nevertheless typically
One can see the large (up to 35%) discrepancies between the within 10%.
data sources. Values calculated by Moller’s model lie about It is possible to refine the model by using the solubility
average for all of these data and above those of BLOUNTand data to adjust some parameters. This is not warranted because
DICKSON ( 1969). Consistent with this, high pressure solu- ofthe reliability of the data, as displayed by the scatter between
bilities calculated at 500 and 1000 bars (Fig. 15 ) also plot various authors shown in Fig. 15 for 100°C. It also appears
above the data of Blount and Dickson by about the same that accurate and extensive measurements of densities and
amount as for 1 bar. compressibilities of Na2S04 above lOO”C, like those now
The same plot for 150°C (Fig. 16) shows that the agreement existing for CaClz , are certain to improve the present model
is very good for 500 and 1000 bars. In this figure, the two more than any other measurements.
dashed curves are calculated by ignoring the compressibility
contribution to the variation of the activity coefficient with C. Variation of the Activity of Water with Pressure;
pressure. One can see that this contribution introduces only Solubility of Gypsum in Pure Water
a small difference in the calculated solubilities at 500 bars,
but quite a large one at 1000 bars. Figure 17, which is for The variation of the activity of water with pressure is given
2OO“C, shows that the calculated anhydrite solubilities are by
overestimated. At 500 bars, the discrepancy is, at most, 15%,
but it is very large for 1000 bars, increasing with the NaCl = v,(T,P,m)- l%(T,P)
. (44)
concentration. It should be remembered that the data for RT
3278 C. Monnin

The partial molal volume of water can be obtained by dif- partial molal volume of water from the density of the solution
ferentiating the expression of the total volume of the solution and from the partial molal volumes of the solutes, which can
(MONNIN, 1989) with respect to the number of moles of be calculated as previously indicated:
water. It can also be obtained by noting that V, the total
volume of the solution. is equal to the sum of the partial
1000 + Z: mjMi
molal volumes of the components multiplied by their number
pw= -MW i+v#
. (47)
of moles: 1000 P

1000 -
v= CniVt=- M VW+ C PZiVi (45)
I w i+w The calculations indicate that at 25”C, for a 6 M NaCl so-
lution, the partial molal volume of water is equal to 17.8 1
with IV, = 18.016 g/mol. The relation between the total cm 3/mol and that its variation with the concentration of the
volume of the solution and its density p is solution is very small. The quantity ?, - v$ is very small.
Hence, the variation of the activity of water with pressure
1000 + C miMi can be neglected.
ifw Figure 18 shows the very good agreement between calcu-
I/= (46)
P lated and measured solubilities of gypsum in pure water up
to 1000 bars between 40 and 80°C. No data for gypsum
in which A&is the molar mass of the ith solute. By combining ~lubility in NaCl solutions as a function of pressure have
Eqns. (45) and (46), one obtains, in a simple manner, the been found.

I I I I I I i
mco5(
0.0

ao:

a0

FIG. 15. Comparison of calculated (solid lines) anhydrite solubilities in NaCl solutions to the measurements of
MARSHALLet al. ( 1964; triangles), BLOCKand WATERS( 1969; squares) and BLOUNTand DICKSON( 1969; filled
circles) at IOOT, and 1, 500, and 1000 bars.
Effect of pressure on activity coefficients 3279

species between 1 and 300 bars has been found to be very


“ccl%
F
I I I I I 1
small. MILLERO(1979,1983) presents an extensive discussion
on the respective contributions of r and I? to the variation
of the activity coefficients with pressure.
Another example of the relevance of pressure corrections
for the activity coefficients of aqueous solutes can be found
in the work of LANGMUIR and MELCHIOR ( 1984). In their
study of the geochemistry of alkaline earth sulfates in the
natural brines of the Palo Duro (Texas) basin, these authors
built a high temperature-high pressure model for the solubility
of these minerals which ignores the variation of activity coef-
ficients with pressure. The sole pressure effect included in
their model is the variation of the ~lubility products of the
minerals. They have calculated the solubility products of these
sulfates at high pressure by two methods. In the Gist, they
directly obtained the KSpfrom the solubility in pure water
measured by BLOUNT and DICKSON ( 1973) and from the
activity coefficients of the aqueous solutes at 1 bar. For an-
hydrite, this leads to

K,(anh) = mcs(T, PI+ ms~,(T, P)*&so,(T, PO). (48)


The second method is based on Eqn. (43) using the volume
and compressibility data of MILLERO (1982).LANGMUIR

--F-d-- m

FIG. 16. Comparison of calculated anhydrite sohtbilities in NaCl “co&


solutions at 150°C and various pressures to the data of BLOW and
DKKSON ( 1969; pressure in bars). The dashed curve is calculated
with the compressibility contribution to the variation of the activity
coefficient of CaSO, equal to zero. The plain curve includes this
contribution.
0.02

XI. DISCUSSION: TEMPERATURE AND PRESSURE


CORRECTIONS FOR THE ACI’MTY COEFFICIENTS OF
AQUEOUS SOLUTES To 200°C AND 1 ICRAR

WEARE (1987)reports that the level of accuracy required


from a model to be able to distinguish between minerals at
equilibrium with a given natural water and those over or
undersaturated, is about 0.02 pK units. In other words, such
a model must predict solubilities within 5% of the measured
values. In the previous sections of this paper, the accuracy
of the solubility predictions of the present model has been 0.01
estimated to be about 10% up to 1000 bars. It is interesting
to quantify the variations of the activity coefficients of the
aqueous solutes with respect to temperature and pressure in
order to test their relevance in regard to this accuracy. As an
illustrative example, we can choose the case of a fiction
solution containing 3 moles of NaCl and 0.01 moles of
CSOd, and calculate the change in the activity coefficients
of the solute from surface conditions (25”C, 1 bar) to those
that one may encounter in an oil field ( 14O”C, 300 bars).
The results are reported in Table 3. One can see that the
activity coefficients decrease with temperature and that this cl
decrease is in part compensated by an increase in pressure. mNUCI
The variation of y with P is between 4 and 14% depending
RG. 17. Comparison of calculated anhydrite solubilities (solid lines)
on the salt. It increases with the charges of the ionic com- in NaCl solutions at 200°C and at high pressures to the data of
ponents of the electrolyte. The change in the distribution of BLOUNTand DICKSON( 1969).
3280 C. Monnin

FIG. 18. Comparison of model prediction of gvpsum solubihty in pure water at high pressure between 40 and 80°C
to the data of BLOUNTand DICKSON( 1973).

and MELCHIOR( 1984) note a difference of 0.02-0.03 between solutes depend more on the type of solute than on the specific
the values of log Ksp for gypsum and anhydrite obtained by solute itself. The system Na-Ca-Cl-S04-Hz0 contains salts
the two different methods, while the agreement is good for representing the major classes of electrolytes: l-l (NaCl),
barite and celestite. For a solution at equilibrium with an- I-2 ( Na2 SO.,), 2- 1 (CaCl* ) , and 2-2 ( CaSO, ) . Therefore, the
volumetric properties of these solutes as reported here may
hydrite at 60% and 500 bars, the model leads to log z
be used to approximate those for other salts of the same type
N 0.014, in which y is the activity coefficient of CaSO, at up to 200°C.
500 bars and ye is that for 1 bar. This quantity induces a
variation in log iu, obtained by Eqn. (48) of 0.028, which ~c~~~~~e~g~e~~~-Thisis CN~/INSU/DBT contribution no. 188.
Financial support was provided by PIRSEM (ARC “GCothermieen
amounts to the systematic difference observed by ~ngmuir
terrain ~imen~i#“) and INSU (th&me “Fluides, min&aux et
and Melchior between the two values of the anhydrite sol- cinbique” of the program “untrue et Bilan de la Terre”). I am
ubility product. Barite and celestite solubilities in pure water very grateful to Jaimey Hovey for his close readingof the manuscript.
are much lower than those of gypsum or anhydrite. For very Thanks are also due to Roland Hellmann for his help with the English
low concentrations in pure water, the excess partial molal language.
volume P - P” is very small. The effect of pressure on the Editorial handling: F. J. Miller0
activity coefficient of BaS04 ( aq ) or SrS04 ( aq ) in pure water
at equilibrium with barite or celestite is then likely negligible. REFERENCES
In an earlier paper ( MONNIN, 1989 ) , it was noted that the
AKSMANOVICM. and KREYJ. ( 1970) Dichte w@?rigerLdsungen
pressure corrections on the activity coefficients of aqueous
von Magnesiumchlorid und Calciumchlorid. Chem. Eng. Tech.
42(24), 1568-1570.
ANANTHASWAMYJ. and ATKINSON G. ( 1984) Thermodynamics of
Table 3 - Salute activity coefficients for a concentrated electrolyte mixtures. 4. ~~-~bye-H~ckel limiting
solution containing 3 moles of NaCl and slopes for water from 0 to 100” and from 1atm to 1 kbar. J. Chem.
Eng. Data 29, 81-87.
0.01 moles of C&O, per kg of water.
BEYERR. P. and STAPLES B. R. ( 1986) Pitzer-Debye-Hiickellimiting
slopes for water from 0 to 350” and from saturation to 1 kbar. J.
Sot. Chem. H(9), 749-764.
BIRCHF. ( 1963) Compressibility: elastic constants. In Handbook oj
~~ Physicai Constants(ed. S. P. CLARK);Memoir 97, pp. 97-114.
The Geological Society of America.
BLOCKJ. and WATERS0. B., JR. (1968) The CaSO.,-Na,SO,-NaCl-
Hz0 system at 25 to 100”. J. Chem. Eng. Data 13, 336-344.
BLOUNT C. W. and DICKSON F. W. ( 1969) The solubility ofanhydrite
in NaCl-Hz0 from 100 to 450” and 1 to 1000 bars. Geochim.
(a) mean electrolyte Ca::..+ SO::f,.m.
Cosmochim. Acta 33.227-245.
Effect of pressure on activity coefficients 3281

BLOUNT C. W. and DICKSONF. W. ( 1973) Gypsum-anhydrite equi- International Critical Tables ( 1928) McGraw Hill.
libria in systems CaS04-HZ0 and CaS04-NaCl-HrO. Amer. Min- KELL G. S. ( 1975) Density, thermal expansivity and compressibility
erals 58, 323-33 1. of liquid water from 0 to 150”: correlations and tables for atmo-
BOGATYKHS. A. and EVNOVICHI. D. ( 1972) Density of aqueous spheric pressure and saturation reviewed and expressed on 1968
solutions of lithium chloride, lithium bromide and calcium chlo- temperature scale. J. Chem. Eng. Data 20(l), 97-105.
ride. Russian J. Appt. Chem. 38,932-933. KHAIBULLINI. K. and NOV~KOVB. E. ( 1973) Thermodynamic study
BRADLEYD. and PITZERK. S. ( 1979) Thermodynamics of electro- of aqueous and vapor solutions of sodium sulfate at high temper-
lytes. 12. Dielectric properties of water and Debye-Hiickel param- atures. Teplojiz. Vysokihk Tempera&r 11,320-327 (in Russian).
eters to 350°C and 1 kbar. J. Phys. Chem. 83, 1599-1603. KOROSIA. and FABUSSB. M. ( 1968) Viscosities of binary aqueous
CHEN C. T., EMMETR. T., and MILLEROF. J. ( 1977) The apparent solutions of NaCI, KCI, Na2S04 and MgS04 at concentrations and
molal volumes of aqueous solutions of NaCl, KCI, Na2S04 and temperatures of interest in desalination processes. J. Chem. Eng.
MgSO., from 0 to 1000 bars at 0, 25 and 50°C. J. Chem. Eng. Data 13, 548-552.
Data 22(2), 201-207. KUMARA. ( 1986 ) Densities and apparent molal volumes of aqueous
CHEN C. T., CHEN J. H., and MILLEROF. J. (1980) Densities of concentrated calcium chloride solutions from 50 to 200” at 20.27
NaCl, MgClr, Nar SO, and MgS04 aqueous solutions at 1 atm bars. J. Sol. Chem. E(5), 409-412.
from 0 to 50°C and from 0.00 1to I .5 m. J. Chem. Eng. Data 25, KUMAR A. and ATKINSONG. ( 1983) Thermodynamics of concen-
307-3 10. trated electrolyte mixtures. 3. Apparent molal volumes compres-
CONNAUGHTONL. M., HERSHEYJ. P., and MILLEROF. J. (1986) sibilities and expansivities of NaCl-CaC& mixtures from 5 to 35°C.
PI’T properties of concentrated aqueous electrolytes. V. Densities J. Phys. Chem. 87,5504-5507.
and apparent molal volumes of the four major sea salts from dilute KUMAR A., ATKINSONG., and HOWELLR. D. ( 1982) Thermody-
solutions to saturation and from 0 to 100”. J. Sol. Chem. 15( 12), namics of concentrated electrolyte mixtures. II. Densities and
989-1002. compressibilities of aqueous NaCl-CaClr at 25’C. J. Sol. Chem.
CONNAUGHTONL. M., MILLEROF. J., and PITZER K. S. ( 1989) 11( 12), 857-870.
Volume changes for mixing the major sea salts: equations valid to LANGMUIRD. and MELCHIORD. ( 1984) The geochemistry of Ca,
ionic strength 3.0 and temperature 95°C. J. Sol. Chem. 18( 1l), Sr, Ba and Ra sulfates in some deep brines from the Palo Duro
1007-1018. Basin, Texas. Geochim. Cosmochim. Acta 49,2423-2432.
DEL GROSSOV. A. and MADER C. W. ( 1972) Speed of sound in Lo SURDOA., AUOLA E. M., and MILLEROF. J. (1982) The PVT
pure water. J. Acoust. Sot. Amer. 52( 5), 1442-1446. properties of concentrated aqueous electrolytes. I. Densities and
DICKSONF. W., BLOUNTC. W., and TUNELL G. ( 1963) Use of apparent molal volumes of NaCI, Na2S04, MgCl, and MgSQ
hydrothermal solution equipment to determine the solubihty of solutions from 0.1 mol/kg to saturation and from 273.15 to 323. I5
anhydrite in water from 100 to 275°C and from 1 bar to 1000 K. J. Chem. Therm. 14,649-662.
bars pressure. Amer. J. Sci. 261,61-78. MACDONALDR. W. and NORTHN. A. ( 1974) The effect of pressure
ELLIS A. J. ( 1967) Partial molal volumes of MgClr, CaC12, SrClr on the solubility of CaCOs, CaFr and SrSO, in water. Canadian
and BaQ in aqueous solutions to 200”. Chem. Sot. (London) J. J. Chem. 52,3181-3186.
1967A, 660-664. MARSHALLW. L., SLUSHERR., and JONESE. V. ( 1964) Solubility
ELLISA. J. ( 1968) Partial molal volumes in high temperature water. and thermodynamic relationships for CaS04 in NaCl-HZ0 from
III. Halide and oxyanion salts. Chem. Sot. (London) J. 1%8A, 40 to 200”, 0 to 4 molal NaCI. J. Chem. Eng. Data 9, 187-191.
1138-1143. MIKHAILOVI. G., SAVINAL. I., and FEOFANOVG. N. ( 1957) The
FABUSSB. M., KOROSIA., and HUQ S. ( 1966) Densities of binary ultrasonic velocity and compressibility of strong electrolyte solu-
and ternary aqueous solutions of NaCl, Na2S04 and MgSO.,, of tions. Leningrad Univ. Vestkik. Ser. Fiz. Chim. 195714). . ,, 25-42
seawater and seawater concentrates. J. Chem. Eng. Data 11,324- (in Russian).
331. MILLEROF. J. ( 1979) Effects of temperature and pressure on activity
FISHER F. H. and Fox A. P. ( 1979) Divalent sulfate ion pairs in coefficients. In Activity Coefficients in Electrolyte Solutions (ed.
aqueous solutions at pressures to 2000 atm. J. Sol. Chem. 8(4), R. M. P~TCKOWICZ), Vol. II, pp. 63-152. CRC Press.
309-328. MILLEROF. J. ( 1982) The effect of pressure on the solubility of
GARVIND., PARKERV. B., and WHITEH. J. ( 1987) CODATA ther- minerals in water and seawater. Geochim. Cosmochim. Acta 46,
modynamic tables. Selection for some compounds of calcium and 1l-22.
related mixtures: a prototype set of tables. CODATA Series on MILLEROF. J. ( 1983) Influence of pressure on chemical processes
Thermodynamic Properties. Hemisphere Publ. in the sea. In Chemical Oceanography (eds. J. P. RILEY and R.
GATESJ. A. and WOD R. H. ( 1985) Densities of aqueous solutions CHESTER),pp. 2-88. Academic Press.
of NaCI, KCI, NaBr, LiCl and CaClr from 0.05 to 5.0 mol/kg and MILLEROF. J. and KNOX J. H. ( 1973) Apparent molal volumes of
0.1013 to 40 MPa at 298.15 K. J. Chem. Eng. Data 30( I), 44- aqueous NaF, Na2S04, KCI, KZS04, MgCI, and MgSO., solutions
49. at 0 and 50°C. J. Chem. Eng. Data 18(4), 407-41 I.
GATESJ. A. and WOOD R. H. ( 1989) Density and apparent molar MILLEROF. J., WARDG. D., and CHETIRIUNP. V. ( 1977a) Relative
volume of aqueous CaC& at 323-600 K. J. Chem. Eng. Data 34, sound velocities of sea salts at 25°C. J. Acoust. Sot. Amer. 61(6),
53-56. 1492-1498.
GREENBERGJ. P. and MOLLERN. ( 1989) The prediction of mineral MILLEROF. J., GOMBARF., and OSTERJ. ( 1977b) The partial molal
solubilities in natural waters: a chemical equilibrium model for volume and compressibility change for the formation of the calcium
the Na-K-Ca-Cl-S04-Hz0 sytem to high concentration form 0 to sulfate ion pair at 25°C. J. Sol. Chem. 6(4), 269-280.
250°C. Geochim. Cosmochim. Acta 53,2503-25 18. MILLEROF. J., VINOKUROVAF., FERNANDEZM., and HERSHEY
HAAR L., GALLAGHERJ. S., and KELL G. S. ( 1984) NBS/NRC J. P. ( 1987) PVT properties of concentrated electrolytes. VI. The
steam tables: thermodynamic and transport properties and com- speed of sound and apparent molal compressibilities of NaCl,
puter programs for vapor and liquid states of water in SI units. Na, SO,, MgClr and MgSO., solutions from 0 to 100°C. J. Sol.
Hemisphere Publ. Chem. 16(4), 269-284.
HARVIEC. E., MOLLERN., and WEAREJ. H. ( 1984) The prediction MOLLERN. ( 1988) The prediction of mineral solubilities in natural
of mineral solubilities in natural waters: the Na-K-Ca-Mg-H-Cl- waters: a chemical equilibrium model for the Na-Ca-Cl-SO.,-HZ0
S04-OH-HCOs-C03-CO*-Hz0 system to high ionic strengths at system to high temperature and concentration. Geochim. Cos-
25°C. Geochim. Cosmochim. Acta 48,723-75 1. mochim. Acta 52,821-837.
HELIXSONH. C. and KIRKHAMD. H. ( 1976) Theoretical prediction MONNINC. ( 1987) Densities and apparent molal volumes of aqueous
of the thermodynamic properties of aqueous electrolytes at high CaCIZ and MgC&. J. Sol. Chem. 16( 12), 1035-1048.
pressures and temperatures. III. Equation of state for aqueous spe- MONNIN C. ( 1989) An ion interaction model for the volumetric
ties at infinite dilution. Amer. J. Sci. 276,97-240. properties of natural waters: density of the solution and partial
3282 C. Monk

molal volumes of electrolytes to high concentration at 25°C. Geo-


chim. Cosmochim. Acta 53, 1 l77- 1188. l&
MONNINC. and GALLINIERC. ( 1988) The solubility of celestite and
barite in electrolyte solutions and natural waters at 25°C: a ther-
modynamic study. Chem. Geol. 71,283-296.
PABALAN R. T. and PITZER K. S. ( 1987) Apparent molar heat ca-
pacity and other thermodynamic properties of aqueous KC1 so-
lutions to b&h temperatures and pressures. J. Chem. Eng. Data
33,354-362.
PEPINOV R. I., LOBKHOVAN. V., and Z~KHRABBEKOVA
G. ( 1985)
Plotnost votznich rastvorov sulfata natriia v chirokom intervalle
parametrov sostoiania. TeploJiz. Vys. Temp. 23(2), 399-402 (in
Russian).
PERMAN E. P. and URRY W. D. (1929) The compressibility of
aqueous solutions. Royal Sot. (London) Proc. Al%, 44-78.
PITZERK. S. ( 1987) Thermodynamic model for aqueous solutions
of liquid-like density. In Thermodynamic Modeliing of Geological
Materials: Minerals, Fluids and Metals (eds. I. S. E. CARMICHAEL
and H. P. EUGSTER); Reviews in Mineralogy 17, pp. 143-176.
Mineralogical Society of America.
ROBIE R. A., HEMINGWAYB. S., and FISHERJ. R. ( 1979) Ther-
modynamic properties of minerals and related substances at 98.15
K and 1 bar ( 10s Pa) pressure and at higher temperatures. Geol.
Surv. Bull. 1452.
ROGERSP. S. Z. and PITZERK. S. ( 1982) Volumetric properties of -1
sodium chloride solutions. J. Phys. Chem. Ref: Data 11( I), 15-
81. I I I
2
J-m
SKINNER B. J. ( 1963) Thermal expansion. In Handbook of Physical 1
Constants (ed. S. P. CLARK); Memoir 97, pp. 15-96. The Geo-
logical Society of America. FIG. A 1.Variation of the apparent molal isothermal (circles) and
TANGERJ. C., IV and HELGE~ONH. C. ( 1988) Calculation of the adiabiatic (triangles) compressibilities (in lo4 cm3/mol/bar) of
thermodynamic and transport properties of aquoeus species at high aqueous solutions of calcium chloride versus the square root of the
pressures and temperatures: revised equation of state for the stan- molality (filled signs: MILLEROet al., 1977a; open signs: KUMARet
dard partial molal properties of ions and electrolytes. Amer. J. Sci. al., 1982).
288, 19-98.
UEMATXI M. and FRANCKE. U. ( 1980) Static dielectric constant
of water and steam. J. Phys. Chem. Ref Data 9, 1291-1330.
WEAREJ. H. (1987) Models of mineral solubility in concentrated for CaCl* (aq) have recently been critically evaluated by GARVINet
brines with application to bled observations. In Thermodynamic al. ( 1987). C, values can be easily calculated from the coefficients
Modeliing of Geological Materials: Minerals, Fluids and Melts of Pitxer’s equations given by these authors.
(eds. I. S. E. CARMICHAEL and H. P. EUGSTER);Reviews in Min- On Fig. A 1, the apparent adiabatic and isothermal compressibilities
eralogy 17, pp. 143-176. Mineralogical Society of America. are plotted versus the square root of the molality. The apparent adi-
abatic compressibilities have been calculated from the density and
sound speed data of Mn.LzRo et al. ( 1977a) and KUMARet al. ( 1982).
APPENDIX The difference between &, and @,rsis not large, as already indicated
by MILLEROet al. ( 1987) for sodium and magnesium chlorides and
Conversion of the Adiabatic Compressibility of Calcium sulfates. A fit of these data to Eqn. ( 10) leads to the following values
Chloride to its Isothermal Counterpart of the parameters:
The adiabatic (or isentropic) compressibility KS is related to the
isothermal one through

an expression in which p is the density of the system, C, its heat


capacity, and (I its coefficient of thermal expansion defined by The value of a, determined here compares well with the one
already determined by MILLEROet al. (1977a) (-94.66 10e4) by a
I av
a=-

v (1
-

ar p’
(A-2)
different method. KUMAR et al. (1982) included the parameter
BGt in the analysis of their data. It was not required here and leads
to overfitting the data, which drastically atfects the value of the stan-
a can be obtained by a simple differentiation with respect to tem- dard compressibility.
perature, of the expressions for the specific volume of the solution, Finally, the use of interaction parameters determined from sound
along with those for the standard molal volume of CaCl,(aq) and speeds does not introduce any significant errOr in the calculation of
for the various interaction parameters. Heat capacity measurements isothermal compressibilities.

You might also like