Chemical Engrrweriny Science, Vol. 48. No. 18. pp. 3 149-3158, 1993. ooo-25@9/93 woo + 0.

00
Printed in Great Britain Q 1993 Pergamo” Press IAd

CORRELATION AND PREDICTION OF MINERAL

SOLUBILITIES IN THE RECIPROCAL SALT SYSTEM
(Na+, K’)(Cl-, SO:-)-Hz0 AT 0-100°C

HENRIK NICOLAISEN, PETER RASMUSSEN’ and JENS M. SORENSEN

lnstitut for Kemiteknik, Technical University of Denmark, DK-2800 Lyngby, Denmark

(First received 23 November 1992; accepted in revised form 22 March 1993)

Abstract-A simplified version of the extended UNIQIJACPDebyePHiickel model is used for correlation
and prediction of mineral solubilities in the reciprocal salt system (Na+, K+)(CI-, SO:-)-Hz0 at
temperatures from 0 to 100°C. In this work the model is modified in order to contain only binary
parameters with no concentration dependence. The UNIQUACPDebyeeHiickel SE-model is used for
calculation of both ionic- and water-activities in aqueous solutions containing two or more of the four ions
Na*, K*, Cl- and SO:- at concentrations from infinite dilution to saturation and temperatures from 0 to
100°C. In addition to the activity coefficient model, an expression for calculation of solubility products for
various salts in the reciprocal salt system is presented. The results show that the extended UNIQUAC
model is able to give a good representation of ternary solubility diagrams and a fair representation of
vapor-liquid data, in the whole concentration range from infinite dilution to saturation, using the same set
of interaction parameters. Furthermore, it is shown that the model is able to predict solid--liquid equilibria
in the quaternary system (Nat, K+)(CI-, SO: )-Hz0 over a broad temperature range.

INTRODUCTlON treatment plants. The important industrial process

Calculations of phase equilibria for solutions contain- fractional crystallization may be designed by graphi-
ing electrolytes are of great import&ce for research cal methods in phase diagrams obtained from experi-
and design of equipment in the chemical industry. An mental solubility data, as described by Fitch (1970).
important subject in this area is correlation and pre- The main advantage of using a model instead of
diction of salt precipitation from multiion aqueous graphical methods is that it is possible to optimize the
solutions. Salt precipitation (SLE) calculations are design in a much more efficient way by means of
important for design and operation of industrial cry- temperature and concentration operating conditions.
stallizers. They are also important for processes where The disadvantage of using a model for such purposes
precipitation is to be avoided. Furthermore, SLE cal- is that model uncertainties could give rise to er-
culations for aqueous electrolyte solutions are applied roneous process design.
in sciences such as geochemistry and metallurgy. Be- In this work we have included the influence of the
cause of its great importance, many authors have temperature on the precipitation of various solids
treated the subject of correlation and prediction of from aqueous electrolyte solutions. The model used
mineral solubilities in aqueous electrolyte solutions. for this purpose is a simplified version of the extended
Examples of previous works are Wood (1975), which UNIQUAC model of Sander (Sander, 1984; Sander et
used a modified Harned’s rule (Harned and Owen, al., 1986). As shown by Sander (1984), this model is
1958) for calculation of solid-liquid equilibria in able to give good predictions of the influence of the
the systems NaCl- KClGMgCl,-CaCI,-H,O and temperature on both VLE and SLE for electrolyte
NaCI-MgSOq-HZ0 at 25”C, and the extensive work solutions, with a linear temperature-dependence of
of Harvie and Weare (1980). In this work the Pitzer the parameters.
model (Pitzer, 1973, 1975; Pitzer and Mayorga, 1973,
1974; Pitzer and Kim, 1974) was used for prediction of SOLll%LtQUID EQUILIBRIA IN AQUEOUS ELECTROLYTE
mineral solubilities in the salt system (Na+, K*, SOLUTIONS
Mg 2+, Ca* l) (Cl-, SO:-) at 25°C.
Precipitation of a salt, which consists of v + cations
In spite of the extensive work performed earlier in
M”, v- anions X’ and n crystal water, from
this field, the majority of the existing thermodynamic
a saturated aqueous electrolyte solution can be rep-
models are unable to give a good description of the resented by the following equilibrium reaction:
temperature dependence on phase equilibria for elec-
trolyte solutions, as shown by Zemaitis et al. (1986). v+ M’+ (aq) + v-X”- (aq) + nH20 &
A predictive result indicates that the model may be
M,, Xv_. nH,O (4. (1)
used for design of industrial processes involving elec-
trolytes such as fractional crystallization, evaporation The criterion for solid-liquid equilibrium is that the
paths in natural brines and design of wastewater- chemical potential of the salt is equal to the sum of the
chemical potentials of the constituents of the salt:

‘Author to whom correspondence should be addressed.

3149
 3t50 HENRIK NK-OIAISFN tv al.

The chemical potential of water can be defined as by is given by

fiw = p: + RTln a, = & + RTln (y,+.xw) (3) ahp

where a, is the water activity,
water activity coefficient,
yw is the symmetric
where y,,, + 1 as x, + 1 and If the standard
C-1
heat capacity
dT #,,“;

c: is assumed
= CT.
inde-
(10)

& is the standard chemical potential for water. For pendent of the temperature, and equal to the value at
the ions, we adopt the unsymmetric convention, 25°C. eq. (9) can be integrated to give
where y;f;, --* 1 asCmi-+O:

pi = ~7 + RTln(yi~,mi) (4)

where yiTm is the unsymmetric molal activity coeffi-

-c~,[ln($)+$- I] (11)
cient for ion i, and ~7 are the standard chemical
potentials for the solutes.
where subscript 0 refers to the values at 298.15 K.
The solid phase consists of the pure salt; therefore,
The hp- and cy-values for the ions and the water are
a Salt = 1 and so /1,.ll = &i,. With these definitions, the
obtained from the NBS tables (Wagman et al., 1968,
equilibrium conditions (2) can be written as

This term is defined as the solubility product K for

the salt

1. (6)
v+pLM” + v-&z + n/L; - L&+x ” . “11ZO
K M,+x,_.~H~o = exp
RT

Therefore, the solubility problem is reduced to solu- 1969,1971, 1981; Parker et al., 1971; Schumm, 1973),
tion of the following equation while the values for the salts are found by fitting to
(Y$;“‘.m rnM~+)‘+(y~~ ,mmx~~)Y- a; experimental solubility data.

--K M,* x._ , nH20= 0. (7) EXTENDED UNIQUAC

Equation (7) is solved for the salt molality m (mol The extended UNIQUAC equation by Sander
salt/kg H,O), in order to obtain the solubility of the (Sander, 1984; Sander et al., 1986) is a local composi-
salt M,, X,_ , nHZO. tion mode1 based on the original UNIQUAC equa-
tion (Abrams and Prausnitz, 1975; Maurer and
mM=.
m=_=!!E Prausnitz, 1978) combined with a Debye-Hiickel
(8)
v+ “_ term. The extended model was originally developed
The standard chemical potentials ~7 and py for the for calculation of salt effects on vapor-liquid equilib-
ions and the water at 25°C may be found in data ria in mixed solvent systems.
compilations such as the thermodynamic tables from The extended UNIQUAC equation consists of
the National Bureau of Standards (Wagman et al., three contributions to the total excess Gibbs energy:
1968, 1969, 1971, 1981; Parker et al., 1971; Schumm a Debye-Hiickel term, a combinatorial term and a re-
1973). Values of the standard chemical potentials of sidual term. The unsymmetric molar excess Gibbs
simple salts may also be found in such data compila- function on mole fraction basis can, therefore, be
tions, while py-values for more complex salts (double written as
salts and salt hydrates) are normally not available in gE*/RT= gi&/RT+ g:*lRT+ g,E”/RT. (12)
such tables.
In this work the standard chemical potentials for The Debye-Hiickel term gives the contribution to
the ions and for water at 25°C are obtained from the the long-range electrostatic forces in the solution to
NBS tables, while the py-values for the solid phases the total excess Gibbs energy.
are fitted to experimental solubility data. Sander (1984) uses the following expression, which
At temperatures different from 25°C we obtain the originally was suggested by Fowler and Guggenheim
py-values using the Gibbs-Helmholtz equation (1949) for the electrostatic contribution to the excess
(Atkins, 1986) Gibbs energy:

&LY/ T h? y,E&,IRT =

1
(9)
( i?T > P.-j 2, TZ.
_ x,Mw4$ In(1 +ha)--ba+$ (13)
The temperature dependence of the standard enthalpy L
 Correlation and prediction of mineral solubilities 3151

where x, and M, are, respectively, the mole fraction In the original version of the extended UNIQUAC
and the molar mass ( = 0.0 18015 kg/mol) of water and model (Sander, 1984), the interaction parameters be-
h = l.50(kg/mol)i’2. A is the Debye-Htickel con- tween ions and water were assumed both temper-
stant, which is given by ature- and concentration-dependent, as suggested by
d”2 Skjold-Jerrgensen (1980). The concentration depend-
ence of the parameters requires an additional set of
A = c @7-)3/Z
interaction parameters. Therefore, the concentration
c = 1.3287x 10m5 K3/2 m3/2mol-1’2 (14) dependence of the parameters are neglected in this
work, in order to reduce the total number of para-
where d and D are, respectively, the density (kg/m3) meters in the model:
and the dielectric constant of water. When the phys-
ical properties of water are introduced in eq. (14), we uiw = up, + &, (T - 300). (22)
obtain the following expression for the Debye-
The water and ion activity coefficients are obtained by
Hiickel constant as a function of the temperature T: partial molar differentiation of the g”-expressions:

A = 1.131 + 1.335x 10-3x(T- 273.15)

+ 1.164x 1O-5 x(T- 273.15)‘.

lnyt = [a(nga~~T’]p,T,n,+,.
(23)
(15)
Using eq. (23) we obtain the following contributions
I is the ionic strength (mole/kg H,O):
to the symmetric activity coefficient for the water and
the unsymmetric activity coefficients for the ions.
I = k? m,zF = -$Z$ x,zf/x,M, (16) The activity coefficient for the water and for the
1 I ions can then be calculated from
where mi is the molality (mole/kg H20) and z; is the lny,=lny,D~“+lny$+lny,R
charge number of ion i. The short-range interactions (24)
are described by the UNIQUAC equation, which lny: = In y:TDmH+ lny:~’ + In y:.“.
consists of two terms.
The unsymmetric activity coefficients r: for the ions
The first term is the combinatorial, or entropic,
are obtained from the symmetric activity coefficients
term. This term takes the deviation from ideality, due
yi using the relationship
to differences in size and shape of the species in the
system into account. The combinatorial term is lny: = Iny, - Iny? (25)
given as
where y,F is the infinite dilution activity coefficient of
ion i in pure water.
s:IRT= ~xkIn(~k/xk)-~C4kxk1n(Ok/8k)
The expressions for the three contributions to the
k k
activity coefficient of the water and of the ions are
(17)
given below.
where z is the coordination number, which arbitrarily
is set equal to 10. Parameters I#J~and @k give the
Debye-Htickel term:
volume and surface area fractions:
Water: In yz-” =
(18)
l+bJri--- 2ln(l+ b*) .
l+bd 1
where rk and qr are volume and surface area para-
meters for the ions and water in the system.
(26)
The second part of the short-range contribution is Ions: _
the residual, or enthalpic, term. This term takes the ]ny:.D-H= -,y,?A-. JI
short-range molecular energetic interactions into l+ha
account:
Combinatorial term:
g:IRT= -~wh~n(~xk!bk~)~ (19)
Water:
The parameter $kl is given by In Y$ = In (&lx,) + 1 - &Jx, - : qw

erl=exp( -y) (20) x Dn (&J&I + 1 - 4.JLl. (27)

Ions:
where ukI and ul, are interaction energy parameters.
The ukI ( = ua) parameters between ionic species in ]n yi* .’ = In (di/Xi) - +i/Xi - In (MA + rilru
the solution are considered temperature dependent:

ukl = u,9 + U:j (T - 300). (21)

-
z
Zj 4i

[
In (4ilei) - Qi/ei - 1 n(z)+#
3152 HENRIK NICOLAISEN er al.

Residual term: perimental SLE-data for aqueous electrolyte solu-

tions at various temperatures. The solubility data in
Water:
the data bank are primarily obtained from the ex-
tensive data compilations by Linke and Seidell (1965)
and Silcock (1979). The experimental equilibrium data
used for the parameter estimation are extracted dir-
ectly from the data bank. The VLE-data used for the
estimation of parameters were obtained from the
VLE-data compilations by Hamer and Wu (1972) and
by Goldberg (1981).
The model parameters and the thermodynamic
functions for the salts were estimated by minimization
of the following objective function
In the original UNIQUAC equation of Abrams
NVLE
and Prausnitz (1975), the volume and surface area r Y>.crp
*

(
Y calf -
F=x w, *
parameters r, and qk were calculated from physical i= I Y+ .erp >
properties. Sander (1984) found that very inaccurate
results were obtained when re- and q,-values cal-
culated from physical properties were used for electro-
lyte solutions. The rlr- and q&-parameters are here where y + is the molal salt activity coefficient and msa’
assumed to be equal and considered as adjustable is the salt molality of the saturated solution. NVLE is
parameters in the extended UNIQUAC equation. the total number of binary (2 ions in water) experi-
The ion and water activity coefficients calculated mental VLE data points for the four salts NaCl, KCl,
from the extended UNIQUAC model (24) are rational Na2S04 and KZSO, used for parameter estimation,
activity coefficients (on mole fraction basis), while the NSLE is the number of ternary solubility data (3 ions
activity coefficients required for solution of the solu- in water). The estimation of parameters for the recip-
bility equation (7) is on a molal basis. rocal system (Na+, K+)(Cl-. SOi-)-Hz0 included
The rational activity coefficients given by the 223 binary VLE and 340 ternary SLE datapoints.
UNIQUAC equation can be converted into molal Weighing factor for data set number i is wi. The
activity coefficients by the following relation weighing factors have been chosen by trial and error
to give a good representation of all the calculated
phase diagrams. The weight factor is necessary to give
a uniform representation of all the ternary salt sys-
where the summation is over all ionic species. tems, because of a large variation in the number of
experimental data points for the different ternary salt
ESTIMATION 0F PARAMETERS systems. Systems where very few experimental equilib-
In order to solve eq. (7) for the solubflity of a given rium data were available were, therefore, given a relat-
salt, one requires data for the interaction parameters ively high weight, while systems containing many ex-
in the activity coefficient model and values for the perimental data were given a low weight. The purpose
standard chemical potential, enthalpy and heat capa- of this work was correlation and prediction of salt
city for the salt. These data were estimated by simul- solubilities in aqueous solutions, the experimental
taneous adjustment to experimental vapor ‘liquid VLE data used for parameter estimation were, there-
equilibrium data and solubility data. For this purpose fore, given a relatively low weight in the objective
a computer based data bank has been developed. The function. It was here found suitable to weight the
data bank contains both experimental VLE- and ex- experimental SLE data with values between 1 and 10.

Table 1. Standard chemical potentials, enthalpies and heat capacities at 25°C for ions
and salts in the reciprocal salt system (Na’, K+)(CI-, SOi-)-Hz0
v V v
cpi
Name Formula (l&Ol) &lOl, (J/mol K)

Water Hz0 - 237.129 - 285.830 15.29

Sodium ion Na+ -261.905 - 240.120 46.40
Potassium ion K+ -2X3.270 - 252.380 21.80
Chloride ion cl- -131.228 - 167.159 - 136.40
Sulfate ion so: - -744.530 - 909.270 - 293.00
Halite NaCl ~384.401 - 412.870 SO.50
Sylvite KCI -409.608 ~ 432.848 51.30
Arcanite K,SO, - 1321.630 - 1427.980 131.46
Mirabilite Na,SO,. IOH, - 3646.730 - 43 14.680 460.20
Thenardite Na,SO, - 1210.650 ~ 1381.020 128.20
Aphthitalite NaK,(SO,), - 2622.660 - 2827.490 - 72.23
 Correlation and prediction of mineral solubilities 3153

The objective function (30) was minimized using 1.20

a modified Marquardt method for non-linear least-
square fitting (Fletcher, 1971). 1.00
The result of the parameter estimation is shown in
four tables. Table 1 contains values for the standard
chemical potentials, enthalpies and heat capacities at 0.80

25°C for ions and salts in the reciprocal salt system.

The thermodynamic values for the salts are estimated y 0.60
l
from experimental solubility data, while the values for
the ions and water were obtained from the NBS tables 0.40
(Wagman et al., 1968, 1969, 1971, 1981; Parker et al.,
1971; Schumm, 1973). In cases where ,up-, hp- or c:~-
0.20
values for a salt were available in the NBS tables, it
was ensured that the deviation between the estimated
values and the tabulated values was less than 10%. 0.00
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7. 10
Table 2 contains the UNIQUAC volume parameters m (moles/kg HaO)
(ri) and surface area parameters (qi). The ri- and qi-
values for the four ions are estimated, while the values Fig. I. Calculated (-) and experimental (A, n , n or 0)
for water are calculated from the physical properties (Hamer and Wu, 1972; Goldberg, 1981) mean molal activity
coefficients for NaCI, KCI, Na,SO, and K,SO, at 25°C.
of water as described elsewhere (Maurer and
Prausnitz, 1978). Tables 3 and 4 contain the two
binary UNIQUAC parameters, which are used for
calculation of the energetic parameter uij [eq. (21)]. activity coefficient for the four salts NaCI, KCI,
Some of the binary parameters are arbitrarily set NaZS04 and K,SO, in water at 25°C was calculated.
equal to either 2500 and 0, in order to reduce the The results of the calculations are shown in the
number of parameters fitted. figures.
Figure 1 shows the result of the vapor-liquid equi-
RESULTS AND DISCUSSION
librium calculation at 25°C. The mean molal salt
activity coefficient is calculated at concentrations
The ternary and quaternary SLE diagrams for the
from infinite dilution to saturation for each of the four
reciprocal salt system (Na+, K+) (Cl -, SOi- )-HZ0
salts NaCl, KCI, Na2S04 and K,SO,. The calculated
have been calculated at selected temperatures between
salt activity coefficients are compared with the corres-
0 and 100°C. In addition to this, the mean molal salt
ponding experimental data of Hamer and Wu (1972)
and Goldberg (1981). As it appears from the figure,
a good agreement between calculated and experi-
Table 2. UNIQUAC volume parameters (ri) and surface
area parameters (y;)
mental values are obtained. For the activity coefficient
of NaCl there is a minor deviation in the concentra-
H,O Na+ K+ Cl_ so:- tion range from 1 mol/kg to about 5 mol/kg, while
a good agreement is obtained in the diluted area and
r, 0.92 5.83 2,95 1.63 0.12
1.40 5.83 0.72
in the near-saturation area. At temperatures different
cli 2.95 1.63
from 25°C we obtain similar results. At 110°C the

Table 3. UNIQUAC interaction parameters u: (K)

Na* KC cl- so:; H,O

Na+ 2500.00
K+ 7747.81 2500.00
Cl- 2067.02 924.12 2500.00
so: - 1403.23 1002.91 2656.14 2500.00
ho 1847.43 62.65 857.41 2237.03 0.00

Table 4. UNIQUAC interaction parameters u:~

Na+ Kf Cl so: - ‘-‘,O

Na’ 0.000
K+ ~ 71.259 0.000
cl- 8.067 4.026 0.000
so:- 2.476 2.366 39.677 OX00
H,O 0.872 0.037 5.469 11.369 0.000

CES 48:18-D
3154 HENRIK NICOLAISEN et al.

result for the NaCI-Hz0 system is excellent, while

NaCl-Na2S04--Hz0 at 0°C
deviations about 10% are seen for the same system at
0°C in the concentration range from 1 mol/kg to 040-
about 5 mol/kg. The observed deviations between cal-
culated and experimental salt activity coefficients are
due to the fact, that the VLE data were given a very
low weight at the estimation of parameters. At calcu-
lations we observed that very good results could be
obtained in the whole concentration range from in-
finite dilution to saturation at all temperatures be-
N&TO,. lOHe
tween 0 and loO”C, if the model parameters were
estimated directly from experimental vapor-liquid
eqmhbrium data. IhC,
It may, therefore, be concluded that the extended r/l
N

UNIQUAC model, with the parameters given in the

Tables 3 and 4, can be used for VLE calculations for
2 0 00
0
the system (Na+, K+)(Cl-, SO:-)-Hz0 at ternper- 0 00 NaCl
2.00 (moles/kg
4.00 HeO)
6.00 s’
atures from 0 to about 100°C with a satisfactory
Fig. 3. Calculated (-) and experimental (A) (Silcock, 1979)
result. salt solubilities in the ternary system NaCI-Na,SO,-H,O at
Ln Fig. 2 the solubility diagram for the ternary salt 0°C.
system NaCl-KCl-HZ0 at temperatures from 0 to
100°C is shown. The calculated solubilities are com- NaCl-Na&O,-H,O at 25°C
pared with experimental data of Linke and Seidell
(1965) and Silcock (1979). The deviations between
2’oo2 l NadO.. lOH,O
calculated and experimental solubilities are within n1.60 -
experimental accuracy, except for the solubility of 0
pure KC1 at 0°C which is underestimated by approx- 2
imately 10% by the model. From the four curves it ho120 -
appears that there is a fine agreement between the <
calculated and experimental invariant points-the z
point where both NaCl and KC1 are precipitating ~o.eo -
simultaneously-for all the investigated temperatures. .E
From Fig. 2 it appears that no salt hydrates or
so.40 -
double salts may be precipitated from an aqueous r/I
solution containing Nat-, K*- and Cl--ions in the N

temperature interval from 0 to 100°C; only the two iz

0.00 ,I,-
simple salts NaCl and KCI will precipitate. The solu- 0.00 2.00 4.00 8.00 D

bility diagrams for the ternary system NaCl- NaCl (moles/kg HsO) ’
Na2S04-H,O, which are shown in Figs 3-6, are Fig. 4. Calculated (-) and experimental (A) (Silcock, 1979)
therefore more interesting from a scientific point of salt solubilities in the ternary system NaCILNa,SO,-H,O at
25°C.
NaCl-KCl-Hz0 NaCl-NazS04--Hz0 at 50°C
800 , I
7.00

E;.s.oo
LJY 4’ooi
5.00
M
Coo
s
~3.00

0.00 I I I
0.00 100 2.00 3.00 4.00 5.00 6.00 7.00 0
NaCl (mol/kg HeO) NaCl (moles/kg HaO) a
Fig. 2. Calculated (-) and experimental (A, n , l or 0) Fig. 5. Calculated (-_) and experimental (A) (Silcock, 1979)
(Sitcock, 1979) salt solubilities in the ternary system salt solubilities in the ternary system NaCI-Na,SO,-H,O at
NaCtKCIPH,O at 0, 25, 50 and 100°C. 50°C.
 Correlation and prediction of mineral solubilities 3155

NaCl-NazS04-Hz0 at 100°C 1 .oo

Na&04-KgS04-Hz0 at 25””L
3.20 T_

‘1 -0.60
0
EC

*o.zo 1

Na.SO,,lOH,O
E
/
b?
2
0 00 000 I I
I
7
0.00 2.00 4 00 6.00 O.OD 0.50 1.00 I.50 2.00 2.50 3 0
NaCl (moles/kg HaO) NaaSO, (moles/kg HaO)

Fig. 6. Calculated (-) and experimental (A) (Silcock, 1979) Fig. 7. Calculated (-) and experimental (A) (Silcock, 1979)
salt solubilities in the ternary system NaCl-Na,SO,-H,O at salt solubilities in the ternary system Na,SO,-K,SOt-Ha0
100°C. at 25°C.

view. In this system possibility for the formation of NazS04-QSO,-Hz0 at 50°C

three different solid phasessodium chloride (halite), 1.204
sodium sulfate (thenardite) and sodium sulfate with 10 KS’.

crystal water (mirabilite))-exists. As it appears, the

formation of either Na2S04 or Na2S04, 10H20 is
dependent upon the temperature. At 0°C only solid
NaCl or solid Na,SO,. lOH,O may be formed. If the
temperature is raised to 17.9”C, the third salt Na2S04
may also precipitate. If the temperature is increased
further, only two different solid phases, NaCl and
Na2S04, may be formed.
Figures 3-6 show that the extended UNIQUAC I NapSO.
model is able to give a good representation of the
solidPliquid phase equilibria in the system NaCl-
Na2S04-Hz0 in the whole temperature range from
0 to 100°C. It is seen, that the appearance and disap-
000 / I
0.00 100 2 00 3.00 4.6 0

pearance of the various solids are correlated very NaaSO, (moles/kg HaO)
nicely by the model.
Fig. 8. Calculated (-) and experimental (A) (Silcock, 1979)
Table 5 shows a comparison of calculated and
salt solubilities in the ternary system Na,SO,-K,SO,-Hz0
experimental invariant points for the same system at at 50°C.
0~100°C. The experimental data are average vaIues of
data taken from the extensive data compilation by
Silcock (1979). From the table it appears that the The SLE phase diagrams for the system Na2S04-
calculated invariant points represent well the meas- KzS04-Hz0 at the temperatures 25, 50, 75 and
ured values considering the experimental uncer- 100°C are shown in Figs 7-10. Owing to the insuffi-
tainties. cient experimental information for this system, the

Table 5. Experimental and calculated invariant points for the system NaCI-Na,SO,-H,O at 0-100°C

* f%ac1 4kSO4 %11SO~

eNaC1 (Calc.) WP.) (Calc.) Solid phases

0 5.98 5 0.04 6.06 0.13 f 0.009 0.13 NaCl + Na,SO,. iOH,

25 5.59 * 0.09 5.51 0.70 f 0.03 0.73 NaCl + Na,SO,
3.37 + 0.08 3.10 1.49 f 0.011 1.56 NaCl + Na,SO,. IOH,O
50 5.71 * 0.09 5.73 0.53 * 0.02 0.58 NaCl + Na,SO,
100 6.33 * 0.10 6.27 0.46 + 0.009 0.42 NaCl + Na,SO,

‘mi: mol i/kg H,O; riii: average m,

*u: standard deviation of experimental data.
3156 HENRIK NICOLAISEN et al.

Na2S04-KzS04-Hz0 at 75°C KCl-K,&30q-Hz0

1.40 1.50

I.... 25’C
1.20 . . . . . 50°C
00000 100°C

%oo
X

20 80 90

‘;; <o.,o
-
010.60
c I
z
E
z
,O.‘iO
B
d -0.40 -
*
g 0.20
E
ST
000 T
k?? KC1
0.00 0.50 loo 150 2.00 2.50 3.00 3.50 4. 60 0.00 1 I - -I
Na,SO, (moles/kg HzO) 0.00 200 4.00 6.00 8.00
KC1 (moles/kg HzO)
Fig. 9. Calculated (-_) and experimental (A) (Silcock, 1979)
salt solubilities in the ternary system Na,SO,-K,SO,-H,O Fig. 11. Calculated (-) and experimental (B, l or 0)
at 75°C. (Silcock, 1979) salt solubilities in the ternary system
KCI-K,SO,-H,O at 25, 50 and 100°C.

Na2S04-KzS04-Hz0 at 100°C
-I

NaZS04 (moles/kg
250
H20)
3.00
;;..jr.r
-yr-y-J 0.40
X SO.'_
0.80 0.80 1.oo

Fig. 10. Calculated (-) and experimental (A) (Silcock, 1979) Fig. 12. Calculated (-) and experimental (A) (Silcock, 1979)
salt solubilities in the ternary system Na,SO,- K,SO,-H,O phase diagram for the reciprocal system (Na+, K+)(Cl-,
at 100°C. SO:-)-Hz0 at 25°C.

(N~+,K+)(C~-,S~,'-)-H~~ at 100°C
L.00

calculated phase diagram for the system could n_ot be

compared with experimental data at temperatures
below 25°C. As is clear from the diagrams, we obtain 0 80

a good correlation of the solubility of the complex

double salt NaK3(S0& by the extended UNIQUAC
0.60
model. The good result for the calculation at 100°C
should also be noted. This result is especially satisfac-
tory because very few experimental data, for the para-
+a
so.40
meter estimation, were available at this temperature.
Figure 11 shows that satisfactory results for the salt
system KCI-KZS04~~Ha0 at 25-100°C are also ob- 0.20

tained. KC, K.S%

The SLE phase diagrams for the reciprocal system i
(Na+, K*)(Cl-, SOi-)-H,O at 25 and 100°C on
0.00
t
J
0 20 0.40
I I
0.60
I
0.80
i 1.00
a water-free basis are shown in Figs 12 and 13. The Xso*‘-
values on the horizontal axis are the anion mole
Fig. I 3. Calculated (-) and experimental (A) (Silcock, 1979)
fractions of sulfate ions, defined in eq. (31). The values phase diagram for the reciprocal system (Na+, K*)(CI-,
on the vertical axis are the cation mole fractions of SOi-)-Hz0 at 100°C.
 Correlation and prediction of mineral solubilities 3157

sodium ions in the saturated solution. ri UNIQUAC volume parameter of spe-

cies i
30:- mh+
X so:- = 1
X Na* = R the universal gas constant
2 mcl- + msof- mN.+ + mK+. t temperature, “C
(31) T temperature, K
A more detailed description of this type of diagram r, reference temperature, K
may be found elsewhere (Nyvlt, 1977). u, ua, ut UNIQUAC interaction parameters
The two figures show that the model is able to give Wi weighing factor
XI mole fraction of species i
predictions of multicomponent phase equilibria over
anion, cation mole fraction defined in eq.
a broad temperature range with a satisfactory result. xi

The calculated diagrams are predicted using the ex- (3 1)

z UNIQUAC coordination number
tended UNIQUAC model, with parameters obtained
from binary and ternary VLE and SLE data. As is =i charge number of ion i
clear, the model is able to predict the disappearance of
Greek letters
the solid phase of Na,SO,. 10H20 when the temper-
symmetric rational activity coefficient of
ature is increased from 25 to 100°C. Yi
species i
Y* unsymmetric rational activity coefficient
CONCLUSION of species i
The applicability of the extended UNIQUAC Yi.* symmetric molal activity coefficient of
model for correlation and prediction of mineral species i
solubilities has been tested. The model contains only *
Y;,_ unsymmetric molal activity coefficient of
binary parameters which are linearly dependent on species i
the temperature. Oi surface area fraction of species i
The simplified model was able to give good results Pi chemical potential of species i
for the calculation of binary vapor-liquid equilibria in stoichiometric coefficient.
spite of the fact that the main emphasis was on ternary Zi volume fraction of species i
SLE data when the model parameters were estimated. I/lij UNIQUAC parameter, defined in eq.
The result of calculations of mineral solubilities (20)
for the four ternary systems NaCI-KCI-H20,
Na#O,-K,SO,-H,O, NaCI-Na,SO,-H,O and Subscripts
KCI-KzS04-Hz0 confirm that the model contains C combinatorial
the necessary mathematical flexibility required for D-H DebyeeHiickel
correlation of salt solubilities at various temperatures. R residual
Furthermore, it is demonstrated that the model, with W water
parameters obtained from the binary VLE data and + cation
ternary SLE data, can be extended to give good pre- - anion
dictions of mineral solubihties in multicomponent
systems. Superscripts
C combinatorial
D-H Debye-Htickel
NOTATION E excess property
A Debye-Hiickel constant m molal quantity
a, waler activity E* unsymmetric excess property
6 Debye-Hiickel parameter R residual
CP molar heat capacity 0, *, V reference states
D dielectric constant of water cc infinite dilution
d density of pure water
F objective function
9 molar Gibbs function REFERENCES
h molar enthalpy Abrams, D. S. and Prausnitz, J. M., 1975, Statistical thermo-
I ionic strength, mol/kg HZ0 dynamics of liquid mixtures: a new expression for the
K solubility product, defined in eq. (6) Gibbs energy of partly or completely miscible systems.
molality of species i, mol/kg Hz0 A.1.Ch.E. J. 21(l), 116-128.
mi Atkins, P. W., 1986, Physical Chemistry, 3rd Edition. Oxford
MW molar mass of water University Press, Oxford.
ni mole number of species i Fitch, B., 1970, How to design fractional crystallization. Ind.
n total mole number Engng Chem. 62,633.
P pressure Fletcher, R., 1971, A modified Marquardt subroutine for
PSSl non-linear least squares. Harwell Report, AERE R. 6799.
water saturation pressure Fowler, R. H. and Guggenheim, A. E., 1949, Statisrical
4iw UNIQUAC surface area parameter of 77wrmodynamics. Cambridge University Press, Cam-
species i bridoe.
3158 HRNRIK NW

Goldberg, R. N., 1981, Evaluated activity and osmotic coef- Sander, B., 1984, Extended UNIFAC/UNIQUAC models
ficients for aqueous solutions: thirty-six uni-bivalent elec- for (1) gas solubilities and (2) electrolyte solutions. Ph.D.
trolytes. J. phys. Chem. Rel: Data 10(3), 671-764. thesis: institut for Kemiteknik, DTH.-
Hamer, W. J. and Wu, Y.-C., 1972, Osmotic coefficients and Sander, B., Fredenslund, Aa. and Rasmussen, P. 1986, Calcu-
mean activity coefficients of uni-univalent electrolytes in lation of vapor-liquid equilibria in mixed solvent/salt sys-
Water at 25°C. J. phys. Chem. Ref Data I(4). 1047-1099. tems using an extended UNIQUAC equation. Chem.
Harncd, H. S. and Owen, B. B., 1958, The Physical Chemistry Engng Sci. 41(5), 1171-1183.
of Electrolyte Sohdions, 3rd Edition. Sander, B., Fredenslund, Aa. and Rasmussen, P. 1986, Calcu-
Harvie. C. E. and Weare. J. H.. 1980. The brediction of lation of solid-liauid ecluilibria in aaueous solutions of
mint&l solubilities in’ natuial waters: ‘the Na K nitrate salts using an &tended UNIQUAC equation.
Mg-Cl-SO&-H,0 system from zero to high concentration Chem. Ensna Sci. 41(5), 1197-1202.
at ?YC. G&hi&d Cosmochimica Acta-&, 981-997. Schumm, R. H., Wagman, D. D., Bailey, S. M., Evans, W. H.
Linke, W. F. and Seidell, A., 1965, Soluhilities of Jnnrgunic~ and Parker, V. B., 1973, National Bureau of Standards,
and Metal Organic Compounds. American Chemical Technical Note 270-7.
Society, Washington, DC. Silcock. H. L.. 1979. Soluhilities of Inoraonir and Oraonir
.I
Maurer, G. and Prausnitz, J. M., 1978, On the derivation and Combaunds.‘Perga’mon, Oxford. .’ ”
extension of the UNIQUAC equation. Fluid Phase Equi- Skiold-Jorgensen, S., 1980, On the UNIFAC and
libria 2, 91-99. UNIQCAC models. Ph.D. thesis, Institut for Kemiteknik,
Nyvlt, J., 1977, Solid-Liquid Phase Equilibria. Elsevier, DTH.
Amsterdam. Wagman, D. D., Evans, W. H., Parker, V. B., Halow, I.,
Parker, V. B., Wagman, D. D. and Evans, W. H., 1971, Bailev. S. M. and Schumm. R. H.. 1968. National Bureau
National Bureau of Standards, Technical Note 270-6. of Standards, Technical N&e 270-3.
Pitzer, K. S., 1973, Thermodynamics of electrolytes. 1. Theor- Wagman, D. D., Evans, W. H., Parker, V. B., Halow, I.,
etical basis and general equations. J. phys. Chem. 77(2), B&y, S. M. and Schumm, R. H., 1969, National Bureau
268-277. of Standards, Technical Note 270-4.
Pitter, K. S. and Mayorga, G., 1973, Thermodynamics of Wagman, D. D., Evans, W. H., Parker, V. B., Halow, I.,
electrolytes. II. Activity and osmotic coefficients for strong Bailey, S. M. and Schumm, R. H., 1971, National Bureau
electrolytes with one or both ions univalent. J. phys. Chem. of Standards, Technical Note 270-X
77( 19), 2300-2308. Wagman, D. D., Evans, W. H., Parker, V. B., Schumm, R. H.
Pitzer, K. S. and Mayorga, G., 1974, Thermodynamics of and Nuttall, R. L., 1981, National Bureau of Standards,
electrolytes. 111. Activity and osmotic coefficients for 2-2 Technical Note 270-8.
electrolytes. J. Sol. Chem. 3, 539-546. Wood, J. R., 197.5, Thermodynamics of brine-salt equi-
Pitzer, K. S. and Kim, J. J., 1974, Thermodynamics of elec- libria-l. The systems NaCI-KCI-MgCI,-CaCl,-H,O
trolytes. IV. Activity and osmotic coefficients for mixed and NaClpMgSO,-Hz0 at 25°C. Ceochimica Cos-
electrolvtes. J. Am. Gem. Sot. 96. 5701-5707. mochimica Acta 39, 1147-1163.
Pi&r, K.-S., 1975, Thermodynamics of electrolytes. V. Ef- Zemaitis, J. F., Clark, D. M., Rafal, M. and Striver, N. C.,
fects of higher-order electrostatic terms. J. Sol. Chew 4, 1986, Handbook of Aqueous Electrolyte Thermodynamics.
249-265. American Institute of Chemical Engineers, New York