The Journal of Supercritical Fluids, 1992,5, 283-295 283

Calculatiowof Interaction Parameters for Binary

Solid-SCF Equilibria Using Several EOS and -
Mixing Rules?
Angel C. Caballero, Luz N. Hernindez, and L. Antonio Esthez*

University of Puerto Rico, Department of Chemical Engineering,

Mayagiiez, PR 00681-5000

Received October 8, 1991; accepted in revised form June 18, 1992

Binary interaction parameters (BIPs) have been computed for the description of solid-supercritical-
fluid equilibrium using solubility data. Several combinations of cubic equations-of-state and mixing rules
have been studied, computing parameters and average errors for over 30 systems for which experimental
data were available in the literature. The computation was done by defining an objective function, and
minimizing it by a suited algorithm - Fibonacci (golden section) algorithm to find one binary interaction
parameter and the Nelder and Mead (Simplex) method to find IIVOinteraction parameters. Equations-of-
state (EOS’s) chosen for this study were Redlich-Kwong (RK), Soave (SRK), and Peng-Robinson
(PR), whereas the mixing rules used were quadratic (van der Waals) and cubic. Good agreement was
found between computed values and the few values found in the literature. Results using one interaction
parameter for the three EOS indicate that, globally, there is no significant difference in the prediction abil-
ity of these EOS. Differences among the EOS were found when the errors for individual systems were
compared. These differences were explained in terms of the ability of the EOS to predict specific molecu-
lar interactions that occur in these systems. For the PR-EOS, two mixing rules were compared. The cor-
responding results show no substantial difference for these mixing rules. Finally, the results for the PR-
EOS using one BIP were compared to those for which two BIPs were used. A marked reduction in the
errors was observed when the second BIP was included in the analysis.

Keywords: supercritical fluids, solid-SCF equilibria, cubic equations-of-state, binary interaction parameters, mixing rules

INTRODUCTION Reid,5 the use of these equations to deal with solid solutes
Cubic equations-of-state (EOS) have been widely implicitly assumes the solid phase to be a subcooled liq-
employed to predict solubilities of solids in supercritical uid, and as such, the integration step used when equating
fluids. Most of these equations, for example, the fugacities proceeds essentially from a vapor to a liquid
Redlich-Kwong,’ the Soave,* and the Peng-Robinson3 phase. In spite of this, these equations have been found
EOS, are essentially modifications made to the original to correlate binary solid-SCF experimental data quite well
van der Waals EOS4 due to the latter’s inability to accu- in regions away from criticality, especially data for non-
rately predict properties of compressed gases and liquids. polar symmetric systems.
The Peng-Robinson3 EOS, for instance, was proposed as Cubic equations-of-state account only for dispersion
a modification to the Soave EOS2 due to the poor predic- ,forces between the components of a mixture. Specific
tion capabilities of liquid densities exhibited by the latter. chemical interactions such as hydrogen bonding and acid-
These equations-of-state and their temperature-dependent base type interactions, and complexities in the molecular
functions were originally derived to describe VLE sys- interaction fields due to large differences in structure, size,
tems. It has been suggested that these equations-of-state shape, polarity, and energies of the components as evi-
may not be appropriate to describe systems in which solid denced by large differences in their critical temperatures,
solutes are involved. As pointed out by Schmitt and can only be accounted for indirectly through the inclusion
of adjustment or correction factors in mixing and combi-
i Paper presented at the 2nd International Symposium on nation rules which give the composition dependence of
Supercritical Fluids, May 20-22, 1991, Boston, MA, USA. the equations-of-state. These correction factors are known

0896-8446/92/0504-0283$4.00/O 0 1992 PRA Press

284 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, NO. 4, 1992

as binary interaction parameters (BIPs), and they measure BIP or one pair of average BIPs applicable to a range of
the deviation from the behavior expected in nonpolar experimental temperatures and pressures. This way of re-
symmetric systems. The most commonly used of the porting BIP values might not be appropriate if these val-
mixing and combination rules are the conventional van ues are intended to be used for predictive purposes. From
der Waals (VDW) rules. They express the mixture cohe- the work of other researchers it is known that, at least k,,
sive energy parameter, amix, and the mixture repulsive en- is temperature dependent. This dependency is strongest
ergy parameter, bmix, of the EOS in terms of a quadratic for those solid compounds with low melting points and/or
composition dependence of pair parameters. The binary solubilities above y = 0.01 at the highest temperatures
interaction parameter, k,, measures the deviation from ge- and pressures.1o The report of one average BIP applicable
ometric intermolecular attractions assumed for the unlike to a range of experimental temperatures ignores the tem-
cohesive energy parameter, aij, while the binary interac- perature dependency of the parameter, and implicitly at-
tion parameter, hc (variously called ~6 or 1Gby other au- tributes the variations of the parameter exclusively to ran-
thors), measures the deviation from arithmetic inter- dom errors. Calculated solubility values are also ex-
molecular repulsions assumed for the unlike repulsive en- tremely sensitive to the values of the binary interaction
ergy parameter, bij. Both kii and hij can be either positive parameters. Small variations in the BIP values may cause
or negative. A physical interpretation has been ascribed large deviations in the calculated solubilities from their
to negative k, values as being indicative of the presence experimental values. An average BIP value may not ade-
in the mixture of specific chemical interactions such as quately represent the experimental data for all temperatures
hydrogen bonding.6 Interpretation of h, values is more included within the range. It is also known that the EOS
difficult since there are not sufficient data on this parame- does not fit the data equally well at all temperatures and
ter to establish trends in its values; most of the re- pressures. Regarding the second binary interaction param-
searchers have followed the example of Peng-Robinson3 eter, report of average hii values offers limited information
and have set hij equal to zero in their regressions of to help establish trends in the values of this parameter.
experimental data. Recent work in the use of cubic EOS to represent
Even though setting 12, equal to zero has been solid-SCF equilibria has focused on the need to develop
common practice in the use of cubic EOS to predict solu- new mixing rules and to cope with the limitations im-
bilities, it has been demonstrated that both interaction pa- posed by the use of conventional corresponding states
rameters are needed when cubic EOS are used to predict theory based on critical properties. It is generally recog-
phase behavior. McHugh et a1.7 used the PR-EOS with nized that the conventional VDW mixing rules are too
two interaction parameters fitted to the P-T trace of the simple to describe adequately the composition dependence
SLG curve to generate the phase diagram for the naphtha- of the EOS in mixtures of vastly different molecules.
lene-xenon system. They found that it was not possible These highly asymmetric mixtures comprise the majority
to obtain simultaneously a good fit of both the entire of the solid-SCF systems of interest. Since the fugacity
SLG curve and the upper critical endpoint (UCEP). coefficients are calculated out of the composition depen-
Furthermore, while the use of only one adjustable parame- dence of the EOS, the choice of mixing rule is crucial in
ter (the parameter corresponding to the attractive term of determining the chemical potentials. The selection of
the EOS, kij) allowed them to obtain a reasonably good mixing rule is especially important because the expres-
prediction of the SLG curve, the second adjustable param- sion for the chemical potential contains the derivative of
eter (the parameter corresponding to the repulsive term of the mixing rule, which tends to magnify any errors in the
the EOS, hij) was needed to obtain a good prediction of mixing rule itself.” Rao and Mukhopadhyay12 have pro-
the UCEP- It was found that changing the kij values had posed a covolume dependent mixing rule for the attraction
a large effect on the predicted UCEP temperature and a energy parameter, amix, of the PR- and Soave EOS, ex-
smaller effect on the predicted UCEP pressure, while pressed in terms of quadratic composition dependence of
changing the hij values had a large effect on the predicted uijlbii2’3. The mixture repulsive energy parameter, bnli,,
UCEP pressure and a smaller effect on the predicted is expressed by the simple quadratic mixing rule with 11,
UCEP temperature. The work of McHugh et aL7 corrob- set to zero. Results show a reduction in the sensitivity of
orates the earlier findings of Dieters and Schneider,s who the calculated solubility values to the binary interaction
calculated phase diagrams for binary mixtures at high parameter, kg, and an overall improvement of 3% in
pressures using the RK-EOS and also recommended the %AARD over the case in which the conventional VDW
use of two interaction parameters to fit high-pressure mixing rules are used.
phase behavior. The lack of reliable critical data for many solid so-
Recently Lee et a1.9examined the Redlich-Kwong,’ lutes of interest, especially complex organic molecules,
the Soave,2 and the Peng-Robinson3 EOS to evaluate poses another problem when using cubic EOS to model
their capability for correlating solubility data of solids in solid-SCF systems. The usual approach has been to use
compressed gases. They correlated available experimental several estimation methods.” The use of estimation
data for 35 binary systems using both one and two binary methods puts an additional burden on the correction fac-
parameters. Results were reported in terms of one average tors or BIPs since they must correct not only the inade-
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 285

quacies of the EOS, but also the uncertainties introduced part, the PR-EOS with two binary interaction parameters,
by the estimation methods. To avoid the use of estimated kiz and h12, is used to correlate the experimental data.
properties, Schmitt and Reid5 used a modified Peng- Two mixing rules are used: the conventional VDW
Robinson EOS in which the pure component parameters (quadratic) rule and a cube-root mixing rule originally
for the solid solutes, a2 and b2, were treated as variables. proposed by Viswanath et a1.‘s,t9 and corrected by
The pure component parameters for the SCF solvent, a, Caballero and EstCvez.Zo It is hoped that the availability
and bl, were calculated from their critical properties as of data for h12 will help to develop trends in the values of
usual. They used the conventional VDW mixing and this parameter. The use of a second binary interaction pa-
combination rules with kt2 and h,2 set to zero, that is, no rameter in cubic EOS will also help other researchers to
binary interaction parameters were used. The modified make a fairer assessment of the performance of more
Peng-Robinson EOS with its two adjustable parameters complex models that contain more than one adjustable pa-
correlated the solubility data better than the classical rameter when compared to the performance of a cubic
Peng-Robinson EOS with a single binary interaction pa- EOS.
rameter, kt2, and with the a2 and b2 parameters calculated
from known critical properties. Dohrn and Brunnert5 EQUATIONS-OF-STATE
have develop empirical correlations for the calculation of The equations-of-state for which binary interaction
pure-component parameters of the Peng-Robinson EOS parameters were calculated can be written in the form orig-
and for the acentric factor w. These correlations are of inally proposed by Schmidt and Wenze121for a generalized
two types - specific correlations for a group of four-parameter cubic EOS
compounds, such as alkanes, aromatic naphthenic
compounds, and carboxylic acids, and general all-fluid N2a
p= NRT (1)
correlations. The only input data needed are the liquid V-(Nb) - V2 +u(Nb)V+w(Nb)2 .
molar volume of a fluid at 20 “C, vL2a, and the normal
boiling point, Tb. Even though these correlations apply
For more than one component the parameters a, b, u, and
to liquid-vapor systems, it would be interesting if the
w refer to the mixture. These parameters usually vary
approaches of Schmitt and Reid5 and Dohm and Brunner15
with temperature, and all may be composition dependent.
could be combined to develop empirical correlations to
This composition dependence is given by the specific
calculate the pure-component parameters for solid solutes
mixing rules used. Parameters a and b are given by
in supercritical solvents. An attempt has been made by
Bartle et al.tO to establish an empirical correlation based
on critical properties of the solute and solvent for the R R2T2
a = a,a; a, = a ’ (2)
binary interaction parameter ktz of the Peng-Robinson PC
EOS. The correlation applies to solid solutes in
supercritical CO2 at temperatures close to 308 K. and
Recent interest in supercritical fluid research has
centered on the behavior of supercritical fluid mixtures
b&,,F
near critical points. The formation of macroscopic clus- c
ters of solvent about the solute has been observed to occur
at points of criticality resulting in an increase in solubil- where a and b for the mixture are usually determined by
ity of the solute in the solvent.16 This kind of behavior the conventional VDW mixing rules. For a binary mix-
is beyond the scope of behavior capable of being predicted ture,
by a conventional cubic EOS. More complex models
have been developed that take into account the clustering
amix = al ,x12 + a22x22 + 2a,2x1x2 (4)
phenomenon. An excellent review on general modeling
of supercritical mixtures, including the use of cubic EOS,
and
has been made by Johnston et a1.t7 In spite of their limi-
tations, cubic EOS exhibit a good simplicity/accuracy
balance which makes them particularly attractive for engi- emix = b,lx,2 + b22x22 + 2b,,x,x2
neering purposes.
The present work examines three cubic EOS for where Xi is the mole fraction of component i, and aii and
their ability to correlate binary solid-SCF solubility data. bii are given by eqs 2 and 3. The combination rules for
The equations are the Redlich-Kwong,’ the Soave,2 and the calculation of al2 and b12 are
the Peng-Robinson3 EOS. The binary interaction param-
eter, kt2, was calculated for 31 systems and for over 100 (6)
isotherms for which experimental data were available.
The conventional VDW mixing and combination rules and
were used for the first part of this work. For the second
286 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992

TABLE I
Equations-of-State

Cubic EOS u W Q;2, ab a

Redlich-Kwong’ 1 0 0.42748 0.08664

Soave2 1 0 0.42748 0.08664 [l + (0.480 + 1.574w - w2)( 1 - &)I2

Peng-Robinson3 2 -1 0.45724 0.07780 [l + (0.37464 + 1.54226~ - 0.26992w2)(1 - A)]’

Equation 11 may be integrated in a general form when eq

(1-h12) (7) 1 is used.22 The result is

where k12 and h12 are the binary interaction parameters. In ei =%(Z-l)- In (Z-B)+AQi (12)
When h12 is set equal to zero, eq 5 simplifies to

where the expression used to evaluate Qi depends on the

value of the discriminant A defined by
i=l

A=u2-44~. (13)
A cube-root mixing rule for the amix was recently
proposed by Viswanath et a1.*8,*9 and corrected by If A # 0. then
Caballero and EstCvez.20 It is given by

a 1,3_ 44i,)1’3 +x2(a22M2Y3

nm -
XlMl ‘I3 + x2M2”3

+ $[(:+l)A-(5~-2j,)] (14)
+ (JiGGy3x1x2(1 -42) (9)

whereas for A = 0,
where Mi is the molecular weight of component i. The
bki, is calculated using the conventional VDW (quadratic)
mixing and combination rules. Qi = L[~+(+++++l] (1%
The three cubic equations-of-state employed in this
work, along with their parameters, are summarized in
Table I. where

SOLUBILITY OF SOLIDS IN SCF

The solubility of a solid species i in a supercritical
‘; qgi -,i++ (16)
solvent may be expressed as
and

(10)
si=(wi-w)+2w+. (17)

where 4; is the fugacity coefficient which is given by

The expression used to evaluate the parameter L in eqs 14
and 15 also depends on the value of A. If A > 0,
In&=&j -
-&zJ”,Nj -J#- In z . (11)
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 287

TABLE II
Physical Properties of the Supercritical Solvents

Solvent M PC TC 0 Reference
PW 09

chlorotrifluoromethane 104.459 3.92 301.95 0.108 23

carbon dioxide 44.010 7.38 304.25 0.225 23
ethane 30.070 4.88 305.35 0.098 23
ethylene 28.054 5.03 282.40 0.089 13
fluoroform 70.013 4.88 299.15 0.275 23

TABLE III
Physical Properties of the Solutes

Solute M PC TC w A B 103 x PO’ Reference

(MW W 09 ( m3/mol)

benzoic acid 122.124 4.50 752.00 0.6200 14.408 4618.1 0.0965 9

naphthalenee 128.174 4.05 748.15 0.3020 13.583 3733.9 0.1103 23
2,3-dimethylnaphthalene 156.23 3.22 785.00 0.4240 14.0646 4302.5 0.1547 9
2,6-dimethylnaphthalene 156.23 3.22 777.00 0.420 1 14.4286 4419.5 0.1547 9
1,4-naphthoquinonee 158.16 4.12 792.15 0.5750 14.735 4739.4 0.1112 23
hexamethylbenzenee 162.276 2.38 752.00 0.4980 13.1336 3855.0 0.1526 9

anthracene
fluorene 178.234
166.23 3.12
2.99 869.30
821.00 0.3531
0.4070 12.147
14.2046 4397.6
4561.8 0.1426
0.1393 t::
phenanthrene 178.24 3.17 882.55 0.3299 14.631 4873.4 0.1512 (“)
acridinee 179.22 3.19 883.15 0.4980 13.721 4740.1 0.1783 23
hexachloroethane 236.74 3.45 714.60 0.1630 10.6322 2600.94 0.1132 27#
triphenylmethanee 244.34 2.24 863.00 0.5760 14.7858 5228.0 0.2409 9

e Estimated values
(*) P,, T,, and w from reference 24; A and B from reference 25; Pi from reference 26.
# ’ A*&dB calculated from P vs. i data in reference 27.

lated. The tables also show the references from which the
L=& ln[ ~~:~~~~j] (18) data were taken. The Antoine constants for the solid so-
lutes of Table III correspond to the Antoine equation writ-
ten as
ifA=O,
log psat=A-$
2
L=- (19)
2Z+uB
where Psat is given in Pa and Tin K.
and, if A ~0, Once Pisat,Visa’,and & have been determined, eq 10
can be used to compute the solubility yi at the given P
and T.
L= --& tan-’ (z+t) . (20)
BIPs FROM EXPERIMENTAL DATA
If experimental solubility data are available for a
In the previous equations, underlined variables with sub- given binary, solid-SCF system, then the solubility can
script i represent composition derivatives defined as partial be predicted from a model using optimum binary interac-
molar properties, and they are tabulated elsewhere2* for tion parameters obtained by minimizing an objective
the EOS used in this work. function which measures the deviations between experi-
Tables II and III summarize the physical properties mental and predicted values. The objective function used
of the five supercritical solvents and the twelve solid so- in this work is the root-mean-square (rms) of the absolute
lutes, respectively, for whose systems BIPs were calcu- deviations, which is defined as
288 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992

TABLE IV
Computed k12 and Errors when h12 = 0

Solid Solvent T N-P klz AVERR % Reference

w> RK SRK PR RK SRK PR

benzoic acid co2 45 5 -0.1752 0.0176 0.0167 6.1 I Il.68 13.15 28

benzoic acid co2 55 5 -0.1867 0.0020 0.0022 10.51 6.29 6.46 28
benzoic acid co2 65 5 -0.1941 -0.01 I I -0.0095 15.35 0.67 1.09 28
benzoic acid W-4 45 5 -0.2544 -0.0534 -0.0477 II.49 14.62 15.93 28
benzoic acid W, 55 5 -0.2600 -0.0625 -0.0545 8.13 13.07 14.35 28
benzoic acid (2-4 65 5 -0.2722 -0.077 I -0.0672 4.20 10.77 11.78 28
naphthalene co, 35 13 0.0280 0.1003 0.0980 7.16 10.06 13.13 23
naphthalene co, 36 7 0.0289 0.1016 0.0989 6.80 4.83 4.80 29
naphthalene co, 45 I3 0.0243 0.0985 0.0977 10.40 12.96 17.51 23
naphthalene co2 55 7 0.0102 0.0977 0.0970 15.76 10.85 15.80 29
naphthalene co2 55 15 0.0161 0.0983 0.0992 25.19 14.97 21.29
naphthalene co2 55 22 0.0141 0.0980 0.0980 22.65 13.60 19.83 ;;
naphthalene C&4 12 4 -0.0563 0.0176 0.0196 2.92 I .94 2.80 30
naphthalene C2H4 25 5 -0.0652 0.0128 0.0168 2.29 9.90 13.16 30
naphthalene Cd-L, 25 IO -0.0663 0.0079 0.0099 10.96 18.33 19.45
;‘,
naphthalene CA 25 I5 -0.0658 0.01 I I 0.0146 7.36 15.57 18.60
naphthalene C2H4 35 j 6 -0.0723 0.0122 0.0173 3.76 9.78 13.34 30
naphthalene GH, 45 17 -0.0855 0.01 13 0.0173 22.48 19.12 24.04 30
naphthalene W-4 45 5 -0.0795 0.0002 0.0035 4.46 15.21 15.10
;;
naphthalene ‘2% 45 22 -0.0844 0.0108 0.0173 21.53 23.59 28.40
naphthalene C2H4 50 21 -0.1113 0.0146 0.0181 58.66 26.50 24.59 30
naphthalene W-4 50 4 -0.0839 0.0060 0.0097 9.54 20.96 20.85
;*;
naphthalene Cd-4 50 25 -0.1081 0.0146 0.0181 62.26 28.19 26.24
naphthalene (3-h 20 7 -0.0321 0.0433 0.0427 17.77 18.98 20.39 25
naphthalene (3% 25 7 -0.0330 0.0439 0.0457 5.95 7.28 9.68 25
naphthalene C2H6 35 6 -0.0453 0.037 I 0.0386 9.52 16.34 18.88 25
naphthalene C2H6 35 15 -0.0415 0.0409 0.0448 9.22 15.27 19.99
naphthalene C2H6 35 21 -0.0427 0.0404 0.0437 9.75 16.19 20.92 G
naphthalene C2H6 45 8 -0.0490 0.0404 0.0442 6.67 19.83 24.84 31
naphthalene C2H6 45 12 -0.053 1 0.0395 0.0446 8.29 13.56 19.01
naphthalene C2H6 45 20 -0.0513 0.0404 0.0448 10.34 16.99 21.49 (‘;
naphthalene C2H6 55 14 -0.08 I7 0.0413 0.0448 34.81 14.09 16.1 I 23
naphthalene CHF, 35 6 0.0258 0.0912 0.0956 I I .40 17.29 21.26 23
naphthalene CHF, 45 6 0.0256 0.0895 0.0956 7.83 17.58 21.90 23
naphthalene CHF, 55 6 0.0270 0.0897 0.0970 4.74 15.55 20.23 23
naphthalene CCIF, 35 7 0.0168 0.0833 0.0895 8.83 10.87 14.26 23
naphthalene CCIF, 45 7 0.0185 0.0846 0.0919 8.07 12.10 16.24 23
naphthalene CCIF, 55 7 0.0170 0.0829 0.0919 4.14 10.52 15.25 23
2,3-dimethylnaphthalene co2 35 5 -0.01 I3 0.1033 0.0989 4.81 2.84 4.59 28
2,3-dimethq-inaphthalene co2 45 5 -0.0055 0.1074 0.1042 2.20 9.44 12.46 28
2,3-dimethylnaphthalene co2 55 5 -0.0005 0.1104 0.1078 1.74 9.42 12.08 28
2,3-dimethylnaphthalene Cd-b 35 6 -0.1087 0.0194 0.0230 11.38 2.12 3.55 28
2,3-dimethylnaphthalene C2H4 45 6 -0.1190 0.0141 0.0190 20.37 3.74 1.19 28
2,3-dimethylnaphthalene G-6 55 6 -0.1444 0.0064 0.0120 42.52 2.28 I.1 I 28
2,6-dimethylnaphthalene (332 35 5 -0.0046 0.1063 0.1019 4.95 5.71 6.72 28
2,6-dimethylnaphthalene co2 45 5 -0.0082 0.1025 0.0992 5.82 4.20 5.76 28
2,6-dimethylnaphthalene co2 55 5 -0.0058 0.1030 0.1010 7.50 8.46 9.70 28
2,6-dimethylnaphthalene C2H4 35 6 -0.1040 0.0184 0.0226 8.38 2.10 3.10 28
2,6-dimethylnaphthalene C2H4 45 6 -0.1 I25 0.0126 0.0181 15.82 4.57 2.62 28
2,6-dimethylnaphthalene C2H4 55 6 -0.1232 0.0082 0.0150 21.60 2.00 0.45 28
1,4-naphthoquinone co2 45 6 -0.1199 0.0572 0.0566 6.94 14.96 17.69 23
1,4-naphthoquinone co2 55 6 -0.1 I81 0.056 1 0.0566 3.78 12.45 15.73 23
I ,4-naphthoquinone (332 70 6 -0.1202 0.05 I9 0.0539 6.51 17.04 19.88 23
I ,4-naphthoquinone C2H6 35 7 -0.0486 0.1 142 0.1155 13.65 14.19 18.34 23
I ,4-naphthoquinone C2H6 45 7 -0.0537 0.1081 0.1107 7.90 9.90 13.74 23
1,4-naphthoquinone C2H6 55 7 -0.0568 0.1036 0.1074 12.96 16.35 19.30 23
I ,4-naphthoquinone C2H6 70 7 -0.0548 0.1019 0.1074 6.29 9.63 11.81 23
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 289

TABLE IV
(continued)

Solid Solvent T N-P kl2 AVERR % Reference

(“(3 RK SRK PR RK SRK PR

1,4-naphthoquinone CHF, 45 6 -0.1592 0.0159 0.0214 22.00 29.73 32.62 23

1,4-naphthoquinone CHF, 55 6 -0.1573 0.0119 0.0185 14.06 13.55 17.09 23
1,4-naphthoquinone CCIF, 45 6 -0.0380 0.1187 0.1231 19.20 22.71 25.89 23
1,4-naphthoquinone CCIF, 55 6 -0.04 15 0.1116 0.1175 8.66 13.68 17.16 23
hexamethylbenzene co2 30 7 0.0725 0.1973 0.1849 41.96 43.51 36.61 25
hexamethylbenzene co2 50 10 0.0793 0.1977 0.1888 58.61 49.94 43.39 25
hexamethylbenzene co2 70 10 0.0685 0.1862 0.1820 69.06 55.99 47.41 25
hexamethylbenzene c2H4 25 9 -0.0157 0.1222 0.1204 57.26 56.30 43.97 25
hexamethylbenzene c2H4 45 9 -0.0214 0.1171 0.1 178 65.65 61.85 46.53 25
hexamethylbenzene C2H4 70 6 -0.0214 0.1169 0.1237 57.64 37.33 25.62 25
fluorene co2 30 7 -0.0132 0.0921 0.0868 26.17 26.92 17.00 25
fluorene co2 35 6 -0.0146 0.0903 0.0864 22.09 20.67 12.39 25
fluorene co2 50 9 -0.0143 0.0885 0.0868 22.96 15.19 7.32 25
fluorene co2 70 8 -0.0232 0.0795 0.0833 33.16 16.21 8.42 25
fluorene C2H4 25 6 -0.1049 0.0093 0.0152 22.48 20.39 11.59 25
fluorene C2H4 45 7 -0.1155 0.0016 0.0111 32.15 25.75 15.47 25
fluorene C2H4 70 11 -0.1352 -0.0084 0.0042 39.58 23.96 12.18 25
anthracene (332 30 4 0.0256 0.1042 0.0992 26.94 27.39 18.90 25
anthracene co, 35 5 0.0250 0.1027 0.0968 22.22 20.65 18.20 24
anthracene co2 45 6 0.0245 0.1001 0.0959 10.61 5.57 5.86 24
anthracene co2 50 10 0.0150 0.0910 0.0901 23.57 15.43 9.28 25
anthracene co2 70 9 0.0035 0.0744 0.0767 24.14 10.30 7.10 25
anthracene C2H4 50 7 -0.0702 0.0113 0.0214 14.98 12.00 6.44 31
anthracene C2H4 70 9 -0.0738 0.0037 0.0146 14.23 11.79 10.04 31
anthracene C2H4 85 13 -0.0903 -0.0159 -0.0013 8.84 7.22 7.30 31
anthracene C2H6 30 6 -0.0353 0.0483 0.0522 20.13 20.26 12.67 25
anthracene C2H6 35 4 -0.0353 0.0490 0.0530 23.78 23.60 15.25 25
anthracene C2H6 50 10 -0.0495 0.0347 0.0404 16.92 15.79 9.91 25
anthracene C2H6 70 7 -0.0672 0.0159 0.0243 13.33 11.18 8.31 25
phenanthrene co2 30 8 0.0084 0.0797 0.0762 15.06 15.31 8.67 25
phenanthrene co2 45 5 0.0088 0.0782 0.0744 7.12 7.40 8.06 28
phenanthrene co2 50 6 0.0069 0.0776 0.0773 15.73 10.39 7.31 25
phenanthrene co2 55 5 0.0084 0.0753 0.0729 7.33 4.77 5.78 28
phenanthrene co2 65 5 0.0129 0.0758 0.0749 9.54 3.65 5.23 28
phenanthrene (32 70 7 0.0214 0.0877 0.0901 19.90 9.28 9.40 25
phenanthrene C2H4 25 11 -0.0747 -0.0007 0.0016 5.75 7.20 9.22
phenanthrene C2H4 25 14 -0.0462 0.0258 0.0294 94.62 95.57 98.05 :*:
phenanthrene C2H4 45 11 -0.0795 -0.0057 -0.0005 7.77 8.12 8.88 31
phenanthrene C2H4 45 5 -0.0817 -0.0075 -0.0025 4.36 4.65 6.07
phenanthrene C2H4 45 18 -0.0643 0.0093 0.0150 65.60 67.52 69.46 ;:
phenanthrene C2H4 55 5 -0.0948 -0.0199 -0.0135 19.28 12.83 10.39 28
phenanthrene C2H4 65 5 -0.0519 0.0182 0.0265 73.51 76.98 78.39 28
phenanthrene C2H4 70 8 -0.0844 -0.0132 -0.0046 12.79 5.14 4.74
phenanthrene C2H4 70 9 -0.0566 0.0150 0.0250 87.24 91.50 93.39 f*;
phenanthrene C2H6 30 6 -0.0338 0.0407 0.0413 7.47 7.25 5.26 25
phenanthrene C2H6 40 4 -0.0608 0.0176 0.0208 6.88 5.80 3.80 25
phenanthrene C2H6 60 7 -0.0508 0.0268 0.0333 12.97 11.39 11.11 25
acridine co2 35 6 -0.0079 0.1259 0.1208 12.56 10.02 7.52 23
a&dine co2 45 8 -0.0122 0.1204 0.1166 6.67 2.74 7.00 23
acridine co2 55 7 -0.0129 0.1178 0.1157 20.68 24.03 25.40 23
acridine co2 70 7 -0.0141 0.1118 0.1116 15.92 5.42 5.56 23
acridine C2H6 35 6 -0.0289 0.1078 0.1071 23.92 24.81 28.85 23
acridine C2H6 45 10 -0.0320 0.1045 0.1065 13.64 15.99 21.18 23
acridine C2H6 55 9 -0.0546 0.0850 0.0885 13.40 14.75 16.22 23
acridine C2H6 70 7 -0.0640 0.0764 0.0818 7.53 10.62 13.67 23
acridine CHF, 45 6 -0.0274 0.1030 0.1063 16.26 22.84 27.64 23
acridine CHF, 55 6 -0.0270 0.0992 0.1036 9.85 16.50 20.65 23
290 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992

TABLE IV
(continued)

Solid Solvent T N=P klz AVERR % Reference

w> RK SRK PR RK SRK PR

acridine CClF, 45 5 -0.0018 0.1273 0.1314 14.18 17.45 20.79 23

acridine CClF, 55 5 -0.0079 0.1199 0.1255 10.49 15.50 19.03 23
hexachloroethane co2 35 5 0.0989 0.1131 0.1078 8.90 7.77 3.24 28
hexachloroethane co, 45 5 0.0901 0.1047 0.1010 14.33 7.02 2.56 28
hexachloroethane co2 55 5 0.0782 0.0950 0.0924 43.03 43.80 43.08 28
triphenylmethane co2 30 9 -0.0619 0.1078 0.0992 25.63 25.86 14.20 25
triphenylmethane CO2 40 6 -0.0601 0.1084 0.1003 16.43 12.84 6.59 25
triphenylmethane co2 50 7 -0.0578 0.1095 0.1049 21.46 11.26 11.08 25
triphenylmcthane C2I-b 30 8 -0.1213 0.0614 0.0599 17.92 15.83 7.31 25
triphenylmethane C,H, 40 6 -0.1164 0.0643 0.0614 17.32 12.70 8.05 25
triphenylmethane C,H, 50 6 -0.1231 0.0583 0.0548 9.10 5.45 7.47 25

average error for all isotherms 18.86 17.36 17.06

(*) combined data from references at the same temperature.

NP 2

ICC Yi,calc - Yi,exp 1

(22)

in which N, is the number of experimental data points.

Two optimization algorithms are used to minimize OF.
For the first part of this work, in which one binary inter-
action parameter, k,z, is calculated, the golden section
(Fibonacci) algorithm is used. For the second part, in
which two binary interaction parameters, k12 and h12, are
calculated, the Nelder and Mead minimization (Simplex)
algorithm is used. The quality of the fitting of the EOS 15 20 25

to the experimental solubility data is measured in terms of Pressure [MPal

a percentage average error, %AVERR, defined as

NP Figure 1. Comparison of the prediction, using different

EOS, to the experimental values32of solubility (expressed as
mole fraction) of naphthalene in fluoroform at 35 ‘C.
, I
C(Yi,cak
i=l
- Yi.enp)2
table, were as follows: 18.86% for RK 17.36% for SRK,
and 17.06% for PR. On the basis of the overall percent-
AVERR% = 100 v NP
Ye*P age average errors we may conclude that there is no signif-
icant difference between the three EOS analyzed in their
OF N, ability to correlate experimental solubility data. This re-
=lOO-OF
-=lOO N
P
. (23) sult contrasts with the result obtained by other au-
Y exP thors,g,33 who found that the Redlich-Kwong EOS corre-
Yi,exp
c lated experimental solid-SCF solubility data better than
i=l
the SRK and the PR-EOS. In Figure 1, predictions by
RESULTS AND DISCUSSION the three EOS are compared to experimental solubilities
Table IV shows the optimum k12 values calculated of naphthalene in fluoroform at 35 ‘C, showing that for
for each of the EOS tested when no interaction parameter this system the RK-EOS is indeed better than the other
for bmix is used. Overall %AVERR’s were calculated two EOS. Haselow et a1.33investigated experimental su-
from the data in Table IV for each of the three EOS to de- percritical solubility data for 31 systems, 14 of which
termine which equation gave the best description of solid- were correlated using known critical properties, while for
SCF equilibria. Results, shown in the bottom line of the the remaining 17 systems, estimated critical properties
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 291

TABLE V
Binary Interaction Parameters and Errors for Two Mixing Rules

Solid Solvent T Nexp Quadratic Mixing Rule Cubic Mixing Rule Reference

(“c> kl2 42 % Error kl2 hl2 % Error

benzoic acid 45 5 -0.0796 -0.2252 4.05 0.8857 -0.2583 3.81 28

benzoic acid 55 5 -0.0177 -0.0463 5.83 0.9951 -0.0810 5.25 28
benzoic acid 65 5 -0.0171 -0.0178 0.44 0.9875 -0.0659 0.39 28
benzoic acid 45 5 -0.1738 -0.3153 7.86 0.9861 -0.3491 7.54 28
benzoic acid 55 5 -0.1895 -0.3346 2.40 0.9504 -0.3869 2.25 28
benzoic acid 65 5 -0.1796 -0.2750 0.92 0.9544 -0.3533 1.02 28
naphthalene 35 13 0.0580 -0.075 1 5.55 1.2005 -0.0584 3.68 23
naphthalene 35 22 0.055 1 -0.0809 6.62 1.1978 -0.0621 4.32 32
naphthalene 36 7 0.0838 -0.0284 3.56 1.2559 0.0077 2.86 29
naphthalene 45 13 0.0489 -0.0859 4.86 1.1895 -0.0618 3.44 23
naphthalene 45 20 0.0486 -0.0867 5.21 1.1875 -0.0642 3.79 32
naphthalene 55 7 0.0534 -0.0657 2.29 1.2112 -0.0160 1.63 29
naphthalene 55 15 0.0532 -0.0686 3.28 1.2047 -0.0274 2.17
naphthalene 55 22 0.0540 -0.0662 3.34 1.2077 -0.0220 2.42 ;;
naphthalene 55 22 0.0532 -0.0692 4.00 1.2049 -0.0277 2.78 32
naphthalene 12 4 0.0027 -0.0339 1.69 1.3389 -0.0229 1.47 30
naphthalene 12 19 -0.05 13 -0.1392 10.85 1.2424 -0.1498 9.47 32
naphthalene 25 5 -0.0397 -0.1077 1.19 1.2561 -0.1255 1.10 30
naphthalene 25 10 -0.0859 -0.2126 6.45 1.1644 -0.2668 5.43
naphthalene 25 15 -0.05 13 -0.1321 6.08 1.2323 -0.1599 5.12 ;t
naphthalene 25 14 -0.0548 -0.1319 8.37 1.2271 -0.1600 7.37 32
naphthalene 35 6 -0.03 12 -0.0834 1.31 1.2735 -0.0947 1.07 30
naphthalene 35 16 -0.0386 -0.0912 7.41 1.2536 -0.1158 6.09 32
naphthalene 45 17 -0.0284 -0.0657 6.09 1.2784 -0.0742 4.45 30
naphthalene 45 5 -0.0639 -0.1339 4.45 1.1643 -0.2482 3.60
naphthalene 45 22 -0.0319 -0.0707 6.38 1.2700 -0.0854 4.83 ;‘,
naphthalene 45 16 -0.0345 -0.0693 6.05 1.2663 -0.0866 4.85 32
naphthalene 50 21 -0.0090 -0.0263 3.72 0.9722 -0.4901 16.53 30
naphthalene 50 4 -0.0505 -0.1021 3.23 1.2030 -0.1856 2.03
naphthalene 50 25 -0.0087 -0.026 1 4.26 0.9843 -0.4753 18.02 ;;
naphthalene 20 7 -0.0494 -0.1982 8.80 1.2140 -0.2132 7.20 25
naphthalene 25 7 0.0062 -0.0753 4.32 1.3 1 18 -0.0590 3.45 25
naphthalene 35 6 -0.0347 -0.1339 5.15 1.2492 -0.1337 3.82 25
naphthalene 35 15 -0.0233 -0.1204 5.92 1.2675 -0.1095 4.09
naphthalene 35 21 -0.0266 -0.1257 5.94 1.2640 -0.1138 4.18 5
naphthalene 45 8 -0.0335 -0.1307 6.90 1.2479 -0.1289 3.83
naphthalene 45 12 -0.0134 -0.0895 6.69 1.3005 -0.0466 3.91
naphthalene 45 20 -0.0158 -0.0962 8.57 1.2934 -0.0593 5.44 ?d,
naphthalene 55 14 0.0113 -0.0364 3.10 1.0979 -0.3002 13.41 23
naphthalene 35 6 0.0074 -0.1769 8.81 0.9633 -0.1732 8.27 23
naphthalene 45 6 0.0020 -0.1822 7.63 0.9533 -0.1806 7.36 23
naphthalene 55 6 0.0049 -0.1745 5.37 0.9510 -0.1816 5.08 23
naphthalene 35 7 0.0265 -0.1257 6.75 0.8991 -0.1238 6.35 23
naphthalene 45 7 0.0134 -0.1540 5.12 0.8796 -0.1530 4.94 23
naphthalene 55 7 0.0175 -0.1407 3.20 0.8822 -0.1432 3.09 23
2,3-dimethylnaphthalene 35 5 0.0802 -0.0366 2.64 1.3080 -0.0237 2.51 28
2,3-dimethylnaphthalene 45 5 0.0321 -0.1429 2.33 1.2034 -0.1414 2.21 28
2,3-dimethylnaphthalene 55 5 0.0385 -0.1378 1.56 1.2151 -0.1353 1.45 28
2,3-dimethylnaphthalene 35 6 0.0119 -0.0215 2.22 1.4784 0.0064 1.93 28
2,3-dimethylnaphthalene 45 6 0.0202 0.0022 1.09 1.5156 0.0612 1.28 28
2,3-dimethylnaphthalene 55 6 0.0112 -0.0016 0.98 1.5384 0.1082 2.26 28
2,6-dimethylnaphthalene 35 5 0.0750 -0.0537 4.70 1.2973 -0.0459 4.41 28
2,6-dimethylnaphthalene 45 5 0.0713 -0.0555 3.23 1.29 16 -0.0434 2.98 28
2,6-dimethylnaphthalene 55 5 0.0561 -0.0890 6.39 1.2604 -0.0772 5.86 28
2,6-dimethylnaphthalene 35 6 0.0100 -0.025 1 1.84 1.4670 -0.0085 1.82 28
2,6-dimethylnaphthalene 45 6 0.0200 0.0074 2.35 1.5073 0.0500 2.07 28
2,6-dimethylnaphthalene 55 6 0.0130 -0.0033 0.21 1.5044 0.0579 0.62 28
292 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992

TABLE V
(continued)

Solid Solvent T Nexf’ Quadratic Mixing Rule Cubic Mixing Rule Reference

CC) k12 h12 % Error klz hlz % Error

1,4-naphthoquinone (332 45 6 -0.0157 -0.1524 8.99 1.1226 -0.1671 8.72 23

1,4-naphthoquinone co2 55 6 -0.0116 -0.1417 4.82 1.1263 -0.1615 4.66 23
1,4-naphthoquinone (32 70 6 -0.030 1 -0.1690 3.89 1.0705 -0.2188 3.72 23
1,4-naphthoquinone C2H6 35 7 0.0373 -0.1793 4.91 1.4692 -0.1785 4.88 23
1,4-naphthoquinone C2H6 45 7 0.0460 -0.1469 3.34 1.4839 -0.1458 3.28 23
1,4-naphthoquinone C2H6 55 7 0.0233 -0.1898 9.29 1.4458 -0.1877 9.14 23
1,4-naphthoquinone C2H6 70 7 0.057 1 -0.1124 7.05 1.5041 -0.1069 6.94 23
1,4-naphthoquinone CHF, 45 6 -0.1245 -0.3203 16.18 0.7653 -0.3305 16.47 23
1,4-naphthoquinone CHF, 55 6 -0.0392 -0.1241 10.47 0.8983 -0.1480 10.1 1 23
1,4-naphthoquinone CClF, 45 6 0.0141 -0.2501 13.18 0.8904 -0.2502 13.15 23
1,4-naphthoquinone CCIF, 55 6 0.0406 -0.1741 7.46 0.9319 -0.1741 7.43 23
hexamethylbenzene co2 30 7 0.2810 0.1904 14.21 1.7378 0.1944 13.30 25
hexamethylbenzene 50 10 0.3052 0.2287 5.87 1.8098 0.2567 4.80 25
hexamethylbenzene ::; 70 10 0.3045 0.2165 4.16 1.8142 0.2623 5.78 25
hexamethylbenzene C2H4 25 9 0.2210 0.1841 4.84 1.9305 0.2217 3.75 25
hexamethylbenzene C2H4 45 9 0.2036 0.1493 4.48 1.9146 0.2194 2.15 25
hexamethylbenzene C2H4 70 6 0.1803 0.0932 4.11 1.9526 0.2720 2.55 25
fluorene co2 30 7 0.1243 0.0722 8.27 1.4188 0.0775 7.57 25
fluorene co2 35 6 0.1147 0.0532 7.93 1.3988 0.0598 7.43 25
fluorene co2 50 9 0.1064 0.0367 4.08 1.3820 0.0474 3.67 25
fluorene co, 70 8 0.1069 0.0405 4.63 1.3883 0.0626 3.58 25
fluorene C2H4 25 6 0.033 1 0.0326 9.57 1.5434 0.0395 9.02 25
fluorene C2H4 45 7 0.0435 0.0554 9.44 1.5704 0.0734 8.65 25
fluorene C2H4 70 11 0.0284 0.0369 4.72 1.5581 0.0791 4.14 25
anthracene (32 30 4 0.1545 0.1074 5.77 1.5253 0.1076 5.76 25
anthracene co2 35 5 0.1496 0.1118 14.03 1.5142 0.1121 14.01 24
anthracene co2 45 6 0.0887 -0.0155 5.73 1.3794 -0.0150 5.72 24
anthracene co2 50 10 0.1214 0.0607 4.90 1.4510 0.0612 4.89 25
anthracene co2 70 9 0.0880 0.0221 6.62 1.3759 0.0235 6.60 25
anthracene C2H4 50 7 0.0499 0.0549 3.70 1.6252 0.0562 3.67 31
anthracene C2H4 70 9 0.0377 0.0479 9.53 1.6002 0.0497 9.49 31
anthracene c2H4 85 13 -0.0203 -0.0396 6.79 1.4768 -0.0413 6.76 31
anthracene W& 30 6 0.0886 0.0677 8.86 1.6465 0.0674 8.84 25
anthracene C2H6 35 4 0.1045 0.0969 7.84 1.6782 0.0979 7.82 25
anthracene C2H6 50 10 0.0753 0.0685 6.50 1.6232 0.0702 6.47 25
anthracene C2H6 70 7 0.0356 0.0228 8.00 1.5472 0.0246 7.94 25
phenanthrene co2 30 8 0.0896 0.0254 7.56 1.3845 0.0287 7.27 25
phenanthrene co2 45 5 0.0489 -0.0530 6.82 1.2917 -0.0516 6.64 28
phenanthrene co2 50 6 0.0817 0.0079 7.13 1.3646 0.0124 6.95 25
phenanthrene (332 55 5 0.0465 -0.0553 3.98 1.2847 -0.0541 3.86 28
phenanthrene (332 65 5 0.0445 -0.0643 2.23 1.2793 -0.0625 2.22 28
phenanthrene co2 70 7 0.0804 -0.0182 8.90 1.3611 -0.0111 8.72 25
phenanthrene c2H4 25 11 -0.0357 -0.0809 6.01 1.4441 -0.0865 5.91
phenanthrene C2H4 25 14 -0.2350 -0.5621 77.85 1.0464 -0.5602 76.95 ;t
phenanthrene C2H4 45 11 -0.0318 -0.0672 7.36 1.4477 -0.0778 7.01 31
phenanthrene C2H4 45 5 -0.0340 -0.0678 3.86. 1.4428 -0.0791 3.73
phenanthrene C2H4 45 18 -0.1947 -0.4475 56.38 1.1189 -0.4615 54.55 ;;
phenanthrene C2H4 55 5 0.0380 0.1065 4.57 1.6266 0.1441 4.79 28
phenanthrene C2H4 65 5 -0.2916 -0.6995 59.04 0.9054 -0.7407 55.95 28
phenanthrene c2H4 70 8 -0.0093 -0.0100 4.58 1.499 1 -0.0124 4.32
phenanthrene C2H4 70 9 -0.2438 -0.5656 73.09 1.0075 -0.6034 68.61 +‘,
phenanthrene 30 6 0.045 1 0.0082 5.14 1.5671 0.0138 5.06 25
phenanthrene % 40 4 0.0154 -0.0 109 3.48 1.5111 -0.0040 3.24 25
phenanthrene C;H, 60 7 0.0156 -0.0333 9.44 1.5116 -0.025 1 8.70 25
acridine (32 35 6 0.1207 0.0001 7.48 1.4578 0.0156 7.33 23
acridine (32 45 8 0.0894 -0.0571 2.56 1.3736 -0.0524 2.59 23
acridine co2 55 7 0.0644 -0.1070 22.57 1.3118 -0.1056 22.34 23
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 293

TABLE V
(continued)

Solid Solvent T N-p Quadratic Mixing Rule Cubic Mixing Rule Reference

w> k12 h12 % Error ,412 h12 % Error

acridine co2 70 i 0.0970 -0.0305 4.67 1.3983 -0.0136 4.67 23

acridine C,H, 35 -0.0121 -0.2719 6.08 1.4550 -0.2716 6.03 23
acridine C,H, 45 10 0.0397 -0.1422 9.69 1.5593 -0.1395 9.54 23
acridine C,H, 55 9 0.0530 -0.0741 13.50 1.5876 -0.0673 13.15 23
acridine C,H, 70 7 0.0322 -0.1021 8.22 1.5474 -0.0935 7.94 23
a&dine U-IF, 45 6 0.0148 -0.1974 13.62 0.9824 -0.1979 13.56 23
acridine U-IF, 55 6 0.0342 -0.1496 11.77 1.0192 -0.1504 11.74 23
acridine CClF, 45 5 0.0605 -0.1557 13.04 0.9295 -0.1561 13.01 23
acridine CCIF, 55 5 0.0517 -0.1611 10.76 0.9120 -0.1610 10.73 23
hexachloroethane co* 35 5 0.1202 0.0217 1.95 1.6183 0.0505 2.85 28
hexachloroethane co* 45 5 0.1103 0.0154 1.51 1.6389 0.0968 2.46 28
hexachloroethane co2 55 5 -0.2344 -0.5138 23.27 1.2977 -0.2663 19.63 28
triphenylmethane co2 30 9 0.1269 0.0495 8.87 1.6732 0.0555 8.73 25
triphenylmethane CO, 40 6 0.1047 0.0079 6.35 1.6175 0.0160 5.96 25
triphenylmethane co2 50 7 0.094 1 -0.0191 10.19 1.5876 -0.0123 10.18 25
triphenylmethane C2H6 30 8 0.0728 0.0235 5.71 1.8909 0.0460 5.35 25
triphenylmethane C,H, 40 6 0.0727 0.0225 7.27 1.9042 0.0600 7.19 25
triphenylmethane C2H6 50 6 0.0309 -0.0506 5.04 1.8063 -0.0183 4.77 25

average error for all isotherms 8.07 7.80

(*) combined data from references at the same temperature.

were used. They concluded that, for systems with known teraction constant a12. The polarizability of C2H4 is
critical properties, the Redlich-Kwong EOS appears to 42.52 x 1O-25 cm3, while that of CO2 is 29.11 x 1O-25
give the best overall description of solubility. For the cm3.14 For the same solute, and upon changing the sol-
systems for which the critical properties of the solutes had vent from CO2 to C2H4 at the same temperature, it is to
to be estimated, they found that there was a general in-the be expected that the interaction constant, a12, should be
experimental value for all the EOS used, and, at the same larger for C2H4 than for CO2. The ratio of the polariz-
time, the distinction among the equations as to which ability of C2H4 to that of CO2 is approximately 1.5.
equation correlated the experimental data better became When the ratio of the interaction constant aI2 in C2H4 to
blurred. In the present work the values for the critical that in CO2 is calculated, using the RK EOS, for in-
properties of 5 of the solutes investigated, namely, stance, and for the same solute at the same temperature, it
naphthalene, 1,4-naphthoquinone, hexamethylbenzene, gives 1.l. Since this calculation is based on critical
acridine, and triphenylmethane, were estimated. For the properties, it is very likely that the actual ratio of interac-
remaining 7 solutes, namely, benzoic acid, 2,3-dimethyl- tion constants alis is somewhat larger than that predicted
naphthalene, 2,Gdimethylnaphthalene, fluorene, hex- using critical properties. The increase in total average er-
achloroethane, anthracene, and phenanthrene, known criti- ror may point at the inadequacy of using critical properties
cal properties were used. The fact that estimated critical to predict successfully the interaction constant in mixtures
properties were used for slightly more than half of the that are highly asymmetric.
systems investigated in this work (68 out of a total of While no distinction can be made among the EOS
129) may explain why no significant distinction was for either benzoic acid or 2,3-dimethylnaphthalene in
found among the equations when overall average errors are COZ, the data for benzoic acid in C2H4 appear to be better
used as a criterion of comparison. correlated by the RK or the SRK EOS than by the PR-
When the %AVERR’s for individual systems are EOS; on the other hand, the data for 2,3-dimethylnaphtha-
analyzed, however, a distinction may be made among the lene in C2H4 is best correlated by the SRK or the PR-
equations. For example, benzoic acid and 2,3-dimethyl- EOS. This observation could indicate that, besides the
naphthalene show an increase in total average error when use of correction factors or BIPs, the functional form of
the SCF solvent is changed from CO2 to C2H4. A pos- the attractive term of the EOS plays an important role in
sible explanation for this observation could be related to the ability of the equation to account for the effect of
the difference in the polarizabilities of the two solvents nondispersive-type forces in solid-SCF systems. In sys-
and the effect of this difference on the solute-solvent in-
294 Caballero et al. The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992

ever, the behavior of the fli2 values mirrors that of the k12
values, that is, if k12 increases with increasing tempera-
ture, so does 11,~; if k12 decreases, h12 also decreases.
Overall percentage average errors for the two mixing rules
used were 8.07 for the conventional VDW mixing rule
and 7.80 for the Viswanath et al.is.ig cube-root mixing
rule. From the overall %AVERR’s we can conclude that
the Viswanath et al. cube-root mixing rule performs as
well as the conventional VDW mixing rule in represent-
ing solid-SCF equilibria. However, a closer inspection
shows that for most of the isotherms, the cubic mixing
rule performs slightly better, and there are only a few
10-G cases in which the quadratic rule gives less error. Figure
0 10 20 30 40 50 2 shows a comparison of the prediction by the Peng-
Robinson EOS, using two different mixing rules, to the
experimental values 25 of solubility of hexamethylbenzene
Figure 2. Comparison of the prediction by the Peng- in ethylene at 25 and 70 “C. For this particular system,
Robinson EOS, using two different mixing rules, to the ex-
the cubic mixing rule gives a similar prediction at 25 “C
perimental values25 of solubility (expressed as mole fraction)
of hexamethylbenzene in ethylene at 25 and 70 “C. and a slightly better prediction at 70 “C as compared to
the performance of the quadratic mixing rule.
terns where dispersive forces predominate, no significant
difference is likely to be found in the ability of the equa- CONCLUSIONS
tions to model solid-SCF equilibria. When specific The conclusions drawn from the results of this
chemical forces are present, however, the nature and study are summarized as follows. When only one interac-
strength of these forces, as determined by the type of sol-- tion parameter is used, the three EOS tested give similar
vent, will influence the ability of these equations to results in terms of average errors. However, differences
model the system. This ability will depend on how well among the EOS are found when the errors for individual
the functional form of the attractive term of the EOS can systems are compared. These differences can be explained
account for the effect of these chemical forces. in terms of the ability of the EOS to predict specific
Most of the k12 values in Table IV show a decrease molecular interactions that occur in these systems. For
in their values with increasing temperature. This is the the PR-EOS, the quadratic and cubic mixing rules show
trend observed also by Bartle et al.34 The kiz values de- no substantial difference when the average errors are com-
pend on the source of the critical data used to evaluate the pared, although the error for the cubic mixing rule is sys-
pure component parameters of the EOS, the source of the tematically lower than that for the quadratic one. Finally,
experimental solubility data, and the form of objective a marked reduction in the error is observed when the sec-
function used to minimize the deviations between calcu- ond BIP is included in the analysis. The following rec-
lated and experimental values of solubility. Our calcu- ommendation emerges from these conclusions: any of the
lated kt2 values for the PR-EOS show very good agree- three EOS along with any of the two mixing rules tested
ment with the values reported by Schmitt and Reid.35 can be used, as long as two binary interaction parameters
For example, they report k12 values of 0.121 and 0.0452 are included, one for the cohesive energy parameter, amix,
for the systems acridine/COz at 35 “C and naphtha- and one for the repulsive energy parameter, b,r,. From a
lene/C2H6 at 5.5 “C, respectively; our corresponding calcu- practical point of view, the simplest EOS and mixing rule
lated values for the same systems were 0.1206 and are to be used, that is, the Redlich-Kwong EOS and the
0.0449. More recently Bartle et al.‘O reported a k12 value classical van der Waals mixing rules.
of 0.090 for the system fluorene/C02 at 35 “C using a
simplified form of the PR-EOS for dilute solutions; our ACKNOWLEDGMENT
calculated value for the same system using the conven- Financial support by NSF under Grant CTS-
tional PR-EOS is 0.0864. 8902462 is gratefully acknowledged.
Table V summarizes the optimum k12 and hi2 val-
ues calculated for the PR-EOS,, along with the respective REFERENCES
errors. These results show that, as expected, the introduc- (1) Redlich, 0.; Kwong, J. N. S. Chem. Rev. 1949, 44,
233.
tion of a second binary parameter to the PR-EOS greatly (2) Soave, G. Chern. Eng. Sci. 1972, 27,. 1197.
improves the accuracy of the predicted solubility values (3) Peng, D. Y.; Robinson, D. B. I& Eng. Chern. Fun&m.
for most of the systems. No definite trends can be ob- 1976, 15, 59.
served in the behavior of the k12 and hr2 values with in- (4) Van der Waals, J. D. Ph.D. Thesis, Leiden, The
creasing temperature. For any particular system, how- Netherlands, 1873.
The Journal of Supercritical Fluids, Vol. 5, No. 4, 1992 Calculation of Interaction Parameters 295

(5) Schmitt, W. J.; Reid, R. C. J. Chem. Eng. Data 1986, (21) Schmidt, G.; Wenzel, H. Chem. Eng. Sci. 1980, 35,
31, 204. 1503.
(6) McHugh, M. A.; Krukonis, V. J. Supercritical Fluid (22) Milller, E. A.; Olivera-Fuentes, C.; Estevez, L. A. Lat.
Extraction. Principles and Practice; Butterworths: Am. Appl. Res. 1989, 19, 99.
Stoneham, MA, 1985; p 82. (23) Schmitt, W. J.; Reid, R. C. Presented at the AIChE
(7) McHugh, M. A.; Watkins, J. J.; Doyle, B. T.; Annual Meeting, San Francisco, Ca., 1984.
Krukonis, V. J. Ind. Eng. Chem. Res. 1988,27, 1025. (24) Kosal, E.; Holder, G. D. J. Chem. Eng. Data 1987,32,
(8) Dieters, U.; Schneider, G. M. Ber. Bunsenges. Phys. 148.
Chem. 1976,80, 1316. (23 Johnston, K. P.; Ziger, D. H.; Eckert, C. A. Ind. Eng.
(9) Lee, Y. Y.; Kim, H.; Lee, H.; Hong, W.-H. Korean J. Chem. Fundam. 1982,21, 191.
Chem. Eng. 1988,.5, 76. (26) Dean, J. A. Lange’s Handbook of Chemistry, 13th ed.;
(10) Bartle, K. D.; Clifford, A. A.; Shilstone, G. F. Proc. McGraw-Hill: New York, 1987.
2nd Symp. Supercritical Fluids; 1991, p 467. (27) Martinelli, E.; Gribaudo, L.; Campanella, E. An. Assoc.
(11) Rizvi, S. S. H.; Benado, A. L.; Zollweg, J. A.; Daniels, Q&m. Argent. 1987, 75, 491.
J. A. Food Technol. 1986,40, 55. (28) Kumik, R. T.; Holla, S. J.; Reid, R. C. J. Chem. Eng.
(12) Rao, V. S. G.; Mukhopadhyay, M. Proc. Int. Symp. Data 1981, 26, 47.
Supercritical Fluids; 1988, p 161. (29) Mitra, S.; Chen, J. W.; Viswanath, D. S. J. Chem. Eng.
(13) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Data 1988,33, 35.
Properties of Gases and Liquids, 4th ed.; McGraw-Hill: (30) Diepen, G. A. M.; Scheffer, C. A. J. Phys. Chem. 1953,
New York, 1987. 57, 575.
(14) Handbook of Chemistry and Physics, 71st ed.; Lide, D. (31) Johnston, K. P.; Eckert, C. A. AIChE J. 1981, 27,
R., Ed.; CRC Press: Boca Raton, 1990. 773.
(15) Dohm, R.; Brunner, G. Proc. 2nd Symp. Supercritical (32) Tsekhanskaya, Y. V.; Iomtev, M. B.; Mushkina, E. V.
Fluids, 1991, p 471. Russ. J. Phys. Chem. 1964,38, 1173.
(16) Kim, S.; Johnston, K. P. AIChE J. 1987b, 33, 1603. (33) Haselow, J. S.; Han, S. J.; Greenkom, R. A; Chao, K.
(17) Johnston, K. P.; Peck, D. G.; Kim, S. Ind. Eng. Chem. C. Equations of State for Supercritical Extraction;
Res. 1989, 28, 1115. ACS Symposium Series, Chapter. 7, 1986.
(18) Drame, N. R.; Viswanath, D. S. J. J. Martin (34) Bartle, K. D.; Clifford, A. A.; Shilstone, G. F. J.
Symposium, AIChE Spring National Meeting, Paper Supercrit. Fluids 1989,2, 30.
34c, New Orleans, 1988. (35) Schmitt, W. J.; Reid, R. C. In Supercritical Fluid
(19) Pongsiri, N.; Viswanath, D. S. Ind. Eng. Chem. Res. Technology; Penninger, J. M. L.; Radosz, M.;
1989, 28, 1918. McHugh, M. A.; Krukonis, V. J., Eds.; Elsevier
(20) Caballero, A. C.; EstCvez, L. A. Ind. Eng. Chem. Res. Science Publishers B. V.: Amsterdam, Netherlands,
1991, 30, 601; 1992, 31, 2256. 1985.