Sei sulla pagina 1di 22

ELSEVIER Fluid Phase Equilibria 99 (1994) 219-240

Prediction of liquid -liquid equilibria in ternary mixtures


from binary data
Jayaraman Suresh, Eric J. Beckman*
Department of Chemical and Petroleum Engineering, University of Pittsburgh, Pittsburgh, PA 15261, USA
Received 3 October 1993; accepted in final form 2 April 1994

Abstract

Liquid-liquid equilibria in type-1 and type-II ternary mixtures are predicted from binary data alone by two
methods: (1) Statistical Associating Fluid Theory (SAFT), which models molecular association based on the
first-order perturbation theory developed by Wertheim; (2) a hybrid model, in which a modified form of the
Panagiotopoulos-Reid mixing rule is applied to the physical contribution terms of SAFT. The proposed mixing rule
reduces to the original Panagiotopoulos-Reid mixing rule for binary mixtures, while simultaneously satisfying the
invariant condition when a component is divided into sub-components, and hence is consistent for application to
multicomponent systems. Although the Panagiotopoulos-Reid mixing rule provides accurate representation of phase
equilibria in type-II ternary systems, the prediction of ternary phase behavior in type-1 systems is not satisfactory. The
association model, however, predicts phase behavior in both type-1 and type-II ternary mixtures quite accurately. The
significance of inclusion of hydrogen bond acceptor groups on aromatic rings, due to the presence of z-electrons, is
elucidated in the representation of phase equilibria in binary and ternary mixtures.

Keyworris: Experiments; Data; Association; Liquid-liquid equilibria; Mixing rules; SAFT model

1. Introduction

The availability of a model which accurately represents liquid-liquid equilibria would be


useful in the design of various separation processes (Treybal, 1963) such as extraction and
leaching. While substantial progress has been made over the last decade in the representation of
vapor-liquid equilibria (VLE), the representation of liquid-liquid equilibria (LLE) in associat-
ing mixtures still remains a major challenge. More often, the representation/prediction of LLE
requires empirical modification of thermodynamic models such as the incorporation of temper-
ature-dependent interaction parameters (Leung and Dadakshan, 1987; Narodoslawsky et al.,

* Corresponding author.

037%3812/94/$07.00 0 1994 - Elsevier Science B.V. All rights reserved


SSDZ 0378-3812(93)02520-B
220 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

1987), inclusion of ternary interaction parameters (Nagata, 1989, 1990) or, in equation of state
type calculations, the introduction of asymmetric mixing rules (Huron and Vidal, 1979;
Mathias and Copeman, 1983; Luedecke and Prausnitz, 1985; Panagiotopoulos and Reid,
1986). Prausnitz and De Pablo Juan (1989, 1990) have pointed out that the poor representa-
tion of many thermodynamic models in representing LLE is, in part, due to the fact that
classical models do not perform well near the critical point. Further, he has shown that
transformation of the coordinates of a classical equation of state to non-classical coordinates
improves the representation of LLE of binary and ternary mixtures, including the critical
region.
Alternatively, there have been recent attempts (Nagata et al., 1987; Anderko and Mal-
onowski, 1989; Nagata and Fuzukawa, 1990; Panayiotou and Sanchez, 1991; Economou and
Donohue, 1992; Suresh and Elliott, 1992; Suresh et al., 1994) at representing LLE in mixtures
using models that explicitly account for molecular association through theories such as the
chemical theory (Heidemann and Prausnitz, 1976), the statistical mechanical theory
(Wertheim, 1984a, b, 1986a-c) and the bond counting approach (Levine and Perram, 1968;
Veytsman, 1990). The association parameters in such models can be easily related to molecular
properties, and hence can be obtained from spectroscopic analysis. Although studies on such
models (Anderko, 1990; Economou and Donohue, 1992; Suresh and Elliott, 1992) indicate
that they are capable of representing VLE/LLE in binary mixtures reasonably well, few studies
have been performed for predicting LLE behavior in ternary systems that include a wide
variety of compounds from binary data alone. Such a study would be a stringent test for phase
equilibrium models since because of the following.
(1) LLE are, in general, more sensitive to the molecular details built into the phase
equilibrium models than VLE. As an example, VLE in associating mixtures has been success-
fully correlated by phase equilibrium models that do not explicitly account for molecular
association (Panagiotopoulos-Reid, 1986).
(2) Prediction of phase equilibrium data in ternary mixtures from binary data alone is, in
general, more difficult than representation of binary data due to the lack of any adjustable
parameters.
(3) The simultaneous prediction of a wide variety of ternary mixtures, that include
molecules possessing various number of association sites, is quite challenging.
Hence, an evaluation of the effect of explicit incorporation of the association phenomena for
predicting LLE in ternary mixtures from binary data alone would elucidate the significance of
incorporating realistic molecular details in phase equilibrium models.
In this paper, we evaluate two equation of state models for predicting ternary LLE in
associating mixtures from binary data alone. For both models, the repulsive, dispersion and
covalent-bond chain terms have been taken to be identical and have been obtained from the
SAFT equation of state (Huang and Radosz, 1990, 1991). The difference between the two
models lies in the method used to treat association effects, which are as follows.
(1) A method that explicitly accounts for association through Wertheim’s theory (1984a, b,
1986a-c), which is referred to as the association model (which is simply the SAFT equation of
state as developed by Huang and Radosz (1990)) in the rest of this paper. For details on
Wertheim’s association theory, the readers are referred to Chapman et al. (1990), Huang and
Radosz ( 1990) and Suresh et al. ( 1994).
J. Suresh, E.J. Beckman /Fluid Phase Equilibria 99 (1994) 219-240 221

(2) A method that does not explicitly account for association, but instead treats such
non-ideal mixtures through a newly proposed asymmetric mixing rule for the dispersion energy
term. The mixing rule proposed in this paper is a modified form of the Panagiotopoulos and
Reid (1986) mixing rule, which has been shown to be simple and yet powerful for treating
binary, non-ideal mixtures. Our modification takes into account an inconsistency in the
Panagiotopoulos and Reid ( 1986) mixing rule. This inconsistency (Michelsen and Kistenmacher,
1990) arises from the fact that the Panagiotopoulos and Reid ( 1986) mixing rule for multicom-
ponent mixtures is not invariant to the division of a component into sub-components. Hence, in
this paper, we provide a new mixing rule that reduces to the Panagiotopoulos and Reid (1986)
mixing rule for binary mixtures but is consistent for extension to multicomponent mixtures, i.e.
the case for a ternary where two constituents are identical reduces to the appropriate binary.
This model is referred to as the P-R model in the rest of this paper.
In summary, we study the effect of explicit incorporation of association effects in equation of
state methods for predicting ternary LLE from binary data alone. A new mixing rule is proposed
that reduces to the familiar Panagiotopoulos and Reid (1986) mixing rule for binary mixtures
but is consistent for extension to multicomponent mixtures. Ternary mixtures that include
various types of hydrogen bonding components, such as acetic acid which dimerizes in the pure
state, water and ethanol which can form higher order oligomers, and non-associating diluents,
have been selected in this study.

2. Theory

Statistical Associating Fluid Theory (SAFT) accounts for four types of molecular interac-
tions: ( 1) repulsive; (2) dispersion; (3) chain effects; (4) molecular association. Since we are
primarily interested in correlating the liquid-liquid equilibrium properties of mixtures that
include hydrogen bonding compounds, we have focussed on the two approaches we have
adopted for the treatment of the hydrogen bond contribution:
(1) molecular association;
(2) modification of the P-R mixing rule.

3. Molecular association

The key expression in Wertheim’s association theory is that which defines the association
strength d :
A = K[exp( -H/U) - l]pg(“) (1)
where g@‘)is the pair correlation function at contact = (ZreP - 1)/4~, q = reduced density = pi*,
p = bulk fluid density, U* = volume of a molecule, H = the enthalpy change on hydrogen bond
formation, k = Boltzmann factor, T = temperature, K = bonding volume. An application of Eq.
(1) to associating mixtures would required knowledge of:
( 1) the number of proton donor and proton acceptor groups in the various types of molecules
used in this study;
222 J. Suresh, E.J. t’vckman / Fluid Phase Equilibria 99 (1994) 219-240

(2) the enthalpies (H) and the bonding volumes (K) of the hydrogen bonds formed between
like species and unlike species.
We shall consider the above two aspects sequentially.

3.1. SpeciJication of the number of association sites in each molecule

In this study, alcohols and water are each treated as molecules with two sites (one proton
donor and one proton acceptor site). Such a description for alcohols has been found to be
adequate for the representation of phase equilibrium properties (Ikonomou and Donohue, 1988;
Suresh and Elliott, 1991), while being computationally efficient. As for water, the specification
of the number of sites is based on three considerations.
(1) Although water possesses two proton donor and two proton acceptor sites (a four-site
model), the effective number of sites in water is still not clear. For example, while a four-site
model has been proposed by Ghonasgi and Chapman ( 1993) based on simulation results, a
three-site model has been proposed by Wei et al. (1991) from an analysis of spectroscopic
results, and Buckingham and Fowler (1988) from quantum mechanical calculations.
(2) A two-site model (one positive and one negative site) has been found to be as accurate as
the three-site model in the representation of phase behavior in water-hydrocarbon systems,
while the four-site representation is not found to be accurate (Economou and Donohue, 1992;
Suresh and Elliott, 1992).
(3) A two-site model is computationally more efficient than the three-site or the four-site
models. Hence, the water molecule was assigned two sites in this study. Regarding carboxylic
acids, each molecule was modeled with one association site. Hence acetic acid can dimerize but
cannot form higher order oligomers in its pure state, as observed experimentally (Prausnitz,
1969).
Although it has been well established from spectroscopic studies that water, carboxylic acids
and alcohol molecules form hydrogen bonded entities in pure fluids and in mixtures (Joesten
and Schaad, 1968), few structural studies have been carried out in aromatic acceptors. In most
systems, small, unsaturated hydrocarbons serve as H acceptors in complexes with Lewis acids
(Cheney et al., 1988). Although such bonds are weak, they have been extensively studied by both
experimental (Baiocchi et al., 1983; Hartland et al., 1992) and theoretical (Pople et al., 1982;
Frisch et al., 1983) means. Of particular significance are the H-bonds formed between the
7c-electrons of aromatics and various proton donors. Weak 1: 1 complexes between benzene and
hydrogen halides have been characterized in the gas phase by means of fourier transform
microwave molecular beam techniques (Reid et al., 1981). Further, ab initio studies of
unconventional rc hydrogen bonded dimers between benzene and various proton acceptors
(water, ammonia, hydrogen fluoride, hydrogen sulfide and hydrogen chloride) show weak
association and a flat potential surface where z hydrogen bonding takes place (Cheney et al.,
1988). More recently, Tang et al. (1990) studied the n-type hydrogen bonded systems consisting
of benzene as the proton acceptor and hydrogen fluoride as proton donor and found that the
stabilization energy is around 4.81 kcal mol- ‘, Apart from quantum ab initio studies, the
significance of n-electrons in benzene has also been elucidated in applications of LLE in aqueous
systems by Michel et al. (1989). In spite of the above observations, to our knowledge, no phase
equilibrium model has yet explicitly accounted for the z hydrogen bonds formed by aromatic
J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 223

rings. In this paper, we propose that the increased mutual solubility of water and aromatic
compounds, as compared with water and alkanes or water and naphthenes, is a result of
formation of cross-associated dimers between water and aromatic compounds. We further show
that the incorporation of proton acceptor groups on aromatic compounds improves predictions
of ternary mixtures that include aromatic compounds as one of its constituents.

3.2. Determination of the enthalpies (H) and the bonding volumes (K) of the hydrogen bonds

The association parameters (H and K) for pure components are obtained by optimizing the
errors in the vapor pressures in liquid densities for a range of temperatures (Huang and Radosz,
1990). Unfortunately, the estimation of these hydrogen bond parameters between unlike species
in a mixture is not straightforward as they have to be treated as adjustable parameters. To
minimize the number of adjustable parameters in this study, these bond parameters have been
related to the self-association constants of the two components as follows:
H,j = (H,H,) 1/Z (2)
Kij = (Kii Ka) ’ ‘*Uij
(3)
where i, j are the component indices and aij is a parameter that adjusts the extent of
cross-association between the two species. An aii value of unity would indicate that the
cross-association bonding volume is simply the geometric mean of the self-association bonding
volumes of the two components. For mixtures in which component ‘i’ self-associates and
cross-associates with component ‘j’, while component ‘j’ does not self-associate, an application
of Eq. (2) and (3) would yield a value of zero for the cross-associating enthalpy and bonding
volume, and hence the following alternate relationships have been applied:
Hij = Hti (4)
Kil = K,iaij (5)
In summary, there are two adjustable parameters for SAFT:
(1) a conventional binary parameter (k,) that accounts for two-body interactions, and is
applicable to all binary mixtures;
(2) a cross-section parameter (a,) that adjusts the degree of cross-association between two
species, and is applicable to binary mixtures in which the first component can hydrogen bond
with the second.
As examples, k, and aij are both adjustable for the water + ethanol and water + benzene
systems, whereas k, is the only adjustable parameter employed during modeling of the
water + hexane and ethanol + heptane systems.

4. Modification of the P-R mixing rule

We now discuss modification of the P-R mixing rule for extension to multicomponent
systems. In a mixture, molecular interactions occur between both like and unlike molecules. The
intermolecular energy in a mixture proposed by Panagiotopoulos and Reid (1986) is given by:
Em =&$t)+$ (6)
224 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

where l$,) and E:) are given as follows:

E$) = cc (EVXiXj) (7)

(8)

Efj = (Eil&#)1’2(1 - klj) (9)


where k, ( = k,J is the standard, symmetric binary parameter, and dij is an additional parameter
introduced with di, = dJ = 0 and dii = -&. The summations are carried out from 1 to the
number of components. For binary mixtures, Eq. (8) becomes

~2) = x1x2(d12x1 + d,,x,) (10)

Eq. (10) can alternatively be obtained if we consider the following expression:

(11)

with
Ciii = 0 (12)
C, = - c, (13)
where C,, are temperature-dependent material parameters.
For a binary mixture, Eq. (11) becomes

Ek? = 3x1XZ(G 12x1+ G22x2) (14)


which is identical to Eq. (10) if we equate the following parameters:

& = 3G2
d21 = 3C221

For a ternary mixture, Eq. (11) reduces as follows:

&) = 3Xi%(Ci& + c221x2) + 3xi-%(C113% + CM,%) + 3%x,(C,,,% + (c332x3)

+ 6(7123x1 ~2x3 (15)


If components ‘2’ and ‘3’ are identical, then

cl23 = c22, (16)


C332 = C223 = 0 (17)
and hence Eq. ( 15) is transformed to

&) = 3&,x1(% +x3)2 + 3c112X:(x2 + x3) (18)


J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 225

Recognizing that the case where components ‘2’ and ‘3’ are identical is equivalent to a situation
where component ‘2*’ has been divided into sub-components ‘2’ and ‘3’:
xz*=x*+x 3 (19)
Eq. ( 18) then reduces to
E$) = 3X1Xz*(X,
Cl12 + C*,,xf) (20)
which is identical to Eq. (14). The most important step in the above derivation is given by Eq.
(16), which has often been neglected in the formulation of new mixing rules. For the P-R
mixing rule, C 123is always zero and, hence, the mixing rule for ternary mixtures does not reduce
to that of a binary mixture when two components of the ternary system are identical. For Eq.
( 16) to be satisfied when two or more components in a mixture become identical, we propose a
simple equation for Ccjk:
Cuk= [C~C&-~(C~J
+ CiJ&)/2+ CjirCjkk(cji+ Cjf&)/2+ Ck,CQ/(Ck,i+ CQj)/2]1’3 (21)
It must be noted that Cij~ is invariant to interchanges of component incides. Further, C, can be
positive or negative for any i, j combination. The mixing rule for a binary mixture reduces to the
classical quadratic mixing rule if Cid is equal to zero.
In this paper, we apply this modified P-R mixing rule to the physical contributions of the
SAFT equation of state. In other words, we exclude Wertheim’s explicit association term from
the SAFT equation of state, and instead attempt to model phase equilibria in associating
mixtures by an asymmetric mixing rule for the molecular interaction term, given as follows:

(22)

(23)

where ri is the number of segments in molecule ‘i’, dgk = (dii + dii + dkk)/3, dij = (dii + d,)/2, and
cii = depth of the dispersion square well energy between components ‘i’ and ‘j’.
The adjustable parameters used in this model for binary mixtures are kij and C,. Both
parameters can be temperature dependent, depending on the binary system under study.

5. Results and discussions

The pure component parameters for use with the association model given by Huang and
Radosz (1990) provide excellent vapor pressure correlations and VLE of multicomponent
mixtures. Unfortunately, they do not accurately correlate LLE. This feature has been noted
previously for other models as well (Magnussen et al., 1981; Cha and Prausnitz, 1985); LLE are
more sensitive than VLE to the value of pure component and binary parameters. For example,
while there are several sets of hydrogen bond energy and bonding volume SAFT parameters
which can provide simultaneous representations of VLE and liquid densities of methanol, only
one set of parameters can simultaneously correlate VLE, liquid densities and LLE of mixtures
226 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

Table 1
Revised pure component parameters of some associating species

v* elk r k: AH/k elk No. of sites

Water 9.637 235.84 1.425 0.12000 -3210.0 1.0 2


MeOH 20.095 181.77 1.831 0.04856 - 3000.0 10.0 2
EtOH 26.642 149.72 3.424 0.07350 - 3000.0 10.0 2
Acetic acid a 14.500 290.73 2.132 0.03926 -3941.0 10.0 1

v* = volume of a molecule = voor, voo = segment volume, r = number of segments, E = dispersion energy,
k,* = Bonding volume, AH = enthalpy of hydrogen bonding, e/k = constant, k = Boltzmann’s constant.
a Parameters are identical to those proposed by Huang and Radosz (1990).

Table 2
Comparisons of percentage average absolute deviations of pure component vapor pressures and liquid densities for
the association model and the P-R model over a temperature range of T/T, = 0.5-0.9

Error in vapor pressures Error in liquid densities

Association model P-R model Association model P-R model

Water 3.3 1.8 4.4 6.2


MeOH 3.4 3.6 1.9 3.4
EtOH 1.6 5.4 1.0 4.2
Acetic 2.8 3.1 0.8 5.4
acid

that include methanol as one of its components. Hence, we have found it necessary to determine
the pure component parameters of associating species by correlating LLE of binary mixtures, in
which the second component is preferably non-associating to avoid additional complexities
arising from cross-association networks. The solubility data of methanol + cyclohexane, wa-
ter + pentane, ethanol + hexadecane and acetic acid + nonane systems at 298.15 K have been
used to determine the pure component parameters of methanol, water, ethanol and acetic acid
respectively. Table 1 lists the revised pure component parameters for some associating compo-
nents. Table 2 shows the percentage average absolute deviations in vapor pressures and liquid
densities for the various associating species. Fig. 1 shows that the incorporation of solubility
data in developing the pure component parameters improves the representation of LLE in the
methanol + heptane system. In particular, the liquid-liquid critical point is accurately posi-
tioned by the model, which is an important aspect since an artificial phase separation between
two components at a particular temperature caused by inaccuracies in the liquid-liquid critical
point can seriously deteriorate the quality of predictions of ternary tie-lines in type-1 systems
(Luedecke and Prausnitz, 1985). Not surprisingly, the binary interaction parameter (AZ,),(Table
3) is close to zero.
We now proceed to water + non-aromatic systems. The k, values for alkanes and naphthenes
have all been found to be 0.33, and it is independent of temperature. Fig. 2 shows a solubility
J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 227

.I
*d ‘\
i
330. ,I i
.' i
/' i
i

‘\
P
.i a 0
i
i
F ,i'
i
.i i
i
,;' i
310. i
./ i
/ i
i
i
i
i
i
/f \

0.0 0.5 1.0


X(MeOH)

Fig. 1. Solubility plot of the methanol + heptane system. (0) Experimental data (Sorensen and Arlt (1979). (-)
SAFT correlations with revised methanol parameters. (- . - -) SAFT correlations with methanol parameters taken
from Huang and Radosz (1990).

Table 3
Binary parameters of the association model for various mixtures

References aij k,

Alkane/naphthene + water Brady et al. (1982) 0.000 0.330


Benzene + water Tsonopoulos and Wilson (1983) 0.008 0.232
Xylene + water Andersen and Prausnitz (1978) 0.008 0.243
Toluene + water Brady et al. (1982) 0.008 0.240
Ethylbenzene + water Brady et al. (1982) 0.008 0.270
Ethanol + water Phutela et al. (1979) 0.744 0.000
Ethanol + cyclohexane Nagai and Isii ( 1935) 0.000 0.080
Ethanol + benzene Ellis and Clark (1961) 0.050 0.055
Heptane + benzene Messov et al. (1977) 0.000 0.020
Methanol + benzene Lee (1931) 0.013 0.025
Methanol + heptane Sorensen and Arlt (1979) 0.000 0.045
Methanol + cyclohexane Sorensen and Arlt (1979) 0.000 0.053
Methanol + ethanol Kooner and Fenby (1980) 1.ooo 0.035
Acetic acid + benzene Haughton (1967) 0.000 0.040
Acetic acid + water Acharya and Rao (1947) 0.550 0.030

plot of various hydrocarbons in the water-rich phase, and Fig. 3 shows a plot of the composition
of water in the hydrocarbon-rich phase for propane and butane at 3000 Psia. While SAFT
provides accurate results at temperatures above 328 K, it does not predict a minimum in
hydrocarbon solubility and hence deviates from experimental data at temperatures below 328 K.
This problem has been noted for other models as well, such as APACT (Economou and
228 J. Suresh, E.J. Beckman / Fluid Phase Equilibria 99 (1994) 219-240

io 300 400 500


VW
Fig. 2. Hydrocarbon solubility in water-rich phase at three-phase pressures.(*, - . - -) Experimental and SAFT
correlations for cyclohexane, (0, -) Experimental and SAFT correlations for hexane. ( q, ---) Experimental and
SAFT correlations for n-octane. The experimental data were taken from Tsonopoulos and Wilson (1983).

-0.E
IO

-3.0
‘0 300 400 500
T(K)

Fig. 3. Water solubility in hydrocarbon-rich phase (LLE) at 3000 Psia (*, - . - -) Experimental (Reamer et al.,
1952) and SAFT correlations for n-butane. (0, -) Experimental (Kobayashi and Katz, 1953) and SAFT correlations
for n-propane.

Donohue, 1992). Improvements to the representation of the minimum in hydrocarbon solubili-


ties will probably require the consideration of long-range electrostatic effects (Walsh et al.,
1992), a modification which is not considered in this study.
Fig. 4 shows a plot of the water + benzene system for two cases: (1) proton acceptor in
benzene forms cross-associated dimers with the proton donor in water; (2) benzene does not
J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 229

-1.
IO

a
I" -2.
ZIO

-3.
IO

in
-41 r

‘” 300 400 500


T(K)

Fig. 4. Three-phase equilibria of the water + benzene system. (0) Experimental data (Tsonopoulos and Wilson, 1983).
(-) SAFT correlations assuming benzene can form cross-associated dimers with water. (- - -) SAFT
correlations assuming benzene does not participate in hydrogen bonding. Top curve represents water in the
benzene-rich phase. Bottom curve represents benzene in the water-rich phase.

cross-associate with water. If benzene is assumed to be a non-associating compound, the


solubility of water in the benzene-rich phase is under-estimated, especially at lower temperatures,
owing to the exothermic nature of hydrogen bond interactions. However, the correlation of
water’s solubility in benzene is substantially improved when benzene is assumed to form dimers
with water. Similar results are obtained for other aromatic components as well. The interaction
parameters for the various mixtures are given in Table 3. It should be noted that the aU
parameter is set to the same value for all water + aromatic hydrocarbon mixtures. In other
words, the extent of cross-association is identical for all water + aromatic hydrocarbon mixtures,
and is not adjustable for these systems.
The pure component parameters for the P-R model are obtained from the SAFT equation of
state by setting the enthalpy and bonding volume of the hydrogen bonds to zero. Table 4 lists

Table 4
Pure component parameters of water, ethanol and methanol for the P-R model

V* r elk
Water 9.505 272.36 4.694 1.0
MeOH 19.043 191.74 6.173 10.0
EtOH 26.011 166.18 8.396 10.0
Acetic 7.1020 269.72 4.235 10.0
acid
230 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

Table 5
Binary parameters of the P-R model for various mixtures

Species (1) + species (2) T (K) E 122 k 12

Water + benzene 337.15 -0.1368 0.452


Ethanol + water 337.15 0.000 - -0.080
Ethanol + cyclohexane 298.15 -0.055 0.310
Ethanol + benzene 337.15 - 0.030 0.270
Heptane + benzene 305.95 0.000 0.020
Methanol + benzene 305.95 - 0.033 0.260
Methanol + heptane 305.95 - 0.048 0.253
Methanol + cyclohexane 298.15 -0.065 0.300
Methanol + ethanol 298.15 0.000 0.000
Acetic acid + water 353.15 0.000 0.125
Acetic acid + benzene 353.15 -0.040 0.030

the pure component parameters of water, methanol, ethanol, and acetic acid. Table 2 lists the
error in vapor pressures and liquid densities for the various pure components. The pure
component properties of components that do not self-associate are identical to the SAFT
parameters proposed by Huang and Radosz (1990).
Mathias et al. (1991) applied the P-R mixing rule to a modified version of the Peng-
Robinson equation of state (Mathias and Copeman, 1983) and found that the deviations in the
distribution coefficients (the ratio of mole fractions in the two phases) were less than 10% for
several complex binary mixtures. Similar results were observed in our application of the P-R
mixing rule to the physical contribution of the SAFT equation of state, and hence we have not
provided the details of these calculations. However, it is useful to note that two temperature-
dependent parameters are necessary to describe the phase behavior over a wide range of
temperatures, as compared with a single, temperature-independent parameter required by the
association model. The interaction parameters for various binary mixtures are given in Table 5.
We now proceed to analyze the ternary mixtures. Five mixtures have been chosen in this
study.
(1) A non-aqueous type-II mixture that does not include benzene: hexane + heptane +
methanol. This mixture includes self-associated hydrogen bonds, but no cross-associations.
(2) A non-aqueous type-1 mixture that does not include benzene: ethanol + methanol +
cyclohexane. This mixture includes self-associating as well as cross-associating components, and
a diluent.
(3) A non-aqueous type-1 mixture that includes benzene: heptane + benzene + methanol. This
mixture includes a component that can both self-associate and cross-associate, a component that
cannot self-associate but can cross-associate, and a diluent that can neither self-associate nor
cross-associate.
(4) An aqueous type-1 mixture that includes benzene: water + ethanol + benzene, which is
similar to mixture (3) but does not include a diluent.
J, Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 231

Table 6
Percentage average absolute deviations of VLE pressures for the association and P-R models

Species (1) + species (2) Association P-R

Ethanol + water 2.6 1.8


Ethanol + cyclohexane 0.9 4.6
Ethanol + benzene 3.0 4.0
Heptane + benzene 2.7 2.7
Methanol + benzene 1.8 1.8
Methanol + ethanol 1.9 5.9
Acetic acid + water 3.8 3.4
Acetic acid + benzene 4.0 1.9

(5) A type-1 mixture that includes water, benzene and a molecule which forms only dimers in
its pure state, acetic acid.
It should be noted that the ternary analyses are predictions, i.e. all interaction parameters
were obtained from binary analysis alone. For partially miscible or immiscible systems, the
interaction parameters are obtained by matching the phase compositions. For miscible systems,
a VLE analysis is performed and compared with the experimental data (Table 6). The
interaction parameters for the various binary systems are given in Tables 3 and 5 for the
association and the P-R models, respectively. For the P-R model, the interaction parameters
were obtained at the temperature at which the ternary analysis is performed, since they are
strong functions of temperature for highly non-ideal binary mixtures (Mathias et al., 1991).
Figs. 5 and 6 show plots of the binodal curve and the distribution ratio of hexane between the
two liquid phases for a non-aqueous type-II system: heptane -I-hexane + methanol at 305.95 K.
Where the association model accurately predicts the phase behavior of this system, the P-R

-0 0.2 0.4 0.6 0.8 1

X(3)
Fig. 5. Ternary phase diagram for the heptane( 1) + hexane(2) + methanol(3) system at 305.95 K. ( n) Experimental
tie-lines (Wittrig, 1979). (-) SAFT. (. . .) P-R model.
232 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

DISTR. RATIO
0.4 (

_._
0 0.2 0.6 0.8

Fig. 6. Distribution ratio of hexane between the heptane-rich and methanol-rich phases at 305.95 K. X(2) is the mole
fraction of hexane in the methanol-rich phase. (0) Experimental data (Wittrig, 1979). ( -) SAFT. (...) P-R
model.

n
"0 0.2 0.4 0.6 0.8 1
X(3)
Fig. 7. Ternary phase diagram for the methanol( 1) + ethanol(2) + cyclohexane(3) system at 298.15 K. ( W) Experi-
mental data (Nagata and Ohta, 1983). ( -) SAFT. (. . .) P-R model.

model accurately predicts the binodal curve (Fig. 5) although the representation of the
distribution ratio of hexane between the two phase (Fig. 6) is not accurate. A similar analysis
of a non-aqueous system but which is of type I, ethanol + methanol + cyclohexane system, shows
that while the binodal curve (Fig. 7) and the distribution ratio of methanol between the two
phases (Fig. 8) are accurately predicted by the association model, the P-R model is grossly in
error relative to its performance for the type-II system. These results are in compliance with
previous observations that type-II systems are relatively easier to model than type-1 systems (Cha
and Prausnitz, 1985) and, hence, do not constitute a stringent test for phase equilibrium models.
It should be noted that the P-R model is checked to ensure that it does not falsely indicate a
liquid-liquid phase split for the ethanol + cyclohexane system, and hence any deficiency in
predicting the ternary system cannot be attributed to false binary phase separations.
J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 233

------_____
v
‘, -----------my________
‘\
4 \
a
e
d
\ %.D
zE %..,
%.
2
-......
0

-.. 0

~‘~~.._._,_o
0

Fig. 8. Distribution ratio of ethanol between the methanol-rich and cyclohexane-rich phases at 298.15 K. X(2) is the
mole fraction of ethanol and the cyclohexane-rich phase. (0) Experimental data (Nagata and Ohta, 1983). (- - -)
SAFT. (- . - . -) P-R model.

0 0.2 0.4 0.6 0.8 1

X(3)
Fig. 9. Ternary phase diagram for the heptane( 1) + benzene( 2) + methanol( 3) system at 297.95 K. ( n) Experimental
data (Wittrig, 1979). (-) SAFT assuming benzene can form cross-associated dimers with water. (. . .) SAFT
assuming benzene does not participate in hydrogen bonding.

We now proceed to systems that include an aromatic component. Fig. 9 shows a plot of the
phase diagram of the heptane + benzene + methanol system at 279.95 IL The prediction of
ternary phase behavior of this system is substantially improved if benzene is assumed to form
cross-associated dimers with methanol. If benzene is not assumed to form dimers with methanol,
a higher distribution ratio for benzene between the heptane-rich phase and the methanol-rich
phase results, as shown in Fig. 10, due to lower solubility of benzene in the methanol-rich phase.
Fig. 11 shows a plot of the same system but at a temperature of 305.95 K. In this plot, the
association model is compared with the P-R model. Although both models are capable of
representing binary phase behavior with equal accuracy, the association model is superior to the
P-R model for ternary predictions. Again, the distribution ratio of benzene between the
234 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

00 0.1 0.2 0.3


X(2)

Fig. 10. Distribution ratio of benzene between the heptane-rich and methanol-rich phases at 279.95 K. X(2) is the
mole fraction of benzene in the methanol-rich phase. (0) Experimental data (Wittrig, 1979). (- - -) SAFT assuming
benzene can form cross-associated dimers with water. (- . - -) SAFT assuming benzene does not participate in
hydrogen bonding.

0.1 -

0.05 -

0
0 0.2 0.4 0.6 0.8 1

X(3)
Fig. Il. Ternary phase diagram for the heptane( 1) + benzene(2) + methanol(3) system at 305.95 K. ( n) Experimen-
tal data (Wittrig, 1979). ( -) SAFT. (. .) P-R model.

heptane-rich and the methanol-rich phases is grossly overestimated by the P-R model (Fig. 12).
Fig. 13 shows a plot of ternary predictions for the system water + acetic acid + benzene at
353 K. We have four types of hydrogen bonds in this system: (1) self-association of water to
form oligomers; (2) dimerization of acetic acid; (3) cross-association between acetic acid and
water; (4) cross-associated dimerization between benzene and a proton donor. Fig. 14
shows that the distribution ratio of the acetic acid between the two liquid phases predicted
by the P-R model is grossly in error while the association model is accurate. Similar results
are obtained for a more commonly encountered system: ethanol + water + benzene system at
337.15 K (Figs. 15 and 16).
J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240 235

4J

--__ ---__
--__ --__
--__ --.

3.

g
$ L._.:
.2 b.,&
F .-A_.
cr,
0 -*-._*
-0..
%._.
--4*

1.

O\
0.00 0.05 0.10 0.15
X(2)

Fig. 12. Distribution ratio of benzene between the heptane-rich and methanol-rich phases at 305.95 K. X(2) is
the mole fraction of benzene in the methanol-rich phase. (0) Experimental data (Wittrig, 1979). (- - -) SAFT.
(- . - -) P-R model.

0.25

0
0 0.2 0.4 0.6 0.8 I
X(3)
Fig. 13. Ternary phase diagram for the water( 1) + acetic acid(2) + benene( 3) system at 353.15 K. ( W) Experimental
data (Tagliavini et al., 1955). (-) SAFT. (. .) P-R model.

6. Conclusions

We have demonstrated in this paper that the association model can predict ternary phase
equilibria for mixtures that include water, hydrocarbons, alcohols and carboxylic acids from
binary data alone. A revised set of pure component parameters have been proposed for alcohols
and water to provide accurate simultaneous representation of vapor pressures, liquid densities,
VLE and LLE in multicomponent mixtures. The significance of the proposed z-type hydrogen
236 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

0.00 0.25 0.50


X(2)
Fig. 14. Distribution ratio of acetic acid between the water-rich and benzene-rich phases at 353.15 K. X(2) is the mole
fraction of acetic acid in the benzene-rich phase. (0) Experimental data (Tagliavini et al., 1955). (- - -) SAFT. (-
- . -) P-R model.

X(2)
0.5 -

0.25

0
0 0.2 0.4 0.6 0.8 1
X(3)
Fig. 15. Ternary phase diagram for the water( 1) + ethanol(2) + benzene(3) system at 337.15 K. ( n) Experimental
data (Morachevskii and Belousov, 1958). ( -) SAFT. (- . - . -) P-R model.

bonding between benzene and other proton donor molecules has been elucidated in the
representation of phase behavior in water + aromatic hydrocarbon mixtures. A universal binary
interaction parameter is applicable for all water + alkanes and water + naphthenes and is valid
for temperatures above 328 K.
The phase behavior in ternary systems has been predicted by two models: (1) SAFT, which
explictly recognizes association; and (2) the P-R mixing rule applied in conjunction with the
physical contributions of SAFT (excluding explicit association effects). Ternary mixtures that
include various types of hydrogen bonding components, such as acetic acid which dimerizes in
J. Suresh, E.J. Beckman / Fluid Phase Equilibria 99 (1994) 219-240 237

0.0 0.1 0.2 0.3 0.4


X(2)

Fig. 16. Distribution ratio of ethanol between the water-rich and benzene-rich phases at 337.15 K. X(2) is the mole
fraction of ethanol in the benzene-rich phase. (0) Experimental data (Morachevskii and Belousov, 1958). (- - -)
SAFT. (- . - -) P-R model.

the pure state, benzene which possesses only a proton acceptor group, water and ethanol, which
can form higher order oligomers, and non-associating diluents, have been selected in this study.
Before performing this analysis, the P-R mixing rule was modified to satisfy the invariant
condition when a component is divided into sub-components. The formulation is general and
can be applied to any equation of state for extending the P-R mixing rule to multicomponent
systems. Although the P-R mixing rule provides excellent representation of phase behavior in
complex binary mixtures and accurately predicts phase equilibrium in type-II ternary systems,
the prediction of ternary phase behavior in type-1 systems is not satisfactory. Furthermore, the
P-R model has two temperature-dependent binary interaction parameters as compared with one
temperature-independent parameter for the association model. The association model predicts
phase behavior in both type-1 and type-II ternary mixtures quite accurately. The incorporation
of hydrogen bond acceptor groups on aromatic rings (the z-electrons) substantially improves
predictions of the phase behavior in ternary systems that include at least one aromatic
component.

References

Acharya, M.V.R. and Rao, M.N., 1947. Vapor-liquid equilibrium of non-ideal solutions. Trans. Indian Inst. Chem.
Eng. 29: 1.
Anderko, A. and Malonowski, S., 1989. Calculation of solid-liquid, liquid-liquid, and vapor-liquid equilibria by
means of an equation of state incorporating association. Fluid Phase Equilibria, 48: 223.
Andersen, T.F. and Prausnitz, J.M., 1978. Application of the UNIQUAC equation to calculation of multicomponent
phase equilibria. 2. Liquid-liquid equilibria. Ind. Eng. Chem., Process Des. Dev., 17: 561.
238 J. Suresh, E.J. Beckman 1 Fluid Phase Equilibria 99 (1994) 219-240

Baiocchi, F.A., Williams, J.H. and Klemperer, W.J., 1983. Molecular beam studies of hexafluorobenzene, trifl-
uorobenzene, and benzene complexes with hydrogen fluoride. J. Chem. Phys., 87: 2079.
Brady, J.C., Cunningham, J.R., Wilson, G.M., 1982. Water Hydrocarbon Liquid-Liquid-Vapor Equilibrium
Measurements at 530°F. Gas Petroleum Association Research Report-62, Houston, TX.
Buckingham, A.D. and Fowler, P.W., 1988. Electrostatic model for hydrogen-bonded dimers: donor-acceptor scale
for hydrogen halides and pseudohalides. J. Mol. Struct., 189: 203.
Cha, T.H. and Prausnitz, J.M., 1985. Thermodynamic method for simultaneous representation of ternary vapor-liq-
uid and liquid-liquid equilibria. Ind. Eng. Chem., Process Des. Dev., 24: 551.
Chapman, W.G., Gubbins, K.E., Jackson, G. and Radoz, M., 1990. New reference equation of state for associating
liquids. Ind. Eng. Chem. Res., 29: 1709.
Cheney, B.V., Schultz, M. W. and Richards, W.G., 1988. Hydrogen-bonded complexes involving benzene as an H
acceptor. J. Am. Chem. Sot., 110: 4195.
Economou, I. G. and Donohue, M. D., 1992. Equation of state with multiple associating sites for water and
water-hydrocarbon mixtures. Ind. Eng. Chem. Res., 31: 2388.
Ellis, S.R.M. and Clark, M.B., 1961. Phase equilibrium of ethanol-benzene system. Chem. Age. India, 12:
377.
Frisch, M.J., Pople, J.A. and Del Bene, J.E., 1983. Hydrogen bonds between first-row hydrides and acetylene. J.
Chem. Phys., 91: 4063.
Ghonasgi, D. and Chapman, W.G., 1993. Theory and simulation for associating fluids with four bonding sites.
Mol. Phys., 291: 79.
Hartland, G.V., Henson, B.F., Venturo, V.A. and Felker, P.M., 1992. Ionization-loss stimulated Raman-spec-
troscopy of jet-cooled hydrogen-bonded complexes containing phenols. J. Phys. Chem., 96: 1164.
Haughton, CO., 1967. Vapor-liquid equilibrium in benzene-acetic acid system. Br. Chem. Eng., 12: 1102.
Heidemann, R.A. and Prausnitz, J.M., 1976, A van der Waals type equation of state for fluids with associating
molecules. Proc. Natl. Acad. Sci. U.S.A., 73: 1773.
Huang, S.H. and Radosz, M., 1990. Equation of state for small, large, polydisperse, and associating molecules. Ind.
Eng. Chem. Res., 95: 7127.
Huang, S.H. and Radosz, M., 1991. Equation of state for small, large, polydisperse, and associating molecules:
extension to fluid mixtures. Ind. Eng. Chem. Res., 30: 1994.
Huron, M.J. and Vidal, J., 1979. New mixing rules in simple equations of state for representing vapor-liquid
equilibria of strongly non-ideal mixtures. Fluid Phase Equilibria, 3: 225.
Ikonomou, G.D. and Donohue, M.D., 1988. Extension of the associated perturbed anisotropic chain theory to
mixtures with more than one associating component. Fluid Phase Equilibria, 39: 129.
Joesten, M.D. and Schaad, L.J., 1968. In: A.K. Covington and P. Jones (Eds.), Hydrogen-Bonding Solvent
Systems. Taylor and Francis, London.
Kobayashi, R. and Katz, D.L., 1953. Vapor-liquid equilibria for binary hydrocarbon-water mixtures. Ind. Eng.
Chem., 45: 440.
Kooner, Z.S. and Fenby, D.V., 1980. Vapor pressure study of the deuterium exchange reaction in methanol-
ethanol system: equilibrium constant determination. Aust. J. Chem., 33: 1943.
Lee, S.C., 1931. Partial pressure isotherms. J. Phys. Chem., 35: 3558.
Leung, R.W.K. and Dadakshan, A., 1987. Prediction of LLE using UNIFAC for organic acid-water-toluene
systems. Ind. Eng. Chem. Res., 26: 1593.
Levine, S. and Perram, J.W., 1968. In: A.K. Covington and P. Jones (Eds.), Hydrogen-Bonded Solvent Systems.
Taylor and Francis, London.
Luedecke, D. and Prausnitz, J.J., 1985. Phase equilibria for strongly non-ideal mixtures from an equation of state
with density dependent mixing rules. Fluid Phase Equilibria, 22: 1.
Magnussen, T., Rasmussen, P. and Fredunsland, A., 1981. UNIFAC parameter table for prediction of liquid-liquid
equilibria. Ind. Eng. Chem., Process Des. Dev., 20: 331.
Mathias, P.M. and Copeman, T.W., 1983. Extension of the Peng-Robinson equation of state to complex mixtures:
Evaluation of the various forms of the local composition effect. Fluid Phase Equilibria, 13: 91.
Mathias, P.M., Klotz, H.C. and Prausnitz, J.M., 1991. Equation of state mixing rules for multicomponent mixtures:
The problem of invariance. Fluid Phase Equilibria, 67: 31.
J. Suresh, E.J. Beckman / Fluid Phase Equilibria 99 (1994) 219-240 239

Messov, V., Pfestorf, R. and Kuchenbecker, D., 1977. Thermodynamic studies on solvent/n-paraffin systems. III.
The benzene/n-heptane and benzene/n-decane systems. Z. Phys. Chem. (Leipzig), 258: 24.
Michel, S., Hooper, H.H. and Prausnitz, J.M., 1989. Mutual solubility of water and hydrocarbons from an equation
of state. Need for an unconventional mixing rule. Fluid Phase Equilibria, 45: 173.
Michelsen, M.L. and Kistenmacher, H., 1990. On composition dependent interaction coefficients. Fluid Phase
Equilibria, 58: 229.
Morachevskii, A.G. and Belousov, V.P., 1958. Three-phase equilibrium in the system benzene-ethanol-water.
Vestn. Leningr. Univ., Fiz., Khim., 13: 117.
Nagai, J. and Isii, N.J., 1935. Vapor pressures of mixture of ethanol and cyclohexane. Sot. Chem. Ind. Jpn., 38: 86.
Nagata, I., 1988. Liquid-liquid equilibrium for ternary mixtures containing two alcohols and one saturated
hydrocarbon. Thermochim. Acta, 127: 337.
Nagata, I., 1989. Modification of the extended UNIQUAC model for ternary and quaternary liquid-liquid
equilibrium calculation. Fluid Phase Equilibria, 51: 53.
Nagata, I., 1990. Simultaneous correlation of ternary vapor-liquid equilibria and liquid-liquid equilibria. Ther-
mochim. Acta, 64: 119.
Nagata, I. and Fuzukawa, M., 1990. Thermodynamics of associating solutions. Ternary liquid-liquid and vapor-
liquid equilibria for solutions containing acetic acid, water, and one non-associating component. Thermochim.
Acta, 170: 243.
Nagata, I. and Ohta, T., 1983. Liquid-liquid equilibrium for systems acetonitrile-benzene-water, acetonitrile-
toluene-cyclohexane and methanol-ethanol-cyclohexane. J. Chem. Eng. Data, 28: 256.
Nagata, I., Miyamoto, K., Allesi, P. and Kikic, I., 1987. Prediction of thermodynamic equilibrium properties from
mutual solubility data. Thermochim. Acta, 120: 63.
Narodoslawsky, M., Huemer, H. and Moser, F., 1987. Liquid-liquid equilibria and densities of the coexisting
phases of methanol-water-dichloromethane system in the temperature range between 20 and 90°C. Monatsh.
Chem., 118: 1129.
Panagiotopoulos, A.Z. and Reid, R.C., 1986. Multiphase high pressure equilibrium in ternary aqueous systems.
Fluid Phase Equilibria, 29: 525.
Panayioutou, C. and Sanchez, I.C., Hydrogen bonding in fluids: equation of state approach. J. Phys. Chem., 95:
1009.
Phutela, R.C., Kooner, Z.S. and Fenby, D.V., 1979. Vapor pressure study of deuterium exchange reactions in
water-ethanol system: equilibrium constant determination. Aust. J. Chem., 32: 2353.
Pople, J.A., Frisch, M.J. and Del Bene, J.E., 1982. Hydrogen bonds between hydrogen halides and unsaturated
hydrocarbons. Chem. Phys. Lett., 91: 185.
Prausnitz, J.M., 1969. Thermodynamics of Fluid Phase Equilibria. Prentice-Hall, Englewood Cliffs, NJ.
Prausnitz, J.M. and De Pablo Juan, J., 1989. Liquid-liquid equilibrium for binary and ternary systems including
the critical region. Transformation to non-classical coordinates. Fluid Phase Equilibria, 50: 101.
Prausnitz, J.M. and De Pablo Juan, J., 1990. Thermodynamics of liquid-liquid equilibria including the critical
region. Transformation to non-classical coordinates using revised scaling. Fluid Phase Equilibria, 59: 1.
Reamer, H.H., Sage, B.H. and Lacey, W.N., 1952. Phase equilibria in hydrocarbon systems. Ind. Eng. Chem., 44:
609.
Reid, W.G., Campbell, E.J., Henderson, G. and Flygare, W.H., 1981. Identification and structure of the benzene-
hydrogen chloride complex from microwave spectrum. J. Am. Chem. Sot., 103: 7670.
Sorenson, J.M. and Arlt, W., 1979. Liquid-Liquid Data Collection - Binary Systems. DECHEMA, Frankfurt.
Suresh, S.J. and Elliott, J.R., Jr., 1991. Application of a generalized equation of state for associating mixtures. Ind.
Eng. Chem. Res., 30: 524.
Suresh. S.J. and Elliott, J.R., Jr., 1992. Multiphase equilibrium analysis via an equation of state for associating
mixtures. Ind. Eng. Chem. Res., 31: 2783.
Suresh, S.J., Enick, R.M. and Beckman, E.J., 1994. Phase behavior of Nylon 6/trifluoroethanol/carbon dioxide
mixtures. Macromolecules, 27: 348.
Tagliavini, G., Arich, G. and Biancani, M., 1955. Miscibility diagram and partition equilibrium in the system
water-benzene-acetic acid. Ann. Chim. (Rome), 45: 292.
240 J. Suresh, E.J. Beckman / Fluid Phase Equilibria 99 (1994) 219-240

Tang, T.H., Hu, W.J., Yan, D. and Cui, Y.P., 1990. A quantum mechanical study on selected a-type hydrogen
bonded systems. THEOCHEM (Eng.), 66: 319.
Treybal, R.E., 1963. Liquid Extraction, 2nd edn. McGraw-Hill, New York.
Tsonopoulos, C. and Wilson, G.M., 1983. High temperature mutual solubilities of hydrocarbons and water. AIChE
J., 29: 990.
Veytsman, B.A., 1990. Are lattice models valid for fluids with hydrogen bonds? J. Phys. Chem., 94: 8499.
Walsh, J.M., Koh, C.A. and Gubbins, K.E., 1992. Physical theory for fluids of small associating molecules. Fluid
Phase Equilibria, 76: 49.
Wei, S., Shi, Z. and Castleman, A.W., Jr., 1991. Mixed cluster ions on a structure probe: Experimental evidence for
clathrate structures of water-hydrogen ion clusters” J. Chem. Phys., 94: 3267.
Wertheim, M.S., 1984a. Fluids with highly directional attractive forces. I. Statistical thermodynamics. J. Stat. Phys.,
35: 19.
Wertheim, MS., 198413.Fluids with highly directional attractive forces. II. Thermodynamic perturbation theory and
integral equations. J. Stat. Phys., 35: 35.
Wertheim, M.S., 1986a. Fluids with highly directional attractive forces. III. Multiple attraction sites. J. Stat. Phys.,
42: 459.
Wertheim, M.S., 1986b. Fluids with highly directional attractive forces. IV. Equilibrium polymerization. J. Stat.
Phys., 42: 477.
Wertheim, M.S., 1986c. Fluids of dimerizing hard spheres and fluid mixtures of hard spheres and dispheres. J. Chem.
Phys., 85: 2929.
Wittrig, T.S., 1979. Ph.D. Thesis, University of Illinois, 1977. In: J.M. Sorenson and W. Arlt (Eds.), Liquid-Liquid
Data Collection - Binary Systems. DECHEMA, Frankfurt.

Potrebbero piacerti anche