Ind. Eng. Chem. Process Des. Dev., Vol. 17, No.
3, 1978 333
Estimation of Solvent Activities in Polymer Solutions
Using a Group-Contribution Method
Takeru Oishi and John M. Prausnitz"
Chemical Engineering Department, University of California, Berkeley, California 94720
A recently proposed group-contribution method, UNIFAC, derived for mixtures of ordinary liquids, is extended to
polymer solutions. The extension provides a free-volume correction as suggested by Flory's equation-of-state
theory for polymer solutions. The modified UNIFAC method can be used for estimating activities of solvents in
polymer solutions when no experimental data for the mixture are available. For 13 binary solvent polymer sys-
tems, calculated and observed solvent activities agree to about f10%.
Introduction
Industrial processing of polymers is often concerned with
polymer solutions. Rational design of processing equipment
may require estimates of solvent activities in polymer mix-
tures; such estimates are particularly important for processes
directed a t polymer devolatization as required to meet envi-
ronmental, health, and safety regulations. This work discusses
a molecular thermodynamic method for making such esti-
mates when no experimental equilibrium data for the mixture
are available. Comparison with typical experimental results
indicates that solvent activities in a variety of concentrated
polymer solutions can often be estimated with an uncertainty
of no more than &lo%.
The estimation method is based on the UNIFAC correlation
which follows from the concept of group contributions, as
discussed previously (Fredenslund et al., 1977). The acronym
UNIFAC denotes UNIQUAC functional-group activity
coefficient. The UNIFAC correlation is based on a semiem-
pirical model for liquid mixtures called UNIQUAC (Universal
quasi-chemical activity coefficient) as discussed elsewhere
(Abrams and Prausnitz, 1975).
Group Contributions
The method presented here is based on the well-known
group-contribution concept which has been successful for
estimating a variety of pure-component properties such as
liquid densities, heat capacities, and critical constants. The
basic idea is that whereas there are thousands of chemical
compounds of interest in chemical technology, the number
of functional groups which constitute these compounds is
much smaller. Therefore, if we assume that a physical prop-
erty of a fluid is the sum of contributions made by the mole-
cule's functional groups, we obtain a possible technique for
correlating the properties of a very large number of fluids in
terms of a much smaller number of parameters which char-
acterize the contributions of individual groups.
Any group-contribution method is necessarily approximate
because the contribution of a given group in one molecule is
not necessarily the same as that in another molecule. The
fundamental assumption of a group-contribution method is
additivity; the contribution made by one group is assumed to
be independent of that made by another group. This as-
sumption is valid only when the influence of any one group in
a molecule is not affected by the nature of other groups within
that molecule.
For example, we would not expect the contribution of a
carbonyl group in a ketone (say, acetone) to be the same as
that of a carbonyl group in an organic acid (say, acetic acid).
On the other hand, experience suggests that the contribution
0019-7882/78/1117-0333$01.00/0
of a carbonyl group in, for example, acetone, is close to (al-
though not identical with) the contribution of a carbonyl
group in another ketone, say, 2-butanone.
Accuracy of correlation improves with increasing distinction
of groups; in considering, for example, aliphatic alcohols, in
a first approximation no distinction is made between the po-
sition (primary or secondary) of a hydroxyl group, but in the
second approximation such distinction is desirable. In the
limit, as more and more distinctions are made, we recover the
ultimate group, namely, the molecule itself. In that event, the
advantage of the group-contribution method is lost. For
practical utility, a compromise must be attained. The number
of distinct groups must remain small but not so small as to
neglect significant effects of molecular structure on physical
properties.
Extension of the group-contribution idea to mixtures is
extremely attractive because, while the number of pure fluids
in chemical technology is already very large, the number of
different mixtures is still larger, by many orders of magnitude.
Thousands of multicomponent liquid mixtures of interest in
the chemical industry can be constituted from perhaps 30,50,
or a t most 100 functional groups.
The UNIFAC Model
The fundamental idea of a solution-of-groups model is to
utilize existing phase equilibrium data for predicting phase
equilibria of systems for which no experimental data are
available. In concept, the UNlFAC model follows Derr and
Deal's (1969, 1973) ASOG model, wherein activities in mix-
tures are related t o interactions between structural groups.
UNIFAC entails the following: suitable reduction of experi-
mentally obtained activity data to obtain parameters char-
acterizing interactions between pairs of structural groups in
nonelectrolyte systems, and use of these parameters to predict
activities in other systems which have not been studied ex-
perimentally but which contain the same functional
groups.
Derr and Deal (1969, 1973) separate the molecular activity
coefficient into two parts: one part provides the contribution
due to differences in molecular size and the other provides the
contribution due to molecular interactions. The first part is
arbitrarily estimated by using the athermal Flory-Huggins
equation; the Wilson equation, applied to functional groups,
is chosen to estimate the second part. Much of the arbitrari-
ness is removed by combining the solution-of-groups concept
with the UNIQUAC equation. First, the UNIQUAC model per
se contains a combinatorial part, essentially due to differences
in size and shape of the molecules in the mixture, and a re-
sidual part, essentially due to energy interactions. Second,
0 1978 American Chemical Society
334 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978

functional group sizes and interaction surface areas are in- There are two adjustable parameters, ~ i and
j ~,i
troduced from independently obtained pure-component,
molecular structure data.
Abrams and Prausnitz (1975) show that the UNIQUAC
equation can give good representation of both vapor-liquid
and liquid-liquid equilibria for binary and multicomponent
mixtures containing a variety of nonelectrolytes such as hy-
where u,; refers to the potential energy of an i-j pair.
drocarbons, ketones, esters, water, amines, alcohols, nitriles,
In eq 2, 5 , and 6, the mole fraction is not used; instead we
etc.
use as the primary composition variable wi, the weight fraction
of component i. The summations in eq 6 are over all compo-
Activities from UNIQUAC and UNIFAC
nents, including component i.
In mixtures of ordinary liquids, it is customary to define an The segment fraction &’ is similar to the volume fraction.
activity coefficient yi for component i Pure component parameters ri’ and qi’ are, respectively,
measures of van der Waals volume and surface areas; the units,
Ti = a,/xi (1)
however, are not per mole but per unit mass, here chosen as
where a stands for activity and x for mole fraction. The ac- grams.
tivity, in turn, is defined as the ratio of the fugacity of com- The two adjustable binary parameters ~ i and j ~ , i appearing
ponent i in the mixture to that in a standard state, usually in eq 6 must be evaluated from experimental phase equilib-
chosen to be pure liquid i at system temperature and pressure. rium data.
At modest pressures where gas-phase nonidealities can be In the UNIFAC correlation, the combinatorial part of the
neglected, the activity of i is the ratio of the partial pressure UNIQUAC activities, eq 5 , is used directly. Only pure-com-
of i to the vapor pressure of pure liquid i a t the same tem- ponent properties enter into this equation. Parameters ri’ and
perature. It is the activity, a function of temperature and qi’ are calculated from the sum of the molar group volume and
composition, which we require in process design. Previous group area parameters, Rk and Q k , given in Table I.
publications on UNIQUAC and UNIFAC discuss calculation ,
of activity coefficient y. (7)
For polymer solutions the mole fraction is an awkward unit
of concentration because the molecular weight of polymer is where Vk“),always an integer, is the number of groups of type
so much larger than that of solvent. As a result, activity h in molecule L and M , is the molecular weight of component
coefficient y, defined above, is not useful to such solutions. i. Group parameters Rk and Qk are obtained from the van der
Therefore, toward applying UNIQUAC and UNIFAC to Waals group volume and surface areas v w k and Awk given by
polymer solutions, we must first discuss how these equations Bondi (1968)
must be modified to yield activities, rather than activity
coefficients. Rk = Vuk/15.17; Qk = A,k/(2.5 X lo9) (8)
In a binary mixture, the UNIQUAC equation for the ac- The normalization factors 15.17 and 2.5 X lo9 are those given
tivity of solvent 1 in polymer 2 is by Abrams and Prausnitz (1975);they correspond to a -CH2-
lnal = Inalc + halR
(2)
unit in polyethylene.
combinatorial residual The residual part of the activity, eq 6, is replaced by the
solution-of-groups concept. Instead of eq 6, we write
The combinatorial part is given by
In aiR = ZkUk“) [h r k - In rk“)]
(9)
all groups
where r k is the group residual activity (or activity coefficient)
t
+-Mlql’ln---Mlq1’
2 41/ 2
81’ z
(1 -$) ( 3 ) and r k 1,) is the group residual activity (or activity coefficient)
of group k in a reference solution containing only molecules
of type i.
Since In this treatment, can be considered to be either activity
or activity coefficient because
-
I - - =41&’
X 1
(1 - -); 2 - 42’ (4) a, = a,Ca,R = X l y , C T , R (10)
, follows that a I R = yIR.
Since uLc= x I y L c it
+ 42’ + - Mlql’ In 7
- -
2 81’
In a l c In 41’ In eq 9, the term In r k ( I ) is necessary to attain the normal-
2 41 ization that a, 1 as w, 1.
The group activity coefficient r k is found from an expres-
sion similar to eq 6
where m stands for the number of segments per polymer In r k = MkQk’[l - hl (Zm8,’*,k)
molecule divided by the number of segments per solvent - Zm(@m’*km/zn@n’*nm) (11)
molecule. The residual part is given by
Equation 11 also holds for In r k “ ) In eq 11, 8,’ is the area
fraction of the group m , and the sums are over all different
groups. M , is the molecular weight of functional group m, and
The coordination number z is set equal to 10. These equations Qk’ is the group area parameter per gram such that Qk’ =
use two composition variables; they are the surface fraction Qk/Mk 8,’ is calculated in a manner similar to that for 8,‘
8i’ and the segment fraction &’ Qm’ Wm
e,‘ = = e,
C Qn’wn
n
 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978 335

Table I. Group Volume and Surface Area Parameters

Group namea No. Rk Qk Sample group assignment
1 CH3 1 0.9011 0.848 Butane: 2 CH3,2 CH2
CHz 2 0.6744 0.540
“CHz” CH 3 0.4469 0.228 Isobutane: 3 CH3,l CH
C 4 0.2195 0.000 2,2-Dimethylpropane: 4 CH3, 1 C
2 CHz=CH 5 1.3454 1.176 1-Hexene: 1 CH3,3 CHp, 1 CHz=CH
CH=CH 6 1.1167 0.867 2-Hexene: 2 CH3,2 CH2,l CH=CH
“C=C” CH=C 7 0.8886 0.676 2-Methylbutene-2: 3CH3 1 CH=C
CHz=C 8 1.1173 0.988 2-Methylbutene-1: 2CH3 1CH2, lCHz=C
3 ACH 9 0.5313 0.400 Benzene: 6 ACH
“ACH” AC 10 0.3652 0.120 Styrene: 1 CHz=CH, 5 ACH, 1 AC
4 ACCH3 11 1.2663 0.968 Toluene: 5 ACH, 1 ACCH3
“ACCH2” ACCHz 12 1.0396 0.660 Ethylbenzene: 1 CH3,5 ACH, 1 ACCHz
ACCH 13 0.8121 0.348 Cumene: 2 CH3,5 ACH, 1ACCH
5 CHzCHzOH 14 1.8788 1.664 1-Propanol: 1 CH3,l CHzCHzOH
CHOHCH3 15 1.8780 1.660 2-Propanol: 1 CH3,l CHOHCH3
“CCOH” CHOHCHz 16 1.6513 1.352 3-Octanol: 2 CH3,4 CH2,l CHOHCH2
CH3CHzOH 17 2.1055 1.972 Ethanol: 1 CH3CHzOH
CHCHzOH 18 1.6513 1.352 2-Methyl-1-propanol: 2 CH3,l CHCHzOH
6 CH30H 19 1.4311 1.432 Methanol: 1 CH30H
7 HzO 20 0.92 1.40 Water: 1 HzO
8 ACOH 21 0.8952 0.680 Phenol: 5 ACH, 1ACOH
9 CH3CO 22 1.6724 1.488 Ketone group is 2nd carbon;
2-Butanone: 1 CH3, I CH2,l CH3CO
“CHzCO” CHzCO 23 1.4457 1.180 Ketone group is any other carbon;
3-Pentanone: 2 CH3,l CH2,l CHzCO
10 CHO 24 0.9980 0.948 Acetaldehyde: 1 CH3, 1 CHO
11 COOCH3 25 1.9031 1.728 Butyl acetate: 1CH3,3 CH2,l COOCH3
“COOC” COOCHz 26 1.6764 1.420 Butyl propanoate: 2 CH3,l CH2,l COOCHz
12 CH30 27 1.1450 1.088 Dimethyl ether: 1 CH3,l CH30
CHzO 28 0.9183 0.780 Diethyl ether: 2 CH3, 1CH2, 1 CHz0
“CHz0” CH-0 29 0.6908 0.468 Diisopropyl ether: 4 CH3,l CH, 1 CH-0
FCH20 30 0.9183 (1.1) Tetrahydrofuran: 3 CH2,l FCHzO
13 CH z m z 31 1.3692 1.236 n-Propylamine: 1 CH3,l CH2,l CHzNHz
“CNHz” CHNHz 32 1.1417 0.924 Isopropylamine: 2 CH3,l CHNHz
CH3NHz 33 1.5959 1.544 Methylamine: 1 CH3NHz
14 CHzNH 34 1.2070 0.936 Diethylamine: 2 CH3, 1 CHz, 1 CHzNH
“CNH” CHNH 35 0.9795 0.624 Diisopropylamine: 4 CHa, 1 CH, 1 CHNH
CH3NH 36 1.4337 1.244 Dimethylamine: 1 CH3,l CH3NH
15 ACNHz 37 1.0600 0.816 Aniline: 5 ACH, 1ACNH2
16 CH3CN 38 1.8701 1.724 Acetonitrile: 1 CH3CN
“CCN” CHzCN 39 1.6434 1.416 Propionitrile: 1 CH3, 1 CHzCN
17 COOH 40 1.3013 1.224 Acetic acid: 1 CH3,l COOH
“COOH” HCOOH 41 1.5280 1.532 Formic acid: 1 HCOOH
18 CHzCl 42 1.4654 1.264 Butyl chloride: 1 CH3, 2 CHz, 1 CHzCl
“CC1” CHCl 43 1.2380 0.952 Isopropyl chloride: 2 CH3, 1 CHCl
cc1 44 0.7910 0.724 tert-Butyl chloride: 3 CH3, 1 CC1
19 CHClz 45 2.0606 1.684 1,l-Dichloroethane: 1 CH3,l CHClz
“CC12” CHzClz 46 2.2564 1.988 Dichloromethane: 1 CHzClz
CClZ 47 1.8016 1.448
20 CHC13 48 2.8700 2.410 Chloroform: 1 CHC13
“CC13” CC13 49 2.6401 2.184 l,l,l-Trichloroethane: 1 CH3, 1 CC13
21 cc4 50 3.3900 2.910 Carbon tetrachloride: 1 CC14
22 ACCl 51 1.1562 0.844 Chlorobenzene: 5 ACH, 1 ACCl
23 CH3NOz 52 2.0086 1.868 Nitromethane: 1 CH3N02
“CN02” CHNOz 53 1.5544 1.248 Nitropropane-2: 2 CH3,l CHNOz
CHzNOz 54 1.7818 1.560 Nitropropane-1: 1 CH3,l CH2,l CHzN02
24 ACNOz 55 1.4199 1.104 Nitrobenzene: 5 ACH, 1 ACNOz
25 CSZ 56 2.0570 1.650 Carbon disulfide: 1 CS2
a Note: Subgroups (for example CHz=CH and CH=CH) within the same main group (here C=C) have different values for Rk
and Q k but identical group interaction parameters, aij.
338 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978

Table 11. Binary Systems and Exgerimental Conditions

~ ~~~~~~~ ~

No.
Compn of
Temp, Polymer (2) range, data Lit.
System "C Solvent (1) no. av mol wt vol% (2) points ref
I 25 Benzene Polyisobutylene (4 X 104)k 62-95 11 a
I1 25 Cyclohexane Polyisobutylene (4 X 104)k 39-85 8 b
I11 25 n-Pentane Polyisobutylene (4 X lo4) 33-96 9 C
IV 30 Toluene Polyheptene (2.2 X 105) 16-99.9 21 d
V 70 Benzene Polyethylene oxide (6 X lo5) 22-87 8 e
50.3 21-61 5 e
70 Benzene Polyethylene oxide (5.7 X lo3) 21-96 9 e
45.8 21-62 5 e
VI 69.9 Benzene Polypropylene oxide (5 X lo5) 48-90 6 f
60.2 28-76 7 f
47.2 28-85 11 f
VI1 70 Methyl ethyl ketone Polystyrene (2.9 X lo5) 53-90 6 g
25 40-93 11 g
VI11 80 Toluene Polystyrene (2.9 X lo5) 27-70 3 g
60 40-78 3 g
25 21-95 11 g
IX 25 Propyl acetate Polystyrene (2.9 X lo5) 44-92 10 h
x 50 Acetone Polystyrene (1.6 X lo4) 48-90 7 h
25 54-94 7 h
XI 30 Ethyl acetate Polyvinyl acetate (1.1 x lo5) 48-86 6 i
30 Ethyl acetate Polyvinyl acetate (9 x 103) 38-86 8 i
XI1 30 Acetone Polyvinyl acetate (1 x 106) 47-81 4 i
30 Acetone Polyvinyl acetate (1.1 X 105) 42-87 6 i
30 Acetone Polyvinyl acetate (9 x IO3) 45-84 5 1
XI11 30 Benzene Polyvinyl acetate (1.1x 105) 36-83 11 i
XIV 60 Hexane Hexadecane 32-99 9 I
50 32-99 9 i
40 32-99 9 I
30 32-99 9 j
a Eichinger, B. E., Flory, P. J., Trans. Faraday SOC.,64,2053 (1968). Eichinger, B. E., Flory, P. J., Trans. Faraday SOC.,64,2061
(1968). Eichinger, B. E., Flory, P. J., Trans. Faraday SOC.,64,2066 (1968). Tait, P. J. T., Livesey, P. J., Polymer, 11,359 (1970).
e Booth, C., Devoy, C. J., Polymer, 12,309 (1971). f Booth, C., Devoy, C. J., Polymer, 12,320 (1971). g Bawn, C. E. H., Freeman, R.
F. J., Kamaliddin, A. R., Trans. Faraday Soc., 46,677 (1950). Bawn, C. E. H., Wajid, M. A,, Trans. Faraday SOC.,52,1658 (1956).
1 Matsumura, K., Katayama, T. Kagaku Kogaku, 38,388 (1974). J McGlashan, M. L., Williamson, A. G., Trans. Faraday SOC.,57,

588 (1961). Viscosity-averagemolecular weight.

where W , is the weight fraction of group m in the mixture.
The group interaction parameter qmnis given by
(
qmn= exp - u m ~ ~ u n n =) exp - (a,,/T)
where Urn,is a measure of the energy of interaction between
groups m and n. The group-interaction parameters am,, and
anm (two parameters per pair of groups) are the parameters
which must be evaluated from experimental phase equilibri-
um data. Note that amn has units of degrees Kelvin and that
amn # an,. Parameters umn and an,, obtained from a data
base using a wide range of experimental activities, are given
by Fredenslund e t al. (1977).
The functional groups considered in this work are those
given in Table I. Whereas each group listed has its own values
of R and Q, the subgroups within the same main group (for
example, subgroups 1,l; 1,2; 1,3; and 1,4) are assumed to have
identical group energy-interaction parameters.
Modification for Polymer Solutions
The UNIFAC-UNIQUAC model is based on a two-liquid
lattice theory of liquid mixtures which does not explicitly take
into account changes in free volume caused by mixing. I t has
now been well established by numerous authors, notably by
Flory (1970) and by Patterson (1969), that in polymer solu-
tions free-volume effects (also called equation-of-state con-
tributions) are important. In ordinary liquid mixtures remote
from critical conditions, the components are, to a rough ap-
proximation, equally expanded and therefore free-volume
effects are usually secondary. In polymer-solvent solutions,
however, since the polymer molecules are much more tightly
packed than the solvent molecules, free-volume effects are far
from negligible. T o apply UNIFAC to polymer solvent mix-
tures, we rewrite eq 2
Inai = Inaic + InaiR + InaiFV
(14)
combinatorial UNIFAC free volume
residual
where superscript FV stands for free volume.
For mixtures of ordinary (modest molecular weight) liquids,
In aiFVmakes only a small contribution which is usually
negligible in comparison with other approximations in UNI-
FAC. However, for mixtures of solvents and polymers, In aiFV
is often significant. We find that when UNIFAC (without
free-volume correction) is applied to polymer solutions, pre-
dicted solvent activities tend to be lower than those observed
experimentally. This is consistent with the theory of Flory and
Patterson since the free-volume contribution to the activity
of solvent is expected to be positive.
T o find an expression for In aiFV, we use the equation-of-
state theory proposed by Flory with the simplification that
Flory's parameter Xlz is zero. This parameter takes into ac-
count the difference in potential energies
x12 - 2a12 - a11 - 1/22
 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978 337

Table 111. Comparison of Calculated and Experimental Solvent Activities

Av Av error
error UNIFAC with
Temp, UNIFAC free-volume
System "C Solvent (1) Polymer (2) only, % correction, %
I 25 Benzene Polyisobutylene 33 3
I1 25 Cyclohexane Polyisobutylene 24 5
I11 25 n-Pentane Polyisobutylene 38 8
IV 30 Toluene Poly heptene 14 8
V 70 Benzene Polyethylene oxide (6 X lo5) 7 2
50.3 4 1
70 Benzene Polyethylene oxide (5.7 X lo3) 8 2
45.8 4 0.4
VI 69.9 Benzene Polypropylene oxide 9 6
60.2 6 4
47.2 6 4
VI1 70 Methyl ethyl ketone Polystyrene" 19 5
25 15 7
VI11 80 Toluene Polystyrenea 12 3
60 24 11
25 17 10
IX 25 Propyl acetate Polystyrenea 17 2
X 50 Acetone Polystyrene" 21 7
25 20 6
XI 30 Ethyl acetate Polyvinyl acetateb (1.1X 105) 22 8
30 Ethyl acetate Polyvinyl acetateb (9 x 103) 23 11
XI1 30 Acetone Polyvinyl acetateb (1 x 106) 21 8
30 Acetone Polyvinyl acetateb (1.1x 105) 22 8
30 Acetone Polyvinyl acetateb (9 x 102) 19 4
XI11 30 Benzene Polyvinyl acetateb (1.1x 105) 18 7
XIV 60 Hexane Hexadecane 11 6
50 11 6
40 11 6
30 11 6
" Polymer density is taken from data in toluene solution. Polymer density is taken from data in benzene solution.

Table IV. ComDarison of Calculated and ExDerimental Table VI. Comparison of Calculated and Experimental
Activities for Cyclohexane (1) in Polyisobutylene (2) at Activities for Acetone (1) in Polyvinyl Acetate (2) at 30 "C
25 o c ( M B(2) =4 x 104) ( M (2) = 1.1 x 105)
Vol Vol
IOOa lUF- (alUFV/
% (2) looale b lOOalU (alulale) vd (alUFV/ale) 9/0 (2) looale 100alU (alu/ale) 100alUFV ale)
39.0 96.0 87.7 (0.91) 93.2 (0.97) 41.8 93.4 85.6 (0.92) 90.7 (0.97)
55.9 87.6 73.2 (0.84) 83.7 (0.96) 61.0 81.7 68.4 (0.84) 78.0 (0.96)
66.0 78.4 61.1 (0.78) 74.6 (0.95) 74.0 68.5 52.2 (0.76) 64.1 (0.94)
78.4 61.0 42.6 (0.70) 57.5 (0.94) - /-- 62.7 41.9 (0.67) 53.8
85.2 46.3 30.9 (0.67) 44.4 (0.96) (0.86)
Viscosity average. ale = activity of solvent (experiment).
Q
87.0 50.9 30.9 (0.61) 41.7 (0.82)
alu = activity of solvent (UNIFAC only). alUFV= activity of
solvent (UNIFAC with free-volume correction).
At low pressures, following Flory (1970), we obtain for the
solvent
Table V. Comparison of Calculated and Experimental
Actifities for Propyl Acetate (1) in Polystyrene (2) a t 25
o c (M,, (2) = 2.9 x 105)

Vol
% (2) IOOale 1OOalU (alu/ale 100alUFV(alUFV/ale)
44.5 99.0 92.5 (0.94) 99.6 (1.01) The reduced volume 6 is given by
54.5 95.0 85.8 (0.90) 96.3 (1.01)
62.2 90.6 78.2 (0.86) 91.4 (1.01) I3 = u/u* (17)
66.7 88.0 72.7 (0.83) 87.2 (0.99)
76.6 76.3 58.9 (0.77) 75.3 (0.99) where u1 = volume of solvent per gram, V I * = hard-core vol-
83.1 61.8 44.2 (0.72) 60.0 (0.97) ume of solvent per gram, and 3 c l = number of external degrees
of freedom per solvent molecule. Subscript M refers t o the
mixture.
where q i j is a parameter characteristic of the potential energy T h e reduced volume for the solvent is given by
between a segment belonging to molecule i and a segment
belonging to molecule j . But it is precisely this difference in
potential energy which is calculated by UNIFAC. Therefore,
Flory's equation with Xlz = 0 gives us the desired free-volume where b is a proportionality factor of order unity. T o simplify
contribution. the calculations, we assume t h a t the volume of the liquid
338 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978

c
a
Z 0.6 -
0)
o'8 t
m
L
0

-4-
21
c

2 0.4-

0
- - Experiment
A UNIFAC only
0.2 - 0 UNIFAC with Free-
Volume Correction
Volume Correction

fip,(21 = 40,000 Mn(21: 5700

0
0 0.2 0.4' 0.6 0.8 1.0 '0 0.2 0.4 0.6 0.8 1.0
Volume Fraction o f Polymer Volume Fraction of Polymer
Figure 1. Calculated and experimental activities of benzene (1)in Figure 2. Calculated and experimental activities of benzene (1) in
polyisobutylene (2) at 25 "C. polyethylene oxide (2) a t 70 "C.

I. O W ] The results indicate that the free-volume correction pro-

i
0.8
h especially at high polymer concentration, are often subject to
Experiment
A UNIFAC only
0 UNIFAC with Free-
Volume Correction
M(21 = 110,000
0 0.2 0.4 0.6 0.8 1.0
Volume Fraction of Polymer
Figure 3. Calculated and experimental activities of ethyl acetate (1)
in polyvinyl acetate (2) a t 30 "C.
mixture is additive. Then the reduced volume for the mixture
is given by
U l W l + u2w2
EM =
15.17b(rl'wl + rz'wz)
where u1 and u2 are, respectively, the volume per gram of
solvent and polymer and w is the weight fraction.
Upon comparing calculated w-ith experimental activities
for solvents in polymer solutions, we find that agreement is
best when b = 1.28. For the solvents considered here, c is set
equal to 1.1as suggested by earlier results (Flory, 1970; Beret
and Prausnitz, 1975). For other solvents, especially those
containing large molecules, a large value of c may be need-
ed.
T o illustrate how the calculations are performed, an illus-
trative example is given in the Appendix.
Results
Table I1 gives details for 14 binary systems studied here.
Table I11 gives calculated and experimental activities for these
systems. In these calculations c1 was taken as 1.1for all sol-
vents. For solvents significantly heavier than those shown in
Table 11,a somewhat larger value of c1 should be used.
vides a substantial improvement in the predicted activities.
These predictions, it must be recalled, contain no adjustable
binary parameters.
Tables IV-VI and Figures 1-3 give detailed results for six
solvent-polymer systems. In all cases the activity calculated
from UNIFAC alone (no free-volume correction) is too low.
The free-volume correction always raises the solvent's activity.
Agreement with experiment is usually within &lo%.In some
cases agreement is worse, but the significance of such dis-
agreement is difficult to assess because the experimental data,
appreciable error.
Conclusion
Experimental data for polymer-solvent systems are not
plentiful; the binary systems studied here are primarily
nonpolar although a few polar polymers and polar solvents are
included in Table 11. It appears that the modified UNIFAC
method gives good estimates of solvent activities for most
nonpolar systems and at least fair estimates for polar systems.
As discussed elsewhere (Oishi, 1977),solvent activities can also
be calculated for ternary (or higher) mixtures.
A priori estimates for solvent activities in polymer systems
may be helpful for engineering calculations in process design,
especially in polymer devolatilization operations. Pertinent
computer programs are available from the authors.
Acknowledgment
For financial support, J.M.P. is grateful to the National
Science Foundation and to Union Carbide Corporation.
During his stay at Berkeley, T.O. was the recipient of a fel-
lowship from Asahi Chemical Industry Co., Ltd.
Appendix. Illustrative Calculation Using UNIFAC
with Free-Volume Correction
We calculate the activity of benzene in the benzene (1)-
polyisobutylene (2) system a t 25 "C. The volume fraction of
benzene is 0.1.
density of benzene (pl) = 0.87382 g/cm3
density of polyisobutylene (p2) = 0.91693 g/cm3
w(1) =
-
0.1 0.87382
= 0.09575
+ -
0.1 * 0.87382 0.9 0.91693
w@) = 0.90425
 Ind. Eng. Chem. Process Des. Dev., Vol. 17, No. 3, 1978 339

First step (combinatorial part) ~ 1 ,=

3 61.13 K; ~ 3 , =
1 -11.12 K

rl' =
6 * 0.5313
= 0.04081
\E1,3 = exp (- -) = 0.81463; \ E ~ =, ~1.038
78.11
* 0'400
ql' = ___ - 0.03073 1 - In (0.70126 - 0.81463
78.11
+
P(2 0,9011 0.6744 + 0.2195) = 0.04806 + 0.22328 * 0.81463 + 0.07546)
r2/ =
P 56.10 0.07546
where P is the degree of polymerization +
(0.76126 0.2232810.81463 0.07546 +
- +
P ( 2 * 0.848 0.54)
92' =
P 56.10 -
+
= 0.03986
(0.70126 0.22328)1.038
+
0.70126 0.22328 0.07546 1.038
In r3,g(1)
=0
= 0.056
+ - 1
0.04081 * 0.09575
41' = = 0.0825
+
0.04081 0.09575 0.04806 * 0.90425 In u l R = 6(0.056 - 0) = 0.336
42' = 0.9175 Third Step (Free-Volume Part)
=
0.03073 0.09575 - = 0.07547 1
81'
- +
0.03073 0.09575 0.03986 * 0.90425 0.87382
GI= = 1.4442
02' = 0.92453 1.28 0.04081 15.17
In a l C= In (0.0825) + 0.9175 + 5 - 78.11 ~0.03073 0.09575 + 0.90425
~~

X In (0.07547/0.0825) - 5 - 78.11 - 0.03073

0.87382 0.91693
=
DM
+ -
X (1 - E) = -1.53
1.28(0.04081 * 0.09575 0.04806 0.90425)15.17
= 1.1914
1.4442113 - 1
In alFV = 3.1.1 In
Second Step (Residual Part of UNIFAC) 1.1914113- 1
15.03 2
W1,i = 0.90425 * ____ = 0.48452
- (--
- 1.1 14442 1)/(1-- 1
56.10 11914 1.44421,s) = 0'529
14 03
W1,z = 0.90425 * -= 0.22614
In a1 = -1.53 + 0.336 + 0.529 = -0.665
56.10 a1 = 0.515 (calculated)
W1,4 = 0.19358
a l = 0.533 (experimental)
W3,9 = 0.09575
Literature Cited
0.848
0.48452 * - Abrams, D.S . , Prausnitz, J. M., AlChEJ., 21, 116 (1975).
15.03 Beret, S., Prausnitz, J. M., AlChE J., 21, 1123 (1975).
8'1,l = = 0.70126 Bondi, A., "Physical Properties of Molecular Crystals, Liquids and Glasses",
0.848 0.54 Wiley, New York, N.Y., 1968.

15.03
+
0.48452 - 0.22614 * -
14.03
Derr, E. L., Deal, C. H., 1. Chem. E. Symp. Ser. No. 32, 3, 40 (1969).
Derr, E. L., Deal, C. H., Adv. Chem. Ser., No. 124, 11 (1973).
Flory, P. J., Discuss. Farday Soc., 49, 7 (1970).
0.400
+ +
0.19358 * 0 0.09575 * -
13.02
Fredenslund, A., Grnehling, J., Michelsen. M. L., Rasmussen. P., Prausnitz, J.
M., lnd. Eng. Chem. Process Des. Dev., 16, 450 (1977).
Oishi, T., M.S. Thesis, University of California, Berkeley, Calif.. 1977.
Patterson, D., Macromolecules, 2, 672 (1969).
EY1,z = 0.22328
Received for reuielv July 4, 1977
f3'3,g = 0.07546 Accepted January 23, 1978