You are on page 1of 15

[lIMl[

EUlLlllIA
ELSEVIER FluidPhaseEquilibria 121 (1996) 125-139

A modified NRTL equation for the calculation of phase equilibrium


of polymer solutions
You-Ting Wu, Zi-Qiang Zhu *, Dong-Qiang Lin, Le-He Mei
Department of Chemical Engineering, Zhejiang University, Hangzhou 310027, China
Received 12 June 1995; accepted 11 January 1996

Abstract

A local composition model based on the lattice theory, which is a modified non-random two-liquid (NRTL)
equation, is developed for the representation of the Helmholtz free energy of polymer solutions. A simple
temperature dependence of the model parameters is introduced to account for the oriented interactions without
changing the general formulation of the free energy of mixing. The model provides a flexible thermodynamic
framework for both correlating and predicting the phase equilibrium of polymer solutions. Special attention is
paid to the representation of phase behavior of homologous polymer solutions and systems with or without
oriented interactions.

Keywords: Model; Activity coefficient;Local composition; Vapor-liquid equilibrium; Liquid-liquid equilibrium; Polymer
solutions

1. I n t r o d u c t i o n

A quantitative understanding of the thermodynamic behavior in polymer systems has enormous


practical value. The proper design and control of polymer reaction and processing operations usually
require a knowledge of the phase equilibria. In many cases, mixtures involving strongly interactive
species need to be understood. The solutions of water soluble polymers are an example and are of
great interest in the biochemical engineering and pharmacological industry where in many cases water
is the only solvent. In such systems, strong, local interactions (or the hydrogen-bonding effect)
between the species play a most important role and make it harder to understand such systems.
Besides, the simultaneous representation of phase behavior of homologous polymer solutions is
undoubtedly important when the experimental data are not readily available, and should receive
further study in the literature.

* Corresponding author.

0378-3812/96/$15.00 © 1996ElsevierB.V. All rights reserved


PII S0378-3812(96)03015-4
126 Y.-T. Wu et a l . / F l u i d Phase Equilibria 121 (1996) 125-139

A variety of G E models have been developed during the last 50 years. Among them the
Flory-Huggins (F-H) equations [1,2] and the empirical F - H X models [3] have been the most
popular expressions for correlating experimental equilibrium data of polymer solutions. However,
these models have not successfully described the phase behavior of systems with strong interactive
species, and of systems which only differ in polymer molecular weight. Recently, several models
partly concerning the above two cases have been proposed. Ochs and Cabezas [4] developed a
non-lattice coordination theory to describe simultaneously aqueous solutions of polyethylene glycols
(PEGs). Good agreement with the experimental data was obtained. However, an adsorption model has
to be incorporated in the calculation of very concentrated range, and their theory can not easily be
extended to multicomponent polymer solutions. Elbro and coworkers [5] proposed a F-H-type
equation for the entropy of mixing. After combining it with the energy contribution taken from the
UNIQUAC model [6] for the free energy of mixing, the prediction of vapor-liquid equilibrium(VLE)
of nearly athermal polymer solutions was achieved. Later, Chen [7] suggested a correlative G E model,
which used a combination of F-H-type expression for the entropy of mixing and the non-random
two-liquid (NRTL) theory [8] for the local composition contribution. He concluded that the model
parameters are independent of temperature, chain length and polymer concentration. More recently,
Vetere [9] linked the model parameters to the difference between the Hildebrand solubility parameters
of polymer and solvent, and used this rule to predict the VLE of polymer solutions. As to the
description of systems with strong, local interactions, the double lattice model proposed by Hu and
coworkers [10], the soivation model of Yu and coworkers [11], and the generalized lattice-fluid
model with specific interactions of Sanchez and Balazs [12], are among the most successful. The
successful ideas from the above models are incorporated into our model.
In this work, a local composition model is derived for the Helmholtz free energy of mixing, and is
applied to the calculation of phase behavior of homologous polymer solutions with or without
oriented interactions.

2. Model development

Our aim here is to establish expressions for the Helmholtz free energy of mixing for polymer
solutions. The model development is shown to begin with binary systems. Under the lattice condition,
each molecule of solvent (1) occupies one site and each molecule of flexible polymer (2) occupies r 2
sites, where r 2 is the number of segments in the polymer molecule relative to r I = 1 for the solvent.
Then the canonical partition function Q at temperature T is given by

Q = ~ Q°QNRexp( - E~,/kT) (1)

where k is the Boltzmann constant, E,~ is the potential energy at configuration a , Q0c is the
combinatorial factor at the athermal limit of the solution, and QNR aCCOUntS for the entropic
contribution of non-random mixing caused by the thermal effect. For pure components, QNR = 1, then
the partition functions of pure components are

Qi= ~_~Q°iexp(-E~,i/kT ), (i=1,2) (2)


c~
Y.-T. Wu et al./Fluid Phase Equilibria 121 (1996) 125-139 127

Using the approximation of the maximum term in the partition functions, the Helmholtz free energy
of mixing, A A, can then be calculated as follows
AA
- lnQ+lnQl +lnQ2
kT
QOc E~ - E~ .1 - E, ,2
= - In In QNR +
QOc.l QOc.2 kT

AS c ASNR AE AA c AANR
- - - + - - = - - + - - (3)
k k kT kT kT
where A S, A E and A A are the changes of entropy, energy and Helmholtz free energy, respectively,
and the subscripts have the same meaning as above. AA c = - T A S ~ and ASNR, AE and AANR may
be correlated by the Gibbs-Helmholtz equation.
There are many expressions of AS c in the literature. The F - H entropy of mixing and that of
Guggenheim [13] are among the most familiar. However, the F - H expression uses a mean-field
approximation and overestimates the contribution of the configurational entropy. This is one of the
main causes of the x parameter in the models being both temperature and composition dependent.
Guggenheim's expression is an exact one; but it has to involve another kind of pure substance
parameter (surface parameters) and has to preset the value of the coordination number Z (Z = 10 for
the UNIQUAC model), which seems cumbersome to use. Much more recently, Freed and coworkers
[14,15] developed a complicated lattice cluster theory for polymer solutions, which is an exact
mathematical solution of the F - H lattice. They expanded the entropic corrections to the F - H
mean-field approximation in powers of the inverse lattice coordination number and volume fraction of
the polymer segment. In order to make the model simple and practical, we truncate their expression at
its first correction, then the resulting expression for A A c is

A Ac - qbl lnqbl + ln+2 '(' + 1)2 +1+2 (4)


Nrk~ rI r2 -Z rI r?
where
Nr= Nir I +N2r2, qbj = N i r l / N r , +2 = 1 - ( b l

and N~ and N 2 are the molecule numbers of solvent (1) and polymer (2), respectively. The first two
terms are originally from the F - H expression, the third one accounts for the correction to the F - H
mean-field approximation, which can also be understood as the contribution of solution structure and
size dissimilarity between two components or as the local composition effect from the inner
connections of the polymer chain. Fig. 1 compares Eq. (4) with the F - H equation, Guggenheim's
expression and the original Freed's theory. The results show that Eq. (4) improves the F - H equation
considerably.
As to the calculation of A ANR, we propose following the approximated expression by incorporat-
ing the two-fluid theory and the local composition concept (see Appendix)

- - ~_ ¢!

NqkT X'X2 X, + XzG2, X2 + X,G'f2 (5)


128 Y.-T. Wu et a l . / F l u i d Phase Equilibria 121 (1996) 125-139

-0,0 11

.
"'~ -0.20 q4
!\ \ \\x:
,/i
/'! i

'\ /i
< ','\ ¢ i
~. / I

O.O0 0.20 0.40 0.60 0.80 1.00


~2
Fig. 1. Athermal Helmholtz free energy of mixing from lattice theories for a cubic lattice(Z= 6), r 2 = 100, - - - ,
Flory-Huggin; .... Guggenheim; - - , Bawendi-Freed; - - - - - - , this work, Eq. (4).

where

1 ) Z ~'ji - - 8 i i
G'j'i = e x p - ~ r j i , "rJi= 2 kT

and "r and e are energy parameters, and have similar meanings as in the NRTL model. X~ is the
effective mole fraction of segment of species i

X,=N,q,/Nq, X 2=l-X,, N q = N l q , +N2q 2

and qi means effective segment number of species i, and may be correlated to r i as in the usual way,

qi=[ri(Z-Z)+ 2]/Z=ri[1-Z(1-1/ri)/Z ] (i=1,2) (6)

The lattice postulation in the lattice theory is an approximation. The exact value of the coordination
number is not clearly known. Therefore, we define the structural factor of the solution, 1 / Z , as the
non-random factor c~, as done in the N R T L model. After inserting Eq. (4) and Eq. (5) into Eq. (3), the

(,,)2 ,2c,2)
general formulation of A A is obtained

AA
= ~bl ln<b, -+- <b2 lnqb2 -t- c~ (J),(~2 -'}- - - M I X 2 +
NrkT r] r2 rI r2 Nr X 1 q- X2G21 X2 + XIGI2
(7)

where

G 2 , = e x p ( - - o t q 2 1 ), G,2=exp(--o~1"12 )
Y.-T. Wu et al./Fluid Phase Equilibria 121 (1996) 125-139 129

and qi is rewritten into the following expression

qi = ri[1 - 2a(1 - 1/ri) ] (6a)

The activity of solvent in the system can also be derived as

(r,) 1 1 ~+q,X~ 'r21G21


+
"rl2GI2 ]
lna I = ln~l + 1 -- ~ ~2 -t- o~r I rl r2
(X 1 + X2G21) 2 (X 2 + XiGl2) 2
(8)
Eq. (7) and Eq. (8) are our working equations; the extension to multicomponent mixtures is
straightforward. It can be regarded as a modified NRTL model. Besides, it can also frequently go
back to the F - H x model (when a = 0).
It should be noted that the non-random factor c~ is a structural factor of solution and assumed to be
independent of temperature and composition. The value of c~ is typically between 0 and 0.4, and
should be carefully selected for different kinds of solutions. At present, the general rule proposed in
the NRTL model may be used, or c~ should be fitted to the experimental data.

3. Systemswith oriented interactions


The philosophy we have adopted is to generalize the local composition model to include oriented
interactions while trying to preserve its simplicity. The followed methodology is quite similar to that
of Sanchez and Balazs [12]. We improve and generalize it to satisfy our model.
To ensure that our argument is reasonable, it is assumed that the oriented interactions do not
change the general form of the partition function, Q, and therefore, do not change the form of Eq. (7);
but the interaction energies between segment-segmen t pairs will be affected, and therefore, the
interaction parameters, x. This is equivalent to the assumption that the interaction energies, e, in the
partition function should be replaced by the interaction free energies, f, while considering the oriented
interactions. Thus, the problem is altered to how to calculate f from the lattice theory.
We introduce a secondary lattice to calculate f. In this secondary lattice, the total lattice number is
equal to the total number of i-j pairs, Nji, in the primarily lattice. The i-j pairs can be strong with
energy (eji +~ji) or weak with energy e:;, If 0j/ of the Nji interaction pairs are strong and (1 - 0ji)
are weak, then the total potential energy, Eji, between j-i pairs is

Eji = OjiNji(eji + ~ji) -.I-(1 - Oji)NjiEji = Nji(13ji-~- Oji~ji ) (9)

If an oriented j-i interaction occurs in only one unique way, let q:i be the number of ways that the
non-oriented j-i interaction occurs. Otherwise, q j; means the ratio of the statistical degeneracies of
the two states. Then the partition function for this secondary lattice, Q~i, is given by

O)i = ~_,O'c.jiexp[-Nji(eji + Oji~ji)/kT ] (10)


oji
130 Y.-T. Wu et al./Fluid Phase Equilibria 121 (1996) 125-139

Where
• (l Oj,)Uj,
Njil qji
O'c'ji= ( O j i N j i ) ! [ ( 1 - O j i ) N j i ] ! (1 + q j i ) N''

The combinatorial factor, Q~. ji, is calculated in the absence of energetic preference. Thus, the free
energy associated with these j - i interactions is given by

NjiL. i = -- k T lnQ'2i (11)

Approximating Q)i by the generic term in the sum, and minimizing fj~ with respect to Oji yield
1 -F qji
fji = 8ji 4- ~ji -F k T In 1 -I- qji e x p ( ~ j i / k T ) (12)

Eq. (12) is in accordance with that of Sanchez and Balazs [12]; however, it is not convenient to use.
So, further simplification is needed. If the oriented interactions are weak enough, the right-hand side
of Eq. (12) can be twice expanded using the Taylor method. After truncation at its first two terms, the
approximated fji is

-fji- = - - eJi + 1 8ji qji ( Sji ] 2 (13)


kT kT (1 + qj,) kT 2(1 + q~,)2 ~ k~J

By applying the same treatment to f,i and fJT' and using f to replace e in the expression of "rj~ and "r~j
respectively, the resulting expressions are as follows

"rii = aSli)( T o / T ) + a~2i)(T o l T ) 2 (14a)

'rq = a ~ ) ( T o / T ) + a ~ ) ( T o l T ) 2 (14b)

Z
a(.! ) = eJi e l i + /T o (14c)
kT o 1 + q~i 1 + qii

a ~2) =
jl (lad)
2 2(1 + qj,)~ 1 kTo ] 2(1 + q.)~ I kVo ] ]
ag),j and aft),J can be expressed in similar forms as a{.!
)J, and "ji-(2)by replacing subscript i with j, and j
with i. To is the reference temperature, To = 298.15 K, and the adjustable model parameters, a °) and
a (2), are assumed to be independent of temperature and composition. If the temperature range is not
wide, the oriented interactions between i-i and j - j pairs can be further neglected. Then

~2) _ _~2) (15)


aji -- uij
Y.-T. Wu et a l . / F l u i d Phase Equilibria 121 (1996) 125-139 131

c~l c'q

. . . .

~E
0 0
(..)~.,~.,

N ~q

O ~ O ~ O 0 0 ~ O O 0

I I I

C'q ~ ~ O0

~ I~ . ',,O O0 ~

-- d d o ~ -- o
I I I I I

d ~ c5 o d o d, 0
O

x
J O0

t¢3

ddd o o o o od d~ odd od d
I I I I I I I II II I II II I I
_m
I
ddd o o o o od dd ddo do d

t-<
00 eq eq oo
o

O
O O~ ~ "~" I~- ~
t'M

O 0 0 ~,0 O ~ O O O O

O oo O O O

e.,

~ E
--z
~o~
e-, ¢- ~ E 2 >" "~

"~ o ~

o ~.
-6 N
132 Y.-T. Wu et al./Fluid Phase Equilibria 121 (1996) 125-139

~'xl ¢~l ¢~1 -- c'q ¢'q ¢'q

d d d..~ d ~ d~.

-o
Ox

~3

II c~c~ r~
C'q ~ t'xl
-- M 0 - - -- 0

©
~t r..-

8
o -- 0
I I
x,O
0
t",l ~
"t-
O o.
O

O
I I
,.c:

--
,9-
o.
x._.,

oo U'~ O x,O

O 0

O O O 00
. . . . . . 0
o oo 9• oo, o o
d b4
-- -- ¢~. -- ~ (*q --
c~ 0
---: ,--:.

o~ Oh
,.v
e-
o -- -- xO ~-- ~ .O
oo e..,
Ox
e-. ~N m m m m ~
I I
. . . . . . ~ ' ~
o6 N~NN~N =
Ox O O~ Ox m m ~ m N ~ "-N ~D

o~ oo O. o. o . o . o . E
o0 "~cq 0o
-- -- C'q
o"5 oo
E
.=_
O O O ~ O O O o o o o o o
ox -- o'h -- ~pq t-'h ~O O
g
t-'
©

._~
I ,z:

~.8
O.. 0~0
Y.-T. Wu et a l . / F l u i d Phase Equilibria 121 (1996) 125-139 133
4. Results and discussion

The experimental phase equilibrium data of some polymer solutions are used to test the model. In
the calculation, the pure substance parameter, r2, is calculated from the molar volumes of components
or from number average molecular weight (M--~,), and the volume fraction is calculated from weight
fraction (W 2) by using the weight average molecular weight (M----~w if the polymers have large
polydispersity. Otherwise, we take M w = M n. Besides, the non-random factor, c~, is not treated as an
adjustable parameter, but previously selected. The two (without oriented interactions, = t~J
i~(2) = 0 ) uij-(2)
or three (with oriented interactions, a(2
I j ) = a(~
j i ) 4: 0) parameters left are fitted to experimental data of

1.00

)
0.80

0.60 ~k~

t~ 04.0 exp data a 297.6 K

020 pred. (~1

0,00 l i T ; l [ l l ; l ~ l i l r l ;~1 ~ l l l ; I ; I ; l l l l l l :111 llllllr Jil


0.00 0.20 0.40 0.60 0.80 1.00
W2
i.oo

0.80 ~ )
0.60

exp data o 303.1 K "~\


0.40 a 323.1 K \\\
pred. x 338.1 K

O.20

0.00
ooo ...... 6'56 . . . . . . 6'16 . . . . . . 6156. . . . . . 6'~6 . . . . . ~'.oo

W2
Fig. 2. Water activity in PEG-water solutions. (a) PEGI50; (b) PEG300. The data for graphs (a) and (b) are from
Herskowitz and Gottlieb ([17,24,25]) and Malcolm and Rowlinson ([26]), respectively.
134 Y.-T. Wu et a l . / F l u i d Phase Equilibria 121 (1996) 125-139

1.00

~ 3 ( o ) ,

0.80

0.60

exp data o 293.1 K


o.4o

0.20

0.00
0.00 0.20 0.40 0.60 0.80 1.00

1.00
W2
0.80

0.60

6 exp date a 313.1 K


0.40
x 333.1 K

pred.

0.20

0.00 I'I'IIIIJILII~III'IIIIIIIIll liar I'l '111 Illllllll


0.00 0.20 0.40
W2
0.60 0.80 1.00

Fig. 3. Water activity in PEG-water solutions. (a) PEGI500; (b) PEG6000. The data for graphs (a) and (b) are from
Herskowitz and Gottlieb ([25]).

one or two systems using the least-squares technique, and are used to predict other systems in the
sequence.
The results of correlating and predicting the VLE data for different kinds of polymer solutions are
reported in Tables 1 and 2, and Figs. 2 - 4 show the predictions for some PEG aqueous solutions. In
general, agreement between the calculated and experimental values of solvent activity is satisfactory,
especially for non-aqueous solutions. For PEG aqueous solutions, the parameters are first correlated
from PEG200 solution, and we use these parameters to predict the VLE of other PEG solutions.
Unfortunately, the results are only good for PEGs with molecular weights of less than 600, relatively
large deviation occurs for others. The analysis shows that this may result from the strong effect of the
Y.-T. Wu et ul./Fluid Phase Equilibria 121 (1996) 125-139 135

1.00

0.98

0.96
pred.

0.94 1 ......... ~, ...... ~......... q........ r .........


0.00 0.10 0.20 0.50 040 0 50

W2
Fig. 4. Water activity in PEG-water solutions. The data for PEG1000 (©) and PEG8000 (x) are from Ochs and coworkers
([27]), and those for PEGI000 (+) and PEG4000 (A) are from Lin and coworkers ([28]). All experimental data were
measured at 298.1 K.

end group in the PEG molecule when the molecular weights are relatively low. Therefore, the
parameters are re-fitted to the experimental data of PEG3000 and PEG5000 solutions to predict the
remaining systems.
For some well-studied systems, a comparison is made between the present work and the results
obtained by applying other predictive models from the literature, such as the UNIFAC-FV model of
Oishi and Prausnitz [16], the methods proposed by Vetere [9], Elbro and coworkers [5] and
Herskowitz and Gottlieb [17]. Although these models are completely predictive, they suffer inaccu-
racy in the predicted solvent activities, especially for P E G - w a t e r systems. Therefore, the present
method provides an accurate way of correlating and predicting the VLE of polymer solutions, and has
its own advantage of convenience in use.
Table 2 shows both VLE data representations with and without oriented interactions. For
polypropylene glycol/methanol systems, the oriented interactions actually do little for the data
representation, while for strongly interactive systems they are essential to the calculation at different
temperatures. For P E G - w a t e r systems, the deviation without oriented interactions is nearly twice the
deviation with oriented interactions.
To further examine the oriented interactions, we calculate the binodals of PEG-water systems. The
condition for liquid-liquid equilibrium is the equality of component activities in both phases, and the
condition for critical points is written
02A A 03A A
02~b----~- = 03~b----~ - 0 (16)

Three adjustable parameters, a(211), a~2), a~2) ( = a(2~)), are obtained by simultaneously fitting to the
critical point(s) and the data of one tieline (sometimes). Fig. 5 shows results from PEG-water
systems. It can be seen that three parameters are enough to represent the coexistence curve of the
PEG8000-water system which only has a lower critical solution temperature (LCST), while for the
136 Y.- T. Wu et al. / Fluid Phase Equilibria 121 (1996) 125-139

500.00 !
4 S(o)

4 J

450.00
5~
~- 5(b)

400.00 ~L~__...;;~_~.._.~.~._19,....~

35000 i i
0.00 0.10 o.~o o.~o 0.4-0 0.50
W2
Fig. 5. Binodals of polyethylene glycol aqueous solutions. (a) Mn = 2180, r 2 = 98.5, a = 0.25. - - - - - - , Calculated with
the model parameters: a~l,) = 2.0446, a~ ) = 1.9224, a~22) = a~21) = - 1.5462. - - , Calculated with the model parame-
ters: a~11) = 1.9254, a~12) = 1.7969, ",2~(2)_- -2J~(2)= - 1.3443 (for UCST only, T > 465 K); a~ll) = 2.2083, a~22) = 2.0906, a~ ) =
a~2~) = - 1.7970 (for LCST only, T < 465 K); A, Calculated critical points. © exptl, points from Saeki and Coworkers [29].
(b). M n = 8000, r 2 = 363.4, a = 0.25. - - , Calculated with the model parameters: a~]j) = 3.9476, a~ ) = 9.7054,
a~ ) = a~22~
) = -8.0578; *, calculated critical point; [3, experimental points from Saeki and coworkers ([29]).

P E G 2 1 8 0 - w a t e r s y s t e m w h i c h has an u p p e r critical solution temperature ( U C S T ) and a L C S T , three


parameters yield a n a r r o w e r c o e x i s t e n c e curve. H e n c e , w e calculate the u p p e r part and the l o w e r part
o f the c o e x i s t e n c e c u r v e o f the P E G 2 1 8 0 - w a t e r system, respectively; g o o d results are obtained.

5. Conclusions

A simple local c o m p o s i t i o n m o d e l , w h i c h is a m o d i f i e d N R T L m o d e l , is presented f o r evaluating


the t h e r m o d y n a m i c properties o f p o l y m e r solutions, and appears to be reliable both for correlating and
predicting the V L E o f h o m o l o g o u s p o l y m e r solutions. T h e simple temperature d e p e n d e n c e o f the
m o d e l parameters is introduced to a c c o u n t for oriented interactions and to aid the calculations at
different temperatures.
F o r strongly interactive s y s t e m s , the oriented interactions are essential to the calculation o f phase
behavior, especially to the calculation o f L C S T with a relatively low n u m b e r o f adjustable parameters.

6. List of symbols

a interaction p a r a m e t e r defined in Eqs. ( 1 4 a - d ) , or activity o f species


A free e n e r g y o f m i x i n g
Y.-T. Wu et a l . / Fluid Phase Equilibria 121 (1996) 125-139 137

potential energy at configuration et


f interactive free energy between segment-segment pairs
G binary parameter in the NRTL equation
k Boltzmann constant
M molecular weight
N number of molecules, or number of segment-segment pairs
q effective segment number of polymer, or ratio of statistical degeneracies of two states
Q canonical partition function, or combinatorial factor in the partition function
r number of segments per molecule
S entropy
T absolute temperature
X effective mole fraction of segments
Z coordination number in the lattice theory

6.1. Greek Letters

ot non-random factor in the NRTL model


8 oriented interaction parameter between segment-segment pairs
E interaction energy between segment-segment pairs
0 mole fraction of segment-segment pairs with oriented interactions
"1" binary interaction parameter defined in the NRTL equation
+ volume fraction

6.2. Subscripts

C combinatorial factor
i,j any species or segments
q, tt,jj segment-segment pairs
NR contribution from non-random-mixing
1,2 solvent and polymer, respectively

6.3. Superscripts

0 athermal limit
! tV
notation for distinguishment
(1),(2) notation for distinguishment

Acknowledgements

This work is sponsored by the National Natural Science Foundation of China.


138 Y.-T. Wu et al. / Fluid Phase Equilibria 121 (1996) 125-139

A p p e n d i x A . C a l c u l a t i o n o f A A NR

We calculate A ANR from A E. The reference states for the equation of A E are pure liquid for the
solvent and a hypothetical segment aggregation state for the polymer segments. In this hypothetical
aggregation state, all segments are surrounded by segments of the same type. Following the similar
derivation of the well-known local-composition equations, the effective local mole fractions X;i and
Xii of segments j and i in the immediate neighborhood of a central segment i are related by the
following Boltzmann-type equation
Xji Xjexp(-eji/kT)
-- = (A1)
Xii Xiexp( - e i i / k T )
According to the two-fluid theory and after the simple algebraic manipulation, A E can thus be written
as
Z
A E= ~Nq[ XIXz,(e2, - e,,) + X2X,z(el2- e22)] (A2)

(
= NqX'X2 X l + X2G'2,
"r21G21 -+-
X 2 -t- XIGt,2
(A3)

where
G2,' = e x p ( - 2 u r 2 , ) , G'12=exp(-2url2 )
There are three methods to compute A ANR from A E:
1. if we neglect the contribution of ASNR to A ANR, and take the factor 2 / Z as the non-random
factor, Eq. (A3) becomes the extended form of the original NRTL model;
2. if we insert Eq. (A3) into the Gibbs-Helmholtz equation and take a complete integration, the
resulting expression for A ANR becomes an extended Wilson equation;
3. if the energy parameters, 'r;i, are assumed to have small values, the right-hand side of Eq. (A3) can
be twice expanded into a Taylor series.

_NqkT=X,
_ X2 (%1 + ' r , 2 l - ~ ( X , "r2, +X2"r~2 ) - X, X2('r~, + ' r l32 ) + . . . (14)

Using the Gibbs-Helmholtz equation and integrating, an approximated equation is obtained

- - N I X 2 ('I"21-1-- TI2 ) -- (XIT2I +Xz'r22) - ~ X, X2('r~, +'r~2) + ... (15)

( "rzlG'21 + "r12G'12 ) (16)


~XIX2 X 1 -t- X~zv21t';"" X2-t.-XIGI2"

Eq. (A6) also has a similar form to the NRTL model. In fact, we only need to ensure that (1/Z)'rji is
small enough in deriving Eq. (A6). Fortunately, this can be easily realized from our knowledge that
most solutions have high coordination numbers. Herein, Eq. (A6) is adopted to express A ANR.
Y.-T. Wu et al./Fluid Phase Equilibria 121 (1996)125-139 139

References

[1] P.J. Flory, J. Chem. Phys., 9 (1941) 660.


[2] M.L. Huggins, J. Chem. Phys., 9 (1941) 440.
[3] R. Koningsveld and L.A. Kleintjens, Macromolecules, 18 (1985) 243-252.
[4] L.R. Ochs and H. Cabezas, Jr., Fluid Phase Equilibria, 88 (1993) 1-12.
[5] H.S. Elbro, A. Fresdenslund and P. Rasmussen, Macromolecules, 23 (1990) 4707-4714.
[6] D.S. Abrams and J.M. Prausnitz, AIChE J., 21 (1975) 116-128.
[7] C.C. Chen, Fluid Phase Equilibria, 83 (1993) 301-312.
[8] H. Renon and J.M. Prausnitz, AIChE J., 14 (1968) 135-142.
[9] A. Vetere, Fluid Phase Equilibria, 97 (1994) 43-52.
[10] Y. Hu, X. Ying, T. Wu and J.M. Prausnitz, Fluid Phase Equilibria, 83 (1993) 289-300.
[11] M. Yu, H. Nishiumi and J.S. Arons, Fluid Phase Equilibria, 83 (1993) 357-364.
[12] I.C. Sanchez and A.C. Balazs, Macromolecules, 22 (1989) 2325-2331.
[13] E.A. Guggenheim, Mixtures, The Oxford University Press, Amen House, London, 1952, PP. 183-258.
[14] M.G. Bawendi and K.F. Freed, J. Chem. Phys., 88 (1988) 2741-2756.
[15] J. Dudowicz, K.F. Freed and W.G. Madden, Macromolecules, 23 (1990) 4803-4819.
[16] T. Oishi and J.M. Prausnitz, Ind. Eng. Chem. Process Des. Dev., 17 (1978) 333-339.
[17] M. Herskowitz and M. Gottlieb, J. Chem. Eng. Data, 29 (1984) 450-452.
[18] A. Nakajima, II. Yamakawa and I. Sakarada, J. Polym. Sci., 35 (1959) 489-495.
[19] K. Matsumara and T. Katayama, Kagagu Kogaku, 38 (1974) 388-392 (Japan).
[20] C. Brooth and C.J. Devoy, Polymer, 12 (1971) 309.
[21] S.G. Canagaratna, D. Margerison and J.P. Newport, Trans. Faraday Soc., 62 (1966) 3058.
[22] C.E.H. Bawn and R.D. Patel, Trans. Faraday Soc., 52 (1956) 1664.
[23] M.L. Lakhanpal and B.E. Conway, J. Polym. Sci., 46 (1960) 75-92.
[24] M. Herskowitz and M. Gottlieb, J. Chem. Eng. Data, 29 (1984) 173-175.
[25] M. Herskowitz and M. Gottlieb, J. Chem. Eng. Data, 30 (1985) 233-234.
[26] G.N. Malcolm and J.S. Rowlinson, Trans. Faraday Soc., 53 (1957) 921-931.
[27] L.R. Ochs, M. Kabiri-Badr and H. Cabezas, Jr., AIChE J., 36 (1990) 1908-1912.
[28] D. Lin, L. Mei, Z. Zhu and Z. Han, Fluid Phase Equilibria, 18 (1996)241-248.
[29] S. Saeki, N. Kuwahara, M. Nakata and M. Kaneko, Polymer, 17 (1976) 685-689.

You might also like