Models For Polymer Solutions

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

Computer Aided Property Estimation for Process and Product Design

G.M. Kontogeorgis and R. Gani (Editors)


© 2004 Elsevier B.V. All rights reserved. 143

Chapter 7: Models for Polymer Solutions

Georgios M. Kontogeorgis

7.1 INTRODUCTION - AREAS OF APPLICATION

Knowledge of phase equilibria of polymer systems (solutions, blends, ...) is of interest to the
design of a variety of processes related to polymers. Some examples are shown in the table
below.

Table 1. Applications of polymer thermodynamics.

Application Properties/ Phase Equilibria Involved


Solvent devolatilization after polymerisationPolymer-solvent VLE
Selection of solvents for paints and coatingsPolymer-solvent LLE, often mixed
solvents
Emissions from paint production Polymer-solvent VLE, solvent activities
Polymer recycling via physico-chemical Polymer-solvent LLE
methods (selective dissolution)
Product design Systems with co-polymers and polymer
blends
Design of flexible polymer pipes carrying Gas solubilities and diffusivities in
subsea oil polymers
Finding alternative plasticizers to PVC Compatibility (miscibility) of PVC -
plasticizers
Deposition of polymer thin films using rapid Polymer-SCF (SGE)
expansion from supercritical solution (RESS)
Separation of proteins via aqueous two-phase LLE of polymer-water-protein often in
systems presence of electrolytes

This table shows a variety of systems and types of phase equilibria, which are of interest
in the many practical situations where polymer thermodynamics plays a key role. For this
reason, many different models have been developed for polymer systems and often the
situation may seem rather confusing to the practising engineer. Polymer solutions and blends
are complex systems: frequent existence of liquid-liquid equilibria (UCST, LCST, closed
loop, etc.), the significant effect of temperature and polymer molecular weight including
polydispersity in phase equilibria, free-volume effects and other factors may cause
difficulties. The choice of a suitable model will depend on the actual problem and depends,
specifically on:
144

- type of mixture (solution or blend, binary or multieomponent,...)


- type of phase equilibria (VLE, LLE, SLLE, gas solubility,...)
- conditions (temperature, pressure, concentration)
- type of calculations (accuracy, speed, yes/no answer or complete design,...)

This chapter focuses mostly on simple activity coefficient models for polymers, which can
be applied to a wide range of applications. These are group-contribution based (UNIFAC)
models, which account for some special effects in polymer systems such as free-volume
differences. Some few recommendations on the vast but a bit confusing literature on
equations of state for polymers will be also provided at the end of the chapter. Since most
models often perform better for VLE than for LLE, indirect techniques are widely applied
e.g. for solvent selection. These are briefly summarized in the next section.

7.2 CHOICE OF SOLVENTS

A summary of some rules of thumb for predicting polymer-solvent miscibility, with focus on
the screening of solvents for polymers, is presented here. These rules are based on well-
known concepts of thermodynamics (activity coefficients, solubility parameters) and some
specific ones to polymers (Flory-Huggins parameter). Then, a brief discussion of some of the
concepts involved is included. It can be roughly said that a chemical (1) will be a good
solvent for a specific polymer (2), or in other words the two compounds will be miscible if
one (or more) of the following 'rules of thumb' are valid ' :

i. If the polymer and the solvent have 'similar hydrogen bonding degrees:

1
{cm3} (1)

ii. If the polymer and the solvent have very different hydrogen bonding degrees:

V 4 fo. - *J2 Y + fe, - Sp2 f + (Shl - 8h2 f <R (2)


where R is the Hansen-solubility parameter sphere radius.

iii. Q" < 6 (the lower the infinite dilution activity coefficient of the solvent, the greater
the solvency of a chemical). Values of the infinite dilution activity coefficient above
10 indicate non-solvency. In the intermediate region, it is difficult to conclude if the
specific chemical is a solvent or a non-solvent.
145

iv. Xn - 0-5 (the lower the Flory-Huggins parameter value, the greater the miscibility,
or, in other words, the greater the solvent's capacity of a specific chemical). Values
much above 0.5 indicate non-solvency.

7.2.1 The Rules of Thumb Based on Solubility Parameters

They are widely used. The starting point (in their derivation / understanding) is the equation
for the Gibbs Free-Energy of mixing:

AGmix=AH-TAS (3)

A negative value implies that a solvent/polymer system forms a homogeneous solution i.e.
the two components are miscible. Since the contribution of the entropic term ( - TAS) is
always negative, it is the heat of mixing term that determines the sign of the Gibbs energy.
The heat of mixing can be estimated from various theories e.g. the Hildebrand regular
solution theory for non-polars systems, which is based on the concept of the solubility
parameter. For a binary solvent(l)/polymer(2) system, according to the regular solution
theory:

AW = <p,(p2V{d,-d2)2 (4)

where p,is the so-called volume fraction of component i. This is defined via the mole
fractions x; and the molar volumes V;, as (binary systems):

"=^vo (5)

According to Eq. 4, the heat of mixing is always positive. For some systems with specific
interactions (hydrogen bonding) the heat of mixing can be negative and Eq. 4 does not hold.
Thus, the regular solution theory is strictly valid for non-polar/slightly polar systems, without
any specific interactions.
According to Eqs. 3 and 4, if solvent and polymer have the same solubility parameters, the
heat of mixing is zero and they are thus miscible at all proportions. The lower the solubility
parameter difference the larger the tendency to be miscible. Many empirical rules of thumb
have been proposed based on this observation. Seymour1 suggests that if the difference of
solubility parameters is below 1.8 (cal/cm3)1'2, Eq. 1, then polymer and solvent are miscible.
Similar rules can be applied for mixed solvent - polymer systems, which are very
important in many practical applications, e.g. in the paints and coatings industry and for the
separation of biomolecules using aqueous two-phase systems. The solublity parameter of a
mixed solvent is given by the equation:

8 = £<P,8, (6)
/
146

Barton3"4 provides empirical methods based on solubility parameters for ternary solvent
systems.
Charles Hansen introduced the concept of three-dimensional solubility parameters, which
offer an extension of the regular solution theory to polar and hydrogen bonding systems.
Hansen observed that when the solubility parameter increments of the solvents and polymers
are plotted in three-dimensional plots, then the 'good' solvents lie approximately within a
sphere of radius R (with the polymer being in the centre). This can be mathematically
expressed as:

^<Sin - 8dl )2 + {Spl - 8p2 )2 + (Shi - 8h2 )2<R (2)

where subscript 1 denotes the solvents and subscript 2 the polymer. The quantity under the
square root is the distance between the solvent and the polymer. Hansen found empirically
that a universal value 4 should be added as a factor in the dispersion term to approximately
attain the shape of a sphere. This universal factor has been confirmed by many experiments.
Several other two-dimensional plots have been proposed, which employ all three
contributions e.g. 8p-8h,Sh-8d,5p-8d or even combined plots such as the use of
8V -8h,8v = J8d +d2p plots suggested by van Krevelen2. With few exceptions good
solvents lie within the circle of radius R, which mathematically can be expressed as:

^8rl-8r2f+(Sh]-8h2y <R (7)

The justification for this plot lies in the fact that, of the three solubility parameter increments,
the dispersion one varies the least and, via this average way, it can be treated together with
the polar increment. The hydrogen bonding increment is very important and it is thus
accounted for separately in Eq. 7.
The Hansen method is very valuable. It has found widespread use particularly in the paints
and coatings industry, where the choice of solvents to meet economical, ecological and safety
constraints is of critical importance5. It can explain some cases in which polymer and solvent
solubility parameters are almost perfectly matched and yet the polymer won't dissolve. The
Hansen method can also predict cases where two non-solvents can be mixed to form a
solvent. Still, the method is approximate, it lacks the generality of a full themiodynamic
model for assessing miscibility and requires some experimental measurements. The
determination of R is typically based on visual observation of solubility (or not) of 0.5 g
polymer in 5 cm3 solvent at room temperature. Given the concentration and the temperature
dependence of phase boundaries, such determination may seem a bit arbitrary. Still the
method works out pretty well in practice, probably because the liquid-liquid boundaries for
most polymer/solvent systems are fairly 'flat'. A recent review of the Hansen method with
extensive tables of solubility parameters is available6.
147

7.2.2 The Rule of Thumb Based on the Infinite Dilution Activity Coefficient

Since in several practical cases concerning polymer/solvent systems, the 'solvent' is only
present in very small (trace) amounts, the so-called infinite dilution activity coefficients are
of importance. On a molar and weight basis, they are defined as follows:

r? =lim,,^o r,
nr m
=" m
(xy\ (8)

The weight-based infinite dilution activity coefficient, Q", which can be determined
experimentally from chromatography, is a very useful quantity for determining good
solvents. Low values (typically below 6) indicate good solvents, while high values (typically
above 10) indicate poor solvents according to rules of thumb discussed by several
investigators7"9. The derivation of this rale of thumb is based on the Flory-Huggins model,
discussed in the next section, 7.2.3.
This method for solvent selection is particularly useful because it avoids the need for
direct liquid-liquid measurements and it makes use of the existing databases of solvent
infinite dilution activity coefficients, which is quite large (e.g. the DECHEMA and DIPPR
databases).
Moreover, in the absence of experimental data, existing thermodynamic models (such as
the Flory-Huggins, the Entropic-FV and the UNIFAC-FV discussed later, section 7.3) can be
used to predict the infinite dilution activity coefficient. Since, in the typical case today,
existing models perform much better for VLE and activity coefficient calculations than
directly for LLE calculations, this method is quite valuable and successful, as shown by
sample results in Table 2.
The thermodynamic models UNIFAC-FV (U-FV), Entropic-FV (E-FV), UNIFAC and
GCLF, shown in Table 2, are used for obtaining the infinite dilution activity coefficient for
several PVC systems . We can see that, with few exceptions, these models in combination
with the rule of thumb mentioned above can identify which chemicals are good solvents for
PVC and which are not. Similar results have been presented elsewhere ' for selecting
solvents for paint polymers such as PBMA and PMMA.
This rule of thumb makes use of either experimental or predicted, by a model, infinite
dilution activity coefficients. However, the results depend not only on the accuracy of the
model, but also on the rule of thumb, which in turns depends on the assumptions of the Flory-
Huggins approach. A thermodynamically more correct method is to employ the activity -
concentration (aw) diagram, as shown in figures 1 and 2. Results for the PVC systems are
shown in Table 2. The two plots have been generated with UNIFAC-FV. The maximum
indicates phase split, while a monotonic increase of activity with concentration indicates a
single liquid phase (homogeneous solutions).
148

Table 2. Observed and predicted weight fraction activity coefficients at infinite dilution and
ai(w), activity-weight fraction, for different PVC (Mn=50 000)-solvent systems at 298 K.

Nr, Chemicals Experimental (E-FV) (U-FV) <CCLF) UNIFAC


s/ns
Mansen '92 Fred '75 Uansen '(JI

Cl? a,(iv) (V a,(w) £V a,(«) £V a,(w)


1 M omit; h loraben/cn c S 5.OX S 3.08 S 4.54 s 2.45 S
2 Chloroform Ns 2.98 S 4.44 S 4.54 s 2.11 s
Dichlorom ethane S 3.79 s 1.12 s 4.5 2.64 s
4 Nitrocthane Ns - • - - - 5.67 s
5 \:Ah\\ aeelale S 6.15 S 6.46 s 9.47 1.02 3.78 s

6 1.4-1 )io\il!k' S 12.2 s 19.08 1.34 5.76 s 14.5 1.22


7 Isopropanol Ns 48.41 1.17 40.62 1.18 8.15 s 40.62 1.18
X Meihunol Ns 70,72 1.23 94.5 1.4 9.48 s 61.81 1.05
y Henzj 1 alcohol Ns 23.2 S 19.97 1.05 10.14 s 15.42 s
10 Tetrahydrofurane S - - - 6.h7 - 7.32 1.01
II Toluene S 2.54 S 19.9 1.72 9.16 s 2.87 s
12 Benzene S 4.64 S 4.47 s 8.23 s 3
13 Acetone Ns 6.44 S 9.5 s K.45 s 6.02
14 MI{K (a) S S.M S 6.63 s 9,35 s -
15 MEK(b) s 5.35 S 7.51 s 9.72 s - -
16 Methyl Isjbulyl s 4.93 S 6 s 11.78 1.05 - -
Kelone (a)
17 Methyl Isobulyl s 4.72 S 6.21 s 12.3 1.04 4.49 s
Ketone (bl
18 Di-n-Propyl IZlher Ns 1 3 58 1.03 17.1 1.13 23.49 1.31 9.52 1.01
19 n-Nonane Ns iK.xy 1.24 16.27 1.08 29.K9 1.43 11 57 1.02
20 n-Ociane Ns 17.72 1.2 15.72 1.06 27.34 1.36 10,72 1.0!
21 n-Pentane Ns 15.98 1.2 15.88 1.02 22.56 1.22 8.71 s
22 n-Heptane Ns 16.72 1.16 15.31 1.05 25.24 1.3 9.96 s
23 o-Xylene S 11.26 1.17 5.16 s 10.2 1.02 3.85 s
24 1,2-dichloroben/ene S 2.06 s 1.58 s 6.22 s 2.1)1 s

25 Butyl aery laic Ns 3.67 S 8.65 1.02 12.78 1.08 4.08 s


26 F.thylcne di-chloride S 4.1') s 4.32 s 5.22 s 3.22 s
27 Amyi acetate S 6.42 s 6.43 5 8.41 1.01 22 s
28 Carbon disuliide Ns - - - - 4.6 s 7.14 s

29 Vinyl chloride Ns 2.17 s 3.93 S 3.8 s •

.in Acctonilrile Ns - - - - - - 10.51


Wrong answers 4 6 3 4 13 12 5
Total 26 26 26 26 28 28 26 26
The dark colour indicates the cases where the rule of thumb can be applied, while the light colour
indicates cases where 6<XV <8. and thus no conclusion can be made. The numbers in the a-w
columns indicate maximum on graphs of solvent activity vs. solvent weight fraction curves (meaning
insoluble chemicals), a and b means that there are two ways of defining the compound.
149

Figure 1. Solvent-activity diagram of n- Figure 2. Solvent-activity diagram of


heptane and PVAC at 300 K with UNIFAC- chloroform and PBMA at 300 K with
FV. The maximum indicates a phase split. UNIFAC-FV. Monotonically increasing line
indicates solubility at all concentrations.

7.2.3 The Rule of Thumb Based on the FIory-Huggins Model

The Flory-Huggins (FH) model for the activity coefficient, proposed in the early 40's by
Flory and Huggins12, is a famous Gibbs free energy expression for polymer solutions. For
binary solvent-polymer solutions and assuming that the parameter of the model, the so-called
FH interaction parameter %{1 is constant, the activity coefficient is given by the equation

x, x,
(9)
( n
= In — + 1 YPi+XvVl
x, \ r)
where cp, can be volume or segment fractions and r is the ratio of the polymer volume to the
solvent volume V2/V1 (approximately equal to the degree of polymerization).
Using standard thermodynamics and Eq. 9, it can be found that for high molecular weight
polymer-solvent systems, the polymer critical concentration is close to zero and the
interaction parameter has a value equal to 0.5. Thus, a good solvent (polymer soluble in the
solvent at all proportions) is obtained if Xn - 0-5, while values greater than 0.5 indicate poor
solvency. Since the Flory-Huggins model is only an approximate representation of the
physical picture and particularly the FH parameter is often not a constant at all, this empirical
150

rule is certainly subject to some uncertainty. Nevertheless, it has found widespread use and
its conclusions are often in good agreement with experiment. This can be demonstrated by a
socially important example, the choice of suitable (miscible) plasticizers with PVC10. Typical
results are shown in Tables 3 and 4 and figure 3. The infinite dilution activity coefficients
calculated by E-FV and U-FV are also shown in these Tables.

Table 3. Classification of the solvent power of various plasticizers from different calculation
(thermodynamic) methods and experiments (dilute solution viscosity, apparent melting
temperature, equilibrium swelling).
Plasticizer X a Dilute- App. £V (298K) £V (298K) Equilibrium n,"
soln. Melting (E-FV) (U-FV) swelling (350K) (350K)
viscosity temperature at 350 K (E-FV) (U-FV)
DOS 0.62 0.8 DOA DOS 3.85 7.46 DOS 6.43 6.68
DOA 0.48 1.4 DOS DOA 2.30 13.7 DOA 4.03 7.35
DOP 0.05 2.4 DOP DOP 2.31 4.22 DOP 4.31 3.87
BBP 0.17 2.6 BBP BBP 2.00 2568 BBP 3.16 19.11
DBP 0.04 3.4 DBP DBP 2.12 3.95 DBP 3.4D 3.54

The a and x values are taken from Bigg13. The a values are simply (l-%)/MW. (MW is the
molecular weight).

Table 4. Classification of the solvent power of various phthalates from different calculation
(thermodynamic) methods.
Plasticizer Mw X (323K) a 3i K (323K)
(g/mol) (E-FV) (U-FV)
Didecyl phthalate DDP 446.7 0.98 3.75 4.64
0.56
Dimethyl phthalate DMP 194.20 2.27 3.31 6.96
0.56
Diethyl phthalate DEP 222.2 2.61 2.89 4.82
0.42
Dihexyl phthalate DHP 334.5 3.38 2.77 3.63
-0.13
Dioctyl phthalate DOP 390.9 2.53 3.14 3.98
0.01
Dibutyl phthalate DBP 278.3 3.73 2.64 3.73
-0.04

The a and % values are taken from Doty and Zable14.


151

Figure 3. Dependency of the experimentally determined FH interaction parameter values


and Bigg's alpha values on molecular weight of phthalates. The % values are taken from
Barton4.

There are several, still rather obscure issues about the Flory-Huggins model, which we
summarize here together with some recent developments:

1. There is no single rigorous widely-accepted extension of the FH model to


multicomponent systems. Several extensions have been proposed, but (at least) one
Xi2-value is required per binary.
2. It has been shown that, unfortunately, the FH parameter is typically not a constant and
should be estimated from experimental data. Usually it varies with both temperature
and concentration, which renders the FH model useful only for describing
experimental data. It cannot be used for predicting phase equilibria for systems for
which no data is available. Moreover, when fitted to the critical solution temperature,
the FH model cannot yield a good representation of the whole shape of the miscibility
curve with a single parameter.
3. Accurate representation of miscibility curves is possible with the FH model using
suitable (rather complex) equations for the temperature and the concentration-
dependence of the FH-parameter15"16.
4. In some cases, a reasonable value of the FH parameter can be estimated using
solubility parameters via the equation:

X, 2 =Xs+X* = 0 . 3 5 + ^ ( 8 , - 8 2 ) 2 (10)
152

Eq. 10, without the empirical 0.35 factor, is derived from the regular solution theory.
The constant 0.35 is added for correcting for the deficiencies of the FH combinatorial
term. These deficiencies become evident when comparing experimental data for
athermal polymer and other asymmetric solutions to the results obtained with the FH
model. A consistent underestimation of the data is observed, as discussed extensively
in the literature17, which is often attributed to the inability of the FH model in
accounting for the free-volume differences between polymers and solvents or between
compounds differing significantly in size such as n-alkanes with very different chain
lengths. The term, which contains the "0.35 factor", corrects in an empirical way for
these free-volume effects. However, and although satisfactory results are obtained in
some cases, we cannot generally recommend Eq. 10 for estimating the FH parameter.
Moreover, for many non-polar systems with compounds having similar solubility
parameters, the empirical factor 0.35 should be dropped.
5. Recently, Lindvig el a/.18 proposed an extension of the Flory-Huggins equation using
the Hansen solubility parameters for estimating activity coefficients of complex
polymer solutions.
(Pl CPi o
lny^ln^ +l - ^ + x ^
Xi Xi

(11)
2
Xl2 =0.6-^[(5 dl _ 5d2 )2 +o.25(5pl -8p2f +0.25(8,,, Sh2) ]
In order to achieve that, Lindvig et al.18, as shown in Eq. 11, have employed a
universal correction parameter, which has been estimated from a large number of
polymer-solvent VLE data. Very good results are obtained, especially when the
volume-based combinatorial term of FH is employed, as summarized in Table 5.

Table 5. Average absolute deviations between experimental and calculated activity


coefficients of paint-related polymer solutions using the Flory-Huggins/Hansen
method and three group contribution models. From Lindvig et a/.18. The second
column presents the systems used for optimization of the universal parameter (358
points of solutions containing acrylates and acetates). The last two columns show
predictions for two epoxy resins.

Model % AAD (systems % AAD Araldit 488 % AAD Eponol-55


in database)
FH/Hansen, 22 31 28
Volume (Eq. 11)
FH/Hansen 25 — —
Segment
FH/Hansen 26 — —
Free-Volume
Entropic-FV 35 34 30
UNIFAC-FV 39 119 62
GC-Flory 18 29 37
153

6. Based on the Flory-Huggins model, several techniques have been proposed for
interpreting and for correlating experimental data for polymer systems e.g. the so-
called Schultz-Flory (SF) plot. Schultz and Flory19 have developed, starting from the
Flory-Huggins model, the following expression, which relates the critical solution
temperature (CST), with the theta temperature and the polymer molecular weight:

where \m = — is \'ls m e entropic parameter of the FH model (Eq.10) and r is the

ratio of molar volumes of the polymer to the solvent. This parameter is evidently
dependent on the polymer's molecular weight. The SF plot can be used for correlating
data of critical solution temperatures for the same polymer/solvent system, but at
different polymer molecular weights. This can be done, as anticipated from Eq.12
because the plot of 1/CST against the quantity in parentheses in Eq. 12 is linear. The
SF plot can also be used for predicting CST for the same system but at different
molecular weights than those used for correlation as well as for calculating the theta
temperature and the entropic part of the FH parameter. It can be used for correlating
CST/molecular weight data for both the UCST and LCST areas. Apparently different
coefficients are needed.

7.3 THE FREE-VOLUME ACTIVITY COEFFICIENT MODELS

7.3.1 The Free-Volume Concept

The Flory-Huggins model provides a first approximation for polymer solutions. Both the
combinatorial and the energetic terms need substantial improvement. Many authors have
replaced the random van-Laar energetic term by a non-random local-composition term such
as those of the UNIQUAC, NRTL and UNIFAC models. The combinatorial term should be
extended/modified to account for the free-volume differences between solvents and
polymers.
The improvement of the energetic term of FH equation is important. Local-composition
terms like those appearing in NRTL, UNIQUAC and UNIFAC provide a flexibility, which
cannot be accounted for by the single-parameter van Laar term of Flory-Huggins. However,
the highly pronounced free-volume effects should always be accounted for in polymer
solutions.
The concept of free-volume (FV) is rather loose, but still very important. Elbro8
demonstrated, using a simple definition for the free-volume (Eq.13), that the FV percentages
of solvents and polymers are different. In the typical case, the FV percentage of solvents is
greater (40-50%) than that of polymers (30-40%). There are two exceptions to this rule; water
and PDMS: water has lower free-volume than other solvents and closer to that of most of the
polymers, while PDMS has quite a higher free-volume percentage, closer to that of most
154

solvents. LCST is, as expected, related to the free-volume differences between polymers and
solvents. As shown by Elbro8, the larger the free volume differences the lower the LCST
value (the larger the area of immiscibility). For this reason, PDMS solutions have a LCST,
which are located at very high temperatures.
Many mathematical expressions have been proposed for the FV. One of the simplest and
successful equations is20'21:

Vf=V-V'=V-Vw (13)

originally proposed by Bondi20 and later adopted by Elbro et al.21 and Kontogeorgis et al.22
in the so-called Entropic-FV model (described in 7.3.3). According to this equation, FV is
just the 'empty' volume available to the molecule when the molecules' own (hard-core or
closed-packed V*) volume is substracted.
The free-volume is not the only concept, which is loosely defined in this discussion. Even
the hard-core volume is a quantity difficult to define and various approximations are
available. Elbro et al.21 suggested using V*=VW, i.e. equal to the van der Waals volume (Vw),
which is obtained from the group increments of Bondi and is tabulated for almost all existing
groups in the UN1FAC tables. Other investigators23'24 interpreted somewhat differently the
physical meaning of the hard-core volume in the development of improved free-volume
expressions for polymer solutions, which employ Eq. 13 as basis, but with V values higher
than Vw. Table 6 shows that, due to the closed packed structure of molecules, a higher value
of the hard-core volume would have been expected e.g. around 1.2-1.3 Vw. Indeed,
investigations for athermal polymer systems (without any energetic interactions) demonstrate
that the optimum results with Entropic-FV (discussed below) and for both the solvent and
polymer activities are obtained when V*=1.2Vw (Fig.4). This observation regarding the
magnitude of hard-core volume related to Vw has helped not only in developments of the
Entropic-FV model, but as shown in Figure 5, also in understanding problems of hard-core
volume theories such as the one proposed by Guggenheim. This particular hard-core volume
theory has been often used in models for estimating diffusion coefficients for polymeric
systems.

Table 6. Values of Packing Density and of the ratio V'/Vw=Vo/Vw for various packing of
fluids, as well as for various fluid families23.

Structure/Compound Packing Density (po) v7v w =v«/v w


Open-packed cubic structure of spheres 0.52 1.92
Closed-packed cubic structure of spheres 0.74 1.35
Open-packed arays of infinite cylinders 0.78 1.27
Close-packed arays of infinite cylinders 0.90 1.11
Polyethylene 0.76 1.31
Most organic compounds 0.7....0.78 1.43 .... 1.28
Random densely packed mixture of up to 0.8 down to 1.25
spheres with log normal size distribution
155

Figure 4. Percentage deviation between experimental and calculated solvent infinite dilution
activity coefficients, versus the a-parameter in the free-volume expression of the Entropic-FV
model (Vf=V-aVw). From Kouskoumvekaki et al.23.

Figure 5. Plot of the ratio V /Vw calculated from Guggenheim's hard-core volume equation
(V* = 0.286Vc) as a function of the van der Waals volume Vw for n-alkanes. The critical
volume (Vc) is obtained from two different sets of experimental data, those by Teja and those
by Steele. Only those data by Teja have been verified by independent investigations based on
molecular simulation. The plot shows that, using these "correct" data, the V /Vw plot does not
follow the physically expected trend. (Modified from Kontogeorgis et al.24).

The original UNIFAC model does not account for the free-volume differences between
solvents and polymers and, as a consequence of that, it highly underestimates the solvent
activities in polymer solutions2122'25. On the other hand, the various modified UNIFAC
156

versions (Lyngby and Dortmund, see chapter 4), which use exponential segment fractions,
are also inadequate for polymer solutions. Although, their combinatorial terms are more
satisfactory for alkane systems, they fail completely for polymer-solvent systems and as
shown26 they significantly and systematically overestimate the solvent activities. Although
these UNIFAC models are not adequate for polymer solutions, the problem seems, however,
to lie more in the combinatorial term rather than the residual (energetic) term. In other words,
improvements are required especially for describing the free-volume effects, which are
dominant in polymer solutions.

7.3.2 The UNIFAC - FV model

Various modifications - extensions of the classical UNIFAC approach to polymers have been
proposed. All these approaches attempt to include the FV effects, which are neglected in the
UNIFAC combinatorial term. All of them employ the energetic (residual) term of UNIFAC.
The most well-known is the UNIFAC-FV model by Oishi and Prausnitz25:

lny, =lny™m(J + l n y r +W" (14)

The combinatorial and residual terms are obtained from original UNIFAC. An additional
term is added for the free-volume effects. An approximation but at the same time an
interesting feature of UNIFAC-FV, and other models of this type, is that the same UNIFAC
group-interaction parameters - i.e. those of original UNIFAC- are used. No parameter re-
estimation is performed. The FV term used in UNIFAC-FV has a theoretical origin and is
based on the Flory equation of state:

lnyr=3c,.ln[li^)j-C,{f^-lTl-^]1 (15)
where the reduced volumes are defined as:

bVhw
(16)
- = wyl+w7y1
m
b{WlVLw+w2Vlw)

In Eq.16, the volumes Vj and the van der Waals volumes are all expressed in cnrVmol. Wj is
the weight fraction.
In the UNIFAC-FV model as suggested by Oishi and Prausnitz25 the parameters Cj (3CJ is
the number of external degrees of freedom) and b are set to constant values for all polymers
and solvents (CJ=1.1 and b=1.28). The performance is rather satisfactory, as shown by many
investigators, for a large variety of polymer-solvent systems. Some researchers have
suggested that, in some cases, better agreement is obtained when these parameters are fitted
to experimental data27.
157

Originally, the UNIFAC-FV model was developed for solvent activities in polymers. It
could be expected that the model (Eqs.14-16) is also valid for estimating polymer activities.
However, such an application of UNIFAC-FV is rather problematic28. It has been shown23
that the performance of UNIFAC-FV in predicting the activities of heavy alkanes in shorter
ones is not very good. Such problems limit the applicability of UNIFAC-FV to cases where
the polymer activity is also of importance such as liquid-liquid equilibria for polymer
solutions. Indeed, to our knowledge, UNIFAC-FV has not been applied to polymer-solvent
LLE.

7.3.3 The Entropic-FV model

A similar but somewhat simpler approach to UNIFAC-FV, which can be readily extended to
multicomponent systems and liquid-liquid equilibria, is the so-called Entropic-FV model
proposed by Elbro et al.2i and Kontogeorgis et al.22:

\nli=\n1';omb-fv+\nyrjes

fv fv
lnyfo^=ln^ + l-^
*/ */
(1?)
fv _ xMjv ><i{Vi-Vwi)

i i

lny|"es -^UNIFAC (chapA)

As can been seen from Eq. 17, the free-volume definition given by Eq.13, is employed. The
combinatorial term of Eq. 17 is very similar to that of Flory-Huggins. However, instead of
volume or segment fractions, free-volume fractions are used. In this way, both combinatorial
and free-volume effects are combined into a single expression. The combinatorial - FV
expression of the Entropic-FV model is derived from Statistical Mechanics, using a suitable
form of the generalised van der Waals partition function.
The residual term of Entropic-FV is taken by the so-called 'new or linear UNIFAC model,
which uses a linear-dependent parameter table29:

amn=amnJ +amn2(T-T0) (18)

This parameter table has been developed using the combinatorial term of the original
UNIFAC model. As with UNIFAC-FV, no parameter re-estimation has been performed. The
same group parameters are used in the "linear-UNIFAC" and in the Entropic-FV models.
A common feature for both UNIFAC-FV and Entropic-FV is that they require the
volumes of solvents and polymers (at the different temperatures where application is
required). This can be a problem in those cases where the densities are not available
experimentally and have to be estimated using a predictive group-contribution or other
158

method, e.g. GCVOL30'31 or van Krevelen methods. These two estimation methods perform
quite well and often similarly even for low molecular weight compounds or oligomers such
as plasticizers, as shown in figure 6 for the family of phthalates.
Both UNIFAC-FV and Entropic-FV, especially the former, are rather sensitive to the
density values used for the calculations of solvent activities.

Figure 6. Volumes of different phthalates calculated by GC-VOL and van Krevelen,


compared to the experimental volumes taken from Ellington32 at 293.15 K (From Tihic10)

7.3.4 Results and Discussion

The UNIFAC-FV and Entropic-FV models have been widely applied to polymer solutions
and some typical applications are shown in tables 7-9, figures 7-12 and discussed in this
section, and figures 1 -2 and tables 2-4 in the previous sections. Some of the most recent
applications are included, while attention is paid to the advantages and shortcomings of the
models. In some cases, comparisons with two group-contribution equations of state (GC-
Flory and GCLF) are presented.

Vapor-Liquid Equilibria

Both models have been extensively applied to vapour-liquid equilibria (VLE) - solvent
activities in polymers and other size-asymmetric systems, including infinite dilution
conditions for binary polymer solutions22'27'33, VLE for co-polymer/solvent systems34, solvent
activities in dendrimer solutions35, VLE for a large variety of polar and hydrogen-bonding
systems36'37, VLE for paint-related polymer solutions including commercial epoxy resins37'11,
and recently also for VLE of ternary polymer-mixed solvent systems38.
159

Table 7 shows the performance of the models for infinite dilution activity coefficients for
some polyisoprene (PIP) solutions, while comparisons for complex systems (from a recent
comparative study, Lindvig et al.37) are shown in Table 8. Finally, results for some ternary
polymer-mixed solvent systems are shown in Table 9.

Table 7. Prediction of infinite dilution activity coefficients for PIP systems with three predictive
group contribution models. Experimental values and calculations are at 328.2 K

PIP systems Exper. value Entropic-FV GC-Flory UNIFAC-FV


+acetonitrile 68.6 47.7(31%) Ng 52.3 (24%)
+acetic acid 37.9 33.5 (12%) 50.0 (32%) 17.7(53%)
+cyclohexanone 7.32 5.4 (27%) Ng 4.6 (38%)
+acetone 17.3 15.9(8%) 10.5 (39%) 13.4 (23%)
+MEK 11.4 12.1 (6%) 7.5 (35%) 10.1 (12%)
+benzene 4.37 4.5 (2.5%) 2.8 (37%) 4.4 (0 %)
+1,2 dichloroethane 4.25 5.5 (29%) 6.6 (55%) 6.5 (54%)
+CC14 1.77 2.1 (20%) 1.6(11% 1.8(0%)
+1,4 dioxane 6.08 6.3 (4%) Ng 5.9 (2%)
+tetrahydrofurane 4.38 4.9(14%) Ng 3.9 (10%)
+ethylacetate 7.47 7.3 (2 %) 4.4(41%) 6.6(11%)
+n-hexane 6.36 5.1 (20%) 3.8 (39%) 4.6 (27%)
+chloroform 2.13 3.00(41%) 2.6 (24 %) 2.6 (20%)
ng: no groups available
Table 8. Percentage deviation between calculated and experimental solvent activity
coefficients from various thermodynamic models

Ntnpdarsdverts Ftiar adverts r-J/dqpi bcrdng schrais

FCLMVBR BV UFV QCF UNFA: EV LFV CCF UNRflC BV LFV QCF LNFyOC
FCL 24 99 14.1 224 ***** ***** ***** ***** ***** ***** ***** *****
***** ***** ***** ***** ***** *****
FDVB 169 17.2 17.8 ***** ***** *****
FBZE 53 25 425 31 154 632 324.8 628 ***** ***** ***** *****
FED 53 63 4.4 11.3 458 34.2 337 358 ***** ***** ***** *****
RVM\ 183 183 25 224 7.6 7.6 88 121 ***** ***** ***** *****
FKIE 462 466 433 47.7 51.1 44.3 34.7 450 ***** ***** ***** *****
***** *****
FTO 24 23 1.0 7.2 ***** ***** 28 1.4 31 23
FVA; 35 4.0 63 131 36 31 134 20.8 64 86 240 11.7
FWE 7.4 Q8 56 96 366 11.0 9.6 190 ***** ***** ***** *****
A/.%da/. 137 14.1 7.2 163 222 21.7 51.5 27.9 35 28 7.3 4.2
160

Table 9a: Average logarithmic deviations (xlOO) between experimental and predicted vapor
phase mole fractions for some ternary polymer- mixed solvent systems. The color indications
divide the deviations into the following groups: Grey: less than 20 %. Light grey: 20 - 50 %.
Dark grey: above 50 %.

Polymer Solvent SAFT FFV/UQ FH Pa-Vc FFV UFV GCI.F FH/Ha


PMMA Butanone 49 47 48 52 51 45 37
Toluene 222 153 154 196 194 137 89
PS Benzene 23 22 24 23 14 14 23 22
Toluene 40 37 41 39 20 20 37 36
PS Toluene 6 4 4 5
Ethylbenzene 14 14 13 12 15
PS Toluene 18 9 17 8 9 9 14 17
Cvclohexanc 9 3 8 2 3 4 5 8
PS Chloroform 33 33 31 23 34 27 34 27
Carbontetrachloride 37 37 34 22 37 27 37 28
PS. 7 = 373.15 K Styrcne 23 25 27 30 26
Ethylbenzene 41 45 51 60 | 48
PS, 7 = 393.15 K Styrcne 17 20 21 23 20
Ethylbenzene 26 32 34 39 32
PS. T = 413.15K Styrene 9 10 11 12 10
Ethylbenzene 13 15 17 21 16

Table 9b: Average logarithmic deviations (xlOO) between experimental and predicted
pressures for some ternary polymer-mixed solvent systems. Color indications as in Table 9a.

System SAFT EFV/UC. FH Ri-Ve EFV UFV GCLF FFPFfe


PMM4-butanone-tolLme 16 15 16 16 | 36 15 17
PS-benzHie-toluene 7 4 12 7 8 14 14 7
PS-toluene-ethylbenzene 2 1 8 2
9z, 9z,
PS-toliHK-cyclohsxare 14 8 2 11
PS-cHcrofaTrKarbortetrachloride 14 16 11 4 17 19 5 52
PS-styrene-etlTylbenzere, T=373.15 K 32 31 34 31 33
PS-styrene-ethylberEerE, T=393.15K 32 30 35 30 33
PS-styrene-ethylbenzerB, r=413.15K 32 28 34 26 32

Figures 7 and 8 also show recent results 35 , demonstrating the performance of the models for
dendrimer solutions, including the sensitivity of the calculations to the density value
employed.
161

Figure 7. Experimental and predicted activities of methanol in the dendrimer PANAM-G2


with the UNIFAC-FV and the Entropic-FV models35. Results are shown using experimental
and predicted densities (Reprinted with permission).

Figure 8. Experimental and predicted activities of acetone in the dendrimer A4 with the
UNIFAC-FV and Entropic-FV models33. Results are shown using experimental and predicted
densities (Reprinted with permission).

Overall, we can conclude that the Entropic-FV and UNIFAC-FV models, especially the
former, provide satisfactory predictions of solvent activities, even at infinite dilution, for
complex polar and hydrogen bonding systems including solutions of interest to paints and
coatings, and rather satisfactory predictions when mixed solvents are present. Some more
specific comments can be made from comparative investigations for different types of
systems.
162

Athermal systems
The articles cited in this section include investigations, which compare the performance of
the UNIFAC-FV, Entropic-FV and several more recent free-volume equations for athermal
systems. In such systems the energetic effects are zero or very small and they can be thus
used for testing the combinatorial and free-volume terms of the models. Although the
database used in the various investigations is not always the same, it typically consists of
solutions having components differing significantly in size but which do not exhibit energetic
interactions. Examples of these nearly athermal systems are solutions of polyethylene and
polyisobutylene with alkanes (only solvent activities are available), alkane solutions (where
both the activity of light and heavy-chain alkanes are available), polystyrene/ethylbenzene,
polyvinyl acetate/vinyl acetate as well as "pseudo" experimental data for polymer activities
generated with molecular simulation techniques.
In their recent review, Pappa et a/.33, considered over 200 experimental datapoints for
athermal polymer solutions at intermediate concentrations and about 100 points at infinite
dilution and compared the Entropic-FV and Zhong-Masuoka39 models. They found that the
Entropic-FV formula yields lower error than the Zhong term, though the latter does not
contain any volume terms (9% vs. 16%). Other literature comparisons40 also agree that the
free-volume models with the volume-containing terms perform better than those models
requiring no volume information. Thus, Entropic-FV, Flory-FV and related models provide a
good basis for building a full thermodynamic model for polymers. UNIFAC-FV seems to
offer no advantage over the simpler approaches and seem to be more sensitive1122'37 to the
volume values employed compared to simpler free-volume equations.
Despite the overall successful performance of Entropic-FV and UNIFAC-FV models for a
large number of systems and types of phase equilibria, it has been shown over the last years
by a number of researchers23'40"43, that the combinatorial/free-volume terms of both the
Entropic-FV and UNIFAC-FV models have a number of deficiencies:

i. The solvent activities in athermal polymer solutions are systematically


underestimated by, often, 10% (in the case of Entropic-FV) or more (for UNIFAC-
FV). For athermal systems, the residual term is zero. Such an underestimation
cannot be entirely attributed to the small interaction effects present in such systems.
ii. The activities of heavy alkanes in short-chain ones, available from SLE
measurements, are in significant error, especially as the size difference increases.
Due to the lack of experimental data on polymer activities, such SLE data can help
test the models' applicability for the activities of heavy-compounds.
iii. The performance of the models is rather sensitive to the values used for the polymer
density.

Numerous investigations and developments of new combinatorial/free-volume terms have


been reported over the last 5 years. The general conclusions that can be drawn are:

i. The activities of alkane solvents in either alkane or athermal polymer (PE, PIB)
solutions are very satisfactorily predicted (much better than with the Entropic-FV
formula) by some more recent modified free-volume equations e.g. Chain-FV, p-
FV and R-UNIFAC. However, these models cannot be extended to multicomponent
163

systems. This is a serious limitation for multicomponent systems. The Flory-FV


and a recently developed model23 do not suffer from this limitation.
ii. Volume-based models perform better than those not including volume-containing
terms.
iii. The UNIFAC-FV expression, the first free-volume equation proposed, which is
derived from the theory of Flory, is not as successful for athermal systems
compared to more recent simpler equations. This may be due to the values of the
parameters b and c employed in this model. Fitting these parameters may improve
the performance of the UNIFAC-FV term. The results with this model seem
particularly sensitive to the density values employed.
iv. All models perform clearly less satisfactorily for the activities of heavy alkanes in
short-chain ones, especially as the size-asymmetry increases. Models without free-
volume corrections such as UNIFAC, and Flory-Huggins are particularly poor in
these cases. Unfortunately, such activity coefficient measurements, which could
bee used for testing the performance of the models for the activities of polymers,
are scarce. Direct measurements for polymer activities have not been reported.
Molecular simulation studies can offer help in this direction43.

Non-polar and slightly polar systems


Numerous results (predictions and correlations) are available for such systems23'33'40"43. Many
models perform satisfactory even when pure predictions are considered. In a recent
comparison, Pappa et al.33 showed that Entropic-FV performs better than the Zhong-Masuoka
model (11% vs. 20%). In the review by Lee and Danner44, GCLF (Group-Contribution
Lattice Fluid Equation of State) and Entropic-FV perform similarly for non-polar systems
(15%), but GCLF appears to perform better for the weakly polar ones. This is attributed to
problems of the Entropic-FV model for systems containing polyacrylates and
polymethacrylates with acetates. UNIFAC-FV has an average error of 23% for these types of
systems and GC-Flory of 20%.

Water-soluble polymers and other hydrogen-bonding systems


Predictions have been provided for some hydrogen bonding systems with a number of
models. Pappa et a/.33 report an average deviation of 26% (in VLE) with both Entropic-FV
and Zhong-Masuoka models, which is higher than the deviations observed for non-polar and
polar systems. Lee and Danner's44 comparison revealed that Entropic-FV is the best model
for strongly polar solvents (23%), followed by GCLF (28%) and GC-Flory (31%). UNIFAC-
FV does not seem to be very successful for such complex systems (mean deviation 65%). In
some recent investigations ' , several of these well-known group contribution models
(Entropic-FV, UNIFAC-FV, GC-Flory) have been tested for VLE of paint-related systems.
These are systems of polyacetates, polyacrylates, polymethacrylates, epoxies and a variety of
solvents (non-polar, polar, hydrogen bonding, water). The performance of the models is
overall similar with the Entropic-FV and GC-Flory being overall better than UNIFAC-FV in
most situations, in agreement to the investigations reported earlier. Some results are shown in
Table 8.
164

Co-polymer solutions
Comparisons for co-polymer systems are not extensive, although several VLE data for
solvent/co-polymers are available. Bogdanic and Fredenslund34, Pappa et a/.33 and Lee and
Danner44 have presented comparisons for such systems, using the models Entropic-FV, GC-
Flory, Zhong-Masuoka, UNIFAC-FV and GCLF. In their comparison, Pappa et a/.33 found
that both Entropic-FV and Zhong-Masuoka models perform similarly for these systems with
a deviation around 20%. Similar overall performance for the Entropic-FV, GC-Flory and
UNIFAC-FV models was observed by Bogdanic and Fredenslund ', though the various
models perform different for specific co-polymer systems. For example, the Entropic-FV
model has problems in the presence of chloro-groups. GCLF is also shown to be quite
successful for a number of co-polymer solutions in mostly non-polar/slightly polar solvents.

Polymer-mixed solvent systems


In a recent investigation38, a database for ternary VLE systems (polymer-mixed solvents) has
been compiled and used for evaluating the performance of several group contribution models
(Entropic-FV, UNIFAC-FV, and GCLF). The performance of these predictive models,
though inferior to the binary systems, can be considered quite satisfactory, considering also
the experimental uncertainties involved in these measurements. The experimental
measurements of solvent activities in mixed solvent/polymer systems are not easy and may
be often associated with significant errors (J.M.Prausnitz, 2000. Personal Communication).

Liquid-Liquid and Solid-Liquid Equilibria

The application of free-volume models to liquid-liquid and solid-liquid equilibria of polymer


solutions is much more limited compared to VLE, and only Entropic-FV has been widely
used in such cases, and for both polymer solutions and blends. Entropic-FV has been applied
to SLE of asymmetric alkane solutions , LLE for binary polymer solutions and polymer
blends46, SLLE for semicrystalline polymer/solvents47, as well as for LLE for ternary
polymer-solvent-solvent (and solvent-antisolvent) systems48.
Figures 9-12 show typical LLE results for binary and ternary polymer-solvent solutions
with Entropic-FV. Entropic-FV can be readily applied to liquid-liquid and solid-liquid
equilibria and can predict all types of phase diagrams present in polymeric systems (UCST,
LCST, hourglass-type) e.g. the results in figure 9 for PS/acetone. However, the results are of
qualitative than of quantitative value in most cases. A difference of 10-30 degrees should be
expected in the predictions. The performance of Entropic-FV seems rather system-specific,
e.g. for polyethylene/octylphenols the difference in UCST is 5-10 °C while for
polyethylene/octanol it is approximately 40 °C. In the cases of polymer blends46, where free-
volume effects are not very important, the model deviates substantially from experimental
data, although it can predict the UCST-type behavior. Compared to the other free-volume
models, Entropic-FV may be considered as the most successful and widely used extension of
UNIFAC to polymers.
Besides binary polymer-solvent LLE, Entropic-FV has been also applied to ternary
polymer-solvent LLE48, and compared to the Holten-Andersen et al. equation of state (a
165

previous version of the GC-Flory EoS). The polymer/mixed solvent systems considered
include both two solvent-solvent and solvent-anti-solvent systems. The comparison was
limited to eight PS-two solvent systems (benzene/acetone, benzene/methanol,
methylcyclohexane/acetone, toluene/acetone, MEK/acetone, ethyl acetate/acetone,
NNDMF/cyclohexane) and one PMMA system (with chlorobutane/butanol-2) for which full
data are available. Qualitatively good results are obtained with both models, especially
Entropic-FV (despite the fact that all group interaction parameters were based on low
pressure VLE of non-polymeric systems). Typical results are shown in figures 11 and 12.
Finally, in one of the very few works reported on the prediction of solid-liquid-liquid
equilibria47 for polymer solutions, the Entropic-FV and UNIFAC models have been shown to
yield similar results for SLLE.

Figure 9. Correlation of PS/acetone LLE with Entropic/FV model (using GCVOL for the
density of the polymer)45. The numbers (4800, 10300, 19800) correspond to the molecular
weight of the polymer. (Reprinted with permission)
166

Figure 10. PS/cyclohexane LLE prediction with various predictive group contribution
models43 (Reprinted with permission).

Figure 11. Ternary LLE for PS(300000)/benzene/methanol at T=298.15 K48. (Reprinted with
permission)
167

Figure 12. Ternary LLE for PS(300000)/methyl cyclohexane/acetone, T=298.15 K48.


(Reprinted with permission)

7.3.5 The Entropic-FV/UNIQUAC model

Both UNIFAC-FV and Entropic-FV are group contribution models. This renders the models
truly predictive, but at the same time with very little flexibility if the performance of the
models for specific cases is not satisfactory. An interesting alternative approach is to employ
the UNIQUAC expression for the residual term. This Entropic-FV/UNIQUAC model has
been originally suggested by Elbro et a/.8'17'21 and has shown to give very good results for
polymer solutions if the parameters are obtained from VLE data between the solvent and the
low molecular weight monomer (or the polymer's repeating unit).
The Entropic-FV/UNIQUAC model has been recently further developed and extended
independently by two research groups49"51. Both VLE and LLE equilibria are considered but
the emphasis is given to LLE. Very satisfactory results are obtained as can be seen for two
typical systems in figures 13 and 14. It has been demonstrated that the Entropic-
FV/UNIQUAC approach can correlate both UCST/LCST and closed loop behavior49'50 and
even show the pressure dependency of critical solution temperatures (UCST and LCST)31.

7.3.6 Extension to Semi-crystalline Polymers and Swelling

When highly crystalline or cross-linked polymers are considered, e.g paints after drying,
rubbers, polyolefins, the effects of cross-linking and crystallinity should be considered
because they affect the solubility. Cross-linking and crystallinity are often visualized as
'similar' (in some sense) phenomena and are described with the same theories: crystalline
regions are assumed to act as 'physical or giant cross-links'.
168

Figure 13. Correlation of LLE for PBMA/MEK system49. • Exp.data (Mw=200000 g/mol);
correlation

Figure 14. Correlation and prediction of LLE for HDPE/1-dodecanol system49.


A Exp.data (Mw=60700 g/mol), correlation
o Exp.data (Mw=77800 g/mol), prediction
• Exp.data (Mw=21300 g/mol), — — prediction
• Exp.data (Mw=6800 g/mol), prediction
169

Crystalline and cross-linked polymers do not dissolve (with a few exceptions) in solvents but
only swell. Swelling equilibria is thus important. To account for the crystalline/cross-linking
effect, an additional factor (elastic term) is typically required in thermodynamic models. Two
popular theories to account for this effect are the Flory-Rechner12:

lnaf=^%>'3 (19)
M
c
where:
p a is the density of the amorphous polymer
Vi is the molar volume of the solvent
Mc is the molecular weight between cross-links

and the Michaels-Hausslein53 equation:

p CT;
, . £h' -(K)]-»»
In a, =— F—T r-^S (20)
[3/(2ftp2)-l]
where:
Tm is the melting point temperature
f is the fraction of elastically effective chains in amorphous regions
As observed from these equations, both theories introduce at least one extra parameter,
which needs to be determined from experimental data: the molecular weight between cross-
links Mc and the fraction of elastically effective chains f. They have been combined with
free-volume models ' '"" and they have been applied to semicrystalline polymer/solvent
systems. The results are satisfactory but they are not predictive: the Mc and f parameters
should be estimated from experimental data. However, the swelling of cross-linked polymers
can be estimated with such equations.
7.3.7 Extension of Free-Volume Models to Gas Solubilities in Elastomers

Thorlaksen et al.56 have recently combined the Entropic-FV term with Hildebrand's regular
solution theory and developed a model for estimating gas solubilities in elastomers. The so-
called Hildebrand - Entropic FV model is given by the equation:

lny,=lny? + lny™'v (21)

'* R-T

lny^' = t n ^2_ + 1_^2_ ( 2 2 )

X, X7
170

where :
81 = solvent solubility parameter, 82= gas solubility parameter
X2 = gas mole fraction in liquid/polymer.
02 is the 'apparent' volume fraction of solvent, given by:

x2 -V'-
2
O,=
r -V1' + x • VL

and O2Ff is the 'free-volume' fraction given by:

2
~x2-(y<--vrhx]-{v<--vr)
Vj is a hypothetical liquid volume of the (gaseous) solute.

; f< VJ--[Sl-5iy-Oii
— = ^-exp — '- (23)

A £

f2 is the fugacity of the gas and f'2 is the fugacity of the hypothetical liquid, which can be
estimated from the equation:

f1 4 74547
ln^- = 3.54811—: + 1.60151-T -0.87466 -T2 +0.10971 -T3 (24)
Pc Tr
Finally, the gas solubility in the polymer is estimated from the equation:

; f> (vl--{5,-S
2
2y-O] , Of , Of)
— = l^.exp\ ^ 1 11 L + in^^ + i i_\ (25)
r *8 I R-T x x \
J 2

Calculations showed that the hypothetical gas "liquid" volumes are largely independent to
the polymer used, and moreover, for many gases (H2O, O2, N2, CO2 and C2H2) these are
related to the critical volume of the gas by the equation:

V!,- = 1.776Vc -86.017 (26)

Very satisfactory results are obtained as shown in Table 10 and Figure 15.
171

Table 10. Summary of the performance of the models tested at T = 298 K; P = 101.3 kPa
Errors associated with models for predicting gas solubilities in polymers5 .

Hildebrand- Hildebrand-
Hildebrand/
Polymer Gas Michaels/Bixler Tseng/Lloyd Entropic FV Entropic FV -
-1 2

PIP N2 14.7 73 3.9 -7.9 -4.6


o2 -16.1 -4 14 10.8 11.8
Ar -32.5 -23 - 29.4 -22.2
CO, -3.2 13 4.5 8.7 4.6
PIB N2 -2.5 - 6.8 3.1 5.0
o2 -6.1 - -1.7 -8.3 1.7
Ar - - - - -
CO2 32.8 - -1.9 41.1 35.2
PBD N2 22.3 - 8.1 8.1 12.6
o2 14.9 - -6 8.7 10.8
Ar 12.1 - - 111.1 24.0
CO, -9.7 - -4.6 0.4 -4.0
PDMB N2 - - -23 -7.5 -3.1
o2 - - -32 -16.8 -15.9
Ar - - - - -
CO, - - -24 2.3 -2.2
PCP N2 58.1 - 49 -7.0 -4.2
o2 43.7 - 60 -1.4 -1.4
Ar - - - - -
CO, 8.8 - 27 -13.3 -17.1
AAD: 19.8 28 18 16.8 10.6

Hildebrand Entropic-FV-1: The liquid volume of the gas is determined from its relationship
with the critical volume, Eq. (26).
Hildebrand Entropic-FV-2 : The average hypothetical liquid volume of a gas is used
172

Figure 15. The solubility of several gases in PIP as a function of temperature predicted by the
Hildebrand Entropic-FV model.

7.4 EQUATIONS OF STATE FOR POLYMERS

Many equations of state (EoS) have been proposed for polymers, for a recent review see
Kontogeorgis . Both cubic equations of state (vdW, SRK, PR, SWP) and non-cubic ones
esp. GCLF and SAFT have been applied with success. In many of the cubic EoS, mixing
rules based on the EoS/GE principle have been employed using one of the previously
described activity coefficient models (FH, E-FV, U-FV) in the mixing rules. In this case, the
behavior of the activity coefficient model at low pressures is reproduced. In other cases
(cubic EoS, SAFT) a single interaction parameter is used or group-based parameters (GCLF)
are employed.
173

Numerous applications of these EoS have been reported: low and high-pressure VLE,
Henry's law constants and LLE for systems including polymers, co-polymers and blends.
Examples of what is possible with such EoS are shown in figures 16-20: LLE for polymer
solutions with a cubic EoS58 and VLE and LLE for binary and ternary systems with a
recently developed modified SAFT EoS59"61.
This is still a very active open area of research and is difficult to recommend a specific
approach. A serious problem with all EoS for polymers which, in our view, has not been
adequately addressed as yet is the way to get the EoS parameters for polymers. Methods
employed for low molecular weight compounds (see chapters 5 and 6) cannot be used since
critical properties and vapour pressure data are not available (have no meaning) for polymers.
Numerous indirect methods57 have been employed using volumetric data and additional
information, often phase equilibria data for mixtures of polymers with low molecular weight
compounds. Such methods may be necessary since use of volumetric data alone do not seem
to provide polymer EoS parameters useful for phase equilibrium calculations. Use of phase
equilibria data, on the other hand, may render the parameters of pure polymers sensitive to
the type of information employed. A thorough investigation on methods to obtain meaningful
polymer parameters for equations of state will significantly improve and enhance the
applicability of equations of state for polymers. A first effort towards this direction has been
recently reported for the simplified PC-SAFT equation of state62.

Figure 16. Correlated UCST phase diagrams with the van der Waals equation of state for
PS/cyclohexane at various molecular weights38. A single interaction parameter is used (per
molecular weight).
174

Figure 17. Predicted and correlated UCST phase diagrams with the van der Waals equation of
state for PBMA(11600)/n-pentane:>8. A single interaction parameter is used. The predicted
interaction parameter is based on a simple correlation with the solvent's molecular weight.

Figure 18. VLE prediction with the PC-SAFT equation of state for PVAC(170000)-acetone at
three temperatures (the interaction parameter is set equal to zero). The PC-SAFT developed
by von Solms et a/.59 is employed.
175

Figure 19. LLE with the PC-SAFT equation of state for PS-methylcyclohexane using a single
interaction parameter.

Figure 20. Ternary LLE with the PC-SAFT equation of state for the ternary system
PS(300000)-acetone-methylcyclohexane. The binary parameters are regressed from the
binary systems.
176

LIST OF ABBREVIATIONS

BR Butadiene Rubber
CST Critical Solution Temperature
DBP dibutyl phthalate
DDP didecyl phthalate
DEP diethyl phthalate
DHP dihexyl phthalate
DMP dimethyl phthalate
DOA dioctyl adipate
DOP dioctyl phthalate
DOS dio(2-ethylhexyl) sebacate
EoS Equation of State
Exper./exp. Experimental
EFV Entropic-FV
FH Flory-Huggins (model/equation/interaction parameter)
FV free-volume
GC Group Contribution (method/principle)
GC-Flory Group Contribution Flory Equation of State
GCF Group Contribution Flory Equation of State
GCLF Group Contribution Lattice Fluid
GCVOL Group Contribution Volume (method for estimating the density)
HA Holten-Andersen equation of state
Ha Hansen solubility parameters
HDPE High Density polyethylene
LCST Lower Critical Solution Temperature
LLE Liquid-Liquid Equilibria
MEK methyl ethyl ketone
MW molecular weight
NRTL non-random two liquid activity coefficient model
Pa-Ve Panayiotou-Vera equation of state
PBD Polybutadiene
PDMB Polydimethylbutadiene
PBMA Polybutyl methacrylate
PDMS Polydimethylsiloxane
PE Polyethylene
PEO Polyethylene oxide
PIB Polyisobutylene
PIP Polyisoprene
PMMA Polymethyl methacrylate
PR Peng-Robinsoon equation of state
Pred. predicted
PS Polystyrene
PVAC Polyvinyl acetate
PVC Polyvinyl chloride
177

SCF supercritical fluid


SF Schultz-Flory plot
SGE solid-gas equilibria
SRK Soave Redlich Kwong equation of state
S WP Sako Wu Prausnitz equation of state
SLE Solid-liquid equilibria
SLLE Solid-liquid-liquid equilibria
UCST Upper Critical Solution Temperature
U-FV UNIFAC-FV
UNIFAC Universal Functional Activity Coefficient
UQ Uniquac
vdW van der Waals equation of state
vdWl f van der Waals one fluid (mixing rules)
VLE Vapor-Liquid Equilibria
VOC Volatile Organic Content

REFERENCES

1. R. B. Seymour, Plastics vs. Corrosives, SPE Monograph Series, Wiley, 1982


2. Van Krevelen, Properties of polymers. Their correlation with chemical structure; their
numerical estimation and prediction from additive group contributions, Elsevier,
1990.
3. A.F.M. Barton, Handbook of solubility parameters and other cohesion parameters,
CRC Press, 1983.
4. A.F.M. Barton, CRC Handbook of polymer-liquid interaction parameters and
solubility parameters, CRC Press, Boca Barton, FL, 1990.
5. J. Bentley and G.P.A. Turner, Introduction to Paint Chemistry and Principles of Paint
Technology, 4th edition, Chapman and Hall, 1998.
6. CM. Hansen, Hansen Solubility Parameters. A User's Handbook, CRC Press, 2000.
7. J. Holten-Andersen and K. Eng, Progress in Organic Coatings, 16 (1988) 77.
8. H.S. Elbro, Phase Equilibria of polymer solutions - with special emphasis on free
volumes, Ph.D Thesis. Department of Chemical Engineering, Technical University of
Denmark, 1992.
9. J. Klein and H.E. Jeberien, Makromol. Chem., 181 (1980) 1237.
10. A. Tihic, Investigation of the miscibility of plasticizers in PVC, B. Sc. Thesis.
Department of Chemical Engineering, Technical University of Denmark, 2003.
11. Th. Lindvig, M.L. Michelsen and G.M. Kontogeorgis, AIChE J., 47(11) (2001) 2573.
12. P.J. Flory, J. Chem. Phys, 9 (1941) 660.
13. D. C. H. Bigg, J. Appl. Polym. Sci., 19 (1975) 3119.
14. P.M. Doty and H.S. Zable, J. Polymer Sci, 1 (1946) 90.
15. C. Qian, S.J. Mumby and B.E. Eichinger, Macromolecules, 24 (1991) 1655.
16. Y.C. Bae, J.J. Shim, D.S. Soane and J.M. Prausnitz, J. Appl. Polym. Science, 47
(1993)1193.
17. J.M. Prausnitz, R.N. Lichtenthaler and E.G.D. Azevedo, Molecular thermodynamics
of Fluid Phase Equilibria. Prentice-Hall International. 3rd Edition, 1999.
178

18. Th. Lindvig, M.L. Michelsen and G.M. Kontogeorgis, Fluid Phase Equilibria, 203
(2002) 247.
19. A.R. Schultz and P.J. Flory, J. Amer. Chem. Soc, 75 (1953) 496.
20. A. Bondi, Physical Properties of Molecular Crystals, Liquids and Glasses; John Wiley
and Sons: New York, 1968
21. H.S. Elbro, Aa. Fredenslund and P. Rasmussen, Macromolecules, 23 (1990) 4707.
22. G.M. Kontogeorgis, Aa. Fredenslund and D.P. Tassios, Ind. Eng. Chem. Res., 32
(1993)362.
23.1. Kouskoumvekaki, M.L. Michelsen and G.M. Kontogeorgis, Fluid Phase Equilibria,
202(2) (2002) 325.
24. G.M. Kontogeorgis, LA. Kouskoumvekaki and M.L. Michelsen, Ind. Eng. Chem.
Res., 41(18) (2002) 4848.
25. T. Oishi and J.M. Prausnitz, Ind. Eng. Chem. Process Des. Dev., 17(3) (1978) 333.
26. G.M. Kontogeorgis, Ph. Coutsikos, D.P. Tassios and Aa. Fredenslund, Fluid Phase
Equilibria, 92(1994)35.
27. J.R. Fried, J.S. Jiang and E. Yeh, Comput. Polymer Science, 2 (1992) 95.
28. L.A. Belfiore, A.A. Patwardhan and T.G. Lenz, Ind. Eng. Chem. Res., 27 (1988) 284-
294.
29. H.K. Hansen, B. Coto and B. Kuhlmann, UNIFAC with lineary temperature-
dependent group-interaction parameters, IVC-SEP Internal Report 9212, 1992.
30. H. S. Elbro, Aa. Fredenslund and P. Rasmussen, Ind. Eng. Chem. Res., 30 (1991)
2576.
31. E.C. Ihmels and J. Gmehling, Ind. Eng. Chem. Res., 42(2) (2003) 408-412.
32. J.J. Ellington, J. Chem. Eng. Data, 44 (1999) 1414.
33. G.D. Pappa, E.C. Voutsas and D.P. Tassios, Ind. Eng. Chem. Res., 38 (1999) 4975.
34. G. Bogdanic and Aa. Fredenslund, Ind. Eng. Chem. Res., 34 (1995) 324.
35.1. Kouksoumvekaki, R. Giesen, M.L. Michelsen and G.M. Kontogeorgis, Ind. Eng.
Chem. Res., 41(19) (2002) 4848.
36. G.M. Kontogeorgis, Aa. Fredenslund, I.G. Economou and D.P. Tassios, AIChE J., 40
(1994)1711.
37. Th. Lindvig, L.L. Hestkja;r, A.F. Hansen, M.L. Michelsen and G.M. Kontogeorgis,
Fluid Phase Equilibria, 663 (2002) 194.
38. Th. Lindvig, I. G. Economou, R.P. Danner, M.L. Michelsen and G. M. Kontogeorgis,
Modelling of multicomponent vapor-liquid equilibria for polymer-solvent systems,
Fluid Phase Equilibria (in press).
39. C. Zhong, Y. Sato, H. Masuoka and X. Chen, Fluid Phase Equilibria, 123 (1996) 97.
40. E.N. Polyzou, P.M. Vlamos, G.M. Dimakos, I.V. Yakoumis and G.M. Kontogeorgis,
Ind. Eng. Chem. Res., 38 (1999) 316-323.
41. E.C.Voutsas, N. S. Kalospiros and D. P. Tassios, Fluid Phase Equilibria, 109 (1995)
1.
42. G.M. Kontogeorgis, G.I. Nikolopoulos, D.P. Tassios, D.P and Aa. Fredenslund, Fluid
Phase Equilibria, 127 (1997) 103.
43. G.M. Kontogeorgis, E.C. Voutsas and D.P. Tassios, Chem. Eng. ScL, 51 (1996) 3247.
44. B.-C Lee and R.P. Danner, AIChE J., 42 (1996) 837.
179

45. G.M. Kontogeorgis, A. Saraiva, Aa. Fredenslund and D.P. Tassios, Ind. Eng. Chem.
Res., 34 (1995) 1823.
46. V.I. Harismiadis, A.R.D. van Bergen, A. Saraiva, G.M. Kontogeorgis, Aa.
Fredenslund, Aa. and D.P. Tassios, AIChE J., 42 (1996) 3170.
47. V.I. Harismiadis and D.P. Tassios, Ind. Eng. Chem. Res., 35 (1996) 4667.
48. G.D. Pappa, G.M. Kontogeorgis and D.P. Tassios, Ind. Eng. Chem. Res., 36 (1997)
5461.
49. G. Bogdanic and J. Vidal, Fluid Phase Equilibria, 173 (2000) 241-252.
50. G. Bogdanic, Fluid Phase Equilibria, 4791 (2001) 1-9.
51. G.D. Pappa, E.C. Voutsas and D.P. Tassios, Ind. Eng. Chem. Res., 40 (2001) 4654.
52. P. J. Flory and J. Rechner, J. Chem. Phys., 11 (1943) 521.
53. M. J. Michaels and R.W. Hausslein, J. Polym. Sci. C , 10 (1965) 61.
54. J. S. Yoo, S. J. Kim and J. S. Choi, J. Chem. Eng. Data, 44 (1999) 16.
55. S. J. Doong and W. S. W. Ho, Ind. Eng. Chem. Res., 30 (1991) 1351.
56. P. Thorlaksen, J. Abildskov and G.M. Kontogeorgis, Fluid Phase Equilibria, 211
(2003) 17.
57. G. M. Kontogeorgis, Ch. 16: Thermodynamics of polymer solutions, in Handbook of
Surface and Colloid Chemistry, 2nd ed., CRC Press, 2003.
58. V.I. Harismiadis, G.M. Kontogeorgis, A. Saraiva, Aa. Fredenslund and D.P. Tassios,
Fluid Phase Equilibria, 100 (1994) 63-102.
59. N. v. Solms, M. L. Michelsen and G. M. Kontogeorgis, Ind. Eng. Chem. Res., 42(5)
(2003) 1098.
60.1.A.Kouskoumvekaki, N.v.Solms, M.L.Michelsen and G.M.Kontogeorgis, Fluid Phase
Equilibria, 215(1) (2004) 71-78.
61. Th.Lindvig, M.L.Michelsen and G.M. Kontogeorgis, Ind. Eng. Chem. Res, 43(4) (2004)
1125-1132.
62.1. A. Kouskoumvekaki, N. v. Solms, Th. Lindvig, M. L. Michelsen and G. M.
Kontogeorgis, A novel method for estimating pure-component parameters for polymers:
Application to the PC-SAFT equation of state, Ind. Eng. Chem. Res. (submitted).
This page is intentionally left blank

You might also like