You are on page 1of 11

Fluid Phase Equilibria 194–197 (2002) 1179–1189

Activity coefficients and diffusivities of solvents in polymers


Suojiang Zhang a,b,∗ , Akio Tsuboi b , Hiroaki Nakata b , Takeshi Ishikawa b
a
Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100080, PR China
b
Mitsubishi Chemical Corporation (MCC), DERC, Kurashiki 712-8054, Japan
Received 17 April 2001; accepted 8 November 2001

Abstract
Activity coefficients and diffusivities of small molecules in phenol resin were determined by using inverse gas
chromatography (IGC) technique with packed column. The Romdhane–Danner approach was used to measure
the diffusivities for solvents in phenol resin. Since the thickness of the polymer film on the support materials
has a significant effect on the value of diffusivity, it was determined by using a more accurate method based on
Braun–Guillet approach. The experimental data of activity coefficients and diffusivity at infinite dilution for three
systems have been determined. The PolyNRTL model was adopted to represent the measured activity coefficient
data. Also, the elution model of the packed columns and the estimations of model parameters were evaluated to
obtain the diffusivity. © 2002 Elsevier Science B.V. All rights reserved.
Keywords: Measurement; Model; Activity coefficient; Diffusion coefficient; Polymer; Solvent; Inverse gas chromatography

1. Introduction

Knowledge of equilibrium and mass transfer properties is of considerable importance in a wide variety
of manufacturing operations of polymers, such as polymerization, devolatilization, vacuum/gas stripping
and drying [1–3]. Accurate measurements of the activity coefficients and diffusivities for small molecules
in polymer melts or solutions have been a very difficult but important issue for many years. Conventional
methods for measuring them rely on bulk equilibration and gravimetric sorption/desorption experiments
[4]. These techniques, however, become very difficult to apply to the polymer–solvent systems when
the solvent is present in vanishingly small amounts, or at temperatures in the vicinity of or below the
glass transition temperature. Moreover, the experimental time may become very long when the value of
the diffusivity is small. On the other hand, the inverse gas chromatography (IGC) method is a fast and
reliable technique for measurement of the activity coefficients and diffusivities, particularly in the highly
polymer concentrated region. This region is critically interesting in the polymer industry, particularly in
the manufacturing of polymer films and coating/drying operations. Furthermore, the increasingly tighter

Corresponding author. Fax: +86-1-06256-1822.
E-mail addresses: zhangsj 0@yahoo.com, sjzhang@home.ipe.ac.cn (S. Zhang).

0378-3812/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 3 8 1 2 ( 0 1 ) 0 0 7 7 2 - 5
1180 S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189

environmental and health criteria regarding the volatile residuals in the polymer products have induced
the increased interests in the activity coefficients and diffusivities at infinite dilution for volatile solvents
in polymer melts. In fact, the activity coefficient and diffusivity data at infinite dilution provide the most
valuable information in terms of molecule–molecule interactions as compared with those in the finite
solvent concentration region. It has been demonstrated that the equilibrium and diffusion performance in
the finite concentration region could be reliably predicted via the polymer GE models and the free volume
theory [3] from the experimental data at infinite dilution alone.
In this work, the packed column IGC technique was adopted for determining the activity coefficients and
diffusivities in polymers. The key point in the measurement of activity coefficients in polymers is to make a
thin and uniform coating film of polymer on the support material and to ensure the equilibrium between the
gas and polymer phases to be fully achieved. Comparatively, a more complicated procedure is involved in
the determination of diffusivity. The retention time and peak area of small molecules must be determined
over a variety of flow rates at each temperature as well as over a variety of temperatures at a fixed flow rate.
The Romdhane–Danner approach [3] was applied here to the data analysis to obtain the diffusivities. Since
the thickness of the polymer film on the stationary phase has a significant effect on the measured accuracy
of diffusivity, it was determined by using a more accurate method based on Braun–Guillet approach [5].
The activity coefficients and diffusivities for three systems, water + phenol resin, NA + phenol resin and
NB + phenol resin, were determined (note: the detailed name of NA and NB cannot be opened here due to
the secret reason; if interested, please contact the authors directly). The PolyNRTL model was adopted to
represent the measured activity coefficient data. Also, the elution theory of the packed chromatographic
column and the model parameter estimations for obtaining the diffusivities were evaluated.

2. Experimental

A schematic diagram of the IGC apparatus used for determination of the activity coefficients and
diffusivities in polymers is presented in Fig. 1. Although it is possible to obtain both the activity coefficient
and diffusivity data simultaneously, the experiments were carried out separately for determination of
activity coefficients and diffusivities in this work, because the actual processes are operated in a variety
of conditions in terms of the temperatures and concentrations, where either the equilibrium phenomena
or mass transfer behavior is critical.

2.1. Measurement of activity coefficient

All the measurements were carried out on a gas chromatograph equipped with a thermal conductivity
detector. Helium was used as the carrier gas. The carrier gas flow rate was measured by means of a soap
bubble flowmeter. The temperatures of the injection block and the detector were set about 50 ◦ C above
the column temperature to avoid condensation in the injector and detector.
The chromatographic column is 3.2 mm o.d., 1.80 m long stainless steel tube, packed with Chromosorb
P (60–80 mesh) which was coated with phenol resin (M w = 5000). The percentage loading of the coated
polymer is 3.98 wt.%. The flow rate of the carrier gas was set to be 20 cm3 /min in all the experiments,
which has been proven to be sufficient to guarantee the equilibria between the vapor and polymer phases.
Small amount of solvent (<0.1 ␮l) was injected through the rubber septum of the injection port into the
carrier gas using 1 ␮l syringe. Usually, about 0.7 ␮l of air was injected along with the liquid sample in order
S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189 1181

Fig. 1. A schematic diagram of the IGC apparatus used for measurement of activity coefficients and diffusivities in polymers:
A, needle valve; B, preheating coil; C, D, tempering coils; E, TCD detector; F, GC column; G, injector; H, water saturator; I,
soap bubble flowmeter; TS 1, GC air thermostat; TS 2, water thermostat bath; TI, TIC, temperature sensor or control.

to determine the dead volume/time. Retention time data were found to be independent of solvent sample
size. At each temperature, three replicate experiments were carried out to ensure that the measured results
were reproducible at a fixed set of conditions. In this work, the activity coefficients at infinite dilution
(Ω ∞ ) were determined for water, NA and NB in phenol resin (M w = 5000) in the temperature region
from 94 to 115 ◦ C.

2.2. Measurement of diffusivity

The specification of the chromatography and detector assembly was the same as that in the measurement
of activity coefficients. Two chromatographic columns (columns A and B) were used throughout all the
experiments in the measurement of diffusivity. One column (A) is 3.2 mm o.d., 1.95 m long stainless
steel tube which was packed with Chromosorb P (60–80 mesh) with the coated polymer percentage of
10.51 wt.%. The other column (B) is 3.2 mm o.d., 0.51 m long stainless steel tube packed with glass beads
(60–80 mesh) with the coated polymer percentage of 0.5 wt.%. Column B was used for determination of
the thickness of the polymer film coated on the support material. At each temperature, the measurements
with the column A were carried out over a range of flow rate from 20 to 60 cm3 /min. Whereas, the
measurements with column B were carried out over a range of temperature from 45 to 90 ◦ C at the fixed
flow rate (50 cm3 /min) of the carrier gas. Three replicate experiments were done at each flow rate or
temperature to ensure the reproducibility of the measured results.
In this work, the diffusivity data for water, NA and NB in phenol resin (M w = 5000) at the temperatures
of 55 and 74 ◦ C were determined. The diffusivity for toluene in polystyrene (PS, M w = 68,700) at
the temperature of 140 ◦ C was also determined in this work in order to validate this apparatus. The
value of diffusivity for toluene in PS is 8.45 × 10−9 cm2 /s, which is in good agreement with the value,
8.61 × 10−9 cm2 /s, in the literature [3].
1182 S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189

3. Results and discussions

3.1. Activity coefficient

3.1.1. Data reduction


The generalized equations relating the measured data to activity coefficients of solvent in polymer are
given as follows:
M2 273.15R 1
Ω1∞ = γ1∞ = (1)
M1 vgo M1 ϕ1s P1s

F Pfm − Pws 273.15 3 (Pin /Pout )2 − 1


vgo = (tR − tA ) (2)
w2 760 Tfm 2 (Pin /Pout )3 − 1
where R is the gas constant, M1 the molar mass of solvent, w 2 the mass of polymer on the inert carrier
material packed in the column, tR the retention time of the solvent, tA the dead time required for an inert
gas to pass through the column, F the carrier gas volume flow measured by the soap bubble flowmeter
at the pressure, Pfm and temperature, Tfm , Pws the saturated vapor pressure of water at Tfm , Pin and Pout
are the column inlet and outlet pressure, respectively, and P1s is the saturated vapor pressure of the pure
solvent. The saturated vapor pressures (Ps ) of water, NA and NB were calculated by use of Antoine
constants taken from the Dortmund Data Bank (DDB). The saturated fugacity coefficients of the solvents
ϕ1s were calculated according to the Soave equation of state [6]; the necessary critical properties and
acentric factors were also taken from DDB.
The measured activity coefficients at infinite dilution for three systems are presented in Table 1. A plot
of Ω ∞ versus T is shown in Fig. 2. It can be seen that the activity coefficient values of water are much
higher than those of NA and NB, probably due to the large difference in the properties between water
and phenol resin (M w = 5000). Also, it can be seen that the activity coefficients of NA and NB in phenol
resin are much less dependent on the temperature.

3.1.2. Modeling of activity coefficient


PolyNRTL model was adopted here for representation of the activity coefficients at infinite dilu-
tion. The activity coefficient of specie I is described as the sum of the Flory–Huggins combinatorial
entropy-of-mixing term and the NRTL energetic term [7]:

ln γI = ln γIFH + ln γINRTL (3)

Table 1
The measured activity coefficients (Ω ∞ ) for water, NA and NB in phenol resin (M w = 5000) at infinite dilution
Temperature (◦ C) Solvent

H2 O NA NB
94.7 30.7 17.9 15.1
104.5 29.3 17.6 15.0
114.1 28.5 17.4 15.0
S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189 1183

Fig. 2. Activity coefficients at infinite dilution (Ω ∞ ) for water, NA and NB in phenol resin (M w = 5000), measured by IGC
method.

    φJ 
φI
ln γIFH= ln + 1 − rI (4)
xI J
rJ
 
j Xj Gj s τj s
 Xj Gj s  
k Xk Gkj τkj
lnγI =s = 
NRTL
+  τj s −  (5)
k Xk Gks j k Xk Gkj k Xk Gkj

where
nI rI
φI =  (6)
J nJ rJ
xI ri,I
Xi =   (7)
J j xJ rj,J

Gji = exp(−αji τji ) (8)


gji − gii
τji = (9)
RT
where α is the nonrandomness factor and fixed to be 0.2 in this work, because the value of α, in the range
of 0.2–0.3, has no significant impact on the behavior of the model. The variable, r is the polymerization
degree. The species i and j represent solvent molecules or segments. The symbols I and J stand for the
polymers or solvents. The quantities of gj i and gii are, respectively energies of interaction between j–i
and i–i pairs of species. The binary interaction parameters, τ ij and τ j i , are usually determined by fitting
the experimental data.
1184 S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189

Table 2
The estimated binary interaction parameters of PolyNRTL between PR (phenol resin) and solvent (water, NA and NB)
Binary parameter PR (i)–water (j) PR (i)–NA (j) PR (i)–NB (j)

aij −7.7644 −7.1302 −7.3458


aj i 0.0 0.0 0.0
bij 552.37 59.184 200.31
bj i 0.0 0.0 0.0
α ij 0.2 0.2 0.2
ri a 53 53 53
a
Here, ri denotes the polymerization degree of phenol resin (M w = 5000).

The estimated parameters of PolyNRTL for three studied systems are presented in Table 2. A comparison
between the correlated results and the experimental values is graphed in Fig. 2. The agreement is very
good as seen in Fig. 2. In fact, such determined parameters of PolyNRTL can be applied to predict the
VLE behavior in the finite concentration region. More detailed discussion about this issue, however, is
omitted here considering the limited pages.

3.2. Diffusivity

3.2.1. Elution model for packed columns


A brief description of the packed chromatographic column model is necessary for determination of
diffusivity by IGC method. Assuming that the thickness of polymer film in the packed column, df , is
uniform and the thermodynamic equilibrium between the gas and polymer phases is instantaneous. The
diffusion coefficients of the small molecule in the gas and polymer phases are assumed to be DL and Dp ,
respectively. Surrounding the coated particle is a gas film characterized by an external fluid film mass
transfer coefficient, kf . The real time–concentration profile is treated by employing the moment analysis
algorithm. After correcting for the gas compressibility, an equation relating the diffusion coefficient of
small molecule in polymer is given by Romdhane and Danner [3] as follows:
 
µ∗2 2τ Dmo 2 k df2 2 k KRp Rp3 − Rs3
L 2 = 2λdp f + jf + v+ fv (10)
µ1 v 3 (1 + k)2 Dp 3 (1 + k)2 kfo j Rp3
and
 
Kεp
µ1 = ta 1+ (11)
εg
where
∞
tc(t) dt
µ1 = 0∞ (12)
0 c(t) dt
∞
(t − µ1 )2 c(t) dt
µ∗2 = 0  ∞ (13)
0 c(t) dt
In Eqs. (10) and (11), L is the column length, v the velocity of the carrier gas, Dmo the diffusion coefficient
of small molecule in the gas phase at the outlet, λ the packing characteristic factor, τ the tortuosity factor
S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189 1185

which takes into account the irregular pattern along the particles with the usual value to be 0.6–1.0., κ
the capacity ratio defined by the equation κ = Kεp /εg , where K is the equilibrium partition coefficient
between the gas and polymer phases, ε p and ε g are the fractional volumes of the stationary phase and
mobile phase, Rs and Rp are the radius of the support particle and the polymer film, and dp the diameter
of the support particle, and ta is the retention time of air. The left side of Eq. (10), L(µ2 ∗ /µ21 ), is usually
referred to as the height equivalent to a theoretical plate (HETP).
After a series of reduction, a mathematical manipulation of Eq. (10) gives:
Hmt j
= Cp + Cg0 (14)
v f
and
µ∗2 j B0 2
Hmt = L 2f
− Aj − j (15)
µ1 v
where j and f are the compressibility factors of James–Martin and Giddings et al. [3], respectively, A
the multipath factor, A = 2λd p , B0 the longitudinal diffusion term, B0 = 2τ D mo , Cp and Cg0 are the
resistances to mass transfer in the polymer and gas phases, and their expressions are given by the following
equations:
2 k df2
Cp = (16)
3 (1 + k)2 Dp
 
2 k KRp Rp3 − Rs3
Cg0 = (17)
3 (1 + k) kfo
2 Rp3

3.2.2. Model parameter estimation


A very complicated procedure is involved in the data analysis [3,8]. Total five parameters of K, B0 , A,
df , and Cp must be estimated in order to obtain the diffusivity.
1. The first moment (µ1 ) and the second central moment (µ∗2 ) were determined by using Eqs. (12) and
(13). Then the value of K was obtained from a slope of a plot of µ1 versus ta according to Eq. (11).
2. The value of Dmo was estimated by the Fuller–Schettler–Giddings equation [3], and the value of B0
was then calculated. Notice that the value of τ was set to be 1 in all the cases, since its variation did
not affect the results appreciably [3].
3. The value of A was obtained from the high velocity data where Eqs. (14) and (15) simplify to:
µ∗2
L = Af + Cv (18)
µ21
4. The thickness of the polymer film was determined through the same procedure following Braun and
Guillet [9]. Dodecane was chosen as the probe solvent, since its retention mechanism was known to be
restricted to surface adsorption. The retention volume of dodecane through the column B packed with
the glass beads which was coated with the polymer of 0.5 wt.% was measured at several temperatures
below the glass transition temperature of phenol resin (M w = 5000) at the fixed flow rate (50 cm3 /min)
of the carrier gas. The adsorption equilibrium constant, Ka0 and the adsorption heat .Ha were then
obtained.
1186 S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189

By combing the retention volume data for the solvent through the column A over a variety region
of temperatures, the specific surface area Ssp of the Chromosorb P in the column A was obtained. The
polymer film thickness, df , was then calculated using the following relationship:
x
df = (19)
Ssp ρp (1 − x)

where x is the percent loading of polymer on the support, and ρ p is the density of polymer at the
measured condition.
5. The value of Cp was obtained by regressing Hmt /v against j/f according to Eq. (14).
6. The diffusivity, Dp , was calculated from Eq. (16).

3.2.3. Experimental results


The measured GC data were treated following the procedure described in Section 3.2.2. The measured
GC quantities have been proven to be highly consistent with the theoretical relationships based on the
Romdhane–Danner approach [3]. For example, the variation of µ1 was shown to be linearly dependent
on ta as seen in Fig. 3, which is consistent with the theoretical relationship of Eq. (11). Also, the linear
relationships between Hmt /v and j/f have been observed, which are in consistence with the theoretical
Eq. (14). These figures, however, are not presented here due to the limited pages.
The measured diffusivity data for three systems at the temperatures of 55 and 74 ◦ C are presented in
Table 3. It is found that the values of diffusivities decrease in the order H2 O > NA > NB.
In order to apply the measured diffusivity data to the simulation of the manufacturing processes of phenol
resin, and also to extrapolate the diffusivity data at infinite dilution to the region of finite concentration,
the Vrentas–Duda [10] free volume theory can be applied [11]. Detailed discussions about the modeling
and prediction of the diffusivity, however, are beyond the scope of this manuscript.

Fig. 3. Plots of µ1 vs. ta at 55 ◦ C (a) and 74 ◦ C (b).


S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189 1187

Table 3
The measured diffusivities (Do ) for water, NA and NB in phenol resin (M w = 5000) at infinite dilution
Solvent Do × 1010 (cm2 /s)

55 ◦ C 74 ◦ C

H2 O 3.1167 3.2901
NA 0.8965 0.9380
NB 0.5590 0.6333

4. Conclusions

The IGC method is a fast and reliable technique for determination of the activity coefficients and
diffusivities for small molecules in polymers, particularly in the region with negligible amounts of small
molecules. The key point in the measurement of activity coefficients in polymers is to make a thin and
uniform polymer film on the support material and to ensure the equilibrium between gas and polymer
phases to be fully achieved. Whereas, the measurement of solvent diffusivity involves a more complicated
procedure. The retention time and peak area of small molecules must be determined over a variety of flow
rates at each temperature as well as over a variety of temperatures at a fixed flow rate. In this work, the
Romdhane–Danner approach was applied to the data analysis, and the thickness of the polymer film on
the support material was determined by using a more accurate method based on Braun–Guillet approach,
since it has a significant effect on the measured accuracy of the diffusivity.
In this work, the activity coefficients and diffusivity for water, NA, and NB in phenol resin (M w = 5000)
were determined by use of IGC technique with packed column. The PolyNRTL model was adopted to rep-
resent the measured activity coefficient data. The PolyNRTL parameters estimated from the activity co-
efficient data at infinite dilution can be used to fairly predict the VLE behavior in finite concentration range.
In regard to the diffusivity, the elution model of packed columns and the model parameter estimations
were evaluated. The measured GC quantities were shown to be highly consistent with the theoretical rela-
tionships based on the Romdhane–Danner approach. The obtained diffusivity data at infinite dilution can
be used to predict the temperature and concentration dependence of diffusivity by the free volume theory.

List of symbols
A multipath factor (cm)
B0 axial dispersion term (cm2 /s)
c solute concentration in the gas phase (mol/cm3 )
C overall resistance to mass transfer, the coefficient in Eq. (18)
Cg0 resistance to mass transfer in the gas phase (s)
Cp resistance to mass transfer in the polymer phase (s)
df polymer film thickness (cm)
dp particle diameter (cm)
DL longitudinal diffusion coefficient (cm2 /s)
Dmo solute diffusion coefficient in the mobile phase at the outlet pressure (cm2 /s)
Dp solute diffusion coefficient in the polymer phase (cm2 /s)
f Giddings compressibility factor
1188 S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189

F flow rate of carrier gas (cm3 /s)


g interaction energy
G Gibbs energy
Hmt effective plate height represented by Eq. (15)
j James–Martin compressibility factor
kfo fluid film mass transfer coefficient at the outlet pressure (cm/s)
K equilibrium partition coefficient of the polymer–solvent system
L length of the column (cm)
M molar mass of solvent or polymer
Ps saturated vapor pressure
Pin column inlet pressure
Pout column outlet pressure
r number of segments per polymeric species, degree of polymerization
R gas constant
Rp radius of the polymer film (cm)
Rs radius of the support particle (cm)
Ssp specific surface area of the support material
ta retention time of air (s)
tR retention time of solvent (s)
T temperature (K)
v average linear velocity of the carrier gas (cm/s)
w mass of polymer on the support material
x liquid phase mole fraction based on segments
X effective local mole fraction

Greek letters
α NRTL nonrandomness factor
εg fractional volume of the mobile phase
εp fractional volume of the stationary phase
φ free volume fraction
ϕ1s saturated fugacity coefficient of solvent
γ activity coefficient, mole fraction scale
κ capacity ratio represented by the equation, κ = Kεp /εg
λ packing characterization factor
µ1 first temporal moment (s)
µ∗2 variance of the concentration distribution (s2 )
ρp density of polymer
τ ij , τ ii NRTL binary interaction parameter
Ω activity coefficient, mass fraction scale

Superscripts
FH Flory–Huggins term
NRTL local composition contribution term
∞ infinite dilution state
S. Zhang et al. / Fluid Phase Equilibria 194–197 (2002) 1179–1189 1189

Subscripts
i, j, k any species, solvents or segments
I, J any species, solvents or polymers
p polymer species
s solvent species

References

[1] J.M. Caruthers, K.C. Chao, V. Venkatasubramanian, R.S.S. Kioa, C.R. Novenario, A. Sundaram, Handbook of Diffusion
and Thermal Properties of Polymers and Polymer Solutions, DIPPR/AIChE, New York, 1998.
[2] J.S. Vrentas, J.L. Duda, AIChE J. 25 (1979) 1–24.
[3] I.H. Romdhane, R.P. Danner, AIChE J. 39 (1993) 625–635.
[4] F. Timinlioglu, P.K. Surana, R.P. Danner, J.L. Duda, J. Polym. Sci. Part B. Polym. Phys. 35 (1997) 1279–1290.
[5] J.M. Braun, J.E. Guillet, Macromolecules 8 (1975) 882–888.
[6] G. Soave, Chem. Eng. Sci. 27 (1972) 1197–1203.
[7] C.C. Chen, Fluid Phase Equilib. 83 (1993) 301–312.
[8] W. Jiang, H. Liu, S. Han, Acta Phys. Chim. Sinica 15 (1999) 668–672.
[9] J.M. Braun, J.E. Guillet, Macromolecules 8 (1975) 882–886.
[10] J.S. Vrentas, J.L. Duda, J. Polym. Sci. Part B Polym. Phys. 15 (1977) 403–416.
[11] J.M. Zielinski, J.L. Duda, AIChE J. 38 (1992) 405–415.

You might also like