Professional Documents
Culture Documents
Chen 1992
Chen 1992
ABSTRACT Different phase transitions (LCST,UCST, VL, and VLL) in binary, ternary, and, quaternary
olefin-polyolefin systems are found to be easily tuned with density. Polyolefin is alternating poly(ethy1ene-
propylene), PEP for short. The PEP phase transitions are quantitatively related to pressure, temperature,
solventcomposition,and itasize. For example,supercritical ethylene drastically reduces the mutualsolubilities
of comonomers and PEP and hence increases the minimum pressure needed for complete miscibility. This
ethylene antisolvent effect is modeled with statistical associating fluid theory (SAFT). SAFT also predicts
that LCST and UCST curves can merge into a single U-LCST curve upon increasing the size difference
between the polymer and solvent. Furthermore,one internally consistent set of SAFT parameters can correlate
LCST, VL, and VLL equilibria simultaneously, which has not been demonstrated before. SAFT is verified
experimentallywith new phase equilibrium data obtained in an optical variable-volume cell for ternary PEP
solutions in propylene + ethylene, l-butene + ethylene, and l-hexene + ethylene.
3. Statistical Associating Fluid Theory (SAFT) Dip are 24 nonzero universal constants determined by Chen
SAFT is an equation of state that explicitly accounts and Kreglew~ki.~~ ulk is the segment energy which is a
for nonspecific (repulsive, dispersion) and specific (asso- temperature-dependent characteristic parameter in the
ciation due to hydrogen bonding) interactions among equation of statem and 7 = 0.7405. In order to extend eq
molecules. I n a simplified physical picture, a pure SAFT 6 to mixtures, we use the van der Waals one-fluid theory
Macromolecules, Vol. 25,No. 19, 1992 Density-Tuned Polyolefin Phase Equilibria. 2 4989
(vdW1) mixing rules as 2.5
2.0
where I n/
0 10 20 30 40
Molar Mass, kglmolc
06k -
26k
400 - 15k
IOk
-
150
-
a a
5.0k
-
4
m
Y
2 300 -
Y
y 100 w
a
3 3
v) v)
v)
w g 200 - 700 -
p:
a a
a
50
LOO - -
0 0
-150 -100 -50 0 50 LOO I50 200 250
TEMPERATURE ("C) TEMPERATURE ("C)
Figure 2. Experimental datal6and theoretical predictions (ki, Figure 4. Experimentall6and theoreticalp-7' phase boundaries
= 0) for PEP790 + propylene and PEP790 + 1-butene. SAFT for PEP + propylenefor different PEP molecular weights. Using
predicts quantitatively the various types of phase transitions. the binary interaction parameter correlation developed for this
SAFT predicts a small VLL region for the PEP790 + 1-butene system, theoretical calculations are shown for various PEP mo-
system that is not observed experimentally. In addition, SAFT lecular weights. For the PEPlOk + propylene system, SAFT
slightly overpredicts the critical point of the solvent. The LL to predicts a steep UCST curve with a UCEP around -80 O C .
L phase boundary is also referred to as LCST in the text.
500
1 ' 1 ' 1 ' l ' 1 ' 1 ' 1 '
400
LOO
0
0 0 20 40 BO 60 100
-150 -100 -50 0 LOO 50 150 200 250
MOLECULAR WEIGHT (XIO-3)
TEMPERATURE ('C)
Figure 5. LCST pressure dependence on the polymer molec-
Figure 3. Experimentallsand theoreticalp-Tphase boundaries ular weight for PEP + propylene at various temperatures from
for PEP + propylene for different PEP molecular weights. A experimental datal6and SAFT calculations.
molecular weight dependent binary interaction parameter, ki,, is
needed to correlate the phase behavior quantitatively. As the for the PEP26k mixture a t approximately 10 "C. The
molecular weight of the polymer increases the phase boundary
shiftsto higher pressures, and finallythe UCST and LCST curves U-LCST behavior for PEP26k is later confirmed in a
merge. separate glass tube experiment.16
SAFT can correlate the experimental data shown in
which is not observed experimentally. However, the SAFT Figure 3 with the use of one binary parameter kij that
predictive power for these systems is far better than that depends on, and can be correlated against, the polymer
of the Flory-Patterson approach.16 molecular weight. The kij correlations used in this work
As the molecular weight of the polymer increases, the are shown in Table IV.
LCST curve shifts to lower temperatures and higher These kij correlations allow for theoretical predictions
pressures, as shown in Figure 3 for propylene + PEPS where no experimental data are available, e.g., for pro-
having different molecular weights. The region of com- pylene + PEPlOk and propylene + PEPlBk, as shown
plete miscibility (above the LCST boundary) shrinks, and in Figure 4. The purpose for these calculations is to explore
eventually the LCST and UCST curves merge for pro- the UCST demixing, which cannot be observed experi-
pylene + PEP96k into a U-LCST curve. In addition, for mentally because of low-temperature equipment limita-
PEP790 and PEP5.9k, a dew-point-like phase transition tions, and to quantify the molecular weight dependence
is observed, whereas for PEP26k and PEP96k the phase of the U-LCST minimum. For propylene + PEPlOk,
transition is of the bubble-point type. While SAFT SAFT predicts a separate UCST curve with an UCEP
correctly predicts the U-LCST minimum for PEP96k, it (upper critical end point) a t approximately -80 "C. As
is somewhat off in predicting the slope of the UCST branch shown in Figure 4, SAFT can predict a transition from
of this curve. In addition, it predicts a U-LCST minimum separate UCST and LCST curves (propylene + PEPlOU
4992 Chen et al, Macromolecules, VoE. 25, No. 19, 1992
-
2501-:--l--i-i
200
-m
400
m
-
2
w
150
-2
Y
300
p: s
3
v)
y LOO
zy 200
m m
a a
50 LOO
a T=200°C -
- SAFT
0 I b I I I I I , . L 0 -150 -100 -50 0 50 LOO 150 200 250
0
0 20 40 eo 80 LOO
MOLECULAR WEIGHT ( X I O - ~ ) TEMPERATURE ("C)
Figure 6. LCST pressure dependence on the polymer molec- Figure 8. ExperimentaP and theoretical P-Tphase boundaries
ular weight for PEP + 1-butene at various temperatures from for PEP96k + propylene + 1-butene. In the absence of 1-butene,
experimental datal6 and SAFT calculations. the UCST and LCST curves merge. As the 1-butene concen-
tration increases, the LCST boundary shifts to lower pressures
250, , and higher temperatures. For 65 w t % 1-butene in the system,
1 ' 1 '
separate UCST and LCST curves are predicted.
300
1 ' 1 ' 1 ' 1 ' 0 i '
2
Y
150
200oc
J
a
v)
100
ma
50
790
0
TEMPERATURE ("C) 50 t
Figure 7. ExperimentallGand theoreticalP-T phase boundaries
for PEP + 1-butene for different PEP molecular weights. A
0
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.7
molecular weight dependent binary interaction parameter, k,, is POLYMER WEIGHT FRACTION
needed to correlate the phase behavior quantitatively. Figure 9. SAFT-calculated P-x phase diagrams for the PEP +
1-butenesystem at 200 O C . The phase boundary shifts to higher
to a single U-LCST curve (propylene + PEP15k) upon pressure and the calculated critical point to lower polymer
increasing the polymer molecular weight. Furthermore, composition for higher molecular weight polymers.
increasingthe polymer molecular weight increases the tem-
perature and pressure of the U-LCST minimum, as shown observed for 1-butene+ PEP in Figure 6, where the LCST
in Figure 4 and Table V. pressure is shown at the same three temperatures as in
It should be mentioned that SAFT calculations become Figure 5.
extremely sensitive to small changes in kij for high mo- However, in general, the phase behavior of the 1-butene
lecular weight polymers. For example, for the propylene + PEP system is qualitatively different from that of the
+ PEP96k system, the kij correlation gives a value of 0.027. propylene + PEP system, at least for the high molecular
If one sets ki, = 0, then SAFT predicts that the LCST weight polymers, as shown in Figure 7. For all the systems
curve extends all the way to the vapor pressure curve. examined, the LCST curve intersects with the vapor
Such high sensitivity to small changes in k, was also pressure curve, and no U-LCST type of phase behavior is
observed by Liu and Prausnitz,26 who used the PHCT detected. As for propylene, increasing the polymer mo-
equation of state. lecular weight decreases the one-phase region. In other
The effect of the polymer molecular weight on the pro- words, the LCST boundaries are shifted to lower tem-
pylene + PEP LCST pressure is shown at three different peratures and higher pressures. SAFT results are in a
temperatures in Figure 5. The LCST pressure is very very good agreement with the experimental data. As for
sensitive to the polymer molecular weight below 30 0oO propylene, only one molecular weight dependent binary
and, especially, below 20000. For higher molecular parameter (ktj given in Table 4) is used.
weights, however, the LCST pressure is very weakly de- This binary parameter, derived from the experimental
pendent on the polymer molecular weight. A similar mo- data for propylene + PEP and 1-butene + PEP binary
lecular weight dependence of the LCST pressure is systems, without further readjustment, is also used to
Density-Tuned Polyolefin Phase Equilibria. 2 4993
500
1 ' 1 ' 1 ' 1 ' I ' I ' I ' I ' I
- SAFT I VL+L
- A L L + L (BP) (3 VLL+LL
400
h
a
2 300
w
Y
a
g
P
a
200
t
LOO
0
40 80 120
TEMPERATURE ("C)
180 200
-I
240
loo
0 i&&;.j
-50 0
50 LOO
TEMPERATURE ("C)
150
12.2 r1Z Eth
200 250
Figure 10. Experimental and theoreticalp-T phase boundaries Figure 12. Experimental and theoreticalp-Tphase boundaries
for PEP26k + 1-butene + ethylene at different ethylene for PEP26k + 1-hexene + ethylene at different ethylene
concentrations. For 23.4 wt % ethylene in the mixture, SAFT concentrations. The binary parameter, kp~,used for these
predicts that the UCST and LCST merge. calculationsis the same as that used for the PEP26k + 1-butene
so0
+ ethylene system shown in Figure 10.
I ' I i I ' I ' I
I. T=lOO°C 500
1
I I I
0 T=150°C A T=lOO°C
400 0 T=150°C
400 I T=200°C
P
2 300 - - SAFT
I
w
w
P
a
fn
13 200
E 13 200
;'""
a
P
a
LOO
100
0
0 6 IS
10 20 21
I30
l/ J
0
ETHYLENE WEIGHT PERCENT 0 10 20 30 40
Figure 11. LCST pressure for PEP26k + 1-butene+ ethylene ETHYLENE WEIGHT PERCENT
dependence on the ethylene concentration at constant temper- Figure 13. LCST pressure for PEP26k + 1-butene+ ethylene
ature. Experimental data and SAFT calculations. dependence on the ethylene concentration at constant temper-
ature. Experimental data and SAFT calculations.
predict the phase behavior of the ternary propylene +
+
1-butene PEF96k system, shown in Figure 8. In all the Table IV
SAFT kifCorrelationP
cases, the polymer composition is 15wt % . In the absence
of 1-butene in the system, i.e., for propylene + PEP96k, system A(') A(2) A(3)
the U-LCST phase behavior is observed. As the 1-butene PEP + propylene 0.027 4.02209 1.034 X lo4
concentration increases, the binary solvent becomes a PEP + 1-butene 0.005 -0.03499 7.483 X
"better" solvent because 1-butene acts like a cosolvent. As PEP67k + n-hexane -0.02 0.0 0.0
a result, the one-phase region increases (the LCST phase n-hexane + propylene 0.063 0.0 0.0
boundary shifts to lower pressures and higher tempera- 0 ki, = A(') + A@)exp(-A@)Mp). Mp = PEP molecular weight in
tures). Eventually, for a solvent containing 65 wt % 1- g/mol.
butene (on a polymer-free basis), SAFT predicts that the
U-LCST curve splits and that the separate UCST and point (filled circle) shifts to lower polymer compositions
LCST curves intersect with the vapor pressure curve. upon increasing the polymer molecular weight. These
While the scope of this work is limited to the systems trends are qualitatively correct. However, our preliminary
having a constant polymer concentration (around 15 wt experimental data suggest that the predicted P-x curves
%), we make an attempt to predict an LCST dependence should be more flat. This issue is being addressed in our
on the polymer concentration in 1-butenesolutions. Figure current work.
9 illustrates a preliminary, SAFT-generated P-x phase In a ternary system of 1-butene + ethylene + PEP26k,
diagram for 1-butene + PEP at 200 OC. As the polymer ethylene acts like a powerful antisolvent because it
molecular weight increases, the phase boundary shifts to drastically reduces the 1-butene + PEP26k density, which
higher pressures, and thus the two-phase region (of limited is shown in Figure 10. As the ethylene concentration
miscibility) increases. In addition, the calculated critical (polymer-free basis) increases, the phase boundary shifts
4994 Chen et al. Macromolecules, Vol. 25, No. 19, 1992
PEPMW
96k
26k
15k
Table V
SAFT-Calculated U-LCST Minima for the PEP
Propylene System
Tmin
44.0
10.0
-20.0
("C) Pmin
322.2
251.8
116.4
+
(bar)
-
m
0 8.0 x t % Prop
Prop ) ~ 1
510k U-LCST minimum reaches the VLL curve <
m
Y
W
m 4 0
to higher pressures. For example, addition of 7.4 wt 5% 7
v)
ethylene to the system shifts the LCST by more than 80
.
W
v)
bar. A further increase of the ethylene concentration to 0:
a
23.4 wt % results in a U-LCST-type of behavior, as
predicted by SAFT at 23.4 wt ?& . No experimental data 20
J
30.0 r l X E l h
miscibility. Furthermore, for 35.5 wt % ethylene (polymer-
free basis), the UCST and LCST curves merge into a single
.
U-LCST curve. 200: - SAFT
SAFT quantitatively correlates the LCST curves up to
24.3 w t % using two adjustable parameters, k ~ (ethylene
p
+ PEP26k) and k ~ (1-hexene
p + PEP26k). However, K
2 150 -
the k ~ isp determined based on the 1-butene + ethylene
20.0 w t Z Elh
v /