Cosolvency Effects on Copolymer Solutions at High Pressure BRUCE M. HASCH, MELCHIOR A. MEILCHEN, SANG-HO LEE, and MARK A. McHUGH* Department of Chemical Engineering, johns Hopkins University, Baltimore, Maryland 21 21 8 SYNOPSIS Cloud-point data to 180°C and 2800 bar are presented for polyethylene, poly( methyl ac- rylate), and two poly(ethy1ene-co-methyl acrylate) copolymers (10 and 31 mol % methyl acrylate) in propane and chlorodifluoromethane with two cosolvents, acetone and ethanol. The addition of small amounts of either cosolvent to the copolymer-solvent mixtures shifts the cloud-point curve to lower pressures and temperatures, as both cosolvents provide favorable polar interactions with the acrylate group in the backbone of the copolymer. Ethanol has a larger effect than acetone since ethanol hydrogen bonds to the acrylate group. However, if the concentration of ethanol is increased above ca. 10 wt %, it self- associates and reverts to antisolvent behavior, forcing the copolymer out of solution. For nonpolar polyethylene-propane mixtures, the polar cosolvents behave as traditional an- tisolvents. In poly (methyl acrylate) -chlorodifluoromethane mixtures, both polar cosolvents enlarge the single-phase region. The cloud-point curves for the ( co ) polymer-propane- acetone mixtures are modeled reasonably well using the Sanchez-Lacombe equation of state with two adjustable mixture parameters. No attempt is made to model the mixtures that exhibit hydrogen bonding. 0 1993 John Wiley & Sons, Inc. Keywords: copolymer solutions, cosolvency effects on miscibility of copolymers in co- solvent systems, pressure-temperature relations for cloud point of copolymers in cosolvent systems INTRODUCTION While many studies have investigated the effect of liquid cosolvents on the solubility behavior of polar homopolymers, few have been performed on copol- ymer-cosolvent mixtures at high pressures. It is worthwhile to mention briefly some of the studies done within the past 20 years on the solubility be- havior of polar homopolymers in liquid cosolvent mixtures since there are direct corollaries to the high-pressure work reported here with polar /non- polar copolymer-cosolvent mixtures. Wolf and Blaumg demonstrate that the use of cosolvents that hydrogen bond to the polymer cause large changes in the region of miscibility. For example, small amounts of 2-butanol depress the upper critical so- lution temperature (UCST) of the poly (methyl methacrylate) (PMMA) -chlorobutane system by as * To whom correspondence should be addressed. Journal of Polymer Science: Part B: Polymer Physics, Vol. 31, 429-439 (1993) 0 1993 John Wiley & Sons, Inc. CCC 0%37-6266/93/040429-11 much as 70°C as the 2-butanol hydrogen bonds to the basic acrylate group in the backbone of the poly- mer. However, subsequent addition of butanol to the mixture causes an increase in the UCST as the butanol self-associates and forms long-chain mul- timers which greatly decrease the solvent power of the mixture. At low cosolvent concentrations poly- mer-cosolvent hydrogen bonding is a favorable in- teraction that promotes miscibility, but as the con- centration of cosolvent is increased, cosolvent-co- solvent hydrogen-bonding interactions overwhelm cosolvent-polymer interactions and decreases the region of miscibility. Many authors report findings for other polymer-solvent-cosolvent systems that are consistent with this behavior.'-' More modest changes in solubility behavior are observed if a polar cosolvent is used that does not hydrogen bond. For example, when 4-heptanone is added to PMMA- chlorobutane mixtures, the UCST decreases mono- tonically with heptanone concentration over only a 10°C range.g Cowie and McEwen" have shown that low mo- lecular weight polystyrene (PS) , which is insoluble 429
Transcript
Cosolvency Effects on Copolymer Solutions at High Pressure
BRUCE M. HASCH, MELCHIOR A. MEILCHEN, SANG-HO LEE, and MARK A. McHUGH*
Department of Chemical Engineering, johns Hopkins University, Baltimore, Maryland 21 21 8
SYNOPSIS
Cloud-point data to 180°C and 2800 bar are presented for polyethylene, poly( methyl ac- rylate), and two poly(ethy1ene-co-methyl acrylate) copolymers (10 and 31 mol % methyl acrylate) in propane and chlorodifluoromethane with two cosolvents, acetone and ethanol. The addition of small amounts of either cosolvent to the copolymer-solvent mixtures shifts the cloud-point curve to lower pressures and temperatures, as both cosolvents provide favorable polar interactions with the acrylate group in the backbone of the copolymer. Ethanol has a larger effect than acetone since ethanol hydrogen bonds to the acrylate group. However, if the concentration of ethanol is increased above ca. 10 wt %, it self- associates and reverts to antisolvent behavior, forcing the copolymer out of solution. For nonpolar polyethylene-propane mixtures, the polar cosolvents behave as traditional an- tisolvents. In poly (methyl acrylate) -chlorodifluoromethane mixtures, both polar cosolvents enlarge the single-phase region. The cloud-point curves for the ( co ) polymer-propane- acetone mixtures are modeled reasonably well using the Sanchez-Lacombe equation of state with two adjustable mixture parameters. No attempt is made to model the mixtures that exhibit hydrogen bonding. 0 1993 John Wiley & Sons, Inc. Keywords: copolymer solutions, cosolvency effects on miscibility of copolymers in co- solvent systems, pressure-temperature relations for cloud point of copolymers in cosolvent systems
INTRODUCTION
While many studies have investigated the effect of liquid cosolvents on the solubility behavior of polar homopolymers, few have been performed on copol- ymer-cosolvent mixtures at high pressures. It is worthwhile to mention briefly some of the studies done within the past 20 years on the solubility be- havior of polar homopolymers in liquid cosolvent mixtures since there are direct corollaries to the high-pressure work reported here with polar /non- polar copolymer-cosolvent mixtures. Wolf and Blaumg demonstrate that the use of cosolvents that hydrogen bond to the polymer cause large changes in the region of miscibility. For example, small amounts of 2-butanol depress the upper critical so- lution temperature (UCST) of the poly (methyl methacrylate) (PMMA) -chlorobutane system by as
* To whom correspondence should be addressed. Journal of Polymer Science: Part B: Polymer Physics, Vol. 31, 429-439 (1993) 0 1993 John Wiley & Sons, Inc. CCC 0%37-6266/93/040429-11
much as 70°C as the 2-butanol hydrogen bonds to the basic acrylate group in the backbone of the poly- mer. However, subsequent addition of butanol to the mixture causes an increase in the UCST as the butanol self-associates and forms long-chain mul- timers which greatly decrease the solvent power of the mixture. At low cosolvent concentrations poly- mer-cosolvent hydrogen bonding is a favorable in- teraction that promotes miscibility, but as the con- centration of cosolvent is increased, cosolvent-co- solvent hydrogen-bonding interactions overwhelm cosolvent-polymer interactions and decreases the region of miscibility. Many authors report findings for other polymer-solvent-cosolvent systems that are consistent with this behavior.'-' More modest changes in solubility behavior are observed if a polar cosolvent is used that does not hydrogen bond. For example, when 4-heptanone is added to PMMA- chlorobutane mixtures, the UCST decreases mono- tonically with heptanone concentration over only a 10°C range.g
Cowie and McEwen" have shown that low mo- lecular weight polystyrene (PS) , which is insoluble
429
430 HASCH ET AL.
in either liquid acetone or diethyl ether, is readily soluble in a mixture of the two solvents. A single- phase region is formed covering 170°C between the upper and lower critical solution temperatures when 33 vol % diethyl ether is added to a PS-acetone mixture. Wolf and Blaum" expanded this study by investigating the effect of pressure on the PS-ace- tone-diethyl ether system. At low pressures, high molecular weight PS ( M , = 110,000) is also not sol- uble in either liquid acetone or diethyl ether, how- ever, at temperatures near 2OOOC and pressures of ca. 400 bar, PS will dissolve in either solvent. Adding acetone to PS-diethyl ether mixtures monotonically reduces the pressure needed at a given temperature to obtain a single phase. The explanation for the cosolvency effect of acetone is twofold. First, acetone decreases the free volume of the solvent, or equiv- alently, increases the solvent density since acetone is more dense than diethyl ether especially at 200°C where diethyl ether is ca. 7°C above its critical tem- perature. Second, acetone adds considerably to the polarity of the mixed solvent since its dipole moment is more than twice that of diethyl ether.
While the effect of cosolvents on copolymers has been studied using light scattering and viscometry, very little attention has been paid to the contribution of a cosolvent to the global phase behavior of co- polymer-solvent mixtures. One exception is the poly (ethylene- co-vinyl acetate) ( EVAc) -ethylene- vinyl acetate system, which has been thoroughly in- vestigated by RAtzsch and co-workers.12-16 They first determined cloud-point curves in ethylene for EVAc copolymers of varying acetate content.12 The cloud- point curves shift to lower pressures by as much as 250 bar as the vinyl acetate content in the backbone of the copolymer is increased from 3 to 20 mol %. This behavior is consistent with the results of Hasch and co-workers,17 who found that the pressure of poly (ethylene- co-methyl acrylate ) -ethylene cloud- point curves decreased when the acrylate content was increased from zero to 18 mol %. The decrease in the cloud point pressures can be rationalized as a consequence of the favorable increased polar in- teractions between ethylene, which has a quadrupole moment, and vinyl acetate or methyl acrylate, which both have dipole moments. Ratzsch and co- w o r k e r ~ ~ ~ then did studies using vinyl acetate as a cosolvent at concentrations equal to its concentra- tion in the backbone of the copolymer. The cosolvent studies showed that vinyl acetate as a cosolvent can lower the pressure of the cloud-point curve by as much as 450 bar. Again, this is not surprising since the copolymer is comprised of nonpolar ethylene and polar vinyl acetate repeat units, which should make
it readily soluble in solvent mixtures consisting of a polar and a slightly polar component.
Meilchen et al. recently reported experimental results on the phase behavior of the poly (ethyl- ene- co-methyl acrylate ) -solvent-cosolvent system, where the acrylate content in the copolymer was fixed at 36 mol %, the solvents were propane and chlorodifluoromethane (F22) , and the cosolvents were acetone and ethanol.18 While both cosolvents increased the region of miscibility of the copolymer in propane at low cosolvent concentration, ethanol self-associates at high alcohol concentrations, which changes the polarity of the solvent and causes the polymer to precipitate out of solution. At low co- solvent concentrations, ethanol increases the region of miscibility by a larger amount than an equal quantity of acetone since ethanol can hydrogen bond to the acrylate group in the backbone of the copol- ymer. However, because acetone does not self-as- sociate, much more acetone can be added to the so- lution, which results in a larger single-phase region compared to that obtained with ethanol. In F22, the effect of both cosolvents was diminished since F22 can hydrogen bond to both the cosolvent and the acrylate group in the copolymer.
The present study, which is an extension of our two previous publications, 17~18 presents experimental information on the effect of acetone and ethanol as cosolvents on the phase behavior of four ethylene- methyl acrylate copolymers varying in chemical composition from 100 mol % ethylene to 100 mol % methyl acrylate. Table I lists the properties of the solvents used in this study, propane and F22. Note that F22 has a critical temperature and pressure close to those of nonpolar propane, although it pos- sesses a large dipole moment and a critical density that is more than twice that of propane. The most important difference between these two solvents is that F22 can hydrogen bond to the acrylate group in the backbone of the copolymer while propane does not. Although both cosolvents, acetone and ethanol, are very polar, ethanol can form hydrogen bonds with both itself and the basic acrylate group in the backbone of the copolymer while acetone, which is a basic molecule, does not hydrogen bond either to itself or to the acrylate group in the copolymer. Using these two cosolvents, it should be possible to ascer- tain the importance of polarity and hydrogen bond- ing on the phase behavior, especially when these cosolvents are used with propane. Table I1 lists the properties of the polyethylene, poly( methyl acry- late), PMA, and two poly (ethylene- co-methyl ac- rylate) copolymers (10 and 31 mol % methyl acry- late, EMAgo,lo and EM&9/31, respectively) used in
COSOLVENCY EFFECTS 43 1
Table I. Properties of Propane, Chlorodifluoromethane, Acetone, and Ethanol"
Critical Density Dipole Moment Component T, ("C) P, (bar) (g/cm3) (Debye)
this study. The notation used throughout the paper is EMAA,B where A and B represent the respective mole fractions of ethylene and methyl acrylate pres- ent in the polymer. The polarity of the copolymer increases as the acrylate content of the copolymer increases since each methyl acrylate unit is equiv- alent to a methyl propionate unit with a polariza- bility of 87.9 X lopz5 cm3 and a dipole moment of ca. 1.7 Deb~e."-'~ The copolymers are assumed to be statistically random, neglecting the possibility that acrylate units can group in a nonrandom fash- ion in the backbone of the copolymer. Since poly- ethylene and EMA90,10 are partially crystalline, the crystallization boundary can intrude on the fluid- phase portion of the diagram at temperatures near the melting point of the polymer. Cloud-point mea- surements (liquid + liquid --* fluid transitions) will be terminated at the crystallization boundary of the polymer. Cloud-point curves are obtained at a fixed concentration of ca. 5.5 wt % polymer, which should be reasonably close to the true mixture-critical
EXPERIMENTAL
The experimental apparatus and experimental pro- cedures used to obtain the cloud-point data pre- sented in this work are described in detail else- where.24 The cloud-point pressures are measured with a Heise gauge accurate to within f2.8 bar and
are reproducible within +5 bar. The cloud-point temperatures are accurate to within +0.2"C.
MATERIALS
Propane (CP grade, 99.0% minimum purity) was obtained from Linde Corporation and chlorodifluo- romethane (99.8% minimum purity) was obtained from Matheson Gas Products. Acetone (HPLC grade, 99.9% minimum purity) and ethanol (HPLC grade) were obtained from Aldrich Chemical Com- pany, Inc.
The copolymers and polyethylene were kindly donated by DuPont Corporation. Poly (methyl ac- rylate ) was obtained from Scientific Polymer Prod- ucts, Inc.
RESULTS
EMA-Propane-Cosolvent Mixtures
The cloud-point data will be presented in order of increasing polarity of the polymer, starting with the results for nonpolar polyethylene. Figure 1 shows the experimental results for the polyethylene-pro- pane-cosolvent systems. As mentioned previously, the cloud-point curves in Figure 1 are terminated at the crystallization boundary so that only liquid + liquid + fluid transitions are shown. Consider first the effect of ethanol on the cloud point behavior.
Table 11. Physical Properties of the Polymers Used in This Study
Figure 1. Effect of ethanol and acetone on phase be- havior of polyethylene in propane. The polymer concen- tration is 5 wt %.
At ca. 1.0 wt %, ethanol has no effect on the cloud point pressure. This low concentration of ethanol apparently does not increase the solvent density of propane sufficiently to improve the quality of the mixed solvent. Also, the small increase in favorable dispersion and induced dipolar interactions between ethanol and polyethylene apparently does not out- weigh strong ethanol-ethanol interactions, which include dispersion, dipolar, and hydrogen-bonding forces. In fact, if the ethanol concentration is in- creased to ca. 5.0 wt %, polyethylene falls out of solution. Since there are no polar groups in the backbone of polyethylene, ethanol hydrogen bonds to itself, forming polar ethanol multimers that have an effective dipole moment that is much greater than that of an isolated ethanol molecule. As the amount of ethanol multimers increases, the polarity of the solvent increases in a nonlinear manner and reduces the solvent quality of the mixed solvent sufficiently to force nonpolar polyethylene to precipitate out of solution.
Figure 1 also shows that adding 27 wt % acetone reduces the cloud-point pressure by ca. 50 bar. In this instance, acetone contributes to the solvent quality in two ways. It contributes favorable dis- persion and induced dipolar interactions between itself and the polyethylene, it increases the solvent density, and thus it reduces the free volume differ-
ence between the solvent and the polymer. However, as the concentration of acetone is further increased to 42 wt %, acetone-polyethylene interactions are far outweighed by polar acetone-acetone interac- tions, which results in a dramatic increases the cloud-point pressure needed to dissolve polyethyl- ene. If the concentration of acetone is further in- creased to 62 wt %, polyethylene falls out of solution. The negative slope exhibited by the 42 wt % acetone cloud-point curve is probably a result of the dipole- dipole interactions between acetone molecules, which diminish with increasing temperature, 25 so that the dominant interactions at high temperatures are, once again, favorable dispersion and induced dipole interactions between acetone and polyeth- ylene.
Figure 2 shows the effect of ethanol and acetone on the cloud-point behavior of EMAw/lo-propane mixtures. The cloud-point data are terminated at ca. 65"C, which is the crystallization boundary for this partially crystalline copolymer. Before describ- ing the differences between the effects of acetone and ethanol cosolvents on the phase behavior, we first compare the cloud-point behavior of the polyethylene-propane-acetone system with the EMAgo/lo-propane system. This comparison allows us to determine the difference in the magnitude of
800
LIQUID t LlQUlU
60 70 80 90 100 110 120 130 TEMPERATURE ("C)
Figure 2. Effect of ethanol and acetone on phase be- havior of poly (ethylene-co-methyl acrylate) (90 mol % / 10 mol %) in propane. The polymer concentration is 5 w t % .
COSOLVENCY EFFECTS 433
the effect of adding a polar substance either to a nonpolar solvent or to the backbone of a nonpolar polymer. As previously described, Figure 1 shows that adding ca. 27 wt % acetone (or equivalently, ca. 26 mol % ) to nonpolar propane, decreases the pressure of the polyethylene-propane cloud-point curve by ca. 50 bar relative to the base case of no acetone. Figure 2 shows that adding only 10 mol % methyl acrylate to the backbone of nonpolar poly- ethylene increases the pressure of the EMASO/lo- propane cloud-point curve by ca. 150 bar relative to the polyethylene-propane-acetone case over the temperature range of 90-130°C. The difference be- tween these two cases is actually even more exac- erbated since, not only are less than half as many moles of a polar group added to the backbone of polyethylene, the strength of the dipole moment of methyl acrylate in the backbone is only ca. 60% as strong as that of acetone. Evidently, when the polar repeat units are placed in the backbone of the poly- mer, their location and proximity to one another are sufficiently constrained that polymer segment-seg- ment interactions are greatly enhanced.
At first glance, ethanol and acetone influence the phase behavior of the EMAsO/lo-propane system in a similar manner. Both lower the cloud-point pres- sure and reduce the slope of the cloud-point curves, although, at low concentrations, the decrease in cloud-point pressures caused by ethanol is greater than that caused by an equivalent amount of ace- tone. For example, at lOO"C, 2.7 wt % ethanol re- duces the cloud-point pressure by ca. 100 bar, 4.7 wt 5% reduces the cloud-point pressure another ca. 60 bar, but 7.0 wt % causes the polymer to fall out of solution. At low concentrations nearly twice as much acetone is needed to achieve the same reduc- tion in pressure. However, because acetone does not self-associate, larger amounts can be added to the solution without precipitating the polymer. The ef- fect of 10 mol % methyl acrylate in the backbone of the polyethylene is again evident if the polyethyl- ene-propane-acetone curves are compared to those of the EMASO/lo-propane-acetone system. For the polyethylene-propane case, acetone at concentra- tions up to ca. 27 wt ?6 decreases the cloud-point pressure, but at acetone concentrations greater than 27 wt %, the cloud-point pressure increases, even beyond the values found for the base case of no ace- tone. For the EMAsO/lo-propane case, the cloud- point pressure always decreases as acetone is added to the solution to concentrations as high as 40 wt %. Evidently, the small number of favorable ace- tone-acrylate polar interactions is sufficient to bal- ance acetone-acetone interactions that would have
- 1500
LLI 20001 n. 500
0 L 2000
- 1500 - 2
LIQUID +
LIQUID
LIQUID + LIQUID
0 ~ I ~ I ~ I ~ I ~ ' ~ I ' I ~ I ~ I ~ I ~ I ~ I ~ 35 55 75 95 115 135 155
TEMPERATURE ("C)
Figure 3. Effect of ethanol and acetone on phase be- havior of poly(ethy1ene-co-methyl acrylate) (69 mol % / 31 mol % ) in propane. The polymer concentration is 5 wt %.
caused the cloud-point curve to shift to higher pres- sures.
Figure 3 shows the experimental results for the EMA6s~31-propane-cosolvent systems. The en- hanced effect of an increased amount of polar group in the backbone of the polymer is evident by com- paring the differences in pressures and temperatures for the cloud-point curves of the EMAsO/lo-propane and the EMA6s/31-propane systems. Not only is the EMA6s/31-propane cloud-point curve more than 1000 bar higher than the less polar EMA90/lo-pro- pane case, it also turns up abruptly at ca. 140°C. To dissolve EMA6s/31 in propane, it is necessary to op- erate a t temperatures high enough to reduce the po- lar segment-segment interactions within the poly- mer to make the polymer accessible to the solvent. To a first approximation, elevated pressures are needed at the high temperature to increase the den- sity of propane and reduce the free volume difference between the polymer and solvent, and thus make the polymer soluble in nonpolar propane.
The effect of ethanol on the phase behavior of the EMA6s/31-propane is shown in the upper graph of Figure 3. Ethanol a t 2.7 wt % shifts the cloud- point curve by ca. 20°C at an arbitrarily chosen pressure of 1800 bar. Increasing the ethanol con- centration to 5.4 wt % results only in an additional
434 HASCH ET AL.
ca. 10°C increase in the miscibility region. And in- creasing the ethanol concentration to 9.3 wt % in- creases the region of miscibility by an additional ca. 12°C. If the ethanol concentration is increased above ca. 10 wt %, the polymer precipitates out of solution. One reason for the nonlinear effect of ethanol on the phase behavior is that hydrogen bonding is a chemical form of molecular interaction that requires a certain number of specific interaction sites. At low concentrations, ethanol hydrogen bonds to the ac- rylate groups, which outnumber the ethanol mole- cules available to self-associate. As the acrylate groups become saturated with ethanol molecules, any further addition of ethanol results in ethanol- ethanol hydrogen bonding which changes the nature of the mixed solvent to a highly polar environment that causes the copolymer to fall out of solution.
In contrast to chemical forces (hydrogen bond- ing), physical forces that occur between acetone and methyl acrylate do not become saturated. The ad- dition of 6 wt % acetone to the EMA6g,31-propane system reduces the cloud-point temperature by about 25°C) while it takes only 2.7 wt % ethanol to shift the cloud point curve a similar amount. At low cosolvent concentrations ethanol is the stronger co- solvent because of hydrogen bonding. However, as Figure 3 shows, significantly more acetone can be added to the system because it does not hydrogen bond. Notice that 26 wt % acetone increases the region of miscibility by over 100°C and also reduces the cloud-point pressure by ca. 800 bar compared to the base case of no acetone. This large shift in the cloud point curve is much greater than that observed for the EMAgo/lo and the polyethylene cases with equivalent amounts of acetone. It is apparent that the addition of polar acrylate groups in the backbone of the polymer has a large impact on the location of the cloud-point curve.
Although PMA is soluble in pure acetone, it was not possible to dissolve PMA in neat propane or with up to 30 wt % acetone at conditions as extreme as 2850 bar and 180°C. No experiments were at- tempted with ethanol as a cosolvent since PMA does not dissolve in pure ethanol.
EMA-Chlorodifluoromethane-Cosolvent Mixtures
Since chlorodifluoromethane (F22) is a polar solvent capable of hydrogen bonding and dissolving polar substances, the cloud-point curves with this solvent will be presented in order of decreasing polarity, starting with PMA. Figure 4 shows the effect of ethanol and acetone on the cloud-point pressures of PMA-F22 mixtures. Notice that now the cloud-
I ' I ' I . I ' I ' I ' I ' I ' I '
10.9 0 18.9
FI.UIU
60 811 1uO 120 140 16U TEMPERATURE (OC)
Figure 4. Effect of ethanol and acetone on phase be- havior of poly(ethy1ene-co-methyl acrylate) (69 mol % / 31 mol % ) in chlorodifluoromethane. The polymer con- centration is 5 wt %.
point curves have positive slopes typical of many polymer-liquid solvent mixtures, probably because F22 has liquidlike density even at temperatures above its critical temperature, *' as compared to pro- pane, which requires high pressure to achieve high density. Also, the data in Figure 4 show that only very modest pressures are needed to obtain a single phase at temperatures greater than 70°C. The sin- gle-phase region is expanded considerably with both cosolvents, although the effect of ethanol is slightly greater than that of acetone. For example, a t 75 bar, ethanol at 10 wt % shifts the cloud-point curve to higher temperatures by ca. 30"C, and at 40 wt % ethanol shifts the cosolvent-free cloud-point curve by ca. 85°C. Equal amounts of acetone shift the cloud-point curves by ca. 25 and ca. 65"C, respec- tively. It is not surprising that both cosolvents have similar effects in F22, since F22 hydrogen bonds to methyl acrylate and to both acetone and ethanol. This is consistent with the study of Joesten and S ~ h a a d , ~ ~ which shows that F22 hydrogen bonds with ethyl acetate, a molecule that is structurally similar to methyl acrylate. This means that F22, which is in excess compared to the cosolvent con- centration, will hydrogen bond to PMA and reduce the impact of ethanol. Also, any "free" ethanol will hydrogen bond to F22, which reduces the possibility
COSOLVENCY EFFECTS 435
. . . . .
- 200 / FLUID / 1
60 80 100 120 140 160 180
Figure 5. Effect of ethanol and acetone on phase be- havior of poly ( methyl acrylate) in chlorodifluoromethane. The polymer concentration is 5 wt %.
of ethanol to self-associate at high ethanol concen- trations and to exhibit "antisolvent" characteris- tics."
Figure 5 shows the effect of cosolvents on the phase behavior of EMA69/31 in F22. The large dif- ference between the solvent power of propane and F22 is demonstrated by comparing the cloud-point curve for the EMA69/31-F22 system with that for EMA69/31-propane system in Figure 3. Although EMA69/31 readily dissolves in F22 at temperatures below ca. 140°C and at pressures below ca. 300 bar, this copolymer is not soluble in propane below 135°C and 1800 bar.
Both ethanol and acetone increase the single- phase region of the EMA69/31 system by shifting the cloud-point curves of the EMA6g/31-F22 systems to higher temperatures in a manner similar to their effect with PMA. The effect of acetone on the phase behavior of EMA69/31 is also slightly weaker than that of ethanol. It is interesting that the magnitude of the effect of both cosolvents is virtually the same for both EMA69/31 and PMA even though EMA69/31 contains 69 mol % ethylene repeat units.
It was not possible to dissolve EMAWIlo or poly- ethylene in F22, with or without cosolvent, a t 2000 bar and 140°C. Since F22 is a very dense solvent even in its critical region, the lack of solubility of polyethylene-rich polymers in F22 is probably a re-
sult of the high, positive interchange energy asso- ciated with replacing F22-F22 interactions with polymer-F22 interactions.
One of the main differences in cosolvent contri- butions in propane compared to F22 is that ethanol will self-associate in propane and cause the polymer to fall out of solution, while the amount of self-as- sociation is limited in F22 because ethanol can hy- drogen bond with the excess F22 in solution. A non- linear relationship exists between the number of po- lar sites in the copolymer and the cloud-point pressure of that copolymer in a polar or nonpolar solvent. Adding small amounts of methyl acrylate to polyethylene-rich copolymer causes the copoly- mer to become very polar, and shifts the cloud-point curves with propane to high temperatures and very high pressures. At the other end of the copolymer composition spectrum, it was possible to dissolve EMA69/31 in polar F22 even though two thirds of the polar groups in the backbone of poly (methyl acry- late) were replaced with nonpolar ethylene groups. If the number of methyl acrylate groups was further decreased to 10 mol %, the copolymer became in- soluble in F22.
MODELING
The Sanchez-Lacombe equation of state28 is used to model the phase behavior of the homopolymer and copolymer-solvent-cosolvent systems used in this study. Since the Sanchez-Lacombe equation of state does not directly account for hydrogen bonding, only polymer-propane-acetone mixtures will be modeled. The pertinent equations are given in detail elsewhere.24 Only the basic procedure for the mod- eling is described here.
The experimental data to be modeled in this study consist of two-phase systems at an overall polymer concentration of 5 wt %. The thermodynamic equa- tions that need to be solved in systems consisting of three components and two phases are
where p represents the chemical potential, subscript i represents the polymer and the two solvents, and the superscripts L and u represent the liquid and the gas phases, respectively. The Sanchez-Lacombe equation is used to derive the chemical potential along with the following mixing rules. The close- packed molar volume ugix is
436 HASCH ET AL.
with the cross term u; given as an arithmetic mean of the two pure component characteristic volumes,
where q,, is a fitted mixture parameter. We consider this parameter as a rough measure of the free volume differences between the mixture components. The volume fractions @i and 4j are defined as
Table 111. Characteristic Pure-Component Parameters for the Solvents and Polymers Used in This S t ~ d y ’ ~ , ~ ~
T* P* P* Component (K) (g/cm3) (bar)
Propane C hlorodifluoromethane Ethanol Acetone Polyethylene
where m is the mass fraction and p is the charac- teristic close-packed mass density of a mer. The mixing rule for the characteristic interaction energy egix of a mer of mixture is
with
(12) * * 0.5 t; = ( t i i t j j ) (1 - kij)
where k, is a mixture parameter that is interpreted as a correction factor for the specific binary inter- actions between components i and j not accounted for in the simple geometric-mean average of e ; .
The mixing rule for the number of sites r,ix oc- cupied by a molecule of the mixture is given by
where ri represents the number of sites a molecule occupies in the lattice.
At equilibrium, the chemical potential of each component present in each of the phases must be the same in each phase. If the polymer fractions were truly “monodisperse,” the cloud point for the ternary mixture would be the intersection of the two- phase dome on the ternary phase diagram at the overall mixture c~ncentration.~’-~~ However, cloud points are calculated, ignoring molecular weight distributions since properly accounting for the dis- tributions is expected to be of secondary importance relative to the inherent limitation of the equation of state to account accurately for polar interactions.
It should also be noted that the calculated cloud- point curves are very sensitive to the values used for the pure component parameters of the copoly- mer, which are estimated. The complete pressure- temperature trace of a given cloud-point curve for a ternary mixture is obtained by calculating ternary phase diagrams over a range of pressures and tem- peratures.
The pure-component characteristic parameters for each of the components are given in Table 111. Before calculating the cloud points for ternary mix- tures, it is necessary to determine the binary mixture parameters ki, and qi,, for each of the three binary mixtures that comprise the ternary mixture. These binary mixture parameters are compiled in Table IV. The two mixture parameters for the propane- acetone system are obtained by determining the best fit of the vapor-liquid equilibrium data shown in Figure 6.32 The binary interaction parameters used to obtain this fit are k, = 0.030 and qg = 0. The results in Figure 6 show that the Sanchez-Lacombe equation does only a fair job representing the phase behavior of this system even with two mixture pa- rameters. The calculated critical mixture pressures are about 6% higher than the experimental values, and the predicted and experimental critical mixture compositions differ by approximately 20%. The equation of state does a reasonable job matching the bubble point line; however, the dew point line is overestimated particularly in the region close to the mixture critical point. It is not surprising that the Sanchez-Lacombe equation of state breaks down in the critical region since this is a mean-field theory. If ki, is decreased from 0.030 to zero, the two-phase region is shifted to higher propane concentrations and it is decreased by lowering the bubble point pressures slightly. Increasing the value of qij from zero to 0.030 has the same effect on the calculated curves as decreasing kij.
COSOLVENCY EFFECTS 437
Table IV. Phase Equilibrium Calculations
Mixture Parameters Used for the Polymer-Propane-Acetone
Propane Acetone Polyethylene EMAw/io EM&g/a
Propane Acetone
Propane Acetone
kij
0.030 0.000 -0.020 0.023 0.043 0.034 0.000
Iij
0.000 See Figure 7 See Figure 7 -0.002 0.000 0.000 0.000
The two mixture parameters for the polymer- propane systems were determined by fitting cloud- point data as described in a previous p~b1ication.l~ For the EMA69/31-propane system, a good fit of the cloud-point curve is obtained with constant values of ki, and q i j of 0.023 and -0.002, respectively. For the EMAw/lo-propane system, it is possible to obtain a good fit of the cloud-point curve if kG is set equal to -0.020 and q i j is permitted to vary with temper- ature as shown in Figure 7. For the polyethylene- propane system, kij is set equal to zero, and qij is permitted to vary with temperature as also shown in Figure 7. We expected the value of ki, to be very close to zero for the ethylene-rich polymers because propane is structurally similar to an ethylene repeat unit. Hence, the magnitude of segment-segment, segment-propane, and propane-propane interaction energies will be similar. However, a nonzero, tem- perature-dependent value of qi, is expected for these ethylene-rich polymers based on packing arguments and the thermal expansion differences between each
10 p - 0.0 0.2 0.4 0.6 0.8 1.0
Prooane. Mole Fraction
Figure 6. Phase behavior of the acetone-propane sys- tem. Symbols represent experimental data, and solid lines represent calculations with the Sanchez-Lacombe equa- tion of state with k , = 0.030 and qi, = 0.000.
of the polymers and propane. Note that the two curves in Figure 7 are somewhat parallel.
Unfortunately, it is not possible to determine values for the two mixture parameters for the poly- mer-acetone mixtures studied here because, to the best of our knowledge, there are no binary data available in the literature for these mixtures. To minimize the number of adjustable parameters, q~ was set equal to zero for all of the polymer-acetone systems. Since EMA69/31 is slightly soluble in ace- tone at atmospheric pressure and elevated temper- atures, ki, was set equal to zero. With these param- eter values the model predicts that EMA69/31-ace- tone will be miscible a t elevated pressures in the temperature range investigated in this study. Con- versely, EMAgo/lo and polyethylene are only slightly soluble in highly polar acetone even at elevated temperatures and pressures. Therefore, the value of kij for these two polymers in acetone was adjusted until two phases were predicted in the tempera- ture and pressure ranges investigated. For the
Figure 7. Temperature dependence of ~ i j for the poly- ethylene-propane and EMAgol lo-propane systems. For EMAWll0, k , is set equal to -0.020. For polyethylene, kij is set equal to zero.
438 HASCH ET AL.
800- rn FLUID -
. 4 2 . 3 ~ 1 %
LIQUID i LIQUIU 0 26.9 200 -
90 100 110 120 130 TEMPERATURE ("C)
Figure 8. Phase behavior of the poly (ethylene)-pro- pane-acetone system. Symbols represent experimental data, and solid lines represent calculations using the San- chez-Lacombe equation of state. The mixture parameters used in the calculations are given in Table IV.
EMAsO,lo-acetone system kij was set equal to 0.034; for the polyethylene-acetone system k , was set equal to 0.043.
Figure 8 shows the predicted cloud-point curves for the polyethylene-propane-acetone system. Note that the cloud-point curves predicted for both zero and 26.9 wt % acetone are in fair agreement with experimental data. However, the Sanchez-Lacombe equation of state predicts that the cloud-point curve decreases in pressure at 42.3 wt % acetone when, in fact, experimentally the curve shifts to higher pres- sures. Even with other values of k, for polyethylene- acetone, it is not possible to model all three curves in Figure 8, which suggests that the Sanchez-La- combe equation of state can handle the density en- hancement attributed to the liquid cosolvent, but that it cannot handle the cosolvent polar interac- tions, even with temperature-dependent parameters.
1000
800 m 10.4 FLUID
5.1 wt% \ 10.4~1% >
4 1 . 4 ~ 1 % mm 0 LIQUID+ LIQUID
60 70 80 yo iuo 110 120 130 TEMPERATURE ("C)
Figure 9. Phase behavior of the poly(ethy1ene-co- methyl acrylate ) (90 mol % / 10 mol % ) -propane-acetone system. The symbols represent experimental data, and the solid lines represent calculations using the Sanchez- Lacombe equation of state. The mixture parameters used in the calculations are given in Table IV.
Figure 10. Phase behavior of the poly(ethy1ene-co- methyl acrylate) (69 mol%/31 mol % )-propane-acetone system. The symbols represent experimental data, and the solid lines represent calculations with the Sanchez- Lacombe equation of state. The mixture parameters used in the calculations are given in Table IV.
Figure 9 shows the predicted cloud-point curves for the EMASO/lO-propane-acetone system. The cloud-point curve without acetone is well repre- sented. In addition, the model appears to predict the cloud-point pressures at 41.4 wt % acetone quite well. However, the predicted rate of decrease in cloud-point pressure with acetone concentration is essentially linear when, in fact, the rate of decrease quickly diminishes as acetone concentration in- creases above ca. 5 wt %. Adding acetone to the co- polymer-propane mixture increases the density of the solvent, which shifts the cloud-point curve to lower pressures and increases the number of polar interactions between acetone and the acrylate repeat units in the copolymer, which initially also shifts the cloud-point curve to lower pressures, but even- tually shifts the curve to higher pressures as the number of acetone-acetone interactions dominate the interchange energy. On the basis of the results shown in Figure 8, the Sanchez-Lacombe equation is expected to account for the density effect but not the polarity effect. It is probably fortuitous that the calculated curve for 41 wt % acetone is in agreement with experimental data since the location of the ex- perimental cloud point curve in P-T space is a result of both density and polarity effects while the cal- culated curve is predominantly a result of a density effect.
Figure 10 shows the results for the EMA6s/31- propane-acetone phase equilibrium calculations. Again, we find that the calculated cloud-point curves at high and low concentrations of acetone are in reasonable agreement with the experimental curves. However, the predicted cloud-point pressures for the intermediate concentration of acetone, 5.9 w t %, fall below the observed curve.
COSOLVENCY EFFECTS 439
CONCLUSIONS
The addition of a polar cosolvent to ethylene-methyl acrylate copolymer-solvent mixtures enlarges the region of miscibility in both polar and nonpolar sol- vents. While a larger effect per mole of cosolvent can be observed when a hydrogen-bonding cosolvent is used, excess quantities of hydrogen-bonding co- solvent added to the mixture will cause the polymer to precipitate out of solution. The magnitude of the cosolvent effect is related to the number of polar sites in the copolymer, particularly with hydrogen- bonding cosolvents. However, the effect of a hydro- gen-bonding cosolvent is mitigated if a hydrogen- bonding solvent is used.
Because of the strong polar interaction between acetone molecule and a polar copolymer segment, it is difficult to model the data using a mean-field equation of state with random-mixing rules. Inter- estingly, cloud-point data can be modeled reasonably well a t zero and at high cosolvent concentrations, but the model breaks down at intermediate cosolvent concentrations, where the specific cosolvent-polar site interactions have a larger effect on the phase behavior than is predicted by the equation of state. Attempts to model polymer-solvent-cosolvent phase behavior with hydrogen-bonding cosolvents have not been addressed, and will require a modification of the equation of state such as that proposed by Pan- ayiotou and Sanchez.34
The authors acknowledge partial support of this project by the National Science Foundation under Grant CTS- 9122003. The authors also thank the reviewers of the manuscript, who provided technical insight that has made this a better paper.
REFERENCES AND NOTES
1.
2.
3.
4.
5.
6.
7.
8.
J. M. G. Cowie and J. T. McCrindle, Eur. Polym. J., 8,1185 (1972). J. M. G. Cowie and J. T. McCrindle, Eur. Polym. J., 8,1325 (1972). J. M. G. Cowie and I. J. McEwen, Macromolecules, 7 , 291 (1974). J. M. G. Cowie and I. J. McEwen, Polymer, 24,1449 (1983). 1. A. Katime, P. M. Sasia, and B. Eguia, Eur. Polym. J., 24,1159 (1988). M. D. Blanco, J. M. Teijon, and I. A. Katime, Eur. Polym. J., 26,249 (1990). M. Palaiologou, F. Viras, and K. Viras, Eur. Polym. J., 2 4 , 1191 (1988). F. Viras and K. Viras, J. Polym. Sci.: Part B: Polym. Phys., 2 6 , 2525 ( 1988).
9. B. A, Wolf and G. Blaum, Polym. Sci.: Part B: Polym. Phys., 13,1115 (1975).
10. J. M. G. Cowie and I. J. McEwen, J. Chem. Soc., Far- aday Trans., 70 , 171 (1974).
11. B. A. Wolf and G. Blaum, Makromol. Chem., 177 , 1073 (1976).
12. M. T. Ratzsch, P. Wagner, Ch. Wohlfarth, and D. Heise, Acta Polymerica, 33 , 463 (1982).
13. M. Ratzsch, R. Findeisen, and V. S. Sernov, Z. Phys. Chem., Leipzig, 2 6 1 , 995 (1980).
14. Ch. Wohlfarth, P. Wagner, M. T. Ratzsch, and S. Westmeier, Acta Polymerica, 33 , 468 ( 1982).
15. M. T. Ratzsch, Ch. Wohlfarth, and S. Gleditzsch, Acta Polymerica, 34 , 340 ( 1983).
16. Ch. Wohlfarth, P. Wagner, D. Glindemann, M. Volk- ner, and M. T. Ratzsch, Acta Polymerica, 35 , 498 (1984).
17. B. M. Hasch, M. A. Meilchen, S.-H. Lee, and M. A. McHugh, J. Polym. Sci.; Polym. Phys. Ed., in press.
18. M. A. Meilchen, B. M. Hasch, S.-H. Lee, and M. A. McHugh, Polymer, 33 , 1922 (1992).
19. R. C. Reid, J. M. Prausnitz, and B. E. Polling, The Properties of Gases and Liquids, 4th ed., McGraw-Hill, New York, 1987.
20. R. M. Masegasa, I. Hernandez-Fuentes, E. A. Ojalvo, and E. Saiz, Macromolecules, 12, 862 (1979).
21. K. J. Miller and J. A. Savchik, J. Am. Chem. Soc., 101 , 7206 (1979).
22. G. Allen and C. H. Baker, Polymer, 6,181 (1965). 23. C. A. Irani and C. Cozewith, J. Appl. Polym. Sci., 31 ,
1879 (1986). 24. M. A. Meilchen, B. M. Hasch, and M. A. McHugh,
Macromolecules, 24,4874 ( 1991). 25. J. M. Prausnitz, R. N. Lichtenthaler, and E. G. de
