Fluid Phase Equilibria 168 Ž2000. 165–182 www.elsevier.nlrlocaterfluid Vapor–liquid equilibrium in polymer–solvent systems with a cubic equation of state Vassiliki Louli, Dimitrios Tassios ) Laboratory of Thermodynamics and Transport Phenomena, Department of Chemical Engineering, National Technical UniÕersity of Athens, 9, Heroon Polytechniou Str., Zographou Campus, Athens 15780, Greece Received 8 July 1999; accepted 9 December 1999 Abstract A cubic equation of state ŽEoS., the Peng–Robinson ŽPR. one, is extended to polymers by using a single set of energy Ž a. and co-volume Ž b . parameters per polymer fitted to experimental volume data. Excellent results for the volumetric behavior of the polymer up to 2000 bar pressure are obtained. The EoS is applied to the correlation of low pressure vapor–liquid equilibrium ŽVLE. data for a variety of polymer solutions by employing only one interaction parameter in three mixing rules: the van der Waals one fluid ŽvdW1f. — using the Berthelot combining rule ŽB. for a i j , and the arithmetic mean ŽAM. and Lorentz ŽL. ones for bi j — the Zhong and Masuoka ŽZM.; and the modified Huron–Vidal 1st order ŽMHV1., by coupling the cubic EoS with the Flory–Huggins ŽFH. model. Best results are obtained with the ZM mixing rule and the same holds for the extrapolation with respect to temperature and polymer molecular weight ŽMW.. Low errors in correlation and extrapolation are also obtained with the other mixing rules, but phase split is observed in certain cases. The AbbreÕiations: AM, arithmetic mean combining rule; a-PP, atactic polypropylene; B, Berthelot combining rule; cC6, cyclohexane; EB, ethylbenzene; EoS, equation of state; GC-Flory, group contribution Flory ŽEoS.; GCLF, group contribution lattice fluid ŽEoS.; HDPE, high-density polyethylene; i-PP, isotactic polypropylene; KHFT, method of Kontogeorgis et al.; LDPE, low-density polyethylene; L, Lorentz combining rule; MHV1, modified Huron–Vidal 1st order mixing rule; MW, molecular weight; MEK, methyl–ethyl–ketone; nC x, n-alcane with x carbon atoms; NP, number of data points; OS, method of Orbey and Sandler; OBC, method of Orbey et al.; PBD, cis-1,4-polybutadiene; PcHMA, polyŽcyclohexyl methacrylate.; PDMS, polyŽdimethylsiloxane.; PEA, polyŽethyl acrylate.; PEMA, polyŽethyl methacrylate.; PEO, polyŽethylene oxide.; PIB, polyisobutylene; PMA, polyŽmethyl acrylate.; PMMA, polyŽmethyl methacrylate.; PMP, polyŽ4-methyl1-pentene.; PoMS, polyŽ o-methylstyrene.; PR, Peng–Robinson; PS, polystyrene; PSCT, perturbed soft chain theory; PVAc, polyŽvinyl acetate.; PVC, polyŽvinyl chloride.; PVME, polyŽvinyl methyl ether.; SAFT, statistical associating fluid theory; vdW, van der Waals; vdW1f, van der Waals one fluid mixing rule; VLE, vapor–liquid equilibrium; ZM, Zhong and Masuoka mixing rule ) Corresponding author. Tel.: q30-1772-3232; fax: q30-1772-3155. 0378-3812r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 3 8 1 2 Ž 9 9 . 0 0 3 3 9 - 8 166 V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 performance of other methods for polymer parameters evaluation is also examined. q 2000 Elsevier Science B.V. All rights reserved. Keywords: Cubic equation of state; Mixing rule; Vapor–liquid equilibrium; Polymer solutions 1. Introduction In recent years, the phase equilibrium in polymer solutions has attracted the interest of the scientific community due to their fundamental importance in a number of process applications, such as polymer devolatilization, polymeric membrane–separation processes, vapor phase–photografting, optimum formulation of paints and coatings, etc. w1x. In order to design such processes, the successful description of the vapor–liquid equilibrium Ž VLE. behavior in solvent–polymer systems is required. This need has led to the development of several activity coefficient models, which are mainly modifications of the Flory–Huggins Ž FH. equation w2,3x Ž e.g., Entropic-FV w4x, UNIFAC-FV w5x, etc.. , and various equations of state Ž EoS; e.g., group contribution Flory wGC-Floryx w6x, group contribution lattice fluid wGCLFx w7x, perturbed soft chain theory wPSCTx w8x, statistical associating fluid theory wSAFTx w9x, Sanchez–Lacombe w10x, Panayiotou–Vera w11x, etc.. . The latter have the advantage of being applicable to both vapor and liquid phases in contrast to the former models. However, in the last few years, cubic EoS have been extensively used as well, because they provide accuracy comparable to the other models and their simplicity makes them attractive to the practicing engineer. The first attempt to apply a cubic EoS in polymer solutions has been made by Sako et al. w12x, who introduced a three-parameter cubic EoS, where apart from the energy Ž a. and co-volume Ž b . parameters, Prigogine’s external degrees of freedom parameter c is employed for the extension of the equation to polymers. However, the evaluation of these parameters is quite complicated for both the solvent and the polymer and thus, this approach does not retain the simplicity of a cubic EoS. On the other hand, Kontogeorgis, Harismiadis, Fredenslund and Tassios Ž KHFT. w13x proposed in 1994 the use of the van der Waals ŽvdW. EoS with a very simple method for the calculation of the attractive and repulsive parameters of the EoS. According to this method, the a and b of the polymer are fitted to two experimental volumetric data at essentially zero pressure. The extension to polymer solutions was achieved by using the classical van der Waals one fluid mixing rule Ž vdWf1. , the arithmetic mean ŽAM. combining rule for bi j and the Berthelot one for a i j with only one binary interaction parameter, l i j . In a series of papers, this approach was applied with satisfactory results to: Ž1. the correlation and prediction of VLE w13,14x; Ž2. the correlation and prediction of LLE in polymer solutions and blends w15–17x; and Ž 3. the correlation and prediction of Henry constants w18x. This simple method, however, has the disadvantage of predicting rather poor volumetric behavior at high pressures Ž as shown in Fig. 1a and b for polyisobutylene wPIBx and polystyrene wPSx, respectively. and unrealistically high vapor pressures Ž as shown in Fig. 2. for the pure polymer. The latter causes problems in the correlation of polymer–solvent VLE data involving small molecular weight ŽMW. polymers, as it leads to the appearance of polymer in the vapor phase. Parameter values with this method and the Peng–Robinson ŽPR. EoS are presented in Table 1 for five polymers. Orbey and Sandler ŽOS. w19x, in 1994, fitted the two parameters of the PRSV EoS for the pure polymer to its volumetric data at the temperature range of interest by assuming that its vapor pressure is equal to 10y7 MPa. For each MW, a different set of parameters is obtained Ž see Table 1. , which makes the method time-consuming, while the predicted volumetric behavior is rather poor Ž see Fig. 1a V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 167 Fig. 1. Ža. Percent error in volume Ž DV %. of PIB Ž50,000. vs. pressure at 338.15 K, for all the methods considered in this study for polymer parameters evaluation ŽOBC is not included, because N DV %N )90%.. Žb. Percent error in volume Ž DV %. of PS Ž290,000. vs. pressure at 395.15 K, for all methods considered in this study for polymer parameters evaluation ŽOBC is not included, because N DV %N )90%.. and b. and the vapor pressures are rather high, especially at high temperatures Ž Fig. 2. . In order to correlate the VLE in polymer solutions, they employed the mixing rule proposed by Wong and Sandler w20x by coupling the EoS with the FH G E model, and employing two binary interaction Fig. 2. Predicted vapor pressure of PIB Ž50,000. vs. temperature, for all methods considered in this study for polymer parameters evaluation. 168 PIB ŽT range: 326.15–383.15 K. PS ŽT range: 388.15–469.15 K. PEO ŽT range: 361.15–497.15 K. arMW brMW MW arMW brMW MW 1170 1350 2700 40,000 45,000 50,000 66,500 100,000 10 6 2.25=10 6 All MWs 848,250 843,750 823,500 748,000 744,750 742,500 733,495 722,900 654,700 629,550 450,893 1.0350 1.0340 1.0320 1.0230 1.0224 1.0221 1.0215 1.0210 1.0160 1.0120 0.9614 870,350 858,005 830,550 823,725 820,752 813,078 784,160 0.9260 0.9220 0.9200 0.9180 0.9175 0.9160 0.9120 600,000 750,000 0.9020 194,000 580,000 0.8420 14,600 770,000 0.8902 454,347 0.8486 351,985 0.7863 287,543 0.7197 376,848 0.8510 a Žcm6 barrmol 2 . and b Žcm3rmol.. brMW MW arMW PVME ŽT range: 303.15–471.15 K. MW 10,300 15,700 49,000 63,000 70,300 97,200 290,000 arMW PVAc ŽT range: 308.15–373.15 K. brMW MW arMW brMW V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 Table 1 PR EoS parameters for the pure polymer calculated with Kontogeorgis et al. Ž italic . and OS methods Žvolumetric data from Tait equation with the Rodgers parameters w26x. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 169 parameters, k i j and x ŽFlory interaction parameter.. Recently, Zhong and Masuoka w21x simplified the aforementioned mixing rule by setting A`E equal to zero and obtained satisfactory results by employing the method of Orbey and Sandler for polymer parameters a and b and only one binary interaction parameter, i.e., k i j . Kalospiros and Tassios w22x proposed a method of calculation of polymer parameters a and b based on the free–volume theory w23,24x, and employed a predictive EoS-G E model, which coupled the t-mPR EoS with Entropic-FV G E model w4x. But although they were led to good results and low vapor pressure, P s , for pure polymer, the necessity of properties like the glass transition temperature of the polymer, its volume at this temperature, and its thermal expansion coefficient in the glassy state makes the method not very practical. Finally, Orbey, Bokis and Chen Ž OBC. w25x suggested the use of a common set of critical properties for all polymers, which are independent of the polymer MW Ž Tc polymer s 1800 K, Pc polymer s 10 bar.. They employed the modified Huron–Vidal 1st order ŽMHV1. mixing rule by coupling the SRK EoS with the FH model, and obtained very satisfactory results in the description of VLE behavior. The predicted volumetric behavior of the polymer is very poor, however, with very large errors, over 50%, for high MWs Žthis is why it is not included in Fig. 1a and b. . The predicted P s values are also high, especially at high temperatures, as shown in Fig. 2. In this study, we evaluate the polymer parameters a and b of a cubic EoS, in this case, the PR one, by fitting them to pure polymer PVT data and assuming that arMW and brMW are independent of MW. We proceed then with the correlation of VLE data for a great variety of non-polar and polar polymer–solvent systems using three different mixing rules, requiring only one adjustable parameter. Table 2 PR EoS parameters of some representative polymers calculated with the method proposed in this work Žvolumetric data from Tait equation with the Rodgers parameters w26x. Polymer T range ŽK. P range Žbar. arMW brMW AAE% in V HDPE LDPE PS PVAc PMMA PIB PEO PVME PMA PEA PEMA PVC PoMS PcHMA PBD PDMS PMP i-PP a-PP 413.15–476.15 394.15–448.15 388.15–469.15 308.15–373.15 387.15–432.15 326.15–383.15 361.15–497.15 303.15–471.15 310.15–493.15 310.15–490.15 386.15–434.15 373.15–423.15 412.15–471.15 396.15–471.15 277.15–328.15 298.15–343.15 514.15–592.15 443.15–570.15 353.15–393.15 0–1960 0–1960 0–2000 0–800 0–2000 0–1000 0–685 0–2000 0–1960 0–1960 0–1960 0–2000 0–1800 0–2000 0–2835 0–1000 0–1960 0–1960 0–1000 1,280,756 1,373,984 1,315,409 1,847,343 1,277,496 2,307,400 2,278,342 1,219,650 1,132,607 1,030,010 1,185,273 998,549 1,414,377 1,234,992 942,530 1,021,986 2,645,412 1,292,409 1,627,639 1.2066 1.1991 0.9549 0.8428 0.8407 1.0882 0.9497 0.9801 0.8664 0.9037 0.9050 0.7154 0.9746 0.9085 1.0549 0.9968 1.3607 1.2444 1.1716 2.77 2.62 2.17 1.38 1.77 1.40 2.48 2.92 3.11 3.34 2.15 1.85 1.95 2.14 2.35 2.10 2.47 3.65 1.77 a Žcm6 barrmol 2 . and b Žcm3rmol.. AAE%sÝabsŽ Vcal yVexp .r Vexp r NP =100. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 170 The possibility of extrapolation with respect to temperature, polymer MW and solvent concentration is also discussed. Finally, results with the aforementioned methods of polymer parameter estimation ŽKHFT, OS and OBC. are included for comparison purposes. 2. Evaluation of EoS parameters for polymers In this study, the evaluation of the energy Ž a. and co-volume Ž b . parameters of the pure polymer in the PR EoS: RT a Ps y 2 , Ž1. Vyb V q 2 bV y b 2 is obtained by fitting the available PVT data with a single set of Ž arMW. and Ž brMW. parameters for all MWs. The source of the required PVT data is the Tait equation, with the parameter values proposed by Rodgers w26x. Zhong and Masuoka w27x used a similar approach but their parameters were MW-dependent. Parameter values for several polymers, along with the percent absolute average error in volume ŽAAE%. , are presented in Table 2, while typical volume errors Ž DV %. as a function of pressure are presented graphically in Fig. 1a and b, for PIB and PS. Very satisfactory results are obtained, which are as expected better than those of the aforementioned methods Ž KHFT, OS, OBC. . Furthermore, the proposed method predicts reasonably low P s values ŽFig. 2.. 3. Correlation of VLE data 3.1. SolÕent parameters The PR EoS parameters of the solvents examined in this study are determined from: a s 0.45724 b s 0.0778 Ž RTc . Pc RTc Pc 2 a Ž Tr . , Ž2. , Ž3. where Tc and Pc are the pure compound critical temperature and pressure, respectively, obtained from Daubert and Danner w28x, and Tr is the reduced temperature Ž TrTc . . For the quantity a Ž Tr . , two cases can be distinguished depending on the polarity of the solvent. For non-polar compounds, the a Ž Tr . is related to the acentric factor v : ( / a Ž Tr . s 1 q m 1 y Tr ž 2 , Ž4. m s 0.384401 q 1.52276 v y 0.213808 v 2 q 0.034616 v 3 y 0.001976 v 4 . Eq. 5 was proposed by Magoulas and Tassios w29x and provides better predictions at low P s. For polar compounds, the Mathias–Copeman w30x expression is used: ( / ( / a Ž Tr . s 1 q c1 1 y Tr q c2 1 y Tr ž ž 2 ( / q c3 1 y Tr ž 3 2 , Ž5. Ž6. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 171 where the values of c1, c 2 , c 3 are evaluated by fitting the pure compound vapor pressure data obtained from the correlation of such data by Daubert and Danner w28x. In this study, the c1, c 2 , c 3 values reported by Kalospiros and Tassios w22x were employed. 3.2. Mixing rules In order to extend the PR EoS to polymer–solvent systems, the following mixing rules are considered. 3.2.1. ÕdW1f mixing rule am s Ý Ý x i x j ai j i bm s Ý j i Ý x i x j bi j , Ž7. j with the combining rules for the cross energy Ž a i j . and co-volume Ž bi j . parameters below: Berthelot rule ŽB.: ( ai j s ai a j bi j Ž1 y k i j . , (b b Ž 7.1 . i j AM rule: bi j s bi q bj , 2 Lorentz Ž 1r3. rule ŽL. : bi j žb s 1r3 q bj1r3 i Ž 7.2 . 3 / , Ž 7.3 . 8 where x iŽ j. is the mole fraction and k i j is the binary interaction parameter. Note that, as Harismiadis et al. w14x have explained, the Berthelot combining rule is preferable to the geometric mean for this type of systems. 3.2.2. ZM mixing rule am RT where: sQ Qs Ý i DD bm s 1 y DD Ý xi x j j ž a by RT / Q 1 y DD , ij , Ž8. DD s Ý x i i ai bi RT , and: a ž b y RT / 1 s ij 2 ai aj ž b y RT / q ž b y RJ / Ž1 y k . . i j ij Ž 8.1 . Actually, the only difference between this mixing rule and the Wong–Sandler one is the absence of excess Helmholtz free energy at infinite pressure, A`E , which was set equal to zero in this case. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 172 Table 3 Overall correlation results for various polymer–solvent systems ŽExperimental data are taken from Wen et al. w32x. The number of the experimental data is given in the parenthesis. Polymer Solvent T range ŽK. Overall AAD% vdW1f ŽBrAM. vdW1f ŽBrL. ZM MHV1 PIB Ž4=10 4 –2.25=10 6 . nC4, nC5, nC6, nC8 Ž122. 298–338 Ž4=10 4 –10 5 . cC6, benzenea , toluenea , EB a Ž129. 298–338 Average error 2.07 3.81 2.94 2.37 3.07 2.72 2.48 3.29 2.00 5.60 2.24 4.45 PIB (1170, 1350, 2700) nC5 (42) 298 – 328 8.72 35.22 3.92 6.05 PS Ž10,300–2.9=10 5 . cC6, benzene, toluene, EB Ž111. acetone, MEK, CHCl3 Ž61. 288–353 298–343 0.74 4.43 2.59 1.00 3.16 2.08 0.90 1.10 3.52 2.84 2.21 1.97 PEO Ž5700–4=10 6 . Ž6000, 5=10 6b . Average error benzene Ž50. CHCl 3 Ž12. 319–424 298–333 1.88 5.40 3.64 4.33 3.56 3.95 1.41 2.45 6.37 5.94 3.89 4.20 PVAc Ž194,000. Ž50,500–158,000. Average error CHClb3 Ž7. benzenea Ž33. 333 303–333 4.96 1.98 3.47 5.53 2.22 3.88 5.04 4.91 1.78 1.98 3.41 3.45 PDMS Ž89,000. nC6 a Ž24. CHCl 3 Ž8. 298–308 303 0.80 1.00 0.90 8.85 10.8 9.83 0.89 2.48 1.56 0.57 1.23 1.53 PVC Ž34,000. toluenea Ž8. 316 4.95 3.29 5.65 0.91 PVME Ž14,000. Ž14,000, 14,600. Average error CHCla3 Ž13. benzenea , toluenea Ž60. 298 298–343 4.28 0.65 2.47 5.67 1.79 3.73 5.97 5.01 0.63 1.18 3.30 3.10 Average error Average error a b Experimental data taken from Wohlfarth w33x. Experimental data taken from Gupta et al. w34x. 3.2.3. MHV1 mixing rule a bRT s Ý xi i ai 1 q RTbi A1 ž GE RT q Ý x i ln i b bi / b s Ý x i bi , Ž9. i where G E is the excess Gibbs free energy and A1 s y0.623 for PR EoS. In this case, the cubic EoS is coupled with the FH model w2,3,31x, which is: GE RT s x 1 ln f1 x1 q x 2 ln f2 x2 q xf 1 f 2 Ž x 1 q x 2 z . , Ž 10. with: f1 s x1 x1 q x2 z and f2 s x2 z x1 q x2 z , Ž 11. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 173 where 1 is the solvent, 2 is the polymer, f is the volume fraction and z is the degree of polymerization, i.e., z s MWpolymerrMWsegment w25x. The only interaction parameter here is the x found in the FH model. 4. Results Bubble point pressure calculations are carried out for seven polymers: PIB, PS, polyŽ ethylene oxide. wPEOx, polyŽvinyl acetate.wPVAcx, polyŽ dimethylsiloxane. wPDMSx, polyŽ vinyl chloride. wPVCx and polyŽvinyl methyl ether. wPVMEx with a variety of polar and non-polar solvents, using the aforementioned mixing rules. The VLE data are taken from Wen et al. w32x and Wohlfarth w33x, while those for PEO Ž5,000,000. –chloroform and PVAc Ž 194,000. –chloroform are from Gupta and Prausnitz w34x. In each case examined, the average absolute deviation Ž AAD%s ÝabsŽ Pcal y Pexp .rPexprNP = 100. in bubble point pressure is reported. 4.1. Correlation Overall correlation results using the EoS polymer parameters proposed here are presented in Table 3, while typical ones are shown graphically in Figs. 3–7. The following comments summarize our observations on the obtained results. Ž1. Very satisfactory results are obtained with all mixing and combining rules, with typical errors for non-polar solvents of 2–3% and a little higher for polar ones Ž Table 3, and Figs. 3–5.. Ž2. Satisfactory results are also obtained for the chloroform Ž CHCl 3 . systems, where hydrogen bonding exists, as Table 3 and Fig. 6 indicate. Fig. 3. Correlation of the bubble point pressure for the system PIB Ž50,000. – nC6 Ž k i j vdW1fŽBrAM. s 0.620r0.627r0.640; k i j vdW1fŽBrL. s 0.541r0.543r0.546; k i j ZM s 0.917r0.917r0.922; x s1.068r1.112r1.083 for T s 298.15r313.15r338.15 K, respectively.. 174 V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 Fig. 4. Correlation of the bubble point pressure for the system PVAc Ž158,000. –benzene Ž k i j vdW1fŽBrAM. s 0.612r0.619; k i j vdW1fŽBrL. s 0.542r0.543; k i j ZM s 0.909r0.91; x s 0.667r0.652 for T s 313.15r333.15 K, respectively.. Ž3. Low error in pressure does not, however, guarantee good correlation as suggested by the results obtained with the vdW1f Ž BrAM. mixing rule for PIB-benzene in Fig. 7. Phase split is predicted at Fig. 5. Correlation of the bubble point pressure for the system PS Ž290,000. –methyl–ethyl–ketone wMEKx Ž k i j vdW1fŽBrAM. s 0.521r0.557; k i j vdW1fŽBrL. s 0.449r0.475; k i j ZM s 0.894r0.915; x s 0.853r0.926 for T s 298.15r343.15 K, respectively.. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 175 Fig. 6. Correlation of the bubble point pressure for the systems PEO Ž6000. –chloroform and PVAc Ž194,000. –chloroform Ž k i j vdW1fŽBrAM. s 0.540r0.610; k i j vdW1fŽBrL. s 0.323r0.512; k i j ZM s 0.705r0.870; x sy0.485r0.026, for each system, respectively.. 313.15 and 338.15 K, even though the corresponding AAD% in bubble point pressure are 4.96% and 4.81%, respectively. For this system, the MHV1 mixing rule also predicts phase split. This is Fig. 7. Correlation of the bubble point pressure for the system PIB Ž45,000. –benzene Ž k i j vdW1fŽBrAM. s 0.613r0.624r0.634; k i j vdW1fŽBrL. s 0.557r0.565r0.564; k i j ZM s 0.946r0.949r0.944; x s1.047r1.080r1.034 for T s 298.15r313.15r338.15 K, respectively.. 176 V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 observed for other systems as well, like PIBŽ 10 6 . –nC5 Ž298.15, 313.15 K. and PIBŽ10 5 . –cyclohexane ŽcC6. Ž338.15 K., where both the vdW1f ŽBrAM. and MHV1 mixing rules lead to phase split. No such phase split has been observed with the ZM mixing rule. Ž4. The results for systems involving PIB of very low MW suggest that the vdW1f Ž BrL. mixing rule fails completely, while the ZM mixing rule gives very satisfactory results. Ž5. The x values obtained with the MHV1 mixing rule are very often higher than 0.5, suggesting — in terms of the conventional use of the FH G E model — phase separation. This, of course, is not necessarily the case as this MHV1 model suffers from the same deficiency as the original MHV1 model, where UNIFAC is the G E model: it fails to reproduce the G E model coupled with it, as pointed out by Boukouvalas et al. w35x and explained by Kalospiros et al. w36x. Ž6. The mixing rule of ZM appears to give the best overall results, which is surprising — especially for polar systems — considering the assumption involved in its development. 4.2. Extrapolation We examine next the possibility of using the interaction parameter obtained from a given set of VLE data for extrapolation purposes, since this is of great importance in the case of a process design when a limited number of experimental data are available. Satisfactory results for extrapolation with respect to temperature— for reasonable temperature ranges Žabout 508C. — are obtained, as shown in Table 4 and Fig. 8 for the PEO Ž4 = 10 6 . –benzene Table 4 Prediction of VLE with respect to temperature and polymer MW using the interaction parameter values of the lowest available temperature or MW System T range ŽK. vdW1f ŽBrAM. vdW1f ŽBrL. ki j AAD% ki j AAD% ki j AAD% x PIB Ž2.25=10 6 . – nC5 Ž21. PIB Ž10 5 . –cC6 Ž19. PS Ž49,000. –cC6 Ž24. PS Ž63,000. –benzene Ž20. PS Ž2.9=10 5 . –toluene Ž6. PS Ž2.9=10 5 . –CHCl 3 Ž11. PS Ž15,700. –acetone Ž7. PEO Ž4=10 6 . –benzene Ž23. PVME Ž14,600. –benzene Ž5. PVME Ž14,600. –toluene Ž11. Overall error a 308–328 313–338 303–338 303–333 333–353 323.2 323.2 361–423 343.2 343.2 0.669 0.606 0.484 0.488 0.466 0.505 0.554 0.640 0.444 0.413 6.17 6.98 1.46 10.16 16.28 10.84 13.59 18.14 3.75 6.26 9.36 0.618 0.541 0.335 0.404 0.387 0.412 0.488 0.556 0.299 0.285 2.61 1.78 1.21 3.30 4.23 3.42 2.66 4.82 0.85 2.14 2.70 0.962 0.917 0.787 0.828 0.822 0.816 0.933 0.916 0.713 0.682 2.48 1.99 1.28 4.52 5.21 7.02 1.39 4.03 2.47 3.29 3.37 0.997 0.832 0.215 0.217 0.474 0.084 0.980 1.085 0.466 0.636 5.45 3.89 1.13 3.99 3.24 4.56 2.29 5.19 1.56 2.71 3.40 PIB Ž10 6 , 2.25=10 6 . – nC5 Ž17. PIB Ž5=10 4 , 10 5 . –cC6 Ž21. PS Ž290,000. –toluene Ž11. PVAc Ž150,000. –benzene Ž7. Overall error b 298.2 298.2 298.2 303.2 0.671 0.603 0.457 0.598 6.96 3.02 6.18 3.51 4.92 0.601 0.528 0.372 0.520 4.51 2.82 4.43 3.12 3.72 0.957 0.912 0.817 0.891 5.26 1.85 1.76 3.12 3.00 1.287 0.901 0.344 0.481 11.07 5.25 4.82 1.98 5.78 a ZM MHV1 AAD% For all the systems, the lowest temperature reported is 298 K, apart from PS–benzene Ž288 K., PEO–benzene Ž348 K. and PVAc systems Ž323 K.. b For PIB, the lowest MW system is the one of 4=10 4 , while for PS and PVAc, they are those of 49,000 and 5=10 4 , respectively. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 177 Fig. 8. Prediction of the bubble point pressure for the system PEO Ž4=10 6 . –benzene at various temperatures using the interaction parameter values obtained from the lowest temperature data. system. Best results are given by ZM and vdW1f ŽBrL. mixing rules, since the other two occasionally lead to a phase split, such as the cases of PIB Ž 2.25 = 10 6 . –nC5 Ž313.15 K. and PIB Ž10 5 . –cC6. Fig. 9. Prediction of the bubble point pressure for the system PIB– nC5 at various MWs using the interaction parameter values obtained from the lowest MW. 178 Table 5 Typical correlation results for some representative systems of this study using all the studied methods for determining EoS polymer parameters The number of the experimental data is given in the parenthesis. System Overall error in pressure KHFT vdW1f ŽBrAM. PIB Ž1170,1350, 2700. – nC5 Ž42. PIB Ž50,000. – nC6 Ž24. PIB Ž45,000. –benzene Ž29. PS Ž10,300, 49,000, 290,000. –toluene Ž35. PS Ž290,000. –CHCl 3 Ž22. PEO Ž6=10 5 . –benzene Ž13. PVAc Ž194,000. –CHCl 3 Ž7. PVME Ž14,600. –toluene Ž34. 298–328 298–338 298–338 298–353 298–323 323–343 333 323–343 1.43 a 3.27 4.42 0.73 5.52 0.52 5.03 0.45 OS vdW1f ŽBrL. ZM MHV1 4.00 a 2.99 1.43 0.92 1.37 2.21 7.65 0.77 1.51a 4.21 1.43 1.12 5.99 0.62 5.03 0.46 3.49 a 2.66 6.46 0.64 4.68 2.32 4.89 1.26 This study PIB Ž1170, 1350, 2700. – nC5 Ž42. PIB Ž50,000. – nC6 Ž24. PIB Ž45,000. –benzene Ž29. PS Ž10,300, 49,000, 290,000. –toluene Ž35. PS Ž290,000. –CHCl 3 Ž22. PEO Ž6=10 5 . –benzene Ž13. PVAc Ž194,000. –CHCl 3 Ž7. PVME Ž14,600. –toluene Ž34. a 298–328 298–338 298–338 298–353 298–323 323–343 333 323–343 8.73 2.99 5.70 0.55 5.04 0.29 4.96 0.46 vdW1f ŽBrAM. 3.53 3.08 4.93 0.62 5.25 0.36 4.94 0.45 vdW1f ŽBrL. ZM MHV1 14.17 3.02 1.54 1.18 1.59 1.67 5.50 0.78 1.52 4.19 1.44 1.12 5.98 0.62 5.03 0.46 1.55 2.67 6.50 0.62 4.71 2.40 4.91 1.27 14.00 30.38 11.51 25.55 39.57 35.68 110.7 5.25 2.03 4.26 1.42 1.13 5.99 0.62 5.03 0.45 1.41 2.88 4.31 1.17 6.54 1.04 6.33 0.71 OBC 35.22 2.69 2.06 1.55 1.82 1.48 5.53 1.12 3.92 3.98 1.73 1.12 5.99 0.62 5.04 0.45 6.05 2.77 6.44 0.63 4.78 2.44 4.91 1.20 D y vdW1f ŽBrAM. s 4.1Ey4, D y vdW1f ŽBrL. s 2.0Ey4, D y ZM s 3.1Ey4, D y MHV1 s 3.2Ey4. 4.36 24.62 9.98 24.29 32.53 35.57 82.82 4.05 V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 T range ŽK. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 179 The same observations apply to the extrapolation with respect to polymer MW ŽTable 4., when high values of it are involved Ž) 10,000., where the MW does not affect significantly the VLE Ž Fig. 9.. At last, the extrapolation with respect to solvent concentration from a limited number of data: those in the lowest and those in the highest solvent concentration range may lead to uncertainties. 4.3. The effect of polymer parameters Table 5 presents results for some representative systems using the polymer parameters determined by four methods: the proposed, the KHFT, the OS and the OBC ones, combined with the mixing rules considered here: vdW1f ŽBrAM., vdW1f ŽBrL., ZM and MHV1. The first three methods for polymer parameters determination give similar and good results in all cases, with the exception of the low MW PIB systems. There, the KHFT parameters lead to significant amounts of polymer in the vapor phase for the system PIB–nC5. This is due to the high polymer P s values predicted by them, as shown in Fig. 2, and the low solvent mole fractions — due to the low polymer MW — which make more significant the influence of the polymer EoS parameters to the VLE correlation. That is why this phenomenon is not observed for the high MW polymers. For the same low MW systems, the vdW1f Ž BrL. mixing rule gives high errors in pressure, especially with the parameters of this study. The OBC parameters give poor results Ž high errors in pressure and unrealistic values for the interaction parameter. when used with the vdW1f Ž BrAM. and vdW1f ŽBrL. mixing rules, as shown in Fig. 10 for the system PS Ž290,000. –chloroform. Moreover, their performance deteriorates with increasing polymer MW due to the unrealistic high value of Pc . On the other hand, they give good results with the ZM mixing rule, but the interaction parameters assume unrealistically high absolute Fig. 10. Correlation of the bubble point pressure for the system PS Ž290,000. –chloroform at 298.15 K using the OBC polymer parameters. 180 V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 values ŽFig. 10.. The parameters do provide, however, satisfactory results when used with the MHV1 mixing rule. 5. Conclusions A cubic EoS, in this case the PR one, is applied to the modeling of VLE of polymer–solvent systems. Polymer EoS parameters, arMW and brMW, are obtained by fitting polymer PVT data and provide excellent results in the description of the pure polymer behavior up to very high pressures. Correlation of VLE data for a variety of non-polar and polar systems— including hydrogen bonding ones— is carried out by using three mixing rules: the vdW1f one Ž with the Berthelot combining rule for a i j and the AM and Lorentz combining rules for bi j . ; the one of ZM and the MHV1 proposed by Orbey et al., all involving only one interaction parameter. All mixing rules give low errors in pressure, but phase split may be predicted on occasion with the vdW1f Ž BrAM. and MHV1 mixing rules, while the vdW1f mixing rule with the Berthelot–Lorentz combining rules for a i j and bi j , respectively, fails when very small polymer MWs are encountered. Best overall results are obtained with the ZM mixing rule, especially since no phase split is observed with it. Extrapolation with respect to temperature and polymer MW Ž) 10,000. is satisfactory, especially when the ZM and vdW1f Ž BrL. mixing rules are employed, since the vdW1f Ž BrAM. and MHV1 mixing rules lead on occasion to pressure maxima. Three other methods of evaluating the polymer EoS parameters, those of Kontogeorgis et al. ŽKHFT., Orbey–Sandler Ž OS. and Orbey et al. Ž OBC. , were also considered. They all give poorer PVT predictions than the proposed method, especially the OBC one which considers the same parameters for all polymers, and they also predict substantial polymer P s values. Correlation of binary VLE data, using the mixing rules considered here with the polymer parameter values of KHFT and OS methods, gives satisfactory results, except when very small MW polymers are involved, where the KHFT method may lead to significant polymer amounts in the vapor phase. The OBC parameters give poor results with the vdW1f mixing rule, unrealistic interaction parameter values with the ZM mixing rule, but satisfactory results with the MHV1 one. List of symbols a energy parameter of a cubic EoS Žbar cm6rmol 2 . b co-volume parameter of a cubic EoS Žcm3rmol. c1 , c 2 , c 3 pure compound parameters as defined in Mathias–Copeman expression G Gibbs free energy A Helmholtz free energy ki j binary interaction parameter in Eqs. 7.1 and 8.1 m parameter of Eq. 4 P pressure Ž bar. R universal ideal gas constant Žs 83.14 bar cm3rmol K. T absolute temperature ŽK. V molar volume Žcm3rmol. V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 x, y w 181 mole fraction weight fraction Greek symbols a Ž Tr . energy parameter temperature correction x Flory interaction parameter v acentric factor Superscripts E excess s saturation state Subscripts c cal exp iŽ j . ij m r critical property calculated value experimental value component in a mixture cross-parameter Ždefined by the combining rules. mixture reduced property References w1x w2x w3x w4x w5x w6x w7x w8x w9x w10x w11x w12x w13x w14x w15x w16x w17x w18x w19x w20x w21x w22x w23x R. Gupta, J. Prausnitz, Ind. Eng. Chem. Res. 35 Ž1996. 1225–1230. P.J. Flory, J. Chem. Phys. 9 Ž1941. 660–661. P.J. Flory, J. Chem. Phys. 10 Ž1942. 51–61. G.M. Kontogeorgis, Aa. Fredenslund, D.P. Tassios, Ind. Eng. Chem. Res. 32 Ž1993. 362–372. T. Oishi, J.M. Prausnitz, Ind. Eng. Chem., Process. Des. Dev. 17 Ž1978. 333–339. J. Holten-Andersen, P. Rasmussen, Aa. Fredenslund, Ind. Eng. Chem. Res. 26 Ž1987. 1382. M.S. High, R.P. Danner, AIChE J. 36 Ž1990. 1625. W.O. Morris, P. Vimalchand, M.D. Donohue, Fluid Phase Equilib. 32 Ž1987. 103–115. S.H. Huang, M. Radosz, Ind. Eng. Chem. Res. 29 Ž1990. 2284–2294. I.C. Sanchez, R.H. Lacombe, J. Phys. Chem. 80 Ž1976. 2528. C.G. Panayiotou, J.H. Vera, Polymer J. 14 Ž1982. 681–694. T. Sako, A.H. Wu, J.M. Prausnitz, J. Appl. Polym. Sci. 38 Ž1989. 1839–1858. G.M. Kontogeorgis, V.I. Harismiadis, Aa. Fredenslund, D.P. Tassios, Fluid Phase Equilib. 96 Ž1994. 65–92. V.I. Harismiadis, G.M. Kontogeorgis, Aa. Fredenslund, D.P. Tassios, Fluid Phase Equilib. 96 Ž1994. 93–117. V.I. Harismiadis, G.M. Kontogeorgis, A. Saraiva, Aa. Fredenslund, D.P. Tassios, Fluid Phase Equilib. 100 Ž1994. 63–102. A. Saraiva, G.M. Kontogeorgis, V.I. Harismiadis, Aa. Fredenslund, D.P. Tassios, Fluid Phase Equilib. 115 Ž1994. 73–93. V.I. Harismiadis, A.R.D. van Bergen, G.M. Kontogeorgis, Aa. Fredenslund, D.P. Tassios, AIChE J. 42 Ž1996. 3170–3180. S. Bithas, N.S. Kalospiros, G.K. Kontogeorgis, D.P. Tassios, J. Polym. Eng. Sci. 36 Ž1995. 254–261. N. Orbey, S.I. Sandler, AIChE J. 40 Ž1994. 1203–1209. D.S.H. Wong, S.I. Sandler, AIChE J. 38 Ž1992. 671. C. Zhong, H. Masuoka, Fluid Phase Equilib. 123 Ž1996. 59–69. N.S. Kalospiros, D.P. Tassios, Ind. Eng. Chem. Res. 34 Ž1995. 2117–2124. M.L. Williams, R.F. Landel, J.D. Ferry, J. Am. Chem. Soc. 77 Ž1955. 2701. 182 w24x w25x w26x w27x w28x w29x w30x w31x w32x w33x w34x w35x w36x V. Louli, D. Tassiosr Fluid Phase Equilibria 168 (2000) 165–182 J.J. Aklonis, W.J. MacKnight, in: Introduction to Polymer Viscoelasticity, 2nd edn., Wiley, New York, 1983. H. Orbey, C.P. Bokis, C. Chen, Ind. Eng. Chem. Res. 37 Ž1998. 1567–1573. P.A. Rodgers, J. Appl. Polym. Sci. 48 Ž1993. 1061–1080. C. Zhong, H. Masuoka, Fluid Phase Equilib. 144 Ž1998. 49–57. T.E. Daubert, R.P. Danner, Physical and Thermodynamic Properties of Pure Compounds: Data Compilation, Hemisphere, New York, 1989. K. Magoulas, D.P. Tassios, Fluid Phase Equilib. 56 Ž1990. 119–140. P.M. Mathias, T.W. Copeman, Fluid Phase Equilib. 13 Ž1983. 91–108. M.L. Huggins, Ann. N.Y. Acad. Sci. 43 Ž1942. 431–443. H. Wen, H.S. Elbro, P. Alessi, Polymer Solution Data Collection Part 1: Vapor–Liquid Equilibrium, DECHEMA Chemistry Data Series, DECHEMA, Frankfurt, 1991. C. Wohlfarth, Vapor–Liquid Equilibrium Data of Binary Polymer Solutions, Elsevier, Amsterdam, 1994. R.B. Gupta, J.M. Prausnitz, J. Chem. Eng. Data 40 Ž1995. 784–791. C. Boukouvalas, N. Spiliotis, P. Coutsikos, N. Tzouvaras, D. Tassios, Fluid Phase Equilib. 92 Ž1994. 75–106. N.S. Kalospiros, N. Tzouvaras, Ph. Coutsikos, D.P. Tassios, AIChE J. 41 Ž1995. 928–937.
© Copyright 2026 Paperzz