2548
Ind. Eng. Chem. Res. 2004, 43, 2548-2555
SEPARATIONS Pervaporation of Alcohols and Methyl tert-Butyl Ether through a Dense Poly(2,6-dimethyl-1,4-phenylene oxide) Membrane J. P. G. Villaluenga,* P. Godino, M. Khayet, B. Seoane, and J. I. Mengual Department of Applied Physics I, Faculty of Physics, Complutense University of Madrid, 28040 Madrid, Spain
Pervaporation separation and swelling experiments of methanol/methyl tert-butyl ether mixtures were performed by using a flat-sheet, dense poly(2,6-dimethyl-1,4-phenylene oxide) membrane prepared from a chloroform casting solvent. The effect of the feed composition on the membrane performance was described by the permeability of the components, the separation factor, and the sorption selectivity. The effect of the feed temperature on the separation was also studied, and the Arrhenius activation energy for permeability was determined over the whole range of feed composition. Pervaporation operations offered a maximum methanol selectivity of 5.4 and an overall normalized flux of 110 kg‚µm/m2‚h. Pervaporation and sorption experiments were also conducted using ethanol, propanol, butanol, and octanol to study the effect of the size of the alcohol molecule on the transport across the membrane. 1. Introduction In pervaporation processes, a liquid mixture is in contact with one side of a dense membrane, and the permeate is removed as a vapor from the opposite side of the membrane. The mass transport through the membrane, which is generally described by the solution/ diffusion model,1 involves three steps: sorption, diffusion, and desorption.2 Additional resistances due to concentration polarization in the feed side and due to the mass transfer toward the vacuum trap in the permeate side should be considered. Thus, permselective properties of the membrane can be explained in terms of differences in the relative solubility and diffusivity in the membrane of the liquid mixture components. Applications of pervaporation can be included in three general categories:3 dehydration of organics,4-6 removal of organic pollutants from dilute aqueous solutions,7,8 and separation of organic/organic mixtures.9-12 Methanol/ methyl tert-butyl ether (MTBE) is one example of an organic/organic mixture that can be separated, and there are numerous papers dealing with this possibility13-38 (see Table 1). MTBE is an octane enhancer used as a replacement for toxic lead and lead compounds in car fuel. The production process of MTBE includes catalytic reaction of methanol and isobutene (iC4) at moderate temperature and pressure and extraction of MTBE from the effluent stream of the reactor. The latter stream contains unreacted methanol and C4’s, MTBE, and some byproducts. The presence of methanol and C4’s reduces the octane number expected from MTBE use. Therefore, the removal of these compounds is essential for MTBE. The existing technologies to separate and purify MTBE are cumbersome and costly, mainly because methanol forms azeotropic mixtures * To whom correspondence should be addressed. Tel.: (34) 91 394 4454. Fax: (34) 91 394 5191. E-mail: juanpgv@ fis.ucm.es.
Table 1. Results of Studies on the Separation of Methanol/MTBE Mixtures by Pervaporation
polymer CTA PPO-OH PAA/PVA silicalite PSS/Al2O3 Na Y zeolite PANHEMA PANMAC CS agarose SA/chitosan CA PVA/PAA PVA/SSA CTA CA HEC/agarose PVAc/alumina PVP/alumina SEC/pDMDAAC CS a
methanol normalized in the feed temp flux (wt %) (°C) (kg‚µm/m2‚h) selectivity ref 19 21 20 50a 14.3 10 15 15 20 10 25 20 20 20 20 20 ∼30 ∼30 ∼30 12.3 20
40 22 35 30 25 40 30 30 25 30 40 50 50 30 30 22 30 20 20 50 25
∼6.3 18.4 12.2 55.2 0.945 4.2 0.115 0.09 1.5 0.01 ∼0.15 ∼7.5 ∼0.2 ∼0.26 ∼33 1.9-3.4 0.47 ∼0.5b ∼3b 10.4 ∼8
∼90 7.7 45 ∼9.0 1200 6500 2561 773 50 ∼106 133-∞ ∼16 ∼4000 ∼2095 ∼80 246-46 2.3 × 105 ∼15 ∼10 349 ∼200
13 14 15 16 17 18 19 19 20 21 22 23 24 24 25 26 27 28 28 29 30
vol %. b kg/m2‚h.
with both MTBE and iC4. Pervaporation seems to offer a simple and economical alternative to the present separation process. The present study tries to investigate the performance of a poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) membrane in a pervaporation process for the separation of methanol/MTBE mixtures. PPO is a glassy polymer at room temperature, having high permeability to gases because of the absence of polar groups in the polymer backbone. The high free volume and the ease of the rotational motion of the phenyl rings contribute to the high gas diffusivity exhibited by dense PPO films. In addition, PPO possesses excellent mechanical and thermal properties and is resistant to several chemicals
10.1021/ie034299g CCC: $27.50 © 2004 American Chemical Society Published on Web 04/20/2004
Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004 2549 Table 2. Chemical Formula, Molecular Weight, Molar Volume at 20 °C, Diffusional Cross Section, Boiling Point, and Density for the Liquids Used in This Work CH3OH C2H5OH C3H7OH C4H9OH C8H17OH (CH3)3COCH3
M (g/mol)
V (cm3/mol)
σ (Å2)a
Tb (°C)
F (g/cm3)
32.04 46.07 60.10 74.12 130.23 88.15
40.49 58.37 74.81 91.53 157.47 119
17.6 21.8 25.2 28.2 39.1 40.0
64.6 78.2 82.3 117.7 180 55-56
0.7914 0.7893 0.7809 0.8098 0.8193 0.740
a Values of σ were calculated by using the following expression: i σi ) Vi/NAdi, where Vi is the molar volume, NA is Avogadro’s number, and di is the hard-sphere diameter.
including acids, bases, and alcohols. This latter property makes the PPO membrane a good candidate for the pervaporative methanol/MTBE separation because the chemical resistance and morphological deterioration is a major challenge. Moreover, to improve its permselectivity, PPO can be modified by various electrophilic substitutions including bromination, sulfonation, nitration, and carboxylation.39,40 In a previous work,41 PPO dense membranes were prepared using chloroform and 1,1,2-trichloroethylene as casting solvents. The membranes were characterized by X-ray diffraction, tapping-mode atomic force microscopy, and contact angle measurements. It was found that the membranes prepared from chloroform exhibited a better pervaporation performance than the PPO membranes prepared from 1,1,2-trichloroethylene when methanol/ethylene glycol mixtures were used. Moreover, the PPO membranes are better than those used for methanol/ethylene glycol separation such as cellophane and copolymerized acrylonitrile membranes, referring to both flux and methanol selectivity. In this paper, the PPO membrane prepared from chloroform was used for the separation of methanol/MTBE mixtures by pervaporation. The influence of operating conditions such as the feed mixture concentration and temperature on the separation performance of the membrane has been studied. In addition, the influence of permeate properties on the pervaporation and sorption properties of the membrane was investigated by using liquids such as methanol, ethanol, propanol, butanol, octanol, and MTBE. 2. Experimental Section 2.1. Materials. PPO powder of an intrinsic viscosity of 1.57 dL/g in chloroform at 25 °C and a density of 1.04 g/cm3 was supplied by General Electric. Analyticalgrade chloroform, purchased from Aldrich Chemicals, was used as the solvent in the preparation of the PPO membrane. Methanol, ethanol, 2-propanol, 1-butanol, and 1-octanol were supplied from Merck. MTBE was purchased from Fluka. Physicochemical properties of solvents are given in Table 2. All of the chemicals were of analytical purity grade with concentrations of 9799%. No further purification was carried out for the solvents used in the pervaporation and sorption experiments. 2.2. Membrane Preparation. In this study, the concentration of the PPO casting solution was 4 wt % in chloroform. A total of 4 mL of the solution of PPO was spread smoothly over a leveled glass plate inside an O ring of about 10 cm inner diameter made of stainless steel. The casting ring was then covered with a filter paper to keep out dust. After 24 h at a temperature of 25 °C, the membrane was removed very
cautiously from the glass plate by immersing the whole plate in a water bath. After drying at 25 °C for 24 h in a fume hood, the membrane was further dried for 72 h in a vacuum to remove the last traces of solvent. The average thickness of the obtained film was 37.7 µm. 2.3. Sorption Experiments. The swelling properties of the membrane in the liquids were measured gravimetrically. Preweighed dry PPO membranes were immersed for 48 h in a closed bottle containing either methanol, ethanol, 2-propanol, 1-butanol, 1-octanol, MTBE, or a mixture of methanol and MTBE. The bottle was placed in a thermostated bath at 30 °C. Then, the swollen membrane was taken out of the liquid, wiped out with tissue paper, and weighed again. The increment in weight is equal to the weight of the liquid sorbed by the membrane. Each swelling experiment was carried out at least two times. 2.4. Pervaporation Experiments. The experiments were carried out in the pervaporation apparatus described elsewhere.42 The effective membrane surface area was 28 cm2. The feed was circulated over the membrane sample through meander-type channels. The feed pressure was kept at 1.0 bar. The permeate stream was evacuated by means of a vacuum pump, and the permeate was collected alternatively in two traps cooled by liquid nitrogen. The downstream pressure was monitored by a pressure transducer and was maintained below 0.1 Torr. The fluxes were determined by weighing the sample collected within a predetermined time interval. In the case of methanol/MTBE mixtures, the feed and permeate compositions were determined by measuring their refractive index with an Abbey-type refractometer model 60/ED. The pervaporation selectivity of the membrane was studied in terms of the separation factor, R, which is defined as
R)
(w1/w2)permeate (w1/w2)feed
(1)
where w are the weight fractions of the relevant component in the feed and permeate, respectively. Indexes 1 and 2 refer to methanol and MTBE, respectively. Sample history and relaxation phenomena are very important in membrane processes where glassy polymers, such as PPO, are used.43,44 In the present work, thermal and chemical preconditioning of the PPO membrane prior to the pervaporation experiments were conducted by methanol permeation experiments at 50 °C. In each operation run, several samples were collected to determine the total flux as well as the composition of the permeate. The alternative use of two cold traps allowed the permeate to be continuously sampled without interrupting the experiment. It was observed that the membrane exhibited initially high fluxes, which then gradually moved to a steady-state value. The data given in this paper correspond to the final steady-state values. 3. Results and Discussion 3.1. Pure Compound Measurements. Table 3 shows the permeation rate, expressed as the total normalized flux, and the solubility, expressed in terms of the weight gain, of methanol, ethanol, propanol, butanol, octanol, and MTBE in the PPO membrane. The analysis of the influence of the compound size expressed
2550 Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004 Table 3. Normalized Pervaporation Flux, Solubility, and Concentration-Averaged Diffusion Coefficient of Pure Compounds
CH3OH C2H5OH C3H7OH C4H9OH C8H17OH CH3OC(CH3)2CH3
J (kg‚µm/m2‚h)
S (g/g)
Di (10-11 m2/s)
35.3 31.2 8.28 3.39 1.34 2.04
0.173 0.161 0.122 0.101 0.071 0.057
5.4 4.8 1.7 0.9 0.5 0.64
Di )
as the liquid molar volume shows that, in the case of alcohols, their solubility in the PPO membrane decreases with increasing size of the molecule. In fact, a regression analysis indicates that the mean correlation coefficient between the solubility and molar volume is 0.93. This behavior suggests that the sorption of alcohol in the membrane depends not only on the affinity between the penetrate and polymer but also on the molecular volume of the penetrate. Burshe et al.45 observed a similar behavior when testing poly(vinyl acetate) (PVA) membranes using water/alcohol mixtures. They found that, in the case of alcohols, the polarity decreases with an increase in the carbon number and, therefore, the extent of sorption also decreases. From Table 3, it is seen that the solubility of MTBE is smaller than those of alcohols. In particular, if the solubility of MTBE is compared with those of butanol and octanol, because the MTBE molar volume is between the values of butanol and octanol, its solubility is lower than those of both because of lower polarity. Table 3 also shows that the permeation rate across the membrane with alcohol is molecular size. It must be mentioned that methanol and ethanol fluxes are significantly higher than the fluxes of the other alcohols. On the contrary, MTBE flux is smaller than those of the alcohols, with the exception of octanol. The examination of pervaporation and sorption data suggests that the solubility of the compounds in the PPO membrane is not the only characteristic responsible for the observed different permeation rates. An estimation of the diffusivity of the permeants in the membrane was made by assuming the validity of Fick’s law in our system. The classic Fick’s law can be expressed as follows:
Ji ) -Di
dCi dx
(2)
where Di is the Fickian diffusion coefficient of component i and Ci is the penetrate concentration in the membrane. The diffusion coefficient depends usually on the penetrant concentration in the membrane, and it also changes across the membrane as a function of the distance.46 Because the latter information could not be obtained with accuracy in our system, we expressed Fick’s law as follows:
Ji ) Di
Ci,f - Ci,p l
Although the film thickness changes in the pervaporation experiments, it was assumed that the thickness of the PPO membrane is the value of the dry sample. Consequently, an apparent diffusion coefficient in PPO was calculated by using the following equation:
(3)
where Di is a mean diffusion coefficient of component i, l is the membrane thickness, and Ci,f and Ci,p are the concentrations of component i at the membrane feed and permeate sides, respectively. In the present study, the permeate side was kept at sufficiently reduced pressure (less than 0.1 Torr) and Ci,p was assumed to be zero.
J il Ci,f
(4)
The obtained apparent diffusion coefficients are given in Table 3. It should be pointed out that, although the apparent diffusion coefficient is not the true diffusion coefficient, it is presented to compare the diffusivities of the different compounds. As far as the alcohols are concerned, there is a noticeable difference in their diffusivities in the membrane; i.e., the diffusion coefficient of methanol is 1 order of magnitude higher than that of octanol. In addition, the values of Di of methanol and ethanol are significantly higher than those of the other alcohols. This fact suggests that the diffusivity is important for permeation rates. On the other hand, it is known that the alcohol molecules are polar, with the ability to form hydrogen bonds through their hydroxyl groups. This implies the capability of the formation of alcohol clusters in the PPO membrane. For example, Nguyen et al.47 observed the formation of alcohol clusters in a poly(dimethylsiloxane) membrane, and they concluded that clustering increased the solubility in the membrane, although the diffusivity was decreased. The clustering probably occurs more with the smaller alcohols, so if alcohol clustering occurs in the PPO membrane, a lower diffusivity for alcohols with lower molecular size would be expected. However, diffusivities of methanol and ethanol in the PPO membrane are larger than those of higher alcohols. On the other hand, an analysis of the effect of shape and size of the penetrate molecule on the diffusivity can also be made in terms of a diffusional cross section of the penetrants, σi, given in Table 2. A regression analysis shows that there was not a linear correlation between Di and σi, but there is a noticeable reduction in the diffusivity in the PPO membrane for molecules with a diffusional cross section larger than approximately 2225 Å2. In fact, diffusivity in the membrane is dependent on both the size of the penetrant and the polymer structure. PPO is a semicrystalline polymer. The presence of crystalline entities in the PPO membrane, which can be regarded as impermeable to the solvent transport, reduces the configurational mobility of the nearby chains and increases the path length of diffusion via tortuosity. The results obtained suggest that the restriction to the diffusion produced by the microstructure is more important for alcohols with a molecular size larger than some critical value. 3.2. Influence of the Feed Composition on the Membrane Performance for Methanol/MTBE Mixtures. The effect of the feed composition on the flux and selectivity was investigated over the whole mixture concentration range for methanol/MTBE mixtures at 30 °C. Figure 1 shows the total flux and methanol selectivity as a function of the feed mixture composition. It can be seen that the flux increases with the methanol content in the feed. For example, the flux goes from 7.3 to 18.4 kg‚µm/m2‚h when the methanol weight fraction is increased from 0.14 to 0.69. The PPO membrane is selective toward methanol, with a selectivity factor ranging between 2.6 and 5.4. The variation of the
Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004 2551
Figure 1. Normal flux and methanol selectivity for methanol/ MTBE mixtures vs methanol content of the feed. The lines are only a visual guide.
Figure 3. Apparent permeability coefficient of methanol (b) and MTBE (2) vs methanol content of the feed. The lines are only a visual guide.
better understanding of the process is achieved if the permeabilities of the permeants are considered instead of their permeation rates. The permeability of the membrane toward the components, which is an intrinsic membrane transport property, is characterized by the permeability coefficient. Such a coefficient can be determined if the flux of component i through the membrane is expressed as follows:48
Ji ) Pi
fi,f - fi,p l
(5)
where Pi is the apparent permeability coefficient and fi,f and fi,p are the fugacities of component i in the feed and permeate sides, respectively. The fugacities can be expressed as follows: Figure 2. Partial fluxes of methanol (b) and MTBE (2) vs methanol content of the feed. The lines are only a visual guide.
selectivity upon feed composition follows a complicated pattern, and the typical “tradeoff” between flux and selectivity is not observed in this case, because for methanol-rich mixtures, both flux and selectivity increase with the methanol content. It should be pointed out that the minimum in selectivity is due to a maximum in the MTBE flux, as will be discussed later. On the other hand, a comparison between the pervaporation performance of PPO and the values reported for other membrane materials indicates that the fluxes can be considered satisfactory, but the selectivity is rather low. The partial fluxes of methanol and MTBE, which were calculated by using both the total flux and permeate composition data, are plotted in Figure 2 as a function of the methanol content in the feed. The methanol flux increases with its content in the feed, whereas the MTBE flux initially keeps nearly constant and finally decreases. This behavior can be explained as follows: methanol molecules occupy most of the dissolution sites available in the polymer matrix, so the MTBE solution is severely restricted, and thus MTBE permeates slower, which results in an increase of the methanol selectivity of the PPO membrane, as observed in Figure 1. However, Wijmans48 has shown that the analysis of the membrane performance in terms of its flux is not recommended because this quantity includes not only the membrane properties but also the effect of the operating conditions in the pervaporation process. A
fi,f ) γixipi°
(6)
fi,p ) yipp
(7)
where γi is the activity coefficient, pi° is the saturation vapor pressure of component i at temperature T of the feed, and pp is the total pressure in the permeate. Values of Pi for each component were calculated using the following expression:
Pi )
Jil Jil ≈ γixipi° - yipp γixipi°
(8)
An important parameter needed to calculate the permeability coefficient is the activity coefficient. In the present work, values of γi for methanol and MTBE at 30 °C were taken from Coto et al.49 as a function of the methanol mole fraction in their mixture, and pi° were calculated with the extended Antoine equation. The apparent permeability coefficients of methanol and MTBE, which are shown in Figure 3 as a function of the variation of the methanol content in the feed, exhibit a very nonideal behavior. As can be seen, the permeability of methanol increases steadily as the methanol content in the feed increases. In contrast, the permeability of MTBE shows a nonmonotonic behavior, with a maximum at about 50/50 wt % feed composition. According to the solution/diffusion model, the permeability of a dense PPO membrane is determined by the solubility and diffusivity of the permeants. Because both
2552 Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004
Figure 4. Solubility for methanol/MTBE mixtures (9), methanol (b), and MTBE (2) vs methanol content of the feed. The lines are only a visual guide.
sorption and diffusion are dependent on the liquid composition in the membrane, also the permeability characteristics of the membrane are influenced by the composition. The analysis of the sorption properties of the membrane toward methanol/MTBE mixtures can shed light on the behavior observed for the permeability. The sorption of a binary liquid mixture in a polymer can be characterized by the overall sorption, which represents the total amount of liquid inside the membrane. The results of the swelling measurements with mixtures of methanol and MTBE in PPO are presented in Figure 4 as a function of the methanol content in the mixtures. It can be observed that the higher the methanol content in the feed, the more the membrane is swollen. As an example, the degree of swelling goes from 0.06 for pure MTBE to 0.207 for pure methanol. The composition of the liquid sorbed in the PPO membrane was estimated by a model derived from the Flory-Huggins theory,50,51 which makes a statistical calculation of the different configurations of a polymer/solvent system upon some assumptions on molecular shapes and interactions. The theory enables sorption modeling using a single parameter, which is a reflection of the polymer/solvent affinity. The lower the value of this parameter, the higher the sorption and thus the higher the affinity. Furthermore, the model was extended by Mulder52 to describe the sorption of binary liquid mixtures. According to this model, at a given volume fraction of liquid i in the feed, vi, the volume fraction in the membrane, φi, can be calculated by the following equations:
ln
() ()
()
φ1 v1 φ2 - ln ) (r - 1) ln - g12(v1 - v2 + φ2 φ2 v2 v2 ∂g12 ∂g12 - v1v2 (9) φ1) - φ3(χ13 - rχ23) + u1φ2 ∂u1 ∂v2
Figure 5. Volume fraction of methanol (b) and MTBE (2) in the liquid sorbed in the membrane vs methanol content of the feed. The lines are only a visual guide.
fraction of component i in the sorbed liquid:
ui )
φi 1 - φ3
i ) 1, 2
(11)
Values of the weight fraction of polymer were obtained from the swelling experiments by using mixtures of methanol and MTBE at different concentrations. The volume fraction of polymer was easily calculated from its weight fraction data by assuming that there is no change of volume upon mixing and the penetrant in the ternary mixture has the same density as that in the liquid binary mixture. The interaction parameter between the liquids and polymer, which were assumed to be concentration-independent, were calculated from swelling experiments by using either pure methanol or pure MTBE. On the other hand, the liquid/polymer parameter, χi3, can be expressed as follows:52
χi3 ) -
[ln(1 - φ3) + φ3]
(12)
φ32
The value of χi3 for methanol (1.6) is smaller than that of MTBE (2.5), indicating that the methanol is more soluble than MTBE in the PPO membrane, as was mentioned in section 2.1. The binary liquid interaction parameter was calculated using the following expression:
g12 )
(
)
x1 x2 ∆GE 1 x1 ln + x2 ln + x1v2 v1 v2 RT
(13)
where xi is the mole fraction of component i, ∆GE is the excess free energy of mixing, and R is the gas constant. For methanol/MTBE mixtures, values of ∆GE as a function of the mixture composition at 30 °C were taken from Coto et al.49 Thus, the dependence of g12 on v1 was expressed by a fourth-order polynomial relationship
(10)
g12 ) 2.532v14 - 6.356v13 + 6.222v12 - 3.406v1 + 1.715 (14)
where g12 is the liquid/liquid interaction parameter, χi3 is the polymer/liquid parameter, indexes 1 and 2 refer to the methanol and MTBE, respectively, index 3 refers to the PPO polymer, and ui is the relative volume
Values of absolute, φi, and relative, ui, volume fractions of methanol and MTBE in the PPO membrane were calculated from eqs 9-11 and are given in Figures 4 and 5. It is clearly shown that methanol is preferen-
φ1 + φ2 + φ3 ) 1
Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004 2553
Figure 6. Normalized flux for methanol/MTBE mixtures vs feed temperature with different methanol volume fractions: 1 ((), 0.9 (9), 0.7 (2), 0.5 (×), 0.3 (*), 0.15 (b), 0 (+).
tially sorbed from its mixtures with MTBE by the membrane, even for methanol-poor mixtures. Preferential sorption of methanol is probably due to the fact that its polarity is close to that of the membrane polymer and to its lower molar volume. As an example, for a liquid feed mixture with a methanol volume fraction of 0.15, the model predicts that the methanol volume fraction in the sorbed liquid is 0.86. It is also observed that the amount of methanol in the membrane slowly increases with its content in the feed mixture after a volume fraction of 0.15. On the other hand, Figure 3 shows that methanol permeability follows the same trend as the amount of methanol sorbed by the membrane (Figures 4 and 5): it increases with the methanol content in the feed. This fact suggests that the permeability variation may be associated with the solubility of the permeants in the membrane. In addition, it may be stated that a swelling effect or plasticization should be taking place in the membrane, which would enhance the methanol diffusivity, because the permeability increases about 7 times over the concentration range studied. At low methanol content, because of the low affinity between PPO and MTBE, the presence of methanol reduces the amount of MTBE sorbed by the membrane compared with the case where MTBE is pure, as can be seen in Figure 4. When the methanol content in the feed increases, the decrease of the amount of MTBE sorbed in the membrane is less pronounced. Because the sorption behavior of MTBE does not account alone for the variation of its permeability, as shown in Figure 3, it may be stated that the kinetic coupling effect53,54 caused by the presence of methanol may occur. In fact, the increased swelling of the membrane by the sorbed methanol molecules results in increased polymer segmental motions. This facilitates the diffusion of the larger MTBE molecules and therefore increases the permeability of MTBE. At higher 50 wt % methanol in the feed, the decrease of the solubility of MTBE in the PPO membrane seems to prevail over the effect of the enhanced diffusivity, which results in a decrease of the permeability. 3.3. Influence of the Feed Temperature on the Membrane Performance for Methanol/MTBE Mixtures. The effect of temperature on the membrane performance was investigated over a temperature range of 30-50 °C. Figures 6 and 7 show plots of the total
Figure 7. Methanol selectivity for methanol/MTBE mixtures vs methanol content of the feed at 30 °C (9), 40 °C (b), and 50 °C (2). The lines are only a visual guide. Table 4. Activation Energy (kJ/mol) for the Permeability Coefficient of Methanol and MTBE in Mixtures with Different Methanol Volume Fractions v1 v1 EP(methanol) EP(MTBE)
1.0
0.9
0.7
0.5
0.3
0.15
0
32.6
20.9 86.2
15.8 55.2
6.9 38.8
31.7 61.1
21.5 64.5
71.3
normalized flux and methanol selectivity, respectively, as a function of the feed temperature for pure methanol and MTBE and their mixtures. As can be seen, the flux increases with the feed temperature. This may be attributed to the fact that, although the amount of solvent sorbed by the membrane decreases with temperature, this is more than offset by the increase of the penetrant diffusivity with increasing temperature. The diffusion process is facilitated at higher temperatures because of the greater kinetic energy of the penetrant and the larger available free volume in the polymer matrix.15,55 On the assumption that Arrhenius’ law is valid, the activation energy of the pervaporation process can be calculated from Arrhenius-type plots of the flux. However, the energy values thus obtained include the effect of temperature not only in the permeability coefficient but also in the driving force.56 In the present study, activation energy values were obtained from Arrhenius plots of the permeability coefficient of each component. Thus, the activation energy for permeation is regarded as a combination of the activation energy of diffusion and the enthalpy of sorption of the permeants in the membrane. The data presented in Figure 6 were used to determine the apparent permeability coefficient by using eq 8. The saturation vapor pressures at different temperatures were calculated by means of the extended Antoine equation. Values for the activity coefficient of methanol/MTBE mixtures at different temperatures were found in the literature or estimated following Wilson and NRTL methods.49,57-60 The activation energy values obtained for the permeability are presented in Table 4. It should be noted that the fitting of the permeability coefficient data to straight lines on the Arrhenius-type plots is rather poor, as reflected by the correlation coefficients obtained, which lie in the range of 0.90-0.99. As can be seen, the activation energy of MTBE is higher than that of methanol. This explains the fact that the methanol selectivity is reduced by
2554 Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004
increasing the feed temperature, as can be observed in Figure 7; i.e., the partial fluxes of MTBE were more affected by the temperature than those of methanol. With regard to the effect of the feed composition on the activation energy, it is observed that, for MTBE-rich mixtures, the activation energy of methanol and MTBE decreases with the methanol content in the feed mixture, whereas, for methanol-rich mixtures, the increase in the methanol concentration gives rise to an increase in the activation energies of both components. Temperature variation may affect the polymer relaxation associated with the membrane swelling/plasticization and the interaction between methanol, MTBE, and the PPO polymer, and it may also cause the fusion of the less perfect crystallites in the polymer matrix. Because of the fact that the effect of these phenomena on the solution and diffusion of methanol/MTBE mixtures may be very complex, at the moment we have no plausible explanation for the variation of the activation energy with composition. According to some authors,61-63 it can be speculated that the increase in swelling of the PPO membrane at increasing methanol content results in a lowering of the energetic barrier for diffusion owing to the possible plasticization by the methanol. 4. Conclusions The pervaporation separation for the mixture of methanol and MTBE by using a PPO membrane was investigated. The membrane preferentially permeated methanol over the whole range of composition. The permeability coefficient of methanol increases with the methanol content in the feed. This may be attributed to the membrane swelling, if not plasticization, caused by methanol. The permeability of MTBE showed a nonmonotonic behavior with the feed composition. The total amount of liquid sorbed in the PPO membrane increases with the concentration of methanol. This fact indicates that methanol has a higher solubility in PPO than MTBE. The Flory-Huggins model predicts that the membrane has preferential sorption toward methanol over the whole range of feed composition. Both sorption and diffusion processes account for the permselective properties of the membrane. The temperature dependence of the overall flux followed an Arrhenius relationship over the concentration range of 0-100% methanol. The activation energies for the permeation of methanol and MTBE were 32.6 and 71.3 kJ/mol, respectively. Pervaporation and sorption measurements by using five different alcohols showed that the permeation rate, solubility, and diffusivity increased with the compound molecular size. It can be inferred that pervaporation of pure compounds is governed by both sorption and diffusion processes in the PPO membrane. Acknowledgment Financial support from the University Complutense of Madrid under Project PR78/02-10993 is gratefully acknowledged. Literature Cited (1) Wijmans, J. G.; Baker, R. W. The solution-diffusion model: a review. J. Membr. Sci. 1995, 107, 1. (2) Feng, X.; Huang, R. Y. M. Liquid separation by membrane pervaporation: a review. Ind. Eng. Chem. Res. 1997, 36, 1048.
(3) Fleming, H. L.; Slater, C. S. Applications and Economics. In Membrane Handbook; Ho, W. S. W., Winston, W. S., Eds.; Chapman and Hall: New York, 1992; Chapter 10. (4) Wesslein, M.; Heintz, A.; Lichtenthaler, R. N. Pervaporation of liquid mixtures through poly(vinyl alcohol) (PVA) membranes. I. Study of water containing binary system with complete and partial miscibility. J. Membr. Sci. 1990, 51, 169. (5) Jiraratanon, R.; Chanachai, A.; Huang, R. Y. M.; Uttapap, D. Pervaporation dehydration of ethanol-water mixtures with chitosan/hydroxyethylcellulose (CS/HEC) composite membranes: I. Effect of operating conditions. J. Membr. Sci. 2002, 195, 143. (6) Maeda, Y.; Kai, M. Recent progress in pervaporation membranes for water/ethanol separation. In Pervaporation Membrane Separation Processes; Huang, R. Y. M., Ed.; Elsevier: Amsterdam, The Netherlands, 1991; Chapter 9. (7) Karlsson, H. O. E.; Tra¨gårdh, G. Pervaporation of dilute organic-water mixtures. A literature review on modelling studies and applications to aroma compound recovery. J. Membr. Sci. 1993, 76, 121. (8) Bo¨ddeker, K. W.; Bengtson, G. Selective pervaporation of organics. In Pervaporation Membrane Separation Processes; Huang, R. Y. M., Ed.; Elsevier: Amsterdam, The Netherlands, 1991; Chapter 10. (9) Garcı´a-Villaluenga, J. P.; Tabe-Mohammadi, A. A review on the separation of benzene/cyclohexane by pervaporation processes. J. Membr. Sci. 2000, 169, 159. (10) Moon, G. Y.; Pal, R.; Huang, R. Y. M. N-Acetylated chitosan membranes for the pervaporation separation of alcohol/toluene mixtures. J. Membr. Sci. 2000, 176, 223. (11) Lipnizki, F.; Hausmanns, S.; Ten, P. K.; Field, R. W.; Laufenberg, G. Organophilic pervaporation: prospects and performance. Chem. Eng. J. 1999, 73, 113. (12) Marx, S.; Gryp, P.; Neomagus, H.; Everson, R.; Kleizer, K. Pervaporation separatiin of methanol/tert-amyl methyl ether mixtures with a commercial membrane. J. Membr. Sci. 2002, 209, 353. (13) Yang, J. S.; Kim, H. J.; Jo, W. H.; Kang, Y. S. Analysis of pervaporation of methanol-MTBE mixtures through cellulose acetate and cellulose triacetate membranes. Polymer 1998, 39, 1381. (14) Doghieri, F.; Nardella, A.; Sarti, G. C.; Valentini, C. Pervaporation of methanol-MTBE mixtures through modified poly(phenylene oxide) membranes. J. Membr. Sci. 1994, 91, 283. (15) Park, H. C.; Ramaker, N. E.; Mulder, M. H. V.; Smolders, C. A. Separation of MTBE-methanol mixtures by pervaporation. Sep. Sci. Technol. 1995, 30, 419. (16) Sano, T.; Hasegawa, M.; Kawakami, Y.; Yanagishita, H. Separation of methanol/methyl-tert-butyl ether mixture by pervaporation using silicalite membrane. J. Membr. Sci. 1995, 107, 193. (17) Chen, W.; Martin, C. R. Highly methanol-selective membranes for the pervaporation separation of methyl tert-butyl ether/ methanol mixtures. J. Membr. Sci. 1995, 104, 101. (18) Kita, H.; Inoue, T.; Asamura, H.; Tanaka, K.; Okamoto, K. Chem. Commun. 1997, 45. (19) Ray, S. K.; Sawant, S. B.; Pangarkar, V. G. Development of methanol selective membranes for separation of methanolmethyl tertiary butyl ether mixture by pervaporation. J. Appl. Polym. Sci. 1999, 74, 2645. (20) Nam, S. Y.; Lee, Y. M. Pervaporation separation of methanol/methyl tert-butyl ether through chitosan composite membrane modified with surfactants. J. Membr. Sci. 1999, 157, 63. (21) Yoshikawa, M.; Yoshioka, T.; Fujime, J.; Murakami, A. Pervaporation of MeOH/MTBE through agarose membranes. J. Membr. Sci. 2000, 178, 75. (22) Kim, S.; Lim, G.; Jegal, J.; Lee, K. Pervaporation separation of MTBE (methyl tert-butyl ether) and methanol mixtures through polyion complex composite membranes consisting of sodium alginate/chitosan. J. Membr. Sci. 2000, 174, 1. (23) Cao, S.; Shi, Y.; Chen, G. Influence of acetylation degree of cellulose acetate on pervaporation properties for MeOH/MTBE mixture. J. Membr. Sci. 2000, 165, 89. (24) Rhim, J.; Kim, Y. Pervaporation separation of MTBEmethanol mixtures using cross-linked PVA membranes. J. Appl. Polym. Sci. 2000, 75, 1699. (25) Niang, M.; Luo, G. A triacetate cellulose membrane for the separation of methyl tert-butyl ether/methanol mixtures by pervaporation. Sep. Purif. Technol. 2001, 24, 427.
Ind. Eng. Chem. Res., Vol. 43, No. 10, 2004 2555 (26) Tabe-Mohammadi, A.; Villaluenga, J. P. G.; Kim, H. J.; Chan, T.; Rauw, V. Effects of polymer solvents on the performance of cellulose acetate membranes in methanol/methyl tertiary butyl ether separation. J. Appl. Polym. Sci. 2001, 82, 2882. (27) Yoshikawa, M.; Yoshioka, T.; Fujime, J.; Murakami, A. Pervaporation of methanol/methyl tert-butyl ether mixtures through agarose/hydroxyethylcellulose blended membranes. J. Appl. Polym. Sci. 2002, 86, 3408. (28) Yoshida, W.; Cohen, Y. Ceramic-supported polymer membranes for pervaporation of binary organic/organic mixtures. J. Membr. Sci. 2003, 213, 145. (29) Schwarz, H.; Apostel, R.; Paul, D. Membranes based on polyelectrolyte-surfactant complexes for methanol separation. J. Membr. Sci. 2001, 194, 91. (30) Huang, R. Y. M.; Moon, G. Y.; Pal, R. Chitosan/anionic surfactant complex membranes for the pervaporation separation of methanol/MTBE and characterization of the polymer/surfactant system. J. Membr. Sci. 2001, 184, 1. (31) Farnand, B. A.; Noh, S. H. Pervaporation as an alternative process for the separation of methanol from C4 hydrocarbons in the production of MTBE and TAME. AIChE Symp. Ser. 1989, 85, 89. (32) Chen, M. S. K.; Markiewicz, G. S.; Venogupal, K. G. Development of membrane pervaporation TRIM process for methanol recovery from CH3OH/MTBE/C4 mixtures. AIChE Symp. Ser. 1989, 85, 82. (33) Cao, S.; Shi, Y.; Chen, G. Pervaporation separation of MeOH/MTBE through CTA membrane. J. Appl. Polym. Sci. 1999, 71, 377. (34) Bangxiao, C.; Li, Y.; Hailin, Y.; Congjie, G. Effect of separating layer in pervaporation composite membrane for MTBE/ MeOH separation. J. Membr. Sci. 2001, 194, 151. (35) Eliceche, A. M.; Daviou, M. C.; Hoch, P. M.; Uribe, I. O. Optimisation of azeotropic distillation columns combined with pervaporation membranes. Comput. Chem. Eng. 2002, 26, 563. (36) Ho¨mmerich, U.; Rautenbach, R. Design and optimization of combined pervaporation/distillation processes for the production of MTBE. J. Membr. Sci. 1998, 146, 53. (37) Gonza´lez-Gonza´lez, B.; Ortiz-Uribe, I. Mathematical modeling of the pervaporative separation of methanol-methyl tertbutyl ether mixtures. Ind. Eng. Chem. Res. 2001, 40, 1720. (38) Zhou, M.; Persin, M.; Sarrazin, J. Methanol removal from organic mixtures by pervaporation using polypyrrole membranes. J. Membr. Sci. 1996, 117, 303. (39) Chowdhury, G.; Kruczek, B.; Matsuura, T. Polyphenylene Oxide and Modified Polyphenylene Oxide Membranes: Gas, Vapor and Liquid Separation; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2001. (40) Khulbe, K. C.; Matsuura, T.; Lamarche, G.; Kim, H. J. The morphology characterization and performance of dense PPO membranes for gas separation. J. Membr. Sci. 1997, 135, 211. (41) Khayet, M.; Villaluenga, J. P. G.; Godino, M. P.; Mengual, J. I.; Seoane, B.; Khulbe, K. C.; Matsuura, T. Development and application of dense poly(phenylene oxide) membranes for pervaporation of methanol-ethylene glycol mixtures. J. Membr. Sci. 2004, submitted for publication. (42) Villaluenga, J. P. G.; Khayet, M.; Godino, P.; Seoane, B.; Mengual, J. I. Pervaporation of toluene/alcohol mixtures through a coextruded linear low-density polyethylene membrane. Ind. Eng. Chem. Res. 2003, 42, 386. (43) Marin, M.; Kalantzi, K.; Gilbert, H. Pervaporation process: membrane conditioning and experimental mass transfer analysis. J. Membr. Sci. 1992, 74, 105. (44) Rautenbach, R.; Ho¨mmerich, U. Experimental study of dynamic mass-transfer effects in pervaporation. AIChE J. 1998, 44, 1210.
(45) Burshe, M. C.; Sawant, S. B.; Joshi, J. B.; Pangarkar, V. G. Sorption and permeation of binary water-alcohol systems through PVA membranes crosslinked with multifunctional crosslinking agents. Sep. Sci. Technol. 1997, 12, 145. (46) Sun, Y.; Chen, Y.; Wu, C.; Lin, A. Pervaporation for the mixture of benzene and cyclohexane through PPOP membranes. AIChE J. 1999, 45, 523. (47) Nguyen, Q. T.; Favre, E.; Ping, Z. H.; Ne´el, J. Clustering of solvents in membranes and its influence on membrane transport properties. J. Membr. Sci. 1996, 113, 137. (48) Wijmans, J. G. A simple predictive treatment of the permeation process in pervaporation. J. Membr. Sci. 1993, 79, 101. (49) Coto, B.; Wiesenberg, R.; Pando, C.; Rubio, R. G.; Renuncio, J. A. R. Vapor-liquid equilibrium of the methanol-tert-butyl methyl ether (MTBE) system. Ver. Bunsen-Ges. Phys. Chem. 1996, 100, 482. (50) Flory, P. Thermodynamics of high-polymer solutions. J. Chem. Phys. 1942, 10, 51. (51) Huggins, M. Thermodynamic properties of solutions of long-chain compounds. Ann. N.Y. Acad. Sci. 1942, 43, 1. (52) Mulder, M. H. V. Thermodynamic principles of pervaporation. In Pervaporation Membrane Separation Processes; Huang, R. Y. M., Ed.; Elsevier: Amsterdam, The Netherlands, 1991; Chapter 4. (53) Drioli, E.; Zhang, S.; Basile, A. On the coupling effect in pervaporation. J. Membr. Sci. 1993, 81, 43. (54) Fleming, H. L.; Slater, C. S. Pervaporation: theory. In Membrane Handbook; Ho, W. S. W., Winston, W. S., Eds.; Chapman and Hall: New York, 1992; Chapter 8. (55) Ray, S. K.; Sawant, S. B.; Joshi, J. B.; Pangarkar, V. G. Methanol selective membranes for separation of methanolethylene glycol mixtures by pervaporation. J. Membr. Sci. 1999, 154, 1. (56) Feng, X.; Huang, R. Y. M. Estimation of activation energy for permeation in pervaporation process. J. Membr. Sci. 1996, 118, 127. (57) Toghiani, R. K.; Toghiani, H.; Venkateswarlu, G. Vaporliquid equilibria for methyl tert-butyl ether + methanol and tertamyl methyl ether + methanol. Fluid Phase Equilib. 1996, 122, 157. (58) Vetere, A.; Miracca, I.; Cianci, F. Correlation and prediction of the vapor-liquid equilibria of the binary and ternary systems involved in MTBE synthesis. Fluid Phase Equilib. 1993, 90, 189. (59) Segovia, J. J.; Martı´n, M. C.; Chamorro, C. R.; Villaman˜a´n, M. A. Excess thermodynamic properties of binary and ternary mixtures containing methyl 1,1-dimethylethyl ether (MTBE), n-heptane, and methanol at T ) 313.15 K. J. Chem. Thermodyn. 1999, 31, 1231. (60) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. Properties of Gases and Liquids, 4th ed.; McGraw-Hill: New York, 1988. (61) Ghosh, I.; Sanyal, S. K.; Mukherjea, R. N. Pervaporation of methanol-ethylene glycol with cellophane membrane: some mechanistic aspects. Ind. Eng. Chem. Res. 1988, 27, 1895. (62) Jiang, J. S.; Greenberg, D. B.; Fried, J. R. Pervaporation of methanol from a triglyme solution using a nafion membrane: 1. Transport studies. J. Membr. Sci. 1997, 132, 255. (63) Nam, S. Y.; Chun, H. J.; Lee, Y. M. Pervaporation separation of water-2-propanol mixture using carboxymethylated poly(vinyl alcohol) composite membranes. J. Appl. Polym. Sci. 1999, 72, 241.
Received for review December 9, 2003 Revised manuscript received March 10, 2004 Accepted March 17, 2004 IE034299G
