Separation of toluene–methanol mixtures by pervaporation using filled elastomeric membranes

Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

Contents lists available at ScienceDirect

Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice

Separation of toluene–methanol mixtures by pervaporation using filled elastomeric membranes Paramita Das, Samit Kumar Ray∗ Department of Polymer Science and Technology, University of Calcutta, 92, A.P.C Road, Kolkata 700009, India

a r t i c l e

i n f o

Article history: Received 14 September 2015 Revised 17 February 2016 Accepted 23 March 2016 Available online 24 April 2016 Keywords: Elastomeric membrane Pervaporation Filler Vulcanization Permeability Diffusion coefficient

a b s t r a c t Hybrid type elastomeric membranes were prepared by mixing natural rubber (NR) with 10 wt% of (of rubber) three different types of organophilic fillers, i.e., carbon black, organophilic clay and organophilic zeolite. These elastomeric membranes were cured (crosslinked) by semi efficient vulcanization technique using sulfur (S) and zinc diethyl dithio carbamate (ZDC), an ultra fast room temperature curing accelerator to obtain three membranes designated as NRCB, NRC and NRZ, respectively. The membranes were characterized by crosslink density, FT-IR, SEM analysis, and mechanical properties and used for pervaporative removal of low concentration of toluene from its mixtures with methanol at different feed toluene concentrations and temperatures. The NRCB membrane was found to show the best result with respect to toluene flux and selectivity over the entire feed concentration of 3–40 wt% of toluene in methanol. © 2016 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction For pervaporative dehydration hydrophilic membranes are used while for pervaporative recovery/separation of a low concentration of an organic from water, an organophilic membrane may be used. However, for organic–organic separation the organophilic membrane should selectively permeate one organic while the other organic should be less permeable. In pervaporation separation takes place by a preferential sorption and diffusion mechanism. The preferential sorption depends on the relative solvent–membrane interaction which may be evaluated in terms of relative solubility parameter values of the solvents with respect to the membrane polymer. Thus, for organic–organic separation, the membrane material should be chosen in such a way that the solubility parameter of the organic to be preferentially permeated is very close to the solubility parameter of the membrane material [1]. Further, the organic to be preferentially permeated should have also smaller kinetic diameter than the other organic for preferential diffusion through the membrane. In the last ten years several polymer membranes were reported [2–17] for pervaporative separation of alkane–aromatic mixtures like separating thiophene from octane or heptane simulating model gasoline [2,3,4], toluene from n-heptane [5–8] and also for separation of aromatic-alcohol mixtures like benzene–methanol [9] and toluene–methanol [10–19].

∗

Corresponding author. Tel.: +91 3323508386; fax: +91 33 351 9755. E-mail address: [email protected] (S.K. Ray).

In most of the above cases, clay filled hybrid type glassy [5–7,9,10] or elastomeric membranes have been tried [1,16,17] because of its integrity and high selectivity. However, the hybrid membranes prepared by incorporating clay in to a glassy thermoplastics show poor flux. On the other hand the membranes made from elastomers yield high flux but poor selectivity. In fact, the elastomeric membranes show preferential organic sorption but its high free volume and the flexibility of its long polymer chains results in high flux but poor selectivity. For improving selectivity, elastomeric membranes were suitably crosslinked [15], filled with adsorptive clay [16,17] and also blended with another glassy polymer [18,19]. Among the various organic mixtures, toluene-methanol binary mixtures are required to separate in many petrochemical and pharmaceutical industries. Both methanol selective [13,14] and toluene selective [5–12,15–19] membranes were reported for this separation. However, toluene forms an azeotrope with methanol at a composition of around 32% (w/w) toluene at atmospheric pressure [18]. Accordingly, separation of low concentration of toluene from its mixture with methanol using a toluene selective hybrid membrane is industrially more significant than selective methanol separation from this binary mixture using a methanol selective membrane. In our previous works hybrid type membranes prepared by incorporating carbon black in natural rubber and crosslinking these filled elastomer with sulfur were used for removal of organics from water [20] and also for the separation of aromatic-alcohol mixtures [15,16]. In continuation of this work, in the present case natural rubber was crosslinked with sulfur and zinc dithio carbamet (ZDC) by semi efficient

http://dx.doi.org/10.1016/j.jtice.2016.03.043 1876-1070/© 2016 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

90

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

method after incorporating three kinds of adsorptive fillers, i.e., N330 carbon black, bentonite clay and also zeolite. These hybrid membranes were used for separation of toluene-methanol mixtures over the feed concentration of 3–40 wt% toluene in methanol. 2. Experimental 2.1. Materials Solvents used for this study, i.e., toluene and methanol of high purity analytical grade were purchased from M/s. E. Merck (India) Ltd, Mumbai. Natural rubber (NR) of grade RSS-4, sulfur, zinc dithio carbamate (ZDC, accelerator) and carbon black filler of N330 grade (BSS No. 325 sieve ∼90%, Iodine Absorption Number 82 g/Kg, pore density 22 lb/ft³) were kindly given by Tyre Corporation of India, West Bengal. Smectone clay C (BSS No. 200 sieve ∼90%, moisture content-4.0 %, specific gravity-1.7) and zeolite (moisture content4.9%, specific gravity-2.2, thickness-4.5 μm) fillers were kindly supplied by Amrfeo pte. Ltd., Kolkata. 2.2. Preparation of filled and crosslinked natural rubber membranes The filled natural rubber membranes were prepared by similar method as reported elsewhere [13]. Natural rubber (NR) of RSS4 grade was masticated and cut in to small pieces. It was then swelled in toluene for 24 h followed by an addition of the required amounts (10 wt%) of the three different organoselective fillers in portions with mechanical stirring of the dispersion for 24 h. Sulfur (crosslinker) and ZDC (accelerator) were then added in 1:1 ratio and the dispersion was stirred for another 8 h. From this rubber dispersion membrane was cast on a smooth and clean glass plate and it was air dried overnight at ambient condition. The air dried membrane on a glass plate was then taken in a hot air oven and the membrane was cured (crosslinked) at 110 °C for 40 minutes. The cured membrane on glass plate was then cooled, immersed in cold water and the membrane was peeled out from the glass plate. Preparation of these filled NR membranes is shown in Scheme-1. The thickness of the filled NR membranes was maintained at ∼160 micron. The thickness was measured by Test Method ASTM D 374 using a standard dead weight thickness gauge (Baker, Type J17). 2.3. Membrane characterization 2.3.1. FTIR spectroscopy The FTIR spectra of the membrane samples were recorded on a Perkin Elmer (model-Spectrum-2, Singapore) spectroscope using a thin film (∼10 μm) of the membrane. 2.3.2. Mechanical strength The tensile strength (T.S.) and elongation at fracture (E.A.F) of the filled NR membranes were determined in by an Instron-Tensile tester (Instron 4301, Instron Limited, England). The experiment was performed according to ASTM D 882-97. 2.3.3. Morphology of the filled NR membranes by scanning electron microscopy (SEM) The membrane samples were coated with gold (Au). The morphology of the unfilled and the filled membranes were analyzed by SEM (Scanning electron Microscope, model no. S3400N, VP SEM, Type-II, made by Hitachi, Japan) with the accelerating voltage set to 15 kV. 2.3.4. Measurement of crosslink density by chemical and mechanical method The crosslink density of the filled NR membranes was measured by both chemical and mechanical method.

2.3.4.1. Measurement of crosslink density by chemical method. The cured natural rubber membrane was weighed and then immersed in pure toluene and kept for 72 h till equilibrium sorption was attained. When the sorption equilibrium was attained, there was no further increment of weight between two successive readings of sample weights at different times. Chemical crosslink densities of the cured membrane were determined from the following Flory– Rehner Eq. (1) [21].



1 ln(1 − vr ) + vr + χ v2r v= Vs v1r /3 − 1 vr



(1)

2

where v, Vs and Vr is the crosslink density, molecular volume and volume fraction, respectively, of the rubber membrane while χ is the interaction parameter between the solvent and the membrane. 2.3.4.2. Measurement of crosslink density by mechanical method. The crosslink density of the cured membrane was also determined by mechanical method using the following Eq. (2) based on the kinetic theory of elasticity in its simplest form [21].

  1 σ = RT v λ − 2 λ

(2)

Here, σ is the tensile stress required to extend the crosslink rubber sample to an extension ratio of λ, R is the universal gas constant, T is the absolute temperature and v is the number of crosslinks per cubic cm of the membrane sample. 2.4. Sorption studies The transport of solutes through the membrane is governed by a preferential sorption and diffusion due to the concentration gradient from the bulk feed to the downstream side of the membranes. The relative separation performances of the membranes were evaluated by pervaporation (PV) experiments. However, PV is a dynamic process combining both sorption and diffusion. The evaluation of the membranes by sorption is important to find out the operating condition for pervaporation [22]. 2.4.1. Total and partial sorption For sorption experiments the membrane samples of known weights were immersed in several toluene-methanol mixtures of known concentrations and allowed to equilibrate for 96 h at 30 °C. Each sample was weighed periodically until no weight change was observed. After sorption, these membranes were taken out from the solution and weighed after the superfluous liquid was wiped out with a tissue paper. Total sorption % (St ) of toluene and methanol mixtures by the membranes (%) is obtained as

St =

wi − wd x100 wd

(3)

where wd and wi are the weight of the membranes before and after immersion in toluene–methanol mixtures. 2.4.2. Sorption selectivity The membrane phase composition of toluene and methanol was determined by analyzing the sorbed liquid from the swollen membrane. The swollen membranes were taken in a conical flask which was connected to a vacuum pump through a cold trap immersed in a liquid nitrogen cryo can as shown in Scheme 1f. By applying vacuum, the liquid, i.e., total amount of toluene and methanol sorped by the swollen membrane were collected in the cold trap and analyzed by a digital refractometer (Anton Paar, model – AbbematHP, precision up to 5 decimal). From the mass of total sorption and

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

91

Scheme 1. Preparation of membranes for sorption and pervaporation experiment, a. natural rubber (NR, RSS-4 grade), b. mastication in 2 roll mill, c. rubber solution in toluene, d. filled rubber dispersion in toluene, e. cast membrane (i) NRC (10% clay), (ii) NRCB (10% carbon black), (iii) NRZ (10% zeolite), f. sorption experiments, g. pervaporation experiments.

corresponding toluene content of the membrane, sorption selectivity (α s ) of the membrane for toluene was calculated using the following Eq. (4).

αS =

mtoluene mmethanol xtoluene xmethanol

(4)

Here, mi and xi are the weight fraction of component i (toluene) in membrane and feed, respectively. 2.5. Permeation studies Permeation studies were carried out by pervaporation experiments in a batch stirred PV cell [23] with adjustable downstream pressure that was maintained at a low pressure of 1 mm Hg as shown in Scheme 1g. Effective membrane area (A) in contact with the feed solution was 19.6 cm2 and the volume of the feed compartment was 150 cm3 . The toluene–methanol mixtures in contact with the membrane were allowed to equilibrate for around 3 h for the first experiment and 1 h for the subsequent experiments with different feed compositions. The PV experiment was performed at a constant temperature by circulating constant temperature water around the jacket of the PV cell for different feed compositions and for a particular feed concentration at four different temperature i.e. 30 °C, 40 °C, 50 °C and 60 °C. At steady state permeates were collected in traps immersed in liquid nitrogen. The results for PV separation of toluene–methanol mixtures were reproducible, and the

errors inherent in the PV measurements were less than 1.0%. The wt. of the permeate was determined by a digital electronic balance. Permeation flux (J) was calculated by dividing the amount of total permeate (W) by the time (t) of experiment and the area of the membrane (A) from the following Eq. (5).

J=

W At

(5)

The amount of toluene present in the permeate was determined in a similar way as in the case of membrane phase composition for sorption experiment by the Abbey type digital refractometer. The separation factor (PV) of toluene was expressed as ytoluene

αPV = yxmethanol toluene

(6)

xmethanol

Here, yi and xi are weight fraction of component i (toluene) in permeate and feed, respectively. 2.5.1. Permeability and intrinsic membrane selectivity The driving force normalized flux or permeability of the solvents was obtained from the thickness (L) of the membrane and the vapor pressure differential of the solvents between feed and permeate sides using the following Eq. (7) [24]

Pi =

Ji L

(x f i γi psat − y pi p p ) fi

(7)

92

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

where xfi and ypi are the mole fraction of the component i on the feed (liquid phase) and the permeate (vapor phase) side, pfi sat is the saturated vapor pressure of component i on the feed side obtained using Antoine’s equation and pp is the permeate pressure in PV process. Ignoring the very low pressure on permeate side, Eq. (7) reduces to

Pi =

Ji L f

(7.a)

where fugacity,

f = x f i γi psat fi

(7.b)

Intrinsic membrane selectivity (α s ) is defined as the ratio of partial permeability

αs =

Pi Pj

(8)

3. Results and discussion 3.1. Preparation of rubber membranes 3.1.1. Grades of rubber The natural rubber used for this study is a natural polymer obtained from rubber tree containing more than 90% cis-1, 4polyisoprene (Scheme 1a) along with naturally occurring resins, proteins sugars etc. The ribbed smoked sheet or RSS-4 grade is obtained by employing aluminum partition vertically into the coagulation tank. RSS-4 grade was selected as it contains minimum amount of non-rubber impurities.

3.1.2. Mastication of natural rubber The ‘nerve’ of the high molecular weight rubber was broken by passing the rubber sheet continuously through a small two roll mill (6 in. diameter) as shown in Scheme 1b. As the rubber sheet passes through the roll, the moving rubber band is cut with a knife at ambient condition. This process of ‘mastication’ breaks down the high molecular weight rubber to a lower molecular weight fragments in the presence of oxygen. The masticated rubber easily goes to solution.

3.1.3. Making rubber solution Natural rubber being a high molecular weight polymer produces highly viscous solution in toluene even with a low concentration and a highly viscous solution is difficult to cast on the glass plate as a membrane material. On the other hand two low concentration of a rubber in toluene produces very thin membrane of poor tensile strength. By repeated trial and error methods 3 wt% of masticated rubber in toluene was found to produce membrane of acceptable thickness (∼160 micron) for filled membrane.

3.1.4. Crosslinking reaction The crosslinking of natural rubber with sulfur in the presence of accelerator, also called vulcanization results in the formation of monosulfide, polysulfide as well as cyclic linkages between rubber chains. Amongst these linkages cyclic linkage of sulfur with a rubber chain only gives intramolecular bond with no restriction in permeability while polysulfide linkages of C–Sx –C type (C is carbon of a rubber chain, x is an index or number of S atoms in a polysulfide linkages) gives free mobility of chain segments. The longer the bridge links in C–Sx –C, i.e., the higher the number of x in this

type of polythioether crosslinks, the easier it is for the individual chains [25] to move when the vulcanized or crosslinked rubber membranes are subjected to any stress like pressure differential in the pervaporation process resulting in an easy permeation of the solvents. On the other hand crosslinking with much stronger and rigid monosulfide bond produces maximum restriction in the rubber matrix which is likely to give high selectivity of the desired solvent during pervaporation. The ratio of these different types of crosslinking can be varied by changing the sulfur to accelerator ratios. In the present work ‘semi efficient’ type vulcanization was employed since this kind of vulcanization with sulfur: accelerator weight ratio of 1:1 has been reported [25] to give the maximum of monosulfide linkage and hence organic selectivity during pervaporation. Like conventional vulcanization, the sulfur crosslinking or vulacanization reaction for the present membranes could not be carried out in a compression mould at a definite pressure and temperature since the cast membranes were too thin to be cured in a compression mould. Instead vulcanization reactions were carried out by an open cure of the cast rubber film in a hot-air-oven at the cure temperature without any pressure. Open cure of rubber in the absence of any pressure demands higher doses of sulfur and accelerator. In this work, sulfur and accelerator concentration was thus fixed at 6 wt% each for semi efficient vulcanization.

3.1.5. Choice of accelerator For the present work an ultra fast room temperature curing accelerator, i.e., zinc dithio carbamate (ZDC), was used. This accelerator was chosen as it has been reported to produce less of polysulfide bridges [25]. This accelerator also causes rapid curing at room temperature. However, in this present system of open cure in the absence of any pressure higher temperature and longer curing (crosslinking) time is required for getting optimum crosslinking density. The curing time was 40 min at 160 °C in a hot air oven. The breaking of the sulfur crosslinking or ‘reversion’ was encountered when the curing time was more than 40 min at this temperature.

3.1.6. Choice of filler In the present work three types of filler, i.e., carbon black of N330 grade, micro size bentonite clay and zeolite filler were used. The carbon black N330 filler was earlier reported to improve mechanical strength and selectivity of natural rubber [16,20], SBR [16,26] and EPDM rubber [27]. The bentonite clay was earlier found to improve organic selectivity of glassy polymers [28] while zeolite has been extensively used for improving membrane properties of PDMS rubber [29]. For comparison, 10 wt% of each filler was used since in the absence of mixing in a two roll mill, it is difficult to incorporate more than 10 wt% of bentonite clay and zeolite in the rubber matrix. The three kinds of vulcanized and filled natural rubber membranes containing 10 wt% carbon black, organophilic bentonite clay and zeolite, designated as NRCB, NRC and NRZ, respectively, were used for sorption and pervaporation of 3–40 wt% toluene in methanol.

3.2. Characterization of the membranes 3.2.1. FTIR spectra The FT-IR spectra of the vulcanized natural rubber (unfilled) membrane are shown in Fig. 1. The natural rubber is observed to show all of its characteristic absorption spectra including absorption due to stretching and bending of the bonds formed by crosslinking reactions, i.e., a strong absorption band due to C–S at 740 cm−1 corresponding to rubber–sulfur crosslinking and a weak absorption at S–S at 468 cm−1 [30].

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

93

Fig. 1. FTIR of unfilled natural rubber (NR0).

3.2.2. Mechanical strength The tensile properties, i.e., modulus at 100% elongation, tensile strength and elongation at fracture of the NRCB, NRC and the NRZ membranes are shown in Table 1. The high tensile strength and elongation indicates strong filler–polymer interaction. Amongst the three membranes, the carbon black incorporated NRCB membrane shows the highest tensile strength since it is a reinforcing filler. The other two membranes show lower tensile strength than NRCB but elongation of these membranes are higher than NRCB membrane. However, the tensile strength and the elongation at break of the three membranes are high enough for use in pervaporation. The solvent resistance properties of the membranes were studied by evaluating its tensile properties after immersion in pure toluene and also in pure methanol for 1 week. The results are also presented in Table 1. From the results it is evident that the membranes retain its mechanical stability in the solvents. It is observed that modulus or tensile strength of the membranes decreases to some extent (around 10%) in toluene though the solvent swollen membranes show higher elongation. It is also observed that the change in mechanical properties of the methanol swollen membranes is marginal which may be because of the toluene selectivity of the membranes. The mechanical properties of the membranes used in sorption and pervaporation experiments of 3–40 wt% toluene in methanol are likely to be intermediate between its mechanical properties in pure (100 wt%) methanol and pure toluene.

3.2.3. SEM The SEM of the three filled membranes is shown in Fig. 2. The SEM of the unfilled natural rubber is also shown (Fig. 2a) for comparison. The unfilled membrane shows dense featureless morphology even at a high magnification of 10,0 0 0x as expected. For observing the distribution of filler in the rubber matrix, the SEMs of the filled elastomers were carried out at a lower magnification of 1500 x. The three kinds of micron sized fillers, i.e., carbon black, bentonite and zeolite are observed to be more or less evenly distributed in the matrix of the rubber membrane without any significant agglomeration indicating good filler–rubber compatibility.

3.2.4. Crosslink density The crosslinking of the natural rubber by sulfur was measured in terms of ‘chemical’ crosslink density. This chemical crosslink

density was measured by both chemical and physical method. In the chemical method the rubber membrane was immersed in pure toluene and the crosslink density was measured from the swelling data using the Flory–Rehner Eq. (1). The value of interaction parameter between toluene and the natural rubber was obtained from literature [31]. Similarly, the chemical crosslink density of the membranes was also determined by ‘physical method from tensile data using Eq. (2) based on the kinetic theory of elasticity. However, there is also physical crosslinking in the filled rubber [16] due to filler–rubber interactions. The physical crosslinking due to only filler was determined by subtracting chemical crosslink density of the unfilled NR0 membrane from the crosslink density of the filled NRCB, NRC and NRZ membrane. These values are shown in Table 2. It is observed that the types of filler influences both chemical crosslinking, i.e., sulfur crosslinking and also filler–rubber interaction, i.e., physical crosslinking. Thus, for the same amount (10 wt%) of filler in natural rubber, both chemical and physical crosslinking show the following trend, i.e., NRCB> NRC>NRZ indicating the same order of filler–rubber interaction.

3.3. Sorption studies 3.3.1. Effect of feed concentration on sorption isotherm The effect of feed toluene concentration on (a) total sorption and sorption selectivity for toluene and (b) partial sorption at 30 °C is shown in Fig. 3a and b, respectively. Similar sorption isotherms were also observed at higher temperature. From these figures it is evident that the filled membranes show high sorption and sorption selectivity for toluene. It is observed that the total sorption or partial toluene sorption shows similar trend, i.e., it increases with an increase in feed toluene concentration while for the same feed concentration sorption increases in the following order NRZ>NRC>NRCB. The sorption selectivity for toluene follows the opposite trend, i.e., it increases in the reverse order of NRCB>NRC>NRZ. Thus, for the feed concentration of around 3– 40 wt% toluene in methanol, the NRCB, NRC and NRZ membrane shows a sorption of 13–42%, 15–46% and 16–49%, respectively and a toluene selectivity of 55–10.6, 50–7.6 and 48–5.6, respectively. This result is in a good agreement with the crosslink density of the membrane which also follows the same trend of sorption selectivity, i.e., NRCB>NRC>NRZ as observed in Table 2. An increase

94

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

Table 1 Mechanical properties of the membranes. Membrane

NRCB NRC NRZ

Modulus at 100% elongation/TS, unused dry membrane

Modulus at 100% EAF% unused dry elongation/TS after one membrane week immersion in methanol (MPa)

EAF% after one week immersion in toluene

EAF% after one week immersion in methanol

(MPa)

Modulus at 100% elongation/TS, after one week immersion in toluene (MPa)

1.04/19.5 0.935/17.5 0.88/16.5

0.85/16.3 0.78/14.5 0.73/12.3

0.96/18.2 0.89/16.7 0.82/11.8

438.7 527.5 531.5

428.3 511.5 515.5

425.7 509.4 512.4

Fig. 2. SEM of (a) NR0 (unfilled), x 10k, (b) NRCB, x 1.5k (c) NRC, x 1.5k (d) NRZ, x 1.5k. Table 2 Crosslink density of the membranes. Membrane

NR0 NRCB NRC NRZ

Chemical crosslink density by physical method x 105

Chemical crosslink density by chemical method x 105

x 105

x 105

0.58 1.82 1.36 1.27

0.523 1.72 1.19 1.12

0 1.197 0.667 0.597

0 1.24 0.78 0.69

in the crosslink density imposes more restriction in the permeation resulting in higher sorption selectivity. However, at higher feed concentration (> 10 wt% toluene in feed) because of the high swelling and plasticization the sorption also increases significantly while sorption selectivity for toluene decreases. From Fig. 4 a and b it is also observed that the sorption isotherm closely resembles the Type III sorption of Rogers [32], i.e., in this case toluene-methanol

Physical crosslink density by chemical method

Physical crosslink density by physical method

interaction is more than the toluene-membrane or methanolmembrane interaction and thus these solvents form cluster in the membrane matrix. From Fig. 4b it is observed that the methanol sorption is much lower than toluene sorption and above around 5 wt% toluene in feed it becomes almost constant. The membranes preferentially sorb toluene since the solubility parameter of toluene (18.2 MPa0.5 ) is more close to the solubility parameter of

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

95

a 70

0.55

Total sorption (g/g membrane)

50 0.35

40

0.25

30 20

0.15

Sorption selectivity For toluene (-)

60

0.45

10

0.05

0 0

10

20

30

40

50

Feed conc (wt% toluene in methanol) NRCB total sorp NRCB sel

NRC total sorp NRC sel

NRZ total sorp NRZ sel

0.5

0.16

0.4

0.13

0.3

0.1

0.2

0.07

0.1

0.04

0

Methanol sorption (g/g membrane)

Toluene sorption (g/g membrane)

b

0.01 0

10 20 30 40 Feed conc (wt% toluene in methanol)

NRCB toluene sorp NRCB methanol sorp

NRC toluene sorp NRC methanol sorp

50 NRZ toluene sorp NRZ methanol sorp

Fig. 3. Effect of feed concentration on (a) total sorption and sorption selectivity for toluene, (b) partial sorption at 30 °C.

natural rubber (16.5 MPa0.5 ) than methanol (29.6 MPa0.5 ). Similar trend of sorption results were reported for carbon black filled SBR [16] and EPDM rubber [25] for toluene–methanol mixtures. 3.4. Permeation studies 3.4.1. Effect of feed concentration 3.4.1.1. Membrane phase and permeate concentration. The effect of feed toluene concentration on its membrane phase and permeate concentration at 30 °C is shown in a McCabe–Thiele type xy diagram in Fig. 4a. The membrane phase and permeate concentration was determined from the sorption and pervaporation (PV) experiments, respectively. Some of the similar vapor–liquid equilibrium data of toluene–methanol mixture is also shown in Fig. 4a for comparison. From the figure it is observed that the membranes show measurable separation in both sorption and pervaporation. It is also observed that for the comparable feed concentration the separation of toluene from methanol by sorption and pervaporation is much higher than its VLE data without any azeotrope. 3.4.1.2. Flux and separation factor. The effect of feed concentration on (a) total flux and separation factor for toluene and (b) partial

toluene and methanol flux is shown in Figs. 4b and 5a, respectively. The total or partial flux follows the same trend of sorption, i.e., it also increases with an increase in the feed concentration. It is observed that at and above 10 wt% toluene in feed there is a significant increase in total or partial toluene flux. This may be because of the plasticization of the toluene selective rubber membranes at higher feed toluene concentration. It is observed that for the same feed concentration the flux also follows the same order of sorption, i.e., it increases in the order: NRZ>NRC>NRCB while the separation factor follows the opposite order. Thus, for around 3–40 wt% toluene in feed the NRCB, NRC and the NRZ membrane shows a total flux of 0.026–0.321, 0.0281–0.342 and 0.0290– 0.351 kg/m2 h, respectively and a toluene selectivity (separation factor for toluene) of 149–103, 141–90 and 134–85, respectively. The decrease in flux or increase in toluene selectivity from NRCB to NRZ may be ascribed to an increase in the crosslink density in the same order. From Fig. 5a it is also observed that methanol flux is much lower than the toluene flux and methanol flux remains marginally constant at and above 10 wt% toluene in the feed. The low methanol flux may be attributed to the wide difference of solubility parameter of methanol from the rubber membrane resulting in very low sorption of methanol as also observed in Fig. 3b. However, the kinetic diameter of methanol (0.38 nm) [33] is much smaller

96

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

a

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0

190 170 150 130 110 90 70

Separation Factor For toluene (-)

2

(kg/m h)

Total Flux

b

50 0

10

20

30

40

50

Feed conc (wt% toluene in methanol) NRCB flux

NRC flux

NRZ flux

NRCBsel

NRCsel

NRZsel

Fig. 4. Effect of feed concentration on (a) membrane and permeate concentration, (b) total flux and separation factor for toluene at 30 °C.

than the kinetic diameter of toluene (0.61 nm) [34]. Hence, the organophilic rubber membranes also show some methanol permeation as observed in Fig. 5a. It is interesting to note in Fig. 4b that the decrease in toluene selectivity is around 23–25% up to around 10 wt% feed toluene concentration but there after the change in toluene selectivity is marginal up to the maximum feed concentration. In fact, at high feed concentration the toluene selectivity decreases for most of the elastomeric membranes because of its extensive swelling. However, in the present case the adsorptive filler not only increases selectivity by increasing physical crosslink density but it also prevents the membrane from excessive swelling to retain even high selectivity at higher feed concentration of toluene. 3.4.1.3. Permeability and intrinsic membrane selectivity. The variation of permeation rate (flux) or selectivity of the membranes with feed concentration as shown in the above Figs. 4b and 5a may be due to the effect of feed concentration on the vapor pressure of the permeating solvents or on the intrinsic membrane properties. To study the effect of feed concentration only on the intrinsic membrane properties the solvent vapor pressure is normalized to obtain the membrane permeability (flux times membrane thickness divided by fugacity) and intrinsic membrane selectivity. The varia-

tion of fugacity of toluene and methanol with feed concentration is shown in Fig. 5b while the variation of (a) total solvent permeability and intrinsic membrane selectivity and (b) partial permeability with feed concentration is shown in Fig. 6a and b, respectively. The results of these figures clearly indicate strong dependence of the concentration on intrinsic membrane properties. It is observed that unlike flux, the total or toluene permeability decreases with feed concentration up to around 10 wt% toluene in feed and thereafter permeability increases significantly with feed concentration. In fact, the rate of increase of vapor pressure of toluene (Fig. 5b) more than offset the rate of increase of flux up to around 10 wt% feed toluene concentration. Thus, permeability decreases with concentration. However, at and above 10 wt% of toluene in feed the flux increases significantly due to the plasticization of the membranes as also observed in Fig. 4a and b. At and above this feed concentration the rate of increase of vapor pressure or fugacity (Fig. 5b) for toluene is marginal resulting in an increase in the rate of permeability. From Fig. 6b it is observed that like methanol flux, methanol permeability is also much less than toluene permeability over the entire feed concentration and thus, the membranes show high membrane selectivity for toluene over the entire feed concentration as also observed in Fig. 6a. The methanol permeability is also observed to follow the similar trend

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

97

a 0.01

0.4

0.005

0.2

2

2

0.15

(kg/m h)

0.25

Methanol Flux

0.0075

0.3

(kg/m h)

Toluene Flux

0.35

0.0025

0.1 0.05

0

0

0

10

20 30 40 Feed conc (wt% toluene in methanol)

NRCB toluene NRCB methanol

50

NRC toluene NRC methanol

NRZ toluene NRZ methanol

24

3.5

22

3

20

2.5

18

2 16

1.5

14

Fugacity of toluene (kPa)

1

Fugacity of methanol (kPa)

0.5

methanol (kPa)

4

Fugacity of

toluene (kPa)

Fugacity of

b

12

0

10 0

10 20 30 40 Feed conc. (wt% toluene in methanol)

50

Fig. 5. Effect of feed concentration on (a) partial flux and (b) fugacity at 30 °C.

of methanol flux, i.e., initially it increases and at and above 5 wt% toluene in feed it becomes marginally constant. The low methanol permeability may also be attributed to very high vapor pressure and hence fugacity of methanol as observed in Fig. 5b.

3.4.1.4. Diffusion coefficient. Based on Fick’s 1st law the flux of a solvent i through a pervaporation membrane may be expressed as [35]

Ji = −Di

dϕim dl

(10)

(11)

Here the six model parameters ψ ii , ψ ij , ψ jj , ψ ji , Di 0 and Dj 0 describes [37] the permeation of the solvent mixtures through the membrane for limited swelling and the plasticization coefficient ψ ab is plasticization of component b on the permeation of component a. Substituting Eqs. (10) and (11) on Eq. (9) and integrating with the boundary condition of ϕ = ϕ u (concentration at upstream side of the membrane) at l = 0 and ϕ = ϕ d at l = L and also assuming zero concentration on the downstream side of the membrane (ϕ d = 0) because of very low pressure,

(9)

Where Di is diffusion coefficient of i through the membrane, ϕim is its concentration in terms of volume fraction in the membrane and l is position variable across the thickness (L) of the membrane. At a constant feed temperature the diffusion coefficient of the solvents i and j (toluene and methanol) depends on its concentration in the membrane and based on solution diffusion model it may be expressed as [36]

Di = Di0 exp(ψii ϕim + ψi j ϕ jm )

D j = D j0 exp(ψ j j ϕ jm + ψ ji ϕim )

 Ji =

 Jj =

D i 0 ρm (ψii − ψi j )L

 

D j0 ρm (ψ j j − ψ ji )L





exp(ψii ϕiu + ψi j ϕ ju ) − exp(ψi j )



(12)



exp(ψ j j ϕ ju + ψ ji ϕiu ) − exp(ψ ji )

(13)

For transport of solvent self diffusion is more effective than coupled diffusion [37] i.e., ψ 11 > ψ 12

98

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

80

700 650 600 550 500 450 400 350 300 250 200 150

Total Permeability (Barrer)

70 60 50 40 30 20 10 0

10

20

30

40

Membrane selectivity For toluene(-)

a

50

Feed conc (wt% toluene in methanol) NRCB perm NRCB MS

NRZ perm NRZ MS

70 65 60 55 50 45 40 35 30 25 20

0.2

0.15

0.1

0.05

Methanol permeability (Barrer)

Toluene Permeability (Barrer)

b

NRC perm NRC MS

0

0

10

20

30

40

50

Feed conc (wt% toluene in methanol) NRCB toluene NRCB methanol

NRC toluene NRC methanol

NRZ toluene NRZ methanol

Fig. 6. Effect of feed concentration on (a) permeability and intrinsic membrane selectivity for toluene, (b) partial permeability at 30 °C.

Hence, the above Eq. (12) may be written as



Ji =

D i 0 ρm (ψii − ψi j )L

 

or,

Ji +

D i 0 ρm = (ψii − ψi j )L





exp(ψii ϕiu + ψi j ϕ ju ) − 1

D i 0 ρm (ψii − ψi j )L





(14)



exp(ψii ϕiu + ψi j (1 − ϕiu )

(15) Considering very low value of

 Ji =

D i 0 ρm (ψii − ψi j )L

or,



ln Ji = ln

 

Di0 ρm (ψii −ψi j )L

with respect to Ji



exp(ψii ϕiu + ψi j (1 − ϕiu )

D i 0 ρm (ψii − ψi j )L

(16)

 + [ψii ϕiu + ψi j (1 − ϕiu )]

(17)

Similarly, for component j (methanol)



ln J j = ln

D j0 ρm (ψ j j − ψ ji )L



+ [ψ j j ϕiu + v ji (1 − ϕiu )]

(18)

The density (ρ m ) and thickness (L) of the membrane was obtained experimentally while the membrane phase concentration

(volume fraction) of i (toluene) and j (methanol) on the upstream side of the membranes (ϕ iu and ϕ ju ) were obtained from the sorption experiments. The linear plotting of the logarithmic of experimental partial flux data (ln J) against upstream side membrane phase concentration of the solvent (ϕ iu or ϕ ju ) yield the values of the four plasticization coefficients (ψ ii , ψ jj , ψ ij and ψ ji ) and the diffusion coefficient of the solvents at infinite dilution (Di 0 and Dj 0 ). These linear fittings were carried out in Origin software (Origin 8) based on Levenberg–Marquardt (L–M) algorithm where parameter values of a model are adjusted in an iterative process using chi square (χ 2 ). The validity of the model was evaluated in terms of the regression coefficient (r2 ) and non linear χ 2 values. For a good fitting, r2 should be close to unity and χ 2 will be low. [38]. The fitting of the experimental data to Eqs. (17) and (18) is shown in Fig. 7a while the values of the model parameters, i.e., self (ϕ ii and ϕ jj ) and coupled plasticization coefficients (ϕ ij and ϕ ji ) and diffusion coefficients at infinite dilution (Di 0 and Dj 0 ) at 30 °C along with the statistical parameters are shown in Table 3. It is observed that the values of the diffusion coefficient at infinite dilution are also in good agreement with the experimental results, i.e., the diffusion coefficient at infinite dilution of toluene (Di 0 ) decreases from NRCB to NRZ while diffusion coefficient at infinite dilution of methanol (Dj 0 ) is lower than Di0 . Similar order of Di 0 and Dj 0 were reported elsewhere [7,15,16,37]. The negative

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

99

ln (partial flux)

a -3.0 -3.2 -3.4 -3.6 -3.8 -4.0 -4.2 -4.4 -4.6 -4.8 -5.0 -5.2 -5.4 -5.6 -5.8 -6.0 -6.2 -6.4 -6.6

NRCBtoluene NRCtoluene NRZtoluene NRCBmethanol NRCmethanol NRZmethanol

0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 ϕ u (-)

b 0.2

1.2

0.15

0.9 0.1 0.6 0.05

0.3

Diffusion coeffcient of Methanol (m2 /s) x 10 12

Diffusion coeffcient of Toluene (m2 /s) x 10 10

1.5

0

0 0

10

20

30

40

50

Feed conc (wt% toluene in methanol) NRCB toluene NRCB methanol

NRC toluene NRC methanol

NRZ toluene NRZ methanol

Fig. 7. (a) Model fitting for diffusion coefficients, (b) effect of feed concentration on diffusion coefficient at 30 °C.

Table 3 Diffusion coefficient at infinite dilution of toluene (Di 0 ) and methanol (Dj0 ) at 30 °C temperature. Membrane

Di 0 x 1010 (m2 /s)

Self/coupled plasticization coefficient (ϕ ii /ϕ ij ) (Dimensionless)

r2 /

χ 2 x 103

Dj 0 x 1012 (m2 /s)

Self/coupled plasticization coefficient (ϕ jj /ϕ ji ) (Dimensionless)

r2 / χ 2 x 103

NRCB NRC NRZ

2.1 1.77 1.35

0.65/−5.36 0.55/−5.11 0.53/−4.11

0.97/8.27 0.98/8.29 0.99/4.8

3.39 6.65 18.9

−4.44/−4.71 −4.47/−4.99 −4.42/−5.9

0.94/0.04 0.92/0.2 0.92/0.23

values of ϕ ij or ϕ ji indicates negligible effect of coupled diffusion on transport of the solvents [37]. The values of the statistical parameter (r2 ∼0.94–0.99 and χ 2 ∼10−3 order) and Fig. 7a clearly shows good fitting of the experimental results to the above two equations based on the solution diffusion models. The diffusion coefficients of the solvents at varied feed concentrations were obtained by putting the values of these six parameters in Eqs. (10) and (11). Accordingly, the effect of the feed concentration on the diffusion coefficient of toluene and methanol at 30 °C is shown in Fig. 7b. The diffusion coefficient of toluene increases with an increase in the feed concentration. At higher feed concentration of toluene, the membranes are plasticized allowing higher rate of diffusion of toluene. Over the entire feed concentration the diffusion coefficient of methanol is observed to be much lower than the diffusion coefficient of toluene resulting in toluene selectivity of the membranes.

3.4.2. Effect of feed temperature The effect of feed temperature on (a) total flux and separation factor for toluene and (b) partial solvent flux is shown in Fig. 8a and b, respectively. It is observed that the total or partial flux increases almost linearly with temperature. The increase in the flux with temperature may be ascribed to an increase in the vapor pressure of the solvents with temperature. Further, at higher temperature the movements of the rubber chains of the membrane also increases resulting in an increased free volume in the membrane matrix which also promote easier permeation of the solvent molecules through the membranes. However, methanol (boiling point 64.7 °C) is much more volatile than toluene (boiling point 110.6 °C) and hence with an increase in the temperature the vapor pressure of methanol increases at a much higher rate than the vapor pressure of toluene as shown in terms of fugacity in Fig. 10a. Thus, at higher temperature the

100

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

a 150

Total Flux (kg/m2h)

0.09

125

0.08 100 0.07 75

0.06 0.05

Separation Factor for toluene (-)

0.1

50 25

35

45

55

65

Feed Temperature (oC) NRCB flux

NRC flux

NRZ flux

NRCB SF

NRC SF

NRZ SF

b 0.015

0.08

0.07 0.01 0.0075 0.06

Methanol Flux (kg/m2 h)

Toluene Flux (kg/m2 h)

0.0125

0.005 0.0025

0.05 25

35 NRCB toluene NRCB methanol

45 55 Feed Temperature (oC)

65

NRC toluene NRC methanol

NRZ toluene NRZ methanol

Fig. 8. Effect of feed temperature on (a) total flux and separation factor for toluene (b) partial flux. Feed conc. 10.8 wt% toluene in methanol.

rate of increase of methanol flux is much higher than the rate of increase of toluene flux. In Fig. 8b, the increase in feed temperature from 30 to 60 °C results in an increase of only 19, 9 and 8% toluene flux but 93, 94 and 95 % methanol flux for NRCB, NRC and NRZ membrane, respectively. In fact, the smaller methanol molecules diffuse at a much faster rate than the bigger toluene molecules at higher temperature. Thus, in Fig. 8a the toluene selectivity is observed to decrease almost exponentially at higher temperature.

3.4.2.1. Permeability and intrinsic membrane selectivity. The effect of temperature on (a) permeability and intrinsic membrane selectivity and (b) partial permeability is shown in Fig. 9a and b, respectively. It is observed that unlike flux, the total or partial permeability decreases almost exponentially with temperature. The increase of vapor pressure of toluene or methanol (Fig. 10a) more than offset the increase of flux for both methanol and toluene. Thus, at higher temperature the permeability of both toluene and methanol decreases. Accordingly, the intrinsic membrane selectivity which is the ratio of the partial permeability of toluene and methanol also decreases at higher temperature.

Table 4 Apparent activation energy (Ep ) for permeation. Membrane

Eact toluene (kJ/mol deg)

r2 (toluene)

Eact methanol (kJ/mol deg)

r2 (methanol)

NRCB NRC NRZ

3.36 1.95 1.74

0.9831 0.9404 0.9833

3.66 4.48 4.91

0.9833 0.9543 0.9527

3.4.2.2. Activation energy for permeation. The apparent activation energy for permeation of toluene and methanol was obtained by the Arrhenious type plot of ln (partial flux) against inverse of absolute temperature (1/T) as shown in Fig. 10b. The apparent activation energy for permeation of toluene (Eact toluene ) and methanol (Eact methanol ) along with the regression coefficients (r2 ) of these linear fittings are shown in Table 4. The values of regression coefficients are observed to be close to unity (0.94–0.98) indicating good fitting of the experimental flux data to these linear trendlines while the values of Eact toluene and Eact methanol are in close agreement with experimental results, i.e., the Eact toluene is lower than Eact methanol .

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

101

a 360

200

330

Permeability (Barrer)

270

Intrinsic membrane selectivity (-)

180

300

160

240 210

140

180 120

150 120

100

90 60

80 25

35

45 55 Feed Temperature (oC)

NRCB perm NRCB MS

NRC perm NRC MS

65 NRZ perm NRZ MS

b

Toluene Permeability (Barrer)

1.75

300 270

1.5

240

1.25

210

1

180

0.75

150 0.5

120

0.25

90

Methanol Permeability (Barrer)

2 330

0

60 25

35

45 55 Feed Temperature (oC)

NRCB toluene NRCB methanol

NRC toluene NRC methanol

65 NRZ toluene NRZ methanol

Fig. 9. Effect of feed temperature on (a) permeability and intrinsic membrane selectivity (MS) for toluene () partial permeability. Feed conc. 10.8 wt% toluene in methanol.

3.5. Effect of membrane thickness on membrane flux and selectivity For any homogeneous membrane flux decreases but selectivity increases linearly with an increase in membrane thickness [16]. However, in the present hybrid membranes the effect of thickness on membrane flux and selectivity will strongly depend on distribution of the adsorptive fillers within the membrane matrix. If the distribution of filler is uniform, the properties of the membrane will also be uniform across its thickness and it will show linear relationship between thickness and flux or thickness and selectivity. From Fig. 11a it is observed that flux and selectivity data at varied membrane thickness of the

three hybrid membranes at 30 °C for 10.8 wt% toluene in methanol show a linear fittings with regression coefficient > 0.95. These results indicate uniform distribution of fillers within the membrane matrix. 3.6. Stability of the membrane To evaluate the chemical stability of the membrane the NRZ membrane was subjected to FTIR analysis before and after use in pervaporation experiments as shown in Fig. 11b. It is observed that there is no significant change in the absorption of the characteristic functional groups of natural rubber at 741 and 465 cm−1 or zeolite

102

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

a 6

90

Fugacity of toluene (kPa) Fugacity of methanol (kPa)

5

80

4

60

50 3 40

2

30

Fugacity of methanol (kPa)

Fugacity of toluene (kPa)

70

20 1 10

0

0 25

30

35

40

45

50

55

60

65

Feed temperature oC

b 0 .8

ln (partial molar flux)

0 .7 0 .6

N N N N N N

0 .5 0 .4 0 .3

R R R R R R

C C Z C C Z

B to lu e n e to lu e n e to lu e n e B m e th a n o l m e th a n o l m e th a n o l

0 .2 0 .1 0 .0 0 3 0 0 0 .0 0 3 0 5 0 .0 0 3 1 0 0 .0 0 3 1 5 0 .0 0 3 2 0 0 .0 0 3 2 5 0 .0 0 3 3 0 1 /T (K

-1

)

Fig. 10. (a) Effect of feed temperature on fugacity, (b) Arrhenius plot for apparent activation energy for permeation.

filler at 1663 and 575 cm−1 of NRZ membrane before and after use in pervaporation experiment. However, the absorption peaks also depends strongly on thickness of the sample and hence ratio of absorption peaks should be considered for comparison. The ratio

of the absorption peak of 1663 cm−1 of rubber to 741 cm−1 of filler changes marginally from ‘1.68’ before pervaporation experiment to ‘1.71’ after the experiment indicating chemical stability of the membrane.

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105

103

a R2 = 0.9809

160

R2 = 0.9962

140

0.08

120

R2 = 0.9758

100

0.06

80

2

R = 0.967

0.04

2

R = 0.994

2

R = 0.9948

60

Separation factor (-)

Total Flux (kg/m2h)

0.1

40

0.02

20 0

0 40

50

60

70

80

90

Membrane thickness (µm)

Total flux NRCB

Total flux NRC

Total flux NRZ

SF NRCB

SF NRC

SF NRZ

b

Fig. 11. Effect of membrane thickness on flux and separation factor. Feed conc. 10.8 wt% toluene in methanol, temperature 30 °C, b. FTIR of NRZ membrane before and after pervaporation.

3.7. Comparison with literature data Several toluene selective membranes of different thicknesses have been reported for its separation from methanol at varied feed concentrations and temperatures including azeotropic composition. The thickness normalized flux and separation factor for toluene of reported membranes are compared with the present membranes in Table 5. It is observed that for the similar feed concentration including azeotropic concentration of toluene in methanol, the present membranes show much higher flux and selectivity than most of the reported membranes. 4. Conclusion Three kinds of filled rubber membranes were prepared by incorporating three different kinds of adsorbent filler, i.e., carbon black, bentonite and zeolite in natural rubber. The filled membranes were crosslinked with 1:1 wt ratio of sulfur and ZDC accelerator by semi efficient vulcanization. The sulfur crosslinking

of the membranes were characterized by FTIR and determined by both chemical and physical method while filler–rubber interaction was evaluated in terms of physical crosslink density. The distribution of filler in the membrane matrix was evaluated by SEM. The carbon black filled membrane showed the highest physical crosslinking. The flux and permeability of the membranes were found to be strongly influenced by feed concentration, feed temperature and crosslink density of the membranes. There was marginal change of tensile properties or absorption bands of the characteristics functional groups of these hybrid membranes before and after pervaporation experiments indicating good mechanical and chemical stability of the membranes. These hybrid membranes showed high flux and toluene selectivity for toluene–methanol binary mixtures. The intrinsic membrane properties, i.e., the permeability and selectivity were observed to decrease drastically at higher temperature. These organophilic membranes may also be effective for pervaporative separation of other organic having similar solubility parameter with this membrane polymer.

104

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105 Table 5 Comparison of the present membranes with other reported membranes for pervaporative separation of toluene from its mixture with methanol. Feed concentration, temperature

Membrane used

Flux (kgμ m/m2 h)

Separation factor for toluene

Ref.

32 wt% toluene in feed, 30 °C 32 wt% toluene in feed, 30 °C 0.5 wt% toluene in feed, 30 °C 10.5 wt% toluene in feed, 30 °C 0.5 wt% toluene in feed, 30 °C 10.5 wt% toluene in feed, 30 °C 32 wt% toluene in feed, 35 °C 32 wt% toluene in feed, 25 °C

Chitosan-10 PET-g-PS SBR-1 NR-3 NR-20

1.25 0.102 0.6 2.3 0.5 2.5 0.4 96.36 113.14 0.38 45.9 44.7 41.1 10.6 11.8 12.5

25 40 162 25 286 150 9.6 2.67 3.66 8.3 95.7 100 109 113.7 105.9 102.7

[10] [12] [15]

32 wt% toluene in feed, 30 °C 32 wt% toluene in feed, 30 °C

10.5 wt% toluene in methanol, 30 °C

PN-10 PDMS PU-PDMS SPA-15 NRZ NRC NRCB NRZ NRC NRCB

[16] [17] [18] [19] Present work

Chitosan-10- chitosan containing 10% zeolite, PET-g-PS-polythylene grafted polystyrene, NR-3 –‘efficiently vulcanized natural rubber, SBR-1-styrene butadiene rubber with conventional vulcanization, NR-20- natural rubber containing 20 wt% N330 carbon black filler, PN-10- polydimethyl siloxane ubber containing 10% MNT type nano clay, PDMS-polydimethyl siloxane rubber, PDMS-PU-polydimethyl siloxane-polyurethane blend, SPA-15- polydimethyl siloxane containing 15 wt% polyimide, NRZ-natural rubber containing 10 wt% zeolite, NRC-natural rubber containing 10 wt% cloisite filler, NRCB-containing 10 wt% carbon black

Acknowledgment The first author is grateful to Department of Science and Technology (DST), Government of India for providing DSTinspire fellowship. Authors are also grateful to Department of Biotechnology, Government of India (DBT, sanction no. BT/PR5757/PID/6/709/2012).

Supplementary Materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jtice.2016.03.043. References [1] Ray SK, Sawant SB, Joshi JB, Pangarkar VG. Development of new synthetic membranes for separation of benzene-cyclohexane mixture by pervaporation-A solubility parameter approach. Ind Eng Chem Res 1997;36:5265–75. [2] Qi R, Wang Y, Li J, Zhao C, Zhu S. Pervaporation separation of alkane/thiophene mixtures with PDMS membrane. J Membr Sci 20 06;269:94–10 0. [3] Liu K, Fang CJ, Li ZQ, Young M. Separation of thiophene/n-heptane mixtures using PEBAX/PVDF-composited membranes via pervaporation. J Membr Sci 2014;451:24–31. [4] Yu S, Jiang Z, Ding H, Pan F, Wang B, Yang J, et al. Elevated pervaporation performance of polysiloxane membrane using channels and active sites of metal organic framework CuBTC. J Membr Sci 2015;481:73–81. [5] Wang N, Ji S, Li J, Zhang R, Zhang G. Poly(vinyl alcohol)–graphene oxide nanohybrid “pore-filling” membrane for pervaporation of toluene/n-heptane mixtures. J Membr Sci 2014;455:113–20. [6] Zhang Y, Wang N, Ji S, Zhang R, Zhao C, Li JR. Metal–organic framework/poly(vinyl alcohol) nanohybrid membrane for the pervaporation of toluene/n-heptane mixtures. J Membr Sci 2015;489:144–52. [7] Aouinti L, Roizard D, Belbachir M. PVC–activated carbon based matrices: A promising combination for pervaporation membranes useful for aromatic–alkane separations. Sep Purif Technol 2015;147:51–61. [8] Kononova SV, Kremnev RV, Suvorova EI, Baklagina YG, Volchek BZ, Uchytil P, et al. Pervaporation membranes with poly(γ -benzyl-l-glutamate) selective layers for separation of toluene–n-heptane mixtures. J Membr Sci 2015;477:14–24. [9] Teli SB, Calle M, Li N. Poly(vinyl alcohol)-H-ZSM-5 zeolite mixed matrix membranes for pervaporation separation of methanol–benzene mixture. J Membr Sci 2011;371:171–8. [10] Patil MB, Aminabhavi TM. Pervaporation separation of toluene/alcohol mixtures using silicalite zeolite embedded chitosan mixed matrix membranes. Sep Purif Technol 2008;62:128–36. [11] Tang J, Sirkar KK, Majumdar S. Permeation and sorption of organic solvents and separation of their mixtures through an amorphous perfl uoropolymer membrane in pervaporation. J Membr Sci 2013;447:345–54. [12] Khayet M, Nasefb MM, Mengual JI. Radiation grafted poly(ethylene terephthalate)-graft-polystyrene pervaporation membranes for organic/organic separation. J Membr Sci 2005;263:77–95.

[13] Singha NR, Kuila SB, Das P, Ray SK. Separation of toluene–methanol mixtures by pervaporation using crosslink IPN membranes. Chem Eng Process 2009;48:1560–5. [14] Pal SM, Pangarkar VG. Acrylonitrile-basedcopolymermembranesfortheseparation of methanol from a methanol–toluene mixture through pervaporation. J Appl Polym Sci 2005;96:243–52. [15] Ray S, Ray SK. Separation of organic mixtures by pervaporation using crosslinked rubber membranes. J Membr Sci 2006;270:132–45. [16] Ray S, Ray SK. Separation of organic mixtures by pervaporation using crosslinked and filled rubber membranes. J Membr Sci 2006;285:108– 119. [17] Garg P, Singh RP, Choudhary V. Pervaporation separation of organic azeotrope using poly(dimethyl siloxane)/clay nanocomposite membranes. Sep Purif Technol 2011;80:435–44. [18] Lue SJ, Ou JS, Kuo CH, Chen HY, Yang TH. Pervaporative separation of azeotropic methanol/toluene mixtures in polyurethane–poly(dimethylsiloxane) (PU–PDMS) blend membranes: Correlation with sorption and diffusion behaviors in a binary solution system. J Membr Sci 2010;347:108–15. [19] Garg P, Singh RP, Choudhary V. Selective polydimethylsiloxane/polyimide blended IPN pervaporation membrane for methanol/toluene azeotrope separation. Sep Purif Technol 2011;76:407–18. [20] Singha NR, Ray SK. Removal of pyridine from water by pervaporation using crosslinked and filled natural rubber membranes. J Appl Polym Sci 2012;124:99–107. [21] Boonstra BB. Reinforcement by fillers. In: Blow CM, Hepburn C, editors. Rubber technology and manufacture. 3rd ed. London: Butterworth Scientific; 1982. p. 269–308. [22] Xiao S, Feng X, Huang RYM. Investigation of sorption propertiesand pervaporation behaviors under different operating conditions for trimesoylchloride-crosslinked PVA membranes. J Membr Sci 2007;302:36–44. [23] Ray S, Ray SK. Dehydration of DMF by pervaporation. Ind Eng Chem Res 2006;45:7210–18. [24] Baker RW, Wijmans JG, Huang Y. Permeability, permeance and selec-tivity: A preferred way of reporting pervaporation performance data. J Membr Sci 2010;348:346–52. [25] Hofmann W. Vulcanization and vulkcanizing agents. 1st ed. London: Maclaren and Sons Ltd.; 1967. p. 15. [26] Singha NR, Ray S, Ray SK. Removal of pyridine from water by pervaporation using filled SBR. Membr J Appl Poly Sci 2011;121:1330–4. [27] Singh NR, Das P, Ray SK. Recovery of pyridine from water by pervaporation using filled and crosslinked EPDM membranes. J Ind Eng Chem 2013;19:2034–45. [28] Kuila SB, Ray SK. Sorption and permeation studies of tetrahydrofuran -water mixtures using Full Interpenetrating Network membranes. Sep Purif Technol 2012;89:39–50. [29] Vane LM, Namboodiri VV, Bowen TC. Hydrophobic zeolite–silicone rubber mixed matrix membranes for ethanol–water separation: Effect of zeolite and silicone component selection on pervaporation performance. J Membr Sci 2008;308:230–41. [30] Silverstein RM, Webster FX. Infrared spectrometry. Spectrometric identification of organic compounds. 6th ed. New York: John Wiley & Sons. Inc; 1996. p. 71–143. [31] Furuta I, Kimura S, Iwama M. Physical constants of rubbery polymers. Polymer hand book. Brandrup J, Immergut EH, Grulke EA, editors. 4th ed, New York, U.S.A.: John Wiley & Sons. Inc.; 1999. pp. V/1–V/7. [32] Rogers CE. Permeation of gases and vapors in Polymers. Polymer permeability. Comyn J, editor. New York: Elsevier Applied Sciences; 1985. [33] ten Elshof JE, Abadal CR, Sekuli J, Roy Chowdhury S, Blank DHA. Trans-

P. Das, S.K. Ray / Journal of the Taiwan Institute of Chemical Engineers 64 (2016) 89–105 port mechanisms of water and organic solvents through microporous silica in the pervaporation of binary liquids. Microporous Mesoporous Mater 2003;65:197–208. [34] Iliyas A, Eic´ M, Hassan M, Niaki Z, Vasenkov S. Towards observation of single– file diffusion using TZLC, diffusion fundamentals the open-access. J Basic Princ Diffus Theory Exp Appl 2005;5:1–3. [35] dePinho MN. Mass transfer in radiation-grafted pervaporation membrane. J Membr Sci 1990;54:131–43.

105

[36] Mulder MHV, Smolders CA. On the mechanism of separation of ethanol/water mixtures by pervaporation. I. Calculations of concentration profiles. J Membr Sci 1984;17:289–307. [37] Mandal MK, Sant SB, Bhattacharya PK. Dehydration of aqueous acetonitrile solution by pervaporation using PVA–iron oxide nanocomposite membrane. Colloids Surf A 2011;373:11–21. [38] Foo KY, Hameed BH. Insights into the modeling of adsorption isotherm systems. Chem Eng J 2010;156:2–10.