Dehydration of 1,4-dioxane through blend membranes of poly(vinyl alcohol) and chitosan by pervaporation

Dehydration of 1,4-dioxane through blend membranes of poly(vinyl alcohol) and chitosan by pervaporation

Journal of Membrane Science 280 (2006) 138–147 Dehydration of 1,4-dioxane through blend membranes of poly(vinyl alcohol) and chitosan by pervaporatio...

381KB Sizes 0 Downloads 182 Views

Journal of Membrane Science 280 (2006) 138–147

Dehydration of 1,4-dioxane through blend membranes of poly(vinyl alcohol) and chitosan by pervaporation夽 D. Anjali Devi a , B. Smitha b , S. Sridhar b , T.M. Aminabhavi a,∗ a

b

Membrane Separations Division, Center of Excellence in Polymer Science, Karnatak University, Dharwad 580003, India Membrane Separations Group, Chemical Engineering Division, Indian Institute of Chemical Technology, Hyderabad 500007, India Received 26 April 2005; received in revised form 4 January 2006; accepted 6 January 2006 Available online 21 February 2006

Abstract Blend membranes prepared from poly(vinyl alcohol) (PVA) and chitosan (CS) were crosslinked with glutaraldehyde and used in the pervaporation dehydration of 1,4-dioxane. Membranes were characterized by Fourier transform infrared spectroscopy (FTIR), thermogravimetric analysis (TGA), differential scanning calorimetry (DSC) and X-ray diffraction (X-RD) to assess, respectively, the intermolecular interactions, thermal stability and crystallinity. Equilibrium sorption studies were carried out in pure liquids and binary mixtures of different compositions of water + 1,4-dioxane mixtures to assess the polymer–liquid interactions. The crosslinked membrane showed a good potential in breaking the azeotrope of 82 wt.% aqueous 1,4-dioxane giving a selectivity of 117 with a reasonable water flux of 0.37 kg/m2 h. The effect of operating parameters such as feed composition, membrane thickness and permeate pressure was evaluated. © 2006 Elsevier B.V. All rights reserved. Keywords: Pervaporation; PVA/CS blend; Dioxane–water azeotrope; Selectivity; Flux

1. Introduction Pervaporation (PV) is an efficient membrane-based process used for the economical separation of liquid mixtures, especially aqueous–organic azeotropes, closely boiling mixtures, isomers, besides heat sensitive and hazardous compounds [1,2]. PV separation is based on the differences in sorption and diffusion properties [3,4] of the permeating molecules. The effectiveness of dehydration of an organic component from its aqueous mixture is generally governed by the chemical nature and structure of the PV membrane. The membrane should possess an excellent mechanical stability in aqueous environment besides having sufficient affinity to water molecules so that water can be preferentially sorbed and transported at the rapid rate affecting an efficient separation. Due to excellent chemical stability, film-forming nature and high hydrophilicity, PVA has been the most widely used PV membrane in PV dehydration studies [5]. However, the deacetylated chi夽

This article is CEPS communication #63. Corresponding author. Tel.: +91 836 2215372/2779983; fax: +91 836 2771275. E-mail address: [email protected] (T.M. Aminabhavi). ∗

0376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2006.01.006

tosan, a hydrophilic biopolymer, can be used to develop blends with PVA, since it can form highly selective and permeable blends with other hydrophilic polymers like PVA. PVA is a crystalline polymer, which gives a low flux for water. In this work, blending of PVA with CS was carried out to improve the flux by reducing crystallinity without significantly altering the selectivity. Additionally, ionic interaction between PVA and CS could possibly improve the separation by reducing the organic sorption due to ‘salting out effect’ as described by Huang [2]. The PVA–CS blend has been successfully utilized for the PV separation of ethanol–water mixtures [6,7], controlled drug release [8,9] and other biomedical applications [10]. Chitosan is expected to make exclusively strong hydrogen bond interactions with water, owing to a close proximity of its solubility parameter value (43.04 J1/2 /cm3/2 ) to that of water (47.9 J1/2 /cm3/2 ) [11,12]. Among the various agents employed to crosslink hydrophilic polymers, glutaraldehyde (GA) has been most commonly used [13]. Hence, the blends of PVA and CS were crosslinked by GA. Even though these membranes were proven materials for separating water from alcohols [2], their performance has not been satisfactory for the dehydration of 1,4-dioxane since it has the tendency to form H-bonds with the membrane.

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

1,4-Dioxane is a well-known organic compound, primarily used in chemical industries as a solvent and stabilizer for chlorinated solvents. It is widely used in pharmaceutical industries and is miscible with water in all proportions. It forms an azeotrope at 18-wt.% of water. There are few reports in the earlier literature on the PV dehydration of 1,4-dioxane. Dehydration of 1,4-dioxane was attempted earlier by Lee et al. [14] using poly(dimethylsiloxane-co-siloxane) membrane. Kimura and Nomura [15] investigated the extraction of 10-wt.% of 1,4-dioxane–water mixture using poly(dimethylsiloxane) membrane and obtained a flux of 0.008 kg/m2 h/10 ␮m with a selectivity of 44. Sridhar et al. [16] used the crosslinked PVA membrane and obtained a flux of 0.16 kg/m2 h/10 ␮m. Recently, Kurkuri and Aminabhavi [17] investigated the PV separation of 1,4-dioxane–water mixture at 30 and 50 ◦ C for 10 wt.% of water using the blend membrane of acrylamide-g-sodium alginate. In another study, Aminabhavi et al. [18] obtained a flux of 0.424 kg/m2 h/10 ␮m with a selectivity of 111 for sodium alginate membrane. Naidu et al. [19] obtained a flux of 0.106 kg/m2 h for the blend membrane of sodium alginate with hydroxyethylcellulose. Wang et al. [20] reported the flux values of 0.318 and 0.417 kg/m2 h with selectivities of 491 and 663, respectively, for the alginate composite membrane in separating 1,4-dioxane–water mixture. In continuation of our ongoing efforts to study the PV separation of aqueous–organic mixtures [13,21–23], we have used for the first time, crosslinked PVA and CS blend membranes for the PV separation of 1,4dioxane + water mixtures. These membranes have offered superior performance over those of the previously studied membranes. 2. Experimental 2.1. Materials PVA with a molecular weight (MW) of 50,000 g/mol was purchased from Loba Chemie, Mumbai. The degree of polymerization of PVA was 1570 ± 50 and saponification degree was 99%. Chitosan was purchased from Aldrich Chemical Co. (Milwaukee, WI, USA), having an average MW of 500,000 with a degree of deacetylation of 84%. 1,4-Dioxane, isopropanol, HCl and glutaraldehyde were all purchased from s.d. Fine Chemicals, Mumbai, India. Glacial acetic acid was purchased from Loba Chemie, Mumbai, India. Deionized water (conductivity = 0.02 ␮S/cm), was generated in the laboratory itself. 2.2. Preparation of membranes A 10 wt.% of PVA solution was prepared in deionized water at 90 ◦ C. A 3 wt.% solution of chitosan in 2% (v/v) aqueous acetic acid solution was prepared and the solution was filtered to remove any undissolved suspended matter. A bubble-free solution was used for casting the films. A homogeneous and nonporous membrane was prepared by casting the polymer solution followed by drying at room temperature and crosslinking with GA up to 120 min. The blend of PVA and CS taken in a composi-

139

tion of 10/10 in wt.% was cast on a perfectly leveled glass plate. The dried membrane was crosslinked using isopropanol–water (90/10, vol%) mixture containing 1 vol% of HCl as a catalyst and 5 vol% of GA as a crosslinker. After crosslinking up to 120 min, membrane was washed with distilled water and dried in an oven to remove the traces of residual solvents used in preparing the membrane. 2.3. Membrane characterization 2.3.1. Fourier transform infrared (FTIR) spectroscopic studies FTIR spectra of the plain chitosan, uncrosslinked PVA/CS and crosslinked PVA/CS blends were scanned in the range between 4000 and 400 cm−1 on a Nicolet-740, Perkin-Elmer283B FTIR spectrophotometer in KBr pellets. 2.3.2. X-ray diffraction (X-RD) analysis Siemens D 5000 powder X-ray diffractometer was used to study the solid-state morphology in powder form. X-rays of ˚ wavelengths were generated by a Cu K␣ source. The 1.5406 A angle of diffraction was varied from 2◦ to 65◦ to identify any changes in the crystal structure and intermolecular distances between intersegmental chains after crosslinking. 2.3.3. Differential scanning calorimetry (DSC) studies DSC thermograms of CS, crosslinked and uncrosslinked PVA/CS blend membranes were obtained on a Perkin-Elmer DSC Model-7. Measurements were performed over the temperature range of 25–400 ◦ C at the heating rate of 10 ◦ C/min in hermetically sealed aluminum pans. The sample pan was conditioned in the instrument before running the experiment. 2.3.4. Thermogravimetric (TGA) analysis Thermal stability of the polymer films was examined using a Seiko 220TG/DTA analyzer from 25 to 600 ◦ C, heated at the rate of 10 ◦ C/min and flushed with nitrogen gas at the rate of 20 mL/min. Samples were subjected to TGA before and after blending to determine thermal stability and decomposition characteristics. 2.3.5. Mechanical properties Mechanical properties of the membranes were evaluated by the Universal Testing Machine (UTM) (Shimadzu, Model AGS10 kNG, Japan) with an operating head load of 5 kN, following the procedure outlined in ASTM D-638 test method. The test length was 5 cm, while the speed of testing was set at the rate of 10 mm/min. Tensile strength was calculated using the equation: Tensile strength =

maximum load cross-sectional area (N/mm2 )

(1)

2.3.6. Swelling characteristics Interaction of the polymer blend membrane with the pure liquid components of the feed mixture was determined by the gravimetric sorption experiments. Weighed samples of circular pieces of polymer films (3-cm diameter) were soaked in pure

140

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

water, 1,4-dioxane as well as their binary mixtures. Membranes were taken out after different soaking times and quickly weighed after carefully wiping the excess adhered liquid drops to estimate the amount of liquid sorbed at a particular time, t. Films were then quickly placed back into the solvent container and the procedure was repeated until the membranes attained the steady state equilibrium as indicated by the observation of constant weight after completion of soaking time. Degree of swelling was calculated as: Degree of swelling =

Ms Md

(2)

where Ms is mass of the swollen polymer in (g) and Md is mass of the dry polymer in (g). The percent sorption was calculated using the equation:   Ms − M d % Sorption = × 100 (3) Md 2.3.7. Determination of ion exchange capacity (IEC) In order to determine the effect of GA on PVA in the blend, ion exchange capacity (IEC) of the blend was estimated. IEC indicates the number of groups present before and after crosslinking, which gives an idea about the extent of crosslinking. Thus, IEC gives the number of milli-equivalents of ions in 1 g of dry polymer. To determine IEC, specimens of identical weights were soaked in 50 mL of 0.01N NaOH solution for about 12 h at the ambient temperature. Then, 10 mL of the sample was titrated against 0.01N H2 SO4 . The sample was regenerated with 1 M HCl, washed with water and dried to constant weight. IEC was then calculated as: IEC =

B − P × 0.01 × 5 m

(4)

where B is sulfuric acid used to neutralize the blank sample (blank titration), P is sulfuric acid used to neutralize the PV membrane (sample titration), 0.01 is normality (N) of sulfuric acid, 5 represents the factor corresponding to ratio of the amount of NaOH taken to dissolve the polymer to the amount used for titration and m represents the sample mass (g). 2.4. Pervaporation experiments PV experiments were carried out in a 100 mL batch mode with an indigenously constructed pervaporation manifold operated at a vacuum level of as low as 0.0666612 × 102 Pa (0.05 mmHg) in the permeate line. Membrane area in the PV cell assembly was 19.8 cm2 . Experimental procedure remained the same as described previously [22,24]. Permeate was collected for 8–10 h. PV experiments were carried out at 30 ± 2 ◦ C and triplicate data were collected using a fresh feed stock solution to check for reproducibility usually within 3% of standard error. The collected permeate sample was weighed after attaining room temperature on a Sartorius electronic balance (accuracy, 10−4 g) to determine the flux and then analyzed by gas chromatography to evaluate the membrane selectivity.

2.4.1. Flux and selectivity In PV, flux, J of a given species, say faster permeating component, i of a binary mixture comprising of i (water) and j (1,4-dioxane) is given by: Ji =

Wi At

(5)

where Wi represents the weight of water in permeate (kg), A the membrane area (m2 ) and t represents the evaluation time (h). In the present study, even though membranes of different thicknesses were fabricated, flux was normalized and reported for a constant thickness of 10 ␮m. This helped in comparing our data with the literature results under identical membrane thickness. Membrane selectivity, α is the ratio of permeability coefficient of water (faster moving component) and 1,4-dioxane (slower moving component), which was calculated from the respective concentrations in feed and permeate sides using the equation: α=

y(1 − x) x(1 − y)

(6)

Here, y is permeate wt.% of water and x is its feed wt.%. 2.5. Analytical procedure Feed and permeate samples were analyzed using a Nucon Gas Chromatograph (GC, Model 5765) installed with a thermal conductivity detector (TCD) and a Tenax packed column of 2 m length. The oven temperature was initially kept at 70 ◦ C and then raised to 210 ◦ C at the heating rate of 25 ◦ C/min. The injector and detector temperatures were maintained at 150 ◦ C each. The sample injection size was 1 ␮L and pure hydrogen was used as a carrier gas at the pressure of 1 kg/cm2 . The GC response was calibrated for this particular column and conditions with known compositions of 1,4-dioxane/water mixtures. Calibration factors were fed into the GC software to obtain the correct analysis for unknown samples. 3. Results and discussion The homopolymer solutions of PVA, CS and their blend solutions were optically clear. No phase separation or precipitation was observed even after keeping the mixture for a longer time at ambient temperature. Scheme 1 represents the structure of homopolymers and the crosslinking reaction. The ionic blending of two polymers is shown in Scheme 1(a). Upon blending, OH group of PVA and NH2 group of chitosan have reacted resulting in the formation of NH3 + · · ·O− bond. Scheme 1(b) represents the structure of crosslinked PVA/CS blend with GA, which suggested the crosslinking of GA with NH2 groups of chitosan. When GA is used as a crosslinking agent, its aldehydic groups have reacted with the amine groups of chitosan as confirmed by FTIR (see Fig. 1). The amount of residual amine and hydroxyl groups of the blend membranes after crosslinking was estimated from the IEC studies (see Appendix A for sample calculations). It was found that the uncrosslinked PVA/CS blend had an IEC of 31.6 mmol/g (31.6 meq/g), whereas the crosslinked polymer exhibited an

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

141

Scheme 1. (a) Ionic blend of PVA and chitosan. (b) PVA/CS blend crosslinked with glutaraldehyde. (c) Crosslinking reaction of poly(vinyl alcohol) with glutaraldehyde showing the formation of homolink.

IEC of 11.8 mmol/g (11.8 meq/g). IEC is equivalent to the total number of free amino groups (considering the fact that amino groups are more reactive than hydroxyl groups). The R NH2 group present in the membrane decreased after crosslinking because some amino groups were consumed due to the reaction of crosslinking agent [25]. Thus, almost 60% of amine groups of the uncrosslinked PVA/CS blend membrane were reacted to give crosslinks in the presence of GA. It is thus possible that there could be still few amine and hydroxyl groups left free for the formation of homolinks as shown in Scheme 1(c). 3.1. Membrane characterization 3.1.1. Fourier transform infrared spectroscopy FTIR spectra of PVA, crosslinked PVA/CS blend, uncrosslinked PVA/CS blend and CS membranes are shown in Fig. 1(a–d). FTIR spectra of plain CS (Fig. 1(d)), shows a broad peak at wavenumbers 1570–1655 cm−1 , which indicates the presence of amides I and II. The wavenumber at 2833 cm−1

corresponds to CH2 stretching, whereas that of free hydroxyl group is observed at 3450 cm−1 . The weakening of the peak at 1150 cm−1 , which is visible in the uncrosslinked blend membrane (Fig. 1(b)) indicates the interaction of GA with the blend membrane. It is therefore proposed that the formation of acetal linkage (C O C) might have occurred, thereby resulting in the weakening of the peak as observed in the crosslinked blend membrane, (Fig. 1(c)). Upon blending CS with PVA, the OH groups of PVA might have interacted with the NH2 group of chitosan or vice versa, which upon crosslinking a reduction in the intensity of OH group was observed as seen in Fig. 1(c). However, a reduction in the size of this peak in the crosslinked blend indicates the intermolecular interactions between PVA and CS chains, which might have disturbed the crystallizability of chitosan in the blend matrix. 3.1.2. X-ray diffraction X-ray diffractograms of (a) PVA, (b) PVA/CS blend and (c) CS are shown in Fig. 2. It is seen that X-RD patterns of all

142

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

the membranes indicate semicrystalline behavior. The diffractograms showed broad peaks around 10◦ of 2θ, indicating an average intermolecular distance of the amorphous part and relatively sharp semicrystalline peaks around 20◦ of 2θ. Two distinct bands giving maxima at 2θ of 7–9◦ and 2θ of 20◦ are related to two types of crystals: crystal I and crystal II. Of these, crystal I, which corresponds to the peak at 9◦ is responsible for the separation, since it comprises the functional groups like NH2 and OH. These groups might have further undergone significant changes after crosslinking. A reduction in effective d-spacing ˚ for uncrosslinked blend (Fig. 2(b)) to 9.31 A ˚ in the from 9.32 A crosslinked blend membrane (Fig. 2(d)) indicates a shrinkage in the cell size or intersegmental spacing, which would improve selective permeation of the membrane.

Fig. 1. FTIR spectra of PVA (a), PVA/CS (b), XPVA/CS (c) and CS (d).

Fig. 2. X-RD diffractograms of PVA (a), PVA/CS (b), CS (c) and XPVA/CS (d).

3.1.3. Swelling characteristics Equilibrium sorption data of the crosslinked PVA/CS membrane in 1,4-dioxane/water mixtures of different compositions are presented in Table 1. Degree of swelling at zero water concentration was low, which steadily increased to 52.40% at the azeotropic composition containing approximately 18 wt.% of water, but swelling has increased with increasing water concentration. Despite crosslinking, blend membranes have interacted with water molecules, showing higher selectivity to water compared to 1,4-dioxane. However, absorption of large amount of water by the PV membrane lead to swelling of the membrane with a subsequent decrease in membrane selectivity due to plasticizing effect of the polymer segments. 3.1.4. Mechanical properties PVA, CS and crosslinked PVA/CS blend membranes exhibited tensile strengths of 51.31, 1.38 and 60.66 N/mm2 , respectively, calculated from the stress–strain curves, which showed an enhancement in mechanical strength due to blending (see data in Table 2). Notice that tensile strength of 1.38 N/mm2 for CS after when blended with PVA of tensile strength of 51.31 N/mm2 went up to 60.66 N/mm2 in the blend. The percent elongation at break (12.0) for crosslinked PVA/CS blend membranes was also higher than those of individual component polymers, viz., PVA (7.5) and CS (2.7). Even though X-RD spectra confirmed a reduction in crystalline components of PVA and CS after blending, extensive intra- and intermolecular hydrogen bonding interactions involving OH group of PVA as well as NH2 and OH groups of CS, besides the formation of an interpenetrating polymer network upon crosslinking, are responsible for such an increase in mechanical strengths of the blend membranes.

Table 1 Pervaporation data of 1,4-dioxane and water mixtures at 30 ◦ C (permeate pressure = 1.333224 × 102 Pa (1 mmHg) and membrane thickness = 50 ␮m) Feed composition

Sorption in feed (wt.%)

Water (x)

1,4-Dioxane (1 − x)

2.85 9.35 17.62 19.56 24.66

97.15 90.65 82.38 80.40 75.34

24.70 31.00 52.40 82.14 93.84

Permeate composition (wt.%) Water (y)

1,4-Dioxane (1 − y)

91.14 97.04 96.13 95.01 90.42

8.58 2.96 3.87 4.99 9.58

Selectivity α

Flux (kg/m2 h)

363 318 116 78.3 28.5

0.075 0.359 0.366 0.450 0.562

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

143

Table 2 Tensile strength and percent elongation of PVA and PVA–CS blend membranes Tensile strength (N/mm2 ) PVA CS XPVA/CS blend

51.31 1.38 60.66

Elongation at break (%) PVA CS XPVA/CS blend

7.5 2.7 12.0

3.1.5. Thermal properties Even though PV experiments were carried out on a benchscale module at ambient temperature, the commercial operation is generally done at higher temperatures, in which a preheater is used to maintain the feed temperature, which may otherwise fall as the enthalpy of vaporization is supplied by the feed itself. Moreover, PV can be integrated with distillation to overcome the azeotropic barrier and the feed to PV could be the distillate, which is available at high temperature. Hence, it is essential to study thermal stability of the membrane at higher temperatures. 3.1.5.1. Differential scanning calorimetry (DSC). Thermal properties of the blend membranes were examined by DSC. It was of particular interest to estimate how the thermal transition in PVA varied with blending, since CS did not show any significant transition in the temperature range of the DSC scans. DSC thermograms are shown in Fig. 3. The homogeneous nature of CS and PVA represented by curves (a) and (b) gave relatively large and sharp melting endotherms with a melting temperature peak (Tm ) observed around 320 and 201–325 ◦ C. However, upon blending, a decrease in endothermic peak was observed. The tendency that Tm of the blend showed 230 ◦ C smaller than that of chitosan, which upon crosslinking, reduced to 201–221 ◦ C; this implies the comparable, but reduced relative crystallinity due to crosslinking. 3.1.5.2. Thermogravimetric analysis (TGA). Thermal degradation of PVA, CS and crosslinked blend of PVA/CS were exam-

Fig. 4. TGA curves of (a) PVA, (b) PVA/CS blends and (c) CS.

ined by TGA and thermograms are displayed in Fig. 4. It was observed that PVA (curve a) exhibited two weight loss stages in the temperature range between 250 and 360 ◦ C followed by the final decomposition of the polymer that began around 460 ◦ C. Weight loss in the first stage could be attributed to splitting of the main polymer chain before its final decomposition. In case of pure CS (curve c), two weight loss stages were observed around 240–280 and 530–560 ◦ C followed by the final decomposition. Thermal degradation of CS occurred at 275 ◦ C followed by the second degradation at 530 ◦ C. The PVA/CS crosslinked blend (curve b) exhibited two weight loss stages in the range of 270–330 and 520–570 ◦ C followed by a final decomposition. The causes for these weight losses are similar to the ones discussed before. An increase in thermal stability of the blend was observed. 3.2. Pervaporation results

Fig. 3. DSC thermograms of (a) CS, (b) PVA, (c) unmodified PVA/CS blend and (d) crosslinked PVA/CS.

3.2.1. Effect of feed composition The relationship between liquid feed composition and PV data was investigated over a wide range of feed mixtures at 30 ◦ C. For this study, membrane thickness and permeate pressure were kept constant at 50 ␮m and 1.333224 × 102 Pa (1 mmHg), respectively. Flux and selectivity data as a function of feed composition are displayed in Fig. 5. PV performance of the crosslinked PVA/CS blend membrane was investigated by varying feed compositions from 2 to 28.53 wt.% of water. The permeation rate of 1,4-dioxane is much smaller compared to water, thus showing that the membrane is permselective to water. Expectedly, a rise in feed composition of water produced an increase in water flux from 0.002 to 0.65 kg/m2 h, a much higher value observed compared to earlier data of the literature for the feed mixtures of this study. Mass transport through hydrophilic blend membranes could occur by the solution-diffusion mech-

144

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

Fig. 5. Effect of feed water composition on PV performance of PVA/CS blend membranes. Fig. 6. Comparison of vapor liquid equilibrium curve with PV data for water and 1,4-dioxane mixtures.

anism [26]. In addition to sorption data of the binary feed mixtures presented in Table 1, we find that the crosslinked membrane showed a high degree of sorption in pure water (121 wt.%) with a relatively negligible sorption (0 wt.%) for pure 1,4-dioxane. Preferential affinity of the membrane towards water might have caused membrane swelling, which might have allowed a rapid permeation of feed molecules through the membrane. The degree of swelling correspondingly increased at the higher feed compositions of water resulting in an enhanced flux. Increased swelling has a negative impact on membrane selectivity, since the swollen and plasticized upstream membrane layer might have allowed some 1,4-dioxane molecules to escape into the permeate side along with water molecules. Therefore, permeate water composition was reduced from 97.04 to 90.42 wt.%, indicating a drop in selectivity from 363 to as low as 28.5, respectively. It is worth mentioning that the membranes of this study are quite promising in dehydrating 1,4-dioxane containing 2–30 wt.% of water. Particularly, the azeotropic composition, viz., 82 wt.% of 1,4-dioxane was broken by the PV membrane, which acted as a third phase. See for example, Fig. 6, wherein PV curve is higher than that of the vapor–liquid equilibrium (VLE) curve throughout the composition range of the feed mixture. Results of the present work are compared with the available literature data in Table 3. We find that improved flux and selectivity results are obtained as compared to other data published in the literature. We have also investigated the effect of permeate pressure and membrane thickness to see whether PV separation properties of the membranes hold good under different operating conditions. Even though flux may be predictable, literature has shown that selectivities would fall at a membrane thickness below the critical value or higher permeate pressures with the latter condition favoring transport of more volatile component of the binary mixture. In view of the close boiling points of water and 1,4-dioxane, these studies are considered worthwhile.

3.2.2. Effect of permeate pressure Permeate pressure was varied from 0.666612 × 102 to 13.33224 × 102 Pa (0.5 to 10 mmHg) in case of crosslinked PVA/CS blend membrane at a constant thickness of 50 ␮m and at the azeotropic feed composition. At lower pressures (high vacuum), the influence of driving force on diffusing molecules inside the membrane was high. This could result in the components to be swept away immediately from the permeate side, thereby resulting in high mass transfer rates. From Fig. 7, we observe that the membrane exhibited considerable lowering of the flux from 0.3201 to 0.014 kg/m2 h as well as a reduction in selectivity from 211 to 13, with increasing permeate pressure from 0.666612 × 102 to 13.33224 × 102 Pa (0.5 to 10 mmHg). Under high vacuum conditions (lower pressures), diffusion through the hydrophilic membrane of this study becomes the rate determining step, and therefore, diffusing water molecules would experience a large driving force, thereby enhancing the desorption rate at the downstream side. On the other hand, lower

Fig. 7. Effect of downstream pressure on membrane thickness.

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

145

Table 3 Comparison of PV performance of the present membranes with literature data Membrane

Temperature (◦ C)

Water feed data (wt.%)

Membrane thickness (␮m)

Flux (kg/m2 h)

Normalized flux (kg/m2 h/10 ␮m)

Selectivity α

Reference author (no.)

PVA/CS crosslinked with GA

30

18

50

0.366

1.830

117

Poly(dimethylsiloxane-co-siloxane) Poly(dimethylsiloxane) TDI crosslinked PVA

– – 30

10 80 18

– – –

– – –

0.108 0.008 0.160

16.3 44 83.8

Anjali Devi et (present work) Lee et al. [14] Kimura et al. [15] Sridhar et al. [16]

NaAlg Copoly-1 Copoly-2

30

10

45

0.058 0.060 0.067

0.261 0.270 0.302

351 82.8 76.7

NaAlg Copoly-1 Copoly-2

50

10

45

0.069 0.071 0.073

0.311 0.319 0.385

82.84 45.88 42.43

NaAlg

30 40 50

10

40

0.106 0.110 0.117

0.424 0.440 0.468

111 72.81 47.25

NaAlg NaAlg-PVA-5 NaAlg-PVA-10 NaAlg-PVA-20

30

10

40

0.180 0.210 0.230 0.273

0.720 0.840 0.920 1.092

120 70 50 40

NaAlg NaAlg + HEC-5 NaAlg + HEC-10 NaAlg + HEC-20

30

10

37.5

0.111 0.106 0.135 0.121

0.413 0.398 0.506 0.453

268 160 127 87

Naidu et al. [19]

SA composite membrane

60 70

10 10



0.318 0.417



491 663

Wang et al. [20]

al.

Kurkuri et al. [17]

Aminabhavi et al. [18]

PVA, poly(vinyl alcohol); CS, chitosan; GA: glutaraldehyde; TDI: toluene diioscyanate; NaAlg, sodium alginate; PEG, poly(ethylene glycol); UFS, ureaformaldehyde-sulfuric acid; HEC, hydroxyethylcellulose; SA, sodium alginate.

vacuum levels reduce the driving force, thus slowing down the desorption of molecules from the membrane matrix. In such cases, vapor pressures of two components of the mixture govern the selectivity of the membrane. 1,4-Dioxane, which has a high vapor pressure, permeates competitively with water molecules, thus lowering the concentration of water in the permeate.

3.2.3. Effect of membrane thickness Effect of varying membrane thickness on PV performance was investigated at a constant feed composition (azeotropic) and permeate pressure (1.333224 × 102 Pa (1 mmHg)) by fabricating membranes of thicknesses ranging from 35 to 115 ␮m. By increasing the membrane thickness, a reduction in flux from 0.732 to 0.0125 kg/m2 h was observed (see Fig. 8). Even though the availability of polar groups of the membrane enhanced with increasing membrane thickness, flux decreased because diffusion was increasingly retarded as the feed molecules move a longer tortuous path to reach the permeate side. However, permeate concentration of water increased from 90.42 to 97.07 wt.%, which means that selectivity increased from 6.41 to 111.4 over the thickness range of 35–115 ␮m. In a PV process, upstream layer of the membrane was swollen and plasticized due to absorption of the feed liquids, but allowed the unrestricted transport of feed components. In contrast, the downstream layer was virtually dry due to the continuous evacuation in permeate side,

which might have formed the restrictive barrier layer thereby, allowing only the interacting and small sized water molecules to transport through the membrane. It is expected that thickness of the dry layer of the membrane could increase with an increase in the overall membrane thickness, which would result in an improved selectivity as observed in the present case.

Fig. 8. Effect of membrane thickness on PV performance.

146

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147

4. Conclusions

References

Crosslinked PVA/CS blends have been prepared. The number of groups crosslinked in PVA/CS blend polymer was identified from IEC studies. Characterization of the crosslinked membranes by FTIR and X-RD confirmed the crosslinking reaction. FTIR spectroscopy revealed the predicted interaction between PVA and the cationic chitosan polymer. Membranes of this study showed adequate thermal and mechanical stability to withstand the PV experimental conditions, while dehydrating 1,4dioxane from its aqueous mixture. The developed membranes could break the azeotrope containing 18 wt.% of water. With increasing feed water compositions, the membrane performance exhibited a reduction in selectivity with an improvement in flux due to increased swelling. However, with increasing membrane thickness, selectivity has improved, but flux decreased. Higher permeate pressure caused a reduction in both flux and selectivity. Pervaporation through crosslinked PVA/CS blend membrane for the feed mixture containing 9.35 wt.% of water could dehydrate up to 97.04 wt.%.

[1] H.L. Fleming, C.S. Slater, Pervaporation, in: W.S.W. Ho, K.K. Sirkar (Eds.), Membrane Handbook, Van Nostrand Reinhold, New York, 1992, p. 105. [2] R.Y.M. Huang (Ed.), Pervaporation Membrane Separation Processes, Elsevier, Amsterdam, 1991. [3] M.H.V. Mulder, C.A. Smolders, On the mechanism of separation of ethanol/water mixtures by pervaporation. Part I. Calculations of concentration profiles, J. Membr. Sci. 17 (1984) 289–307. [4] C.H. Lee, Theory of reverse osmosis and some other membrane permeation operations, J. Appl. Polym. Sci. 19 (1975) 83–95. [5] E. Immelman, R.D. Sanderson, E.P. Jacobs, A.J. van Reenen, Poly(vinyl alcohol) gel sublayers for reverse osmosis membranes: insolubilization by acid-catalyzed dehydration, J. Appl. Polym. Sci. 50 (1993) 1013–1034. [6] V. Dubey, L.K. Pandey, C. Saxena, Pervaporative separation of ethanol/water azeotrope using a novel chitosan impregnated bacterial cellulose membrane and chitosan–poly(vinyl alcohol) blends, J. Membr. Sci. 251 (2005) 131–136. [7] Y.M. Lee, S.Y. Nam, J.H. Kim, Pervaporation of water, ethanol through poly(vinyl alcohol)/chitosan blend membrane, Polym. Bull. 29 (1992) 423–429. [8] J.H. Kim, J.Y. Kim, Y.M. Lee, K.Y. Kim, C.S. Cho, Y.K. Sung, Controlled-release drug delivery through crosslinked poly(vinyl alcohol)/chitosan blend membrane, Polymer (Korea) 15 (6) (1991) 695– 701. [9] G.L. Cynthia Khoo, S. Frantzich, A. Rosinski, M. Sjostrom, J. Hoogstaate, Oral gingival delivery systems from chitosan blends with hydrophilic polymers, Eur. J. Pharma. Biopharm. 55 (2003) 47– 56. [10] M.N.V. Ravikumar, A review of chitin and chitosan applications, React. Funct. Polym. 46 (2000) 1–27. [11] R. Ravindra, K.R. Krovvidi, A.A. Khan, Solubility parameter of chitin and chitosan, Carbohydr. Polym. 36 (1998) 121–127. [12] A.F.M. Barton (Ed.), CRC Handbook of Solubility Parameters and other Cohesive Parameters, CRC Press, Boca Raton, Florida, USA, 1983. [13] M.D. Kurkuri, U.S. Toti, T.M. Aminabhavi, Synthesis and characterization of blend membranes of sodium alginate and poly(vinyl alcohol) for the pervaporation separation of water–isopropanol mixture, J. Appl. Polym. Sci. 86 (2000) 3642–3651. [14] Y.T. Lee, K. Iwamoto, H. Sekimoto, M. Seno, Pervaporation of water–dioxane mixtures with poly(dimethylsiloxane-co-siloxane) membranes prepared by sol–gel process, J. Membr. Sci. 42 (1989) 169– 182. [15] S. Kimura, T. Nomura, Pervaporation of organic substance-water by silicone rubber membrane, Maku (Japan) 8 (1983) 177–183. [16] S. Sridhar, B. Smitha, U.S. Madhavi Latha, M. Ramakrishna, Pervaporation of 1,4-dioxane–water mixtures using poly(vinyl alcohol) membranes crosslinked with toluene–2,4-diisocyante, J. Polym. Mater. 21 (2004) 181–188. [17] M.D. Kurkuri, T.M. Aminabhavi, Pervaporation separation of water/dioxane mixtures with sodium alginate-g-poly(acrylamide) copolymeric membranes, J. Appl. Polym. Sci. 89 (2003) 300–305. [18] T.M. Aminabhavi, B.V.K. Naidu, S. Sridhar, Computer simulation and comparative study on the pervaporation separation characteristics of sodium alginate and its blend membranes with poly(vinyl alcohol) to separate aqueous mixtures of 1,4-dioxane or tetrahydrofuran, J. Appl. Polym. Sci. 94 (2004) 1827–1840. [19] B.V.K. Naidu, K.S.V. Krishna Rao, T.M. Aminabhavi, Pervaporation separation of water + 1,4-dioxane and water + tetrahydrofuran mixtures using sodium alginate and its blend membranes with hydroxyethylcellulose—a comparative study, J. Membr. Sci. 260 (2004) 131–141. [20] X.P. Wang, N. Li, W.Z. Wang, Pervaporation properties of novel alginate composite membranes for dehydration of organic solvents, J. Membr. Sci. 193 (2001) 85–95.

Acknowledgements Professor T.M. Aminabhavi and Miss D. Anjali Devi thank the University Grants Commission, New Delhi (Grant No. F1-41/2001/CPP-II) for financial support to establish Center of Excellence in Polymer Science (CEPS). We thank Dr. M. Ramakrishna, Head of Chemical Engineering Division, IICT for his encouragement. This research is a collaborative effort between CEPS, Dharwad and IICT, Hyderabad under the MoU. Appendix A Calculations of IEC for uncrosslinked PVA/CS and crosslinked PVA/CS membranes Initial conditions: Mass of crosslinked (m) PVA/CS membrane = 0.08 g Mass of plain PVA/CS membrane (m) = 0.03 g Normality of NaOH = 0.01N Normality of H2 SO4 = 0.01N After 12 h: Blank titration (B) = 9.5 mL Sample titration for unmodified PVA/CS membrane (P) = 7.6 mL Sample titration for crosslinked PVA/CS membrane (P) = 7.6 mL Volume of NaOH neutralized was = 10 mL Ion exchange capacity, IEC of the membrane was calculated as: IEC =

blank titration − sample titration × 0.01 × 5 m

D. Anjali Devi et al. / Journal of Membrane Science 280 (2006) 138–147 [21] M.D. Kurkuri, T.M. Aminabhavi, Polyacrylonitrile-g-poly(vinyl alcohol) membranes for pervaporation separation of dimethylformamide and water mixtures, J. Appl. Polym. Sci. 91 (2004) 4091–4097. [22] D. Anjali Devi, B. Smitha, S. Sridhar, T.M. Aminabhavi, Pervaporation separation of isopropanol/water mixture through crosslinked chitosan membranes, J. Membr. Sci. 262 (2005) 91–99. [23] T.M. Aminabhavi, H.G. Naik, Synthesis of graft copolymeric membranes of poly (vinyl alcohol) and polyacrylamide for pervaporation separation of water–acetic acid mixtures, J. Appl. Polym. Sci. 83 (2002) 244–258.

147

[24] S. Sridhar, R. Ravindra, A.A. Khan, Recovery of monomethyl hydrazine liquid propellant by pervaporation technique, Ind. Eng. Chem. Res. 39 (2000) 2485–2490. [25] K. Igrashi, Y. Nakano, Fundamental adsorption properties of chitosan gel particles prepared by suspension evaporation method, J. Appl. Polym. Sci. 86 (2000) 901–906. [26] J.G. Wijmans, R.W. Baker, The solution diffusion model: a review, J. Membr. Sci. 107 (1995) 1–21.