Polyvinyl alcohol as the barrier layer in thin film

Journal of Colloid and Interface Science 338 (2009) 121–127

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Polyvinyl alcohol as the barrier layer in thin film composite nanofiltration membranes: Preparation, characterization, and performance evaluation J.M. Gohil, P. Ray * Reverse Osmosis Discipline, Central Salt and Marine Chemicals Research Institute (CSIR), GB Marg, Bhavnagar, Gujarat, India

a r t i c l e

i n f o

Article history: Received 10 February 2009 Accepted 9 June 2009 Available online 14 June 2009 Keywords: Thin film composite NF membrane Polyvinyl alcohol Maleic acid Flux Preferential salt rejection

a b s t r a c t The present paper describes the preparation, characterization, and performance evaluation of thin film composite (TFC) nanofiltration (NF) membranes having porous polysulfone as base support and polyvinyl alcohol (PVA, degree of hydrolysis 86–87% and molecular weight 125,000) as the final barrier layer. Maleic acid (MA) was used as the cross-linker of PVA. The membranes were characterized by their molecular weight cutoff, FTIR, SEM, and contact angle. The effects of variation of different parameters like concentration of polysulfone, polyvinyl alcohol, maleic acid, and cure time on the membrane performance (flux and rejection of inorganic salts) were studied and the optimum membrane composition was evaluated. From the analysis of flux and rejection data it may be concluded that membranes prepared from 17% PSF, 1% PVA, with MA solution concentration of 0.2% (w/w) cured at a temperature of 125 ± 2 °C for 30 min give the optimum balance of flux and rejection (R). Such membranes show differential rejection among the sulfate and chloride salts. For membranes prepared under optimum conditions the average rejections of NaCl and MgSO4 are 22.8% and 83.8%, respectively; i.e., on an average 60% difference exists between the rejection of MgSO4 and NaCl. The overall trend of rejection by such membranes is RNa2 SO4 > RMgSO4 > RNaCl > RCaCl2 ffi RMgCl2 (R = rejection). The average MWCO of these membranes varies between 250 and 350 Da. Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction Nanofiltration membranes are a relatively new class of pressure-driven membranes with properties between reverse osmosis (RO) and ultrafiltration (UF) membranes [1,2]. They generally offer higher fluxes than RO and significantly better retention than UF for lower molar mass molecules such as sugars, natural organic matters, and even ions. NF membranes are charged membranes and have preferential rejection ability toward multivalent metal ions than the monovalent ones. NF offers several advantages such as low operation pressure, high flux, high retention of multivalent metal ions and organic molecules, relatively low investment, and low operation and maintenance cost, and hence NF membranes have wide scope of applications in water and wastewater treatment [3–6]. Poly (vinyl alcohol), a water-soluble biodegradable polymer, available with different degrees of hydrolysis has immense potential as membrane material because of its hydrophilicity and film forming characteristics [7–11]. The use of heat treated or formalized dense PVA film in desalination has been reported [12]. Greater swelling of PVA membrane in aqueous medium may lead to an

* Corresponding author. Fax: +91 278 2566970/91 278 2567562. E-mail addresses: [email protected], [email protected] (P. Ray). 0021-9797/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2009.06.020

open structure, which affects the membrane performance particularly the membrane rejection [13]. Hence there is a need to balance the hydrophilic and hydrophobic properties of such membranes, which may be achieved by cross-linking [14,15]. PVA may be cross-linked by using multifunctional compounds, such as dialdehydes [16–18], dicarboxylic acids [19,20], and dianhydrides [21], which are capable of reacting with the hydroxyl groups of PVA. Cross-linking improves the stability of such membranes. Although PVA cross-linked by different modes has been used in pervaporation membranes, the use of PVA in nanofiltration membranes is rare. PVA has immense potential as an NF membrane material because of the presence of innumerable numbers of hydroxyl groups in it, which will impart a polar character on the membrane surface. The main target of this work is to adopt a novel technique to prepare thin film composite nanofiltration membranes having porous polysulfone as base support and PVA as the selective barrier layer. It is intended to cross-link the PVA layer with dicarboxylic acid, i.e., maleic acid, which has been established as a good cross-linker of PVA in our previous work [22]. Major emphasis has been given in the optimization of membrane composition and curing conditions of such membranes. Maleic acid crosslinked barrier layer (PVA layer) of these membranes will be negatively charged because of the presence of unreacted hydroxyl/carboxyl groups and at the same time polar ester groups are also formed during the cross-linking reaction. Hence such membranes

122

J.M. Gohil, P. Ray / Journal of Colloid and Interface Science 338 (2009) 121–127

are expected to show preferential rejection toward multivalent ions than the monovalent ions and behave like nanofiltration membranes. Hard water is probably the most common water problem found not only at home but also in the industries causing scaling in boilers, cooling towers, pipelines, and other industrial equipments. Sulfate salt of calcium and magnesium is mainly responsible for the permanent hardness in water. These membranes may be useful for the removal of dissolved sulfates (salts) and partial desalination of brackish water.

ronmentally dried membranes were kept in liquid nitrogen and fractured. The membranes were further dried under vacuum at 30 °C and coated with a thin gold layer using sputter coater.

2. Materials and methods

2.2.2.4. Molecular weight cutoff (MWCO). A mixed solution containing glucose, sucrose, raffinose each of 500 ppm, and glycerol solution (500 ppm) was passed separately through the membranes mounted in RO test kit (effective membrane area 15.2 cm2). The separation experiments were carried out at a pressure of 150 psi. The percentage organic solute (glucose, sucrose, and raffinose) in feed and permeate was analyzed by using Waters2695-SUPELCOGEL C610H (30 cm 7.8 mm i.d.) HPLC columns with guard columns. Total organic carbon analysis was done by Liqui TOC Elemantar for estimation of glycerol. The solute rejection was calculated by using

2.1. Materials PVA, with degree of hydrolysis 86–87%, mol wt 125,000, polysulfone, Udel (mol wt 35,000), dimethyl formamide(AR grade) was from Merck, India; polyester nonwoven fabric (Nordyls TS 100) was from Polymer Group Inc. (France); and chemicals like maleic acid (MA), sodium chloride, magnesium chloride, calcium chloride, magnesium sulfate, sodium sulfate, and glycerol (All AR grade) were from SD Fine Chemicals (India). D-Glucose anhydride (LR grade) was from NICE Chemicals (India), sucrose (LR grade) was from SRL (India), and raffinose pentahydrate (LR grade) was from Loba Chemie (India). Water purified in Millipore water purifier (Milli-Q-gradient) was used.

2.2.2.3. Contact angle (h). The contact angles of resulting PVA TFC and polysulfone membranes were studied by the Wilhelmy plate method using a dynamic contact angle tensiometer (DCAT 21, Data Physics). The reported contact angle value is the average of five measured data.

Cp 100 R ð%Þ ¼ 1 Cf

ð1Þ

where R is the rejection, Cf and Cp are the concentrations of solute in feed and permeate, respectively. The solute rejection above 90% was taken as MWCO of membranes.

2.2. Methods 2.2.1. Preparation of TFC membrane The asymmetric membranes were prepared by a phase inversion technique. PSF solutions of different concentrations (13%, 15%, 17%, and 19% w/w) were prepared in DMF under constant stirring. The dissolution period was 24 h. Solutions were left at room temperature for a period of 2 h for the removal of air bubbles. PSF solutions with different polymer content were casted in the form of film on polyester nonwoven fabric supported on a glass plate with the help of a casting blade having exactly dimensioned slits. The glass plate with the cast solution was kept for 10 s in the ambient condition (temperature 30 ± 2 °C) and then immersed into the gelation bath containing deionized water. The porous polymer film was left in the gelation bath for 60 min. It was then kept in larger quantity of water for 48 h. The thickness of the membranes was 0.15 ± 0.02 mm. Aqueous PVA solutions of different concentrations (0.1%, 0.5%, 1%, 1.5%, and 2% w/v) were prepared by dissolving a definite weight of the polymer in Millipore water. The polymer solution was poured on PSF ultrafiltration membranes (acted as a porous support) mounted on a glass plates and kept in contact for 3 min. Excess PVA solution was drained off and the membranes were dried at ambient temperature (30 ± 2 °C). Cross-linker solution (MA solution of concentration 0.02–1% w/v in water) was then poured on the adsorbed PVA layer and kept for different time periods. The excess MA solution was drained off. The membrane was heated in an oven at a temperature of 125 ± 2 °C for 30 min to complete the curing [22].

2.2.2. Characterization of membranes 2.2.2.1. FTIR. The IR spectra of PSF ultrafiltration and cross-linked PVA TFC membranes were recorded in the middle infrared region using a Perkin Elmer 400 FTIR spectrometer. 2.2.2.2. Scanning electron microscopy (SEM). Cross-sectional and topographical phase morphologies of the membranes were studied using Leo 1400 microscope at 15 kV accelerating voltage. The envi-

2.2.3. Performance evaluation The salt rejection characteristics of membranes were studied in an RO test cell (area 15.2 cm2) by passing aqueous feed solutions of mono and bivalent salts separately through the membranes at 150 psi pressure. Permeate was collected for a definite time period (20 min) and the permeate flux was calculated with the help the equation

F¼

V At

ð2Þ

where F is the flux, A the effective area of membranes, and V is the volume of permeate collected during time period t. The concentrations of inorganic salts in feed and permeate were determined by inductively coupled plasma spectrometry (ICP, Optima 2000 DV model) and chemical analysis (for estimation of Cl ion). The solute rejection was calculated with the help of Eq. (1).

3. Results and discussion 3.1. Membranes characterization 3.1.1. FTIR Fig. 1 compares the IR spectra of PVA TFC membranes (graphs II–IV) with that of polysulfone base support (graph I). The three different PVA membranes were prepared by varying the PVA concentration from 0.5% to 2% (w/v) cross-linked by contact with 1% (w/v) aqueous maleic acid solution. In spectrum I the peaks observed at 1244 and 1488 cm1 correspond to the stretching vibration of C–O–C and CH3–C–CH3 of PSF base support [23]. The distinct broad absorption band at 3200–3500 cm1 is due to the stretching vibration of O–H groups of PVA (spectra II–IV) [24,25]. This type of absorption peak is absent in PSF base support, spectrum I. In the spectra II–IV the peaks at 1706 and 1097 cm1 may be due to the stretching vibration of C@O and C–O of the remaining unhydrolyzed vinyl acetate groups of PVA [26]. As the concentration of PVA increases from 0.5% to 2% (spectra II–IV)

123


Fig. 1. FTIR spectra of base PSF membrane (I) and PVA TFC membranes using PVA-1 solution of concentration, II, 0.5%; III, 1%; IV,2%. Curing condition: 1% MA solution, 125 ± 2 °C, 30 min.

the skeletal aliphatic C–C stretching peak of the PSF backbone chain at 1106 cm1 [23] merged with the C–O stretching band at 1097 cm1 of PVA (absent in spectrum I) [22] and appears as a single broad peak. A small new peak at 1739 cm1 appearing in spectra II–IV may be attributed to the ester carbonyl group formed due to cross-linking of PVA with MA. Enhancement of thickness of the PVA layer (with increase in concentration of PVA from 0.5% to 2%) is reflected in the broadening of the peak area at 3200–3500 cm1 and merging of peak at 1097–1106 cm1. 3.1.2. Contact angle (h) The PVA membranes are in general hydrophilic, and such property is expected to increase with increase in PVA content in the membranes. Again the cross-linking reaction decreases the hydrophilicity of such membranes because of the induced tight network. Such changes in hydrophilicity are reflected in contact angles of such membranes. The contact angle data were collected for PVA 92.5

TFC as well as polysulfone membranes and the results are shown in Fig. 2A and B. PSF membranes show the highest contact angle of 90.5° and it decreases with increases in PVA concentration (Fig. 2A), indicating the enhancement in hydrophilicity of the membranes. The effects of cross-linker dose on membrane contact angle are shown in Fig. 2B. As the concentration of MA solution increases from 0.02% to 1%, the contact angle increases from 86.98 to 88.76, indicating that membranes prepared using 0.02% MA concentration are more hydrophilic than those prepared using 1% MA solution. 3.1.3. Scanning electron microscopy SEM micrographs of base porous polysulfone support membrane and PVA composite membranes are presented in Fig. 3. The bilayer structure of PSF membranes with a top porous polysulfone layer on polyester base support is shown in this cross-sectional micrograph (Fig. 3A). The cross-sectional morphology of the 89.2

A

B

90

Contact angle, (θ).

Contact angle (θ)

88.8

87.5

85

88.4 88 87.6 87.2

82.5

86.8 0

0.5

1

1.5

PVA concentration (% w/v)

2

0

0.5

1

MA solution concentration (% w/v)

Fig. 2. Variation of contact angle with (A) PVA solution concentration (using 0.02% MA) and (B) MA solution concentration (concentration of PVA solution 1%). Membranes cured at 125 ± 2 °C, 30 min.

124


by varying the concentration of MA solution from 0.02% to 1%. It is observed (Fig. 4) that for any particular solute the percentage rejection increases with increase in MA concentration. This is because the enhancement in cross-linker dose results in a tighter network. For loose membranes prepared by using MA solution of 0.02% concentration, it is difficult to achieve >80% rejection with the selected organic compounds in the molecular weight range 90–600 Da. Hence for these membranes the MWCO is >600 Da. For membranes cross-linked with 0.05% MA solution, MWCO is >350 Da where the membranes show P90% rejection. For the other membranes cross-linked with MA solution (concentration 0.1–1%), MWCO lies in the range of 250–350 Da. For the separation of uncharged organic solutes pore dimension is one of the prime decisive factors. Increase in cross-link density results in lesser hydrophilicity and comparatively denser barrier layer. As a result for any organic solute the prohibition by membranes having a tighter barrier layer is comparatively more. 3.2. Optimization of different parameters for membrane preparation The variation of water flux of PSF membranes (used as porous base support for PVA TFC membranes) with PSF concentration was checked at 50 psi and shown in Fig. 5. It is observed that increase in PSF concentration results in densification of membrane structure. Fig. 6 depicts the cross-sectional pore morphology of PSF membranes prepared from 15% and 17% PSF solutions. Membranes prepared from 17% solution are characterized by comparatively dense pore morphology than those pre-

100

Rejection (%)

80

60 0.02% MA

40

0.05% MA 0.2% MA 0.5% MA

20

1% MA

0 0

150

300

450

600

Molecular Weight (Dalton)

PSF layer at a higher magnification (5650) is shown in Fig. 3B. A distinct pore morphological pattern with interconnected pores has been observed in this case. The topographical feature of PVA TFC composite membrane is shown in Fig. 3C (magnification 5710). Comparison of Fig. 3B and C indicates much denser pore morphology for the latter than the former. Partial masking of the polysulfone support by PVA layer is clearly visible in this micrograph, which results in the reduction of the pore size of the membranes.

Fig. 4. Molecular weight vs. % rejection of PVA TFC membranes with variable crosslink density of PVA layer at operating pressure of 150 psi.

Water Flux (lm-2d-1) × 103

Fig. 3. Scanning electron micrographs of PSF base (17% PSF) membranes (A and B) and PVA TFC membranes (C).

10 8 6 4 2

3.1.4. Molecular weight cutoff The rejection ability of the membranes toward different organic solutes varying in their molecular weight from 90 to 600 g/mol was studied. Membranes were varied in their cross-link density

12

14

16

18

20

PS F Concentration (% w/w) Fig. 5. Variation of base membrane flux with PSF concentration.

125


Fig. 6. Cross-sectional pore morphology of PSF membranes prepared from (A) 15% and (B) 17% PSF solutions (9260).

pared from 15% solution and it is observed that water flux of the membranes decrease with increase in PSF concentration. The achieved range of pure water permeability is highly suitable to use these membranes as porous substrate of TFC membranes. In the PVA–MA system, the gel network is caused by the formation of interchain ester linkage between hydroxyl groups of different PVA chains through the carboxyl groups of MA. Selectivity of the PVA active layer depends on the cross-linked density and also on the number of free hydroxyl groups in PVA, which in turn varies with maleic acid concentration. Apart from acting as a cross-linker, the unesterified carboxyl groups of maleic acid will also impart a charged character on the membrane surface. The following sections deal with the studies related to the optimization of membrane composition suitable for NF applications. 3.2.1. PSF/PVA concentration For PVA TFC membranes the flux and rejection depend on the pore morphology of base (PSF support) as well as barrier layer (PVA), which in turn varies with the concentrations of PSF and PVA. The variation of membrane flux and salt rejection properties with variation of PSF and PVA solution concentrations are shown in Figs. 7 and 8, respectively. It is observed that for membranes prepared with same concentration of PVA, increase in PSF concentration results in membranes with lesser flux (Fig. 7) but higher salt (MgSO4) rejection (Fig. 8).

This is quite expected because looser base support results in higher flux due to the facile flow channels of water (Fig. 6). Entrapment of PVA in tight PSF support will give a comparatively more uniform rejection layer than the entrapment of same concentration of PVA in looser PSF support membrane. Hence the probability of polar interaction among the hydroxyl groups on the membrane surface with the ions in the solution will be more for the membranes with tighter base support and as a result salt rejection increases. For any fixed concentration of PSF, as the concentration of PVA solution increases, the flux decreases (Fig. 7). Increase in PVA concentration results in dense barrier layer, which is reflected in gradual fall in membrane flux, however the salt (MgSO4) rejection increases (Fig. 8) with increase in PVA concentration. It may be seen from Fig. 1 that for the same base support membrane (PSF) the higher the concentration of PVA the higher the surface coverage of PSF base membrane, which is reflected by the broadening of the peak area between 3200 and 3500 cm1. The increase in salt rejection with increase in PVA concentration is predominant for the membranes prepared using a loose PSF base matrix. For membranes prepared on loose PSF support (13% PSF concentration) increase in PVA concentration from 0.5% to 2% results in an enhancement of MgSO4 rejection from 27% to 55% whereas for membranes prepared on tight base support (19% PSF concentra-

100 90

MgSO4 Rejection (%)

2500

Flux (lm-2d-1)

2000 0.5% PVA 1% PVA

1500

2% PVA

1000 500

80 70 60 50

0.5% PVA 1% PVA

40

2% PVA

30 20

0 12

13

14

15

16

17

18

19

20

PSF Concentration (% w/w) Fig. 7. Variation of flux with PSF concentration for PVA TFC membranes, operating pressure of 150 psi. MA solution concentration 0.02% (w/v); cure at 125 ± 2 °C for 30 min.

12

13

14

15

16

17

18

19

20

PSF concentration (% w/w) Fig. 8. Variation of MgSO4 rejection with PSF concentration for PVA TFC membranes, operating pressure 150 psi. MA solution concentration 0.02% (w/v); cure at 125 ± 2 °C for 30 min.


3.2.3. Contact time of MA solution with PVA layer Contact time of PVA layer with MA solution plays a dominant role from view point of stability of the rejection layer of such mem-

600

-2 -1

Flux (lm d )

500 400 300 200 100

70

0.02% MA 0.05% MA 0.1% MA 0.2% MA 0.5% MA 1% MA

50 5

10

15

20

25

Time (hours) Fig. 10. Variation of MgSO4 rejection with operation time for PVA TFC membranes with different cross-link densities of PVA layer, operating pressure 150 psi.

branes. It is observed that beyond a contact time of 55–60 s the membrane flux increases and rejection of MgSO4 decreases (Fig. 11). Contact for a longer duration of MA solution with PVA layer results in slow diffusion of PVA from PSF substrate toward the aqueous phase of MA solution. This disturbs the stability of the rejection layer and hence fluxes increases and rejection decreases. 3.2.4. Cure time Cure time of PVA layer is one of the important parameters for the formation of stable selective layer on PSF membrane. Both the flux and rejection of the membranes vary with the cure time of PVA layer (Fig. 12). Enhancement of cure time (keeping the curative dose and curing temperature constant) results in increases of membrane rejection at the cost of flux. Fig. 12 indicates that a cure time beyond 25–30 min produces membranes with almost saturated values in flux and rejection. 3.3. Desalination performance of membranes Maintaining the optimum parameters achieved from the above studies like PSF concentration 17%, PVA concentration 1%, contact time of 50 s with MA solution, curing temperature 125 ± 2 °C, cure time 30 min, different membranes were prepared with varied degree of cross-linking (varying the concentration of maleic acid

MgSO4 Rejection (%)

0.02% MA 0.05% MA 0.1% MA 0.2% MA 0.5% MA 1% MA

90

100

180

95

160

90

140

85

120

80

100

75 80

70

60

65 60

40

55

20

50

0 5

10

15

20

25

Time (hours) Fig. 9. Variation of flux with operation time for PVA TFC membranes with different cross-link densities of PVA layer, operating pressure 150 psi.

-2 -1

3.2.2. Curative (MA solution) concentration TFC membranes have been prepared by varying curative dose, i.e., MA solution concentration (0.02–1%) while keeping the other parameters (PSF concentration 17%, PVA concentration 1%, Cure time 30 min and curing temperature 125 ± 2 °C) constant. Membrane performances (flux and rejection) were studied in continuous mode for duration of 24 h. The variation of flux and rejection with operation time are reflected in Figs. 9 and 10, respectively. It is observed that for membranes prepared using higher crosslinker dose (MA concentration 1%) the flux and rejection remain unaffected with time whereas membranes cured with lower curative dose ( RMgSO4 > RNaCl > RMgCl2 ffi RCaCl2 . Such membranes may be useful for the removal of dissolved sulfates (salts) and partial desalination of brackish water. Acknowledgment

0 0

of brackish water. From the MWCO values and the rejection profile, the developed membranes may be defined as nanofiltration membranes. Comparing the rejection of different chloride salts it is observed that the rejection of NaCl > rejection of MgCl2 and CaCl2. This separation trend indicates that for the same anionic counterpart the rejection of electrolytes depends on cationic charge density and it decreases with an increase in charge density [27].

500

Flux (lm-2d-1)

100

127

0.2

0.4

0.6

0.8

1

MA Solution (% w/v) Fig. 13. Performance study of different PVA TFC membranes using different mono and divalent inorganic salt solutions, operating pressure 150 psi.

solution). Different mono and bivalents salts were selected for the present study. Desalting performances of these membranes along with flux data (150 psi) are shown in Fig. 13. With increases in MA solution concentration from 0.02% to 1%, almost 10 times decline in flux data has been observed. Membranes with highest cross-link density (using 1% MA solution) show 2.5–5 times more rejection than the membranes with lowest cross-link density (0.02% MA solution) and this is valid for all the different salt solutions studied here (Fig. 13). It is also observed that RNa2 SO4 > RMgSO4 > RNaCl > RCaCl2 ffi RMgCl2 (R = rejection). Membrane surfaces are expected to be negatively charged due to the presence of –OH groups of PVA. Because of the higher charge den sity of sulfate (SO2 4 ) ions compare to chloride (Cl ) ions the rejection of sulfate salts are more than those of chloride salts [19]. For membranes prepared using maleic acid solution concentration of 0.2%, the rejection of NaCl and MgSO4 are 22.8% and 83.8%, respectively, i.e., 60% difference or a rejection ratio of 3.67 exists between the rejection of MgSO4 and NaCl. Because of the preferential rejection of sulfate salts compare to chlorides such membranes will be helpful for the removal of hardness as well as partial desalination

The authors are highly thankful to Council of Scientific and Industrial Research (CSIR), India, for financial assistance in the work. References [1] A.A. Hussain, S.K. Nataraj, M.E.E. Abashar, I.S. Al-Mutaz, T.M. Aminabhavi, J. Membr. Sci. 310 (2008) 321. [2] R.W. Baker, Membrane Technology and Applications, second ed., Wiley, Chichester, 2004. [3] B. Van der Bruggen, K. Everaert, D. Wilms, C. Vandecasteele, J. Membr. Sci. 193 (2001) 239. [4] N. Hilal, H. Al-Zoubi, N.A. Darwish, A.W. Mohammad, Sep. Sci. Technol. 40 (2005) 3299. [5] M. Nilsson, G. Tragardh, K. Ostergren, J. Membr. Sci. 280 (2006) 928. [6] W. Zhang, G. He, P. Gao, G. Chen, Sep. Purif. Technol. 30 (2003) 27. [7] T. Hirai, Y. Asada, T. Suzuki, S. Hayashi, J. Appl. Polym. Sci. 38 (1989) 491. [8] N.A. Ibrahim, M.H. Abo-Shosha, Angew. Makromol. Chem. 210 (1993) 7. [9] K. Yamaura, M. Itoh, T. Tanigami, S. Matsuzawa, J. Appl. Polym. Sci. 37 (1989) 2709. [10] W.I. Cha, S.H. Hyon, Y. Ikada, Makromol. Chem. 193 (1992) 1913. [11] G.H. Hsiue, Z.S. Chou, N. Yu, K.P. Hasiung, J. Appl. Polym. Sci. 34 (1987) 319. [12] C.T. Chen, Y.J. Chang, M.C. Chen, A.V. Tobolsky, J. Appl. Polym. Sci. 17 (1973) 789. [13] M.G. Katz, T. Wydeven, J. Appl. Polym. Sci. 27 (1982) 79. [14] R.Y.M. Huang, A. Moreira, R. Notarfonzo, Y.F. Xu, J. Appl. Polym. Sci. 35 (1988) 1191. [15] R.Y.M. Huang, Y.F. Xu, J. Membr. Sci. 43 (1989) 143. [16] W.I. Cha, S.H. Hyon, Y. Ikada, Die Makromol. Chem. 194 (1993) 2433. [17] B. Gebben, H.W.A. van den-Berg, D. Bargeman, C.A. Smolders, Polymer 26 (1985) 1737. [18] V. Macho, M. Fabini, M. Rusina, S. Bobula, M. Harustiak, Polymer 35 (1994) 5773. [19] R.Y.M. Huang, J.W. Rhim, Polym. Int. 30 (1993) 129. [20] R.W. Korsmeyer, N.A. Peppas, J. Membr. Sci. 9 (1981) 211. [21] V. Gimenez, A. Mantecon, V. Cadiz, J. Appl. Polym. Sci. 59 (1996) 425. [22] J.M. Gohil, A. Bhattacharya, P. Ray, J. Polym. Res. 13 (2006) 161. [23] P.S. Singh, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, A.P. Rao, P.K. Ghosh, J. Membr. Sci. 278 (2006) 19. [24] P.S. Kalsi, Spectroscopy of Organic Compounds, second ed., New Age International Limited Publication, New Delhi, 1996. [25] M.K. Lindemann, Encyclopedia of Polymer Science and Technology, vol. 14, Wiley, New York, 1971, p. 149. [26] J.H. Kim, E.J. Moon, C.K. Kim, J. Membr. Sci. 216 (2003) 107. [27] K. Lang, G. Chowdhury, T. Matsuura, S. Sourirajan, J. Colloid Interface Sci. 166 (1994) 239.