Journal of Membrane Science 230 (2004) 1–11

Acrylonitrile-based copolymers containing reactive groups: synthesis and preparation of ultrafiltration membranes Fu-Qiang Nie a , Zhi-Kang Xu a,∗ , Lin-Shu Wan a , Peng Ye a , Jian Wu b,1 a

Institute of Polymer Science, Zhejiang University, Hangzhou 310027, PR China Department of Chemistry, Zhejiang University, Hangzhou 310027, PR China

b

Received 18 July 2003; received in revised form 20 July 2003; accepted 12 October 2003

Abstract Copolymerization of acrylonitrile (AN) and maleic anhydride was conducted by a water-phase precipitation copolymerization process (WPPCP) with K2 S2 O8 –Na2 SO3 as initiator system and the resultant copolymers were used to fabricate ultrafiltration hollow fiber membranes (UHFMs) by a dry–wet phase inversion method. Structures and properties for the UHFMs were studied by the analysis of scanning electron microscopy and the measurements of water permeation, bovine serum albumin (BSA) rejection and breaking strength. Compared with normal solution copolymerization, only poly(acrylonitrile-co-maleic acid)s (PANCMAs) could be obtained with WPPCP due to the hydrolysis of maleic anhydride, however, advantages of WPPCP including high monomer conversion and high copolymer molecular weight were obtained. It was found that increasing the molecular weight of PANCMA or the concentration of casting solution raised BSA rejection and mechanical strength but decreased water flux for the corresponding UHFMs. Membrane structures and properties could also be adjusted by adding additives such as poly(vinyl pyrrolidone) and poly(vinyl alcohol) (PVA) to the casting solution. Ternary phase diagrams of the membrane-forming systems and viscosity data of the casting solutions were used to characterize the characteristics of thermodynamics and diffusion kinetics for the membrane-forming systems. Most promisingly, it was found that the acid groups on membrane surface of PANCMA could be conveniently converted to more reactive anhydride groups by treating the membranes with acetic anhydride/pyridine mixture. Finally, the potentiality for the immobilization of poly(ethylene glycol) (PEG) on membrane surface by the reaction of PEG with anhydride groups to further improve the membrane properties was briefly indicated using Fourier-transform infrared spectroscopy (FT-IR), pure water contact angle and BSA adsorption measurements combining water permeation and BSA rejection determinations. © 2003 Elsevier B.V. All rights reserved. Keywords: Poly(acrylonitrile-co-maleic acid); Ultrafiltration hollow fiber membrane; Phase inversion; Additive; Hydrophilicity

1. Introduction Polyacrylonitrile and acrylonitrile (AN)-based copolymers have been successfully applied as membrane materials in the fields of dialysis, ultrafiltration, enzyme-immobilization, molecular imprinting and pervaporation [1–11]. Ultrafiltration membranes are widely used in the biotechnology industry for the recovery of biological products in steps such as cell broth clarification, cell harvesting, concentration or diafiltration of protein solution prior to separation, and final concentration [12]. Nevertheless, one major problem with ultrafiltration is the loss of permeation flux caused by adsorptive fouling of biological molecules such as proteins ∗

Corresponding author. Fax: +86-571-8795-1773. E-mail address: [email protected] (Z.-K. Xu). 1 Co-corresponding author.

0376-7388/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2003.10.016

on the surface and even inside the pores of typically used hydrophobic membranes such as polyacrylonitrile. Fouling reduces productivity due to longer filtration times and shortens membrane life due to the harsh chemicals necessary for cleaning. What’s more, it can also alter membrane selectivity and lead to significant product loss owing to denaturation. Therefore, the relatively poor hydrophilicity and biocompatibility for this type of membrane limit their further applications in aqueous solution, enzyme-immobilized membrane bioreactor and biomedical usage. Increasing the hydrophilicity of the membrane surface can reduce fouling and improve biocompatibility for the membranes [11–16]. Accordingly, there are many surface modification methods that have been reported to make ideal hydrophilic and fouling-resistant surfaces. Among them, the grafting of hydrophilic monomers on the membrane surface shows some promise [10,12–16]. However, grafting

2

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

polymerizations induced by radical, plasma, electron-beam, ␥-radiation, and ultra-violet may result in production of significant amount of homopolymer or crosslinked polymer. The undesired homopolymer wastes expensive starting materials, and crosslinked polymer is detrimental to membrane filtration since the membrane pores may become blocked. Moreover, the grafting density (number of grafting sites per area) and grafting polymer chain length cannot be determined independently, much less controlled. Copolymerizing hydrophilic monomers, such as acrylic acid, acrylamides, vinyl-pyrrolidone, 2-hydroxyethyl methacrylate, and 4-vinylpyridine with acrylonitrile has also been used to improve the properties of acrylonitrile-based polymeric membranes [3,5–8,11]. It has been recently reported by Hicke et al. [7] that acrylonitrile copolymer containing reactive glycidyl groups can be used to fabricate solvent-resistant and autoclavable membrane. They demonstrated that both the chemical structure and the morphology of the membrane could be stabilized by a post-treatment with ammonia. Our previous work revealed that carbohydrate containing acrylonitrile-based copolymers could be simply synthesized by water-phase precipitation copolymerization process (WPPCP) of ␣-allyl glucoside (AG) with acrylonitrile [17]. Compared with common solution process, main advantages of WPPCP for the synthesis of acrylonitrile-based copolymers include high molecular weight, relative high monomer conversion, and environmental friendship. The previous results also revealed that both the hydrophilicity and biocompatibility of polyacrylonitrilebased membranes could be improved by copolymerizetion of acrylonitrile with vinyl carbohydrates. In order to improve the hydrophilicity and antifouling property of polyacrylonitrile-based membranes, it may be also one effective method to copolymerize acrylonitrile with vinyl monomers containing reactive groups such as maleic anhydride. These anhydride groups can easily undergo a ring opening reaction with nucleophilic reagents which contain hydroxyl or amine groups. For examples, styrene/maleic anhydride copolymers have been reacted with methoxy poly(ethylene glycol) (PEG) to synthesize amphiphilic graft copolymers or thermo-sensitive and pH-sensitive polymers [18,19]. More recently, the reaction of anhydride groups with amine-containing carbohydrates has also been reported to improve biocompatible or biospecific properties for styrene/maleic anhydride and N-vinyl-pyrrolidone/maleic anhydride copolymers [20,21]. In this paper, therefore, copolymerization of acrylonitrile and maleic anhydride was carried out by WPPCP with potassium persulfate-sodium sulfite as initiator system and water as reaction medium. Ultrafiltration hollow fiber membranes (UHFMs) were fabricated from these copolymers by a dry–wet phase inversion process and the influence of fabrication parameters, such as copolymer molecular weight, dope solution concentration and additive addition to casting solution, on membrane structures were studied. Finally, the membranes were treated in acetic anhydride/pyridine mix-

H2C

CH

HC

+

CN

O

CH

C

C

K2S2O8-Na2SO3 H2O, 60 oC, 3 h

O

O Maleic anhydride

Acrylonitrile

CH2

CH

CH

CH

n NC

m COOH COOH

Poly(acrylonitrile-co maleic acid)

Scheme 1. Schematic representative for the copolymer synthesis with WPPCP.

ture followed by the immobilization of PEG. The potentiality of copolymerizing acrylonitrile with maleic anhydride to improve the hydrophilicity of polyacrylonitrile-based membranes was revealed.

2. Experimental 2.1. Materials All chemicals were analytical grade. Acrylonitrile and dimethyl sulphoxide (DMSO) were commercial products and were purified by vacuum distillation before used. Maleic anhydride was used without further purification. Potassium persulfate (K2 S2 O8 ), anhydrous sodium sulfite (Na2 SO3 ), and azobisisobutyronitrile (AIBN) were recrystallized by usual procedure, respectively. Bovine serum albumin (BSA, pI = 4.8, Mw = 66 kDa) was purchased from Sino-American Biotechnology Co. and used as received. Poly(vinylpyrrolidone) (PVP, Mn = 1 × 104 g/mol) and poly(vinyl alcohol) (PVA, Mn = 1700 g/mol) as additives were commercial products and used as received. The nonsolvent selected for the inner and external coagulation bath is ultrafiltrated water. 2.2. Synthesis and characterization of the copolymers The schematic representative for the synthesis of poly(acrylonitrile-co-maleic acid)s (PANCMAs) with WPPCP is shown in Scheme 1. Copolymerization of acrylonitrile and maleic anhydride was performed in a 5 l reactor equipped with mechanical stirrer by varying the molar ratios of total monomer to initiator and acrylonitrile to maleic anhydride. The reaction was continued at 60 ◦ C under nitrogen atmosphere for 3 h and the precipitated copolymer was filtered, washed with excess de-ionized water and ethanol to remove residual monomers. Solution copolymerization of acrylonitrile and maleic anhydride initiated by AIBN in DMSO were carried out at 60 ◦ C with a usual procedure for comparison. The oxygen contents of the copolymers were obtained by elemental analysis (EA1110, CE Instruments) and used to calculate the weight fractions of maleic acid

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

3

and/or maleic anhydride in the copolymers. IR spectra were measured on a Brucker Vector 22 spectrometer. Viscosity measurements were made in a thermostatic water bath at 30 ± 0.1 ◦ C using an Ubbelchde viscometer. Molecular weight of the copolymer was obtained from the relationship for polyacrylonitrile in DMSO at 30 ◦ C [17].

completely remove DMSO and additives. Finally, the hollow fiber membranes were immersed in 50 wt.% glycerin aqueous solutions for 3 days and dried in air at ambient condition.

[η] = 2.865 × 10−2 Mv0.768

To transfer the reactive group on membrane surface from acid to anhydride, the hollow fiber membrane was washed in ethanol for 24 h, in hexane for another 24 h, and air-dried at ambient condition. Then, the dry PANCMA hollow fiber membrane was immersed in acetic anhydride at 90 ◦ C for 60 min and/or in acetic anhydride/pyridine solution (1:1 volume ratio) at 100 ◦ C for 30 min. After that, the membrane was thoroughly washed with acetone and dried in vacuum oven. Finally, it was treated with PEG 200 at 90 ◦ C under nitrogen atmosphere for 96 h. The resulted membrane was washed using large amount of de-ionized water, diethyl ether, and dried in a vacuum oven at 40 ◦ C for 24 h.

where [η] is the intrinsic viscosity and Mv is the viscosity average molecular weight. 2.3. Characterization of the dope solutions for membrane fabrication Intrinsic viscosity ([η]) of PANCMA in DMSO containing additives was determined using dilute solution (1 mg/ml) with an Ubbelohde-type viscometer in a constant temperature bath at 30 ± 0.1 ◦ C. The transparency of solution containing additives at different temperature was estimated visually to reveal the miscibility of acrylonitrile/maleic acid copolymer with various additives in the dope solution. Absorbance values of the dope solutions at 700 nm were also measured to estimate this point. The ternary phase diagrams of membrane formation system, PANCMA/DMSO/H2 O with additive and/or not, were obtained by means of turbidity measurements and theoretical calculations based on experimentally determined parameters of Boom’s linearized cloud point (LCP) relation [22]. Cloud points for DMSO solutions containing a certain amount of PANCMA (0.5–8.0 wt.%) with additive and/or not were determined by a titrimetric method [23,24]. Then, the cloud points for the dope solution at high polymer concentration were calculated on the basis of the linearized cloud point correlation [22]. Binodal lines in the ternary phase diagram were obtained by combining the experimental data of the cloud point measurement with the linearized data in a wide concentration range. 2.4. Preparation of hollow fiber membranes PANCMA was dissolved in DMSO at about 80 ◦ C for 24 h followed by the addition of additive with vigorous stirring until the homogenous polymer solution was formed. Ultrafiltrated water at 25 ◦ C was used as the inner and external coagulation bath for the membrane preparation. All membranes were prepared by the dry–wet phase inversion method at ambient condition (25 ± 1 ◦ C, 65–70% relative humidity). A tube-in orifice type spinneret with the outer diameter 1262 ␮m and the inner diameter 709 ␮m was used. The extrusion rate of the dope solution was controlled at 1.8–2.5 ml/min and the injection rate of the inner coagulant was 0.8–1.2 ml/min. The nascent hollow fibers emerged from the tip of the spinneret and passed through an air gap of 70 mm before entering the external coagulation water bath. The take-up velocity was nearly the same as the free falling velocity of the nascent hollow fibers. The hollow fibers were then washed and kept with ultrafiltration water for 24 h to

2.5. Surface modification of the membranes

2.6. Characterization of the hollow fiber membranes The morphologies of the prepared hollow fiber membranes were inspected by SEM using a Hitachi S-570 Scanning Microscope. To investigate the changes of chemical structure for the copolymer membrane surface, Fouriertransform infrared spectroscopy (FT-IR; Vector 22 FT-IR) with an ATR unit (KRS-5 crystal, 45◦ ) was used. Static water contact angles of the membrane surface were measured by the sessile drop method at 25 ◦ C under an atmosphere of saturated water vapor with a contact angle goniometer (KRÜSS DSA10-MK) equipped with video capture. Bovine serum albumin adsorption was carried out by following method. BSA (5.0 mg) was dissolved in 10 ml Tris–HCl buffer (pH = 8) that can facilitate the hydrophobic interaction and inversely depress the electrostatic binding between the protein and the polymer surface. Hollow fiber membrane with 20 cm2 external surface areas (about 0.212 g, 0.386 ml) and 5 ml BSA solution were introduced into a tube. After the protein solution remained in contact with the sample for 24 h at 30 ◦ C, the membrane was taken out from the protein solution and was further rinsed gently until the surface remained constant. The amount of adsorbed protein was determined by measuring spectrophotometrically the difference between the concentration of albumin in the solution before and after contact with the membrane. The reported data was the mean value of triplicate samples for each copolymer. The filtration experiment of water and BSA solution were performed using an experimental apparatus designed for membrane characterization in our laboratory. Hollow fiber membrane modules with tube and shell configuration were used for these measurements. Each module contained five fibers with the length of 20 cm encased in a glass tube. For each membrane sample, three modules were tested in parallel and an average of their performance was reported. Pure water and/or 500 ppm BSA solution was fed at constant pressure of 0.1 MPa from the inner lumen to the outer surface

4

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

of the hollow fiber, and the filtrate was collected in a flask and weighed. The BSA concentrations in the feed and filtrate were analyzed using a spectrophotometer (Shimadzu, UV-1601) at 280 nm. The solute rejection (R) was calculated by
 Cp R= 1− × 100% (1) Cf where Cp and Cf are the BSA concentration in permeate and feed, respectively. After the ethanol–hexane solvent exchange step, the hollow fiber sample was vacuum dried at 80 ◦ C for 24 h. Then, the mechanical properties of the hollow fiber membrane were examined with a Shimadzu, AGS-500ND tensile test machine. The sample was tested with a gauge length of 50 mm at elongation velocity of 50 mm/min. The breaking strength (S) is the breaking force divided by the cross-section area of the membrane: F S= (2) A where F is the load at the breaking point and A is the cross-section area of the membrane. At least five samples were tested for each reported datum.

3. Results and discussions 3.1. Synthesis and characterization of the copolymers Typical copolymerization results of acrylonitrile and maleic anhydride by water-phase precipitation copolymerization process (WPPCP) are listed in Table 1. Compared with common free radical copolymerization carried out in DMSO using AIBN as initiator, it was found that the total yield and the molecular weight of the copolymer were higher for WPPCP. Water-phase precipitation polymerization was a unique process that could afford the advantage of increasing molecular weight for acrylonitrile-based polymers [17], which, in turn, would benefit the mechanical strength of corresponding polyacrylonitrile-based membranes. When the slightly water-soluble monomer, acrylonitrile, was added to the medium, a small fraction of acrylonitrile dissolved in the continuous aqueous phase where initiator was

(a)

(b)

3100

2700

2300

1900

Wavenumber

1500

1100

700

(cm-1)

Fig. 1. FT-IR spectra for the copolymers of acrylonitrile with maleic anhydride synthesized by WPPCP (a) and solution copolymerization (b).

present. Then, the polymerization took place first in water. After the chains in water grew big enough, they would precipitate from the water medium. At this time, the polymerizations transferred from homogeneous polymerization to heterogeneous polymerization, which meant that the polymerizations conducted both in water and at the surface of the micro-particles precipitated from the water. In such condition, the chances of chain termination and chain transfer became smaller, therefore, the molecular weight and the yield of the polymer increased. However, one obvious limit for the copolymers synthesized by the WPPCP was the relative low content of reactive groups desired to be incorporated in, as can be seen from Table 1. Because maleic anhydride was easy to hydrolyze into maleic acid in water, the copolymerization using the WPPCP method was in fact promoted between acrylonitrile and maleic acid. It could be confirmed by the FT-IR spectra of the copolymers obtained both by WPPCP and solution copolymerization. As can be seen from Fig. 1, the spectrum of the copolymer synthesized by solution polymerization displayed characteristic anhydride peaks at 1860 and 1785 cm−1 . In the spectrum of the copolymer obtained from WPPCP, however, the anhydride peaks disappeared, and instead, the characteristic absorptions of carboxylic acid at 1732 and 1633 cm−1 were observed. In our previous work concerning the copolymerization of acrylonitrile with ␣-allyl glucoside [17], since AG was highly soluble and acrylonitrile was slightly soluble in

Table 1 Typical results for the WPPC of acrylonitrile with maleic anhydride at 60 ◦ C Sample no.

AN/MA

I/M

Yield (%)

[η] (dl/g)

¯ η × 104 (g/mol) M

MA mole fraction in the copolymer (%)

PANCMA-1 PANCMA-2 PANCMA-3 PANCMA-4 PANCMA-5 PANCMA-6a

8.0:1.0 8.0:1.5 8.0:1.0 8.0:1.5 8.0:1.0 8.0:1.0

1:30 1:80 1:100 1:120 1:150 1:80

83 78 81 75 80 37

1.26 2.20 2.54 2.78 3.21 0.19

9.38 20.4 25.1 27.9 33.7 0.47

4.31 5.13 3.36 4.76 3.78 8.25

a

Solution copolymerization in DMSO using AIBN as initiator.

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11 120

200

100

150 80 60

100

40

50 20 0

0

1

3.2. Formation and characterization of ultrafiltration hollow fiber membranes from the PANCMAs Ultrafiltration hollow fiber membranes were fabricated from casting solutions containing DMSO and 13 wt.% of the PANCMA (designated as PANCMA-1 to PANCMA-5 in Table 1) synthesized in this work. The water flux and bovine serum albumin rejection for the membranes are shown in Fig. 2. Because the molecular weight of the copolymer increased from PANCMA-1 to PANCMA-5 (Table 1), the results shown in Fig. 2 indicated that, increasing the molecular weight of PANCMA decreased the water flux gradually and increased the BSA rejection slightly for the corresponding membrane. For PANCMA with high molecular weight, the viscosity of dope solution was higher than that with low molecular weight at the same concentration. High viscosity normally slowed the exchange between solvent and nonsolvent in phase separation process used in this work for membrane preparation, resulting in the decrease of pore size and pore numbers on membrane surface. Accordingly, the water flux decreased and the BSA rejection increased. However, with the exception of molecular weight, the monomer composition, the molecular weight distribution, and even the composition distribution of copolymers should also influence the membrane structures and properties in some extent. This was obviously too complicated to establish the rela-

2

3 4 Sample number

tion between processing parameters and membrane properties. Therefore, we focused our discussions on copolymer PANCMA-2 in the following paragraphs. The effects of PANCMA-2 dope concentration on membrane properties were studied and the results are shown in Figs. 3 and 4, respectively. It can be seen that, at the experimental concentration range, the water flux of the membrane decreased gradually, whereas the BSA rejection increased obviously at first but varied little when the concentration exceeded 13 wt.%. As other spinning conditions were kept similar, increasing dope concentration usually resulted membranes with tight structure, which also led to the increase of mechanical strength for the hollow fiber membranes, as shown in Fig. 4. Concerning the balance of BSA rejection, water flux, mechanical strength and processing convenience, 13 wt.% was selected as the dope concentration of PANCMA-2 for membrane fabrication. In order to control the membrane structure, low molecular weight component or a secondary polymer was frequently used as the additive in the membrane-forming system because it offers a convenient and effective way to develop 100

300

95

2

90

200 85

150 80

100 50 10

11

12

13

14

15

BSA rejection (%)

250

9

5

Fig. 2. Water flux and BSA rejection for UHFMs fabricated from acrylonitrile/maleic acid copolymers. (1) PANCMA-1, (2) PANCMA-2, (3) PANCMA-3, (4) PANCMA-4, (5) PANCMA-5.

350

Water flux (L/m .hr.atm)

BSA rejection

2

Water flux (L/m .hr.atm)

Water Flux

Rejection (%)

water, both the AG content in the copolymers and the AG conversion for WPPCP were higher than those of solution copolymerization in DMSO. Therefore, it was expected that the relative high concentration due to its water-soluble property should favor the incorporation of maleic acid into polymer chains. Nevertheless, the low copolymerization activity might seriously suppress the effect of high monomer concentration. As a result, the content of maleic acid in the copolymer synthesized by WPPCP was lower than that obtained by solution copolymerization.

5

75 16

Concentration of the polymer solution (wt.%) Fig. 3. Effect of casting solution concentration on the water flux and BSA rejection of PANCMA-2 UHFMs.

6

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

Breaking strengthen

(N/cm2)

160

140

120

100

80 9

10

11

12

13

14

15

Concentration of the polymer solution (wt.%) Fig. 4. Effect of casting solution concentration on the breaking strengthen of the PANCMA-2 UHFMs.

200

95

150 90 100 85

50

0

(%)

Water flux

100 BSA rejection Water flux

BSA rejection

(L/m2.hr.atm)

250

80

1

2

3

Fig. 5. Comparison of water flux and BSA rejection for the PANCMA-2 UHFMs prepared with additives. (1) No additive, (2) 4.0 wt.% PVP, (3) 4.0 wt.% PVA.

membranes with high performances [25–27]. In this work, water-soluble polymers, poly(vinylpyrrolidone) (PVP) and poly(vinyl alcohol), were used as additives to modify membrane structures and properties. The compositions of casting solutions for this purpose are listed in Table 2. Permeation results shown in Fig. 5 indicated that, when the total concentration of casting solution was constant as 13 wt.%, with the addition of additives such as PVP and PVA, the water flux

of PANCMA-2 membrane increased and the BSA rejection decreased. The morphologies of cross-section, external surface and inner edge for the hollow fiber membranes were studied with SEM (Figs. 6–8). As can be seen from Fig. 6, the membranes had an asymmetric structure consisting of a dense skin layer and a porous sub-layer. By the addition of 4.0 wt.% PVP or PVA and the decrease of PANCMA-2 concentration from 13.0 to 9.0 wt.%, some pores formed on the outer surface (see Figs. 7a–c) because PVA and PVP were water-soluble additives and would leach out of the membrane to leave pores on membrane surface. However, finger-like macrovoids beneath the membrane surface seems to be suppressed (Figs. 6a–c) in the case of PVP. Furthermore, voids beneath the membrane surface with PVA as additive were larger and more much than that with PVP as additive. These morphologies were consistent with the water permeation and BSA rejection results mentioned in Fig. 5. On the other hand, the formation of macrovoids was effectively suppressed by keeping PANCMA-2 concentration at 13 wt.% and adding PVP to the casting solution, as can be seen from Figs. 6a, d and e. With the increase of PVP from 0 to 4 and then to 7 wt.%, a spongy substructure appeared instead of the finger-like macrovoids (Figs. 6 and 8). Moreover, with the exception of roughness in some extent, no pores were observed (Figs. 7d and e) on the external surface for these membranes by SEM at a magnification of 10000. Therefore, corresponding to these membrane structures, Fig. 9 indicated that the water flux decreased and the BSA rejection increased for the membrane fabricated with the increase of PVP from 0 to 7 wt.% in the casting solution. To understand the structures and properties of the PANCMA-2 UFHFMs mentioned above, the thermodynamics of membrane-forming system was investigated. Ternary phase diagram can be used as a measure of thermodynamic stability for the casting system. In regard of a ternary system consisting of one polymer, one solvent and one nonsolvent, Boom et al. suggested an empirical linearized cloud point (LCP) correlation [22]: ln

φ1 φ2 = b ln +a φ3 φ3

(3)

where φi is the weight fraction of the component i, subscripts 1–3 refer to the nonsolvent, solvent and polymer,

Table 2 Composition of the casting solution for membrane fabrication Solution no.

1 2 3 4 5

PANCMA-2 (wt.%)

13 9 9 13 13 a

DMSO (wt.%)

87 87 87 83 80

PVA (wt.%)

0 4 0 0 0

Value in the brackets was the absorbance of casting solution at 700 nm.

PVP (wt.%)

0 0 4 4 7

Transparencya 25 ◦ C

60 ◦ C

Transparent (0.040) Opaque (0.042) Transparent (0.034) Transparent (0.042) Transparent (0.044)

Transparent Transparent Transparent Transparent Transparent

(0.039) (0.041) (0.033) (0.040) (0.042)

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

7

Fig. 6. Cross-section morphologies of PANCMA-2 UHFMs. (a) PANCMA-2/additive = 13:0, (b) PANCMA-2/PVA = 9:4, (c) PANCMA-2/PVP = 9:4, (d) PANCMA-2/PVP = 13:4, (e) PANCMA-2/PVP = 13:7.

respectively. When only liquid–liquid demixing occurs, this function agrees well with the experimental cloud point. In this relation, only two parameters have to be determined experimentally: the intercept a and the slope b. Therefore, it is possible to complete the cloud point curve with this relation after a few cloud points which are used for the calculation of a and b are measured. Recent studies had revealed that this is a good and simple solution for the establishment of dope and coagulant composition [23,24]. Figs. 10 and 11 show the ternary phase diagram of PANCMA-2/DMSO/H2 O systems at 25 ◦ C. It can be seen from Fig. 10 that the system with PVA as additive was thermodynamically less stability than the system with PVP and/or not, because the binodal line of the system shifted to the polymer–solvent axis and the homogeneous phase region became slightly small. Increasing PVP/PANCMA-2 form 0:13 to 7:13 also shifted the bimodal line to the polymer–solvent axis, as can be seen from Fig. 11. Viscosity of the dope solution for membrane fabrication can severely hinder the exchange rate of solvent and nonsolvent during the phase inversion process, and therefore, it can be used as an important parameter to influence the diffusion

kinetics and the membrane morphology. It has been suggested increasing the viscosity of the final polymeric mixture to better control the spinning of hollow fiber membranes [9]. Table 3 shows the values of intrinsic viscosity ([η]) for PANCMA-2 measured both in neat DMSO and in DMSO containing additives. It was found that the [η] increased with the addition of PVA and/or PVP in solution. It was well recognized that membrane-forming system had small homogeneous region in a phase diagram was thermodynamically less stability and relatively easy to form membrane structure with macrovoids [28]. On the other Table 3 Intrinsic viscosity of the PANCMA-2 with additives in solution at 30 ◦ C Sample PANCMA-2 PANCMA-2/PVA PANCMA-2/PVP PANCMA-2/PVP PANCMA-2/PVP

Intrinsic viscosity (dl/g) = 9:4 = 9:4 = 13:4 = 13:7

2.20 3.68 3.17 3.25 3.38

8

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

Fig. 7. Out surface morphologies of PANCMA-2 UHFMs. (a) PANCMA-2/additive = 13:0, (b) PANCMA-2/PVA = 9:4, (c) PANCMA-2/PVP = 9:4, (d) PANCMA-2/PVP = 13:4, (e) PANCMA-2/PVP = 13:7.

hand, the increase of solution viscosity by additive addition normally inhibited the diffusion exchange between solvent and nonsolvent, resulting in delay phase inversion process, which was negative for the formation of finger-like macrovoids. The morphologies of PANCMA-2 UFHFMs with PVP as additive revealed that the viscosity increase, in turn, the decrease of diffusion exchange between solvent and nonsolvent, was the main factor to control the membrane structure. It was suggested that the overall diffusion exchange of solvent and nonsolvent during the phase inversion process could be inhibited because of high viscosity, even though the systems with additive addition were thermodynamically less stability [26,29]. The precipitation rate decreased with increasing the content of PVP in the casting solution, delayed polymer coagulation occurred, resulting a denser membrane structure. On the other hand, in the case

of PVA as additive, although the increase of viscosity for the casting solution was most obvious, but the thermodynamically less stability played a dominant rule on membrane morphologies. Furthermore, it was observed that (Table 2) PANCMA-2 solutions adding PVP were transparent, while that with PVA was opaque at room temperature. It revealed that the miscibility of PANCMA-2 and PVA was poor, while PVP could disperse well in the PANCMA-2 solution to form a homogeneous phase. Therefore, rapid precipitation (instantaneous demixing) might occur when using PVA as additive. 3.3. Surface modification of PANCMA membranes One of our objectives for this study was to use the reaction of the reactive groups on membrane surfaces with PEG and/or amino sugars to improve membrane performances

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

9

Fig. 8. Inner edge morphologies of PANCMA-2 UHFMs. (a) PANCMA-2/additive = 13:0, (b) PANCMA-2/PVP = 9:4, (c) PANCMA-2/PVP = 13:4, (d) PANCMA-2/PVP = 13:7.

2

100

BSA rejection

Water flux

180 160

95

140 120

90 100 80

85

60

BSA rejection (%)

Water flux (L/m .hr.atm)

200

40 20

80

1

3

2

Fig. 9. Water flux and BSA rejection of the PANCMA-2 UHFMs prepared with different amount of PVP. (1) No PVP, (2) 4.0 wt.% PVP, (3) 7.0 wt.% PVP.

[18–21]. To do such modification, the acids should be transformed to anhydrides because the latter show higher reactivity than the former. The hollow fiber membranes of PANCMA were treated in a 1:1 acetic anhydride/pyridine solution at 100 ◦ C for 30 min and then reacted with PEG 200 at 90 ◦ C for 12 h. Membrane surfaces were analyzed by attenuated total reflection Fourier transform infrared spectroscopy (FT-IR/ATR), pure water contact angle and BSA adsorption measurements. The typical results are shown in Fig. 12 and Table 4, respectively. Compared with the original membrane, the FT-IR/ATR spectrum of the anhydride/pyridine treated membrane (spectrum (b) in Fig. 12) PANCMA-2

PANCMA-2

0.70

0.60 0.4

0.3

0.80

0.75 0.80

0.90

0.1

0.1 0.95

0.95 1.00

0.2

0.85

0.2

0.85 0.90

7.0 wt.% 4.0 wt.% 0.0 wt.%

0.75

0.65 0.70

0.3

0.0

Solvent 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Water Fig. 10. Ternary phase diagram of PANCMA-2/DMSO/H2 O systems at 25 ◦ C. (䊐) PANCMA-2/PVP = 9:4, (䊊) PANCMA-2/PVA = 9:4, ( ) without additive.

1.00

DMSO 0.00

0.0

0.05

0.10

0.15

0.20

0.25

0.30 Water

Fig. 11. Ternary phase diagram of PANCMA-2/DMSO/H2 O systems at 25 ◦ C. (䊐) PANCMA-2/PVP = 13:7, ( ) PANCMA-2/PVP = 13:4, (䊊) without additive.

10

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

Table 4 Comparison of membrane properties

Contact angle (◦ ) Water permeation flux (l/(m2 h atm)) BSA solution permeation flux (l/(m2 h atm)) BSA solution flux decrease after BSA foulinga (%) BSA adsorption (g/m2 ) BSA rejection (%) a

Original acid membrane

Anhydride membrane

PEG 200 modified membrane

55 128.7 ± 7.2 66.9 ± 5.7 48 2.31 ± 0.12 87

65 86.5 ± 4.5 15.6 ± 3.1 82 4.02 ± 0.21 87

33 163.5 ± 8.5 114.4 ± 6.4 30 1.44 ± 0.08 86

Data were calculated from the ratios of BSA solution flux over water flux.

showed absorption peak at 1785 cm−1 corresponding to anhydride. The symmetric C=O stretching absorption band at 1785 cm−1 for anhydride group disappeared in spectrum (c) where the signals corresponding to the C=O stretching vibrations at 1725 and 1620 cm−1 developed instead. The vibration bands corresponding to the C=O stretching of the ester group located at 1579 cm−1 and corresponding to the –CH2 –OH stretching of PEG located at 1076 cm−1 appeared. Furthermore, as can be seen from Table 4, the water contact angle and BSA adsorption on membrane surface as well as the permeation flux decrease after BSA fouling of the membrane were decreased from 55◦ , 2.31 g/m2 and 48% for

-1

1579cm

(c) 1785cm

-1

(b)

Acknowledgements

(a)

3100

4. Conclusions Compared with common solution copolymerization, only poly(acrylonitrile-co-maleic acid) with high molecular weight could be obtained from the copolymerization of acrylonitrile and maleic anhydride with water-phase precipitation copolymerization process. Ultrafiltration hollow fiber membranes containing reactive carboxyl groups were fabricated from the synthesized poly(acrylonitrile-co-maleic acid)s with a dry–wet phase inversion process. Increasing the molecular weight or the casting solution concentration of PANCMA raised BSA rejection and mechanical strength but decreased water flux for the corresponding membrane. Addition additives such as poly(vinyl pyrrolidone) and poly(vinyl alcohol) to the casting solution affected the structures and properties of the membranes effectively. These could be demonstrated on the basis of ternary phase diagrams for the membrane formation systems and viscosity data for the casting solutions. Preliminary results indicated that the PANCMA membrane could be further modified by the reactive ability of the carboxyl groups. Such modifications are on going and the details will be reported in our following paper.

-1

1076cm

PANCMA membrane to 33◦ , 1.44 g/m2 and 30% for PEG modified membrane. The permeation flux of pure water and BSA solution increased from 128.7 to 163.5 l/(m2 h atm) and from 66.9 to 114.4 l/(m2 h atm), respectively, while the BSA rejection was almost kept constant. On the other hand, transforming the acid groups to anhydride ones on membrane surfaces deteriorated temporarily the membrane properties.

2700

2300

1900

1500

1100

700

-1

Wavenumber (cm ) Fig. 12. FT-IR/ATR spectra of the membrane surface. (a) Original PANCMA-2 UHFM, (b) PANCMA-2 UHFM treated with acetic anhydride/pyridine mixture, (c) mixture (b) reacted with PEG 200.

The financial supports of the National Nature Science Foundation of China (Grant no. 50273032) and the High-Tech Research and Development Program of China (Grant no. 2002AA601230) are gratefully acknowledged. We would like to thank Professor You-Yi Xu for allowing us to use the membrane fabrication equipment.

F.-Q. Nie et al. / Journal of Membrane Science 230 (2004) 1–11

References [1] A.A. Bhat, V.G. Pangarkar, Methanol-selective membranes for the pervaporative separation of methanol–toluene mixtures, J. Membr. Sci. 167 (2000) 187. [2] S.K. Ray, S.B. Sawant, J.B. Joshi, V.G. Pangarkar, Methanol selective membranes for separation of methanol–ethylene glycol mixtures by pervaporation, J. Membr. Sci. 154 (1999) 1. [3] F. Trotta, E. Drioli, C. Baggiani, D. Lacopo, Molecular imprinted polymeric membrane for naringin recognition, J. Membr. Sci. 201 (2002) 77. [4] K.W. Broadhead, P.A. Tresco, Effects of fabrication conditions on the structure and function of membranes formed from poly(acrylonitrile-co-vinylchloride), J. Membr. Sci. 147 (1998) 235. [5] B.K. Oh, W.J. Wang, Y.M. Lee, Pervaporation of pyridine–water mixture through poly(acrylonitrile-co-4-styrene sulfonic acid) membrane, J. Appl. Polym. Sci. 59 (1996) 227. [6] T. Godjevargova, V. Konsulov, A. Dimov, N. Vasileva, Behavior of glucose oxidase immobilized on ultrafiltration membranes obtained by copolymerizing acrylonitrile and N-vinylimidazol, J. Membr. Sci. 172 (2000) 279. [7] H.-G. Hicke, I. Lehmann, G. Malsch, M. Ulbricht, M. Becker, Preparation and characterization of a novel solvent-resistant and autoclavable polymer membrane, J. Membr. Sci. 198 (2002) 187. [8] H.Y. Wang, T. Kobayashi, N. Fujii, Molecular imprint membranes prepared by the phase inversion precipitation technique, Langmuir 12 (1996) 4850. [9] M.H. Shinde, S.S. Kulkarni, D.A. Musale, S.G. Joshi, Improvement of the water purification capability of poly(acrylonitrile) ultrafiltration membranes, J. Membr. Sci. 162 (1999) 9. [10] A. Wenzel, H. Yanagishita, D. Kitamoto, A. Endo, K. Haraya, T. Nakane, N. Hanai, H. Matsuda, N. Koura, H. Kamusewitz, D. Paul, Effects of preparation condition of photo-induced graft fillingpolymerized membranes on pervaporation performance, J. Membr. Sci. 179 (2000) 69. [11] D.A. Musale, S.S. Kulkarni, Relative rates of protein transmission through poly(acrylonitrile) based ultrafiltration membranes, J. Membr. Sci. 136 (1997) 13. [12] J. Pieracci, J.V. Crivello, G. Belfort, UV-assisted graft polymerization of N-vinyl-2-pyrrolidinone onto poly(ether sulfone) ultrafiltration membranes using selective UV wavelengths, Chem. Mater. 14 (2002) 256. [13] L.J. Zeman, A. Zydney, Microfiltration and Ultrafiltration: Principles and Applications, Marcel Dekker, New York, 1996. [14] H. Ma, C.N. Bowman, R.H. Davis, Membrane fouling reduction by backpulsing and surface modification, J. Membr. Sci. 173 (2000) 191. [15] J. Pieracci, J.V. Crivello, G. Belfort, Photochemical modification of 10 kDa polyether-sulfone ultrafiltration membranes for reduction of biofouling, J. Membr. Sci. 156 (1999) 223.

11

[16] A. Higuchi, K. Shirano, M. Harashima, B.O. Yoon, M. Hara, M. Hattori, K. Imamura, Chemically modified polysulfone hollow fibers with vinylpyrrolidone having improved blood compatibility, Biomaterials 23 (2002) 2659. [17] Z.-K. Xu, R.-Q. Kou, Z.-M. Liu, F.-Q. Nie, Y.-Y. Xu, Incorporating ␣-allyl glucoside into polyacrylonitrile by water-phase precipitation copolymerization to reduce protein adsorption and cell adhesion, Macromolecules 36 (2003) 2441. [18] X. Yin, H.D.H. Stöver, Thermosensitive and pH-sensitive polymers based on maleic anhydride copolymers, Macromolecules 35 (2002) 10178. [19] S.-S. Hou, P.-L. Kou, Synthesis and characterization of amphiphilic graft copolymers based on poly(styrene-co-maleic anhydride) with oligo(oxyethylene) side chains and their GPC behavior, Polymer 42 (2001) 2387. [20] I. Donati, A. Gamini, A. Vetere, C. Campa, S. Paoletti, Synthesis, characterization, and preliminary biological study of glycoconjugates of poly(styrene-co-maleic acid), Biomacromolecules 3 (2002) 805. [21] R. Auzély-Velty, M. Cristea, M. Rinaudo, Galactosylated N-vinylpyrrolidone-maleic acid copolymers: synthesis, characterization, and interaction with lectins, Biomacromolecules 3 (2002) 998. [22] R.M. Boom, T.V. van den Boomgaard, J.W.A. van den Berg, C.A. Smolders, Linearized cloupoint curve correlation for ternary systems consisting of one polymer, one solvent and one-solvent, Polymer 34 (1993) 2248. [23] L.-Q. Shen, Z.-K. Xu, Y.-Y. Xu, Phase separation behavior of poly(ether imide)/N,N-dimethyl acetamide/nonsolvent systems, J. Appl. Polym. Sci. 89 (2003) 875. [24] W. Albrecht, Th. Weigel, M. Schossig-Tiedemann, K.K. Kneifel, V. Peinemann, D. Paul, Formation of hollow fiber membranes from poly(ether imide) at wet phase inversion using binary mixtures of solvents for the preparation of the dope, J. Membr. Sci. 192 (2001) 217. [25] W.-Y. Chuang, T.-H. Younh, W.-Y. Chiu, C.-Y. Lin, The effect of polymeric additives on the structure and permeability of poly(vinyl alcohol) asymmetric membranes, Polymer 41 (2000) 5633. [26] R.M. Boom, Th. van den Boomgaad, C.A. Smolders, Mass transfer and thermo-dynamics during immersion precipitation for a two-polymer system evaluation with the system PES–PVP–NMP– water, J. Membr. Sci. 90 (1994) 231. [27] R.M. Boom, I.M. Wienk, Th. van den Boomgaad, C.A. Smolders, Equilibrium thermodynamics of a quaternary membrane-forming system with two polymers. 1. Calculations, Macromolecules 27 (1994) 2034. [28] P. van de Witte, P.J. Dijkstra, J.W.A. van den Berg, J. Feijen, Phase separation processes in polymer solutions in relation to membrane formation, J. Membr. Sci. 117 (1996) 1. [29] M.J. Han, S.T. Nam, Thermodynamic and rheological variation in polysulfone solution by PVP and its effect in the preparation of phase inversion membrane, J. Membr. Sci. 202 (2002) 55.