Surface hydrophilization of microporous polypropylene membrane by the interfacial crosslinking of polyethylenimine

Journal of Membrane Science 337 (2009) 70–80

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Surface hydrophilization of microporous polypropylene membrane by the interfacial crosslinking of polyethylenimine Yun-Feng Yang a,b , Ling-Shu Wan a,b , Zhi-Kang Xu a,b,∗ a Institute of Polymer Science, Key Laboratory of Macromolecular Synthesis and Functionalization (Ministry of Education), Zhejiang University, Hangzhou 310027, China b State Key Laboratory of Chemical Engineering, Zhejiang University, Hangzhou 310027, China

a r t i c l e

i n f o

Article history: Received 22 January 2009 Received in revised form 17 March 2009 Accepted 20 March 2009 Available online 28 March 2009 Keywords: Hydrophilicity Microporous polypropylene membrane Polyethylenimine Interfacial crosslinking Dielectric barrier discharge plasma at atmospheric pressure

a b s t r a c t To endow microporous polypropylene membrane (MPPM) with durable surface hydrophilicity, a facile interfacial crosslinking approach was developed, combined with a pretreatment by dielectric barrier discharge (DBD) plasma at atmospheric pressure. The commercially available polyethylenimine (PEI) was coated on MPPM firstly and was sequentially crosslinked with p-xylylene dichloride and quaternized with iodomethane to form a permanently positively charged layer. The physical and chemical changes of the membrane surface were characterized by tensile test, FT-IR/ATR, XPS, and FESEM. The surface hydrophilicity of the modified MPPMs was evaluated by water contact angle and pure water flux measurements. Besides, the influence of surface charge on protein filtration involving flux decline and protein transmission was also investigated in detail. It is found that the optimal time of DBD plasma treatment is ∼30 s. Mass gain for the MPPMs during the interfacial crosslinking can be controlled conveniently by adjusting the PEI concentration from 1.0 to 15 g/L. The surface hydrophilicity can be significantly enhanced and is durable, characterized by the sharp decrease of water contact angle, the double increase of pure water flux and the stability test. The results of protein filtration suggest that the obtained highly hydrophilic and charged membrane surface is resistant to protein fouling. Furthermore, almost 100% of protein transmission indicates that the microfiltration characteristic of MPPMs is unchanged. © 2009 Elsevier B.V. All rights reserved.

1. Introduction It is generally accepted that the separation characteristics of polymer membranes are chiefly determined by the surface physical and chemical properties [1]. A hydrophilic membrane is normally favored for the applications in aqueous system, because the hydrophilic surface can considerably enhance the water permeability and reduce the biofouling of membrane [2]. However, the commercial membranes are mostly manufactured from hydrophobic polymers, attributing to their excellent chemical and mechanical stability. Therefore, surface hydrophilization is reasonably necessary for these hydrophobic membranes [3–7]. Microporous polypropylene membrane (MPPM) is such a typical kind of hydrophobic membrane, which has been widely used owing to the well-controlled porosity, high stability, and low cost of the raw material polypropylene. The intrinsically high hydrophobicity fits it perfectly for membrane distillation. However, it is the hydrophobicity that severely limits its broader application in aqueous solution separation and biomedical fields. So far, various

∗ Corresponding author. Fax: +86 571 8795 1773. E-mail address: [email protected] (Z.-K. Xu). 0376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2009.03.023

surface modification methods have been developed for MPPM to improve the hydrophilicity and functionalize the membrane surface, which largely include plasma treatment [8–10] and surface grafting polymerization of hydrophilic monomers [11–20]. Particularly, the latter has been extensively explored and can be induced by free radical [11], ozone [12], ␥-ray radiation [13], plasma [14,15], UV irradiation [16–18] or surface-initiated ATRP [19,20]. Although these chemical approaches could effectively endow MPPM with durable hydrophilicity, they are mostly limited to lab scale to date because complicated production equipments, strict operation requirements and high cost are usually required. Impregnating or coating the membranes with hydrophilizing agents, such as alcohols [21], surfactants [22], or amphiphilic polymers [23] may be a convenient approach alternatively, but the stability of the resulting hydrophilicity needs improving. Interfacial polymerization/crosslinking is a mature technology to prepare stable thin film composite (TFC) membranes for reverse osmosis [24]. Using interfacial crosslinking, Du and Zhao also fabricated polysulfone composite membrane with a thin charged layer for nanofiltration [25] and gas separation [26]. While there are many reports of interfacial crosslinking, only three have addressed the application to preparation of hydrophilic TFC membrane from MPPM [27–29] due to the poor interfacial adhesion caused by

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

the hydrophobicity of MPPM. Without surface pretreatment, the coating layer was prone to the formation of balloon structure resulting from the interfacial compatibility [27]. Hence, it required great efforts to optimize the experiment condition for achieving uniform and stable coating. To improve the interfacial adhesion, Korikov et al. [28,29] developed a novel pretreatment method with acetone and chromic acid oxidizing solution, and then prepared solvent-resistant and hydrophilic nanofiltration and ultrafiltration membrane by coating an electroneutral polyamide crosslinking layer. In our previous study [30], the more environment friendly dielectric barrier discharge (DBD) plasma at atmospheric pressure [31], characterized with simple equipment (no need for vacuum condition) and easy realization of industrial production, was used for pre-treating the MPPM. Interfacial crosslinking of poly(N,N-dimethylaminoethyl methacrylate), which is self-made in our lab, was then carried out to construct a positively charged and highly hydrophilic surface on MPPM. In this work, the DBD plasma pretreatment was further optimized in view of the treatment effect and mechanical properties. The commercially available polyethylenimine (PEI) was employed alternatively. The crosslinking mechanism and the surface hydrophilicity of the modified MPPMs were well characterized. Especially, the effect of surface hydrophilization on protein filtration was investigated in detail, involving the flux decline behavior, the flux recovery and the transmission of protein through the membranes. 2. Experimental 2.1. Materials MPPM with an average pore size of 0.20 ␮m and a porosity of ∼80% was purchased from Membrana GmbH (Germany), which was prepared by the thermally induced phase separation process. All the membrane samples used were cut into rounds with a diameter of 25 mm and washed by acetone for 0.5 h to remove the impurities adsorbed on the membrane surfaces before drying in a vacuum oven at 40 ◦ C to constant weight. Polyethylenimine (PEI, average Mw 25 kDa) was commercially obtained from Aldrich and was used as received. p-Xylylene dichloride (XDC, 98%, Alfa Aesar) and iodomethane (CH3 I, 99%, Aladdin reagent Co. Ltd.) were used without further purification. Bovine serum albumin (BSA, isoelectric point (pI) 4.8, 67 kDa) and lysozyme (Lys, pI 11.0, 14.4 kDa, >10,000 U/mg) were used as received from Sino-American Biotechnology Co. and Shanghai Bio Life Science & Technology Co. Ltd., respectively. Buffer solutions were prepared from analytical-grade chemicals and ultrapure water (18.2 M) produced from an ELGA Lab Water system. Ethanol, acetone and sodium hydroxide were of analytical grade and used without further purification. 2.2. Fabrication of the membrane surfaces The entire experimental procedure is schematically illustrated in Fig. 1 (left). The DBD plasma at atmospheric pressure setup has been described in our previous study [30]. To achieve pseudoglow or glow discharge mode, only 1% of air was introduced into argon at atmospheric pressure as the discharge gas. The nascent MPPM samples were irradiated at 10 kHz and 3 kV for a given time. The treated membranes were then washed with ethanol for 30 min to remove the low molecular mass oxidized materials produced during plasma irradiation, and dried in air for further use. The interfacial crosslinking and quaternization of PEI was carried out as follows. The plasma-treated MPPM samples were dipped into 5 mL of PEI solution in ethanol with different concentrations for 30 min and then spread on PTFE slices followed by air-drying. Then, the membranes with a thin coating layer of PEI were immersed into

71

5 mL of the crosslinker XDC–acetone solution at given concentrations. The interfacial crosslinking was allowed at room temperature for 6 h. Thereafter, the membranes were further dipped into 0.5 M of CH3 I solution in ethanol (with 0.25 M of NaOH) for the quaternization of PEI, which was conducted at 25 ◦ C for 3 h. Finally, the modified membranes were washed drastically with ethanol by vibration to remove the uncrosslinked PEI. After being dried in a vacuum oven at 40 ◦ C to constant weight, the mass gain (MG, ␮g/cm2 ) was calculated according to Eq. (1). MG =

W1 − W0 Am

(1)

where W0 is the mass of the nascent membrane and W1 the mass of the modified membrane. Am represents the area of the membrane. Each result was the average of at least three parallel experiments.

2.3. Evaluation of mechanical properties after DBD plasma pretreatment The mechanical properties of the membranes after plasma pretreatment were evaluated by tensile test on an INSTRON 5543 tensile tester with a 50 N sensor. The MPPM samples were prepared in rectangular shape with a width of 8 mm and a gauge length of 30 mm. The sample thickness was around 150 ␮m and exactly measured by a spiral micrometer. The test was carried out at a crosshead speed of 2 mm/min at 20 ◦ C and the relative humidity of ∼66%. The data were analyzed by Blue Hill mechanical analyzing software, and the break strength and break elongation were calculated from the tensile curves. Each of data was the average of six measurements.

2.4. Surface characterization To evaluate the surface chemical composition and morphology before and after the modification, attenuated total reflectance Fourier transform infrared spectroscopy (FT-IR/ATR), X-ray photoelectron spectroscopy (XPS), and field emission scanning electron microscopy (FESEM) were used. FT-IR/ATR spectra were collected using Nicolet FT-IR/Nexus470 spectrometer equipped with an ATR accessory (ZnSe crystal, 45◦ ). Sixteen scans were taken for each spectrum at a nominal resolution of 2 cm−1 . FESEM (SIRION100, FEI, USA) was applied to observe the surface morphology of the membranes after sputtered with gold using ion sputter JFC-1100. XPS measurements were performed on a RBD upgraded PHI-5000C ESCA system (PerkinElmer, USA) with Al K␣ radiation (h = 1486.6 eV). In general, the X-ray anode was run at 250 W and the high voltage was kept at 14.0 kV with a detection angle at 54◦ . The base pressure of the analyzer chamber was about 5 × 10−8 Pa. A survey scan (0–1000 eV binding energy range, 93.9 eV pass energy) and a high-resolution scan of all the elements (23.5 eV pass energy) were run for each sample. Binding energies (BEs) were calibrated using the contaminated carbon (C 1s = 284.6 eV). The data were analyzed using the RBD AugerScan 3.21 software provided by RBD Enterprises. Water contact angle (WCA) was determined by the sessile drop method using a CTS-200 contact angle system (Mighty Technology Pvt. Ltd.) at room temperature. Briefly, a water drop (∼2 ␮L) was lowed onto the dry membrane surface with a microsyringe and then the WCA was recorded in equal time intervals (0.5 s) for a period of time. All the WCAs were obtained for more than five different positions on each sample.

72

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

Fig. 1. Schematic illustration of the experimental procedure (left) and corresponding XPS spectra of the MPPM surfaces at each step (right): (a) nascent MPPM; (b) plasma treatment and washing; (c) coating 5 g/L of PEI and crosslinking with XDC; (d) quaternization with CH3 I in NaOH solution.

2.5. Permeation properties A dead-end filtration system with an effective membrane area of 3.9 cm2 was used to investigate the pure water/protein solution permeation properties of the nascent and modified MPPM samples. The filtration protocol is similar to that of Shim et al. [32]. First, a

compaction step by filtering ultrapure water at 0.50 MPa was conducted for 30 min. Then, the pressure was lowered to 0.10 MPa and the flux of ultrapure water (JW ) was measured every 5 min until a constant value reached. Next, 1.0 g/L BSA solution (McIlvaine buffer solution; pH 3.0, 4.8, 7.4; ionic strength 10 mM [33] or Lys solution (phosphate buffer solution; pH 7.4, 11.0, 12.5; ionic strength 10 mM)

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

73

was driven to permeate through the membranes at 0.10 MPa for 30 min, during which the flux decline behavior was recorded and the final flux was denominated as JP . The transmission (T) of protein through the membranes during this filtration process was also investigated, which is defined as T (%) =

Cp × 100 Cf

(2)

where Cp and Cf (mg/mL) are the protein concentration of permeate and feed solution, respectively. Herein, Cf was maintained at 1.0 mg/mL, while Cp was determined at 280 nm using a UV/VIS spectrophotometer (UV-2450, Shimadzu, Japan). To further study the flux recovery properties of this proteinfiltered membrane, the reservoir was replaced with buffer solution (McIlvaine buffer solution of pH 3.0 and ionic strength 10 mM for BSA; PBS buffer solution of pH 7.4 and ionic strength 10 mM for Lys), and the cleaning filtration was performed for 30 min at 0.10 MPa. Thereafter, the reservoir was replaced with ultrapure water again to completely remove the remaining buffer solution and measure the recovery water flux (JR ). The relative flux reduction (RFR) and flux recovery ratio (FRR) were calculated by Eqs. (3) and (4): RFR(%) = FRR(%) =



1−

JP JW



× 100

JR × 100 JW

(3) (4)

2.6. Stability of the interfacial crosslinking layer Since lasting surface hydrophilization is the main objective of this work, two tests were used to estimate the durability of the interfacial crosslinked layer. One is washing the modified membranes with ethanol, water, and the mixture drastically by vibration for as long as 1 month and then measuring the mass loss with analytical balance with 0.01 mg sensitivity. The other is performing three cycles (5 h for each one) of pure water permeation and examining the change of flux, surface WCA and membrane weight. Herein, MPPM samples with a MG of ∼256 ␮g/cm2 were used and each of the data reported was the average of at least three measurements. 3. Results and discussion 3.1. Pretreatment of MPPM with DBD plasma at atmospheric pressure Surface pretreatment is usually necessary for a hydrophobic surface to be coated with a hydrophilic layer because of the difference in surface energy between them. In this study, the MPPM samples were radiated by DBD plasma at atmospheric pressure to introduce polar groups and in turn, to improve the surface adhesion [34]. In our previous work [30], it was found that the membranes were severely etched and the mechanical properties were steeply deteriorated. To alleviate the etching effect, the treatment voltage used here was decreased from 5 kV to 3 kV. In this case, the optimal treatment time was determined by weighing the enhancement of surface polarity and the etching effect. Fig. 2(a) shows the effect of plasma treatment time on the improvement of surface polarity, which is characterized by WCA. As can be seen, 30 s of treatment decreases the WCA significantly, while prolonging the time can only decrease the WCA slightly. It demonstrates the high efficiency of the DBD plasma at atmospheric pressure in our case. The mechanical properties of the treated membranes are presented in Fig. 2(b). Both the break strength and the break elongation reduce rapidly in the first 5 s and then exhibit a persistent and slow decrease. In other words, prolonged treatment will result in more serious etching. Based on the comprehensive consideration of these results, we

Fig. 2. Effect of plasma treatment time on (a) the surface water contact angle (as function of time) and (b) the mechanical properties of the MPPMs.

choose 30 s as the appropriate treatment time for the following experiments. According to Badyal and co-workers [35], such treatment generated a layer of low molecular weight oxidized materials on the polypropylene substrate, which was not firm and could be easily removed by washing with water or other polar solvents. In our case, a deposited layer of such materials was also observed (see Fig. S1 in Supplementary Material (SM)), which accumulated evidently with increasing the treatment time. To eliminate the possible adverse effect of these oxidized materials on the stability of subsequent coating layer, the treated membranes were washed before the interfacial crosslinking. The surface morphologies of the treated membranes after washing are shown in the following text. 3.2. Interfacial crosslinking and quaternization of PEI on the MPPM surface The crosslinking and quaternization of PEI with XDC and CH3 I, respectively, occurred at the solid–liquid interface between the PEI coating layer and the solutions. As a result, a crosslinked layer bearing plenty of charges is fabricated on the membrane surface. Fig. 3 shows the effect of the concentrations of crosslinker XDC and PEI on the MG of membranes after modification. It is found that the MG increases rapidly with the XDC concentration at first and then decreases slightly when the concentration exceeds 3.5 g/L (Fig. 3(a)). This result is consistent with another study, which elucidates that the over high concentration of crosslinker is not helpful to the fixation of PEI [36]. Possibly the high XDC concentration results

74

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

Fig. 4. ATR/FT-IR spectra of the MPPM surfaces after each procedure: nascent membrane (a); coating PEI (b); crosslinking with XDC (c); quaternization with CH3 I with a resulting mass gain of 178.8 (d) and 400 ␮g/cm2 (e).

Fig. 3. Effect of (a) XDC concentration ([PEI] = 5.0 g/L) and (b) PEI concentration ([XDC] = 3.5 g/L) on the mass gain of the modified MPPMs.

in the reaction stoichiometry of 1:1 between the XDC molecules and amino groups of PEI. In this case, the PEI layer is not crosslinked by XDC and the resulting coating is not solvent-resistant. Hence, the XDC solution of 3.5 g/L was used. As shown in Fig. 3(b), the MG increases almost linearly with increasing the PEI concentration from 1 to 15 g/L. It is reasonable since 3.5 g/L of XDC solution can ensure sufficient crosslinking and fixation of the PEI coating layer. By adjusting the PEI concentration, it is convenient to modulate the MG and other related properties, such as surface hydrophilicity and the permeability of water and protein solution. 3.3. Chemical and morphological changes of the membrane surfaces Fig. 4 is the ATR/FT-IR spectra of the nascent and modified MPPMs. For the pure PEI coated membrane (Fig. 4(b)), three additional absorptions are observed, which are centered at 1560, 1631 and 3300 cm−1 and attributed to the primary amine –NH2 , secondary amine –NH– bending vibration and stretching vibration of the two, respectively. After the crosslinking reaction, which is actually an alkylation of amino groups (PEI) by alkyl halides (XDC), the adsorption band of –NH2 bending vibration disappears (Fig. 4(c)). It means that –NH2 is preferentially alkylated [37], whereas whether –NH– is also reacted is not sure because its weak absorption at 1631 cm−1 is overlapped by that of the benzene rings in XDC. Besides, it can be seen from Fig. 4(d) and (e) that the spectra after quaternization does not show obvious change (compared with

Fig. 4(c)), except for a stronger peak at ∼3440 cm−1 due to the inevitable adsorption of water by the more surface charges. To further understand the mechanism of the interfacial crosslinking and quaternization, quantitative analysis of surface chemical composition was achieved by XPS. The survey scan spectra are presented in Fig. 1 (right) and the chemical composition of the membrane surface is summarized in Table 1. After plasma treatment and washing, the peak of O 1s at 533.7 eV is observed, suggesting that the plasma treatment generates oxygen-containing polar groups on the membrane surface (Fig. 1(b)). When the coated PEI was reacted with XDC, the resulting substance HCl will transfer to tertiary amino groups >N– of PEI in our case, owing to the absence of other basic matters (Fig. 1). Accordingly, it can be seen from Fig. 1(c) that both nitrogen and chlorine elements are detected from the membrane surface after interfacial crosslinking. Table 1 (sample 3) shows that the content ratio of [Cl]/[N] is 0.336. It has been reported that the content ratio of –NH2 , –NH– and >N– in PEI is about 1:2:1 and the former two, especially –NH2 would be preferentially alkylated by alkyl halides [38]. Thus it can be concluded that –NH2 are completely reacted and thus converted to –NH–, whereas only ∼8.6% (33.6–25%) of –NH– to total amino groups are alkylated Table 1 Surface chemical composition of the MPPMs calculated from XPS spectra. Membrane samples

1 2 3 4

Molar relative content (%)

Ionization ratio (%) Cl− /(Cl + Cl− )

[C]

[O]

[N]

[Cl]

[I]

100 96.08 80.11 79.27

3.83 3.43 3.23

0.09 12.32 11.92

4.14 0.12

98.9 5.46 –

N+ /(N + N+ )

36.4 43.6

MPPM surfaces after each procedure: (1) nascent membrane; (2) plasma treatment and washing; (3) coating 5 g/L of PEI and crosslinking with XDC; (4) quaternization with CH3 I in NaOH solution.

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

75

Fig. 5. Core-level XPS spectra of Cl 2p (a) and N 1s (b)–(c) of the modified MPPMs. (a) and (b): after crosslinking with XDC; (c) after quaternization with CH3 I in NaOH solution.

in the crosslinking reaction. The quaternization by CH3 I was carried out in NaOH solution, where NaOH was used to remove HCl for promoting the reaction, as verified in Fig. 1(d). Thus, stable positive charges are introduced on the N atoms, rather than unstable ones arising from protonation by HCl. The ratio of quaternization can be calculated by [I]/[N] from Table 1, sample 4. The value is not very high (∼45.8%), which is expected for the branched PEI in which a great amount of amino groups distribute very densely on the polymer backbone [39]. Furthermore, the crosslinking degree and quaternization ratio of the coating layer can be obtained from the high-resolution spectra of Cl 2p and N 1s, as depicted in Fig. 5. The ionized chlorine Cl− 2p core-level spectrum consists of the Cl− 2p3/2 and Cl− 2p1/2 doublet at the BEs of ∼197.0 eV and ∼198.8 eV due to spin–orbit-coupling split. Meanwhile the covalently bonded (C–Cl) Cl 2p core-level spectrum splits into Cl 2p3/2 (200.0 eV) and Cl 2p1/2 (201.8 eV) doublet [40]. The peak intensity ratio of 2p3/2 to 2p1/2 is 2:1. Therefore, the spectra of chlorine species can be resolved into four peaks, as illustrated in Fig. 5(a). The ionization ratio of Cl element, defined as Cl− /(Cl + Cl− ), was calculated and is listed in Table 1. The ionization ratio of 98.6% demonstrates that almost all of XDC contribute to the crosslinking since XDC is reacted by amino groups according to the stoichiometry of 1:2. After the crosslinking with XDC, some amino groups (mainly >N–) are protonated by HCl. Thus, the N 1s spectra can be resolved into N+ (R3 N+ –H Cl− ) and N (>N– and –NH–) components with the BEs of about 400.8 and 398.8 eV [41], respectively (Fig. 5(b)). The content ratio of N+ /(N + N+ ) is ∼36.4%, close to [Cl]/[N] (33.6%). Considering the nearly 100% of ionization ratio of Cl element, the crosslinking degree of PEI can be equated to N+ /(N + N+ ), which ensures the stability of the crosslinked layer. Fig. 5(c) shows the N 1s spectra after sequential quaternization, which can be separated into quaternary ammonium N+ (R4 N+ , 401.5 eV) and N (>N– and –NH–, 398.5 eV) [41]. The quaternization ratio, i.e. N+ /(N + N+ ) is 43.6%, demonstrating a sufficiently positive surface. Morphology changes for the membrane surfaces were examined by FESEM (Fig. 6 ). By comparing Fig. 6(b) with Fig. 6(a), some degree of etching can be observed, which is however slighter than that in our previous work [30]. Fig. 6(c)–(e) illustrates that the membrane surfaces are covered with a crosslinked layer gradually with increasing MG. At low MG, the thin crosslinked layer almost homogenously and conformally adheres to the membrane surfaces and the pore walls, which does not change the high porosity, relatively large pore size and the permeation flux. Predictably, the permeation flux of the modified membranes will be significantly reduced at overhigh MG, which will counteract the improvement of surface hydrophilicity. From the corresponding 10,000 × images (Fig. 6(f)–(j)), these morphologies can be observed more clearly.

3.4. Surface hydrophilicity and water permeability of the MPPMs The hydrophilization effect was carefully studied by WCA measurement. Since some WCAs of modified membranes decrease very rapidly during test, it is more appropriate to record the change of WCA with time instead of a single value of WCA. The WCA on the nascent MPPM is as high as ∼145◦ because of the high surface roughness and only shows a slight decrease with time possibly due to the evaporation of water droplet. For the plasma-treated membrane after washing, as shown in Fig. 7(b), the WCA is ∼130◦ and drops slightly within 300 s, which is very different from the situation without washing (Fig. 2(a)). This phenomenon can be attributed to the removal of the low molecular weight oxidized materials (hydrophilic) generated by the plasma treatment [35]. However, it is verified by the XPS results (Fig. 1(b)) that there are still residual polar groups covalently linked to the membrane surfaces. Besides, the membrane pore structure is altered, larger pore size and higher porosity, due to the etching effect. These two factors lead to the slight decline of WCA compared with the nascent membrane. After modification, the initial WCAs decrease obviously and the water drop soaks into membrane pores more rapidly. It indicates that a highly hydrophilic surface is obtained and the surface porous morphologies of MPPMs are maintained to a large extent. Further, pure water flux of the membranes was measured (Fig. 8). In our experiments, a compaction process at 0.50 MPa was conducted instead of ethanol prewetting to more clearly characterize the effect of surface hydrophilicity on the water permeation flux. The flux of nascent membrane is 1030 ± 120 L/m2 h and is appreciably improved by plasma treatment. The flux doubles when a crosslinked layer of just 78.8 ␮g/cm2 is formed on the membrane surface. It is also found that the water flux increases continuously with the increase of MG and reaches the maximum 2350 ± 90 L/m2 h, whereas further increase of MG causes obvious flux decrease. For a permeation process, the pore size and surface hydrophilicity of membrane are the two key parameters affecting water flux oppositely [18]. When the MG is low, the surface hydrophilicity dominates the enhancement of water flux. With the increase of MG, the membrane pores are gradually blocked, which will then become the decisive factor and thus causes the decline of water flux. However, it should be noted that the membranes are not fully wetted in the prepressuring process at 0.50 MPa. It is because not all membrane pores can be hydrophilized in this modification (especially for small ones) and the water bubble point pressure of small hydrophobic pores (<10 nm) is up to ∼29 MPa, much higher than 0.50 MPa [29]. To achieve complete prewetting, the membranes were immersed in isopropanol in advance. Subsequently, the water fluxes were re-measured to compare with that determined via prepressuring at 0.50 MPa. The experimental procedure and results are shown in Fig. S2 (see SM). It is found that the water

76

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

Fig. 6. FESEM images of the MPPMs: (a) nascent MPPM; (b) plasma-treated MPPM after washing; (c)–(e) modified MPPMs with mass gain of 78.8, 178.7, 400.6 ␮g/cm2 , respectively; (f)–(j) the 10,000× images of the corresponding MPPMs.

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

77

Fig. 6. (Continued ).

fluxes are much larger in this case. Besides, the fluxes also increase at first and then decrease with increasing MG, which is similar to the results from prepressuring at 0.5 MPa. From these results, it can be concluded that the water fluxes of the membranes with an appropriate MG have been enhanced whether the membranes were fully prewetted or not. It indicates that surface hydrophilization can help to improve the water permeability of MPPMs to some extent. 3.5. Protein filtration As demonstrated before, the interfacial crosslinking and quaternization has conferred MPPMs with plentiful surface positive charges, high hydrophilicity and large water permeation flux. Dynamic protein filtration was further conducted to evaluate the effect of surface properties on the separation performance of the modified membranes, where two proteins, acidic BSA and basic Lys were used. The pHs of buffer solutions were set to be 3.0, 4.8 and 7.4 for BSA (pI 4.8) and 7.4, 11.0 and 12.5 for Lys (pI 11.0), at which the proteins are positively charged, neutral and negatively charged, respectively. The flux decline behavior with time, the flux recovery after cleaning and the protein transmission through membranes were investigated in detail and the results are presented in Figs. 9–11, respectively. As shown in Fig. 9, for the nascent membrane, the permeate fluxes all decrease rapidly at the start of filtration (RFR > 70%, see Fig. 10) regardless of the pHs for both BSA and Lys. This may

Fig. 7. Water contact angles of the MPPMs: (a) nascent MPPM; (b) plasma-treated MPPM after washing; (c)–(h) modified MPPMs with mass gain of 78.8, 130.9, 178.7, 256.6, 400.6 or 519.3 ␮g/cm2 .

be ascribed to the adsorption and deposition of proteins onto the membrane surface due to the strong hydrophobic interaction, which is independent of the charge of proteins. However, the situation is different for the modified membranes bearing positive charges. These charges mostly arise from the quaternization of secondary and tertiary amino groups. The surface density of positive charges was determined by titration with fluorescein (Na salt). It is found that the value is ∼1016 charge units/cm2 . This high density charges will play an important role in protein filtration. More information can be obtained from Fig. S3 in SM. In addition, a small part of the residual tertiary amino groups may be protonated at low pH [39] which will slightly contribute to the surface positive charge and then the electrostatic interaction with proteins. At pH 3.0 for BSA and pH 7.4 for Lys, the modified membranes, especially with a MG of 178.8 ␮g/cm2 exhibit a slow and slighter flux decline with time. It could be attributed to the presence of strong electrostatic repulsion providing an energetic barrier for the adsorption interaction between proteins and the membrane surfaces. Thus, the BSA fouling during filtration can be alleviated to some extent [42–44]. It can be seen from Fig. 9(a) and (d) that their RFR are less than 50% at the end of filtration, indicating less fouling. Oppositely, at pH 7.4 for BSA and pH 12.5 for Lys, they are negatively charged and interact with the modified membranes by electrostatic attraction. Consequently, the protein molecules are strongly adsorbed onto both the membrane surfaces and the pore walls [45], resulting in serious narrowing and blocking of membrane pores and the sharp decline of flux initially (Fig. 9(c) and

Fig. 8. Permeate fluxes of pure water for the nascent and modified MPPMs.

78

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

Fig. 9. Permeate flux decline behavior during filtration of protein solution: (a)–(c) BSA solution at pH 3.0, 4.8 and 7.4; (d)–(f) Lys solution at pH 7.4, 11.0 and 12.5; (1)–(5) curves: the modified MPPMs with MG of 0, 78.8, 178.7, 256.6, 400.6 ␮g/cm2 , respectively.

(f)). Moreover, as for BSA, the difference in flux decline behavior between the pH 3.0 and 7.4 should also be ascribed to the easier aggregation at pH over the pI. As reported by Kelly and Zydney, at higher pH the greater ionization of the free thiol groups in proteins causes the formation of intermolecular disulfide linkages and the aggregation [46,47]. Therefore, the resulting RFR reaches 80% or more (Fig. 10(c) and (f)). It is widely accepted that the protein fouling is relatively serious at the pI owing to the fact that the lack of electrostatic protein–protein repulsion increases the possibility of aggregation [48,49]. Nevertheless, it is found from Fig. 9(b) and (e) that the flux decline behavior at the pI is not totally this case. Perhaps the resulting hydrophilic surface has the ability to resist protein fouling to some extent in the absence of electrostatic interaction [2].

After filtration, buffer solutions were used to clean these protein-filtered membranes on line, and the JR was determined (Fig. 10). It is expected that the electrostatic repulsion with the membrane surfaces will promote the removal of proteins. The results show that higher recovery fluxes are achieved for all the modified membranes compared with the nascent one. Moreover, the FRRs are correspondingly higher when electrostatic repulsion exists between the proteins and the membrane surfaces, which is consistent with the flux decline behaviors. Especially, the FRR is as high as 90% for the membrane with a MG of 178.8 ␮g/cm2 at pH 3.0 for BSA, confirming the high resistance to protein fouling induced by electrostatic repulsion. In our case, it is observed the FRR of Lys is lower than that of BSA. It may be caused by the different physicochemical properties of these two proteins, for example,

Fig. 10. Permeate flux of pure water and protein solution: (a)–(c) BSA solution at pH 3.0, 4.8 and 7.4; (d)–(f) Lys solution at pH 7.4, 11.0 and 12.5; (1)–(5) membranes: the modified MPPMs with MG of 0, 78.8, 178.7, 256.6, 400.6 ␮g/cm2 , respectively.

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

79

layer barely results in the rejection of protein. These phenomena also indicate that the microfiltration characteristic of the MPPMs is almost invariant after the formation of a crosslinked layer on the surfaces. Therefore, it is believed that these modified MPPMs can be used for protein microfiltration. 3.6. Stability of the crosslinked layer After intensive washing, the mass loss (∼0.15%) is negligible compared to the MG (per unit weight) of the crosslinked layer (∼7.1%). Moreover, the water flux only slightly decreases from ∼2260 to ∼2100 L/m2 h after three cycles of a 5 h water permeation measurement. Perhaps the decrease is partly due to the compaction effect in each cycle. Besides, the membrane weight and the surface WCA are almost unchanged. These results demonstrate the good stability of the surface crosslinked layer, i.e. durable hydrophilicity. 4. Conclusions

Fig. 11. Transmission of protein through membranes: (a) BSA at pH 3.0; (b) Lys at pH7.4; (1)–(5) curves: the modified MPPMs with MG of 0, 78.8, 178.7, 256.6, 400.6 ␮g/cm2 , respectively.

the solubility and size. In summary, the highly hydrophilic and positively charged membrane surfaces can effectively inhibit protein fouling below the pI of a protein. The MPPM used in this study has an average pore size of 0.2 ␮m, thereby belonging to a kind of microfiltration membrane. To investigate whether its microfiltration characteristic is altered by the modification, the protein transmission through the modified membranes during filtration is determined. Fig. 11 shows the typical results of BSA filtration at pH 3.0 and Lys filtration at pH 7.4 for example (more information can be obtained from Fig. S4 in SM). All curves exhibit the same tendency that the transmission is relatively low at the start of filtration because of the adsorption of proteins, and then the adsorption is saturated and the transmission approximates to 100%. For the nascent membrane, the initial transmission is 67.5–72.5% whatever the pHs and protein types, which is caused by hydrophobic interaction induced adsorption. After modification, the initial transmission through the membranes with different MGs all depend on the pH of protein solution, i.e. the electrostatic interaction between the membrane surfaces and proteins determines the adsorption. As can be seen from Fig. S3 in SM, the initial transmission of both BSA and Lys decreases continuously from electrostatic repulsion, no electrostatic interaction to electrostatic attraction. However, after the adsorption saturation in a short time, all the transmission approaches 100% even for the membranes with high MG. In other words, the electrostatic interaction between the protein and the membrane surfaces or the adsorbed protein

The durable surface hydrophilization of MPPM has been readily achieved by the interfacial crosslinking and quaternization of PEI with XDC and iodomethane after the pretreatment by DBD plasma at atmospheric pressure. The improved interfacial adhesion and sufficient crosslinking verified by XPS results are responsible for the stability of the surface coating layer. Results of protein filtration suggest that these hydrophilic and positively charged membranes can effectively resist protein fouling below the pI of proteins. Furthermore, the protein transmission through the modified membranes is almost 100%, indicating that the microfiltration characteristic has not been altered. Therefore, these membranes are believed to have great potential in separation, purification and clarifying of proteincontaining solutions, e.g. for the recovery of extracellular proteins produced via fermentation and for the removal of bacteria and viruses in the final formulation of therapeutic proteins. Besides, it has been reported that quaternized surface possesses excellent antibacterial performance [50]. Accordingly, maybe these positively charged membranes could inhibit the bacteria growing into biofilm on the surfaces, which means the modified membranes have the anti-biofouling performance. Acknowledgements Financial support from the National Natural Science Foundation of China for Distinguished Young Scholars (Grant no. 50625309) and the Zhejiang Provincial Natural Science Foundation of China (Grant no. Z406260) is gratefully acknowledged. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.memsci.2009.03.023. References [1] L.J. Zeman, A.L. Zydney, Microfiltration and Ultrafiltration: Principles and Applications, Marcel Dekker, New York, 1996. [2] D. Mockel, E. Staude, M.D. Guiver, Static protein adsorption, ultrafiltration behavior and cleanability of hydrophilized polysulfone membranes, J. Membr. Sci. 158 (1999) 63–75. [3] C.Y. Tu, Y.L. Liu, K.R. Lee, J.Y. Lai, Hydrophilic surface-grafted poly(tetrafluoroethylene) membranes using in pervaporation dehydration processes, J. Membr. Sci. 274 (2006) 47–55. [4] D.S. Wavhal, E.R. Fisher, Membrane surface modification by plasma-induced polymerization of acrylamide for improved surface properties and reduced protein fouling, Langmuir 19 (2003) 79–85. [5] G.N.B. Barona, B.J. Cha, B. Jung, Negatively charged poly(vinylidene fluoride) microfiltration membranes by sulfonation, J. Membr. Sci. 290 (2007) 46–54. [6] A. Asatekin, S. Kang, M. Elimelech, A.M. Mayes, Anti-fouling ultrafiltration membranes containing polyacrylonitrile-graft-poly (ethylene oxide) comb copolymer additives, J. Membr. Sci. 298 (2007) 136–146.

80

Y.-F. Yang et al. / Journal of Membrane Science 337 (2009) 70–80

[7] J.A. Koehler, M. Ulbricht, G. Belfort, Intermolecular forces between a protein and a hydrophilic modified polysulfone film with relevance to filtration, Langmuir 16 (2000) 10419–10427. [8] H.Y. Yu, Y. Xie, M.X. Hu, J.L. Wang, S.Y. Wang, Z.K. Xu, Surface modification of polypropylene microporous membrane to improve its antifouling property in MBR: CO2 plasma treatment, J. Membr. Sci. 254 (2005) 219–227. [9] B. Bae, B.H. Chun, D. Kim, Surface characterization of microporous polypropylene membranes modified by plasma treatment, Polymer 42 (2001) 7879–7885. [10] H. Il Kim, S.S. Kim, Plasma treatment of polypropylene and polysulfone supports for thin film composite reverse osmosis membrane, J. Membr. Sci. 286 (2006) 193–201. [11] T. Carroll, N.A. Booker, J. Meier-Haack, Polyelectrolyte-grafted microfiltration membranes to control fouling by natural organic matter in drinking water, J. Membr. Sci. 203 (2002) 3–13. [12] Y. Wang, J.H. Kim, K.H. Choo, Y.S. Lee, C.H. Lee, Hydrophilic modification of polypropylene microfiltration membranes by ozone-induced graft polymerization, J. Membr. Sci. 169 (2000) 269–276. [13] J.S. Kang, J.K. Shim, H. Huh, Y.M. Lee, Colloidal adsorption of bovine serum albumin on porous polypropylene-g-poly(2-hydroxyethyl methacrylate) membrane, Langmuir 17 (2001) 4352–4359. [14] E.Y. Choi, B. Bae, S.H. Moon, Control of the fixed charge distribution in an ionexchange membrane via diffusion and the reaction rate of the monomer, J. Phys. Chem. B 111 (2007) 6383–6390. [15] R.Q. Kou, Z.K. Xu, H.T. Deng, Z.M. Liu, P. Seta, Y.Y. Xu, Surface modification of microporous polypropylene membranes by plasma-induced graft polymerization of alpha-allyl glucoside, Langmuir 19 (2003) 6869–6875. [16] H.M. Ma, R.H. Davis, C.N. Bowman, A novel sequential photoinduced living graft polymerization, Macromolecules 33 (2000) 331–335. [17] A.H.M. Yusof, M. Ulbricht, Polypropylene-based membrane adsorbers via photo-initiated graft copolymerization: optimizing separation performance by preparation conditions, J. Membr. Sci. 311 (2008) 294–305. [18] Q. Yang, Z.K. Xu, Z.W. Dai, J.L. Wang, M. Ulbricht, Surface modification of polypropylene microporous membranes with a novel glycopolymer, Chem. Mater. 17 (2005) 3050–3058. [19] F. Yao, G.D. Fu, J.P. Zhao, E.T. Kang, K.G. Neoh, Antibacterial effect of surfacefunctionalized polypropylene hollow fiber membrane from surface-initiated atom transfer radical polymerization, J. Membr. Sci. 319 (2008) 149–157. [20] Q. Yang, J. Tian, M.X. Hu, Z.K. Xu, Construction of a comb-like glycosylated membrane surface by a combination of UV-induced graft polymerization and surface-initiated ATRP, Langmuir 23 (2007) 6684–6690. [21] H. Molisak-Tolwinska, A. Wencel, Z. Figaszewski, The effect of hydrophilization of polypropylene membranes with alcohols on their transport properties, J. Macromol. Sci., Pure Appl. Chem. 35 (1998) 857–865. [22] P.K. Kang, D.O. Shah, Filtration of nanoparticles with dimethyldioctadecylammonium bromide treated microporous polypropylene filters, Langmuir 13 (1997) 1820–1826. [23] C. Basheer, V. Suresh, R. Renu, H.K. Lee, Development and application of polymer-coated hollow fiber membrane microextraction to the determination of organochlorine pesticides in water, J. Chromatogr. A 1033 (2004) 213–220. [24] R.J. Petersen, J.E. Cadotte, Thin film composite reverse osmosis membranes, in: M.C. Porter (Ed.), Handbook of Industrial Membrane Technology, Noyes Publications, Park Ridge, New Jersey, 1990, Chapter 5. [25] R.H. Du, J.S. Zhao, Properties of poly (N,N-dimethylaminoethyl methacrylate)/polysulfone positively charged composite nanofiltration membrane, J. Membr. Sci. 239 (2004) 183–188. [26] R.H. Du, A. Chakma, X.S. Feng, Interfacially formed poly(N,N-dimethylaminoethyl methacrylate)/polysulfone composite membranes for CO2 /N2 separation, J. Membr. Sci. 290 (2007) 19–28. [27] J.M. Dickson, R.F. Childs, B.E. McCarry, D.R. Gagnon, Development of a coating technique for the internal structure of polypropylene microfiltration membranes, J. Membr. Sci. 148 (1998) 25–36.

[28] P.B. Kosaraju, K.K. Sirkar, Interfacially polymerized thin film composite membranes on microporous polypropylene supports for solvent-resistant nanofiltration, J. Membr. Sci. 321 (2008) 155–161. [29] A.P. Korikov, P.B. Kosaraju, K.K. Sirkar, Interfacially polymerized hydrophilic microporous thin film composite membranes on porous polypropylene hollow fibers and flat films, J. Membr. Sci. 279 (2006) 588–600. [30] Y.F. Yang, L.S. Wan, Z.K. Xu, Surface hydrophilization for polypropylene microporous membranes: a facile interfacial crosslinking approach, J. Membr. Sci. 326 (2009) 372–381. [31] U. Kogelschatz, Dielectric-barrier discharges: their history, discharge physics, and industrial applications, Plasma Chem. Plasma P 23 (2003) 1–46. [32] J.K. Shim, H.S. Na, Y.M. Lee, H. Huh, Y.C. Nho, Surface modification of polypropylene membranes by gamma-ray induced graft copolymerization and their solute permeation characteristics, J. Membr. Sci. 190 (2001) 215–226. [33] P.J. Elving, J.M. Markowitz, I. Rosenthal, Preparation of buffer systems of constant ionic strength, Anal. Chem. 27 (1956) 1179–1180. [34] N.Y. Cui, N.M.D. Brown, Modification of the surface properties of a polypropylene (PP) film using an air dielectric barrier discharge plasma, Appl. Surf. Sci. 189 (2002) 31–38. [35] R.D. Boyd, A.M. Kenwright, J.P.S. Badyal, D. Briggs, Atmospheric nonequilibrium plasma treatment of biaxially oriented polypropylene, Macromolecules 30 (1997) 5429–5436. [36] R.F. Childs, J.F. Weng, M. Kim, J.M. Dickson, Formation of pore-filled microfiltration membranes using a combination of modified interfacial polymerization and grafting, J. Polym. Sci. Polym. Chem. 40 (2002) 242–250. [37] G. Noding, W. Heitz, Amphiphilic poly(ethyleneimine) bases on long-chain alkyl bromides, Macromol. Chem. Phys. 199 (1998) 1637–1644. [38] T.W. Johnson, I.M. Klotz, Preparation and characterization of some derivatives of poly(ethylenimine), Macromolecules 7 (1974) 149–153. [39] J. Yang, J.L. Welby, M.E. Meyerhoff, Generic nitric oxide (NO) generating surface by immobilizing organoselenium species via layer-by-layer assembly, Langmuir 24 (2008) 10265–10272. [40] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers, John Wiley, Chichester, England, 1992, pp. 208–209. [41] A. Haimov, H. Cohen, R. Neumann, Alkylated polyethyleneimine/ polyoxometalate synzymes as catalysts for the oxidation of hydrophobic substrates in water with hydrogen peroxide, J. Am. Chem. Soc. 126 (2004) 11762–11763. [42] I. Gancarz, G. Pozniak, M. Bryjak, Modification of polysulfone membranes 1. CO2 plasma treatment, Eur. Polym. J. 35 (1999) 1419–1428. [43] Z.P. Zhao, Z. Wang, S.C. Wang, Formation, charged characteristic and BSA adsorption behavior of carboxymethyl chitosan/PES composite MF membrane, J. Membr. Sci. 217 (2003) 151–158. [44] Z.F. Fan, Z. Wang, M.R. Duan, J.X. Wang, S.C. Wang, Preparation and characterization of polyaniline/polysulfone nanocomposite ultrafiltration membrane, J. Membr. Sci. 310 (2008) 402–408. [45] D.A. Musale, S.S. Kulkarni, Fouling reduction in poly(acrylonitrile-coacrylamide) ultrafiltration membranes, J. Membr. Sci. 111 (1996) 49–55. [46] S.T. Kelly, A.L. Zydney, Effects of intermolecular thiol-disulfide interchange reaction on BSA fouling during microfiltration, Biotechnol. Bioeng. 44 (1994) 972–982. [47] S.T. Kelly, A.L. Zydney, Mechanisms for BSA fouling during microfiltration, J. Membr. Sci. 107 (1995) 115–127. [48] W.S. Ang, M. Elimelech, Protein (BSA) fouling of reverse osmosis membranes: implications for wastewater reclamation, J. Membr. Sci. 296 (2007) 83–92. [49] X.H. Sun, et al., Characterization and theoretical analysis of protein fouling of cellulose acetate membrane during constant flux dead-end microfiltration, J. Membr. Sci. 320 (2008) 372–380. [50] J.C. Tiller, C.J. Liao, K. Lewis, A.M. Klibanov, Designing surfaces that kill bacteria on contact, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 5981–5985.