Langmuir 2007, 23, 5779-5786
5779
Improving Hydrophilicity and Protein Resistance of Poly(vinylidene fluoride) Membranes by Blending with Amphiphilic Hyperbranched-Star Polymer Yong-Hong Zhao, Bao-Ku Zhu,* Li Kong, and You-Yi Xu Institute of Polymer Science, Key Laboratory of Macromolecule Synthesis and Functionalization (Ministry of Education), Zhejiang UniVersity, Hangzhou 310027, People’s Republic of China ReceiVed January 17, 2007. In Final Form: February 26, 2007 To endow hydrophobic poly(vinylidene fluoride) (PVDF) membranes with reliable hydrophilicity and protein resistance, an amphiphilic hyperbranched-star polymer (HPE-g-MPEG) with about 12 hydrophilic arms in each molecule was synthesized by grafting methoxy poly(ethylene glycol) (MPEG) to the hyperbranched polyester (HPE) molecule using terephthaloyl chloride (TPC) as the coupling agent and blended with PVDF to fabricate porous membranes via phase inversion process. The chemical composition changes of the membrane surface were confirmed by X-ray photoelectron spectroscopy (XPS), and the membrane morphologies were measured by scanning electron microscopy (SEM). Water contact angle, static protein adsorption, and filtration experiments were used to evaluate the hydrophilicity and anti-fouling properties of the membranes. It was found that MPEG segments of HPE-g-MPEG enriched at the membrane surface substantially, while the water contact angle decreased as low as 49° for the membrane with a HPE-g-MPEG/PVDF ratio of 3/10. More importantly, the water contact angle of the blend membrane changed little after being leached continuously in water at 60 °C for 30 days, indicating a quite stable presence of HPE-g-MPEG in the blend membranes. Furthermore, the blend membranes showed lower static protein adsorption, higher water and protein solution fluxes, and better water flux recovery after cleaning than the pure PVDF membrane.
Introduction Membrane fouling is caused by undesired interactionss typically colloids (e.g., proteins or oil droplets in water)swith the membrane materials.1-3 The consequence is a sharp decline in permeate flux and changing solute selectivity with operation time, which has detrimental effects on the efficiency and economics of the membrane process.4 To improve the anti-fouling ability of the membrane, different strategies have been carried out to impart surface hydrophilicity to conventional hydrophobic membrane materials. Basically, these methods include blending with hydrophilic polymers,5,6 surface modification by grafting hydrophilic monomers,7-11 surface coating,12-15 etc. These methods have different main purpose of their own. Both grafting and coating can result in a highly hydrophilic surface, and the latter can also improve the separation selectivity of membranes.
By comparison, blending not only enhances the hydrophilicity, it also can control the membrane structures. As for the blending method, extensive work focusing on the blending of conventional hydrophobic membrane materials with amphiphilic polymers has been reported. These amphiphilic polymers basically include tri-block,16,17 comb,18-20 and branched21-23 polymers. Usually, the hydrophilic component in amphiphilic polymers is poly(ethylene glycol) (PEG) since PEG is well-known for its extraordinary ability to resist protein adsorption, which is thought to arise from its hydrophilicity, large excluded volume, and unique coordination with surrounding water molecules in aqueous solutions.24-27 Other factors that affect the protein resistance ability involve PEG density, chain length, conformation, and charge at the membrane surface.28-31 Therefore, it is reasonable to infer that the anti-fouling abilities of membranes can be improved substantially by increasing the PEG density at membrane surface.
* Corresponding author: Email:
[email protected]. (2) Kim, K. J.; Fane, A. G.; Fell, C. J. D.; Joy, D. C. J. Membr. Sci. 1992, 68, 79. (3) Belfort, G.; Davis, R. H.; Zydney, A. L. J. Membr. Sci. 1994, 96, 1. (4) Chan, R.; Chen, V. J. Membr. Sci. 2004, 242, 169. (5) Huisman, I.; Pra´danos, P.; Herna´ndez, A. J. Membr. Sci. 2000, 179, 79. (6) Nunes, S. P.; Peinemann, K. V. J. Membr. Sci. 1992, 73, 25. (7) Ochoa, N. A.; Masuelli, M.; Marchese, J. J. Membr. Sci. 2003, 226, 203. (8) Singh, N.; Husson, S. M.; Zdyrko, B.; Luzinov, I. J. Membr. Sci. 2005, 262, 81. (9) Nie, F.-Q.; Xu, Z.-K.; Huang, X.-J.; Ye, P.; Wu, J. Langmuir 2003, 19, 9889. (10) Wavhal, D. S.; Fisher, E. R. Langmuir 2003, 19, 79. (11) Ito, Y.; Ochiai, Y.; Park, Y. S.; Imanishi, Y. J. Am. Chem. Soc. 1997, 119, 1619. (12) Ulbricht, M.; Matuschewski, H.; Oechel, A.; Hicke, H.-G. J. Membr. Sci. 1996, 115, 31. (13) Nunes, S. P.; Sforqa, M. L.; Peinemann, K. V. J. Membr. Sci. 1995, 106, 49. (14) Kim, K. J.; Fane, A. G.; Fell, C. J. D. Desalination 1988, 70, 229. (15) Asatekin, A.; Menniti, A.; Kang, S.; Elimelech, M.; Morgenroth, E.; Mayes, A. M. J. Membr. Sci. 2006, 285, 81. (16) Hyun, J.; Jang, H.; Kim, K.; Na, K.; Tak, T. J. Membr. Sci. 2006, 282, 52.
(17) Hancock, L. F. J. App. Polym. Sci. 1997, 66, 1353. (18) Wang, Y.-Q.; Wang, T.; Su, Y.-L.; Peng, F.-B.; Wu, H.; Jiang, Z.-Y. Langmuir 2005, 21, 11856. (19) Hester, J. F.; Banerjee, P.; Mayes, A. M. Macromolecules 1999, 32, 1643. (20) Hester, J. F.; Banerjee, P.; Won, Y.-Y.; Akthakul, A.; Acar, M. H.; Mayes, A. M. Macromolecules 2002, 35, 7652. (21) Park, J. Y.; Acar, M. H.; Akthakul, A.; Kuhlman, W.; Mayes, A. M. Biomaterials 2006, 27, 856. (22) Siegers, C.; Biesalski, M.; Haag, R. Chem. Eur. J. 2004, 10, 2831. (23) Irvine, D.; Mayes, A. M.; Griffith-Cima, L. Macromolecules 1996, 29, 6037. (24) Wang, Y.-Q.; Su, Y.-Q.; Sun, Q.; Ma, X.-L.; Wu, H.; Jiang, Z.-Y. J. Membr. Sci. 2006, 286, 228. (25) Elbert, D. L.; Hubbell, J. A. Ann. ReV. Mater. Sci. 1996, 26, 365. (26) Harris, J. M., Ed. In Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications; Plenum Press: New York, 1992; Chapter 1. (27) Lee, J. H.; Lee, H. B.; Andrade, J. D. Prog. Polym. Sci. 1995, 20, 1043. (28) Kjellander, R.; Florin, E. J. Chem. Soc., Faraday Trans. 1 1981, 77, 2053. (29) Unsworth, L.; Sheardown, H.; Brash, J. Langmuir 2005, 21, 1036. (30) Feng, W.; Brash, J.; Zhu, S. Biomaterials 2006, 27, 847. (31) Szleifer, I. Biophys. J. 1997, 72, 595. (32) Roger, M.; Stephanie, P.; Marcus, T.; Castner, D. Langmuir 2005, 21, 12327.
10.1021/la070139o CCC: $37.00 © 2007 American Chemical Society Published on Web 04/05/2007
5780 Langmuir, Vol. 23, No. 10, 2007
Zhao et al.
Scheme 1. Synthesis of the Amphiphilic Hyperbranched-Star Polymer.
Table 1. Characteristics of Hyperbranched Polyester Boltorn H4036 theoretical core/ monomer ratio
Mtheora (g‚mol-1)
M h wb (g‚mol-1)
M hn (g‚mol-1)
M h w/M hn
1/60
7316
5100
2833
1.8
hydroxyl number (mg KOH‚g-1) 470-500,c 481d
aM theor. is a dendrimer-equivalent molar mass in which all the repeat units are attached to the core molecule. b M h w was determined by size exclusion chromatography (SEC) in 0.2% LiBr/DMF using linear PEG standards. c Data from Perstorp Specialty Chemicals AB. d Remeasured.
Hyperbranched polymers, which possess a highly branched structure and a large number of terminal functional groups,32 can be easily modified by the reaction of the terminal groups. Amphiphilic hyperbranched-star polymers have been successfully synthesized by using hyperbranched polymer as a macro-initiator or by grafting polymer chains to the terminal groups.33,34 Selection of a hyperbranched architecture with a large number of PEG arms as the membrane modifier might be expected to improve the anti-fouling ability of membranes more effectively, since it will result in a high density of PEG chains for each modifier molecule near the membrane surface and prevent bulk crystallization of PEG. Adopting terephthaloyl chloride (TPC) as the coupling agent, an amphiphilic hyperbranched-star polymer was synthesized by grafting methoxy poly(ethylene glycol) (MPEG) to a commercially available hydroxyl-terminated aliphatic hyperbranched polyester (HPE) of the fourth pseudo-generation, which was polymerized using 2,2-bis(methylol)propionic acid (bis-MPA) as the trifunctional AB2 monomer and ethoxylated pentaerythritol (PP50) as the tetra-functional B4 core molecule.35 Via a typical phase inversion route, the synthesized amphiphilic hyperbranched-star polymer was blended with poly(vinylidene fluoride) (PVDF) to fabricate porous membranes. By measuring the surface compositions, water contact angle, static protein absorption, water and foulant solution permeation, etc. of the blended membranes, the efficiency and stability of the synthesized amphiphilic hyperbranched-star polymer in improving the hydrophilicity and protein resistance of PVDF membranes were investigated. Experimental Section Materials. PVDF (FR-904) (M h n ) 380 000) was obtained from Shanghai 3F New Materials Co. Ltd. and was dried at 110 °C for 12 h before use. HPE (Boltorn H40) was purchased from Perstorp (33) Gao, C.; Yan, D. Prog. Polym. Sci. 2004, 29, 183. (34) Weberskirch, R.; Hettich, R.; Nuyken, O.; Schmaljohann, D.; Voit, B. Macromol. Chem. Phys. 1999, 200, 863. (35) Zhai, X.; Peleshanko, S.; Klimenko, N. S.; Genson, K. L.; Vaknin, D.; Vortman, M. Ya.; Shevchenko, V. V.; Tsukruk, V. V. Macromolecules 2003, 36, 3101. (36) Magnusson, H.; Malmstro¨m, E.; Hult, A. Macromolecules 2000, 33, 3099. (37) Certificate of Boltorn H40 Analysis. Perstorp Specialty Chemicals AB, 2001.
Specialty Chemicals AB, Sweden, and dried at 40 °C for 12 h under vacuum before use. The characteristics of HPE are presented in Table 1. The hydroxyl value of HPE was remeasured by the following method: the HPE was dissolved in a mixture of acetic anhydride and pyridine (1/3, v/v) and refluxed at 60 °C for 1 h. Then backtitration was performed with a KOH solution (0.5 M, in deionized water, standardized by potassium hydrogen tartrate, 99%, Acros), and the hydroxyl value was calculated. The remeasured hydroxyl value was used for the calculations in this work. Terephthaloyl chloride (TPC) was used as received from Aldrich. Methoxy poly(ethylene glycol) (MPEG, M h n ) 750) obtained from Acros was dried in vacuo at 60 °C for 48 h before it was used. N,NDimethylacetamide (DMAc, reagent grade) was carefully dried. Toluene and ethanol are reagent grade. Bovine serum albumin (BSA, M h n ) 67 000) was purchased from Aldrich. The phosphate-buffered saline (PBS, 0.01 M, pH 7.4) was prepared by the addition of prepackaged buffer salts (Aldrich) to deionized water. Synthesis of Amphiphilic Hyperbranched-Star Polymer (HPEg-MPEG). The typical synthesis process is shown in Scheme 1. First, 4.5 g (0.022 mol) of terephthaloyl chloride (TPC) was dissolved in 50 mL of dry DMAc in a thoroughly dried 250 mL three-necked flask equipped with a magnetic stirrer under an argon atmosphere. Then a solution of 16.5 g (0.022 mol) dried methoxy poly(ethylene glycol) (MPEG) in 50 mL of dry DMAc was added at a rate of 20 mL‚h-1 into the flask at room temperature. After the completion of the reaction, 2.5 g (0.022 mol of OH-) hyperbranched polyester (HPE) was added to the reaction mixture, and the reaction was kept for another 4 h at room temperature. The amphiphilic hyperbranchedstar polymer HPE-g-MPEG was gradually precipitated out in a cold mixture (-10 °C, 1000 mL) of toluene and ethanol (40/60, w/w), and purified by dialysis using a benzoylated cellulose membrane (Sigma). The residue was dried for 24 h in vacuum (-10 KPa) at 60 °C before it was characterized and used for membrane preparation. 1H NMR spectra of HPE and HPE-g-MPEG were recorded on a NMR spectrometer (Varian VXR-400) in DMSO-d6 solutions at 25 °C with TMS as internal standard. The molecular mass of HPE and HPE-g-MPEG were measured by gel permeation chromatography (GPC, Waters 2690) in dimethylformamide (THF) solution (5 mg/ mL) using PEGs as the standards. Preparation of Membranes. Casting solutions were prepared by dissolving PVDF and HPE-g-MPEG in DMAc. PVDF (10 wt % in casting solutions) is a membrane matrix, and HPE-g-MPEG (1, 2, 3 wt % in casting solutions, respectively) is a membrane modifier. After releasing the bubbles in reduced pressure (-10 KPa) for 30 min, the homogeneous casting solutions were spread on a glass plate with a steel knife at a wet thickness of 160 µm. Then the solution films were immediately immersed in a coagulation bath of deionized water. The formed membranes were peeled off and subsequently washed thoroughly with deionized water to remove residual solvent and dried for 24 h at room temperature before characterization. Characterization of Blend Membranes. X-ray photoelectron spectroscopy (XPS, PHI 5000C ESCA system, PHI Co., USA) was used to investigate the chemical compositions of the blend membrane surfaces. Al KR radiation (1486.6 eV) was used as phonon source and run at a power of 250 W (14.0 kV, 93.9 eV) with an electron takeoff angle of 90° relative to the sample plane. Survey spectra were selected over a range of 0-1100 eV, and core-level spectra of C1s were also collected. To compensate for surface charging effect,
Protein Resistance of PVDF Membranes
Langmuir, Vol. 23, No. 10, 2007 5781
all survey and core-level spectra were referenced to the C1s hydrocarbon peak at 285.0 eV. The surface and cross-section (fractured in liquid nitrogen) morphologies of prepared membranes were inspected by scanning electron microscope (SEM) using a Sirion FE-SEM (FEI, USA) after being coated with gold. The wetting behaviors of pure and blend membranes were investigated using a contact angle system (OCA20, Dataphysics, Germany) equipped with video capture at room temperature. Five measurements were done for each sample to get the average value. BSA Adsorption on the Membrane Surface. BSA was used as a model protein to evaluate the protein-resistant characteristics of the prepared membranes. The studied membranes with about 80 cm2 of external surface area (including top and bottom surfaces) were immersed in ethanol for 10 min followed by PBS solution for 30 min to pre-wet the membrane surface. Then each sample was put into a tube containing 10 mL of BSA solution with various concentrations whose pH was adjusted to 7.4 with phosphate buffer solution. The mixture was incubated at 30 °C for 24 h with a shaking speed of 150 rpm to reach adsorption equilibrium. The amount of the adsorbed protein was determined by measuring the difference between the concentration of albumin in the solution before and after adsorption indicated from the absorbance intensity at 280 nm recorded on a UV-vis spectrometer (HP 8453, USA).37 The reported data were the mean value of triplicate samples for each membrane. Filtration Experiment. Filtration experiments were conducted on 50 mm diameter membranes using a stirred, dead-end filtration cell having an effective filtration area of 19.6 cm2. Under the pressure of 100 KPa and the feed temperature at 25 ( 0.1 °C, the flux of pure water (JW) was obtained from the volume of the permeated water within 1 h. After that, a 1 g/L BSA PBS saline solution was forced to permeate through the membrane at the same pressure for 1 h, and the flux was recorded as JP. The BSA rejection ratio was calculated by the following equation:
(
RBSA (%) ) 1 permeate CBSA
)
permeate CBSA feed CBSA
× 100
feed CBSA
where and are the BSA concentrations in permeate and feed solutions, respectively, measured by UV-vis spectrometer immediately following the 1 h filtration experiment. To confirm the water recovery property of these BSA-permeated membranes, pure water flux (JR) was measured again after cleaning by washing with deionized water for 2 h. The relative flux reduction (RFR) and the flux recovery ratio (FRR) were calculated as follows:
Figure 1. 1H NMR spectra of (a) hyperbranched polyester (Boltorn H40); (b) HPE-g-MPEG in DMSO-d6. Table 2. Molecular Weight Measured by GPC no. of M h n (g‚mol-1) M h w (g‚mol-1) M h w/M h n MPEG arms HPEa HPEb HPEc HPE-g-MPEGc a
2833 2580 2096 13530
5100 6640 4821 39240
1.8 2.57 2.3 2.9
12 ( 1
b
Data from Perstorp Specialty Chemicals AB. Determined by size exclusion chromatography coupled to a multi-angle laser light scattering photometer (SEC-MALS) in a 0.7% LiBr/DMAc solution.40 c Experimental data.
Characterization of Synthesized HPE-g-MPEG. Figure 1a shows the 1H NMR spectrum of hyperbranched polyester (HPE, Boltorn H40). The protons of the methyl group in the terminal, linear, and dendritic repeat units resonate at 1.02, 1.07, and 1.17 ppm, respectively.38 The methylene groups in -CH2OR resonate at approximately 4.2 ppm, whereas the methylene groups in -CH2OH resonate at 3.5 ppm. The signals situated at 4.61 and 4.93 ppm are ascribed to the hydroxyl groups of the terminal and linear repeat units, respectively.39 Figure 1b shows the 1H NMR spectrum of the synthesized HPE-g-MPEG, in which several new signals appear except for the signals attributed to HPE. The chemical shifts at 3.24, 3.54, and approximately 7.90 ppm should
be ascribed to the CH3O-, -CH2O- groups in the grafted MPEG arms, and benzene protons in TPC, respectively. The appearances of the signals situated at 4.61 and 4.93 ppm attributed to the hydroxyl groups of the terminal and linear repeat units in HPE indicate that only part of the hydroxyl groups in HPE are grafted with MPEG arms. The molecular weights measured by GPC are presented in Table 2. The results are compared with the data from Perstorp Specialty Chemicals AB and the results given in the literature.40 It can be seen that all the M h w values are relatively close to the theoretical molar masses, Mtheor. (7316), while the M h n values are considerably lower. Moreover, the GPC results are smaller than those measured by size exclusion chromatography (SEC) and size exclusion chromatography coupled to a multi-angle laser light scattering photometer (SEC-MALS). Due to the fact that the hydrodynamic radius of the hyperbranched polymers in solution is much smaller than that of the linear polymers, the relationship between the hydrodynamic volume and molecular mass of the hyperbranched polymers is different from that of linear polymers.41 As for the discrepancy in molar mass tested by GPC and SEC, it might be mainly induced by the different solvents used in the tests. THF was used in GPC test in this work, while a 0.7% LiBr/DMAc solution was used in SEC measurement in the reference.40 Since the solvation ability of THF is not as
(38) Bradford, M. M. Anal. Biochem. 1976, 72, 248. (39) Komber, H.; Ziemer, A.; Voit, B. Macromolecules 2002, 35, 3514. (40) Ihre, H.; Hult, A.; Fre´chet, J. M. J. Macromolecules 1998, 31, 4061.
(41) Zˇ agar, E.; Zˇ igon, M.; Podzimek, S. Polymer 2006, 47, 166. (42) Mock, A.; Burgath, A.; Hanselmann, R.; Frey, H. Macromolecules 2001, 34, 7692.
( )
RFR (%) ) 1 -
FRR (%) )
JP × 100 JW
JR × 100 JW
Results and Discussion
5782 Langmuir, Vol. 23, No. 10, 2007
Zhao et al.
Figure 2. Survey XPS spectra (a) and core-level XPS spectra of membranes with the HPE-g-MPEG/PVDF ratio of (b) 1/10, (c) 2/10, and (d) 3/10, respectively. Table 3. C1s Component Peak Areas as Percentages of Total Area membrane ID
C-H
C-COO
C-H (PVDF)
C-O (MPEG)
COO-C
COO
C-F
M1 M2 M3
8.28 8.62 9.49
3.39 3.74 3.81
36.42 34.15 31.16
8.71 11.86 16.76
3.39 3.74 3.81
3.39 3.74 3.81
36.42 34.15 31.16
strong as the LiBr/DMAc solution, the HPE molecule in THF might not be able to extend as adequately as in LiBr/DMAc solution, resulting in smaller molecule volume in THF and, consequently, the smaller molar mass. In Zˇ agar et al.’s work,40 they confirmed that the M h n values could better reflect the real molar masses of HPE than M h w values. Thus, M h n values measured by GPC are adopted to calculate the number of MPEG arms of the synthesized HPE-g-MPEG in this work. The calculation is given as follows:
no. of arms )
M h n(HPE-g-MPEG) - M h n(HPE) M h n(MPEG) + M h n(TPC)
where M h n(HPE-g-MPEG), M h n(HPE), M h n(MPEG), and M h n(TPC) are the number average molecular weights of HPE-g-MPEG, HPE, MPEG, and TPC, respectively. Calculations revealed that about 12 MPEG arms are grafted to the HPE molecule, much lower than the theoretical number of hydroxyl groups in each original HPE molecule (e.g., 64). Such low graft degree might be caused by two factors. First, it is difficult for the hydroxyl groups in the interior of the HPE molecule to react with TPCMPEG. Second, part of the peripheral hydroxyl groups of the HPE molecule should be shielded by the grafted MPEG chains and not able to take part in the esterification reactions. Surface Composition of Blend Membranes. The surface compositions of PVDF/HPE-g-MPEG blend membranes were characterized by XPS analysis. Figure 2 shows the survey and core-level spectra of PVDF/HPE-g-MPEG blend membranes with different HPG-g-MPEG content. In the survey spectra (Figure
2a), three major emission peaks can be observed at 285.8 eV for C1s, at 532.1 eV for O1s, and at 689.4 eV for F1s, respectively. The simultaneous increase of the intensity of O1s peak with the HPE-g-MPEG/PVDF ratio suggests the trend of increasing HPEg-MPEG content at the membrane surfaces. To further investigate the chemical compositions of the blend membrane surfaces, the C1s regions of the XPS spectra of the membranes with different HPE-g-MPEG content were fit with seven component peaks representing different chemical environments (Figure 2b-d). The peak centers of the component peaks, referenced to the hydrocarbon peak at 285.0 eV, were constrained as follows: C-H (PVDF), 285.80 eV; C-COO, 285.82 eV; C-O (MPEG), 286.45 eV; COO-C, 286.63 eV; COO, 289.00 eV; and C-F, 290.50 eV.42 The areas of C-H and C-F peaks of PVDF were constrained to be equal, as required by stoichiometry, as well the C-COO, COO-C, and COO peaks of the hyperbranched polyester environments. The percentages of each carbon component calculated from peak area for the membranes are given in Table 3. It can be clearly seen that the content of carbon in C-O (MPEG) increases substantially with the increase of HPE-g-MPEG/PVDF ratio. On the other hand, only a slight increase was observed for the carbon in COO, which belongs to the hydrophobic HPE core, and the contents of carbon in C-O (MPEG) are far larger than those in COO units. Based on the above results, a structure model demonstrating the state of HPE-g-MPEG molecules at the membrane surface is proposed in Figure 3. The formation of this structure should (43) Beamson, B.; Briggs, D. High-Resolution XPS of Organic Polymers; John Wiley & Sons: Chichester, UK, 1992.
Protein Resistance of PVDF Membranes
Langmuir, Vol. 23, No. 10, 2007 5783
Figure 3. Deformation of the HPE-g-MPEG molecule during the immersion precipitation process and the sketch of molecular conformation in the final membrane. Table 4. Comparison of OMPEG/F Atomic Ratio between Theoretical Values and XPS Analysis membrane ID
OMPEG/F atomic ratio (theor.)a
OMPEG/F atomic ratio (XPS)b
M1 M2 M3
0.048 0.096 0.144
0.120 0.174 0.269
a Calculated based on the HPE-g-MPEG/PVDF ratio and the GPC results. b Obtained directly from the ratio of the C-O and C-F peak areas: AC-O/2AC-F.
be reasoned as follows. During the immersion precipitation process, the hydrophilic arms (MPEG) of HPE-g-MPEG are prone to migrate to the membrane surface to minimize the interfacial free energy at the water-PVDF interface.43-45 The migrating of the arms leads to the deformation of the HPE-gMPEG molecule and, consequently, the exposure of the hydrophobic core to the outside. The essential force of such tendency lies in the stronger interaction between PEG chains and waters in coagulation bath. At the same time, the tanglement between hydrophobic core (HPE) and hydrophobic PVDF chains prohibits the HPE-g-MPEG molecule migrating out of the membranes. As a result, the hydrophobic core (HPE) anchors HPE-g-MPEG at the membrane surface, and the hydrophilic arms (MPEG) stretches out of the membranes to form a hydrated polymer layer. A similar structure at an air-water interface has been reported by Zhai et al.34 They synthesized the second generation of HPEs with a variable composition of alkylterminated groups. At an air-water interface, the increase number of attached alkyl tails forces the HPE core to flatten on the surface and submerge in the subphase to allow the alkyl tails to organize at the surface. Also, as has been recently shown, a hyperbranched molecule based on a dendritic polyester core can expand in favorable solvent conditions to nearly twice its original size, demonstrating significant conformational flexibility of hyperbranched cores, which can facilitate structural changes discussed here.46 To further understand the chemical compositions of the blend membranes, the atomic ratios of OMPEG/F were compared between the theoretical values and the XPS analysis results (Table 4). As expected, the OMPEG/F atomic ratio at the membrane surface increases with the increase of HPE-g-MPEG/PVDF ratio, suggesting the increasing density of MPEG chains at the membrane surfaces. More notably, all the OMPEG/F atomic ratios at the membrane surface are higher than those of the corresponding theoretical value, indicating substantial enrichment of the MPEG segments of the HPE-g-MPEG at the membrane surfaces. These (44) Yuan, Y.; Schoichet, M. S. Macromolecules 2000, 33, 4926. (45) Baradie, B.; Schoichet, M. S. Macromolecules 2003, 36, 2343. (46) Chen, N.; Hong, L. Polymer 2002, 43, 1429. (47) Mackay, M. E.; Carmezini, G. Chem. Mater. 2002, 14, 819.
Figure 4. Time dependence of water contact angle of the membranes with different HPE-g-MPEG contents.
results further confirm the reliability of the model proposed in Figure 3. Reasonably, it can be concluded that the MPEG arms stretching out of the membrane surface contribute an effective hydrophilic layer, which will benefit the filtration performance of the membranes in this work. Hydrophilicity of the Blend Membrane. Surface hydrophilicity is one of the most important factors of filtration membrane, and much attention has been paid to it.47 Usually, water contact angle measurements are the most convenient way to assess the hydrophilicity and wetting characteristics of polymer surface. However, such measurements are difficult to interpret for synthetic porous membranes because of capillary forces within pores, contraction in the dry state, heterogeneity, roughness, and restructuring of the surfaces.48,49 Nevertheless, the relative hydrophilicity or hydrophobicity of each sample can be easily obtained by this measurement. Static water contact angles as a function of contact time on the pure and blend membranes are shown in Figure 4. Pure PVDF membrane has the highest initial water contact angle (92°), while the initial water contact angle of the blend membrane with a HPE-g-PEG/PVDF ratio of 3/10 decreases to 49°, indicating a hydrophilic surface. At the meantime, the contact angles on the pure PVDF membrane almost remain stable with the prolongation of the contact time. However, the contact angles on the HPE-g-PEG/PVDF blend membranes reduce gradually within the measurement time. With the increase of HPE-g-PEG content, the decreasing rate of the contact angle increases. The decay of contact angles has been expressed as diffusion-controlled effect by some researchers, and some models have also been established to correlate the change of contact angles during the aging processes.50,51 In our experiments, there should be three major factors that affect the decay of the contact angles. The first factor is the increasing PEG chains coverage of the membrane surfaces, which has been confirmed by XPS. Second, the increase of pore size on the membrane surface makes it easier for water diffusion. Finally, the HPE-g-MPEG in the membrane matrix enhances the wettability of the internal pore channels. Figure 5 is a comparison among the static, advancing, and receding contact angles. It is obvious that the discrepancy in static contact angle (or advancing contact angle) and receding contact angle is enlarged with the increase of HPE-g-MPEG content. This could be ascribed to the improved wettability of the membrane surface. Although the contact angle is affected greatly by the porosity of the membrane surface, the decrease (48) Steen, M. L.; Jordan, A. C.; Fisher, E. R. J. Membr. Sci. 2002, 204, 341. (49) Taniguchi, M.; Pieracci, J. P.; Belfort, G. Langmuir 2001, 17, 4312. (50) Taniguchi, M.; Belfort, G. Langmuir 2002, 18, 6465. (51) Liu, F. P.; Gardner, D. J.; Wolcott, M. P. Langmuir 1995, 11, 2674. (52) Yasuda, T.; Miyama, M.; Yasuda, H. Langmuir 1992, 8, 1425.
5784 Langmuir, Vol. 23, No. 10, 2007
Zhao et al.
Figure 5. Comparison of advancing contact angle (ACA), receding contact angle (RCA), and static contact angle (SCA) of membranes with a HPE-g-MPEG/PVDF ratio of (M0) 0, (M1) 1/10, (M2) 2/10, and (M3) 3/10. Figure 7. BSA adsorption on the membranes with different HPEg-MPEG content.
Figure 6. Water contact angle changes of the membrane with a HPE-g-MPEG/PVDF ratio of 3/10 after shaken in water (60 °C) for different time span.
in the initial contact angle, the increase in the decay rate, and the enlarged discrepancy in advancing and receding contact angles should be, to some extent, a reflection of the hydrophilictity improvement of the membrane surfaces. Stability of HPE-g-MPEG in Blend Membrane. To achieve a high-performance hydrophilic membrane, it is important to sustain the hydrophilicity during the practical application. For a blend membrane containing hydrophilic linear polymer, it is unavoidable that the hydrophilic component diffuses out of the membrane and that the hydrophilicity declines gradually in the filtration process. To investigate the stability of HPE-g-MPEG at the blend membrane surface, the HPE-g-MPEG/PVDF membrane (M3) was chosen as the sample membrane to be continuously shaken in water (60 °C) for different time span. The variation of the water contact angle with shaking time (Figure 6) obviously shows that the water contact angle changes little even the membrane has been leached as long as 30 days in 60 °C water, indicating a quite stable presence of HPE-g-MPEG in the blend membrane. This stability is great important in endowing the PVDF blend membrane with reliable hydrophilicity and filtration performance. In addition, it is necessary to perform stability experiments using typical cleaning chemicals or agents (e.g., NaOH, bleach, etc.), and these experiments are under consideration. Protein Absorption Property. Figure 7 shows the results of BSA adsorption on the pure and blend PVDF membranes. The amounts of BSA adsorbed on the pure PVDF and the blend membrane with a low HPE-g-PEG content (HPE-g-MPEG/ PVDF, 1/10) increase almost linearly with the increase of the BSA concentration, although the amounts of BSA adsorption on the blend membrane smaller than those on the pure PVDF membrane. However, the adsorbed amounts of BSA on the blend membranes with higher HPE-g-PEG contents (HPE-g-PEG/
Figure 8. Permeation fluxes of pure water and BSA solution through the pure PVDF membrane and membranes with a HPE-g-MPEG/ PVDF ratio of (M0) 0, (M1) 1/10, (M2) 2/10, and (M3) 3/10. Table 5. Anti-fouling Properties of the Studied Membranes HPE-g-MPEG/PVD ratio
RBSA(%)
RFR(%)
FRR(%)
0 1/10 2/10 3/10
99 98 96 92
87 64 41 36
22 58 78 89
PVDF, 2/10; 3/10) increase only slightly and keep at a certain low level with the increase of the BSA concentration. The effective reduction in protein adsorption is attributed to the MPEG arms of HPE-g-MPEG that stretch out from the membrane surface into the surrounding aqueous environment, which has been confirmed by the XPS results. According to Halperin’s model,52,53 BSA molecules may penetrate PEG brush in an end-on orientation and deposit on the surface at a low PEG density and thus cause significant adsorption. To obtain effective reduction in protein adsorption, the brush density should be high enough to make the distance between the neighboring PEG chains smaller than the dimensions of the indwelling BSA molecules. This model can be used to explain why the blend membrane with a low HPEg-MPEG content has a relatively high amount of BSA adsorption in this experiment. Filtration Performance. The pure water fluxes (Figure 8) increase dramatically, while the BSA rejection ratio (Table 5) decreases gradually with the increase of HPE-g-MPEG content. As seen from the surface and cross-section SEM images of the prepared membranes (Figure 9), the increase of the HPE-g-MPEG content results in the increase of porosity and pore size of the membrane surfaces, the size of the finger-like voids in the cross(53) Halperin, A. Langmuir 1999, 15, 2525. (54) Leckband, D. E.; Sheth, S. R.; Halperin, A. J. Biomater. Sci., Polym. Ed. 1999, 10, 1125.
Protein Resistance of PVDF Membranes
Langmuir, Vol. 23, No. 10, 2007 5785
Figure 9. Surface (left) and cross-section (right) SEM images of the membranes with a HPE-g-MPEG/PVDF ratio of (a) 0, (b) 1/10, (c) 2/10, and (d) 3/10.
section, and the improvement in connectivity of the internal voids, leading to the substantial increase in the pure water flux and the reduction in BSA rejection ratio. As for the changes of the SEM images of the cross-section, while both the pure and blend
PVDF membranes exhibit macrovoid formation, the size of the macrovoids become larger in the blend membranes with the increase of HPE-g-MPEG content. This result is consistent with previous observation on membranes cast from PVDF/PMMA
5786 Langmuir, Vol. 23, No. 10, 2007
blends5 and may be expected based on ternary diffusion models,54 which relate macrovoid formation to instantaneous liquid-liquid demixing during coagulation. The presence of amphiphilic HPEg-MPEG component in the casting solution might be expected to increase the affinity of the casting solution and precipitant, enhancing solvent-nonsolvent exchange and creating the conditions for instantaneous demixing and associated macrovoid formation. To further investigate the fouling resistance of the prepared membranes, a dynamic protein solution permeation process was conducted in this study, and BAS was used as a model protein. After that, the membranes were cleaned, and pure water fluxes were measured again. The results are compared in Figure 8, and the relative flux reduction (RFR) and flux recovery ratio (FRR) are given in Table 5. Protein fouling of the pure PVDF membrane results in a dramatic loss of the flux (RFR ) 87%) after the foulant solution permeation experiment, suggesting a large amount of BSA protein depositing on the membrane surface. However, the foulant solution flux reduction can be effectively restrained by the introduction of HPE-g-MPEG, and the recovery flux increases significantly with the increase of HPE-g-MPEG content. Concerning the reduction of the BSA rejection rate with the increase of the HPE-g-MPEG content, the differences in concentration polarization should also play a part in the increase in the flux of the BSA solution. For the membranes with a low HPE-g-MPEG content, the density of the MPEG arms of HPE-g-MPEG appearing at the membrane surface is not high enough to prevent the protein molecules from penetrating the MPEG chains and reaching the membrane surface (hydrophobic PVDF), causing irreversible fouling.55 In the case of the high HPE-g-MPEG content, a relatively hydrophilic dense MPEG layer formed at the membrane surface can keep the protein molecules from contacting the hydrophobic PVDF directly. Therefore, most of the protein molecules depositing on the membrane surface can be easily removed off by water washing, and the water flux can be recovered significantly. These results are in consistence with the static BSA adsorption studies. Recycling of the Blend Membrane. For a blend membrane with excellent properties, it should be reused for several runs while maintaining high water flux recovery rate. Therefore, the (55) Smolders, C. A.; Reuvers, A. J.; Boom, R. M.; Wienk, I. M. J. Membr. Sci. 1992, 73, 259. (56) Szleifer, I. M.; Carignano, A. Macromol. Rapid Commun. 2000, 21, 423.
Zhao et al.
Figure 10. Changes of the flux for the blend membrane with a HPE-g-MPEG/PVDF ratio of 3/10 (M3) after different times of filtration experiments.
blend membrane with a HPE-g-MPEG/PVDF ratio of 3/10 (M3) was chosen as the sample membrane to verify its anti-fouling property following the same procedure applied in the filtration experiments for three times. It can be seen from Figure 10 that the pure water flux can be retained at 507 kg/(m2‚h) after three BAS solution filtrations, which is about 82% of the initial pure water flux. These results suggest that the anti-fouling property of the PVDF/HPE-g-MPEG blend membranes can be retained during the filtration process.
Conclusion Amphiphilic hyperbranched-star polymer with about 12 hydrophilic arms in each molecule was successfully synthesized and blended with PVDF to fabricate porous membranes following phase inversion method. The MPEG arms in hyperbranched-star polymer can enrich at the membrane surface and endow the membranes with sufficient and stable hydrophilic property. The substantial coverage of MPEG segments prohibits the protein absorption at the membrane surface effectively. With improved hydrophilicity, the blended membrane exhibit enhanced performance in pure water flux, foulant solution permeability and water flux recovery. These results indicate that the synthesized amphiphilic hyperbranched-star polymer should be a novel candidate for the preparation of hydrophilic filtration membranes. Acknowledgment. Financial support from National “973” Program (Grant 2003CB615705) and “863” Program (Grant 2006AA03Z233) of China are gratefully acknowledged. LA070139O
