Langmuir 2007, 23, 13085-13092
13085
Plasma-Induced Graft Copolymerization of Poly(methacrylic acid) on Electrospun Poly(vinylidene fluoride) Nanofiber Membrane Satinderpal Kaur,*,†,‡ Zuwei Ma,† Renuga Gopal,†,‡ Gurdev Singh,| Seeram Ramakrishna,*,†,‡,§ and Takeshi Matsuura| Nanoscience and Nanotechnology InitiatiVe, Faculty of Engineering, National UniVersity of Singapore, 9 Engineering DriVe 1, Singapore 117576, Singapore, Department of Mechanical Engineering, Faculty of Engineering, National UniVersity of Singapore, 9 Engineering DriVe 1, Singapore 117576, Singapore, DiVision of Bioengineering, Faculty of Engineering, National UniVersity of Singapore, 9 Engineering DriVe 1, Singapore 117576, Singapore, and Industrial Membrane Research Institute, Department of Chemical Engineering, UniVersity of Ottawa, 161 Louis Pasteur Street, P.O. Box 450, Station A, Ottawa, Ontario K1N 6N5, Canada ReceiVed May 10, 2007. In Final Form: August 30, 2007 Electrospun nanofibrous membranes (ENM) which have a porous structure have a huge potential for various liquid filtration applications. In this paper, we explore the viability of using plasma-induced graft copolymerization to reduce the pore sizes of ENMs. Poly(vinylidene) fluoride (PVDF) was electrospun to produce a nonwoven membrane, comprised of nanofibers with diameters in the range of 200-600 nm. The surface of the ENM was exposed to argon plasma and subsequently graft-copolymerized with methacrylic acid. The effect of plasma exposure time on grafting was studied for both the ENM and a commercial hydrophobic PVDF (HVHP) membrane. The grafting density was quantitatively measured with toluidine blue-O. The degree of grafting increased steeply with an increase in plasma exposure time for the ENM, attaining a maximum of 180 nmol/mg after 120 s of plasma treatment. However, the increase in the grafting density on the surface of the HVHP membrane was not as drastic, reaching a plateau of 65 nmol/mg after 60 s. The liquid entry permeation of water dropped extensively for both membranes, indicating a change in surface properties. Field emission scanning electron microscopy micrographs revealed an alteration in the surface pore structure for both membranes after grafting. Bubble point measurements of the ENM reduced from 3.6 to 0.9 um after grafting. The pore-size distribution obtained using the capillary flow porometer for the grafted ENM revealed that it had a similar profile to that of a commercial hydrophilic commercial PVDF (HVLP) membrane. More significantly, water filtration studies revealed that the grafted ENM had a better flux throughput than the HVLP membrane. This suggests that ENMs can be successfully engineered through surface modification to achieve smaller pores while retaining their high flux performance.
Introduction The surface chemical and physical properties of the membrane play a pivotal role in determining the flux and selectivity of a filtration/separation process. Hence, surface modification is a powerful tool in membrane technology, which can be used to enhance the performance of membranes. Polymers suited for membrane applications should preferably be chemically stable and mechanically strong. Traditionally, polymers with the best solvent resistance or those which provide the most convenient pore structure are too hydrophobic for use as a filter in aqueous media.1 Conversely, polymers with the desired active surfaces do not possess adequate mechanical stability and hence cannot be used as a support or base membrane.2 Thus surface modification is frequently employed to combine the attributes of a desirable surface chemistry and adequate mechanical stability. There are several surface-modification techniques available3-5 that can be broadly categorized as physical or chemical processes. * Corresponding authors. E-mail:
[email protected];
[email protected]. † Nanoscience and Nanotechnology Initiative. ‡ Department of Mechanical Engineering. § Division of Bioengineering. | University of Ottawa. (1) Nunes, S. P.; Peinemann, K. V. Membrane Technology in the Chemical Industry; Wiley-VCH: Weinheim, Germany, 2001; pp 39-49. (2) Gopal, R.; Ma, Z.; Kaur, S.; Ramakrishna, S. Surface Modification and Application of Functionalized Polymer Nanofibers. In Molecular Building Blocks for Nanotechnology: From Diamondoids to Nanoscale Materials & Applications; Manssori, A. G., George, T. F., Zhang, G., Assoufid, L., Eds.; Springer: New York, 2006; pp 72-91. (3) Penn, L. S.; Wang, H. Polym. AdV. Technol. 1994, 5, 809-817.
Among these, plasma-induced graft copolymerization6-12 is an efficient and versatile way of introducing a selective polymeric layer on the surface of a hydrophobic membrane. Plasma-induced graft copolymerization treatment is limited only to the surface, and hence, the bulk properties of the membranes are still maintained. The thickness of the modified layer can be controlled up to the angstrom level.13 In addition, it is a powerful technique to transform a membrane with a symmetrical structure to an asymmetrical structure which increases selectivity without increasing hydrodynamic resistance significantly.14 Asymmetric membranes are preferred in membrane technology as they prevent pore plugging, a common drawback of symmetric membranes. Hence, by optimizing the grafting conditions, it could be possible to reduce the size of the pores and develop tighter pored (4) Kato, K.; Uchida, E.; Kang, E. T.; Uyama, Y.; Ikada, Y. Prog. Polym. Sci. 2003, 28, 209-259. (5) Ratner, B. D. Biosens. Bioelectron. 1995, 10, 797-804. (6) Wavhal, D. S.; Fisher, E. R. J. Membr. Sci. 2002, 209, 255-269. (7) Wavhal, D. S.; Fisher, E. R. Langmuir 2003, 19, 79-85. (8) Kang, E. T.; Neoh, K. G.; Shi, J. L.; Tan, K. L.; Liaw, D. J. Polym. AdV. Technol. 1999, 10, 20-29. (9) Hirotsu, T. J. Appl. Polym. Sci. 1987, 34, 1159-1172. (10) Hirotsu, T.; Nakajima, S. J. Appl. Polym. Sci. 1988, 34, 1159-1172. (11) Gancarz, I.; Bryjak, J.; Bryjak, M.; Tylus, W.; Pozniak, G. Eur. Polym. J. 2006, 42, 2430-2440. (12) Gancarz, I.; Pozniak, G.; Bryjak, M.; Frankiewicz, A. Acta. Polym. 1999, 50, 317-326. (13) Chan, C. M.; Ko, T. M.; Hiraoka, H. Surf. Sci. Rep. 1996, 24, 1-54. (14) Keith, S. Handbook of Industrial Membranes, 1st ed.; Elsevier Advanced Technology: London, 1995; pp 3-186.
10.1021/la701329r CCC: $37.00 © 2007 American Chemical Society Published on Web 11/16/2007
13086 Langmuir, Vol. 23, No. 26, 2007
Kaur et al.
Figure 1. Proposed mechanism of plasma induced graft polymerization on the surface of a membrane.
Figure 2. Fiber distribution of ENM.
Figure 3. Tensile strength of heat treated and non heat treated ENM.
Figure 4. Evaluation of grafting density of PMAA on the surface of ENM and HVHP exposed to plasma at different time intervals followed by grafting.
ultrafiltration (UF) and nanofiltration (NF) membranes from microfiltration (MF) membranes.1 Recently, plasma induced graft copolymerization was employed on electrospun nanofibrous membranes (ENMs)15,16 to (15) Ma, Z.; Kotaki, M.; Yong, T.; He, W.; Ramakrishna, S. Biomaterials 2005, 26, 2527-2536.
produce novel affinity membranes. ENMs are fabricated through a process called electrospinning which was patented in 190217,18 and has received renewed interest in recent years.19 An electrospun nanofibrous web possesses attractive properties such as high porosity, an interconnected open pore structure, pore sizes ranging from submicron to several micrometers, super hydrophobicity,20,21 and a large surface area per unit volume.22 These attributes make them highly attractive materials for use in separation technology where they are widely used in air filtration.23 Their usage in liquid filtration is however limited at present. Currently, ENMs possess pores in the micron range and are most suited for the separation of microparticles.24,25 As such they are only applicable as “loose” MF membranes, with a symmetrical structure. To broaden their use in separation technology, the pores have to be reduced further to the submicron level. The goal of this study was to evaluate the application of plasma-induced graft copolymerization as a means to reduce the surface pores of ENMs and develop “tighter” ENM-based filters. In the present study, poly(vinylidene) fluoride (PVDF) was electrospun as the base membrane, and a hydrophilic monomer methacrylic acid (MAA) was grafted on the surface of the ENM to develop an asymmetric membrane, with enhanced flux performance. Concomitantly, grafting was performed on a commercial phase-inverse PVDF membrane. The change in membrane performance of the ENM with the grafted layer was evaluated and compared with the commercial membranes. Experimental Section Fabrication of PVDF ENM. Electrospun fibrous PVDF membranes were prepared using a typical electrospinning setup.22 PVDF was chosen as it is an important material in the field of polymeric membranes, especially for MF and UF applications, due to its high thermal stability and excellent environmental resistance. Fifteen percent wt/vol PVDF (Kynar K-761, Elf Atochem USA) was prepared in N,N-dimethylacetamide (DMAC) and acetone (Merck; 2:3 v/v) mixture at 50 °C. A constant voltage of 15 kV was applied to draw the nanofibers from the prepared solution, using a 0.21 mm spinneret. The solution was electrospun at a rate of 4 mL/h onto a 10 cm2 aluminum collector plate. The distance between the tip of the spinneret (16) Chen, H.; Hsieh, Y. L. Biotechnol. Bioeng. 2005, 90, 405-413. (17) William, J. M. U.S. Patent No. 705,691, 1902. (18) Cooley J. F. U.S. Patent No. 692,631, 1902. (19) Ramakrishna, S; Kazutoshi, F.; Teo, W. E.; Lim, T. C.; Ma, Z. An introduction to electrospinning and nanofibers; World Scientific Pub: Singapore, 2005. (20) Ma, M.; Hill, R. M.; Lowery, J. L.; Fridrikh, S. V.; Rutledge, G. C. Langmuir 2005, 21, 5549-5554. (21) Miwa, M.; Nakajima, A.; Fujishima, A.; Hashimoto, K.; Watanabe, T. Langmuir 2000, 16, 5754-5760. (22) Huang, Z. M.; Zhang, Z.; Y.; Kotaki, M.; Ramakrishna, S. Compos. Sci. Technol. 2003, 63, 2223-2253. (23) Grafe, T.; Graham, K. Polymeric nanofibers and nanofiber webs: A new class of nonwovens. NonwoVen Technol. ReV. 2003, Spring, 51-55. (24) Gopal, R.; Kaur, S.; Ma, Z.; Chan, C.; Ramakrishna, S.; Matsuura, T. J. Membr. Sci. 2006, 281, 581-586. (25) Gopal, R.; Kaur, S.; Chao, Y. F.; Chan, C.; Ramakrishna, S.; Tabe, S.; Matsuura, T. J. Membr. Sci. 2007, 289, 210-219.
Copolymerization of Poly(methacrylic acid)
Langmuir, Vol. 23, No. 26, 2007 13087
Figure 5. SEM Micrographs of (a) nongrafted ENM, (b) top view of ENM exposed to plasma for 90 s, (c) top surface of ENM exposed to plasma for 90 s followed by grafting, (d) bottom surface of ENM after grafting, (e) nongrafted HVHP, (f) top view of HVHP exposed to plasma for 90 s, (g) exposed to plasma for 90 s followed by grafting, and (h) bottom surface of HVHP after grafting. and the collector was fixed at 15 cm. The collection time was set to 1 h. Humidity was controlled at approximately 70% using a dehumidifier. The fibrous membrane obtained was heat treated at 150 °C for 3 h. The thickness of the heat treated membrane was measured using a micrometer screw gauge (Mitutoyo, Japan). The morphology of the ENM was observed using a field-emission scanning electron microscope (FESEM; Quanta 200F, FEI, Netherlands). The diameter of the fibers was determined using the ImageJ software (http://rsb.info.nih.gov/ij/), and 130 measurements were taken to obtain the fiber diameter distribution. Plasma-Initiated Graft Copolymerization. Commercial hydrophobic PVDF membranes (HVHP, Millipore, USA) with a pore size of 0.45 µm and ENM membranes were cut into strips of dimension 1 cm × 2 cm and placed in a March Instruments Incorporated (USA) glow discharge plasma system. The plasma system was equipped with two parallel electrode plates and was supplied with a radio frequency power of 13.6 MHz. The glow discharge was produced at a plasma power of 30 W. Argon gas pressure was adjusted to 275 mtorr. The top surfaces of the membranes were exposed to plasma. The exposure period to plasma was tested under 30, 60, 90, and 120 s. After plasma pretreatment, the membrane surface was exposed to air for approximately 10 min to facilitate the formation of surface oxide and peroxides26,27 on the membrane surface before graft copolymerization was carried out. The membranes were subsequently placed in a glass tube containing aqueous solution of 10% (v/v) MAA (Sigma-Aldrich, Germany) monomer that was purified in vacuo to remove inhibitors that prevent polymerization and sealed with a silicon rubber stopper. The tube containing the membrane and monomer solution was repeatedly vacuumed and purged with argon gas to remove dissolved oxygen. Subsequently, the tube was heated at 80 °C for 1 h to initiate the graft copolymerization. Copolymerization was terminated by removing the silicone stopper and exposing the tube contents to air. The PMAA-grafted membranes were washed with copious amount of deionized (DI) water and left to soak in 0.1 M aqueous NaOH (Merck, Germany) overnight under gentle shaking to physically remove any adsorbed homopolymer or unreacted monomers. The schematic of the plasma induced graft copolymerization method is summarized in Figure 1. The grafting density of PMAA on the surface of ENM and HVHP was evaluated using toluidine blue (TBO) dye method.28,29 The TBO (26) Suzuki, M.; Kishida, A.; Iwata, H.; Ikada, Y. Macromolecules 1986, 19, 1804-1808. (27) Clouet, F.; Shi, M. K. J. Appl. Polym. Sci. 1992, 46, 1955-1966. (28) Uchida, E.; Uyama, Y.; Ikada, Y. Langmuir 1993, 9, 1121-1124. (29) Ma, Z.; Kotaki, M.; Ramakrishna, S. J. Membr. Sci. 2006, 272, 179-187.
dye specifically interacts with the carboxyl groups on the PMAA. The membranes were placed in a TBO (Aldrich, Germany) solution (pH ) 10) for 3 h followed by thorough rinsing with NaOH solution (pH ) 10) to remove any unbound TBO molecules. The stained membranes were then dried before immersing them into 50% acetic acid (BDH, England) which unbinds the TBO from the carboxyl groups of the PMMA. The absorbance of the solution at 620 nm was measured using a UV-vis spectrophotomer (Unicam UV 300, Thermo Spectronic, U.S.A.). The amount of carboxyl groups on the PMMA is directly proportional to the adsorbed TBO amount. A calibration curve of different concentration of TBO against its respective absorbance value was made and with reference to this curve, the grafting density at different plasma pretreatment time was determined. Surface Analysis and Property Measurements. To investigate the changes in chemical structure between the modified and nonmodified membranes, a single bounce horizontal attenuated total reflectance Fourier transform infrared (ATR-FTIR, Thermo Nicolet Avatar 360, U.S.A.) spectrometer was used. The ATR accessory (Avatar OMINI-Sampler accessory) contained a germanium crystal at a nominal incident angle at 45°. Each spectrum was obtained by accumulating 64 scans at a resolution of 4 cm-1. Changes in surface morphology of the nongrafted and grafted membranes were observed using FESEM. Dynamic water contact angle (WCA) measurements were performed on membranes using an Advanced Surface Technologies, Inc., VCA2000 (U.S.A.) video contact angle system. The machine was coupled with a camera enabling image capture at 52 frames/s, and the contact angle was determined from these images. A water droplet of 0.5 µL was dispersed on the membrane surface and the contact angle determined using the system software. The change in static contact angle with respect to time was also measured. The tensile properties of the heat treated and non-heat-treated ENM were tested on a universal testing machine (Instron 3345) equipped with a 100 N load cell. Three strips of length 3 cm × 1 cm were cut from each type of membrane. Membrane Characterization. The pore size distribution and bubble point measurements of all membranes used in this study were determined using a capillary flow porometer (Porous Materials Inc, U.S.A.). The membranes were completely wetted with Galwick (Porous Materials Inc, U.S.A.) and pressure was applied on one side. As the gas pressure increases gradually, there will be no gas flow through the pores until capillary forces are overcome releasing liquid from the pore. The first gas bubble will emerge from the largest pore where the capillary forces are lowest. The pressure at which this occurs is called the “bubble point” pressure. The
13088 Langmuir, Vol. 23, No. 26, 2007
Kaur et al. volume of void spaces × 100% ) volume of bulk membrane volume of Silwick × 100% (2) volume of bulk membrane
Figure 6. ATR-FTIR spectrum of (a) grafted ENM, exposed to plasma for 90 s followed by grafting, and (b) nongrafted ENM.
The mass of Silwick occupying the void spaces in the membrane was measured by subtracting the mass of the membrane measured before and after wetting. The thickness of membrane was measured after absorption of the wetting liquid using a micrometer screw gauge. A simple dead end filtration setup24 was used to evaluate the membrane performance. Based on the optimized grafting conditions, grafting was carried out on membranes with a diameter of 2 cm. A dried membrane was placed in the filtration cell which was then filled with 350 cm3 of DI water. The pressure was gradually increased and the pressure at which the first water droplet was observed at the permeate side was determined. This is defined as the liquid entry pressure of water (LEPw). The membrane was then removed from the cell, wetted with 50% ethanol, and then immersed in DI water for 30 min to ensure that the membrane was completely wet. The wetted membrane was then placed back in the cell for water flux study. The applied pressure was increased gradually and the corresponding water flux measured.
Results and Discussion
Figure 7. ATR-FTIR spectrum of (a) grafted HVHP, exposed to plasma for 90 s followed by grafting, and (b) nongrafted HVHP.
Figure 8. Water contact angle of grafted ENM and grafted HVHP that were initially exposed to plasma for 90 s followed by grafting. relationship between the pore size and the corresponding pressure is given by the Young-Laplace equation R)
2γ cos θ ∆P
(1)
where R is the radius of the pore, ∆P is the differential gas pressure, γ is the surface tension of the wetting liquid, Galwick (γ ) 15.9 dynes/cm), and θ is the wetting angle. The pressure is increased continuously until the liquid is displaced from the smallest pore and there is complete flow of gas through the membrane which equals that through a dry membrane. A pore size distribution was then calculated from this data.30 The membrane was wetted with a wetting liquid, Silwick (γ ) 20.1 dynes/cm). The specific gravity of Silwick at 25 °C is 0.93 g/cm3. The porosity of the membranes was calculated using the formula given below: (30) Porter, M. C. Handbook of Industrial Membrane Technology; Noyes Publications: Park Ridge, NJ, 1990.
Electrospun PVDF ENM. PVDF ENM obtained from electrospinning was comprised of random nonwoven mesh of fibers, giving rise to an interconnected open pore structure. As shown in Figure 2, ∼ 72% of the fibers lie in the range of 200600 nm. The ENMs can be classified as a “tortuous-pore” MF membrane due to a network of interconnecting tortuous flow paths. Generally for such a membrane, the pore openings do not correspond to the limiting pore size within the depth of membranes,30 and hence pore size characterization using the SEM is not suitable. One draw-back of the as-spun ENM was its poor mechanical strength. It had a “cotton” like structure and thus could not function as a filter. To overcome this, the ENM was heat treated under 150 °C for 3 h to improve the membrane’s overall structural integrity. The heat treated membrane became sufficiently rigid to be used for further characterizations. The increase in rigidity is evident from the tensile profile shown in Figure 3. The heat treated membrane had a significantly higher mechanical strength (8.5 MPa) compared to the non-heat-treated ENM (0.4 MPa). Surface-Grafted Membranes. The grafted surfaces of both ENM and HVHP turned orange and hence can be discriminated easily from the nongrafted surfaces. The function of plasma is to generate radicals on the surface of the membranes. The generation of free radicals can happen either through the bombardment of ions and photons which can provide energy higher than the -C-C- or -C-H- bond hence breaking the respective bonds or through elastic or inelastic collision of the electron in the plasma with the polymer resulting in the abstraction of a proton. In inert gas plasmas, the dominant process is hydrogen abstraction.20 When the membranes were exposed to air, these radicals react with the oxygen in the air to form peroxides. When immersed in a degassed MAA solution at 80 °C, these peroxides cleave to form radicals to induce the graft polymerization of MAA monomers. Figure 4 shows the change in graft density with respect to plasma exposure time for both the ENM and HVHP. At 30 s of plasma exposure time followed by grafting, HVHP had the same level of grafting density as ENM. Interestingly, beyond 30 s, the response of the membranes to plasma exposure time greatly varied. The rate of change of grafting density with exposure time for the ENM was significantly larger than that of HVHP. For the
Copolymerization of Poly(methacrylic acid)
Langmuir, Vol. 23, No. 26, 2007 13089
Figure 9. Pore size distribution of ENM.
Figure 10. Pore size distribution of HVHP.
ENM, the graft density increased in a linear manner from 30 (23 nmol/mg) to 90 s (170 nmol/mg) and less gradually from 90 to 120 s (180 nmol/mg). On the contrary, HVHP showed only an increase in graft density from 30 (25 nmol/mg) to 60 s (65 nmol/ mg), and no significant change in graft density was observed with further increasing plasma exposure time. Formation of radicals on the PVDF surface in Ar plasma treatment and the formation of peroxide when the material is
exposed to air have been previously reported.11,34,35 When the membranes were exposed to plasma for 90 s and then grafted, the grafting density was high with an average of 170 nmole/mg (31) Wenzel, R. N. Ind. Eng. Chem. 1936, 2, 988-994. (32) Cassie, A. B. D.; Baxter, S. Trans. Faraday Soc. 1944, 40, 546-551. (33) Singh, A.; Steely, L.; Allock, H. R. Langmuir 2005, 21, 11604-11607. (34) Yang, G. H.; Kang, E. T.; Neoh, K. G. Langmuir 2001, 17, 211-218. (35) Cen, L.; Neoh, K. G.; Kang, E. T. Langmuir 2003, 19, 10295-10303.
13090 Langmuir, Vol. 23, No. 26, 2007
Kaur et al.
Figure 11. Pore size distribution of grafted ENM that was exposed to plasma at 90 s, followed by subsequent grafting.
Figure 12. Pore size distribution of HVLP.
for grafted ENM and 60 nmole/mg for grafted HVHP. On the other hand, when the membranes were not exposed to plasma and then grafted (as a control), the grafting density was very low with an average of 5 nmol/mg for ENM and 3 nmol/mg for HVHP. Therefore, the formation of radicals and peroxide groups on the membrane surface facilitated the grafting. As reported by Ikada et al.,26 the dependence of the peroxide formation on the
glow discharge time does not increase continuously but reaches a maximum at some point. Since, the formation of peroxide groups is directly related to the grafting density, a longer plasma exposure time does not necessarily imply a higher grafting density. This was clearly observed for HVHP, where no significant change in graft density was observed beyond 60 s. In the case of the ENM, although there was a linear increase in graft density from
Copolymerization of Poly(methacrylic acid)
30 to 90 s, beyond 90 s, the grafting density increased only slightly. One possible reason for the difference in grafting densities between the ENM and HVHP is the surface area of the membranes. ENMs by virtue of their small fibers possess a high surface area and hence a larger area for grafting to occur. HVHP, on the other hand, has a much lower surface area. Influence of Grafting on Membrane Surface Morphology and Chemistry. The effect of exposing the surface of the membranes to plasma for 90 s followed by grafting of PMAA was investigated from this point onward. When both ENM and HVHP were exposed to plasma for 90 s, there was no significant change in surface morphology in comparison to their untreated state as seen from Figure 5 (Figure 5a,b for ENM and Figure 5e,f for HVHP). However, upon grafting, the surface of the membranes was significantly altered as seen in Figure 5, panels c and g. In the grafted ENMs, the fibers appear to swell (Figure 5c) resulting in a reduction of the surface pores. Likewise, for the HVHP membranes, a more significant reduction in surface pores was observed after grafting. However, the bottom surface of both membranes that was not exposed to plasma was not affected by the grafting process and this effect can be observed in Figure 5, panels d and h. From the TBO analysis, it was found that the ENM grafting is more extensive than the HVHP grafting. This is most probably due to the grafting in ENM extending within the membrane. The ENM has much larger surface pores and thus more fibers exposed to plasma. As a result, the grafting in ENMs is not confined just to the very top layer of fibers but also includes subsequent underlying layers of fibers which were exposed to plasma. Figure 6 presents the typical ATR-FTIR spectra of the grafted and non-grafted ENM. It is evident that the grafted ENM had additional peaks at ∼3400 and ∼1600 cm-1, corresponding to a hydroxyl and a carbonyl stretch respectively of the carboxyl groups present in PMAA, when compared to nongrafted ENM. Similarly, these additional peaks were present for the grafted HVHP as seen in Figure 7. The water contact angle of the nongrafted ENM was measured to be 132 ( 9°, and that of the HVHP was 124 ( 8°. The increased hydrophobicity for the ENM is due to inherent surface roughness and trapped air pockets.26,27,31-33 However after grafting, the contact angle at time 0 s was reduced to 62° and 67° for the grafted ENM and grafted HVHP, respectively. The contact angle of the grafted ENM decreased rapidly with time and reached zero in a few seconds, whereas for the grafted HVHP contact angle decrease was not as swift as the grafted ENM (Figure 8). The difference between these two grafted membranes seems attributable to the differences in their morphologies. The grafted ENM has a larger pore size and porosity than the grafted HVHP, and as a result, the contact angle of the grafted ENM decreased faster than the grafted HVHP. Water cannot penetrate inside a dry membrane unless a transmembrane hydrostatic pressure that exceeds the liquid entry pressure of water (LEPw) was applied. It was anticipated that having a higher contact angle, ENM would exhibit a higher LEPw than HVHP. However, on the contrary, the LEPws of nongrafted ENM and HVHP were 2 and 14 psi, respectively. This indicates that the amplified hydrophobicity of ENMs was not a true indication of their resistance to wetting. The ENM possess very large surface pores compared to HVHP; thus, a lower amount of pressure is required to overcome the initial surface resistance and allow water to pass through the membrane. After grafting, the LEPw of the ENM dropped to 0 psi, corresponding to the rapid decrease in the water contact angle within 2 s. The LEPw of the grafted HVHP membrane was
Langmuir, Vol. 23, No. 26, 2007 13091
Figure 13. Water flux permeation of ENM, grafted ENM, HVLP, and HVHP.
higher at 3 psi. Although a hydrophilic group was grafted on the surface of the commercial membrane, the pore size played a more dominant factor in determining the LEPw than the surface hydrophilicty. For the ENM, as the pores are generally much larger, the presence of hydrophilic groups had a more significant impact on the LEPw. Influence of Grafting on Pore Size. Prior to grafting, the bubble point of the ENM was 3.58 µm and that of the HVHP was 0.91 µm. The smallest pore detected for the ENM was 0.6 µm, whereas that of the HVHP is 0.2 µm. The pore size distributions of these two membranes are presented in Figures 9 and 10, respectively. From these data, it is obvious that ENM would have a different separation capability than a phase-inverted membrane due to a large difference in their pore size distribution. After grafting, the HVHPs bubble point and its smallest pore were reduced beyond the sensitivity of the porometer. Hence, a comparison of the pore size distribution of the grafted ENM and a hydrophilic commercial 0.45 µm PVDF (HVLP) membrane was made instead. The largest pore (bubble point) of the grafted ENM was detected as 0.88 µm, and that of the HVLP was 1 µm. However, it can be observed from Figures 11 and 12 that the percentage of pores at the bubble point was almost zero. The pore size distribution for the grafted ENM and HVLP was e0.45 and e0.5 µm respectively. For both the membranes, the smallest pore size is 0.05 µm, and the pore size range that occurs at the highest frequency is between 0.3 and 0.35 µm. The results indicate that it is possible to reduce the pore size of the ENM through surface grafting. By optimizing the grafting conditions, it might be even possible to achieve pores in the range of ultrafiltration membranes. Comparison of Water Flux Performance of Grafted and Nongrafted ENMs. According to the flux profile shown in Figure 13, the nongrafted ENM had much better flux performance than the grafted ENM even though grafting increased the hydrophilicty of the membrane. This can be explained in terms of the decreased pore size of the grafted ENM. According to the pore size distribution (Figures 9 and 11), pristine ENM had larger pores and the bubble point was four times more than the grafted ENM. Hence, a smaller pressure was required for the pristine ENM to achieve the same flux. In this situation, the pore size rather than the surface chemistry is the determinant factor for the flux performance. Comparison of Water Flux Performance of Grafted ENM and Commercial Membranes. The flux performance of grafted ENM and commercial membrane is also depicted in Figure 13. When the pore sizes are kept the same and the surface hydrophilicity is varied, the flux performance can be enhanced. This is evident in the case of HVHP and HVLP. Despite having a similar pore sizes (∼0.45 µm), HVLP had a higher flux than HVHP attributed to hydrophilic surface layer. Interestingly, although the pore size distribution of grafted ENM and HVLP
13092 Langmuir, Vol. 23, No. 26, 2007
was very similar and both of them had LEPw of zero, their flux profile is different. Grafted ENM had a higher water flux than HVLP. This can be explained by the difference in the membrane porosity of the membranes. ENM has a porosity of 80%, whereas HVLP has a porosity of 65%. ENMs have a much higher porosity, and this leads to a higher flux as compared to HVLP membranes which have lower porosity. Therefore, through surface modification, specifically plasma copolymerization as shown here, it is possible to develop tighter pored ENMs while retaining their high flux performance characteristics.
Conclusion The surface of ENM was successfully grafted with PMAA by plasma-induced graft copolymerization. The grafted ENM showed
Kaur et al.
a higher grafting density than HVHP, primarily due to its higher surface area for grafting. Through grafting, the bubble point of the ENM was significantly reduced from 3.58 to 0.88 µm. After grafting, the ENM was transformed into a microfiltration membrane, similar in pore-size distribution with a commercial 0.45 µm hydrophilic phase inversed membrane HVLP but with significantly better flux. This indicates that, through grafting techniques, it is possible to develop high flux electrospun nanofiber membranes with reduced pore size. Acknowledgment. This work was supported by the Strategic Research Program Grant (022 109 0059) of the Agency for Science Technology and Research (A*Star) of Singapore. LA701329R
