Surface Modification of Dense Membranes Using Radical Graft

Jun 17, 2010 - Zuckerberg Institute for Water Research and Department of Biotechnology and Environmental Engineering,. Ben-Gurion University of the ...
0 downloads 0 Views 2MB Size
Download PDF
pubs.acs.org/Langmuir © 2010 American Chemical Society

Surface Modification of Dense Membranes Using Radical Graft Polymerization Enhanced by Monomer Filtration Roy Bernstein, Sofia Belfer, and Viatcheslav Freger* Zuckerberg Institute for Water Research and Department of Biotechnology and Environmental Engineering, Ben-Gurion University of the Negev, P.O. Box 635, Beer-Sheva 84105, Israel Received April 30, 2010. Revised Manuscript Received May 30, 2010 Surface graft polymerization is a promising way to modify membranes for improved performance. Redox-initiated graft polymerization of vinyl monomers is a facile and inexpensive method carried out at room temperature in aqueous media; however, its use is often limited by slow kinetics, low surface specificity, and excessive consumption of chemicals on undesired homopolymerization. It is shown that in the case of RO or NF membranes these drawbacks may be eliminated by utilizing the selectivity of the membranes toward monomers and carrying out the polymerization while applying pressure, i.e., under filtration conditions. Concentration polarization that ensues raises the concentration of reagents near the membrane surface and thereby drastically increases the rate of reaction and preferentially directs it towards surface grafting. Grafting experiments using 2-hydroxyethyl methacrylate and other monomers and characterization of modified membranes using permeability measurements, ATR-FTIR, AFM, XPS, and contact angle demonstrate that the required monomer concentrations can be drastically reduced, particularly when a small fraction of a cross-linker is added. As an additional benefit, this approach enables broadening the spectrum of utilizable monomers to sparingly soluble hydrophobic, charged, and macro-monomers, as was demonstrated using sparingly soluble ethyl methacrylate and 2-ethoxyethyl methacrylate and other monomers. Even though the kinetics of the process is substantially complicated by evolution and concentration polarization of oligomeric and polymeric species, especially in the presence of a cross-linker, it is well offset by the benefits of higher rate, specificity, and reduced monomer consumption.

1. Introduction Thin-film composite (TFC) membranes with a polyamide top layer are the most common reverse osmosis (RO) membranes used today for desalination and water treatment. They appear to have an exceptionally favorable combination of characteristics for salt removal, combining high permeability with good salt rejection, robustness, and durability.1 Nevertheless, the propensity to colloidal, organic, and bio-fouling2-4 and poor rejection of noncharged small solutes such as arsenious5 and boric acids6 and many organic pollutants7 present serious drawbacks. They are believed to result from the relative hydrophobicity of the polyamide, which is crucial for good ion exclusion but leads to *Corresponding author. E-mail: [email protected]. (1) Freger, V. Swelling and Morphology of the Skin Layer of Polyamide Composite Membranes: An Atomic Force Microscopy Study. Environ. Sci. Technol. 2004, 38, 3168-3175. (2) Tang, C. Y.; Kwon, Y. N.; Leckie, J. O. Fouling of Reverse Osmosis and Nanofiltration Membranes by Humic acid;effects of Solution Composition and Hydrodynamic Conditions. J. Membr. Sci. 2007, 290, 86-94. (3) Vrijenhoek, E. M.; Hong, S.; Elimelech, M. Influence of Membrane Surface Properties on Initial Rate of Colloidal Fouling of Reverse Osmosis and Nanofiltration Membranes. J. Membr. Sci. 2001, 188, 115-128. (4) Herzberg, M.; Elimelech, M. Biofouling of Reverse Osmosis Membranes: Role of Biofilm-Enhanced Osmotic Pressure. J. Membr. Sci. 2007, 295, 11-20. (5) Zhu, A.; Christofides, P. D.; Cohen, Y. Corrigendum to on RO Membrane and Energy Costs and Associated Incentives for Future Enhancements of Membrane Permeability. J. Membr. Sci. 2010, 346. (6) Sagiv, A.; Semiat, R. Analysis of Parameters Affecting Boron Permeation through Reverse Osmosis Membranes. J. Membr. Sci. 2004, 243, 79-87. (7) Kimura, K.; Amy, G.; Drewes, J. E.; Heberer, T.; Kim, T. U.; Watanabe, Y. Rejection of Organic Micropollutants (Disinfection by-Products, Endocrine Disrupting Compounds, and Pharmaceutically Active Compounds) by NF/RO Membranes. J. Membr. Sci. 2003, 227, 113-121. (8) Paul, M.; Park, H. B.; Freeman, B. D.; Roy, A.; McGrath, J. E.; Riffle, J. Synthesis and Crosslinking of Partially Disulfonated Poly (Arylene Ether Sulfone) Random Copolymers as Candidates for Chlorine Resistant Reverse Osmosis Membranes. Polymer 2008, 49, 2243-2252.

12358 DOI: 10.1021/la1017278

high affinity toward small uncharged solutes and hydrophobic foulants. The improvement may be achieved along different lines, such as search for totally new materials8,9 or making hybrid materials on the basis of the currently used ones. An example of the latter is incorporation of inorganic nanoparticles into the polyamide layer for mitigating fouling10 or increasing flux or salt rejection.11,12 In this case one specific aspect is addressed without losing the other proven advantageous characteristics of the standard material. Another realization of this principle is the surface modification of membranes, which adds a layer of a different material on top of the membrane, which may offer a number of advantages. A very thin extra layer may change the surface characteristics of the membrane, for instance, for improved fouling resistance with a minimal effect on the permeability and salt rejection.13 It can also be used for improving removal of certain contaminants without impairing salt removal. Note that, in order to override selectivity of the original layer, the extra layer must have a certain minimal resistance and hence an appreciable thickness. The improved (9) Zhou, M.; Nemade, P. R.; Lu, X.; Zeng, X.; Hatakeyama, E. S.; Noble, R. D.; Gin, D. L. New Type of Membrane Material for Water Desalination Based on a Cross-Linked Bicontinuous Cubic Lyotropic Liquid Crystal Assembly. J. Am. Chem. Soc. 2007, 129, 9574-9575. (10) Kwak, S. Y.; Kim, S. H.; Kim, S. S. Hybrid organic/inorganic Reverse Osmosis (RO) Membrane for Bactericidal Anti-Fouling. 1. Preparation and Characterization of TiO2 Nanoparticle Self-Assembled Aromatic Polyamide Thin-Film-Composite (TFC) Membrane. Environ. Sci. Technol. 2001, 35, 2388-2394. (11) Jeong, B. H.; Hoek, E. M. V.; Yan, Y.; Subramani, A.; Huang, X.; Hurwitz, G.; Ghosh, A. K.; Jawor, A. Interfacial Polymerization of Thin Film Nanocomposites: A New Concept for Reverse Osmosis Membranes. J. Membr. Sci. 2007, 294, 1-7. (12) Lind, M. L.; Ghosh, A. K.; Jawor, A.; Huang, X.; Hou, W.; Yang, Y.; Hoek, E. M. Influence of Zeolite Crystal Size on Zeolite-Polyamide Thin Film Nanocomposite Membranes. Langmuir 2009, 25, 10139-10145. (13) Rana, D.; Matsuura, T. Surface Modifications for Antifouling Membranes. Chem. Rev. 2010, 481.

Published on Web 06/17/2010

Langmuir 2010, 26(14), 12358–12365

Bernstein et al.

rejection then comes at the expense of permeability, and grafting needs to be optimized to achieve the best trade-off. Since the present study focuses on the grafting procedure rather than membrane performance, the drop in permeability along with direct assessment of the amount of grafted polymer on the surface (e.g., by IR) served as a convenient indicator of the extent of grafting and its uniformity. An appealing way to achieve surface modification is radical graft polymerization, by either growing a polymer from the membrane surface (“grafting from”) or attaching a grown polymer from the solution to the membrane (“grafting to”). The “graft from” process is much more common and starts from reaction of either a vinyl monomer with a primary radical site on the membrane surface or of a vinyl radical and the membrane. Such an event is then followed by addition of monomers to the growing chain, side reactions of electron transfer, and termination by either disproportion or coupling.14 The primary radical can be formed either directly using techniques such as ion beam,15 plasma,16 γ-irradiation, and ozone treatment17 or by electron transfer from a free radical initiator. The latter can be generated by homolytic decomposition of compounds like photoininitiators18 or peroxide19 using heat or light or by a redox reaction between a suitable oxidizer and reductant.14,20 Most of these techniques were used for membrane modification.13,21,22 The redox method used in this work has several important advantages; it is facile and can be used in aqueous media at room temperature without an external activation.20 Belfer et al. successfully applied this method using the persulfate/metabisulfite redox couple for modifying membranes and commercial elements with hydrophilic and charged monomers, resulting in lesser fouling and more efficient cleaning.23-27 However, the redox initiation has the disadvantage of relatively slow kinetics that necessitates the use of high monomer concentrations to reach a substantial (14) Jenkins, D. W.; Hudson, S. M. Review of Vinyl Graft Copolymerization Featuring Recent Advances Toward Controlled Radical-Based Reactions and Illustrated with chitin/chitosan Trunk Polymers. Chem. Rev. 2001, 101, 3245-3274. (15) Liu, F.; Zhu, B. K.; Xu, Y. Y. Improving the Hydrophilicity of Poly (Vinylidene Fluoride) Porous Membranes by Electron Beam Initiated Surface Grafting of AA/SSS Binary Monomers. Appl. Surf. Sci. 2006, 253, 2096-2101. (16) Kim, M.; Lin, N. H.; Lewis, G. T.; Cohen, Y. Surface Nano-Structuring of Reverse Osmosis Membranes Via Atmospheric Pressure Plasma-Induced Graft Polymerization for Reduction of Mineral Scaling Propensity. J. Membr. Sci. 2010. (17) Kato, K.; Uchida, E.; Kang, E. T.; Uyama, Y.; Ikada, Y. Polymer Surface with Graft Chains. Prog. Polym. Sci. 2003, 28, 209-259. (18) Ulbricht, M. Photograft-Polymer-Modified Microporous Membranes with Environment-Sensitive Permeabilities. React. Funct. Polym. 1996, 31, 165-178. (19) Ulbricht, M.; Belfort, G. Surface Modification of Ultrafiltration Membranes by Low Temperature Plasma II. Graft Polymerization Onto Polyacrylonitrile and Polysulfone. J. Membr. Sci. 1996, 111, 193-215. (20) Sarac, A. Redox Polymerization. Prog. Polym. Sci. 1999, 24, 1149-1204. (21) He, D.; Susanto, H.; Ulbricht, M. Photo-Irradiation for Preparation, Modification and Stimulation of Polymeric Membranes. Prog. Polym. Sci. 2009, 34, 62-98. (22) Khulbe, K.; Feng, C.; Matsuura, T. The Art of Surface Modification of Synthetic Polymeric Membranes. J. Appl. Polym. Sci. 2009, 115, 855-895. (23) Belfer, S.; Purinson, Y.; Fainshtein, R.; Radchenko, Y.; Kedem, O. Surface Modification of Commercial Composite Polyamide Reverse Osmosis Membranes. J. Membr. Sci. 1998, 139, 175-181. (24) Belfer, S.; Gilron, J.; Purinson, Y.; Fainshtain, R.; Daltrophe, N.; Priel, M.; Tenzer, B.; Toma, A. Effect of Surface Modification in Preventing Fouling of Commercial SWRO Membranes at the Eilat Seawater Desalination Pilot Plant. Desalination 2001, 139, 169-176. (25) Belfer, S.; Fainshtain, R.; Purinson, Y.; Gilron, J.; Nystr€om, M.; M€antt€ari, M. Modification of NF Membrane Properties by in Situ Redox Initiated Graft Polymerization with Hydrophilic Monomers. J. Membr. Sci. 2004, 239, 55-64. (26) Kim, J. H.; Park, P. K.; Lee, C. H.; Kwon, H. H. Surface Modification of Nanofiltration Membranes to Improve the Removal of Organic Micro-Pollutants (EDCs and PhACs) in Drinking Water Treatment: Graft Polymerization and Cross-Linking Followed by Functional Group Substitution. J. Membr. Sci. 2008, 321, 190-198. (27) Van der Bruggen, B. Comparison of Redox Initiated Graft Polymerisation and Sulfonation for Hydrophilisation of Polyethersulfone Nanofiltration Membranes. Eur. Polym. J. 2009, 45, 1873-1882.

Langmuir 2010, 26(14), 12358–12365

Article

degree of grafting. The modification is also insufficiently selective toward surface polymerization, which results in an excessive consumption of monomers on undesired formation of a homopolymer in solution rather than on the surface. As a remedy to these problems, it is proposed here to utilize the selectivity of RO or NF membranes toward monomers and initiators. In the new method the polymerization is carried out under pressure, i.e., while the monomer solution is being filtered through the membrane. This entails concentration polarization (CP), whereby the concentration of the reagents near the membrane surface (Cm) may increase drastically compared to the feed solution (Cf), i.e., the factor CPF = Cm/Cf may be large. As a result, the reaction is sped up and preferentially occurs near the surface. Compared to regular uniform conditions (CPF ≈ 1), the surface grafting requires smaller amounts of the monomer, cross-linker, and initiator in the bulk in order to reach a desired degree of grafting. The homopolymerization is then significantly reduced due to the much slower kinetics in the bulk. As an additional benefit, this modified approach potentially enables to broaden the range of monomers to include sparingly soluble hydrophobic monomers, charged monomers, and monomers with bulky side groups that are hard to graft in regular way due to vanishingly slow kinetics. The advantages of the new approach are demonstrated in this study using a few monomers commonly used for surface and membrane modification.

2. Experimental Section 2.1. Materials. 2-Hydroxyethyl methacrylate (HEMA), ethylene glycol dimethacrylate (EGDMA), methyl methacryalte (MM), poly(ethylene glycol) methacrylate (PEGMA 360), potassium persulfate, potassium metabisulfite, ethyl methacrylate (EMA), and 2-ethoxyethyl methacrylate (EEMA) were supplied by Aldrich and used without purification. The cross-linker, EGDMA, was dissolved in water for 24 h at very low concentration, 1 g/L water, and stored at 4 °C to be used within 2 weeks. Doubledistilled deionized water was used in all experiments and procedures. The fully aromatic polyamide (PA) membranes ESPA-1 (Hydranautics) was kindly supplied by the manufacturer as a flat sheet and stored at 4 °C. Prior to use, the membranes were washed three times in an ultrasonic bath (10 min), soaked in 50/50 v/v ethanol-water for 1 h, and washed with water. Before modification the membranes were tested at a pressure 20 bar for water permeability using water as a feed and for salt rejection using a 1.5 g/L NaCl solution. The tests were performed in a 150 mL nitrogen-pressurized dead-end stirred cell having a membrane area 12.4 cm2. Permeability was determined by collecting and weighing the permeate and salt rejection Rs = 1 - Λp/Λf was determined from measured electric conductance of feed Λf and permeate Λp.

2.2. Polymerization Procedures and Extent of Concentration Polarization. The kinetics of homopolymerization in the bulk solution was assessed ex situ by measuring the turbidity (as optical density at 600 nm) of solutions of monomers and initiators at different concentrations vs time in a 1 cm thick cuvette using a UV spectrophotometer (Lambda EZ 201 Perkin-Elmer). Water solution of the initiators was used as blank. The CP-enhanced surface polymerization of membranes was carried out in the same dead-end filtration cells as primary testing of flux and salt rejection. The initiators, monomer, and crosslinker were dissolved separately in water, mixed together, and promptly used to fill the dead-end cell, which was then immediately sealed and a trans-membrane pressure was applied. At the end of the modification the membrane was taken out, washed several of times with deionized water, cleaned by shaking for 24 h in 1:1 (v/v) ethanol-water, washed again with water, and examined for salt rejection and permeability. DOI: 10.1021/la1017278

12359

Article

Bernstein et al.

Table 1. Characteristics of Some Acrylic Monomers and Initiators Used in This Study and the Intrinsic Passage (Pin) for ESPA-1 Membrane and CPF Values at No Stirring Conditions Estimated from Results Shown in Figure 1a

a

Values indicated by asterisks were estimated by means of eq 3 using P measured without stirring and Pin as passage at stirring rate 750 rpm.

The factor CPF for the membrane (ESPA-1) was estimated for three uncharged monomers, presented in Table 1, from the monomer passage as a function of stirring rate using the following relation:28 ln

P Pin JV ¼ ln 1 - Pin kd 1-P

ð1Þ

where JV is the volume flux, kd the average mass transfer coefficient, and P = Cp/Cf and Pin = Cp/Cm are the measured and intrinsic monomer passages, respectively, Cp being the permeate concentration. The tests were carried out at constant pressure 20 bar and monomer concentrations in the feed in the low range 15-45 ppm (200-600 μM) to minimize the effect of monomer sorption on flux and to keep the monomer concentration well below the solubility limit. The concentrations Cf and Cp were measured using an Apollo 9000 Teledyne Tekmar total organic carbon (TOC) analyzer. The flux was monitored by collecting and weighing permeate every 1 min and showed no significant change for 6 min that measurmemts lasted. In order to minimize the change of monomer concentration inside the dead-end cell due to filtration, the total volume of permeate did not exceed 8% of the feed volume. The stirring rate was varied for the same solution between 100 and 1000 rpm (minimal CP) to test the effect of polarization. In addition, an experiment without stirring (maximal CP) was performed for some solutes. The dependence of mass transfer coefficient for finite stirring rates ω was assumed to obey the relation used by Opong and Zydney29 kd ¼ k 0 d ω0:567

ð2Þ

where k0 d is a constant specific for the stirred cell and solute used. Pin and k0 d were then determined from the intercept and slope of (28) Bowen, W. R.; Mukhtar, H. Characterisation and Prediction of Separation Performance of Nanofiltration Membranes. J. Membr. Sci. 1996, 112, 263-274. (29) Opong, W. S.; Zydney, A. L. Diffusive and Convective Protein Transport through Asymmetric Membranes. AIChE J. 1991, 37, 1497-1510.

12360 DOI: 10.1021/la1017278

Figure 1. Linearized plots on the basis of eqs 1 and 2 representing passage of different monomers through ESPA-1 membrane versus volume flux (in m/s) and stirring rate (in s-1). Pressure 20 bar. Monomer concentration in the feed is the range 15-45 ppm. ln[P/(1 - P)] versus JVω-0.567 (JV in m/s and ω in s-1), as shown in Figure 1 for three monomers. CPF for any stirring could then be evaluated from measured P as CPF ¼

P Pin

ð3Þ

For HEMA CPF without stirring had a finite value equivalent to eq 2 with ω = 10.6 rpm, for which eq 2 gives kd ≈ 6.2  10-6 m/s. In terms of the thickness of the unstirred layer defined as δ = D/ kd, where D ≈ 5.9  10-10 m2/s is the diffusion coefficient of HEMA, this corresponds to δ ≈ 95 μm (see Supporting Information for details). This thickness is consistent with the well-known practical limit on the maximal thickness of stagnant layers of 50-100 μm imposed by natural convection.30 The stirring rate (30) Gileadi, E., Ed. In Electrode Kinetics for Chemists, Chemical Engineers, and Material Scientists; VCH: New York, 1993; p 597.

Langmuir 2010, 26(14), 12358–12365

Bernstein et al.

Figure 2. ATR-FIR spectrum of ESPA-1 membrane, unmodified and modified using HEMA concentrations 0.02 and 0.05 M under pressure (20 bar) and 0.075 and 0.1 M without pressure. Initiators K2S2O82 and Na2S2O5 4 mM each, reaction time 30 min. ω = 10.6 rpm was then used to evaluate the effective CPF without stirring (see Table 1). 2.3. Surface Characterization of Membranes. The degree of grafting (DG) of modified membranes was determined by attenuated total reflection Fourier transform infrared (FTIRATR) spectroscopy and defined as follows Imon DG ¼ ð4Þ Imem where Imon is the intensity of the 1724-1728 cm-1 band assigned to carbonyl group and characteristic of acrylic monomers and polymers and Imem is the intensity of the 1586 cm-1 band of polysulfone (part of the original membrane). The 1586 cm-1 band usually changes insignificantly upon modification, unless the thickness of the grafted layer is commensurable or exceeds the penetration depth of evanescent IR wave (∼1 μm).31 Figure 2 shows several typical spectra of membranes, original and modified with HEMA for different modification conditions. ATR-FTIR spectra (average of 40 scans at 4 cm-1 resolution) were recorded on a Vertex 70 FTIR spectrometer (Bruker) using a Miracle ATR attachment with a one-reflection diamond-coated KRS-5 element (Pike). The reported DGs are the average of at least five spots on each sample for at least five different samples, the errors being standard deviations. Since physical adsorption of homopolymer could also contribute to Imon, this contribution was evaluated by removing adsorbed or loosely attached polymer using Soxhlet extraction with ethanol for 24 h and measuring DG before and after extraction. Unless stated otherwise, the differences were negligible which indicated that the grafted layer was covalently bonded to the membranes surface. AFM images of modified and nonmodified samples were recorded under water using a Nanoscope 3D Multimode AFM microscope (Veeco) with NP-S cantilevers (Veeco) in the tapping mode. The temperature of the sample chamber (25 C) was controlled and monitored during the scans. The area of elevated features (patches) on the membrane surface was measured by image analysis using Adobe Photoshop and averaged for at least three different images and 30 patches on every image. Statistical significance of the differences was verified by means of the F-criterion using ANOVA and the Post Hoc test at significance level p = 0.05. The contact angles measured with a sessile drop of water using an OCA-20 contact angle analyzer (DataPhysics) equipped with a video camera, image grabber, and data analysis software. Every measurement was repeated and averaged for at least 7 drops (0.25 μL) on each membrane sample.

3. Results and Discussion 3.1. Concentration Polarization of Monomers and Initiators. Table 1 summarizes intrinsic rejection, mass transfer coefficients, (31) Harrick, N. J. In Internal Reflectance Spectroscopy; John Wiley and Sons: New York, 1967.

Langmuir 2010, 26(14), 12358–12365

Article

Figure 3. Degree of grafting and relative permeability (J/J0) of ESPA-1 membrane as a function of transmembrane pressure during modification. Monomer concentration 0.035 M HEMA, initiators K2S2O8 and Na2S2O5 4 mM each. Reaction time 30 min.

and CPF factors for the three monomers in conditions of filtration without stirring. CPF without stirring was calculated using eq 2 with ω = 10.6 rpm (see section 2.2 and Supporting Information). It is seen that the effect of CP increases the concentration of HEMA and EGDMA monomers near the membranes by about an order of magnitude compared to the bulk. The effect was only somewhat smaller for MM due to its smaller size and then higher mass transfer coefficient and higher passage through the membrane (cf. eq 1). In general, the sizes of many acrylic monomers are not very different (apart from macromonomers such as PEGMA); hence, they should have similar mass transfer coefficients and comparable rejections, highly rejected charged and macro-monomers being an exception. The obtained CPF values are then indicative of the CP levels and enhancement of the kinetics expected for many acrylic monomers for the membrane and pressure used. Note that the actual CPF could decrease during grafting due to a drop in flux caused by the increased osmotic pressure and the added resistance of a new polymer layer being formed on the membrane surface. Table 1 also shows CPF for the initiators estimated by taking for Pin the P value measured at stirring rate, 750 rpm. The concentration of initiators strongly increases near the membrane surface as a result of polarization. The much lower value for metabisulfite versus persulfate is mainly due to higher kd and Pin (0.12 vs 0.02) and is apparently the result of hydrolysis and incomplete dissociation of metabisulfite, a relatively weak acid (pKa = 1.89). However, as shown and discussed below, the relation between initiator concentrations and grafting rate is not straightforward; thereby, increased concentration of initiators does not necessarily mean accelerated grafting. 3.2. Effect of Concentration Polarization on Grafting. Figure 3 illustrates the effect of applied transmembrane pressure and resulting polarization on DG and membrane permeability for modification of ESPA-1 membrane with HEMA. It is seen that for the given initial solution composition and polymerization time the grafting exponentially increases with applied pressure. Concurrently, the permeability presented as the ratio between the water fluxes of modified (J) and original (J0) membranes decreases. The modification begins to affect the permeability significantly at pressures 15-20 bar, which corresponds to DG > 0.1. At 30 bar the polyHEMA layer already became so thick that a substantial attenuation of the 1586 cm-1 band of polysulfone of the original ESPA-1 membrane was observed. Such a high DG and the associated loss of flux are obviously too large for foreseen applications, but it well demonstrates how the concentration polarization affects the polymerization rate and the amount of grafted polymer at the surface. DOI: 10.1021/la1017278

12361

Article

Figure 4. Grafting of ESPA-1 with (20 bar) and without pressure for different HEMA concentrations. Initiators, K2S2O8 and Na2S2O5, 4 mM each. Reaction time 30 min, no stirring.

Interestingly, while the permeability dropped upon modification, the salt rejection was not impaired and even rose with DG from 96 ( 1.5% to 97.5 ( 1.5%. Note that, in contrast to the systematic exponential increase of grafting with pressure, the permeability decreased with DG in a nonlinear manner. It dropped already for moderate grafting at 10-15 bar and thereafter decreased less significantly as the DG continued to exponentially increase with pressure. This effect was attributed to the inherent nonuniformity of the membrane32,33 and of the grafting process that proceeds as plugging of “defects”, i.e., more permeable spots of the membrane. It was discussed by Ben-David et al. in the context of modification of a NF membrane for improved rejection of organic pollutants using the present approach34 and was also observed by Louie et al.35 in their experiment on gas permeation through polyamide RO membrane after defect plugging. Figure 4 displays the degree of grafting as a function of concentration of monomer (HEMA) in the absence of a crosslinker. For comparison, the results for regular (no pressure) and CP-enhanced (pressure 20 bar) conditions are shown. It is seen that grafting is about an order of magnitude lower without pressure and shows about linear dependence on the monomer concentration. Substantial grafting (DG g 0.1) in these conditions is only obtained for HEMA concentrations above about 0.1 M, in agreement with the previous reports by Belfer et al.25 However, under 20 bar for otherwise identical conditions DG is higher by an order of magnitude and DG of about 0.1 is reached already at 0.02 M HEMA. It is noteworthy that the increase of DG with HEMA concentration is nonlinear and steeper than in regular conditions, as indicated by a larger exponent, which suggests that the accelerating effect of CP is more complex than merely increasing the concentration of monomer and initiators. The nonlinear kinetics under CP conditions may be explained by continuous evolution of larger oligomeric and polymeric species in the bulk solution. They more readily accumulate at the surface than monomers due to higher rejection by the membrane and slower outward diffusion (smaller kd). They are then likely to continue to react and end up attached to the surface in a manner resembling the “grafting to” approach. Oligomers and small (32) Song, Y.; Sun, P.; Henry, L. L.; Sun, B. Mechanisms of Structure and Performance Controlled Thin Film Composite Membrane Formation Via Interfacial Polymerization Process. J. Membr. Sci. 2005, 251, 67-79. (33) Singh, P. S.; Joshi, S.; Trivedi, J.; Devmurari, C.; Rao, A. P.; Ghosh, P. Probing the Structural Variations of Thin Film Composite RO Membranes obtained by Coating Polyamide Over Polysulfone Membranes of Different Pore Dimensions. J. Membr. Sci. 2006, 278, 19-25. (34) Ben-David, A.; Bernstein, R.; Oren, Y.; Belfer, S.; Dosoretz, C.; Freger, V. Facile Surface Modification of Nanofiltration Membranes to Target the Removal of Endocrine-Disrupting Compounds. J. Membr. Sci. 2010, 357, 152-159. (35) Louie, J. S.; Pinnau, I.; Reinhard, M. Gas and Liquid Permeation Properties of Modified Interfacial Composite Reverse Osmosis Membranes. J. Membr. Sci. 2008, 325, 793-800.

12362 DOI: 10.1021/la1017278

Bernstein et al.

Figure 5. OD at 600 nm versus time for homopolymerization of 0.035 M (solid line) and 0.1 M (dashed line) HEMA solutions. Symbols represent OD of bulk solution after 30 min of membrane modification under 20 bar pressure for the feed initially containing: (a) 0.02 M HEMA, (b) 0.035 M HEMA, and (c) 0.035 M HEMA with 0.2% w/w EGDMA. Initiators K2S2O8 and Na2S2O5, 4 mM each. Table 2. Elemental Composition (in atomic percent) and Ratios Obtained by XPS for ESPA-1 Membrane Modified with HEMA at Different DG DG

C

N

O

O/N

C/O

contact angle (deg)

0 0.07 0.18 0.35 0.85

73.1 72.2 71.3 70.5 67.8

10.2 6.0 4.3 3.0 1.5

16.7 21.8 24.3 26.3 30.5

1.6 3.6 5.7 8.7 20.9

4.4 3.6 2.9 2.7 2.2

50 ( 3 44 ( 3 55 ( 1 65 ( 3

soluble polymers might be too small to scatter light or change solution rheology significantly; however, their solubility drops with size, and large light-scattering aggregates may start forming at some critical size. Simple turbidity or optical density (OD) measurements may then allow a rough comparison of the typical times that takes for the larger species to evolve for given initial conditions.36 Figure 5 displays OD at wavelength 600 nm versus time for 0.035 and 0.1 M HEMA solutions. For comparison, it also shows OD of 0.02 and 0.035 M HEMA bulk solutions after 30 min of actual modification using 20 bar pressure. Without applied pressure 0.1 M solution yielded about the same grafting as 0.02 M solution under 20 bar (cf. Figure 4). A large amount of insoluble polymer quickly formed in the bulk of the 0.1 M solution (dashed curve in Figure 5), yet it could not contribute to surface grafting as much as smaller species in the 0.02 M solution that stayed transparent all the way to the end of reaction (point “a” in Figure 5). In turn, under CP conditions, the 0.02 M solution was far inferior to the 0.035 M solution, in which polymeric aggregates had already evolved at the end of grafting (solid curve and point “b” in Figure 5), leading to additional enhancement observed in Figure 4. Clearly, this nonlinear effect complicates the process but may be well offset by much more efficient utilization of the monomers. 3.3. Surface Characteristics of HEMA-Grafted Membranes. In addition to permeability and ATR-FTIR, some insight into modification with HEMA was gained through three other surface-sensitive methods: contact angle, XPS, and AFM. Table 2 lists the elemental composition and contact angle of the ESPA-1 surface modified with HEMA at different DG measured by XPS. The polyamide layer membrane has a theoretical composition of C:N:O = 75:12.5:12.5; however, XPS, sensitive (36) Flory, P. J. In Principles of Polymer Chemistry; Cornell University Press: Ithaca, NY, 1953.

Langmuir 2010, 26(14), 12358–12365

Bernstein et al.

Article

Figure 6. Tapping mode topographic AFM images under water of ESPA-1 membrane: (a) original, (b) after HEMA modification, DG 0.2, and (c) after HEMA modification, DG 0.8. Image size 10  10 μm2, Z-scale 500 nm.

to the composition of the outmost layer about 1 nm thick, typically yields a proportion closer to 73:10:17,37,38 which is indicative of the presence of carboxylic group in the topmost part of polyamide. As DG increased, the C/O ratio dropped from 4.35 to 2.2. However, the latter value obtained for the highest DG might not be a good indicator of surface coverage, since the C:O ratio for polyHEMA can also deviate from the theoretical value 2:1.39 The content of nitrogen and the O/N ratio seem to be better indicators of the surface coverage with polyHEMA. As DG increased from 0 to 0.85, the N content was gradually reduced from 10% to 1.5% and the O/N ratio rose from 1.6 to 20.9. This indicates a nearly complete surface coverage, but some uncovered patches could remain even at the highest grafting. This appears to agree with the “defect plugging” picture discussed by Ben-David et al.34 Table 2 also displays the results of contact angle measurements, which agrees with the gradual surface compositional changes observed by XPS. The water-air contact angle (CA) measured by the sessile drop method slightly decreased at low DG but then increases at higher DGs. The obtained CA values are in agreement with other reports, in which initially hydrophobic surfaces after grafting with HEMA showed sessile drop contact angles in the range 55°-60°.19,40,41 Although HEMA is considered a hydrophilic monomer and hydrated HEMA-grafted surfaces show much lower contact angles by the captive bubble method, in the sessile drop method the drop is placed on an initially dry surface, which may be more hydrophobic since it is depleted of polar groups that tend to rearrange and face preferentially inward.19,42 At high DG the layer of polyHEMA may then have more freedom for such surface rearrangements, and the measured CA may increase. (37) Kwak, S. Y.; Jung, S. G.; Kim, S. H. Structure-Motion-Performance Relationship of Flux-Enhanced Reverse Osmosis (RO) Membranes Composed of Aromatic Polyamide Thin Films. Environ. Sci. Technol. 2001, 35, 4334-4340. (38) Tang, C. Y.; Kwon, Y. N.; Leckie, J. O. Probing the Nano-and MicroScales of Reverse Osmosis membranes;A Comprehensive Characterization of Physiochemical Properties of Uncoated and Coated Membranes by XPS, TEM, ATR-FTIR, and Streaming Potential Measurements. J. Membr. Sci. 2007, 287, 146-156. (39) Beamson, G. Conformation and Orientation Effects in the X-Ray Photoelectron Spectra of Organic Polymers. J. Electron Spectrosc. Relat. Phenom. 2001, 121, 163-181. (40) Shim, J. K.; Na, H. S.; Lee, Y. M.; Huh, H.; Nho, Y. C. Surface Modification of Polypropylene Membranes by γ-Ray Induced Graft Copolymerization and their Solute Permeation Characteristics. J. Membr. Sci. 2001, 190, 215-226. (41) Jones, D. S.; Lorimer, C. P.; McCoy, C. P.; Gorman, S. P. Characterization of the Physicochemical, Antimicrobial, and Drug Release Properties of Thermoresponsive Hydrogel Copolymers Designed for Medical Device Applications. J. Biomed. Mater. Res., Part B 2008, 85, 417-426. (42) Holly, F. J.; Refojo, M. F. Wettability of Hydrogels I. Poly (2-Hydroxyethyl Methacrylate). J. Biomed. Mater. Res. 2004, 9, 315-326.

Langmuir 2010, 26(14), 12358–12365

From topographic AFM images in Figure 6 it is visible that the membrane morphology of the unmodified and modified membranes was different. While going from unmodified membrane to DG 0.2 and then 0.8, the average area of elevated features (patches) changes from about 0.3 μm2 to 0.75 μm2 and then 0.9 μm2. At the same time the rms roughness rapidly increases from about 80 to 100-110 nm at low and medium grafting and shows little further variation for high grafting. These differences were found statistically significant with P values of at least as low as 0.05. The observed morphological changes seem to indicate that the new material deposited on the surface upon modification tends to form a rather patchy layer, in agreement with the other data. 3.4. Effect of Initiators. The rate of radical polymerization depends on the concentration of the redox initiators in a complex and often nonmonotonous way. Apart from faster initiation, increased concentration of initiators [I] proportionally increases the number of simultaneously growing polymer chains (thus less monomer is available per chain) and the rate of termination increases as [I]2. These effects may be largely enhanced by large CPF values therefore there may be an optimum for achieving both sufficient number and length of the chains. A high concentration of initiators may also increase the osmotic pressure and thus reduce the flux and lower CP for a given applied pressure. Finally, too much initiator may promote more homopolymerization relative to surface grafting.43 Figure 7 summarizes the results that show the effect of the concentration of each initiator on grafting. DG increases steadily but only slightly with potassium persulfate concentration at a constant metabisulfite concentration (4 mM). The effect of persulfate concentration appears even weaker, if one recalls its CP is very large and indeed a reasonable grafting is obtained even for persulfate concentration as low as 0.4 mM. The grafting was however much more sensitive to the metabisulfite concentration, when persulfate level was fixed at 4 mM. The effect was not monotonous, and even though the grafting was negligible at 0.4 and 1 mM K2S2O5, DG showed a sharp maximum at the next lowest concentration 2 mM and decreased for higher concentrations. It appears that it is the metabisulfite concentration that should be mostly considered in relation to the competition between the rates of initiation and termination and monomers available per growing chain. It also points to the need to optimize initiator levels to achieve desired grafting. (43) Kangwansupamonkon, W.; Gilbert, R. G.; Kiatkamjornwong, S. Modification of Natural Rubber by Grafting with Hydrophilic Vinyl Monomers. Macromol. Chem. Phys. 2005, 206, 2450-2460.

DOI: 10.1021/la1017278

12363

Article

Figure 7. Effect of concentration of redox initiators on grafting: cross-hatched bars, DG vs metabisulfite concentration for 4 mM persulfate; filled bars, DG vs persulfate concentration for 4 mM metabisulfite. Monomer 0.035 M HEMA, reaction time 30 min, pressure 20 bar. Asterisks indicate negligibly small DG for designated metabisulfite concentrations and 4 mM persulfate.

Figure 8. DG at different modification time with 0.2% (w/w per monomer) and without cross-linker (EGDMA), membrane ESPA-1, solution 0.035 M HEMA, 4 mM K2S2O8, and Na2S2O5. Pressure 20 bar.

3.5. Effect of Cross-Linker. Polymerization of monofunctional monomers such as HEMA proceeds as a roughly linear addition of monomer and produces predominantly a linear polymer. Disregarding entanglements and branching effects, such polymers are attached to the surface at one end. Such attachment may prove insufficiently strong, and the grafted layer may not be as dense as desired, for instance, if the layer is meant to contribute to rejection of certain solutes. In such cases addition of a crosslinker may strengthen the attachment of the grafted layer, prevent excessive swelling, and improve its sieving (size-excluding) properties. As a 2-functional monomer, cross-linker may also modify the kinetics and, by forming branched and more compact structures, promote faster precipitation and still higher local concentration of the growing polymers at the surface. Figure 8 compares the variation of DG with time of polymerization without and with addition of a cross-linker (EGDMA) in otherwise identical conditions. Notably, EGDMA is only sparingly soluble and then its fraction was small (0.12 mM or 0.2% per monomer weight). Nevertheless, it was expected to concentrate at the surface (CPF ∼ 10, see Table 1), and Figure 8 shows its dramatic effect on the degree of grafting and kinetics. Without the cross-linker grafting became noticeable after about 10 min, simultaneously with a drop in flux, and thereafter 12364 DOI: 10.1021/la1017278

Bernstein et al.

Figure 9. ATR-FTIR spectrum of ESPA-1 modified with 0.7 and 0.85 mM EEMA with 5% EGDMA (w/w per monomer). Reaction time 30 min, initiators K2S2O5 and K2S2O5 4 mM each, pressure 20 bar.

DG approximately linearly increased with time. In contrast, in the presence of cross-linker a noticeable grafting was observed after 7.5 min. Thereafter, DG steeply increased until after 30 min the grafted layer became so thick that the polysulfone bands virtually disappeared from the ATR-FTIR spectrum. The OD of the bulk solution taken at that time was also higher than for solution without EGDMA (point “c” in Figure 5). This example demonstrates that addition of a cross-linker may be highly advantageous for achieving a rapid and robust modification. The very large grafting obtained suggests that the amounts of monomer and cross-linker in solution can be largely reduced, while a significant and faster modification may still be achieved. 3.6. Modification with Different Monomers. The advantages of the proposed approach could be also demonstrated with other monomers. PEGMA is commonly used for PEGylation of surfaces, which lowers proteins adhesion and imparts antifouling properties.44 Because of large size and relatively low fraction of reactive groups reasonable modification usually requires concentrations of at least 0.1 M. In contrast, under pressure significant grafting was readily achieved already at concentrations 0.0075-0.015 M and yielded DG of 0.07-0.2. The large size of PEGMA was clearly beneficial in this case, since it results in high rejection and low kd and hence very high CPF. The modification increased moderately with monomer concentration up to 0.02-0.03 M where the membrane was abruptly covered with a dense gel-like layer. The addition of cross-linker to PEGMA resulted in a still higher DG and formation a thick hydrogel layer on the membrane surface. Notably, no gel was formed in the bulk solution, as commonly occurs at larger PEGMA concentrations without pressure. As for HEMA, the surface modification caused a reduction in the membrane permeability along with improved salt rejection and a similar morphological changes observed by AFM. Surface analysis of the modified PEGMA membrane showed that the CA decreased from 50 ( 1° (ESPA-1) to 35°-38° even at DG as low as 0.1. The C/O ratio at high modification, measured with XPS, changed from 4.36 (ESPA-1) to 2.33, close to the theoretical value 2, similar to what was observed with HEMA. The example of EGDMA (section 3.5) suggests that the CP effect may allow a facile modification using a broad range of sparingly soluble monomers. Such monomers usually require addition of a cosolvent (e.g., an alcohol) to achieve a reasonable monomer concentration and reaction rate, which could be costly (44) Susanto, H.; Ulbricht, M. Photografted Thin Polymer Hydrogel Layers on PES Ultrafiltration Membranes: Characterization, Stability, and Influence on Separation Performance. Langmuir 2007, 23, 7818-7830.

Langmuir 2010, 26(14), 12358–12365

Bernstein et al.

Article

or detrimental to the membranes. However, consumption of chemicals is greatly reduced in the present method with minimal impact on the membrane. As an example, EMA having solubility 0.5% (0.1 M) was grafted onto ESPA-1 using concentrations in the range 4-10 mM. It is important to note that under CP conditions the surface concentration of monomer may exceed its solubility. Thus, the monomer may precipitate and rapidly polymerize forming a loosely bound polymer. It was indeed found that the 1720 cm-1 IR band largely decreased after Soxhlet extraction; i.e., most of the grafted poly-EMA was not permanently bonded to the membrane and removed. However, this problem was resolved by adding a small amount of cross-linker (5-10% per monomer weight), and no loss of graft-polymer was observed after Soxhlet extraction. Similarly, EEMA has solubility in water 1.35% (85 mM) and even at this maximal concentration does not show any grafting without pressure. However, it was readily grafted at concentrations g100 times lower than the solubility limit when pressure was applied, as the 1720 cm-1 band in the ATR-FTIR spectra in Figure 9 indicates. As previously, addition of 5% cross-linker to the monomer assisted in enhancing the kinetics and permanently bonding the grafted layer to the membrane.

4. Conclusions It was demonstrated that many drawbacks of the conventional redox-initiated radical graft polymerization, such as slow kinetics, high reagents consumption, and homopolymerization, could be reduced significantly by utilizing the selectivity of RO membranes toward monomers and carrying out the polymerization while applying pressure, i.e., by filtering the

Langmuir 2010, 26(14), 12358–12365

monomer solution through the membrane. As a result, the reagent concentrations near the membrane surface increased drastically, and the reaction was preferentially directed toward surface grafting. 1-2 orders of magnitude smaller amounts of the monomer and cross-linkers were thus required in order to reach a satisfactory degree of grafting, and competitive homopolymerization could be lessened significantly. Notably, along with drastic enhancement, the kinetics of the process became more complex and was affected in a nonlinear manner by redox initiators and growing oligomeic and polymeric species evolving in the bulk solution. However, as an additional benefit, this modified approach was shown here and previously34 to broaden the monomers spectrum to sparingly soluble hydrophobic, charged, and macro-monomers, whose polymerization in a regular way is often not feasible due to various kinetic and thermodynamic constraints. Addition of a small fraction of cross-linker to such monomers could help to stabilize the grafted layer and improve its attachment to the surface. Acknowledgment. The authors thank R. Fainshtein and A. Mamoutov for the assistance with the ATR-FTIR and TOC measurements. Hydranautics are gratefully acknowledged for supplying membrane samples. This work was supported by a grant from the Water Authority of Israel. Supporting Information Available: Estimation of the diffusion coefficients, equivalent stirring rate and thickness of boundary layer in stagnant conditions. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la1017278

12365