Journal of Membrane Science

Transcription

1 Journal of Membrane Science 327 (2009) Contents lists available at ScienceDirect Journal of Membrane Science journal homepage: Characteristics, performance and stability of polyethersulfone ultrafiltration membranes prepared by phase separation method using different macromolecular additives Heru Susanto a,b,, Mathias Ulbricht a a Lehrstuhl für Technische Chemie II, Universität Duisburg-Essen, Essen, Germany b Department of Chemical Engineering, Universitas Diponegoro, Semarang, Indonesia article info abstract Article history: Received 30 June 2008 Received in revised form 5 October 2008 Accepted 9 November 2008 Available online 24 November 2008 Keywords: Ultrafiltration membrane Membrane preparation Polyethersulfone Polyvinylpyrrolidone Poly(ethylene glycol) Poly(ethylene glycol)-b-poly(propylene glycol)-b-poly(ethylene glycol) Pluronic Polyethersulfone (PES) ultrafiltration (UF) membranes were prepared by non-solvent-induced phase separation (NIPS) method using different macromolecular additives, polyvinylpyrrolidone (PVP), poly(ethylene glycol) (PEG) and poly(ethylene oxide)-b-poly(propylene oxide)-b-poly(ethylene oxide) (Pluronic, Plu). Their effects on membrane structure and their stability in the polymer membrane matrix as well as the resulting membrane performance were systematically compared in order to determine the additive that should be preferred. The investigated membrane characteristics include surface hydrophilicity (by contact angle), surface charge (by zeta potential), surface chemistry (by FTIR spectroscopy), water flux and rejection of macromolecular test substances. Visualization of membrane surface and crosssection morphology was also done by scanning electron microscopy. The membrane performance was examined by investigation of adsorptive fouling and ultrafiltration using solution of bovine serum albumin as the model system. The stability of additive was examined by incubating the membrane in water (40 C) and sodium hypochlorite solution. Modification effects on membrane characteristic as well as performance via blending membrane polymer and macromolecular additive were clearly observed. Overall, the results suggest that Pluronic showed the best behavior in all respects, and, consequently, it should be considered in practical applications Elsevier B.V. All rights reserved. 1. Introduction Ultrafiltration (UF) has become the main focus as promising separation tool in many industrial processes covering fractionation and concentration steps in the food, in pharmaceutical and biotechnological industries, in pure water production and in water and wastewater treatments [1,2]. As consequence of this increasing demand, efforts to improve UF process performance are gaining more and more importance. In general, those efforts include feed pretreatment, advanced membrane and module design, and process condition optimization. However, in many cases, the key for the performance of UF process is the membrane itself. In this regard, three important characteristics for achieving high performance UF membrane are high flux as well as selectivity, low fouling and performance stability for long-term operation. Polymeric membranes prepared by non-solvent-induced phase separation (NIPS) are still dominating commercially available UF membranes. Among them, polysulfone (PSf) and polyethersulfone Corresponding author. Tel.: ; fax: address: heru.susanto@uni-due.de (H. Susanto). (PES) UF membranes are broadly manufactured for industrial applications. Nevertheless, the hydrophobicity of those materials can cause severe fouling problems; therefore, membrane modification is usually done to increase the membrane resistance towards fouling. Three different approaches including (i) membrane polymer modification (pre-modification), (ii) blending of the membrane polymer with a modifying agent (additive), and (iii) surface modification after membrane preparation (post-modification) have been proposed [3]. Even though blending technique can involve significant changes in composition of the casting solution, leading to different membrane structure formed during the phase separation and, consequently, membrane properties can be quite different from the unmodified reference material, it is simple and no additional step is needed during membrane manufacturing. In addition, although stability of the modifying agent in the membrane matrix can be a problem, the effect of hydrophilization of the polymer membrane can clearly be observed. Therefore, this technique seems to be of highest relevance from practical point of view. Polymeric additives (usually hydrophilic polymers) in a casting solution are also used in order to increase both pore size and porosity (poreforming agent) and to suppress macrovoid formation. However, depending on the polymer membrane, solvent and NIPS conditions, the opposite effect can also be observed /$ see front matter 2008 Elsevier B.V. All rights reserved. doi: /j.memsci

2 126 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Polyvinylpyrrolidone (PVP) and poly(ethylene glycol) (PEG) have been intensively used as additives during preparation of PES UF membranes by phase separation methods. The mechanism of PES membrane formation with addition of PVP had well been explained in many publications [4 8]. Addition of PVP into PES NMP (N-methyl-2-pyrrolidone) casting solution could suppress the macrovoid formation via decreasing the effect of delayed demixing [7]. However, opposite effect, i.e., enlargement of the macrovoid structure by addition of PVP into PES DMF (dimethylformamide) was also reported [8]. Increasing water permeability as well as molecular weight cut off (MWCO) of PES membrane by the addition of PVP was another observation [9]. Addition of PVP to PES NMP led to higher water permeability than without PVP but the solute rejection (for PEG) was 15% lower than without PVP. Moreover, it had been found that the characteristic and performance of PES PVP blend membranes were influenced by the concentration and the molar mass of PVP [4,10,11]. The highest product permeation rate examined by using PEG solution was obtained at PVP/PES weight ratio of unity [4]. Observed retention coefficient as well as surface roughness increased with increasing molar mass of PVP [11]. The effect of PVP content on the resulting fouling behavior examined with protein was also investigated; the presence of PVP could to some extent decrease degree of fouling of the resulting membrane [12]. PEG with various molar masses was added during preparation of PES UF membranes [13]. The performance of the resulting membranes was influenced by both molar mass and concentration of PEG. Membranes prepared with higher molar mass of PEG had higher pure water permeation and larger pores. The water permeation increased as concentration of PEG (400 and 600 g/mol) was increased, while solute separation decreased. Apparently, an optimum condition (high flux with acceptable solute rejection) was achieved at PEG concentration of 10 wt%. In addition, differences in surface morphology and roughness could also be detected. A similar study was also performed by using PSf NMP and PSf DMAc (dimethylacetamide) system [14]. The results showed that the increase in PEG molar mass increased the membrane porosity leading to increase in water permeability for both systems. Liu et al. [15] found an optimum PEG content in PES NMP system for increasing water permeability, and further increase in PEG content would decrease the resulting water flux. Moreover, the addition of PEG alone could not suppress macrovoid formation even at high concentration. Only when relatively large amounts of water were added to the dope solution, sponge-like membrane structure was obtained. The effect of PEG/NMP ratio during preparation of PSf UF membranes was investigated [16]. As the PEG content was increased the pure water flux increased while the solute rejection would decrease. Recently, the group of Jiang has intensively proposed the use of a triblock copolymer, poly(ethylene oxide)-b-poly(propylene oxide)- b-poly(ethylene oxide) (Pluronic/Plu), as another attractive additive for preparation of PES UF membranes [17 20]. They observed that Pluronic could improve not only the flux but also the resistance towards fouling. The resulting fouling resistance was significantly influenced by Pluronic content and PEO chain length [19]. It was reported that the apparent protein adsorption amount decreased significantly with increasing the Pluronic content and reached 0 g/cm 2 at concentration of 10.5% [17]. By contrast, the protein rejection and pure water flux were little influenced by the Pluronic content. Very recently, the function of Pluronic as both hydrophilic modifier and pore-forming agent has been confirmed [20]. In sum, three different macromolecular additives, i.e., PVP, PEG and Plu, have frequently been used in the preparation of PES UF membranes including a wide variety in preparation conditions and polymer characteristics. Those additives determine the characteristics as well as the performance of the resulting membrane, and several cases have been reported where high performance membranes could be obtained. However, considerable disagreement was also observed among different reports. We believe that this is because many parameters including characteristics of membrane polymer (e.g., molar mass, concentration in casting solution), solvent (e.g., solubility parameter, viscosity), additive (e.g., molar mass, concentration), and preparation conditions (e.g., temperature, humidity) are involved. Therefore, it is hard to determine the best additive that should be used in preparation of PES UF membrane based on results of previous studies. Torrestiana-Sanchez et al. [10] had briefly compared the use of PVP and PEG as nonsolvents during PES membrane preparation. However, the reported data were very limited to water permeability and protein rejection. Other important parameters such as hydrophilicity, fouling behavior and membrane stability have not yet been compared. In the present study, the effects of those macromolecular additives on characteristics, performance and stability of the resulting PES UF membranes are systematically compared. All preparation conditions, including composition of membrane casting solution as well as molar mass of additive were maintained to be similar. Stability study was included because the immobilization of hydrophilic additives in the polymeric membrane matrix is one of the critical issues from a practical point of view. Thus, the best performing hydrophilic macromolecular additive for preparation PES UF membranes will be known based on the results of this work. 2. Experimental 2.1. Materials Commercial PES (Ultrason E 6020 P) donated by BASF (Ludwigshafen, Germany) was used and dried at 120 C for at least 4 h before use. N-methyl-2-pyrrolidone (NMP) was purchased from Merck (Hohenbrunn, Germany). Polyvinylpyrrolidone (MW 10,000 g/mol) was purchased from Serva Feinbiochemica GmbH&Co. (Heidelberg, Germany). PEG (MW 10,000 g/mol) was purchased from Fluka Chemie AG (Buchs, Germany) and Pluronic F127 (Plu) (MW 12,600 g/mol) was purchased from BASF (Mount Olive, NJ, USA). Bovine serum albumin (BSA) and sodium hypochlorite were purchased from ICN Biomedicals, Inc. (California, USA) and Sigma Aldrich Chemie GmbH (Steinheim, Germany), respectively. Dextrans T-4, T-15, T-35, T-100 and T-200 (the numbers indicating molar mass in kg/mol) were from Serva Feinbiochemica GmbH&Co. (Heidelberg, Germany). Potassium dihydrogen phosphate (KH 2 PO 4 ) and disodium hydrogen phosphate dihydrate (Na 2 HPO 4 2H 2 O) were purchased from Fluka Chemie AG (Buchs, Germany). Potassium chloride (KCl), potassium hydroxide (KOH), and hydrochloric acid (HCl), all of p.a. quality, were purchased from Bernd Kraft GmbH (Duisburg, Germany). Nitrogen gas purchased from Messer Griesheim GmbH (Krefeld, Germany) was ultrahigh purity. Water purified with a Milli-Q system from Millipore was used for all experiments Membrane preparation PES (15 wt%) was dissolved in NMP (75%) with stirring, and different additive with similar molar mass (PVP, PEG or amphiphilic triblock copolymer Plu; 10 wt%) was added to the polymer solution. Attention should be given to PEG because it would not dissolve in the NMP at ambient temperature. Therefore, slight heating ( 45 C) was used during dissolution. This relative high concentration of additive was aimed to see significant effects on the resulting membrane. Polymer solution without an additive was also prepared for control experiments. The homogenous polymer solution was left without stirring until no bubbles were

3 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) observed. The membranes were prepared by using Coatmaster 509 MC, Erichsen Testing Equipment. The polymer solution was cast with a thickness of 200 m using a steel casting knife on a glass substrate (casting speed 25 mm/s) and subjected to humid air (RH = 50 60%) for 1 min. Thereafter, the proto-membrane was solidified in a coagulation bath containing water (20 ± 1 C) for 1 h. The resulting membranes were washed and soaked in the water for 24 h before drying. Drying was sequentially done by immersing in water, water/ethanol, ethanol and hexane Shrinkage measurements Membrane shrinkage was calculated by measuring the area of the initial film before solidification ( proto-membrane ) and the membrane after solidification as well as after drying. These areas were then compared to calculate degree of shrinkage ( plane shrinkage ). In addition, membrane shrinkage was also calculated based on the resulting membrane thickness and initial film thickness ( vertical shrinkage ). ( degree of shrinkage (plane) (%) = 1 A ) (1) A 0 ( degree of shrinkage (vertical) (%) = 1 l ) (2) l 0 where A 0 is the area of proto-membrane, A 1 is the area of membrane after solidification or after drying and l 0 and l 1 are the thickness of membrane before solidification and after drying, respectively. The membrane thickness was measured by Coolant Proof micrometer IP 65, Mutico Co., Japan Compaction, hydraulic permeability measurement, adsorptive fouling and ultrafiltration procedures All experiments were carried out by using a dead-end stirred cell filtration system (Amicon cell model 8010 for compaction, hydraulic permeability measurement and adsorptive fouling, model 8050 for UF experiments) connected to a reservoir ( 450 ml) and pressurized by nitrogen from a gas tank. Membrane compaction was performed by filtration of pure water at 450 kpa for 2 h. The flux profile over time was monitored online gravimetrically. In all further experiments, the membrane was firstly compacted for at least 1 h (this time was enough to achieve steady flux, cf. Section 3.2). Hydraulic membrane permeability was measured at different transmembrane pressures within the range kpa and at least five measurements from different membrane samples were averaged. For static adsorption experiments (adsorptive fouling), a solution of BSA (1 g/l, ph 7 in phosphate buffer) was added to the cell and the outer membrane surface was exposed for 3 h without any flux at a stirring rate of rad/s (300 rpm). Afterwards, the solution was removed, and the membrane surface was rinsed two times by filling the cell with pure water (5 ml) and shaking it for 30 s. Water fluxes before and after exposure were measured at the same pressure (300 kpa). The evaluation of membrane performance was expressed in terms of the relative flux reduction, RFR (Eq. (3)). RFR (%) = J 0 J a 100 (3) J 0 Ultrafiltration experiments at a constant transmembrane pressure (300 kpa) were conducted using a BSA solution (0.1 g/l, ph 7 in phosphate buffer) as the feed. The balance was connected to the PC, the weight of permeate was recorded online, and the flux was calculated. The permeate flux profile over time was investigated. BSA concentrations were determined by measuring its UV absorbance at 280 nm. The apparent BSA rejection was calculated using Eq. (4) Rejection measurement Rejection tests were conducted with a five component mixture of dextrans with molar mass ranging from 4 to 200 kg/mol at a total concentration of 1 g/l. Experiments were performed using a dead-end stirred filtration system (Amicon cell model 8050, Millipore; cf. Section 2.4) at a pressure of 100kPa and a stirring rate of 300 rpm. Around 10 ml of permeate was collected. The compositions of dextran mixtures in the permeate (C downstream ) and feed/retentate (C upstream ) sides of membrane were analyzed using gel permeation chromatography (GPC). The apparent rejection for each molar mass was calculated using the following equation: R = 1 C downstream C upstream (4) 2.6. Membrane morphology The top surface and cross-section morphology of the membranes were observed by using a Quanta 400 FEG (FEI) environmental scanning electron microscope (ESEM) at standard high-vacuum conditions. A K 550 sputter coater (Emitech, U.K.) was used to coat the outer surface of the sample with gold/palladium. For cross-section analysis, the membranes were broken in liquid nitrogen and sputtered for 1.5 min, while for analysis of outer membrane surface, sputtering was done for 0.5 min Contact angle (CA) Sessile drops static CA was measured using an optical contact angle measurement system (OCA 15 Plus; Dataphysics GmbH, Filderstadt, Germany). Five microlitres of water was dropped on the membrane surface from a microsyringe with a stainless steel needle in room temperature (21 ± 1 C). At least seven measurements of drops at different locations were averaged to obtain CA for one membrane sample Zeta potential (ZP) The membrane surface charge was investigated by an outer surface streaming potential measurement. Experiments were carried out in a flat-sheet tangential flow module described in detail in previous study [21]. Before measurement, the membrane was equilibrated by soaking in mol/l KCl solution. The streaming potentials of membranes were measured using mol/l KCl solution within the ph range 3 10 and at a temperature of C. The ZP was calculated using the Helmholtz Smoluchowski equation Surface chemistry The membrane surface chemistry was analyzed by using the instrument Varian 3100 Fourier transform infrared spectroscopy (FTIR) Excalibur series. A total of 64 scans were performed at a resolution of 4 cm 1 and the temperature of C. The Varian s Resolution Pro 4.0 was used to record the membrane spectra versus the corresponding background spectra Stability test The stability of macromolecular additive in polymer membrane matrix was examined by incubating membrane samples in water (20 and 40 C) and sodium hypochlorite solution (active chlorine concentration 400 mg/l) for up to 10 days. Water is usually used for membrane washing before chemical cleaning will be done. Hypochlorite solution had been known as one of the most popular

4 128 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Fig. 1. Plane membrane shrinkage after coagulation and drying for different membranes. PES(only), PES PVP, PES PEG and PES Plu are membranes prepared without an additive, with PVP, with PEG and with Pluronic additives, respectively. The error bars represent standard deviation. chemical cleaning agents to remove irreversible fouling [22 24]. Surface chemistry (by FTIR), surface hydrophilicity (by CA) and water flux measurements were used to investigate changes in membrane characteristics and transport property. The experiments were done by using a different membrane. In water flux stability study, the pure water flux was firstly measured (before soaking) and membranes were then soaked in hypochlorite solution in closed containers for a certain number of days. Thereafter, the membrane was rinsed and the water flux was again measured. 3. Results and discussions 3.1. Shrinkage and membrane thickness Understanding of membrane shrinkage will give information for predicting the area of a non-supported asymmetric membrane product but not for a supported asymmetric membrane. High shrinking tendency will result in tensions within the membrane, and this can have influence on structure of the resulting membranes. It was reported that shrinkage had effect on membrane porosity [25]. Fig. 1 and Table 1 show the plane and the vertical shrinkages, respectively. As clearly seen in Fig. 1 and Table 1, decreases in membrane area and thickness indicate that the PES film shrank not only in the plane view but also in the vertical direction. In plane view, PES PVP membrane showed the highest degree of shrinkage (25%) followed by PES (only). The highest shrinkage of PES PVP might contribute to smaller surface porosity as noticed by its smallest water permeability among the PES membranes prepared with an additive (cf. Section 3.2). Although the membranes have been dried by exchange with organic non-solvent (for PES) method, shrinkage after drying could still be observed. Different with plane view, all membranes prepared using an additive showed similar degree of shrinkage in vertical view. Even though the reason behind this shrinkage phenomenon is still not clearly understood, polymer contraction upon Table 1 Membrane thickness and the resulting vertical shrinkage after drying (polymer solution has been cast with 200 m thickness). No. Membrane Thickness ( m) Degree of shrinkage (%) 1 PES(only) 98 ± PES PVP 130 ± PES PEG 135 ± PES Plu 143 ± 6 29 Fig. 2. Water flux profile over time during compaction at pressure of 450 kpa. solidification may be able to explain [26,27]. It is clearly seen that different macromolecular additive would result in different polymer characteristics and consequently, polymer contraction would also be different. It should also be noted that the total polymer concentration for PES(only) was significantly smaller than the others and consequently, lower viscosity and faster solvent exchange should be obtained. Difference in miscibility between PES and additive for different additives used might also contribute. We have tried to investigate the miscibility by using differential scanning calorimetry (DSC). However, clear explanation could not be drawn from the resulting data. Residual solvent in the sample might have influence on the resulting DSC thermogram. Such phenomena seem to be found also in previously reported literature [18]. Further analysis of this miscibility is now currently under investigation. Overall, PES Plu and PES PEG membranes showed smaller shrinkage than both PES(only) and PES PVP membranes Membrane compaction, water permeability and membrane rejection Compaction a compression of the membrane structure under a transmembrane pressure difference causing a decrease in membrane permeability due to mechanical deformation of the solid polymer is a common phenomenon during application of polymeric membranes [28,29]. In this compaction study, the membranes were pressurized at high pressure (450 kpa) for 2 h. Fig. 2 shows an example of the water flux profile over time during compaction for all membranes. Indeed, gradual decrease in flux over the duration of compaction time was observed for all membranes, and flux reached a steady value after 60 min of compaction. The steady fluxes were 80%, 35%, 40% and 60% of the initial flux for PES(only), PES PVP, PES PEG and PES Plu, respectively. Interestingly, the membrane prepared by using Pluronic as the additive had initially lower flux than the membrane prepared using PEG as the additive, but beyond 30 min duration of compaction it showed higher flux. Because all membranes had asymmetric structure (cf. SEM images), the compaction at high pressure would cause densification of the more porous support layer leading to a thickening of the skin layer (selective barrier). Consequently, thicker membrane would result in lower flux. Such phenomenon had been observed in the previous reported literature [28,30]. Table 2 shows the hydraulic membrane permeability measured after 1 h of compaction. The hydraulic permeability of the wet membranes (never dried) is also included. The membrane prepared without an additive showed the highest hydraulic permeability for both conditions. It is seen that drying process significantly

5 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Table 2 Hydraulic membrane permeability measured after 1 h compaction. No. Membrane Lp (L/m 2 h kpa) after drying Lp (L/m 2 h kpa) before drying 1 PES ± ± PES PVP ± ± PRS PEG ± ± PES Plu ± ± 0.05 decreased the hydraulic permeability of all membranes. The flux reductions after solvent exchange drying were 76%, 95%, 93% and 88% for PES(only), PES PVP, PES PEG and PES Plu membranes, respectively (drying without solvent exchange resulted in much greater flux reductions). The absence of impregnation may contribute to this very large decrease in flux after drying but it should not be the only reason. Nevertheless, the effect of drying on membrane performance was beyond the scope of this paper. Because all relative flux reductions showed similar trend; all further experiments were performed and discussed by using dried membranes. In general, addition of a hydrophilic polymer can facilitate liquid demixing because the system will be closer to phase separation on the one hand, and slow down phase separation by hindering nonsolvent inflow to the polymer solvent mixture (delayed demixing) due to higher viscosity on the other hand. By contrast, the polymer solution without an additive would facilitate faster solvent exchange due to much lower viscosity. It was reported that addition of 10% of PVP or PEG into 20% of PES in NMP solution increased the casting solution viscosity up to more 200% (from 0.70 to 2.75 Pa s) [10]. In this work, the hydraulic permeability data of wet membranes showed that addition of a hydrophilic polymer increased the resulting water flux suggesting that the effect of delayed demixing was less significant compared to the effect of instantaneous liquid demixing. In order to quantify the pore size and its distribution of selective barrier, the rejection of polydisperse macromolecular test substances (dextran) was then measured. Fig. 3 shows that the addition of a macromolecular modifier into polymer solution changed the rejection curve of the membrane. As shown in Fig. 3, the membrane prepared without an additive had a steeper separation curve implying that the pore size distribution is narrower than for the membranes prepared with an additive. In addition, it had smaller MWCO but showed the greatest permeate flux during rejection experiment (data not shown) and water flux (cf. above). It is important to mention that repeating experiments (with selection criteria for the membrane samples, i.e., only membranes that have water permeability in the range of 15% relative to the average values were chosen) showed that most of the results showed similar trends with Fig. 3 and the standard deviations of MWCO were within the range 15 30%. The possible reason for this observation would be that even though the PES(only) membrane had smaller MWCO it had a higher pore density than the membranes prepared with an additive. Comparing the membranes prepared with an additive, it is clearly shown that all membranes showed similar rejection curve trend. The hydraulic membrane permeability, rejection and permeate flux data suggest that the PES Plu membrane had either the highest pore density or largest pore size leading to the highest surface porosity among the membranes prepared with an additive. Comparing PES PVP and PES PEG, it is shown that PES PEG had higher hydraulic permeability. This observation agrees with result reported by Torrestiana-Samschez et al. [10] Membrane morphology Membrane surface morphology as well as cross-section structure was visualized by using SEM. It is important to mention that the SEM observation for membrane surface was repeated for every membrane and similar image was obtained for most of samples. As presented in Fig. 4, all membranes had asymmetric structure consisting of a thin fine porous selective barrier and a much thicker porous sub-structure. The phenomena behind the formation of this typical structure had been explained in many previous publications [6,7]. Cross-section images support the results of membrane thickness measurement (cf. Section 3.1), except for PES PVP membrane. It seemed that part of PES PVP membrane disappeared during sample preparation. Visualization of surface morphology showed that the membrane surface had fine pore structure with dimensions in the nanometer range (<10 nm). It is clearly seen that PES PVP had the roughest surface morphology (it should be noted that the drying process has been done by non-solvent exchange method, thereby collapse of the pore structure should be minimized). Quantification of surface roughness performed by Ochoa et al. [11] showed that the addition of PVP in preparation of PES UF membranes increased the surface roughness. Comparing surface images of PES(only) and PES Plu membranes, it appears that the PES Plu membrane had larger pore size than PES(only). Nevertheless, it seems that the pore density of PES(only) membrane is higher than that of the PES Plu membrane. Thus, smaller pore size but higher pore density for PES(only) compared to PES Plu supports the previous discussion of hydraulic permeability and rejection curve results Membrane surface hydrophilicity As clearly seen in Table 3, PES membrane without an additive had lower CA ( 65 ) than typically measured for non-porous PES film ( 76 ) [31]. Porous structure in the outer membrane surface would be the reason for this difference. Therefore, care should be taken to interpret the CA results because the value is influenced not only by membrane material but also by surface porosity. Indeed, such effect is observed by comparing the CA data of PES(only), PES PEG and PES Plu membranes. The membrane prepared without an additive showed similar CA with the membranes prepared Table 3 Static water contact angle, measured by sessile drop method, of membranes prepared using different additives. No. Membrane Contact angle ( ) Fig. 3. Rejection curve of membranes determined by using a dextran mixture solution with total concentration of 1 g/l (as the feed) and at a pressure of 100 kpa. 1 PES(only) 64.5 ± PES PVP 64.1 ± PES PEG 54.1 ± PES Plu 61.7 ± 2.8

6 130 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Fig. 4. SEM images of the cross-section and membrane surface morphology: from the top to the bottom panel: PES(only), PES PVP, PES PEG and PES Plu, respectively. with a hydrophilic additive (PES PVP and PES PEG). This could be explained because the PES(only) membrane had higher surface porosity as noticed by its higher permeability. Comparing the CA data of PES PEG and PES Plu membranes suggests that the contribution of hydrophobic part of block copolymer in Pluronic could be observed (note that these membranes had similar rejection curve, cf. Fig. 3). It is interesting to note that PES PEG membrane had significantly lower CA compared to other membranes indicating that this was the most hydrophilic membrane, while PES PVP and PES Plu membranes showed similar value.

7 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Fig. 5. Zeta potentials as a function of ph calculated from the tangential streaming potential of the outer surface of different membranes (at mol/l KCl) Membrane surface charge Zeta potential was measured in order to get information on surface charge of the membranes. As presented in Fig. 5, all membrane surfaces had a negative charge over the entire ph range studied, and the absolute values decreased to acidic ph range (cf. Ref. [21] for the reasons behind this phenomenon). The addition of a macromolecular modifier into polymer solution obviously changed the effective surface charge of the PES membrane. Nevertheless, a linear decrease in ZP with increasing ph solution for all membranes reflects typical behavior of a surface that has a negative charge due to anion adsorption from the solution. Both PEG and Plu decreased the surface charge toward smaller absolute value. By contrast, addition of PVP increased the negative charge of membrane. These indicated that adsorption of anions (hydroxyl, chloride) from the solution was more significant for PES PVP than for the other membranes Membranes surface chemistry Figs. 6 and 7 show the IR spectra of macromolecular additives used for membrane preparation and of the resulting membranes, respectively. As expected, all membranes showed typical spectra of PES, i.e., aromatic bands at 1578 and 1485 cm 1 from the benzene ring and C C bond stretch and aromatic ether band around 1240 cm 1. A new significant peak at 1678 cm 1 assigned to a pri- Fig. 7. FTIR spectra of membranes prepared with different macromolecular additives. mary amide stretch was observed for the membrane prepared with addition of PVP. By contrast, no additional peak was observed for the membranes prepared with addition of PEG or Plu. The reason for this result would be overlapping bands of the strongest bands for PEG and Plu (ether) with bands for PES (ether; cf. Figs. 6 and 7). Indeed, a significant increase in transmittance at 1105 cm 1, due to additional intensity of C O bond stretch (from PEG and Plu) was observed and this confirms the presence of the additives in the membrane polymer matrix. Overall, the IR spectra data indicate that changes in surface chemistry were detected after addition of the macromolecular modifiers to the membrane polymer solution Membrane performance based on adsorptive fouling and ultrafiltration The effect of additive on membrane performance was investigated with respect to adsorptive fouling and ultrafiltration. Adsorptive fouling was studied by exposing the outer membrane surface (selective barrier side) to protein (BSA) feed solution. The relative water flux reduction (RFR) was used to identify the extent of adsorptive fouling. As clearly seen in Fig. 8, the membranes prepared with an addition of hydrophilic modifier showed significantly higher resistance towards adsorptive fouling than the membrane prepared without an additive as noticed by their much lower RFR. It Fig. 6. FTIR spectra of the macromolecular additives used. Fig. 8. Relative flux reduction after static adsorption using BSA (1 g/l in phosphate buffer 0.05 M, ph 7, 3 h exposure). The error bars represent standard deviation.

8 132 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) Fig. 9. Normalized flux during ultrafiltration of BSA solutions (0.1 g/l in phosphate buffer 0.05 M, ph 7) at a transmembrane pressure of 300 kpa. Water fluxes after external cleaning with water, relative to initial water flux are also included. should be kept in mind that the effects of adsorptive fouling depend also on the barrier pore size, and the highest flux reductions were found for matching pores and solute sizes (cf. [31]); considering that the pore size distributions were different but still in the same range (cf. Fig. 3), the hydrophobicity of PES seemed to have an additional impact on RFR. This suggests that blending of hydrophilic macromolecular additive with membrane polymer could indeed significantly increase the hydrophilicity of the resulting membrane. PES PEG membrane showed the lowest RFR among the membranes prepared with an additive. This result can be explained by the highest hydrophilicity of this membrane (cf. Table 3). It should be noted that the protein in solution had negative charge and if electrostatic interactions would play a dominant role, PES PVP membrane should have the lowest RFR (cf. Fig. 5). Lower adsorptive fouling for PES membranes prepared with an additive compared to PES(only) membrane was also observed from ZP measurements after adsorptive fouling (cf. Fig. 5). The change in ZP after adsorptive fouling indicates the presence of protein on the membrane surface. The much larger change for PES(only) than for PES Plu suggests that the degree of adsorptive fouling was higher for PES(only) membrane. The positive value of ZP for ph values <4.9 after adsorptive fouling can be explained by considering the IEP of the BSA (ph 4.8), i.e. an excess of positive surface charge has been introduced via the protein. To investigate ultrafiltration performance, dead-end stirred ultrafiltration was performed with constant transmembrane pressure (300 kpa). The results are presented in terms of permeate flux relative to initial water flux (Fig. 9). It was observed that permeate flux dropped rapidly in the beginning of filtration for all membranes. On the one hand, this phenomenon indicates that concentration polarization has taken place. On the other hand, difference in flux profile for the membranes having similar rejection curve (PES PVP, PES PEG and PES Plu, cf. Fig. 3) suggests that fouling also contributed to the permeate flux decline. Water flux measurements after external cleaning using water confirmed that both reversible and irreversible fouling have occurred. Furthermore, higher increase in water flux after external cleaning compared to permeate flux for membranes prepared with an additive implies that reversible fouling was more significant for those membranes than for the membrane prepared without an additive. It is interesting to note that the PES(only) membrane had the highest permeability and a distinctly different pore size distribution (significantly larger sieving for dextran molar masses below about 50 kda which corresponds to the size of BSA, cf. Fig. 3). The higher observed BSA rejection could then be due to fouling (cf. below). Table 4 Apparent protein rejection. No. Membrane Rejection (%) 1 PES(only) 87 2 PES PVP 71 3 PES PEG 75 4 PES Plu 72 Indeed, the presence of hydrophilic macromolecular additive increased the normalized flux indicating that higher resistance towards fouling has been obtained. The membrane prepared without an additive had permeate flux of only 30% relative to the initial water flux, whereas the PES Plu membrane had the highest permeate flux (more than 70%). Of course, the highest initial flux of the membrane without an additive also contributed to the lowest normalized flux but the effect of hydrophilic modifier was quite clear. Interestingly, at the beginning of filtration PES PEG membrane had higher normalized flux than PES Plu membrane but further decrease with filtration time was more significant. The possible reason for this phenomenon would be the stability of the additive in the matrix polymer membrane (cf. Section 3.8). Rejection data presented in Table 4 show that the PES membrane prepared without an additive had the highest protein rejection while all membranes prepared with an additive showed similar protein rejection. This result is in good agreement with rejection curve if the molar mass of BSA is fitted. In general, performance test showed that the membrane prepared with addition of Pluronic as modifier agent showed the best performance, i.e., the lowest flux decline and similar rejection could be obtained Stability test study Because cleaning cannot be avoided in practical application, the performance of membrane prepared with a hydrophilic additive will strongly depend on the stability of the additive in the membrane polymer matrix during membrane cleaning. In this work, the stability of hydrophilic additive in water (20 and 40 C) and sodium hypochlorite solution was investigated. Surface chemistry by FTIR, contact angle and transport property by water flux were used as indicators to evaluate the membrane stability. Fig. 10 shows that no change in CA was observed for the PES(only) membrane after soaking in water (40 C), while slight decrease was observed after incubating in NaOCl solution. Increasing porosity as noticed by increase in water flux (cf. Fig. 12) would be the possible reason for this change. For the mem-

9 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) branes prepared with an additive, soaking in the water 40 C did not change the CA of PES PVP and PES Plu membranes while soaking in NaOCl slightly decreased. Decrease in CA via increasing porosity seemed to be dominant compared to increase in CA via extracting hydrophilic additive during immersing in NaOCl solution. Indeed, significant increase in CA was observed for PES PEG after incubating in water and NaOCl solution even only after 2 days of incubating. This indicates that the stability of PEG in the membrane polymer matrix was quite low. Consequently, hydrophilic character of the resulting membrane would easily be reduced even when only water is used for cleaning. Similar results with respect to stability in water at 40 C were found during stability test in water at 20 C (data not shown). As presented in Fig. 11, slight decrease in IR absorbance for the introduced functional additive (cf. Section 3.6) was observed for all additives used after soaking in both water and NaOCl solution. This indicates that none of the additives was completely stable in the matrix polymer membrane. However, irrespective this loss, significant surface modification could still be observed as noticed by either the appearance of new peak (for PES PVP) or increase in transmittance (for PES PEG and PES Plu) even after soaking for 10 days in all studied potential cleaning agents. Fig. 12 shows the flux stability of the membranes during incubating. It is clearly seen that flux of all membranes did not change after incubating in water (data for 20 C are not shown). Slight decreases in flux for PES PVP and PES PEG membranes were still within the range of experimental error. By contrast, after immersion in sodium hypochlorite solution, significant increase in flux was observed for all membranes including the membrane prepared without an additive. In this case, PES degradation by NaOCl leading to higher porosity was more dominant rather than extracting of hydrophilic additive. ZP measurement after soaking in NaOCl solution (the same concentration as in stability test) supports this observation. The negative charge of both PES(only) and PES Plu membranes significantly increased after soaking. The ZP at ph 9.5 decreased from 11.5 to 22 mv and from 6.1 to 11.5 for PES(only) and PES Plu membranes, respectively. Chain scission of the PES C S bond by attack of NaOCl yielding finally sulfonic acid groups has been evoked [24], while extraction or reduced swelling of hydrophilic macromolecular additive has also been reported [23,32,33]. Fig. 10. Stability test investigated by measuring the contact angle as a function of incubating time. Soaking the membranes in water (20 C) resulted in similar effect with soaking in water 40 C. Fig. 11. Stability test investigated by measuring the IR absorbance of functional additive as a function of incubating time.

10 134 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) prepared with macromolecular additives, whereas the rejection curves examined using mixture of dextrans were similar for all three modified membranes. All membranes had negative surface charge within the entire ph range studied. Surface hydrophilicity measured with CA indicates that the PES PEG membrane was the most hydrophilic membrane; nevertheless, the stability of that macromolecular additive within the membrane polymer matrix was the most crucial problem. Performance evaluation via investigation of adsorptive fouling and ultrafiltration using BSA suggests that PES PEG membrane showed the lowest RFR after static adsorption followed by PES Plu. Ultrafiltration experiments demonstrated that the antifouling effects of PES Plu were the most efficient at similar protein rejection: permeate flux during ultrafiltration using the PES Plu was much higher than using the PES PEG and the PES PVP membranes, and more than 70% of the initial water flux could be recovered after UF just by external cleaning with water. Overall, performance test and stability study suggest that the membrane prepared with the addition of Pluronic as modifier agent showed the best performance as well as the best stability; therefore, it should be considered as additive in practical applications. Acknowledgements The authors thank Dieter Jacobi (Technische Chemie, Universität Duisburg Essen) for the GPC analysis and Smail Boukercha (Anorganische Chemie, Universität Duisburg Essen) for his contribution on SEM observation. We also thank BASF (Germany) for supplying the PES polymer. References Fig. 12. Stability test investigated by measuring the water flux as a function of incubating time. Soaking the membranes in water (20 C) yielded similar effect with soaking in water 40 C. In general, in this stability study, PES Plu membrane showed the best stability followed by PES PVP among the membranes prepared with an additive event though slight removal observed from surface chemistry by FTIR could still be detected. The reason for this phenomenon would be that the hydrophobic part of Pluronic enhances the PES additive interaction. Not only entanglement but also hydrophobic hydrophobic interactions may occur. The possible interaction of Pluronic in PES polymer membrane matrix was described by Jiang et al. [17,20]. The interaction between PES and PVP has been identified through investigation of viscoelastic behavior of their polymer mixtures [4]. The authors proposed that the interaction between PES and PVP could be either between NH C O groups in PVP and O S O groups in PES or between side cyclic groups of PVP and aromatic ring of PES. 4. Conclusions The characteristics, performance and stability of PES membranes prepared by NIPS with three different macromolecular additives, i.e., PVP, PEG and Plu, have been investigated. The characteristic of additive and the preparation conditions were maintained to be similar so that the performance of those additives can be fairly compared. Indeed, the presence of those macromolecular additives in membrane polymer matrix as noticed by FTIR data significantly determined the characteristic as well as performance of the resulting membranes. This observation was confirmed by investigation of PES membrane prepared without an additive as comparison. The PES membrane prepared with addition of Pluronic (PES Plu) showed the highest hydraulic permeability among the membranes [1] M. Cheryan, Ultrafiltration and Microfiltration Handbook, Technomic Publishing Company Inc., Pennsylvania, [2] R.W. Baker, Membrane Technology and Applications, 2nd ed., John Wiley & Sons, Ltd., Chichester, [3] H. Susanto, M. Ulbricht, Polymeric membranes for molecular separations, in: E. Drioli, L. Giorno (Eds.), Membrane Operations. Innovative Separations and Transformations, Wiley VCH, Weinheim, in press. [4] L.Y. Lafreniere, F.D.F. Talbot, T. Matsuura, S. Sourirajan, Effect of polyvinylpyrrolidone additive on the performance of polyethersulfone ultrafiltration membranes, Ind. Eng. Chem. Res. 26 (1987) [5] R.M. Boom, H.W. Reinders, H.H.W. Rolevink, Th. van den Boomgaard, C.A. Smolders, Equilibrium thermodynamics of a quaternary membrane-forming system with two polymers. I. Experiments, Macromolecules 27 (1994) [6] M. Wienk, R.M. Boom, M.A.M. Beerlage, A.M.W. Bulte, C.A. Smolders, Recent advances in the formation of phase inversion membranes made from amorphous or semi-crystalline polymers, J. Membr. Sci. 113 (1996) 361. [7] R.M. Boom, I.M. Wienk, Th. Van den Boomgaard, C.A. Smolders, Microstructures in phase inversion membranes. Part 2. The role of a polymeric additive, J. Membr. Sci. 73 (1992) 277. [8] H.T. Yeo, S.T. Lee, M.J. Han, Role of polymer additive in casting solution in preparation of phase inversion polysulfone membranes, J. Chem. Eng. Jpn. 33 (2000) 180. [9] D.B. Mosqueda-Jimenez, R.M. Narbaitz, T. Matsuura, G. Chowdhury, G. Pleizier, J.P. Santerre, Influence of processing conditions on the properties of ultrafiltration membranes, J. Membr. Sci. 231 (2004) 209. [10] B. Torrestiana-Sanchez, R.I. Ortiz-Basurto, E.B. Fuente, Effect of nonsolvents on properties of spinning solutions and polyethersulfone hollow fiber ultrafiltration membranes, J. Membr. Sci. 152 (1999) 19. [11] N.A. Ochoa, P. Pradanos, L. Palacio, C. Pagliero, J. Marchese, A. Hernandez, Pore size distributions based on AFM imaging and retention of multidisperse polymer solutes: characterisation of polyethersulfone UF membranes with dopes containing different PVP, J. Membr. Sci. 187 (2001) 227. [12] J. Marchese, M. Ponce, N.A. Ochoa, P. Pradanos, L. Palacio, A. Hernandez, Fouling behaviour of polyethersulfone UF membranes made with different PVP, J. Membr. Sci. 211 (2003) 1. [13] A. Idris, N.M. Zain, M.Y. Noordin, Synthesis, characterization and performance of asymmetric polyetehrsulfone (PES) ultrafiltration membranes with polyethylene glycol of different molecular weights as additives, Desalination 207 (2007) 324. [14] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Effect of molecular weight of PEG on membrane morphology and transport properties, J. Membr. Sci. 309 (2008) 209.

11 H. Susanto, M. Ulbricht / Journal of Membrane Science 327 (2009) [15] Y. Liu, G.H. Koops, H. Strathmann, Characterization of morphology controlled polyethersulfone hollow fiber membrane by the addition of polyethylene glycol to the dope and bore liquid solution, J. Membr. Sci. 223 (2003) 187. [16] J.H. Kim, K.H. Lee, Effect of PEG additive on membrane formation by phase separation, J. Membr. Sci. 138 (1998) 153. [17] Y. Wang, T. Wang, Y. Su, F. Peng, H. Wu, Z. Jiang, Remarkable reduction of irreversible fouling and improvement of permeation properties of polyethersulfone ultrafiltration membrane by blending with Pluronic F127, Langmuir 21 (2005) [18] Y. Wang, Y. Su, Q. Sun, X. Ma, X. Ma, Z. Jiang, Improved permeation performance of Pluronic F127-polyethersulfone blend ultrafiltration membranes, J. Membr. Sci. 282 (2006) 44. [19] Y.Q. Wang, Y.L. Su, X.L. Ma, Q. Sun, Z.Y. Jiang, Pluronic polymers and polyethersulfone blend membranes with improved fouling resistant ability and ultrafiltration performance, J. Membr. Sci. 283 (2006) 440. [20] W. Zhao, Y. Su, C. Li, Q. Shi, X. Ning, Z. Jiang, Fabrication of antifouling polyethersulfone ultrafiltration membranes using Pluronic F127 as both surface modifier and pore-forming agent, J. Membr. Sci. 318 (2008) 405. [21] H. Susanto, M. Ulbricht, Influence of ultrafiltration membrane characteristics on adsorptive fouling with dextrans, J. Membr. Sci. 266 (2005) 132. [22] S. Rouaix, C. Causeserand, P. Aimar, Experimental study of the effects of hypochlorite on polysulfone membrane properties, J. Membr. Sci. 277 (2006) 137. [23] I.M. Wienk, E.E.B. Meuleman, Z. Borneman, A. van den Boomgaard, C.A. Smolders, Chemical treatment of membranes of a polymer blend: mechanism of the reaction of hypochlorite with poly(vinylpyrrolidone), J. Polym. Sci. A: Polym. Chem. 33 (1995) 49. [24] E. Arkhangelsky, D. Kuzmenko, V. Gitis, Impact of chemical cleaning on properties and functioning of polyethersulfone membranes, J. Membr. Sci. 305 (2007) 176. [25] L. Wu, J. Sun, Q. Wang, Poly(vinylidine fluoride)/polyethersulfone blend membranes: effect of solvent sort, polyethersulfone and polyvinylpyrrolidone concentration on their properties and morphology, J. Membr. Sci. 285 (2006) 290. [26] P. Menut, Y.S. Su, W. Chinpa, C. Pochat-Bohatier, A. Deratani, D.M. Wang, P. Huguet, C.Y. Kuo, J.Y. Lai, C. Dupuy, A top surface liquid layer during membrane formation using vapour-induced phase separation (VIPS) evidence and mechanism of formation, J. Membr. Sci. 310 (2008) 278. [27] T.A. Tweddle, O. Kutowy, W.L. Thayer, S. Sourirajan, Polysulfone ultrafiltration membranes, Ind. Eng. Chem. Prod. Res. Dev. 22 (1983) 320. [28] V.E. Reinsch, A.R. Greenberg, S.S. Kelley, R. Peterson, L.J. Bond, A new technique for the simultaneous, real-time measurement of membrane compaction and performance during exposure to high-pressure gas, J. Membr. Sci. 171 (2000) 217. [29] L. Brinkert, N. Abidine, P. Aptel, On the relation between compaction and mechanical properties for ultrafiltration hollow fiber, J. Membr. Sci. 77 (1993) 123. [30] K.W. Lawson, M.S. Hall, D.R. Lloyd, Compaction of microporous membranes used in membrane distillation. I. Effect on gas permeability, J. Membr. Sci. 101 (1995) 99. [31] H. Susanto, S. Franzka, M. Ulbricht, Dextran fouling of polyethersulfone ultrafiltration membranes causes, extent and consequences, J. Membr. Sci. 296 (2007) 147. [32] S.H. Wolff, A.L. Zydney, Effect of bleach on the transport characteristics of polysulphone hemodialyzers, J. Membr. Sci. 243 (2004) 389. [33] J.J. Qin, M.H. Oo, Y. Li, Development of high flux polyethersulfone hollow fiber ultrafiltration membranes from a low a critical solution temperature dope via hypochlorite treatment, J. Membr. Sci. 247 (2005) 137.