Surface characterization and performance evaluation of commercial fouling resistant low-pressure RO membranes

Transcription

1 Desalination 202 (2007) Surface characterization and performance evaluation of commercial fouling resistant low-pressure RO membranes David Norberg a, Seungkwan Hong b *, James Taylor a, Yu Zhao a a Civil and Environmental Engineering Department, University of Central Florida, Orlando, FL 32816, USA b Civil and Environmental Engineering Department, Korea University, Anam-dong, Sungbuk-ku, Seoul , South Korea Tel ; Fax ; skhong21@korea.ac.kr Received 31 July 2005; accepted 23 December 2005 Abstract This paper describes the characterization and evaluation of various RO/NF membranes for the treatment of seasonally brackish surface water with high organic contents (TOC 21 mg/l). Twenty commercially available RO and NF membranes were initially evaluated by performing controlled bench-scale flat-sheet tests and surface characterization. Based on the results, four low pressure RO membranes were selected for use in the pilot study. The surface characterization revealed that each of four selected membranes had unique surface characteristics to minimize membrane fouling. Specifically, the LFC1 membrane featured a neutral or low negative surface to minimize electrostatic interactions with charged foulants. The X20 showed a highly negatively charged surface, and thus, was expected to perform well with feed waters containing negatively charged organics and colloids. The BW30FR1 exhibited a relatively neutral and hydrophilic surface, which could be beneficial for lessening organic and/or biofouling. The SG membrane had a smooth surface that made it quite resistant to fouling, particularly for colloidal deposition. In the large-scale pilot study using single element, all of the four membranes experienced a gradual increase in specific flux over time, indicating no fouling occurred during the pilot study. The deterioration of permeate water quality such as TDS was also observed over time, suggesting that the integrity of the membranes might be compromised by the monochloramine used for biofouling control. Keywords: Membrane fouling; RO membranes; Surface charge; Surface roughness; Hydrophobicity; Surface water treatment 1. Introduction In recent years, membranes have become fully or partially integrated into all facilities that *Corresponding author. produce drinking water [1]. This is due to the fact that membrane processes can resolve technically complex and, at times, conflicting requirements related to compliance with multicontaminant regulations [2]. With the tightening of regulations Presented at the conference on Wastewater Reclamation and Reuse for Sustainability (WWRS2005), November 8 11, 2005, Jeju, Korea. Organized by the International Water Association (IWA) and the Gwangju Institute of Science and Technology (GIST) /06/$ See front matter 2006 Published by Elsevier B.V. doi: /j.desal

2 46 D. Norberg et al. / Desalination 202 (2007) in the future, the need for membrane technology such as RO (reverse osmosis) and NF (nanofiltration) will increase significantly. However, the wide use of RO membrane technology in the drinking water industry has been hampered greatly by membrane fouling [3]. The extent and rate of membrane fouling are largely affected by membrane surface characteristics [4,5]. Thus, developing membranes with a better resistance to fouling by modifying membrane surface properties has become an area of increasing significance. A large-scale pilot study was conducted for the treatment of a highly organic brackish water from Lake Monroe in Sanford, Florida. This pilot study consisted of three phases. The preliminary stage involved the selection of four membranes, which are suitable for treatment of this water. The selection process consisted of laboratory bench-scale flat-sheet testing and surface characterization of 20 commercially available membranes. The two pilot study phases involved the evaluation of three pretreatments and the four selected membranes in single element low recovery and two-stage high recovery configurations [6]. This paper primarily focuses on the research work involving the characterization and evaluation of fouling resistant RO membranes selected for this pilot study. Specifically, four commercially available fouling resistant low-pressure RO membranes selected were LFC-1 (Hydranautics), X20 (Trisep), BW30FR1 (FilmTec), and SG (Osmonics), all of which were thin film composite made of polyamide derivatives. The surfaces of these membranes were thoroughly characterized in terms of roughness, charge, and hydrophobicity. In addition, their performance at pilot-scale experiments is briefly discussed in this paper. 2. Experimental methods 2.1. Source water The raw water for this study was collected from Lake Monroe in Sanford, Florida and is highly organic water with high concentrations of dissolved salts (brackish). The raw water has seasonal variations. During the rainy season in Florida (May September), the organics increased significantly due to rainwater runoff, and the total dissolved solid (TDS) decreased due to dilution of the lake water. In the dry season (October April) the organics accordingly decreased and the TDS increased RO/NF membranes Twenty commercially available RO and NF membranes were evaluated to select four membranes suitable for the treatment of high organic brackish surface waters in Florida. These membranes were obtained from major RO/NF manufacturers in the world: Osmonics, FilmTec, Hydranautics, Trisep, and SaeHan. Two of these membranes were made of cellulose acetate (CA) and 18 were thin film composite (TFC) membranes as shown in Table Membrane surface characterization The surface roughness was characterized by atomic force microscopy (AFM) in tapping mode operation. The surface of the membrane was scanned and a topographic image of the surface was stored by the computer. The average roughness is simply the average of the surface height deviations as measured as a distance from the center plane. The roughness was reported as the root mean squared (RMS) of these deviations. Membrane surface charge was evaluated by the zeta potential at the plane of shear and was calculated from streaming potential measurements [7]. The streaming potential was induced by an electrolyte solution flowing across the charged, stationary membrane surface. The ph of the electrolyte was adjusted in order to obtain zeta potential measurements between ph 3 and 11. Lastly, the hydrophobicity of the membrane was determined using contact angle measurements

3 D. Norberg et al. / Desalination 202 (2007) Table 1 Surface characterization of RO/NF membranes Manufacturer Category Membrane RMS (nm) ZP (mv) CA ( ) Osmonics CA (RO) CD* CA (LPRO) CG* TFC (LPRO) AG TFC (LPRO) SG TFC (NF) HL TFC (NF) DK TFC (NF) DL Fluid System TFC (NF) TFC-S TFC (NF) TFC-SR TFC (NF) TFC-SR FilmTec TFC (NF) NF TFC (NF) NF TFC (LPRO) BW30FR Hydranautics TFC (LPRO) LFC TFC (LPRO) ESPA TFC (NF) ESNA SaeHan TFC (NF) BE-FR TFC (LPRO) BL-FR Trisep TFC (LPRO) X TFC (NF) TS by a goniometer. The captive bubble technique was used to determine contact angle because measurements were performed in aqueous conditions Bench-scale flat sheet testing For the initial bench-scale performance evaluation, the raw water collected was coagulated with ferric sulfate and filtered with a 5 μm filter prior to analysis and fouling experiments. The bench-scale testing was performed using a crossflow filtration unit which included a stainless steel, flat-sheet membrane test cell (Sepa CF) manufactured by Osmonics. The dimension of the cell was 14.5 cm length, 9.4 cm width, and 0.86 mm height, providing an effective membrane area of m 2. The feed solution was held in a 20 L cylindrical tank and was kept stirred by a magnetic stirrer. The temperature was maintained at 20 C using a refrigerated recirculator which circulated the coolant through a stainless steel coil submerged in the feed solution. The feed solution was pumped from the reservoir using a Hydracell pump. The permeated flow was monitored by a digital flow meter interfaced with a PC. The crossflow velocity and feed pressure were controlled through adjustment of a by-pass valve and a back pressure regulator Pilot-scale performance evaluation The pilot study involving the treatment of highly organic surface water utilized an integrated membrane system (IMS) consisting of two pretreatments by ferric sulfate coagulation, which fed eight single membrane elements. Specifically, as shown in Fig. 1, the pretreatment processes employed in this study were (1) super pulsator (SP) blanket clarifier followed by dual media gravity filtration and (2) Zenon (ZN) immersed

4 48 D. Norberg et al. / Desalination 202 (2007) Super-P Pretreatment Raw Super-P Filter LPRO Membranes Ferric Coagulant ph Adjustment AntiScalant Monochloramine ZenonUF Pretreatment Raw Ferric Coagulant Zenon UF LPRO Membranes AntiScalant Monochloramine Finished Water Finished Water Fig. 1. Schematic description of treatment process trains for pilot study. micro-filter with ferric coagulation. The pretreatment processes were primarily designed to remove particulate matter, natural organic matter (NOM), and pathogens. RO membranes were utilized to remove dissolved salts and to further achieve higher levels of NOM and pathogen removal. The pilot systems were located at the Sanford wastewater reclamation facilities in eastern Central Florida. The raw source water was obtained from Lake Monroe and supplied to pilot testing systems. Chloramines were added in feed water to prevent biofouling, while anti-scalants were used for scaling control. Four fouling resistant membranes selected were pilot-tested using single elements from April to August in Detailed description of this pilot study is available in a recent article by Zhao et al. [6]. 4. Results and discussion 4.1. Membrane surface characterization and selection A total of 20 low-pressure RO (LPRO) and nanofiltration (NF) membranes were systematically evaluated to determine their selectivity for organic and inorganic compounds. The majority of LPRO membranes tested were fouling resistant membranes, while the NF membranes were designed for high organic removal and softening. The selectivity for organics was ranked based on NPDOC rejection. The selectivity for inorganic compounds was evaluated on the basis of TDS, Ca, Mg, Na, Cl, Br, and SO 4 rejection from CSF (coagulation sedimentation filtration) treated water. The criteria for selection were high organic and inorganic rejection. In addition, membrane surface characteristics were determined to aid membrane selection. The surface charge (ZP: zeta potential) ranged from 4.0 to 19.7 mv at ph of 6.5. The two cellulose acetate membranes (CD and CG) have a low negative charge compared to the thin film composite membranes. The surface roughness (RMS) of the membranes tested ranged from 5.9 to 130.2nm. The cellulose acetate membranes were the smoothest. The hydrophobicity (CA: contact angle) ranged from 38 to 73 for the LPRO and NF membranes. The results of surface characterization are summarized in Table 1. Based on filtrate water quality and surface properties, four fouling resistant LPRO membranes Table 2 Organic (TOC) rejection (%) of RO/NF membranes in bench-scale flat-sheet testing Membrane Feed Permeate Rejection (%) CD CG AG SG BW30FR LFC ESPA BE-FR BL-FR X TS HL DK DL TFC-S TFC-SR TFC-SR NF NF ESNA

5 D. Norberg et al. / Desalination 202 (2007) BW30FR1 LFC 1 SG X 20 Fig. 2. AFM images of fouling resistant LPRO membrane surfaces. were selected for the treatment of high organic brackish surface water. They included BW30FR (FilmTec), X20 (Trisep), LFC1 (Hydranautics) and SG (Osmonics). Despite favorable surface characteristics, the cellulose acetate membranes were not selected because of poor organic rejection as summarized in Table 2. Lastly, the NF specifically designed for color removal, such as DK membrane, was not selected despite high organic rejection, because of their inability to reject dissolved ions (% removal of TDS = 66%). The AFM pictures of selected membranes are shown in Fig. 2 and the surface charge variation as a function of ph is presented in Fig. 3. The surface characteristics of four membranes selected are described more specifically as follows: BW30FR1 (FilmTec): This membrane is specifically designed to resist bio-film formation which is expected to be a primary cause of membrane fouling during surface water treatment. The manufacturer claimed that this membrane LFC-1 BW-30FR X-20 SG Fig. 3. Zeta potential of fouling resistant LPRO membranes under various solution phs. exhibits significantly less loss in productivity and better cleanability than typical thin-film composite polyamide membranes. The surface analysis revealed that the BW30FR1 has a relatively neutral and hydrophilic surface with medium surface roughness.

6 50 D. Norberg et al. / Desalination 202 (2007) X20 (Trisep): This is a thin-film composite membrane featuring polyamide urea, specifically designed for high fouling feed waters. The manufacturer reported that the surface charge of the X20 membrane minimizes fouling of organic substances. The charge measurement by SPA showed a highly negatively charged surface as claimed by the manufacturer. Thus, this membrane is expected to perform well with feed waters containing negatively charged organics and colloids. However, in general, a wide spectrum of foulants with varying degree of surface charge exists in typical source waters. As a result, it is also possible that this membrane may suffer from fouling, particularly through the interactions with positively charged organics. LFC1 (Hydranautics): This is a low fouling composite membrane specifically designed for high fouling feed waters. According to the manufacturer, the LFC1 features neutral surface charge and hydrophilicity which significantly minimize membrane fouling. The surface analysis data showed low negative charge and medium hydrophobicity. The surface roughness of this membrane was also estimated to be in the range of medium to high. SG (Osmonics): This is a thin-film composite brackish water desalting membrane. The manufacturer claimed that the SG membrane has a smooth surface which makes it quite resistant to fouling. The AFM data clearly showed much smaller surface peaks compared to other fouling resistant membranes. However, the contact angle measurements indicated this membrane to be more hydrophobic than the others while it carries surface charges in the range of low to medium Pilot-scale performance evaluation The large-scale pilot study was conducted using four selected LPRO membranes to investigate the feasibility of integrated membrane systems (IMS) for treatment of high organic brackish surface water. As previously described, the source water of the pilot study was collected from Lake Monroe in Sanford, Florida. The water quality fluctuated dramatically. During the rainy season, the runoff with high levels of humic material increased organic concentration (average NPDOC = 21.1 mg/l) significantly. The turbidity decreased steadily in rainy season as a result of dilution of the lake water, while the turbidity increased from 3 to 7 NTU in dry season due to the evaporation of surface water. Total dissolved solid (TDS) also increased in dry season (winter) due to salts introduced to the lake from groundwater intrusion and the evaporation of lake water, ranging from 50 to 750 mg/l. In the pilot study, two different types of pretreatment processes were evaluated: Super-P (SP) and Zenon (ZN). Regardless of seasonal changes in raw water quality, pretreated water quality was kept in an acceptable range as feed water for LPRO membranes. Both treatments showed similar performance except organic removal. Specifically, the average NPDOC of SP was 3.3 mg/l while that of ZN was 6.2 mg/l. The measurements of UV-254 and color also showed higher values for ZN compared to those of SP. The superior organic removal by SP was attributed to low coagulation ph optimized for organic removal, while the coagulation ph of ZN was adjusted to improve iron retention by UF membrane, and as a result, coagulation was Table 3 Average water quality of two pretreatment processes (Super-P and Zenon) Parameter Unit Raw water Super-P Zenon NPDOC mg/l UV 254 cm Turbidity NTU TDS mg/l Fe mg/l SiO 2 mg/l Color cpu Alkalinity mg/l CaCO

7 D. Norberg et al. / Desalination 202 (2007) Table 4 Summary of membrane operating characteristics during pilot study Membrane Feed pressure (kpa) Pressure drop (kpa) Productivity (lmh/kpa) Super-P Zenon Super-P Zenon Super-P Zenon LFCI X SG BW30FR performed at higher ph despite poor organic removal. The average turbidity from both pretreatment was 0.09 NTU. The average TDS of the both pretreated water is higher than TDS of raw water, due to coagulation. Table 3 summarized the average quality of raw and pretreated waters. During four months of pilot study, the membrane systems were operated at a flux of 20.4 lmh and recovery of 70%. A summary of the initial and final observation of the operating characteristics of all four membranes and two pretreatments is presented in Table 4. The feed pressure dropped for all membranes regardless of pretreatment. However, the decrease of feed stream pressure over time was more for Super-P pretreatment than for the ZN pretreatment in all cases. The gradual increase in productivity (i.e. specific flux) over time was observed for all of the fouling resistant membranes, suggesting that no fouling was experienced during operation and/or the membranes might be degraded by monochloramine oxidation. The specific flux increased slightly less for ZN pretreatment, probably due to higher organic loading to LPRO membranes. The membrane degradation was also evidenced by a gradual decrease in TDS rejection [6]. It has been reported that the oxidative membrane degradation by monochloramine could be enhanced at the presence of iron coagulant residuals [8 10]. The exact mechanism of polyamide degradation by monochloramine has not been elucidated clearly yet, and the only speculation discussed in the literature is the formation of radicals from the interaction of iron coagulants and monochloramine, which attack and degrade membrane rejecting layers. 5. Conclusions Based on controlled bench-scale performance evaluation and surface characterization of 20 commercially available RO and NF membranes, four low-pressure RO membranes were chosen for use in a pilot study to treat high organic brackish surface water. They included FilmTec BW30FR, Trisep X20, Hydranautics LFC1, and Osmonics SG. The surface analysis of these membranes interestingly showed that each membrane exhibited one or two unique surface features, that are favorable for minimizing membrane fouling. Their characteristics were found to be neutral or highly negatively surface charge, less hydrophobic, and smooth surfaces, with a wide range of variations. This finding suggested that the selection of commercial fouling resistant membranes should be done carefully by considering foulants characteristics as well as membrane surface properties. The evaluation of these membranes at pilot-scale was performed with two advanced pretreatment processes for the duration of four months. The pretreated water did not cause any significant fouling on all of four membranes. Instead, the enhanced specific flux and reduced TDS removal were observed in

8 52 D. Norberg et al. / Desalination 202 (2007) the pilot study. This observation could be attributable to membrane oxidation by monochloramine that was used to prevent biofouling, indicating that their chemical compatibility should be carefully considered in the process of developing fouling resistant membranes. Acknowledgement This project was sponsored by the St. Johns River Water Management District (Jerry Salsano: SJRWMD project officer), through a subcontract from CH2M Hill (Matt Alvarez: project engineer). References [1] S. Duranceau, Future of membranes, J. Am. Water Works Assoc., 92 (2000) [2] J. Taylor and S. Hong, Potable water quality and membrane technology, J. Lab. Med., 31 (10) (2000) [3] S. Hong and M. Elimelech, Chemical and physical aspects of natural organic matter (NOM) fouling of nanofiltration membranes, J. Membr. Sci., 132 (1997) [4] M. Elimelech, X. Zhu, A.E. Childress and S. Hong, Role of surface morphology in colloidal fouling of cellulose acetate and composite polyamide RO membranes, J. Membr. Sci., (1997) [5] E.M. Vrijenhoek, S. Hong and M. Elimelech, Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes, J. Membr. Sci., 188 (2001) [6] Y. Zhao, J. Taylor and S. Hong, Combined influence of membrane surface properties and feed water qualities on RO/NF mass transfer: a pilot study, Water Res., 39 (2005) [7] A. Childress and M. Elimelech, Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes, J. Membr. Sci., 119 (1996) [8] C.J. Gabelich, T.I. Yun, B.M. Coffey and I.H. Suffet, Effects of aluminum sulfate and ferric chloride coagulant residuals on polyamide membrane performance, Desalination, 150 (2002) [9] C.J. Gabelich, J.C. Franklin, F.W. Gerringer, K.P. Ishida and I.H. Suffet, Enhanced oxidation of polyamide membranes using monochloramine and ferrous iron, J. Membr. Sci., 258 (2005) [10] S. Beverly, S. Seal and S. Hong, Identification of surface chemical functional groups correlated to failure of reverse osmosis polymeric membranes, J. Vacuum Sci. Technol., 18 (4) (2000)