The water quality of the permeates after UF or RO membrane filtration could meet the water reuse ... For CMP wastewater, the potential treatment processes in-.
ENVIRONMENTAL ENGINEERING SCIENCE Volume 25, Number 7, 2008 © Mary Ann Liebert, Inc. DOI: 10.1089/ees.2007.0056
Treatment of Chemical Mechanical Polishing Wastewater for Water Reuse by Ultrafiltration and Reverse Osmosis Separation Lain-Chuen Juang,1,* Dyi-Hwa Tseng,2 He-Yin Lin,2 Chung-Kung Lee,1 and Teh-Ming Liang3 1Green
Environment R&D Center and Department of Environmental Engineering Vanung University Chung-Li, 32061, Taiwan, Republic of China 2Graduate Institute of Environmental Engineering National Central University Chung-Li, 32061, Taiwan, ROC 3Energy and Environment Research Laboratories Industrial Technology Research Institute Hsinchu, 31040, Taiwan, ROC
Received date: February 24, 2007
Accepted date: October 21, 2007
Abstract
This paper describes the treatment of chemical mechanical polishing (CMP) wastewater from a semiconductor plant through membrane-based ultrafiltration (UF) and reverse osmosis (RO) processes to improve the removal efficiency under different water recovery and to determine the possible mechanisms of membrane blocking and rejection. UF pretreatment led to 42.1–46.9% conductivity, 98.1–99.4% turbidity, and 4.5–24.5% total organic carbon (TOC) removal. These contaminants were almost completely rejected after performing subsequent RO processing: the values of the conductivity, turbidity, and TOC were reduced to 6 S/cm, 0.01 NTU, and 1.6 mg/L, respectively. The water quality of the permeates after UF or RO membrane filtration could meet the water reuse standards for tap water, cooling water, boiler water, and the feed for a water purification machine. The variation with time of the permeate flux for the UF process was fitted to the Hermia model to examine the possible blocking mechanisms during the filtration process. The possible rejection mechanisms of the membrane processes are discussed herein in terms of the water quality variation during the membrane-based separation processes. Key words: chemical mechanical polishing; ultrafiltration; reverse osmosis; water reuse; membrane blocking
Introduction
C
MP (CHEMICAL MECHANICAL POLISHING) is an essential technology for global planarization in the manufacture of integrated circuit devices. During this process, a large quantity of ultrapure water is used to clean the surface of the wafer, generating a large quantity of CMP wastewater that is notorious for containing nano-sized particles (primarily SiO2, Al2O3, or CeO2), oxidizing agents, additives (such as NH4OH and KNO3), dispersants/surfactants, buffering inorganic or organic acids, metal complexing agents, corrosion inhibitors, and copper (Steigerwald et al.,
*Corresponding author: Department of Environmental Engineering, Vanung University, Chung-Li, 32061, Taiwan, ROC. Phone: 886 34515811 ext. 55720; Fax: 886 34622232; E-mail: lcjuang@vnu. edu.tw
1997; Li and Miller, 2000). Because CMP wastewater generally has high alkalinity, total solids content, and turbidity (Corlett, 2000; Den and Huang, 2005), it has difficulty meeting discharge standards for water reuse. Furthermore, according to the specifications of the International Technology Roadmap for Semiconductors, the objectives for pure water reuse in fabrication processes must meet 45% water recycling for older plants and more than 80% water recycling capability for new plants (Corlett, 2000). Therefore, there is an urgent need to develop useful and cost-effective reclamation technologies for CMP wastewater to meet the rigorous water reuse standards of industrial processes. For CMP wastewater, the potential treatment processes include chemical coagulation/flocculation (Koopal et al., 1999; Lin and Yang, 2004), electrocoagulation/flotation (Lin and Yang, 2004; Den and Huang, 2005), and membrane filtration (Pan et al., 2005). For the chemical coagulation/flocculation process, wastewater treatment units normally apply high
1091
1092 dosages of coagulants to enlarge the nanoscale particles 100fold to improve their settling and to meet the discharge limit. This process not only boosts the cost of chemical usage but also produces large amounts of sludge. The treatment efficiency of the electrocoagulation/flotation process can be reduced as a result of electrode passivation and the addition of surfactant into the aqueous solutions. To overcome these problems, membrane separation processes have been adopted to treat the CMP wastewater using either a single unit or a post separator (Yang et al., 2003). The major advantages of membrane processes are that the permeate may be reused for industrial process, and that the concentrated metals (such as copper) obtained from the filtration process may also be reused. The main obstacles limiting the application of the membrane processes, however, are decreased permeate flux caused by particle blocking, concentration polarization, and fouling after continuous long-term operation. Depending on the characteristics of the wastewater, the operating conditions, and the required water quality, several pretreatment processes have been used to inhibit or reduce the degree of membrane blocking. Among the traditional pretreatment technologies, coagulation, adsorption, and precipitation are employed most widely (Marcucci et al., 2001; Abdessemed and Nezzal, 2002; Kim et al., 2002; Wong et al., 2002; Shon et al., 2004). Recently, both microfiltration and ultrafiltration (UF) processes have been adopted as pretreatment processes for nanofiltration and reverse osmosis (RO) techniques, with their performance being more stable than those of traditional pretreatment methods (Yalcin et al., 1999; Karakulski and Morawski, 2000; Durham et al., 2001; Qin et al., 2003). The aim of this study was to discuss the performance of UF and RO filtration processes for CMP wastewater—in terms of the removal efficiency of conductivity, turbidity, and total organic carbon (TOC) within different water recovery—with the goal of improving the water quality of the permeate to achieve water reuse standards of industrial processes (e.g., for indirect cooling water, process water, boiler water, and sanitary water). Moreover, we examined the possible blocking and rejection mechanisms of the membrane processes to provide more information regarding the characteristics of the UF- and RObased treatments of CMP wastewater. Materials and Methods The CMP wastewater was obtained from a large semiconductor plant located at Hsinchu Science Park, Taiwan. The components of the CMP wastewater samples were very difficult to ascertain because of the different types and amounts of CMP slurries added during the manufacturing process. In this study, we used conductivity, turbidity, and TOC concentration measurements to represent the levels of inorganic compounds, fine particles, and organic compounds/colloidal materials, respectively, in the CMP wastewater. The conductivity and turbidity were both determined using a Turbiquant 1000IR turbidimeter (Merck, Germany). A Malvern zeta potential analyzer (Zetasizer Nano ZS, Malvern Instrument, UK) was used to determine the size distribution and zeta potentials of the fine oxide particles in the CMP wastewater. The TOC was measured using a TOC-1010 analyzer (O.I. Analytical). A Schambeck-SFD RI2000 gel permeation chromatograph (GPC) equipped with a SB-802HQ column was employed to measure the molecular weight of organic solutes in the CMP wastewater.
JUANG ET AL. The values of the conductivity, turbidity, and TOC of the raw CMP wastewater were 110 S/cm, 108 NTU, and 8.8 mg/L, respectively. The particle size distribution of fine particles was in the range from 40 to 200 nm, concentrated at 70 nm. Although the CMP wastewater had low conductivity, it possessed a high negative zeta potential of 59.19 mV. The organic solute molecular weight in the CMP wastewater, measured using GPC, was 233 Da. In this study, UF and RO membrane modules were used to examine the potential application of these membrane processes to the treatment of CMP wastewater. Both membrane modules were purchased from Aspring Co. (Taiwan). The maximum operating pressure, membrane area, and pore size of the UF (RO) membrane module were 90 (250) psi, 9.57 103 (9.57 103) m2, and 40 (0.3–0.5) nm, respectively. The pore size of the UF membrane was estimated from its surface morphology, obtained using a scanning electron microscope (Hitachi S-800); the pore size of the RO membrane was obtained from the literature (Munir, 1998). The other surface characteristics of the UF and RO membranes are listed in Table 1. The water contact angle, surface roughness, and surface zeta potential were determined using a FACE contact angle meter (CA-D type, Kyowa Kaimenagaku), an atomic force microscope (Nanoscope SPM, Digit Ins.), and measurement of the electromotive potential between two parallel membrane surfaces (R&D Center for Membrane Technology, Chung Yuan University, Taiwan), respectively. As indicated in Table 1, both membranes had negative charges at pH 7–10; both the hydrophobicity and roughness of the RO membrane were larger than those of the UF membrane. Prior to performing the filtration experiments, the UF and RO membranes were both conditioned by passing deionized water through them over 24 h to reach their steady states. The operating parameters of the UF (RO) membrane process were set as follows: pressure, 1 (10) kg/cm2; temperature, 303 (303) K; crossflow velocity, 0.3 (0.1) m/s; testing cycle, 4 (4) h. A schematic representation of the UF and RO membrane processes used in this study is provided in Fig. 1. The feed water for the UF process was taken from the feed tank and purified through a 1-m filter. For the permeate flux experiments of the UF process, both the concentrated and permeate flow were recirculated to the feed tank, which was a closed-loop controlled-pressure system. For experiments of the permeate water quality, the concentrated stream was recirculated to the feed tank and the permeate flow was collected in the permeate tank as the feed water for the RO process. For studies of the water recovery during the UF or RO process, the feed water for the RO process was collected from the permeate stream of the UF process at 60% water recovery. Moreover, this study utilized a background run of 10.8% water recovery without recirculation of both the permeate and concentrated flows. Each permeate sample was collected and analyzed for its conductivity, turbidity, and TOC to examine the degree of removal of inorganic compounds, fine particles, and organic compounds/colloid materials, respectively, in the CMP wastewater. Results and Discussion Permeate flux curve UF crossflow filtration. Figure 2 indicates that the variation, with respect to the filtration time, of the permeate flux for the
TREATMENT OF CHEMICAL MECHANICAL POLISHING WASTEWATER TABLE 1.
BASIC SURFACE CHARACTERISTICS
Membrane (module) UF (PVDF) RO (TFC) aMeasured
OF THE
UF
1093 AND
RO MEMBRANES
Water content angle
Surface roughness (nm)
Zeta potentiala (mV)
59.6 1.7 78.4 1.3
11.58 0.65 104.44 2.61
16.1 to 20.3 26.9 to 32.3
over the pH range of 7–10.
UF procedure displayed exponential-type behavior. The permeate flux variation induced by membrane blockage during the filtration process can usually be divided into three stages (Konieczny, 2002). In the first stage, the pores are partially blocked by particles, decreasing the effective filtration area, resulting in a sharp drop in the permeate flux. Next, a cake layer is formed on the surface of the membrane; hydraulic interruption increases, and the permeate flux decreases gradually with respect to the operation time. Finally, the permeate flux approaches a steady state. Such a blocking mechanism is described well by the Hermia model (Hermia, 1982): dt d2t 2 k dv dv
n
(1)
where t is the filtration time (s), v is the filtrate linear velocity (m/s), and k and n are constants characterizing the blocking mechanisms of filtration process [n 2: complete membrane blocking; n 3/2: standard blocking of pores (inside the pores); n 1: transitory blocking of pores; n 0: cake blocking of pores]. It is noteworthy that the value of the constant k will vary with the value of n (Hermia, 1982; Lee and Clark, 1998; Konieczny, 2002). As indicated in Fig. 2, the experimental flux curve is described poorly by the Hermia model when using a single value of n; the filtration data at the initial and final stages do, however, fit the model well under the conditions of standard blocking (n 3/2, see Fig. 2a) and cake blocking of pores (n 0), respectively. As is ev-
ident from Fig. 2c, the standard blocking of pores appears to be the major blocking mechanism during the initial stages of filtration (within 7 min), while the cake blocking of pores might be the major blocking mechanism for the final filtration stage. The constant k determined from fitting the initial (final) experimental flux data to the standard blocking of pores (cake blocking of-pores) condition was 10.04 m3/2 min1/2 (0.59 107 m6 min). RO crossflow filtration. Figure 3 displays the pattern of permeate flux with respect to the filtration time for the RO procedure performed with pretreatment through both the 1-m filter and the UF membrane. The RO permeate flux declined significantly when UF pretreatment was not applied. Moreover, for the 1-m/UF/RO process, the flux pattern quickly reached the steady state (ca. 20 min), whereas for pretreatment through 1-m filtration only, the permeate flux continued to decline after 240 min. Both the slight variation of flux with time and the rapidity of reaching the steady state for the RO process after UF pretreatment may be ascribed to the fact that UF pretreatment removed 99% of the fine particles from the CMP wastewater. Water quality of permeate UF process. Figure 4 displays the dependence of the conductivity, turbidity, and TOC on the water recovery for the
Return permeate (only operate during flux experiment) Bypass
P
Feed tank
Permeate Tank
Membrane unit
S
Pump P
Concentrate Tank P
: Pressure gauge : Flow control valve
S
FIG. 1.
: Sampling
Schematic representation of the laboratory-scale UF and RO membrane processes performed in this study.
1094
JUANG ET AL. the permeate stream increased slightly upon increasing the water recovery, presumably as a result of the increased concentration of inorganic salts in the permeate. Because the major particle size of the fine particles in the CMP wastewater (70 nm) was larger than the pore diameter of the UF membrane (40 nm), almost all of the fine particles were screened out by the UF membrane. We note that the pore size of the UF membrane was on the same order as the fine particles’ size; therefore, we suspected that irreversible fouling in UF membrane might have occurred. The scanning electron microscopy images of the UF membrane that had been backwashed after the filtration process indicate, however, that the effect of filtration on the UF membrane surface was insignificant. As demonstrated in Fig. 4b, although the increased water recovery did induce a significant increase in the turbidity of the feed water, it had no obvious effect on the turbidity of the permeate. The turbidity removal efficiency of the UF membrane was greater than 98%. For the removal of organic compounds, although the organic compounds present in the wastewater were small (ca. 233 Da), they were partially removed as a result of electrostatic repulsion between their negative charges and that of the membrane surface, hydrophobic interactions with the membrane surface, and the screening mechanism introduced by the cake formed on the membrane surface. Our observation (Fig. 4c) that the TOC of the permeate decreased upon increasing the water recovery supports the occurrence of the screening mechanism. RO process. Figure 5 displays the effects of the water recovery on the conductivity, turbidity, and TOC of both the feed water and the permeate of the RO process pretreated through the 1-m filtration and the UF membrane. The values of the conductivity, turbidity, and TOC were further reduced after the RO process was performed. Moreover, the increased water recovery did not induce a significant change in the water quality of the permeate. It is well known that the removal mechanisms of UF and RO processes for inorganic compounds, fine particles, and organic compounds are distinct. For the RO process, the major mechanism for the removal of inorganic salts is the dissolution/diffusion process. Moreover, because the surface charge of the RO membrane
FIG. 2. (a) Fitting the flux curve of the UF process to the Hermia model with different values of n; (b) fitting the final flux curve of the UF process to the Hermia model with different values of n; (c) fitting the initial and final flux curves of the UF process to the Hermia model with values of n of 3/2 and 0, respectively.
feed water and the permeate of the UF process after pretreatment through 1-m filtration. For the removal of inorganic salts, there was a repulsive effect toward the anions in the wastewater, caused by the negative charge of the membrane (see Table 1). Moreover, the cations in the wastewater were also rejected to sustain the charge neutrality of the solution. As a result, the conductivity of the feed water containing the concentrated permeate from the recirculated flow increased moderately upon increasing the water recovery, as indicated in Fig. 4a. On the other hand, the conductivity of
FIG. 3. Flux curves of the permeate from the 1-m/RO and 1-m/UF/RO processes.
TREATMENT OF CHEMICAL MECHANICAL POLISHING WASTEWATER
1095
of water recovery, as indicated in Fig. 5b. Because the rejection of an organic solute by an RO membrane is closely related to the membrane properties (e.g., the molecular weight cutoff, degree of desalting, and the membrane’s porosity, morphology, charge, and hydrophobicity), the solute properties (e.g., the molecular weight, molecular size, charge, and hydrophobicity), and the feed water chemistry (Bellona et al., 2004), it is difficult for us to comment on the rejection mechanism for the organic compounds subjected to the RO mem-
FIG. 4. Dependence of the water quality of both the feed water (containing the concentrated flow from recirculation) and the permeate on the water recovery for CMP wastewater filtered through the UF membrane; (a) conductivity, (b) turbidity, and (c) TOC. Operating conditions: pressure, 1 kg/cm2; crossflow velocity, 0.3 m/s.
was negative (see Table 1), its repulsion of the anions in the wastewater might also have contributed to the removal of the inorganic salts. The residual conductivity of the permeate was reduced to less than 9 S/cm, as indicated in Fig. 5a. Although the fine particles were nearly all screened out by the UF membrane, applying the RO process could decrease their content further because the pore size of the RO membrane was smaller (3–5 Å) than that of the UF membrane. The turbidity was reduced to 0.01 NTU for each value
FIG. 5. Dependence of water quality of both feed water (containing the concentrated flow from recirculation) and permeate on the water recovery for CMP wastewater filtered through the RO membrane; (a) conductivity, (b) turbidity, and (c) TOC. Operating conditions: pressure, 10 kg/cm2; crossflow velocity, 0.1 m/s.
1096
JUANG ET AL.
FIG. 6. Relationships between the possible removal mechanisms and the water quality of the permeate for both the UF and RO membrane processes.
brane process used in this study when considering only the present data. There is evidence, however, that certain trace organic compounds are not completely removed during RO treatment (Bellona et al., 2004), as indicated in Fig. 5c, which shows that the TOC was reduced to values of 1.44–1.60 mg/L after performing the RO process. Finally, it was found that the surface roughness of the UF membrane before and after using it to filter the CMP wastewater was 38.8 1.1 and 31.2 0.7 nm, respectively, but there was no significant corresponding change in the surface roughness of the RO membrane. These results indicate that the solutes deposited on
CMP wastewater
Conductivity Turbidity TOC
110 S/cm 108 NTU 8.8 mg/L
the valleys of the membrane surface did indeed decrease the UF membrane’s surface roughness (Lee et al., 2004). As discussed above, the variations of both the flux and water quality of the permeate are closely related to the removal mechanisms in operation during the UF and RO processes. Figure 6 summarizes the relationships between the variations of both the flux and water quality of the permeate and the possible removal mechanisms (Bellona et al., 2004). It is evident that the removal efficiencies of the contaminants in the CMP wastewater when using the combination of UF and RO processes increased in the following or-
Permeate
1m
Conductivity Turbidity TOC
110 S/cm 42.7 NTU 8.8 mg/L
Permeate
UF
Conductivity Turbidity TOC
70 S/cm 0.23–0.34 NTU (99.7–99.8%)* 7.9–8.3 mg/L (5.7–10.2%)*
Permeate Conductivity RO
Turbidity TOC
5–6 S/cm (94.5–95.5%)* 0.01 NTU (99.9%)* 1.2–1.6 mg/L (81.8–86.4%)*
*: removal efficiency
Classification of industrial water reuse cooling water, boiling water, and the feed for water purification machines
FIG. 7.
Water quality of the permeate for CMP wastewater subjected to the 1-m filter, UF, and RO processes.
TREATMENT OF CHEMICAL MECHANICAL POLISHING WASTEWATER der: TOC conductivity turbidity. Thus, the use of the membrane-based processes provided highly efficient removal of both turbidity and conductivity, and had good potential for reducing the TOC. Reuse of permeate through different filtration processes As mentioned earlier, one major objective for the treatment of CMP wastewater with UF and RO processes is the potential reuse of the water permeate obtained. Figure 7 displays the variations of the conductivity, turbidity, and TOC of the CMP wastewater during the 1-m, UF, and RO filtration processes and the potential reuse of the permeate as an industrial water source. The 1-m filtration process removed a small amount of the fine particles (from 108 to 42.7 NTU), but it was useless for the separation of organic and inorganic compounds. The UF membrane process removed a large quantity of fine particles (presumably through the screening mechanism) and small amounts of both the organic and inorganic compounds (presumably through electrostatic repulsion/hydrophobic interactions and a repulsive effect to sustain the charge neutrality of the solution, respectively); the conductivity, turbidity, and TOC were reduced to 70 S/cm, 0.34 NTU, and 8.3 mg/L, respectively. Finally, the values of the conductivity, turbidity, and TOC of the permeate after performing the RO process were reduced dramatically, to 6 S/cm, 0.01 NTU, and 1.6 mg/L, respectively. The water quality of the permeates from the UF and RO membrane filtration processes both met the water reuse standards for tap water, cooling water, boiler water, and the feed of water purification machines. Conclusions The treatment of CMP wastewater from a semiconductor plant through UF and RO membranes was investigated with the emphasis placed on the removal of conductivity, the content of fine oxide particles, and TOC. We draw the following conclusions: 1. The major blocking mechanisms of the UF membranebased process at the initial and final stages were the standard blocking of pores (caused by colloidal materials) and the cake blocking of pores (caused by suspended particles), respectively. For use as pretreatment membranes, we found that the UF membrane removed more colloidal material and organic substances than did the 1-m filter. Moreover, because UF pretreatment removed 99% of the fine particles, the permeate flux of the 1-m/UF/RO unit varied only slightly with respect to time. 2. The values of the conductivity, turbidity, and TOC of the CMP wastewater reduced from 110 S/cm, 108 NTU, and 8.8 mg/L, respectively, to 70 (5–6) S/cm, 0.23–0.34 (0.01) NTU, and 7.9–8.3 (1.2–1.6) mg/L, respectively, after treatment using the 1-m filter/UF (1-m/UF/RO) process. The water quality of the permeate from the 1-m/UF and 1-m/UF/RO membrane filtration processes met the water reuse standards for tap water, cooling water, boiler water, and the feed for water purification machines. Acknowledgments This study was performed under the financial support of the National Central University/Industrial Technology Re-
1097
search Institute (grants NCU-ITRI 940301 and NCU-ITRI 951303-5). Author Disclosure Statement The authors declare that no competing financial interests exist. References Abdessemed, D., and Nezzal, G. (2002). Treatment of primary effluent by coagulation-adsorption-ultrafiltration for reuse. Desalination 152, 367. Bellona, C., Drewes, J.E., Xu, P., and Amy, G. (2004). Factors affecting the rejection of organic solutes during NF/RO treatment—A literature review. Water Res. 38, 2795. Corlett, G. (2000). Targeting water use for chemical mechanical polishing. Solid State Technol. 43, 201. Den, W., and Huang, C. (2005). Electrocoagulation for removal of silica nano-particles from chemical–mechanical–planarization wastewater. Colloids Surface A: Physicochem. Eng. Aspects 254, 81. Durham, B., Bourbigot, M.M., and Pankratz, T. (2001). Membranes as pretreatment to desalination in wastewater reuse: Operating experience in the municipal and industrial sectors. Desalination 138, 83. Hermia, J. (1982). Constant pressure blocking filtration laws— Application to power-law non-Newtonian fluids. Inst. Chem. Eng. 60, 183. Karakulski, K., and Morawski, A.W. (2000). Purification of copper wire drawing emulsion by application of UF and RO. Desalination 131, 87. Kim, S.L., Chen, J.P., and Ting, T.P. (2002). Study on feed pretreatment for membrane filtration of secondary effluent. Sep. Purif. Technol. 29, 171. Konieczny, K. (2002). Modeling of membrane filtration of natural water for potable purposes. Desalination 143, 123. Koopal, L.K., Goloub, T., De Keizer, A., and Sidorova, M.P. (1999). The effect of cationic surfactants on wetting, colloid stability and flotation of silica. Colloids Surface A: Physicochem. Eng. Aspects 151, 15. Lee, N., Amy, G., Croué, J.P., and Buisson, H. (2004). Identification and understanding of fouling in low pressure membrane (MF/UF) filtration by natural organic matter (NOM). Water Res. 38, 4511. Lee, Y., and Clark, M.M. (1998). Modeling of flux decline during cross-flow ultrafiltration of colloidal suspensions. J. Membr. Sci. 149, 181. Li, S.H., and Miller, R.O. (2000). Chemical Mechanical Polishing in Semiconductor Processing. San Diego, CA: Academic Press. Lin, S.H., and Yang, C.R. (2004). Chemical and physical treatments of chemical mechanical polishing wastewater from semiconductor fabrication. J. Hazard. Mater. 108, 103. Marcucci, M., Nosenzo, G., Capannelli, G., Ciabatti, I., Corrieri, D., and Ciardelli, G. (2001). Treatment and reuse of textile effluents based on new ultrafiltration and other membrane technologies. Desalination 138, 75. Munir, C. (1998). Ultrafiltration and Microfiltration Handbook. Lancaster, PA: Technomic. Pan, J.R., Huang, C.P., Jiang, W., and Chen, C.S. (2005). Treatment of wastewater containing nano-scale silica particles by dead-end microfiltration: evaluation of pretreatment methods. Desalination 179, 31. Qin, J.J., Oo, M.H., Wai, M.N., Ang, C.M., Wong, F.S., and Lee, H. (2003). A dual membrane UF/RO process for reclamation
1098 of spent rinses from a nickel-plating operation—A case study. Water Res. 37, 3269. Shon, H.K., Vigneswaran, S., Kim, I.S., Cho, J., and Ngo, H.H. (2004). Effect of pretreatment on the fouling of membranes: application in biologically treated sewage effluent. J. Membr. Sci. 234, 111. Steigerwald, J.M., Murarka, S.P., and Guttmann, R.J. (1997). Chemical Mechanical Planarization of Semiconductor Materials. New York: John Wiley.
JUANG ET AL. Wong, F.S., Qin, J.J., Wai, M.N., Lim, A.L., and Adiga, M. (2002). A pilot study on a membrane process for the treatment and recycling of spent final rinse water from electroless plating. Sep. Purif. Technol. 29, 41. Yalcin, F., Koyuncu, I., Oztürk, I., and Topacik, D. (1999). Pilot scale UF and RO studies on water reuse in corrugated board industry. Water Sci. Technol. 40, 303. Yang, G.C.C., Yang, T.Y., and Tsai, S.H. (2003). Crossflow electromicrofiltration of oxide-CMP wastewater. Water Res. 37, 785.
