Você está na página 1de 9

Journal of Membrane Science 267 (2005) 1826

Iron and manganese removal and membrane fouling during UF in conjunction with prechlorination for drinking water treatment
Kwang-Ho Choo , Haebum Lee, Sang-June Choi
Department of Environmental Engineering, Kyungpook National University, 1370 Sankyeok-Dong, Buk-Gu, Daegu 702-701, Republic of Korea Received 7 December 2004; received in revised form 17 May 2005; accepted 24 May 2005 Available online 24 June 2005

Abstract The removal of various levels of iron and manganese along with chlorine dosages from lake water was investigated using different ultraltration (UF) systems in conjunction with an in-line prechlorination step. In particular membrane fouling, caused by oxidized iron and manganese particles, was assessed in depth with visualization of the membrane surfaces. For feedwater containing 1.0 mg/L of Fe and/or 0.5 mg/L of Mn, substantial iron removal was achieved even without the addition of chlorine due to the oxidation of ferrous iron to ferric by dissolved oxygen and the consequent formation of less soluble iron hydroxide particles. Only negligible amounts of manganese removal occurred in the absence of chlorine, but with a dose of chlorine the manganese removal efciency increased markedly and reached a level of more than 80% (corresponding to less than 0.1 mg/L Mn) at a chlorine dosage of 3 mg/L as Cl2 . With a higher dosage of chlorine (e.g., 5 mg/L Cl2 ), there was no signicant increase in the removal of metal ions but more serious membrane fouling occurred. Also, oxidized manganese claimed a greater responsibility for membrane fouling during UF with chlorination. This phenomenon was in close association with the kinetics of manganese oxidation and its oxidized particles deposition inside the pores during backwashing, rather than its accumulation on top of the membrane skin layer. Turbidity and natural organic matter (NOM) removal levels of efciency were enhanced with the addition of chlorine in the presence of iron and manganese because the metal oxides, created by chlorination, could serve as adsorbents. 2005 Elsevier B.V. All rights reserved.
Keywords: Ultraltration; Iron and manganese; Chlorination; Membrane fouling; Backwashing

1. Introduction Ultraltration (UF) technologies recently have become more popular in drinking water treatment, since they can control small pathogenic microorganisms such as viruses very effectively. The use of UF membranes and their hybrid processes is thus considered as an attractive option in solving the issues related to viruses and protozoan oocysts as well as disinfection/disinfection by-products [15]. UF alone, however, is still unable to fully eliminate dissolved inorganic constituents such as iron and manganese that can deteriorate the quality of water with respect to taste and color. Groundwater and some water from the bottom anoxic zones

Corresponding author. Tel.: +82 53 950 7585; fax: +82 53 950 6579. E-mail address: chookh@knu.ac.kr (K.-H. Choo).

of reservoirs often contain such metal ions as reduced forms of Fe2+ and Mn2+ or their complexes with natural organic matter (NOM) [6]. In conventional treatment, the oxidation of iron and manganese was thus carried out using various oxidants such as oxygen, chlorine, ozone, or potassium permanganate, the precipitates of which could be then removed by either clarication or sand ltration, or both. The chemistry of oxidation becomes complicated when background species such as phosphate and fulvic acid are involved, so that even the oxidation of ferrous ion, that can be normally readily oxidized, is retarded [7]. The removal of metals by oxidation followed by depth ltration is greatly dependent on the size of the metal precipitates. In recent years, various treatment technologies have been employed to enhance water quality by removing inorganic contaminants. Ion exchange resins are able to remove many

0376-7388/$ see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.05.021

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

19

inorganic metal ions from drinking water including iron and manganese, and the formation of insoluble metal oxides improves the performance delaying the breakthrough of the column [8]. The removal of manganese from potable water was tested using a biological tricking lter. The removal of up to 94% manganese was achieved at long retention times but it decreased in the presence of ammonia and iron [9]. Tight reverse osmosis (RO) ltration and nanoltration (NF) membranes were used and compared in the tests of the treatment of contaminated water and wastewater containing Mn2+ and Fe2+ [10,11]. Obviously, the RO membrane removes manganese better than the NF one, but it still could not meet the WHO guidelines (0.1 mg/L Mn) [10]. However, in NF for the treatment of chlorine-free pulp mill bleaching efuent, the polymeric NF membrane provided almost 100% removal of manganese and iron, although the ceramic NF membrane only provided a low removal of metals [11]. Membrane cleaning is also an issue because simple water ushing is unable to remove manganese, so a weak, alkaline solution containing NH3 was selected in order to avoid manganese precipitation [10]. A laboratory-made ion exchange UF membrane was tested to remove Fe2+ and Mn2+ , but large amounts of the metal ions (more than 74%) passed through the membrane when humic acid or precipitation were not involved [12]. Microltration (MF) after oxidation using potassium permanganate (KMnO4 ), hydrogen peroxide, and manganese sand ltration was performed to remove iron and/or manganese from groundwater [13,14]. In the case that KMnO4 was used as an oxidizing agent, the MF experiments were conducted only with a suspension containing the already oxidized iron and manganese [13]. Needless to say, there was sufcient removal of the metals, but it seemed unrealistic from a practical point of view for the treatment of drinking water. Manganese sand ltration followed by MF was more promising in the removal of Mn than H2 O2 /MF, whereby both performances relied on a pH of solution and reaction time [14]. Also, membrane fouling still needed to be rectied though levels were reduced with the Mn sands. Manganese and iron were found to cause membrane fouling in the treatment of drinking water and membrane gas transfer applications in addition to NOM [1517]. It was reported, however, that oxidized iron and manganese particles reduced or prevented membrane fouling by NOM while forming a protective coat layer on top of the membrane [3,5,18]. Thus, it is still not clear as to how those metals sometimes exacerbate membrane fouling but sometimes do not. In this paper, therefore, UF with in-line prechlorination was investigated with regard to the removal of iron and manganese from water since little information is available on this approach. The removal of iron and manganese from lake water using bench-scale UF systems, in conjunction with a prechlorination step was evaluated using different concentration levels of iron and manganese and chlorine. During long-term UF experiments at the real water treatment plant, the efciency of turbidity and NOM removal was also examined along with the monitoring of membrane perme-

ability change. To illuminate membrane fouling mechanisms, the characteristics and dynamics of oxidized particles were explored systematically in the laboratory using scanning electron microscopy (SEM).

2. Materials and methods 2.1. Feedwater Water samples used for this study were obtained from the Gosan Water Treatment Plant that receives raw water (RW) from the Unmoon Dam, which is a water supply source for the eastern part of the city of Daegu. The water treatment plant consists of the following typical physico-chemical processes: coagulation/occulation, clarication, rapid gravity ltration, and disinfection with chlorine before distribution. So, ltered water (FW) samples taken from the rapid gravity lter were also used as feedwater for some dead-end UF experiments (see Table 3). Some water samples collected were also moved to the laboratory for further analyses and tests. The key water quality characteristics are given in Table 1. The actual levels of concentration of iron and manganese present in the source water were nearly negligible, so that additional FeCl2 (Wako Pure Chemical Industries, Ltd., Japan) and/or MnCl2 (Daejung Chemicals and Metals Co., Ltd., Korea) were spiked depending on the experimental conditions tested (which are specied in Tables 2 and 3). Through the spiking of each metal into the feedwater, iron and manganese levels were adjusted to approximately 1.0 mg/L as Fe and 0.5 mg/L as Mn, respectively (which are often found in groundwater in Korea). 2.2. UF membranes The UF hollow ber membrane modules used in this study were supplied from Aquasource (France). The membranes were made of cellulose acetate and had a molecular weight cut-off (MWCO) of 100 kDa (corresponding to a nominal pore size of 10 nm). One module had ten bers with an effective surface area of 0.035 m2 . Each single ber in the module was 1.2 m long and had an inner diameter of 0.93 mm. Before use, the membranes were rinsed using distilled water. The at
Table 1 Quality of the water sources tested Parameter pH Temperature ( C) Turbidity (NTU) Alkalinity (mg/L as CaCO3 ) Fe (mg/L) Mn (mg/L) DOC (mg/L) UV254 (cm1 ) SUVA (L/mg m) ND: not detectable Raw water 6.97.3 6.914.5 1.093.0 12.815.4 <0.1 ND 1.393.45 0.0170.33 0.643.69 Filtered water 6.97.3 7.510.9 0.2 12.3 ND ND 1.141.98 0.0140.040 0.713.34

20

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

Table 2 Summary of operating conditions and removal efciency by crossow UF with and without prechlorination Phase Time period (day) Feedwater Source RW RW RW RW RW RW RW RW RW RW RW RW RW RW RW RW pH 7.0 7.3 6.9 7.0 7.0 7.0 7.1 7.1 7.1 7.1 7.2 7.2 7.0 6.8 6.8 6.8 Fe (mg/L) 0.0 0.0 0.9 1.0 1.0 0.7 1.0 1.0 1.1 1.1 0.0 0.0 0.4 1.0 1.0 1.0 Mn (mg/L) 0.0 0.0 0.4 0.5 0.5 0.4 0.5 0.5 0.0 0.0 0.6 0.5 0.6 0.5 0.5 0.5 Turb. (NTU) 2.0 1.1 2.4 4.2 5.0 3.6 5.9 5.9 6.3 6.3 8.7 7.4 8.6 6.3 8.9 9.9 UV254 (cm1 ) 0.022 0.019 0.065 0.045 0.049 0.051 0.059 0.059 0.052 0.052 0.025 0.025 0.025 0.062 0.057 0.050 DOC (mg/L) 1.9 2.2 1.7 2.3 2.8 2.2 2.7 2.7 1.6 1.6 1.6 1.2 1.2 1.2 1.9 1.9 Chorine dose (mg Cl2 /L) Removal (%) Fe Mn Turb. 83 73 93 98 96 96 96 97 96 97 98 98 98 98 98 99 UV254 0 19 66 73 37 47 55 53 34 73 19 31 20 73 74 75 DOC 14 22 30 14 19 28 13 24 27 29 17 27 34 39

A B1 B2 B3 C1 C2 C3 C4 D1 D2 D3 D4 D5 E1 E2 E3

011 1125 2527 2732 3234 3436 3637 3737 3738 3838 3840 4041 4143 4345 4546 4648

0.0 5.0 0.0 5.5 0.0 0.5 1.3 3.0 0.0 3.0 0.0 3.1 0.0 0.0 3.0 4.8

79 13 75 96 84 4 75 5 73 14 82 81 86 90 8 31 66 31 90 2 92 81 91 99

UF membranes (YM 100) used in the laboratory tests were purchased from Amicon (USA). The membranes were made of regenerated cellulose and had an MWCO of 100 kD and an effective ltration area of 28.7 cm2 . 2.3. Membrane system operation The bench-scale UF system used in this work is shown in Fig. 1. The whole system was mainly composed of a feed tank, a UF unit, and a data acquisition system. The system operation and data collection were automated using a programmable logic controller and an on-line computer. For prechlorination, a chlorine solution with 100 mg/L as Cl2 was injected before the feeding pump (P2) using a peristaltic pump (P1). During crossow UF, the circulation pump was operated to maintain a tangential velocity of 1 m/s inside the

hollow bers, but it stopped during dead-end UF. The UF system was designed to operate at a constant ux of 69 L/m2 h (corresponding to 40 mL/min). Periodic backwashing was performed at 1.5 bar using N2 gas with a 40-min cycle as follows: permeation, 39 min and 24 s; rst ush, 3 s; backwash, 30 s; and nal ush, 3 s. The overall water recovery corresponded to 91% of the water injected. The single ber system used in the lab for membrane fouling evaluation was designed in the same manner. To stress the lab system, dead-end UF was conducted at a substantially high ux of 150 L/m2 h. Stirred cell UF experiments were also performed using a 180-mL stirred cell plus an 800-mL reservoir (Amicon 8200, USA). The working volume of the batch unit was 980 mL, while the pressure applied was kept at 1.0 bar using a nitrogen cylinder. The stirring speed was adjusted to 160 rpm using a magnetic stirrer. During stirred ow UF, the permeate mass

Table 3 Summary of operating conditions and removal efciency by dead-end UF with and without prechlorination Phase Time period (day) Feedwater Source RW RW RW RW RW RW RW RW FW FW FW FW FW FW FW pH 6.9 6.6 6.7 6.7 6.5 7.0 6.9 6.9 7.1 7.1 6.8 7.0 7.1 7.1 7.0 Fe (mg/L) 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0 Mn (mg/L) 0 0.6 0.4 0.5 0 0.5 0.5 0.5 0 0.5 0.5 0 0 0.5 0.5 Turb. (NTU) 7 20 12 18 20 21 21 21 0.2 0.2 0.7 0 .5 0.3 0.2 0 .2 UV254 (cm1 ) 0.02 0.05 0.05 0.07 0.08 0.07 0.08 0.09 0.02 0.02 0.02 0.05 0.04 0.02 0.02 DOC (mg/L) 1.5 1.8 1.8 2.2 2.4 2.5 2.8 3.8 1.7 1.2 1.3 2.0 1.6 1.7 1.2 Chorine dose (mg Cl2 /L) Removal (%) Fe 97 97 87 98 100 99 97 99 Mn 36 26 81 7 1 86 5 84 Turb. 95 99 99 99 99 99 31 40 50 68 60 63 36 6 7 16 4 10 6 15 UV254 9 39 55 63 65 47 8 7 0 15 9 7 18 DOC 12 26 16 13 11 24

A1 A2 A3 A4 B1 B2 B3 B4 C1 C2 C3 D1 D2 D3 D4

040 4052 5255 5559 5964 6473 7373 7375 7578 7885 8589 8993 9397 97101 101105

0 0 1 3 0 0 3 0 0 0 3 0 3 0 3

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

21

Fig. 1. Schematic of a bench-scale UF system in combination with a prechlorination step.

was monitored and recorded using an electronic balance and a personal computer to evaluate the change of ux. The specic cake resistance of particles was also measured using the stirred cell unit in an unstirred mode. A cake layer containing 100 mg as Fe and/or Mn was rst formed at 0.2 bar and then the specic cake layer resistance corresponding to the cake resistance per unit mass of particles was determined at a wide pressure range of 0.52.0 bar [19]. 2.4. Analytical methods Feedwater and permeate samples were analyzed for key water-quality parameters such as iron, manganese, turbidity, UV absorbance at 254 nm, and dissolved organic carbon (DOC). The levels of iron and manganese concentration were analyzed using an inductively coupled plasma mass spectrometer (VG Elemental, PlasmaQuad 3, UK). The turbidity measurement was taken on a Hach turbidmeter (2100P, USA). The UV absorbance was determined using a UV/VIS spectrophotometer (Uvicon 942, Kontron, Italy), while the DOC concentration level was measured using a TOC analyzer (Sievers 8200, USA). Feedwater samples were ltered through a 0.45-m membrane (Millipore, USA) to remove particles before the measurements of UV absorbance and DOC concentration level. The chlorine dose was determined by the DPD method using a Hach spectrophotometer and reagent pillows. The surface of the membrane tested was visualized using a eld emission scanning electron microscope (S-4300, Hitach, Japan).

As shown in Fig. 2, the removal efciency of dissolved iron increased very rapidly and reached nearly 100% within 20 min even with a chlorine dose of as little as 0.5 mg/L as Cl2 . The manganese removal efciency was, however, not as effective as that for iron. Since the residual chlorine measured decreased from 1.0 to 0.39 mg/L, some other constituents in raw water were oxidized preferentially at the low chlorine

3. Results and discussion 3.1. Batch chlorination test To determine an appropriate dose of chlorine for the oxidation of iron and manganese, batch tests with feedwater containing 1.0 mg/L Fe and 0.5 mg/L Mn were conducted.

Fig. 2. Removal efciency of: (a) iron and (b) manganese over time at different chlorine dosages.

22

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

dosage. A relatively high chlorine dosage of approximately 2 mg/L as Cl2 was required in order to achieve a desirable manganese removal efciency of greater than 40% (corresponding to the current Korean maximal contamination level of 0.3 mg/L). This result is indicative of large amounts of chlorine required to remove both iron and manganese if manganese exists in the feedwater. Thus, it was necessary to examine the effect of chlorination on UF for the treatment of water containing different levels of iron and manganese. 3.2. Crossow UF with prechlorination Table 2 summarizes the overall performances of the combined prechlorination/crossow UF process under different operating conditions. The concentration of each item in feedwater and permeate, particularly related to Fe, Mn, and DOC, was very low, so small uctuations of their concentration may give negative removal efciency. Although we tried our best to eliminate any errors during experiments, some variations of data were unavoidable. However, it did not change the overall trend of the experimental results. As expected, the combined process operation removed approximately more than 70% of iron, which was slightly dependent on the initial iron concentration. Moreover, substantial iron removal was achieved even without prechlorination (conditions B2, C1, D1, E1). This could be attributed to the oxidation of ferrous iron to ferric by dissolved oxygen and the subsequent formation of iron hydroxide precipitates which were rejected by UF. As a result, the quality of the UF permeate could be guaranteed to meet the regulation criterion of 0.3 mg/L Fe. Manganese was relatively difcult to oxidize but normal, satisfactory manganese removal occurred with chlorination at a larger chlorine dosage of 3 mg/L. In one case (condition D4), however, only 31% of manganese was removed by a dose of approximately 3 mg/L Cl2 and with low chlorine dosages, negligible amounts of manganese was removed. Overall, during the UF treatment with the addition of chlorine, the removal efciency of iron and manganese was well observed at a high level based on the batch tests conducted in the laboratory, but further investigation regarding manganese removal was needed. During the combined process operation, the turbidity and UV removal efciency levels were also monitored. The turbidity was nearly completely removed since UF can reject most of the colloidal particles present in raw water. UV removal was much more substantial than that of DOC. This is possibly due to the oxidative degradation of the aromatic groups of NOM by chlorine that have strong UV absorbance at 254 nm [20]. It is interesting to note that relatively great UV and DOC removal was also observed, particularly when iron was present in feedwater. The above results imply that the material precipitated by oxidation (e.g., ferrihydrite) can play a part in removing NOM from water by sorption. It would be an additional benet of the oxidation of iron and manganese followed by the removal of the precipitates by UF. It was reported elsewhere that iron oxides could remove

Fig. 3. Variation of inlet transmembrane pressure over time in crossow UF. At the points of (*) and CC, physical and chemical cleanings were performed, respectively.

organic substances from river and lake water and wastewater efuent [3,5,21,22]. Fig. 3 displays the change of the transmembrane pressure during the 48-day UF operation in combination with chlorination. In Phase A, there were several signicant increases in pressure, which were caused by membrane lumen clogging alone. After placing a strainer starting from Phase B (refer to Fig. 1), pressure was maintained at a constant level as presented in the early stages of Phase B. A sharp rise in pressure occurred again when the water was chlorinated in the presence of iron and manganese (B3). At rst, this might be ascribed to the deposition of iron and manganese precipitates at the UF membrane. To further clarify it, the chlorine dose was changed stepwise from 0 to 3 mg/L throughout Phase C. The pressure started increasing when a large amount of chlorine was dosed (C4), i.e., membrane fouling seemed to occur when the oxidation of manganese was sped up. In Phases D and E, the increase of pressure always coincided with the increase of manganese removal. Consequently, it could be hypothesized that the deposition of oxidized manganese precipitates claimed a greater amount of responsibility for membrane fouling (which will be discussed later in detail). 3.3. Dead-end UF with prechlorination The efciency of water treatment through chlorination followed by dead-end UF is summarized in Table 3. The iron removal efciency was mostly kept at its highest level regardless of the feedwater quality and whether the water was chlorinated or not, as being in agreement with that of the crossow UF. The characteristic of manganese removal behaved roughly in the same manner. Other parameters such as turbidity, UV254 , and DOC also indicated similar trends as those in the crossow UF when raw water was fed into the system. They deteriorated when the feedwater was switched to ltered water obtained from the sand lters. This is probably due to the extremely low turbidity of ltered water, i.e., large colloidal particles in raw water contributed to the for-

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

23

Fig. 6. Variation of DOC in raw water and permeate over time in dead-end UF. Fig. 4. Variation of transmembrane pressure over time in dead-end UF. At the points of (*) and CC, physical and chemical cleanings were performed, respectively.

mation of secondary dynamic layers on top of the membrane surface while providing additional rejection of turbidity and NOM. Also, the size of the particles in the ltered water from the sand lter is smaller than that in the raw water, so their rejection by UF membrane is of course lower. The negative removal efciencies may occur due to very low level of concentration and its uctuations, but it did not change the overall trend of the experimental results as mentioned above. Changes in transmembrane pressure during the constantux UF operation are shown in Fig. 4. In Phase A, the two sudden leaps in pressure were caused by a malfunction of the in-line strainer, but after xing it there were not any technical problems in the system. When a signicant amount of manganese was removed by chlorination, serious membrane fouling occurred simultaneously (Phase A4) as mentioned above. The pressure, however, began to increase steadily without chlorination (Phase B2), as opposed to the previous results, which seemed to be related to the increase of either turbidity or the NOM levels or both (Figs. 5 and 6). Considering the variations of turbidity during Phases A and C

Fig. 5. Variation of turbidity in raw water and permeate over time in deadend UF.

through D, the turbidity level appeared not to directly affect it. Thus, it was believed that the higher NOM concentration was more responsible for the membrane fouling in Phase B2. Fouling became much worse with chlorination during the rest of Phase B. In addition, the DOC level of UF permeate was sometimes higher than that of raw water as shown in Fig. 6. A similar nding was also reported elsewhere [3]. It is still not clear but it might be caused by the concentration polarization of DOC at the membrane surface resulting in some uctuations in the DOC level of permeate during UF. Another possibility is that chlorine can affect the chemistry of NOM in water leading to the reduction in NOM rejection by UF, which might be in association with the concentration level of colloids in feedwater [20]. In order to minimize the effect of turbidity and NOM on membrane permeability, ltered water was used as feedwater during Phases C through D. With the tests, it was clear that oxidized manganese did affect membrane fouling a lot more than the iron oxides. Nevertheless, there was a question as to why the two inorganic precipitates had a different effect on fouling. On the other hand, the performance of dead-end UF with prechlorination seemed to be more stable in terms of the removal of contaminants and membrane permeability, even though the fouling caused by oxidized manganese was still occurring. It is unclear why dead-end UF is more stable and we need to further examine this phenomenon. But one possible explanation is that during dead-end UF, relatively uniform deposit layers can be formed under even distribution of transmembrane pressure inside the hollow ber leading to higher stability in removal and permeability with backwashing. However, there was a signicant pressure drop (1.0 bar) along the ber during crossow UF, probably resulting in the formation an irregular deposit layer at the membrane. Since the dead-end UF process did not have a circulation pump, its energy consumption would be much less than that of the crossow membrane process. Thus, dead-end UF seemed to be more economical and reliable for water treatment from a practical point of view.

24

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

Fig. 7. Effect of chlorination on membrane permeability during stirred ow UF of feedwater containing different amounts of iron and manganese.

Fig. 9. Effect of chlorination on pressure differential during single hollow ber UF of feedwater containing different amounts of iron and manganese. No backwashing was applied to any of the tests.

3.4. Membrane fouling mechanisms To elucidate the cause of membrane fouling during chlorination/UF, batch stirred cell UF was conducted with different combinations of iron and manganese levels in the feedwater (Fig. 7). Unexpectedly, however, the membrane permeability always increased whenever amounts of oxidized iron and/or manganese were present in feedwater. Also, the manganese oxide particles offered a relatively low specic resistance of the cake compared to the iron oxide particles and their mixture (Fig. 8). Results of the batch tests suggest that manganese particles contribute barely to membrane fouling as compared to the iron ones. This seemed to contradict the ndings from the previous, continuous tests. It could be assumed that the difference between the continuous and batch tests stemmed from modular types such as hollow bers and at sheets. There might be several possibilities related to the geometry and hydrodynamics of the modules: (1) hollow bers can be easily clogged, particularly by manganese oxide particles; and (2) the tangential ow inside the hollow bers may change the deposition of particles and fouling behavior may occur. As shown in Fig. 9, oxidation of the iron and manganese was always benecial with regard to membrane permeability

Fig. 10. Effect of chlorination on pressure build-up during single hollow ber UF along with backwashing. Filtered water containing 0.5 mg/L Mn was fed for all runs. In run 1, no chlorine was added, while 3 mg/L Cl2 was dosed for runs 2 and 3. For run 3, backwashing was done with chlorine-free acidic solution (pH 3).

even when hollow bers were used. This revealed that our reasoning given above turned out to be incorrect. The only possibility left to consider would be backwashing, which had never been referred to as a cause of membrane

Fig. 8. Specic cake resistance of iron and manganese particles and a 1:1 ratio mixture (in mass base).

Fig. 11. Illustration of possible mechanisms of membrane fouling caused by oxidized particles during prechlorination with and without backwashing (BW).

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826

25

Fig. 12. SEM pictures of the UF hollow ber membranes tested: (a) virgin membrane (lumen side); (b) virgin membrane (shell side); (c) used membrane (lumen side); and (d) used membrane (shell side).

fouling but rather it was believed it could prevent it. Surprisingly, as shown in Fig. 10, UF of chlorinated water with normal backwashing aggravated membrane fouling unlike those without backwashing and with backwashing using an acidic solution (pH 3). Thus, it was necessary to explain why normal backwashing increased the level of fouling in this work. Fig. 11 illustrates a possible mechanism of the fouling phenomenon during water treatment by prechlorination/UF. The accumulation of the metal oxide particles on the membrane surface may not bring about severe fouling but rather can improve permeability. This is possible because the oxide particles can sorb NOM, which is also known as a foulant. Similar ndings by other researchers reported that metal oxide adsorption contributed to the reduction in fouling [3,5]. Unlike iron, the speed of the oxidation of manganese is relatively slow (refer to Fig. 2), so some of the manganese particles are still small enough to pass through pores after chlorination. They can grow continuously later on and thereby, plug the pores during the backwashing stage.

If an additional dose of chlorine into the permeate during backwashing is done as it normally does to prevent microbial growth at the membrane, it may accelerate the growth of manganese oxides inside the pores. Consequently, it could be claimed that fouling was affected by the kinetics of the oxidation of manganese and the operation of backwashing. The SEM pictures of the UF membranes used conrmed the deposition of manganese precipitates on the shell side of the membrane during UF with chlorination (Fig. 12). Thus backwashing with an acidic solution can help reduce fouling caused by inorganic precipitation, instead of chlorine addition (run 3 of Fig. 10).

4. Conclusions Prior to UF, chlorine was added to remove iron and manganese from drinking water. The effects of chorine doses and different levels of iron and manganese on treatment efciency

26

K.-H. Choo et al. / Journal of Membrane Science 267 (2005) 1826 [3] Y.J. Chang, K.H. Choo, M.M. Benjamin, S. Reiber, Combined adsorption/UF process increases TOC removal, J. AWWA 90 (5) (1998) 90. [4] S.J. Lee, K.H. Choo, C.H. Lee, Conjunctive use of ultraltration with powdered activated carbon adsorption for removal of synthetic and natural organic matter, J. Ind. Eng. Chem. 6 (2000) 357. [5] K.W. Lee, K.H. Choo, S.J. Choi, K. Yamamoto, Development of an integrated iron oxide adsorption/membrane separation system for water treatment, Water Sci. Technol. Water Sup. 2 (2002) 293. [6] M. Zaw, B. Chiswell, Iron and manganese dynamics in lake water, Water Res. 33 (1999) 1900. [7] A. Wolthoorn, E.J.M. Temminghoff, L. Weng, W.H. van Riemsdijk, Colloid formation in groundwater: effect of phosphate, manganese, silicate and dissolved organic matter on the dynamic heterogeneous oxidation of ferrous iron, Appl. Geochem. 19 (2004) 611. [8] K. Vaaramaa, J. Lehto, Removal of metals and anions from drinking water by ion exchange, Desalination 155 (2003) 157. [9] A. Gouzinis, N. Kosmidis, D.V. Vayenas, G. Lyberatos, Removal of Mn and simultaneous removal of NH3 , Fe and Mn from potable water using a trickling lter, Water Res. 32 (1998) 2442. [10] R. Molinari, P. Arguiro, L. Romeo, Studies on interactions between membranes (RO and NF) and pollutants (SiO2 , NO3 , Mn2+ and humic acid) in water, Desalination 138 (2001) 271. [11] A. Lastra, D. G omeza, J. Romerob, J.L. Francisco, S. Luque, J.R. Alvarez, Removal of metal complexes by nanoltration in a TCF pulp mill: technical and economic feasibility, J. Membr. Sci. 242 (2004) 97. [12] M. Kabsch-Korbutowicz, T. Winnicki, Application of modied polysulfone membranes to the treatment of water solutions containing humic substances and metal ions, Desalination 105 (1996) 41. [13] D. Ellis, C. Bouchard, G. Lantagne, Removal of iron and manganese from groundwater by oxidation and microltration, Desalination 130 (2000) 255. [14] Z. Teng, J.Y. Huang, K. Fujita, S. Takizawa, Manganese removal by hollow ber micro-lter. Membrane separation for drinking water, Desalination 139 (2001) 411. [15] Y.I. Kaiya, K. Fujita, S. Takizawa, Study on fouling materials in the membrane treatment process for potable water, Desalination 106 (1996) 71. [16] W. Xu, S. Chellam, D.A. Clifford, Indirect evidence for deposit rearrangement during dead-end microltration of iron coagulated suspensions, J. Membr. Sci. 239 (2004) 243. [17] C.M. Kinna, D.W. Johnson, Iron fouling in membrane gas transfer applications, J. Membr. Sci. 205 (2002) 45. [18] M.H. Al-Malack, G.K. Anderson, Formation of dynamic membranes with crossow microltratioin, J. Membr. Sci. 112 (1996) 287. [19] S.A. Lee, K.H. Choo, C.H. Lee, H.I. Lee, T. Hyeon, W. Choi, H.H. Kwon, Use of ultraltration membranes for the separation of TiO2 photocatalysts in drinking water treatment, Ind. Eng. Chem. Res. 40 (2001) 1712. [20] T.W. Ha, K.H. Choo, S.J. Choi, Effect of chlorine on adsorption/ultraltration treatment for removing organic matter in drinking water treatment, J. Colloid Interface Sci. 274 (2004) 587. [21] K.H. Choo, S.K. Kang, Removal of residual organic matter from secondary efuent by iron oxides adsorption, Desalination 154 (2003) 139. [22] M. Zhang, C. Li, M.M. Benjamin, Y. Chang, Fouling and natural organic removal in adsorbent/membrane systems for drinking water treatment, Environ. Sci. Technol. 37 (2003) 1663.

and membrane permeability were investigated using batch and continuous membrane systems. Thus, the following conclusions could be drawn: A relatively large dosage (approximately 3 mg/L as Cl2 ) of chlorine was required to remove manganese by means of oxidation followed by UF, while substantial iron removal was possible even without chlorine addition. The formation of the precipitated material by chlorination contributed to an increase in turbidity and NOM removal, probably because of sorption of the oxidized particles. Unlike iron oxide particles, the oxidized manganese caused serious membrane fouling during crossow and deadend UF, along with slight contribution by NOM, while turbidity levels in feedwater, ranging from 0.2 to 90 NTU, had a minimal effect on fouling. Membrane fouling caused by oxidized manganese particles did not occur because of the formation of their cake layers on top of the membrane surface but because the particles that were grown after passing through the pores blocked them up during backwashing. This was an unusual case because backwashing aggravated membrane fouling due to the reaction kinetics of manganese. For iron and manganese removal in drinking water treatment, dead-end UF with prechlorination seemed more feasible than crossow UF. With the former, a relatively stable operation and low pressure were maintained during longtime in situ testing compared to that with the latter. Thus, it was found that tangential ow was not always necessary for a stable UF operation during water treatment.

Acknowledgements This work was supported by the Korea Research Foundation Grant (KRF-2003-042-D20083) and the MOE, Korea. The authors thank Mr. J.H. Lee of the Korea Basic Science Institute and Mr. J.W. Han of Gosan Water Works for their assistance in instrumental analyses and UF system operations, respectively. The provision of laboratory-scale hollow ber UF modules from Aquasource (France) is also appreciated.

References
[1] S.S. Madaeni, A.G. Fane, G.S. Grohmann, Virus removal from water and wastewater using membranes, J. Membr. Sci. 102 (1995) 65. [2] G.K. Pearce, M. Heijnen, J. Reckhouse, Using ultraltration membrane technology to meet UK Cryptosporidium regulations, Membr. Technol. 141 (2002) 6.

Você também pode gostar