Effect on water quality and control of chemically enhanced backwash by-products (CEBBPs) in the adsorption-ultrafiltration process

Currently, there is considerable attention focusing on the pollution problems of everyday potable water sources (Alzahrani et al. 2013; Wei et al. 2016). Because many pollutants, especially organic pollutants, get into the surface water through different ways, the potential risk of water source pollution is becoming more and more prominent; this contaminated surface water is referred to as micro-polluted surface water (Zhang et al. 2015; Kan et al. 2016). However, conventional water purification technology (coagulation – sedimentation – filter – disinfection) has some limits in treating micro-polluted surface water (Wang et al. 2016). Compared with conventional water purification technology, ultrafiltration (UF) and microfiltration (MF), which are representative of low-pressure membrane technology, can simplify the process, improve the utilization rate of raw water and occupy much less processing power of a traditional craft unit (Masindi et al. 2015).

Many studies showed that adsorption combined with membrane can increase the membrane flux and the removal efficiency of organic matter, which is the advantage of the adsorption-UF combination technology (Li et al. 2016). Meanwhile, membrane cleaning is an effective way to control membrane fouling and restore membrane filtration (Gibert et al. 2016).

Membrane cleaning methods include physical and chemical cleaning. The physical cleaning can only restore the reversible membrane fouling and the irreversible membrane fouling can only be accomplished through the chemical cleaning method (Kimura et al. 2016). Owing to the disadvantages of conventional chemical cleaning CIP (clean in place), including the high amount of time required, the complicated operation and the low degree of automation, the online chemical cleaning method, namely chemically enhanced backwash (CEB), has been paid more and more attention by people (Touffet et al. 2015). Compared with CIP, CEB required a relatively lower concentration of chemical reagent, which has a short contact time with membrane and can be conducted at room temperature. The transfer of chemical cleaning reagents to the membrane surface in the cleaning mechanisms was a crucial step. In chemical cleaning, oxidant-type cleaning reagents (such as sodium hypochlorite) with strong oxidation ability can effectively remove organic matter on the membrane surface and in the pores (Porcelli & Judd 2010). The sodium hypochlorite solution can effectively remove organic matter from the fouling of an irreversible membrane and recover the flux in the membrane (Kimura et al. 2004). Furthermore, the NaClO removal effect on organic pollutants under a higher pH value was better, and natural organic matter (NOM) degraded to carboxyl, ketone, and the aldehyde group by NaClO oxidation (Strugholtz et al. 2005).

Chemical cleaning is undoubtedly an effective method to solve membrane fouling, compared with the large amount of membrane fouling research, but there is very little on the study of membrane chemical cleaning. Meanwhile, the potential secondary pollution and the safety of drinking water quality because of membrane chemical cleaning is not yet a cause for concern. The research status of membrane chemical cleaning shows that the main membrane pollution is caused by NOM during drinking water membrane processing and NaClO is widely used for a cleaning reagent (Kan et al. 2016). In the CEB process, membrane cleaning reagents may react with membrane pollution to generate by-products. For instance, NaClO reacts with organic matter in membrane cleaning water to generate disinfection by-products that contain toxicity, mutagenicity, carcinogenicity and teratogenicity components. To ensure the water production rate, membrane filtration water is commonly used as backwash water during CEB and the wash water after CEB will not be discarded but go back to the raw water. Therefore, the potential impact of chemically enhanced backwash by-products (CEBBPs) produced in the cleaning process on water quality needs to be further evaluated.

The overall purpose of this project was to examine the typical types of CEBBPs produced during CEB in the adsorption-ultrafiltration process and analyse the influence of CEB hydraulic parameters (backwash interval, backwash duration, and backwash flux) and a cleaning reagent parameter (cleaning reagent concentration) on typical CEBBPs in the CEB regulation process. Meanwhile, the backwashing parameters of CEB were able to be optimized for the regulation of membrane cleaning technology by analysing the correlativity between the control of CEB parameters and typical CEBBPs concentration, examining the influence of parameters on membrane cleaning, and comparing the change of different membrane cleaning effects under CEB regulation. Human health risk assessment (HRA) was applied to assess the potential adverse health effects from exposure to effluent after the optimal online backwashing of CEB. This research topic will help complement the theory of adsorption/UF and improve the security of drinking water quality.

An adsorption-UF system was used in this study. Powdered activated carbon was utilized for adsorption. The adsorption-UF system is denoted as PAC/UF. A schematic diagram of the adsorption-UF is shown in Figure 1. The hollow fibre modules of a polyvinylidene fluoride (PVDF) UF membrane (Litree, China) UF membrane with 0.1 m² membrane area (20 cm length and 100 fibres) and a nominal pore size of 0.01 μm was used. Raw water was fed into a constant-level tank to maintain the water head for the membrane reactor. Certain doses of PAC were continuously fed into the UF membrane reactor with an effective volume of 1 L. The permeation through the submerged membrane module was continuously withdrawn using a peristaltic pump at a constant flux of 20 L/(m²·h). The trans-membrane pressure (TMP) was continuously monitored with a pressure sensor.

NaClO with an effective chlorine concentration of 10–200 mg/L was used as the cleaning reagent due to its widespread application in UF for membrane cleaning. The CEB hydraulic parameters were regulated by a programmable logic controller (PLC) system programmed with timing, counting and scaling code. The PLC system was used to automatically generate the required profiles and to control the valves, pumps and liquid level gauge of the system. To determine the optimum doses of PAC in the reactor, the influence of PAC doses (0, 2, 4, 6, 8, 10 and 12 mg/L) on the DOC and UV₂₅₄ removal rate of raw water was pre-examined through jar tests. The jar test results of different PAC doses are summarized in Table 1. It was found that when the dose of PAC was 6 mg/L, the DOC and UV₂₅₄ removal were the highest in the tested doses.

To confirm the optimum CEB cleaning parameter, a single factor experiment was used to analyse the influence of backwash duration (BD), backwash interval (BI), backwash flux (BF) and reagent concentration (RC) on CEBBPs concentration and membrane fouling. The single factor experiment of CEB parameters is shown in Table 2. Each level of factors was operated for 72 h.

Fourteen types of volatile halogenated organic compounds (VHOCs) were investigated, including 1,1-dichloroethene (1,1-DCE), methylene chloride (DCM), trans-1,2-dichloroethene (trans-DCE), chlorobutadiene (CBD), cis-1,2-dichloroethene (cis-DCE), chloroform (TCM), tetrachloromethane (CCl₄), 1,2-dichloroethane (1,2-DCA), trichloroethylene (TCE), bromodichloromethane (BDCM), tetrachloroethylene (PCE), dibromochloromethane (DBCM), tribromomethane (TBM), and hexachlorobutadiene (HCBD). The 14 VHOCs were analyzed by a static head-space gas-chromatographic technique (SHS-GC) using a Varian 3380 gas chromatograph equipped with a 63Ni electron-capture detector (ECD) and a Vocol capillary column (30 m × 0.53 mm I.D., film thickness: 3.0 m-Supelco). Quantitative and qualitative analyses of VHOC were managed using Dionex Chromeleon 6.0 software. Calibration was performed according to the external standard method. Standards of VHOCs were prepared with pure analysis reagents (Supelco) in specific concentrations of methanol. The detection limit for each VHOC was 0.1 μg/L. Aliquots of 5 cm³ from the water sample, external standard and control blank (vials with VHOCs-free water) were each placed into a 10 cm³ glass vial, sealed, and heated to a specific temperature (1 h at 37 °C). After incubation, 100 μL of the head-space sample was injected directly into the GC using a gas-tight syringe.

Nine types of haloacetic acids (HAAs) were investigated, including monochloroacetic acid (MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), bromoacetic acid (MBAA), dibromoacetic acid (DBAA), tribromoacetic acid (TBAA), bromochloroacetic acid (BCAA), bromodichloroacetic acid (BDCAA), and chlorodibromoacetic acid (CDBAA). The 9 HAAs were analysed using a modified version of EPA Method 552.3, which involves liquid-liquid extraction of the acids with methyl tert-butyl ether (MTBE), followed by derivatisation of the acids into their corresponding methyl esters using acidic methanol, and subsequent analysis of the HAA methyl esters by GC-MS (Kristiana et al. 2011). All of the tests were conducted at least in duplicate. The relative standard deviations (RSD) for different batches were normally < 10%. All of the aqueous solutions were prepared in Milli-Q water.

A standard health risk assessment method, recommended by the USEPA, assessing the potential adverse health effects from exposure to contaminated water, was applied in this paper (Kavcar et al. 2009). Individuals are exposed to organic pollutants in water by the following main pathways: (1) direct ingestion of water through drinking; (2) incidental ingestion of surface water while swimming; and (3) dermal absorption of contaminants from water adhered to exposed skin. The amount of pollutants ingested via the second pathway is minor, and incidental ingestion rates (IR) while swimming have not been found in the available literature, so this pathway has been ignored in this study.

Water was sampled from the water source of a drinking water treatment plant in northern China during the period of study. During the experiment, the raw water was kept at a temperature in the range of 16.2–18.5 °C and the pH was kept in the range of 7.1‒7.3; other water quality characteristics; for example, turbidity, conductivity, DOC, COD_Mn, UV₂₅₄, SUVA (specific UV absorbance), and zeta potential were in the range of 2.52 ± 0.92 NTU, 82 ± 5 μS/cm, 3.52 ± 1.51 mg/L, 2.98 ± 1.09 mg/L, 0.187 ± 0.009 cm⁻¹, 1.152 ± 0.005 L/mg·m, and −15.6 ± 0.5 mV, respectively.

To determine the main types of CEBBPs, the research used time-of-flight mass spectrometry (TOF-MS) to analyse membrane filtered water quality after CEB. According to the result of the TOF-MS, some VHOCs and HAAs were detected in the filtered water. Thus, the research selected 14 types of VHOCs and 9 types of HAAs to detect which types of material were in the membrane CEB effluent.

To accurately identify the types of CEEBPs generated by the reaction of NaClO and organic matter, an experiment was conducted using a CEB and adopting a high RC (100 mg/L), long BD (10 min), short BI (30 min) and high BF (40 L/(m²·h)). The results showed that 9 types of VHOCs (except 1, 1-DCE, trans-DCE, CBD, cis-DCE and HCBD) and 9 types of HAAs were detected in the membrane effluent. The 9 types of VHOCs can be divided into two classes: THMs (TCM, BDCM, DBCM and TBM) and VHOC₅ (DCM, CCl₄, 1,2-DCA, TCE and PCE). The HAAs can be divided into two classes: HAA₅ (MCAA, DCAA, TCAA, MBAA and DBAA) and HAA₄ (TBAA, BCAA, BDCAA, and CDBAA).

The CEEBPs concentration in the membrane effluent (C_e) without CEB, the concentrations of CEEBPs after CEB at 0 and 24 h (C₀ and C₂₄), the average concentrations of CEEBPs (C_ave) after CEB during 24 h and the standard limit concentration (C_sl), referring to USEPA, are shown in the following table.

As seen from Table 3, some species of CEEBPs had been detected in the membrane effluent before CEB, which indicates that there were certain concentrations of pollutants in the raw water. In addition the concentrations of each species increased greatly even though some species (DCM, CCl₄, 1,2-DCA, MCAA and TCAA) were much higher than the standard limit, indicating that the cleaning agent would react with organic matter in raw water to generate CEEBPs. After CEB, the THMs concentration (17.373 μg/L) was much lower than the standard limit concentration (80 μg/L), indicating that the content of the primary THMs precursor (hydrophobic organic matter) was low in raw water, and the DBCM was the main by-product, accounting for 43% in the THMs. The DCM, CBD and cis-DCE concentration increased significantly and accounted for more than 90% of the total VHOCs in the membrane effluent after CEB. Among the nine identified HAAs, the production of TCAA was highest (1.86 mg/L) and the concentrations of several other materials were relatively high.

As can be seen from Figure 2, the total concentration of CEBBPs reached approximately 3,800 μg/L. The CEBBPs in the membrane effluent reduced rapidly at one hour after CEB and then the concentration stabilized. The concentration of all detected CEBBPs gradually reduced with the extension of the water membrane filtration timer, indicating that CEBBPs had accumulated in the membrane reactor after the CEB, leading to the CEBBPs existing in the membrane effluent. Among the 9 types of VHOCs, the content of four types of THMs proportions was very small (less than 5%), but the VHOC₅ proportion accounted for more than 95%. This indicated that the THMs with strong carcinogenic properties were not the main by-product in the VHOCs generated by the reaction of the NaClO cleaning reagent and organic matter at the time of membrane chemical cleaning. At the initial stage of membrane effluent after CEB, the total concentration of HAAs reached approximately 3,339.48 μg/L. With the extension of time, the average concentration of HAAs showed a linear decreasing trend and finally reached 695.94 μg/L. The total amount of five types of risky carcinogenic substances (HAA₅) showed not much difference with the other four types (HAA₄). Compared to VHOCs, especially THMs, the HAAs were more likely to be generated due to the reaction of the cleaning reagent and organic matter, indicating that the content of primary HAAs precursor (hydrophilic organic matter) was high in raw water and the adsorption process did not lower the hydrophilic organic matter concentration.

To confirm the optimum CEB cleaning parameters, a single factor experiment was used to analyse the influence of backwash duration (BD), backwash interval (BI), backwash flux (BF) and reagent concentration (RC) on the production of CEBBPs and membrane fouling.

The CEBBPs concentrations (C_{2 min}, C_{4 min}, C_{6 min}, C_{8 min}, and C_{10 min}) under different BD (2 min, 4 min, 6, 8 and 10 min), the CEBBPs concentrations (C_{30 min}, C_{60 min}, C_{120 min}, C_{240 min} and C_{480 min}) under different BI (30 min, 60 min, 120, 240 and 480 min), the CEBBPs concentrations (C₅, C₁₀, C₂₀, C₃₀ and C₄₀) under different BF (5 L/(m²·h), 10 L/(m²·h), 20 L/(m²·h), 30 L/(m²·h) and 40 L/(m²·h)), the CEBBPs concentrations (C_{10 mg/L}, C_{25 mg/L}, C_{50 mg/L}, C_{100 mg/L} and C_{200 mg/L}) under different RC (10 mg/L, 25 mg/L, 50 mg/L, 100 mg/L and 200 mg/L) are summarized in Table 4.

The concentration of all types of CEBBPs increased with the increment of BD time. When the BD time reached a certain value, the by-product formation rate significantly decreased and the concentration essentially remained unchanged. This might be because the high concentration of the backwash reagent had almost consumed all the organic matter that could react with NaClO. The higher the amount of backwash reagent entering in the membrane reactor because of the increment of BD, the more CEBBPs were generated in the membrane effluent. In all types of CEBBPs, the HAAs proportion was much higher than VHOCs, reaching more than 93%. This was consistent with the recognition rule of typical CEBBPs. When the BD time was 8 min, the total CEBBPs were approximately 160 μg/L, which was triple the concentration than when the BD time was 2 min. However, under these parameter conditions, the generation amounts of all the by-products were below the standard limit value.

As seen in Table 4, under the same condition of BD time, the amount of CEBBPs generated by CEB, on account of the high-frequency backwash, was much higher than the low-frequency backwash. The amount of CEBBPs when the BI time was 30 min was 14 times higher than when the BI time was 480 min. However, under the same BI time, prolonging the BD time did not make by-product generation rapidly increase. For example, under the condition of a 120 min BI time, when the BD time was increased by 10 min the total CEBBPs was still less than 160 μg/L. This is probably because the frequent backwash not only increased the dose of the backwash reagent, but the CEB backwash reagent was also exposed to the new raw water from the filter in the reactor every time. Specifically, the excessive reagent could react with the organic matter of the new raw water in the filter, and it was difficult to test the complete consumption of the organic matter, in which case the by-product concentration would no longer continue to increase. With a long BI time (e.g., more than 60 min), some kinds of CEBBPs were generated below the detection limit. However, when the BI was shortened, the amount of certain CEBBPs began to increase beyond the standard limit concentration of drinking water. For instance, the DCM concentration and TCAA concentration were higher by approximately 19 and 130 μg/L than the standard limit concentration when the BI time was 60 and 30 min, respectively. This indicated that when the BI was less than or equal to 60 min, the effluent was not in conformity with the drinking water health standards.

The larger the BF, the greater the total amount of CEBBPs generated because the amount of backwash reagent entering into the membrane reactor greatly increased with the larger BF under the same condition of BD, BI and RC. This promoted the reaction of the reagent with the organic matter to generate by-products (Zhang et al. 2016). Among the 21 types of CEBBPs, the content of four types of THMs (proportionally) was very small, less than 1%, and the VHOCs proportion accounted for more than 99%. Meanwhile, under this condition, the amount of all of the generated by-products was below the standard limit concentration, indicating that the PAC in the reactor can adsorb a certain amount of CEBBPs, which reduces the CEBBPs concentration in membrane effluent.

The higher the RC, the greater were all the typical by-products of CEBBPs generated. Utilizing a high concentration of reagents for the backwash greatly increased the total amount of NaClO in the membrane reactor, which promoted its reaction with the organic matter in the water and generated by-product CEBBPs. When the RC was 30 mg/L, the concentration of certain VHOCs was below the detection limit. This was probably because the absorbent in the raw water after adsorption can adsorb the backwash chemical agent, leading to the decrease of the oxidative ability of ClO^- (Xu et al. 2013; Dogan et al. 2015). Under a similarly BF condition, the proportion of four types of THMs among 21 types of CEBBPs was very small, less than 1%, and the HAAs proportion accounted for more than 95% under all the different RC conditions.

Figure 3 shows the total amount of CEBBPs and TMP growth rate under different BD, BI, BF, RC. When the BD time was less than 6 min, the amount of CEBBPs grew slowly. However, when the BD time was between 6 and 10 min, the CEBBPs concentration increased rapidly. This is probably because a small amount of backwash reagent (NaClO) reacted with the organic matter; the CEBBPs concentration increased slowly when the BD time was short. With the increase of BD time, the NaClO concentration gradually increased and the backwash reagent was fully in contact with organic matter in the membrane reactor, which led to a rapid increase in the total amount of CEBBPs. Meanwhile, as seen from the figure, when the device ran for 72 hours without CEB, the K reached 0.132 kPa/h. However, when the BD time increased from 0 to 6 min, the K decreased rapidly, indicating that the CEB can remove the membrane chemical pollutants and effectively reduce membrane fouling. Membrane cleaning was improved with increased BD time. However, when the BD time was greater than 6 min, the K slowed down significantly. After a comprehensive consideration of the effect of BD time on the typical CEBBPs production and membrane cleaning, it was determined that the optimal BD time was 6 min.

When the BI was less than 120 min, the CEBBPs generation rate rapidly decreased with the increase of BI. However, when the BI time was greater than 120 min, the CEBBPs concentration basically remained unchanged, indicating that the frequent use of CEB promoted the formation of CEBBPs. Meanwhile, when the BI time was less than 120 min, the TMP growth increased significantly, but when the BI time was between 120 and 480 min, the growth rate of TMP was slow. When the BI time was greater than 240 min, the growth rate of TMP eventually reached 0.092 kPa/h. After a comprehensive consideration of the effect of BI time on the typical CEBBPs production and membrane cleaning, the optimal BI time was 120 min.

When the BF increased from 0 L/(m²·h) to 30 L/(m²·h), the TMP growth rate was rapidly reduced from 0.132 kPa/h to 0.068 kPa/h. However, when the BF was higher than 30 L/(m²·h), the TMP growth rate decreased slowly, indicating that utilization of a lower BF achieved a better membrane cleaning effect. Meanwhile, when the BF was 40 L/(m²·h), the total CEBBPs concentrations reached approximately 100 μg/L. Because the organic matter in the water was not completely consumed by the quantity of chemical reagents selected in the BF range, the variation curve of the CEBBPs did not show a flat stage and the total amount of typical CEBBPs showed a linear growth trend with the increase of BF. Because all the CEBBPs were below the standard limit, the highest BI possible was chosen for CEB on the premise of the guarantee of a better membrane cleaning effect. After a comprehensive consideration of the effect of BF on the typical CEBBPs production and membrane cleaning, the optimal BF was 30 L/(m²·h).

With the increase of backwash reagent concentration, the overall generation of typical by-products CEBBPs showed an increasing trend. When the backwash reagent concentration was less than 200 mg/L, the total amount of typical CEBBPs showed a linear growth trend with the increase of RC. This is probably because there was redundant organic matter reacting with the backwash reagent to produce CEBBPs and the generation of by-products gradually increased. Meanwhile, when the backwash reagent concentration increased from 0 mg/L to 50 mg/L, the growth rate of TMP quickly reduced from 0.132 kPa/h to 0.060 kPa/h. However, when the backwash reagent concentration was higher than 50 mg/L, the growth rate of TMP decreased slowly, indicating that utilization of a lower concentration of backwash reagent could achieve a better cleaning effect. Although the amount of all generated by-products, except DCM and TCAA, was below the standard limit concentration, a lower concentration of chemical should be chosen as the backwash reagent on the premise of guaranteeing a good membrane cleaning effect. After a comprehensive consideration of the effect of RC on the typical CEBBPs production and membrane cleaning, the optimal RC was 50 mg/L.

To determine the health risk levels posed by CEBBPs, a HRA (Health Risk Assessment) was applied to assess the potential adverse health effects from exposure to the effluent (Viana et al. 2009). In this study, the concentrations of CEBBPs were measured in the membrane effluent after CEB operating under the optimum CEB cleaning parameters and shown in Table 5. The non-carcinogenic HQ (Hazard Quotient) and LCR (Lifetime Carcinogenic Risk) were used for the primary human HRA and are also shown in Table 5.

Under the optimum CEB cleaning parameters, the CEBBPs concentration measured in the effluent after CEB was 172.94 μg/L. The proportion of HAAs in the CEBBPs was approximately 93%, indicating that HAAs should be the main focus in the effluent after CEB. The HRA was only undertaken for pollutants where appropriate toxicity values were available because of the limit of toxicity values for some pollutants. Because some parameters were not available for some kinds of CEBBPs, the HQs for these species are not provided in Table 5. The HQ values for each species of CEBBPs were all less than one; meanwhile, the HI for THMs, VHOC₁₂, HAAs and CEBBPs were 4.50E-05, 4.45E-04, 1.20E-03 and 1.65E-03, respectively. These results show that the concentrations of CEBBPs measured in the effluent under the optimum CEB cleaning parameters in this study pose no health risk to local consumers through ingestion or dermal adsorption. The measured HI may be lower than actual risk because the CEBBPs, which were evaluated in this study, were only a portion of the total by-products in the effluent.

The SFs for the many species of CEBBPs were limited; therefore, the LCRs for only 9 types of CEBBPs were calculated through Equation (8). Under most regulatory regimes, an LCR value over 1.00E-05 indicates a potential carcinogenic risk (Wang et al. 2014). In this study, the LCR values for all kinds of CEBBPs are all below 1.00E-05, indicating that these individual species of CEBBPs in the effluent may not pose a carcinogenic health risk to local consumers based on results shown in Table 5. Meanwhile, many species of CEBBPs did not have appropriate parameter values; therefore, the true RI value of CEBBPs may be over 1.00E-05, indicating that the CEBBPs in the effluent may pose a potential carcinogenic risk to local consumers even though the concentration of the species of CEBBPs is much lower than the standard limit concentration. The LCR value for HAAs (6.68E-06) is very close to the LCR value of CEBBPs (6.78E-06), indicating that the HAAs are the main substances in the effluent after CEB and pose a potential carcinogenic risk to local consumers.

The main conclusions of this study are as follows:

• 1.

  In the process of CEB, the NaClO backwash reagent could react with organic matter to produce CEBBPs, including 9 species of VHOCs and 9 species of HAAs. Compared to VHOCs, especially THMs, the HAAs were more likely to be generated due to the reaction of the cleaning reagent and organic matter, indicating that the content of primary HAAs precursor (hydrophilic organic matter) was high in raw water and the adsorption process did not lower the hydrophilic organic matter concentration.

• 2.

  The CEB removal of membrane foulants can effectively mitigate membrane fouling. Within a certain range, with the increase of BD, the decrease of BI, and the rise of BF and RC, the TMP grew slowly after the membrane operation for a period of time and the K was gradually reduced. After a comprehensive consideration of the influence of a single factor on the CEBBPs formation and membrane cleaning effect, the optimal CEB parameters were 6 min of BD, 120 min of BI, 30 L/(m²·h) of BF and 50 mg/L of RC.

• 3.

  Under the optimum CEB cleaning parameters, the effluent does not pose a non-carcinogenic risk and potential carcinogenic risk to residents. The LCR value for HAAs (6.68E-06) is very close to the LCR value of CEBBPs (6.78E-06), indicating that the HAAs are the main substances in the effluent after CEB. A further purification process should be implemented to ensure the safety of drinking water after CEB, even though the CEB can reduce the membrane fouling effectively and the concentration of species of CEBBPs is much lower than the standard limit concentration.

We are very grateful to the financial support of the National Natural Science Foundation of China (NSFC) under the Grant number No. 51308373.

All relevant data are included in the paper or its Supplementary Information.