Enhanced Permeability, Selectivity, and Antifouling Ability of CNTs

Jan 16, 2015 - Membrane filtration provides effective solutions for removing contaminants, but achieving high permeability, good selectivity, and anti...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/est

Enhanced Permeability, Selectivity, and Antifouling Ability of CNTs/ Al2O3 Membrane under Electrochemical Assistance Xinfei Fan, Huimin Zhao, Yanming Liu, Xie Quan,* Hongtao Yu, and Shuo Chen Key Laboratory of Industrial Ecology and Environmental Engineering (Ministry of Education, China), School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China S Supporting Information *

ABSTRACT: Membrane filtration provides effective solutions for removing contaminants, but achieving high permeability, good selectivity, and antifouling ability remains a great challenge for existing membrane filtration technologies. In this work, membrane filtration coupled with electrochemistry has been developed to enhance the filtration performance of a CNTs/Al2O 3 membrane. The as-prepared CNTs/Al 2O3 membrane, obtained by coating interconnected CNTs on an Al2O3 substrate, presented good pore-size tunability, mechanical stability, and electroconductivity. For the removal of a target (silica spheres as a probe) with a size comparable to the membrane pore size, the removal efficiency and flux at +1.5 V were 1.1 and 1.5 times higher, respectively, than those without electrochemical assistance. Moreover, the membrane also exhibited a greatly enhanced removal efficiency for contaminants smaller than the membrane pores, providing enhancements of 4 orders of magnitude and a factor of 5.7 for latex particles and phenol, respectively. These results indicated that both the permeability and the selectivity of CNTs/Al2O3 membranes can be significantly improved by electrochemical assistance, which was further confirmed by the removal of natural organic matter (NOM). The permeate flux and NOM removal efficiency at +1.5 V were about 1.6 and 3.0 times higher, respectively, than those without electrochemical assistance. In addition, the lost flux of the fouled membrane was almost completely recovered by an electrochemically assisted backwashing process.

■

INTRODUCTION Dwindling water resources and increasingly stringent environmental regulations have stimulated the development of advanced water treatment technologies that can provide a safe and clean water supply in more energy-efficient and environmentally friendly ways.1−3 Membrane filtration is favored over other water treatment technologies because it requires no additional chemicals, relatively low energy consumption and straightforward process handling. 4−6 Although membrane filtration has drawn much attention in various fields, it is difficult for conventional membrane filtration to reject targets of various sizes simultaneously, such as contaminants with sizes smaller than membrane pores. Unfortunately, some of the targets unfavorable for size exclusion separation pose serious threats to ecological systems and human health even at trace levels because of their persistence, toxicity, and bioaccumulation.7 Reducing membrane pore size is a possible approach for achieving high selectivity, but it suffers from a potential flux decline and high operating pressure. Additionally, membrane fouling is a ubiquitous and costly problem for conventional membrane filtration that usually causes a deterioration in membrane performance, especially decreased permeate flux.8−11 Therefore, it is important to develop novel membrane technologies that not only are able to supply high flux and good selectivity but are also affordable and convenient for practical usage. © XXXX American Chemical Society

Carbon nanotubes (CNTs) have been demonstrated to be a promising material for constructing membranes because of their high surface area, high mechanical strength, excellent chemical inertness, and outstanding water-transport properties.12−17 Owing to the overlap and interpenetration of CNTs, as-constructed CNT membranes with meshlike structures (called bucky-paper membranes) usually present plentiful interconnected pores and high porosities (>70%).18−20 These properties further lead to a high permeate flux for CNT membranes even at a low operating pressure. Moreover, CNT membranes were found to be effective in retaining nanoparticles and chemical compounds of different sizes through physical and chemical adsorption.21−23 Although CNT membranes provide unique performances in contaminant removal from aqueous solution, strategies for further improving the performance of CNT membranes still need to be developed. Currently, CNTs exhibit encouraging electrochemical performances in various applications. For example, both the adsorption rate and the adsorption capacity of CNTs can be improved by electrochemical assistance.23,24 The adsorbed Received: August 12, 2014 Revised: January 13, 2015 Accepted: January 16, 2015

A

DOI: 10.1021/es5039479 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

(details in the Supporting Information).30−32 The electroconductivity of the CNTs/Al2O3 membrane was measured with a four-point probe meter (RTS-8; 4 Probes Tech, Guangzhou, China). The interfacial adhesion force was quantified by a scratch test (CAS-01; Rhesca, Tokyo, Japan). The elastic strength and hardness were measured by nanoindentation (XP 100BA-1C nanoindenter; MTS Systems Corporation, Eden Prairie, MN). Cross-Flow Filtration Setup and Filtration Experiments. The cross-flow filtration setup is shown in Figure S1 (Supporting Information). The performance of the CNTs/ Al2O3 membrane was evaluated under different potentials using a CHI-660A electrochemical system (CH Instruments, Shanghai, China) with the CNTs/Al2O3 membrane as the working electrode, a Ti mesh as the counter electrode, and a Ag/AgCl (saturated KCl) electrode as the reference electrode. The distance between the working electrode and the counter electrode was 1 cm. During the filtration processes, the feedwater was pumped into the membrane reactor at a constant cross-flow rate (0.56 m/s) with a peristaltic pump. A plate membrane was used in all experiments, and the membrane area was 7.07 cm2. A constant transmembrane pressure of 0.4 bar was maintained with a vacuum pump. The permeate water was collected and monitored on an analytical balance (BP221S; Sartorius, Göttingen, Germany), which was used to calculate the permeate flux. Filtration of Silica Spheres, Latex Particles, and Phenol under Electrochemical Assistance. To understand the impact of electrochemistry on size exclusion, silica spheres that were comparable in size to the membrane pores were used as the probe during the separation process. Both latex particles and phenol molecules were chosen to investigate the impact of electrochemistry on membrane performance for small target removal. The properties of the silica spheres and latex particles are presented in Table S1 (Supporting Information). The initial amounts of silica spheres and latex particles were 100 mg/L and 4.25 × 1015 L−1 (500 mg/L), respectively. To evaluate the amount of silica spheres or latex particles intercepted on the membrane surface, the membrane was rinsed with 100 mL of ultrapure water. The rinsed silica spheres or latex particles were collected and analyzed. The amount of silica spheres were measured with a turbidity meter (WGZ-1A; Xinrui, Shanghai, China). The amounts of latex particles at different dilution ratios were counted from scanning electron microscopy (SEM) images using ImageJ software (National Institutes of Health, Washington, DC). The initial concentration of phenol in the feedwater was 5 mg/L. Its concentration in the permeated water was determined by high-performance liquid chromatography (HPLC; Waters, Milford, MA). The removal efficiency and flux loss for the filtration of silica spheres, latex particles, and phenol was evaluated after 30 min of operation. The concentration of electrolyte (Na2SO4) in the feed solution was 10 mM. Removal of Natural Organic Matter (NOM) and Membrane Regeneration. An aqueous solution of humic acid (Sigma-Aldrich, St. Louis, MO) with an initial concentration of 10 mg/L was introduced into the membrane module. The effluent sample was collected and analyzed for total organic carbon (TOC) content using a TOC analyzer (TOCVCPH, Shimadzu, Kyoto, Japan). In the membrane regeneration stage, the membrane, after being operated without electrochemical assistance or at 1.5 V for 60 min, was regenerated by backwashing in the absence or

contaminants are able to be decomposed by electrochemical reactions.23,25,26 Furthermore, the regeneration of polluted CNTs can be achieved through electrochemical approaches.27 If these fascinating electrochemical functions can be introduced into CNT membranes, the additional electrochemical functions would promote the physicochemical functions that are not available by size exclusion alone. This might be an alternative approach to solve the aforementioned problems presented in conventional membrane separations. On the other hand, it is difficult for free-standing buckypaper membranes without substrates or reliable stability to be used in practical applications. Although mixed-matrix CNTs/ Al2O3 materials have good mechanical strength,28 the filtration performance of CNT membranes might be sacrificed because the CNTs are randomly blended into the Al2O3 matrix rather than being constructed into a network structure. Membranes with an unsymmetrical layer structure, constructed by coating CNTs on a porous substrate, provide an alternative approach for practical applications, although such membranes would require a sufficient binder to enhance the adhesion force between the CNT layer and the porous substrate. It has been reported that graphene-like carbon from polymer pyrolysis can significantly enhance the mechanical strength of CNT sponges. Thus, it is supposed that the adhesion force between the CNT layer and the porous substrate in an unsymmetrical membrane could possibly be improved by graphene-like carbon. To verify the feasibility of this hypothesis, an electroconductive CNT membrane was prepared by coating CNTs onto a porous Al2O3 substrate (providing mechanical strength for practical applications) by a vacuum filtration−pyrolysis method. The morphology, pore-size tunability, and mechanical stability of the prepared membrane were systematically examined. Filtration performance was tested under electrochemical assistance with several simulated pollutants including silica spheres, latex particles, phenol, and natural organic matter (NOM). Based on these experimental results, various removal mechanisms were investigated such as sieving capability, electrostatic repulsion, electrochemical-enhanced diffusion− adsorption, and electrochemical oxidation. An efficient flux recovery was demonstrated by backwashing under electrochemical assistance.

■

MATERIALS AND METHODS Preparation and Characterization of CNTs/Al2O3 Membrane. In a typical synthesis, a vacuum filtration− pyrolysis process was used to construct a randomly oriented CNT layer on a porous Al2O3 substrate. First, oxidized CNTs [Shenzhen Nanotech Port Co. Ltd.; length = 5−15 μm, purity >97%, diameter = 60−100 nm, Brunauer−Emmett−Teller (BET) surface area = 40−70 m2/g] were dispersed in N,Ndimethylformamide with 0.5 wt % polyacrylonitrile (PAN; Sigma-Aldrich, St. Louis, MO). Then, the CNTs@PAN was vacuum-filtered onto a porous Al2O3 substrate. The resulting CNTs@PAN/Al2O3 membrane was heated at 250 °C for 3 h in air and then pyrolyzed at 1000 °C in a hydrogen atmosphere. After pyrolysis, the membrane was cooled to 400 °C in a hydrogen atmosphere and then to room temperature in air. The membrane morphology was characterized on a Hitachi S-4800 scanning electron microscope (SEM). The pore size distribution of the as-prepared CNTs/Al2O3 membrane was analyzed by liquid−liquid displacement porosimetry (details in the Supporting Information).29 The porosity of the CNTs/ Al2O3 membrane was obtained from microscopic image analysis B

DOI: 10.1021/es5039479 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

the mechanical strength of the CNTs/Al2O3 membrane, nanoindentation and scratch tests were performed. As reported in Table S2 (Supporting Information), the elastic strength and hardness of the CNTs/Al2O3 membrane with Cgr were 1.577 ± 0.08 GPa and 86 ± 29 MPa, respectively, which were approximate 7.9 and 8.1 times higher than those of a membrane without Cgr (elastic strength, 0.2 ± 0.03 GPa; hardness, 8 ± 2 MPa). These results suggest that the mechanical stability and stiffness of the CNTs/Al2O3 membrane were significantly improved by Cgr. The interfacial adhesion strength between the CNT layer and the porous substrate was evaluated in terms of “critical interfacial stress” in a scratch test. The critical load value was about 407 gf for the membrane with Cgr, which was higher than the 275 gf vale for the membrane without Cgr, suggesting that the bonding force between the CNT layer and the porous substrate was also markedly improved by Cgr. The strong adhesion between the CNT layer and the Al2O3 substrate might be due to the formation of C−O−Al bonding (Figure S4, Supporting Information). The improved mechanical strength was further confirmed by ultrasonic shock (40 kHz, 200 W, 30 min; Figure S5, Supporting Information) and scouring (0.84 m/s, 600 min; Figure S6, Supporting Information) experiments. The pore size of the CNTs/Al2O3 membrane was 142 nm (Figure S7, Supporting Information), suggesting that the CNTs/Al2O3 membrane was a microfiltration membrane. The pure water flux was 860 L/m2·h·bar (Figure S8, Supporting Information), and the porosity was 81% (Figure S9 and Table S3, Supporting Information), which were 1.45 and 2.2 times larger, respectively, than the values for a ceramic membrane with the same substrate and comparable pore size and thickness (CM-150; Table S4 and Figure S10, Supporting Information). The electronic conductivity of the CNTs/Al2O3 membrane was 1615 S/m, which ensures this membrane can serve as a platform for combining membrane separation with electrochemistry. Furthermore, the pore size and porosity of the CNTs/Al2O3 membrane can be controlled by adjusting the CNT mass per area (Figure S7 and Table S3, Supporting Information). Removal of Particles Comparable in Size to the Membrane Pores under Electrochemical Assistance. To select an appropriate potential range for CNTs/Al 2 O 3 membrane operation under electrochemical assistance, the electrochemical stability of the CNTs/Al2O3 membrane was tested. The CNTs/Al2O3 membrane was stable (without composition or structural damage) at +1.5 V, whereas electrochemical etching of the CNTs occurred at +2.0 V (Figures S11 and S12, Supporting Information) This result is

presence of electrochemical assistance for 10 min. For membrane regeneration, a water solution containing 10 mM Na2SO4 was poured into the membrane module in end-flow mode. The flux direction was the opposite of that used during filtration. Meanwhile, a negative bias was applied to the membrane. Both the recovered permeate flux and the TOC removal efficiency were investigated to evaluate the membrane recoverability under electrochemical assistance.

■

RESULTS AND DISCUSSION Characterization of CNTs/Al2O3 Membrane. The CNTs/Al2O3 membrane was prepared by coating CNTs on a porous Al2O3 substrate. The morphology of the CNTs/Al2O3 membrane was characterized by SEM. As shown in Figure 1,

Figure 1. (a,c) Low- and (b,d) high-magnification SEM images of a CNTs/Al2O3 membrane: (a,b) top view, (c,d) cross section.

the CNT layer had abundant interconnected pores without cracks. As expected, the CNT layer was adhered on the porous Al2O3 substrate. In addition, various forms of CNTs/Al2O3 membranes including hollow fibers, plates, and tubes could be prepared by this method (Figure S2, Supporting Information). According to previous studies, CNT membranes hold their structures primarily through weak van der Waals interaction, which can be easily destroyed in water flow.33−35 On the other hand, it has been reported that graphene-like carbon (Cgr) from polymer pyrolysis can work as a binder to strengthen the structure of CNT aerogels.36 Therefore, in this work, PAN was employed for CNTs/Al2O3 membrane preparation, because it can be converted into Cgr by pyrolysis. In Figure S3 (Supporting Information), Cgr can be clearly observed at the nodes between the nanotubes. To verify the influence of Cgr on

Figure 2. (a) Silica-sphere removal efficiency of CNTs/Al2O3 membrane. (b) Normalized flux of CNTs/Al2O3 membrane for silica sphere removal. (c) Ratio of silica spheres retained on the membrane surface (operation time of 30 min). C

DOI: 10.1021/es5039479 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 3. (a) Removal efficiency in the absence and presence of electrochemical assistance (gray column indicates the amount of latex particles intercepted on the membrane surface). (b) Dependence of latex particle removal on flux−2/3 without bias and at a bias of +1.5 V (operation time of 30 min).

positively charged with a zeta potential of +7.6 mV in the feedwater. This suggests that an electrostatic repulsion interaction would occur between the silica spheres and the membrane in the presence of a positive bias. When feedwater flowed through the membrane in the presence of a positive bias, some parts of the positively charged silica spheres could be rejected to attach to the membrane surface and then removed by cross-flow. All of these processes led to a significantly improved rejection efficiency and permeate flux of the CNTs/ Al2O3 membrane for removing contaminants with the same polarity and a size comparable to that of the membrane pores in the presence of electrochemical assistance. This theory was further confirmed by filtering silica spheres with a negatively charged zeta potential (pH 9), where the removal efficiency and flux declined dramatically (Figure S13, Supporting Information). Moreover, the improved hydrophility also contributed to the enhanced filtration performance of the CNTs/Al2O3 membrane under electrochemical assistance (Figure S14, Supporting Information). Removal of Particles Smaller than the Membrane Pores under Electrochemical Assistance. It is generally considered that the selectivity of a membrane depends mainly on its pore size. However, it is difficult to remove pollutants with sizes smaller than the membrane pores by size exclusion using conventional ultrafiltration (UF) and microfiltration (MF) membranes. Therefore, one of the major challenges for membrane filtration is to improve the selectivity of the membrane without changing the membrane pore size. In this work, latex particles with a diameter of 57.7 nm (Table S1, Supporting Information) were chosen to investigate the performance of the CNTs/Al2O3 membrane for removing targets smaller than the membrane pores (142 nm). As displayed in Figure 3a, the latex particle removal was 2.24 log (14.9% of the initial logarithm value) for the CNTs/Al2O3 membrane without electrochemical assistance. The low logarithmic removal was consistent with the fact that size exclusion is insufficient for removing contaminants that are smaller than the membrane pores. However, it was noted that the particle removal increased to 4.49 log at +0.5 V, two times higher than that without electrochemical assistance. Moreover, it further increased to 5.59 log at +1.0 V and 6.59 log at +1.5 V. These results suggest that electrochemical assistance improved the membrane performance in removing contaminants that cannot be removed by size exclusion alone. According to previous work, diffusion−adsorption is operative in transporting particles smaller than the membrane pores during the filtration processes.38,39 To understand the removal mechanism

consistent with previous reports that electrochemical etching occurred on CNTs at +1.7 V versus Ag/AgCl electrode.37 Therefore, the performance of the CNTs/Al2O3 membrane under electrochemical assistance was tested at a bias between 0 and +1.5 V versus Ag/AgCl. Silica spheres with a diameter of 140 nm, which was comparable to CNTs/Al2O3 membrane pore size (142 nm), were selected to investigate the performance of the CNTs/ Al2O3 membrane with/without electrochemical assistance. As shown in Figure 2a, 87.2% of the silica spheres were removed by the CNTs/Al2O3 membrane in the absence of electrochemical assistance, whereas the removal efficiency increased to 90.7% at +0.5 V. As expected, the removal efficiency further increased to 94.1% at +1.0 V and 98.9% at +1.5 V. These results demonstrate that the rejection efficiency of silica spheres by the CNTs/Al2O3 membrane could be improved by electrochemical assistance. To further understand the impact of electrochemical assistance on membrane performance, the permeate flux was simultaneously measured during silica sphere separation. The normalized flux of the CNTs/Al2O3 membrane is presented in Figure 2b. A significant flux loss of 35.2% occurred on the CNTs/Al2O3 membrane without electrochemical assistance, whereas the flux loss decreased to 28.8% at +0.5 V. Moreover, the flux loss further decreased to 17.4% at +1.0 V and 2.7% at +1.5 V. These results indicate that the flux loss of the CNTs/ Al2O3 membrane can be significantly inhibited by electrochemical assistance. Usually, the contaminants retained by size exclusion can aggregate on the membrane surface and then lead to flux loss. However, in these experiments, both the removal efficiency and the normalized flux increased under electrochemical assistance. To determine the factors contributing to the enhanced performance, the amount of retained spheres on the membrane surface was analyzed. In the absence of electrochemical assistance, it was found that 70.6% of the removed silica spheres were retained on the membrane surface. This result suggests that the silica spheres were mainly intercepted on the membrane surface by size exclusion in the absence of electrochemical assistance. However, at +0.5 V, the portion of retained spheres on the membrane surface decreased to 53.1%. Interestingly, it further decreased to 21.2% and 5.7% at +1.0 and +1.5 V, respectively. These results demonstrate that size exclusion was not the main factor contributing to silica sphere removal under electrochemical assistance. To understand this phenomenon, the zeta potential of the silica spheres was measured. The results revealed that the silica spheres were D

DOI: 10.1021/es5039479 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 4. (a) Phenol removal efficiency of CNTs/Al2O3 membrane with/without electrochemical assistance (operation time of 30 min). (b) Cyclic voltammogram of CNTs/Al2O3 membrane in the absence and presence of phenol.

increased to nearly 100% at +1.0 and +1.5 V. This abrupt improvement in the removal efficiency indicates that a different removal mechanism might play a role in the removal of phenol at potentials more positive than +1.0 V. To further clarify the removal mechanism under electrochemical assistance, cyclic voltammetry and TOC measurements were conducted to investigate the electrochemical behavior of the CNTs/Al2O3 membrane toward phenol. As presented in Figure 4b, an obvious oxidation peak (+0.83 V) with an onset potential of +0.6 V was observed in the solution with phenol, whereas no oxidation peak appeared without phenol. This result demonstrates that phenol can be electrochemically oxidized at potentials more positive than +0.6 V on the membrane. Meanwhile, the TOC removal efficiency was 39.7% at +0.5 V (Figure S16a, Supporting Information), which is consistent with the phenol removal efficiency under the same conditions. These results illustrate that the phenol removal was dominated by an adsorption process under biases more negative than +0.5 V. According to previous work, phenol adsorption on carbon electrodes can be enhanced by electrochemical assistance.40 Compared with the phenol removal efficiency without electrochemical assistance, the slightly increased phenol removal efficiency at +0.5 V can be explained by an electrochemically enhanced adsorption process. However, the TOC removal efficiency was improved to 71.4% at +1.0 V, and the corresponding phenol removal efficiency was increased to 96.7%. To understand the mismatch between the TOC and phenol removals, the UV−vis absorbance of the permeate water was measured. As shown in Figure S16b (Supporting Information), an adsorption peak of benzoquinone appeared at 242 nm at +1.0 V. This suggests that part of the removed phenol was converted to benzoquinone at +1.0 V, which led to the result that the TOC removal was lower than the phenol removal. When the bias was +1.5 V, the benzonquinone peak disappeared in the UV−vis spectrum, and the TOC removal efficiency increased to 98.4%. This result illustrates that the phenol was almost completely removed by electrochemical oxidation at +1.5 V. NOM Removal under Electrochemical Assistance. NOMs, which are the major precursors of disinfection byproducts, are ubiquitous in surface water and groundwater.41,42 In addition, NOMs have been identified as being among the major foulants that can cause membrane fouling and flux loss. Therefore, humic acid was used as a source of NOM to evaluate the performance of electrochemically assisted CNTs/Al2O3 membrane filtration for NOM removal. For comparison, the filtration performance of a commercial ceramic membrane (CM-100) was also investigated. As shown in Figure

of latex particles, the logarithmic removal of the latex particles as a function of permeate flux was investigated at pressures of 0.4, 0.6, and 0.8 bar (Figure 3b). The good linearity between logarithmic removal efficiency and flux2/3 indicates that the latex particle removal obeys the Levich model for diffusion− adsorption. The CNTs/Al2O3 membrane retained the latex particles through adsorption by the CNT layer during the separation process. However, at +1.5 V, the slope of the fitted line was 7.45 times larger than that without electrochemical assistance. This suggests that the transport of the latex particles was enhanced by electrochemical assistance. Zeta potential analysis revealed that the latex particles were negatively charged (−8.4 mV; Table S1, Supporting Information). Thus, electrostatic attraction would occur between the negatively charged latex particles and the positively charged membrane surface, which could enhance the convective-diffusive transport on the CNTs/Al2O3 membrane. This electrostatic attraction enhanced the convective-diffusive transport on the CNTs/Al 2 O 3 membrane. As a result of the improved mass-transport rate and the attractive electrostatic interaction between the negatively charged latex particles and the positively polarized CNTs/Al2O3 membrane, the adsorption rate and adsorption capability of the CNTs were enhanced, thereby leading to a high removal efficiency of latex particles. These results indicate that contaminants unfavorable for removal by size exclusion can be removed to some extent by the CNTs/Al2O3 membrane with electrochemical assistance if they present a polarity opposite that of the charged membrane. However, the flux was sacrificed during latex particle removal under electrochemical assistance (Figure S15, Supporting Information). Regeneration methods such as electrodegradation or electrodesorption of the adsorbed contaminants are required. Removal of Organic Molecules under Electrochemical Assistance. It is difficult to exclude dissolved organic molecules with microfiltration and ultrafiltration membranes because their pores are much larger than these molecules. In this work, phenol was selected to evaluate the performance of electrochemically assisted membrane filtration using the CNTs/Al 2 O 3 membrane for the removal of chemical contaminants that exist as molecules rather than adsorbed on the particles in neutral water. In the absence of electrochemical assistance, the membrane presented a low phenol removal efficiency of 14.7% (Figure 4a). Because of the much smaller size of the phenol molecule relative to the membrane pore, phenol could not be removed by size exclusion on the CNTs/ Al2O3 membrane. Considering the adsorption ability of CNTs, phenol removal might result from CNT adsorption. At +0.5 V, the removal efficiency increased to 38.2%, and it further E

DOI: 10.1021/es5039479 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 5. (a) TOC removal efficiency and (b) normalized flux for humic acid removal on CNTs/Al2O3 membrane (operation time of 60 min).

chemical behavior of the CNTs/Al2O3 membrane toward humic acid. As shown in Figure S17 (Supporting Information), an oxidation peak (+0.6 V) with an onset potential of +0.5 V was observed for the solution with humic acid, whereas no oxidation peak appeared without humic acid. This result indicates that humic acid could be electrochemically oxidized on the CNTs/Al2O3 membrane at biases more positive than +0.5 V. When the bias was +1.5 V, the accumulation of humic acid on the membrane surface was greatly mitigated, as revealed by the SEM image in Figure 6b. These results indicate that electrochemical oxidation was the major process contributing to the humic acid removal, resulting in a much lower flux loss of 7.4% at +1.5 V. Energy consumption is one of the major concerns for membrane separation technology in practical applications. According to Schäfer et al., the energy cost for microfiltration is approximately 30 Wh/m3 for the removal of NOM (5−12.5 mg/L) from surface water.44 For the case of electrochemically assisted filtration, on one hand, the permeate flux of the CNTs/ Al2O3 membrane was increased by 13.3−35.7% relative to that without electrochemical assistance at biases of +0.5−1.5 V. On the other hand, the electrochemically assisted process could cause additional energy consumption of around