Improved antifouling and antimicrobial efficiency of ultrafiltration membranes with functional carbon nanotubes

Abstract

Aiming to increase the antifouling/antibacterial property and purifying efficiency of a polyethersulfone (PES) membrane, in this study, functional polymer brush grafted carbon nanotubes (p-CNTs) were developed as multifunctional modifiers for membrane modification. Firstly, functional molecules, methyltriethylammonium chloride (MTAC) and poly(ethylene glycol) methyl ether methacrylate (EGMA), were grafted onto the CNTs via surface initiated atom transfer polymerization (SI-ATP); then p-CNT/PES composite membranes were prepared via a phase inversion technique. The modified membranes exhibited improved surface wettability owing to the introduced EGMA brushes. As revealed by protein adsorption and protein solution ultrafiltration experiments, the p-CNTs could improve the membrane antifouling properties significantly, and the bacterial culture results suggested that bacterial adhesion and survival were suppressed by the p-CNTs. Moreover, the p-CNTs enabled the membrane adsorptive activity towards phenolic molecules, which would increase the purifying efficiency of the membranes during water treatment. In general, the fabricated p-CNT/PES composite membranes integrated favorable antifouling ability, high antibacterial activity, and efficient toxin removal ability, which might satisfy diverse separation and purification needs.

---

1. Introduction

Nowadays, the shortage of clean water has become one of the most pervasive problems, and numerous people are suffering from unclean water resources that have been contaminated by bacteria, fungi, pathogens, endocrine disruptors and so on.^1,2 Many technologies have been developed to remove pollutants in waste water.^3–5 As an effective approach, membrane based separation and purification technologies, including microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO), have been widely used to provide sustainable water resources.^6–9 However, there are still some shortcomings that need to be overcome, such as membrane fouling by contaminants.¹⁰ For long-term usage, the hydrophobic nature of membranes makes them non-resistant towards the attachment of contaminants, especially micro-organisms. Non-specific fouling by organisms will eventually lead to the formation of biofilms, which will cover the membrane surface and decline the purifying efficiency.^11,12

To control membrane fouling, extensive efforts have been devoted to designing antibacterial surfaces that can reduce bacterial adhesion and proliferation on the membrane surface and prevent subsequent biofilm formation.^13–15 Antimicrobial agents,^16–18 metal nanoparticles^19–21 and quaternary ammonium compounds (QACs)^22–24 have been explored to achieve this goal. However, some of the antimicrobial agents are too expensive for large-scale application and the massive usage of antimicrobial agents might induce a resistance for pathogens. Metal nano particles, e.g. silver nanoparticles, may not be suitable for long-term usage due to etching by oxygen.²⁵ To this point, QACs, which can interfere with cellular membrane protein function to inhibit the growth of bacteria, are advantageous due to their low cost and long-term bacterial killing activity. Nowadays, QACs have been widely applied to enable interfaces with the ability to kill bacteria and many other micro-organisms.²⁶ Even though bacterial survival and proliferation have been inhibited, the QAC modified surfaces will also be contaminated by the accumulation of dead bacteria, and the QAC surfaces show intensive non-specific protein adsorption, which will also cause severe membrane fouling. Therefore, introducing functional groups to help repel the adhesion of bacteria has been explored to further improve their antibacterial activity; meanwhile, the anchored functional groups could also suppress membrane fouling by proteins.¹⁵ For instance, Yang et al. coated polycarbonate cations and polyethylene glycol (PEG) brushes onto silicone rubber and no bacterial adhesion was noticed even after 14 days.²⁷ Deng’s research also demonstrated that the combination of QACs and PEG units could result in surfaces with robust antibacterial efficiency.²⁸

For membrane modification, bulk blending is a conventional approach, which can be easily processed by adding particular modifiers into membrane casting solutions.^29–31 By using this method, researchers have obtained various functional membranes to fulfil diverse applications. For example, a series of mixed matrix membranes was successfully prepared by incorporating activated carbon into membrane substrates.^32,33 Combining the advantages of adsorption and diffusion, a significantly improved toxin clearance rate was achieved for the membranes, which endowed the membranes with great potential to satisfy purification demands. Compared with activated carbon, carbon nanotubes (CNTs) are a superior absorbent due to their large surface area. However, the fully covered inherent carbons on the CNT surface make them barely soluble in any solvent; hence, the surface treatment of CNTs is always necessary before incorporation into the membrane matrix. The treated CNTs, which contained a large amount of functional groups, showed a remarkable positive influence on membrane performance including increased water permeability and antifouling properties.^34,35

Herein, as a multifunctional modifier for membranes, polymer brush grafted CNTs (p-CNTs) containing QACs and PEG units were synthesized and then applied for the modification of a polyethersulfone (PES) membrane. The p-CNTs were prepared via surface initiated atom transfer polymerization (SI-ATRP), then a p-CNT/PES composite membrane was prepared via a phase inversion technique. Scanning electron microscopy (SEM) images revealed that the p-CNTs could create a porous structure for the membrane, and the water contact angle (WCA) results indicated that the membrane wettability was improved by the p-CNTs. Systemic investigations, including protein solution and sewage mimic solution flux recovery ratio, protein adsorption, bacterial adhesion and micro-organism coverage, revealed that the antifouling and antibacterial properties were enhanced by the p-CNTs. Furthermore, the p-CNTs enabled the membrane to conjugate small toxins, which would also benefit the membrane purifying efficiency.

2. Materials and methods

2.1. Materials

Commercial polyethersulfone (PES, Ultrason E6020P) was purchased from BASF and used as received. Hydroxyl carbon nanotubes (CNT-OH, single wall, outside diameter < 5 nm, length ∼ 15 μm, purity > 90%,and –OH content ∼ 4 wt%) and carbon nanotubes (outside diameter < 5 nm, single wall, length ∼ 15 μm, and purity > 90%) were supplied by Times Nano. Ltd., Chengdu and washed with 10 wt% hydrochloric acid to remove impurities before use. 2-Bromo-2-methylpropionyl bromide (BiBB, 97%, Alfa Aesar) and N,N-dimethylaminopyridine (99%, Aladdin) were used as received. Copper(I) bromide (CuBr, AR, Shanghai Sinpeuo Fine Chemical Co. Ltd.) was washed with acetic acid and ethanol to remove the oxidates before use. Dimethylsulfoxide (DMSO, AR, >99%), methyltriethylammonium chloride (MTAC, with 600 ppm MeHQ), poly(ethylene glycol) methyl ether methacrylate (EGMA, M_w = 475, with 100 ppm MeHQ and 200 ppm BHT), and N,N,N′,N′′,N′′-pentamethyldiethylenetriamine (PMDETA, 99%) were purchased from Aladdin Ltd. Dimethylacetamide (DMAC, AR, >99%, Kelong), chloroform (AR, Kelong), and triethylamine (AR, Kelong) were distilled under vacuum before use. Bovine serum albumin (BSA, fraction V) was obtained from Sigma Chemical Co. Micro BCA™ protein assay reagent kits were the products of PIERCE. Other regents, if not mentioned specifically, were supplied by Aladdin, China.

2.2. Synthesis of macro-initiator CNT-Br

Bromide carbon nanotube (CNT-Br) was synthesized according to Gao’s report.³⁶ Firstly, CNT-OH (2.0 g) was dispersed in 50 mL of chloroform with the assistance of powerful ultrasonication for 30 minutes; then N,N-dimethylaminopyridine (DMAP, 0.4 g, 3.28 mmol) and triethylamine (4.0 mL, 28.6 mmol) were added. 2-Bromo-2-methylpropionyl bromide (BiBB) (6.0 mL, 13.0 mmol) was added drop-wisely at 0 °C for at least 1 h, and then the reaction was carried out at 25 °C for another 48 h. The reaction was ceased by filtration under vacuum. The product was washed by chloroform to remove redundant agents, and then collected by drying under vacuum.

2.3. Synthesis and characterization of CNT-PMTAC and CNT-PEGMA

The polymer brush functionalized CNT (p-CNT) was synthesized via SI-ATRP. The prepared CNT-Br was then employed as a macro-initiator to induce the polymerization of EGMA (or MTAC) in the presence of PMDETA and CuBr as is shown in Table 1. Typically, CNT-Br (0.12 g) and methyltriethylammonium chloride (MTAC, 2.08 g, 10 mmol) were dispersed in dimethylsulfoxide (DMSO, 10 mL) in a Shrek reaction flask. Then, PMDETA (84 μL) and CuBr (29 mg) were added under nitrogen protection. After three cycles of freeze–pump–thaw, the flask was sealed and immersed in a 70 °C oil bath. After 24 hours, the reaction was ceased by vacuum filtration. The product was washed with DMSO and then dialyzed against double distilled water. The samples were collected by lyophilization and named CNT-PMTAC. The EGMA brushes were grafted onto the CNT surface in the same procedure and the product was named CNT-PEGMA. The grafting yield was calculated by the weight change of CNT after polymerization.

Table 1 Components for the synthesis of CNT-PMTAC and CNT-PEGMA

Sample	CNT-Br (g)	MTAC (g)	EGMA (g)	Grafting yield^a
a The grafting yield was estimated by weight changes after polymerization, which was expressed as mmol chain per g CNT.
CNT-PMTAC	0.121	2.08	—	37.09
CNT-PEGMA	0.121	—	4.75	40.61

---

The grafting of the functional polymers was verified by Fourier transform infrared (FTIR) spectra, which were obtained on a Nicolet-560 spectrophotometer (Nicol, US, range: 4000–500 cm⁻¹, resolution: 2 cm⁻¹). ¹H NMR spectra were obtained via a Bruker spectrometer (600 MHz). Transmission electron microscopy (TEM) was used to detect the structure of the CNT on a Tecnai G2 F20S-TWIN transmission electron microscope (FEI Ltd., USA) operated at 200 kV.

2.4. Preparation and characterization of p-CNT/PES composite membranes

The membranes were prepared by a phase inversion technique as is shown in Table 2. A typical process was as follows: firstly, p-CNT was dispersed in DMAC with the assistance of ultra-sonication and then PES was added. The casting solution was vigorously stirred for at least 2 days. After that, the casting solution was spin coated on a glass plate and the membrane thickness was controlled to 60 μm by regulating the spinning rate. Then, the glass plate was immersed in distilled water immediately to obtain the membrane. The membrane blended with CNT-PMTAC was named M-M and the sample blended with CNT-PMTAC and CNT-PEGMA at a weight ratio of 3 : 1 was named M-E1M3, as is shown in Table 2.

Table 2 Preparation of p-CNT/PES composite membranes

Sample	Additives	PES (g)	DMAC (g)
PES	—	16	84
M-C	Pristine CNT, 2 g	16	82
M-M	CNT-PMTAC, 2 g	16	82
M-E1M3	CNT-PMTAC, 1.5 g	16	82
	CNT-PEGMA, 0.5 g
M-E1M1	CNT-PMTAC, 1 g	16	82
	CNT-PEGMA, 1 g
M-E3M1	CNT-PMTAC, 0.5 g	16	82
	CNT-PEGMA, 1.5 g
M-E	CNT-PEGMA, 2 g	16	82
M-E3M1-5	CNT-PMTAC, 1.25 g	16	79
	CNT-PEGMA, 3.75 g

---

Attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR) of the membrane surface was obtained on a Nicolet-560 spectrophotometer (Nicol, US, range: 4000–500 cm⁻¹, resolution: 2 cm⁻¹). X-ray photoelectron spectroscopy (XPS, XSAM800, Kratos Analytical, UK) was used to detect the surface compositions of the membranes. SEM images were obtained by using a JSM-7500F SEM microscope (JEOL, Japan). The contact angles of double-distilled (DI) water on the membranes were measured by a contact angle goniometer (DSA100, KRUSS, Germany).

2.5. Ultrafiltration experiments

The water flux of the membrane was measured to study membrane permeability as reported.^37–39 BSA was dissolved in phosphate buffered saline (PBS) solution (pH 7.4) at a concentration of 1 mg mL⁻¹. To illustrate the ultrafiltration for sewage, mimetic sewage (MS) was prepared by dissolving 0.3 g of bovine serum albumin (BSA), 0.3 g of humic acid (HA), and 0.4 g of sodium alginate (SA) in 1 L of water.³⁹ An air compressor was applied to supply the pressure on the prepared membrane, which was placed in a dead-end ultrafiltration cell. The membrane was pre-compacted by PBS (pH 7.4) for 20 min at 0.06 MPa to get a steady filtration, and then the flux was measured at 0.05 MPa.

The flux (F) was calculated by using the following equation:

F (g m⁻² h⁻¹ mm Hg) = W/SPt (1)

where W is the weight of the permeated solution (g), S is the effective membrane area (m²), P is the pressure applied on the membrane (mm Hg) and t is the time (h).

The BSA rejection ratio, R, was calculated as follows:

R (%) = (1 − C_p/C_b) × 100 (2)

where C_p and C_b stand for the BSA concentrations in the permeated solution and bulk solution, respectively; which were measured on a UV-vis spectrophotometer (UV-1800, Shimadzu) at the wavelength of 278 nm.

After being filtrated by the BSA solution or MS solution, the membrane was rinsed slightly with double distilled water and then the membrane permeability was measured again as described above, to investigate the membrane antifouling property using the flux recovery ratio, which was calculated as follows:

F_RR (%) = (F₂/F₁) × 100 (3)

where F₁ and F₂ (g m⁻² h⁻¹ mm Hg) are the fluxes before and after BSA/MS solution ultrafiltration, respectively.

The observed sieving coefficient (SC_o) of polyethylene glycol (PEG-20 000, 0.1 g L⁻¹) solution of the membranes was measured. The PEG solution was applied to the membranes using the apparatus as described above. To determine the PEG concentration, the solution was firstly reacted with 5.0 wt% BaCl₂ aqueous solution and 0.05 mol L⁻¹ I₂ alcoholic solution, and then the concentration was determined by a UV-vis spectrophotometer (UV-1750, Shimadzu, Japan) at 510 nm. The SC_o was calculated as follows:

SC_o (%) = (1 − C_p-PEG/C_b-PEG) × 100 (4)

where C_p-PEG and C_b-PEG are the PEG concentrations in the permeated solution and bulk solution, respectively.

2.6. Protein adsorption

Protein adsorption of the membranes was measured by using BSA as a model protein. A typical procedure was as follows: the membrane was incubated in BSA solution (1 mg mL⁻¹) at 37 °C for 1 h. After being rinsed with PBS and double distilled water gently, the membrane was immersed in a washing solution (2 wt% sodium dodecyl sulfate) at 37 °C and shaken for 2 h to remove the adsorbed protein. Then, the protein concentration in the washing solution was measured by using the Micro BCA™ Protein Assay Reagent Kit (PIERCE). According to the protein concentration, the adsorbed protein amount was calculated. The experiments were repeated 4 times to get an average value.

2.7. Bacterial adhesion and survival

To investigate the antibacterial efficiency of the membranes, bacterial adhesion on the surfaces was investigated by using Escherichia coli (E. coli, ATCC 6538, Gram negative) and Staphylococcus aureus (S. aureus, ATCC 25922, Gram positive) as model bacteria. The membranes were incubated in a 10⁶ colony forming unit (CFU) bacteria suspension for 12 h and the live/dead bacterial cells were stained using LIVE/DEAD® BacLight Bacterial Viability Kits as instructed. The images were obtained by fluorescence microscopy (DMIRE2, Leica). The density of the adhered bacteria was estimated by at least 6 fluorescence images for quantitative analysis.

2.8. Microbial coverage

The membrane antifouling property was measured by microbial coverage. The membranes were filtrated by polluted water (collected from the Funan river, Chengdu, China) containing various microbes as described in the ultrafiltration experiment. Then, the membranes were slightly rinsed with normal saline (NS) and treated with 2.5 wt% glutaraldehyde at 4 °C for 24 h. The samples were dehydrated by passing them through a series of graded alcohol-NS solutions (25%, 50%, 75%, and 100%) and then the microbial coverage was observed via SEM (JSM-7500F, JEOL, Japan).

2.9. Small toxins adsorption

The small toxins removal of the membranes was measured by using two typical solutions, para-nitro phenol solution (p-NP) and bisphenol A (BPA) solution. Typically, the composite membrane was applied to the toxin solution (0.1 mg mL⁻¹), and the concentration of the solution was monitored by an UV-vis spectrophotometer (UV-1800, Shimadzu) to investigate the adsorption property of the membrane. The adsorption amount (mg toxins per g membrane) was calculated based on the change of solution concentration. At least 4 parallel sample groups were used to regulate the results.

3. Results and discussions

3.1. Synthesis of polymer brush functionalized CNT

In this study, to introduce functional polymer brushes onto the CNT surface, as well as to improve the solubility of CNT, surface-initiated ATRP (SI-ATRP) was applied as is illustrated in Fig. 1(A). From the FTIR spectra in Fig. 1, a peak at 1720 cm⁻¹ was detected on CNT-Br, which was assigned to the carboxyl groups of the grafted BIBB.⁴⁰ In this study, bromide was attached onto the CNTs via the hydroxyl groups on the CNT surface and the acryl bromide groups of BIBB, which is a typical procedure to introduce reactive groups on material surfaces for ATRP. As is illustrated in the synthetic routes in Fig. 1(A), carboxyl groups will be formed after the reaction. Therefore, we believe that bromide has been introduced onto the CNT surface.

Fig. 1 (A) Synthesis of p-CNT by surface initiated atom transfer radical polymerization. (B) Visual images of CNT-OH, CNT-PMTAC and CNT-PEGMA being dispersed in DMAC. (C) FTIR spectra for CNT-OH, CNT-Br, CNT-PMTAC, and CNT-PEGMA.

MTAC is a quaternary ammonium monomer that has been widely used to control bacterial survival on substrate surfaces. EGMA, a functional monomer bearing PEG units as side chains, exhibits good water solubility and has been widely used as a hydrophilic monomer for the synthesis of functional co-polymers.^41–43 Herein, both were selected as functional monomers for membrane modification. By using CNT-Br as a macro-initiator, the EGMA or MTAC units were polymerized onto the CNT surface in the presence of CuBr and PMDETA. As is shown in Table 1, the grafting amounts of MTAC and EGMA were 7.71 g g⁻¹ CNT and 19.29 g g⁻¹ CNT (approx. 37.09 mmol g⁻¹ CNT and 40.61 mmol g⁻¹ CNT), respectively. Meanwhile, from the visual images in Fig. 1(B), it can be clearly observed that the solubility of the CNTs in DMAC was improved after grafting the polymer brushes.

To investigate the chemical structure of the p-CNTs, FTIR spectra were firstly acquired, as is shown in Fig. 1(C). For CNT-PMTAC, strong peaks at 1722 cm⁻¹ and 1180 cm⁻¹ were detected, which were attributed to the ester groups of the MTAC segments. In the spectra of CNT-PEGMA, the peak around 1724 cm⁻¹ was attributed to the carbonyl groups of the EGMA segments, while the peak at 1108 cm⁻¹ was assigned to the ether groups. Additionally, the peaks around 2867 cm⁻¹ in the spectra of the p-CNTs were ascribed to the symmetric and asymmetric modes of the methylene and methine groups of the MTAC and EGMA segments, which confirmed that the MTAC and EGMA brushes were successfully grafted onto the CNTs by SI-ATRP.

¹H NMR spectra for CNT-PEGMA and CNT-PMTAC in D₂O were also obtained. As is shown in Fig. 2(A), for CNT-PMTAC, the signals around 1 ppm, 3.2 ppm, 3.75 ppm and 4.45 ppm were the characteristic peaks of the MTAC units, which proved that the MTAC brushes were successfully grafted onto the CNTs. Meanwhile, for CNT-PEGMA, the signals at 1 ppm, 3.25 ppm and 3.57 ppm confirmed the existence of the EGMA units. To further characterize the polymer brushes on the CNT surface, high resolution-TEM (HR-TEM) images of pristine CNT, CNT-PEGMA and CNT-PMTAC were obtained as are shown in Fig. 2(B). For the pristine CNTs, a typical hollow tubular structure was observed, and for CNT-PEGMA and CNT-PMTAC, clear core–shell structures were observed. Transparent layers were found to be attached on the CNT surface, which were assigned to the polymer brushes. Gathering the results of the visual images, FTIR, ¹H NMR and TEM, it could be concluded that the polymer brushes had been grafted onto the CNT surface via SI-ATRP.

Fig. 2 (A) ¹H NMR spectra for CNT-PEGMA and CNT-PMTAC in D₂O. (B) HR-TEM images of pristine CNT, CNT-PEGMA and CNT-PMTAC, where the polymer layers are pointed out by red arrows. Scale bar: 100 nm. TEM images with higher resolution are shown in Fig. S1, ESI.†

3.2. Preparation and characterization of the membranes

After grafting the polymer brushes on the CNT surface, the composite membranes were prepared by a typical phase inversion technique. To study the effects of the polymer brushes, different ratios of CNT-PMTAC to CNT-PEGMA were used to prepare blended PES membranes as is shown in Table 2. Moreover, to further investigate the effect of the CNT contents, a membrane containing 5 wt% p-CNT was also prepared.

3.2.1. Chemical structure of the membranes. To analyse the chemical structure of the membranes, ATR-FTIR and XPS were applied as is shown in Fig. 3. From the ATR-FTIR spectra, characteristic peaks around 1731 cm⁻¹ were detected on the p-CNT modified membranes, which should be attributed to the C=O groups of the EGMA and MTAC units. In the XPS C 1s spectra, four characteristic signals were detected: the peak at 284.6 eV was attributed to the carbon skeleton of the membrane and the incorporated p-CNT, and the peak around 285.7 eV was attributed to the overlapped peaks of the C–N and C–S groups. The signals around 288.5 eV and 286.3 eV were ascribed to the C=O group and the C–O group, respectively, which were derived from the EGMA and MTAC units. Meanwhile, it was noticed that when increasing the dosage of CNT-PEGMA, the signal for the C–O groups was enhanced. The results of the ATR-FTIR and XPS C 1s spectra indicated that composite membranes with different ratios of CNT-PMTAC to CNT-PEGMA were prepared successfully.

Fig. 3 ATR-FTIR spectra (A) and XPS C 1s spectra (B) of the membranes. The XPS C 1s spectra of neat PES are shown in Fig. S2, ESI.†

3.2.2. Morphologies of the membranes. For the morphological study, SEM images of both the surface and the cross section of the membranes were acquired. From the surface SEM images in Fig. 4(A), it was observed that the pristine PES membrane displayed a smooth surface without any wrinkles; while the surface of the pristine CNT modified membrane (M-C in the figure) displayed more wrinkles and particles (pointed out by red arrows), which were proposed as being generated by the aggregation of the CNTs in the polymer matrix. After being grafted with polymer brushes, little wrinkles were observed on the membrane surfaces, which revealed that the miscibility between the CNTs and PES matrix was improved by the polymer brushes. Typical cross-sectional SEM images are shown in Fig. 4(B), and the detailed SEM images are shown in Fig. S3, ESI.† As shown, all the membranes exhibited typical ultrafiltration membrane structure with a dense skin-layer on the top, followed by a finger-like structure and porous architecture. For the p-CNT/PES membranes, no clear damage on the membrane structure was observed, indicating good miscibility between the p-CNTs and PES matrix. Meanwhile, as observed from the magnified SEM images, porous structures were generated by the p-CNTs, which would benefit the permeability of the membranes.

Fig. 4 (A) FE-SEM images of the surfaces of the membranes. (B) Cross-section SEM images of the pristine and composite PES membranes, where the microstructure of the p-CNTs was pointed out by red arrows. Detailed cross sectional SEM images are shown in Fig. S3, ESI.†

3.3. Water contact angle

The wettability of the membrane has been proven to have a significant impact on membrane performance including water permeability and antifouling properties. Earlier studies have pointed out that on hydrophilic surfaces, less membrane fouling and better water permeability than that on the hydrophobic surface would occur.^44,45 In this study, the hydrophilicity of the samples was assessed by the water contact angle (WCA). As is shown in Fig. 5, the pristine PES membrane owned a typical hydrophobic surface with a water contact angle higher than 75°. After the blending of p-CNT, the contact angles decreased significantly. Moreover, it was noticed that with the increase of EGMA units in the membranes, the contact angle decreased further, and the contact angle of M-E was around 57°. PEGMA is highly water soluble and it can form intermolecular hydrogen bonds with water molecules.⁴⁶ In this study, the p-CNTs can be considered as amphiphilic materials with hydrophobic CNT cores and hydrophilic PEGMA shells. As revealed by many studies, during phase separation, the amphiphilic modifier would migrate to the membrane surface and assemble with the hydrophilic chains directed to the surface while the hydrophobic chains are embedded in the membrane matrix.^30,47,48 During water–DMAC based phase separation, the hydrophilic PEGMA units may migrate to the membrane surface and pore surface, which thus leads to the formation of PEGMA brushes on the surface as is illustrated in Fig. 5(B). Moreover, from the above results of the XPS C 1s spectra of the membrane surface, the emerged peaks for the PEGMA units were detected and the signals for the C–O bonds were enhanced when increasing the dosage of PEGMA, which also verified the enrichment of the PEGMA units on the membrane surface. The MTAC units exhibited limited capability to absorb water molecules, therefore they demonstrated less impact on the membrane wettability than the EGMA units. The WCA changes within 2 minutes are shown in Fig. S4, ESI;† and it was observed that the water contact angles decreased with drop aging for the modified PES membranes, which also suggested that the wettability of the membrane was increased by the p-CNTs.

Fig. 5 (A) Static water contact angles of the membranes. Values are expressed as mean ± SD, n = 6. (B) The alignment of the p-CNTs in the membrane matrix. Due to the high water affinity of PEGMA, the polymer brushes tend to direct into the water phase during membrane formation, while the CNT core would be embedded in the membrane matrix. (C) Visual images of the water drops on the membranes.

3.4. Ultrafiltration experiments

To study membrane permeability, the dead-end filtration method was applied. Herein, the membrane permeability was characterized by water flux, and the antifouling property was revealed by the flux recovery ratio (F_RR), for which a higher F_RR value corresponded to a better antifouling property for the membrane.⁴⁸ Mimetic sewage (MS) solution that contained BSA, humic acid, and sodium alginate was prepared and applied for the F_RR test to illustrate the antifouling property during sewage treatment.³⁹ As is shown in Fig. 6, the pristine PES membrane had the lowest water flux, while the blending of pristine CNT showed no clear effect on water flux. For the p-CNT modified membranes, the water flux increased obviously, and with an increase of EGMA units, the flux increased further. With an increasing amount of p-CNT being blended into the membrane matrix, the water flux increased further. There are several factors that will influence the water permeability of the membranes, especially the membrane hydrophilicity and membrane structure. Hydrophilic surfaces are beneficial for water flow due to the high water affinity, while membranes with bigger pores always exhibit higher water flux due to the large channels for water flow. In this study, the membrane hydrophilicity was proven to be increased by the p-CNTs as indicated by the WCA values. To characterize the membrane pore changes after p-CNT modification, the PEG-20 000 sieving coefficients of the membranes were measured, and the results are shown in Fig. 6(E). It was observed that the blending of the p-CNTs led to a slight decrease in the sieving coefficients of the membrane, indicating that the p-CNTs could create bigger pores for the membrane. In addition, all of the membranes exhibited sieving coefficients higher than 80%. Earlier reports have also pointed out that rigid CNTs are able to generate porous structures for polymeric membranes.^34,49–51 Therefore, the increased water flux should be attributed to the synergistic effects of membrane hydrophilicity and porous structure.

Fig. 6 Dynamic flux of the membranes during a process of two cycles of BSA and MS ultrafiltration. (A and B) the flux for PBS and BSA solution; (C and D) the flux for water and MS; (E) the permeation properties of the membranes. The thickness for the membrane was 60 ± 1 μm.

Even though the water flux was significantly increased, all of the composite membranes showed BSA rejection ratios higher than 96%. For the antifouling property, the pristine PES showed the lowest F_RR values in both the cases of BSA and MS solutions, and the blending of neat CNTs showed no positive effect on the membrane antifouling property. However, for the p-CNT modified membranes, significantly increased F_RR values were noticed, especially for the EGMA incorporated samples. The antifouling property of the membranes was further characterized by protein adsorption and bacterial adhesion, which will be discussed in detail in the following sections.

3.5. Antifouling and antimicrobial efficiency

3.5.1. Protein adsorption. To further study the antifouling property of the membranes, the adsorption amounts of BSA, a typical protein that is related to membrane fouling, were measured. For water treatment, the proteins will block the pores to decrease the membrane permeability; additionally, the adsorbed proteins will also increase the affinity between the membrane and micro-organisms to benefit the adhesion of bacteria, fungi, algae and the like.¹⁴ Therefore, the measurement of the protein adsorption amount is necessary to investigate the membrane antifouling property. As is shown in Fig. 7, the M-M, which contained no PEGMA units, showed high BSA adsorption amounts, which were at the same level as the pristine PES membrane; whilst, with an increase in the CNT-PEGMA dosage, the adsorption amounts decreased significantly. As reported by many other studies, protein adsorption on the material surface is manipulated by many factors, such as surface charge, surface roughness, protein characters and especially the surface water wettability.^52–56 In this study, the improved membrane antifouling property should be attributed to the increased water wettability by the p-CNTs as characterized by the WCA measurements. Moreover, PEG and its derivates have long been studied as antifouling coating materials due to their robust ability to conjugate H₂O to form protective hydration layers.^56,57 The results also suggested that the PEGMA units were essential to protect the surface from being polluted by proteins.

Fig. 7 BSA adsorption amounts on the membrane surfaces. Values are expressed as mean ± SD, n = 4.

3.5.2. Bacterial adhesion and survival. As mentioned above, microbial fouling is one of the most severe drawbacks for membrane based purification technologies, and bacteria are the most prevalent microbes for membrane fouling.^58–60 Herein, we investigated the membrane antibacterial efficiency by observing bacterial adhesion and survival on the membrane surface. Gram-positive E. coli and Gram-negative S. aureus were selected as model bacteria. Typical fluorescence microscopy images with double staining of live/dead bacteria were acquired as is shown in Fig. 8, where the live bacteria were stained green and the dead bacteria were stained red. The details about bacterial adhesion and survival are shown in Fig. S4 and S5, ESI.† The number of live and dead cells was estimated from at least 6 fluorescence images for quantitative analysis.

Fig. 8 (A and C) Typical fluorescence microscopy images of the adhered E. coli and S. aureus (green staining shows live bacteria, while red staining shows dead bacteria). Scale bar: 20 μm. Detailed fluorescence microscopy images are shown in Fig. S5 and S6, ESI.† (B and D) Number of adhered bacteria on the membrane surface, which was estimated from at least 6 fluorescence microscopy images. Values are expresses as mean ± SD.

As shown, a large amount of live bacteria was observed on the pristine PES and pristine CNT/PES membranes. There are several reports that use oxidative derivatives of CNTs as modifiers for the modification of membranes and remarkable achievements have been made.^34,35,61 For instance, the Qian group observed an increased antifouling property of a PVDF membrane by using oxidized CNTs and graphene oxide (GO).³⁴ In the study, CNTs were pre-treated with oxidants like nitric acid and sulfuric acid. After the treatment, abundant oxygenic groups including carboxyl and hydroxyl groups would be generated on the CNT surface and hence greatly increase the water-solubility of the CNTs. When used for membrane modification, the oxidized CNTs served as hydrophilic nanomaterials to increase membrane wettability. Therefore, the membrane water flux and antifouling property could be increased by using oxidized CNTs. But in our study, to compare the effects of polymer brushes on the membrane performance, pure CNTs without functional groups were used as the control. Due to the limited hydrophilicity and functionality, little impact on the membrane performance was observed for the pure CNT modified membrane. In addition, some reports suggested that the bactericidal efficiency of the CNTs was due to a contact killing manner.^62,63 But in this study, since the CNTs are embedded in the PES matrix, direct contact between the CNTs and bacteria has been prevented. Therefore, pure CNTs as modifiers tightly embedded in the membrane matrix may not increase the membrane antibacterial efficiency.

For M-M, the number of adhered live bacteria was decreased significantly, indicating an improved bacterial killing activity due to the MTAC polymer brushes; but there were still abundant dead bacteria on the surface. Meanwhile, with the blending of CNT-PEGMA, the total amounts of the adhered bacteria decreased obviously, which showed a similar tendency with the results of the protein adsorption experiments. The bacterial killing percentages for the membranes were also estimated as is shown in Fig. S7, ESI.† It was noticed that the bacterial killing efficiency was increased with an increased dosage of CNT-PMTAC. M-M showed the highest killing percentages towards both E. coli (89.3%) and S. aureus (82.1%), while, the killing percentages for pure PES were only 9.5% and 3.3% against E. coli and S. aureus, respectively. The mixed membranes of CNT-PMTAC and CNT-PEGMA also exhibited effective killing towards the bacteria. The M-E membrane showed a robust resistance towards bacterial adhesion, but no clear killing towards either E. coli or S. aureus was observed. In this case, the adhered bacteria might proliferate well on the surface and eventually form an intense biofilm. Therefore, to achieve long-term antibacterial efficiency, the mixed membranes of CNT-PMTAC and CNT-PEGMA were more favorable than the neat CNT-PMTAC or CNT-PEGMA modified membranes.

In general, the mixed membranes of MTAC and EGMA owned the best antibacterial adhesion efficiency due to the synergistic effects of antibacterial adhesion and bacterial killing activity. Similar results have also been reported. For example, Yang’s study indicated that the antibacterial activity for the PEG/QACs coating was superior to the bare PEG coating.²⁷

3.5.3. Microbial coverage. In the above ultrafiltration experiments, protein adsorption and bacterial adhesion experiments, it was observed that the p-CNT/PES membranes exhibited much better antifouling and antibacterial activity than the neat PES membranes. Herein, to further study the membrane antifouling property, SEM observations for the microbial coverage were performed. As is shown in Fig. 9(A), massive fouling was noticed on the neat PES surface. For the p-CNT/PES membrane, the fouling was found to be controlled and only a limited adhesion of micro-organisms was noticed on M-E3M1. For quantitative analysis, the coverage was estimated by at least 6 SEM images as is shown in Fig. 9(B). For pristine PES, the coverage of micro-organisms was about 85%. After the blending of CNT-PMTAC, the micro-organism coverage decreased to 28%. When introducing EGMA, the coverage decreased further, especially for M-E3M1, and was below 9%. The results for microbial coverage were consistent with the results of bacterial adhesion and survival, which indicated that the membranes containing both MTAC and EGMA units showed superior antifouling property over those containing only MTAC or EGMA units. Moreover, it was observed that when increasing the dosage of p-CNTs, the antifouling property increased further, which should be attributed to the increased amounts of functional groups in the membrane matrix.

Fig. 9 (A) SEM images of the microbes adhered on the membranes. The adhered microbes were painted for a clearer view. Scale bar: 20 μm. The unmarked images are shown in Fig. S8, ESI.† (B) Microbial coverage on the membranes, which was estimated with at least 6 SEM images. Values are expresses as mean ± SD.

Taking together the results of flux recovery ratio, protein adsorption, bacterial adhesion and microbial coverage, it could be concluded that the p-CNTs could improve the antifouling property of the PES membrane. And the mixed membrane of CNT-PMTAC and CNT-PEGMA exhibited the best antifouling property due to the integrated advantages of bacterial killing and antibacterial adhesion.

3.6. Toxin removal

Nanomaterials, like CNTs, graphene, halloysite tubes, and mesoporous silica, are acknowledged as robust absorbents towards small molecules. The incorporation of absorbents with membranes has led to the development of mixed matrix membranes, which combine adsorption and diffusion in one process and hence display an accelerated toxin removal rate.^64–67 Herein, the adsorption amounts of small phenolic molecules, also known as endocrine disruptors, were investigated, using para nitro phenol (p-NP) and bisphenol A (BPA) as models. As is shown in Fig. 10(A), the adsorption amounts of the membranes were increased by the blending of p-CNT. Meanwhile, from Fig. 10(B), it was noticed that increasing the usage of p-CNT would further increase the toxin removal efficiency of the membranes. For sewage treatment, the removal of small molecular toxins was mainly achieved by reverse osmosis. Due to the poor rejection towards small molecules, the ultrafiltration membrane is not considered suitable for the clearance of small molecular toxins. However, herein, the p-CNTs enabled the membrane to remove small molecules by means of adsorption, which would increase the water purifying efficiency of the PES membrane and broaden its applications in water purification.

Fig. 10 (A) p-NP and BPA adsorption amounts of the membranes. (B) Time-dependent removal of p-NP and BPA for the membranes. Values are expresses as mean ± SD, n = 6.

3.7. State-of-art comparison with other literatures

Research about developing antifouling and antimicrobial membranes to push forward water treatment has gained intense interest and there are several review articles about antifouling and antibacterial materials.^14,15,68 Herein, we’d like to provide a brief summarization of recent progress in the modification of ultrafiltration membranes, as is shown in Table 3.

Table 3 Brief summary of recent progress on membranes with improved antifouling and antibacterial efficiencies. “The control” refers to the pristine membranes in the corresponding report

Matrix	Modifier	Antifouling property	Antibacterial efficiency	Ref.
PVDF	AgNP	FRR for BSA solution: ∼90% (the control: ∼50%)	Bacterial killing: clear inhibitory zones towards E. coli and S. aureus for the modified membrane	69
PES	AgNP/nanogel	BSA adsorption: 3.1 μg cm^−2 (the control: 8.3 μg cm^−2); FRR for BSA solution: 92.8% (the control: 74.3%)	Bacterial killing: E. coli: >97.5% (the control: 1.6%); S. aureus: >92% (the control: 3.2%)	72
PES	QAC-b-PES	BSA adsorption: greatly suppressed (no quantitative data provided); FRR for BSA solution: ∼89% (the control: ∼51%)	Bacterial adhesion: greatly suppressed (no quantitative data provided); bacterial killing: E. coli: 99% (the control: 20%); S. aureus: 99% (the control: 30%)	70
PES	QAC based copolymer	FRR for BSA solution: 95% (the control: 84%)	Bacterial killing: E. coli: 81.8% (the control: 0%)	73
PES	PEGMA, PMTAC	BSA adsorption: 2.9 μg cm^−2 (the control: 14.5 μg cm^−2); FRR for BSA solution: 93.5% (the control: 56.5%)	Bacterial adhesion (cells per cm^2): E. coli: <0.1 × 10^7 (the control: 1.07 × 10^7); S. aureus: <0.1 × 10^7 (the control: 1.41 × 10^7)	71
PVDF/PES	TiO2	FRR for BSA solution: 86.2% (the control: 64.6%)	Bacterial killing: clear inhibitory zones towards E. coli for the modified membranes	74
PVDF	Oxidized CNT, GO	FRR for BSA solution: 80.4% (the control: 29.7%)	Not studied	34
PVDF	GO	BSA adsorption: 3.5 μg cm^−2 (the control: 16.5 μg cm^−2); FRR for BSA solution: 86.3% (the control: 45.0%)	Not studied	75
PES	GO/polyethylenimine	BSA adsorption: ∼28 μg cm^−2 (the control: ∼61 μg cm^−2); FRR for BSA solution: 92.1% (the control: 86.6%)	Bacterial adhesion: greatly suppressed (no quantitative data provided); bacterial killing: 74.9% against E. coli	76
PVDF	CNT-QAC	FRR for BSA solution: 93.6% (the control: 61.4%)	Bacterial killing: E. coli: 92.7% (the control: −10.6%); S. aureus: 95.2% (the control: −10.0%)	77
PES	CNT-PMTAC, CNT-PEGMA	BSA adsorption: 1.9 μg cm^−2 (the control: 14.1 μg cm^−2); FRR for BSA solution: 91.4% (the control: 59.5%)	Bacterial adhesion (cells per cm^2): E. coli: 5.3 × 10^7 (the control: 21.5 × 10^7); S. aureus: 2.1 × 10^7 (the control: 10.3 × 10^7); bacterial killing: E. coli: 99% (the control: 20%); S. aureus: 99% (the control: 30%). Microbial coverage: 9% (the control: 85%)	This study

---

Over the past few decades, remarkable progress has been made to use nanometals and polymeric antibacterial agents to improve the membrane antifouling and antibacterial performance. For nanometals, silver nanoparticles (AgNPs) are one of the most widely used agents owing to their relatively low cost and robust bacterial killing efficiency. Besides, AgNPs can also improve membrane hydrophilicity due to their strong affinity with water molecules. In an earlier study, Li and co-workers successfully improved the antifouling and antibacterial efficiency of PES membranes by using AgNPs.⁶⁹ The F_RR of the membranes increased from 50% to 90% by adding AgNPs. Despite the great success using AgNPs, the shortcoming should not be ignored, which was that the embedded AgNPs might be corroded by oxygen and eventually lose their bacterial killing efficiency. Therefore, AgNPs may not be ideal materials for long-term antibacterial applications.

Quaternary ammonium compounds (QACs) are a positively charged polymer that exhibit strong bacterial killing efficiency.⁶⁸ But when used for membrane modification, one great limitation of QACs is their water-solubility, which makes it easy to wash them away during practical application. Therefore, copolymerizing with a hydrophobic segment is necessary to ensure membrane stability. For instance, the Xu group has synthesized a QAC–PES block copolymer and used it to control membrane fouling. The F_RR of the membrane was increased from 10% to 60% after modification, and the bacterial adhesion was significantly suppressed by this method.⁷⁰ To prevent the elution of QACs in the membrane, we have designed an in situ cross-linking polymerization strategy to introduce MTAC–PEGMA co-polymers into the PES matrix and the obtained membranes exhibited significant improved antifouling and antibacterial properties.⁷¹ The BSA adsorption amounts decreased from 14.0 μg cm⁻² to 2.9 μg cm⁻², while the F_RR for BSA solution increased from 56.5% to 93.5%. The bacterial adhesion on the membrane surface was also significantly inhibited, which was only about 10% of the control sample.

Besides the polymeric agents, numerous efforts have now been addressed on using functional nanomaterials as modifiers for membrane modification. Compared with polymeric modifiers, nanomaterials exhibit well-defined structures and can create porous nanostructures to increase membrane permeability. Pure CNTs are barely soluble in any solvent; therefore, surface treatment is always necessary to improve their solubility. The most typical procedure is oxidation with strong oxidants, like nitric acid and sulfuric acid. After treatment, a large amount of oxygenic groups will be introduced onto the CNT surface, which will confer the CNTs with water-solubility. Therefore, they can serve as a hydrophilic modifier to improve the membrane antifouling property. For example, the Qian group obtained PVDF membranes with a F_RR higher than 70% by using oxidized CNTs, whilst the F_RR for a neat PVDF membrane was only 29.7%.³⁴

In our study, the combined functionality of antibacterial adhesion and bacterial killing would confer the membrane with long-term antifouling properties, which would satisfy long-term usage for waste water treatment. Meanwhile, owing to the nanoscale structure of CNTs, the p-CNT/PES membranes also exhibited high water flux, effective rejection of BSA and adsorptive capability towards small molecules. Given these advantages, the p-CNT/PES membranes may have great potential for water treatment.

4. Conclusions

In this study, aiming to develop functional membranes with robust antifouling and antimicrobial efficiency, polymer brush grafted CNTs (CNT-PMTAC and CNT-PEGMA) were firstly synthesized via surface initiated ATRP, and then p-CNT/PES composite membranes were obtained via a phase inversion technique. ATR-FTIR spectra and XPS spectra confirmed the existence of functional polymer brushes on the membranes, and the SEM observations revealed that the p-CNTs exhibited good miscibility with the PES matrix. Owing to the excellent hydrophilicity of the polymer brushes, the p-CNT/PES composite membranes displayed improved water wettability, as revealed by the water contact angle measurement. The results of systemic ultrafiltration experiments of protein solution and sewage mimic solution indicated that the p-CNT/PES membranes owned much better antifouling properties than the pristine PES membrane. The combination of antibacterial adhesion and bacterial killing enabled the membranes with robust antimicrobial efficiency, as revealed by the bacterial adhesion and microbial coverage on the membrane surfaces. Furthermore, the blended p-CNTs enabled the membranes to remove small toxins by means of adsorption, which might improve the purifying efficiency of the membranes. In general, the p-CNT/PES composite membranes exhibited enhanced antifouling ability in ultrafiltration, efficient adsorption properties of small molecule toxins and high bactericidal capacity, and might serve as promising materials for water purification.

Acknowledgements

We acknowledge financial support by the National Natural Science Foundation of China (No. 51225303, and 51433007), the State Key Laboratory of Polymer Materials Engineering (Grant No. sklpme2015-1-03), and the Sichuan Province Youth Science and Technology Innovation Team (No. 2015TD0001). Dr Chong Cheng acknowledges the financial support of DRS POINT Fellowships of Freie Universitat Berlin.

Notes and references

1. M. A. Shannon, P. W. Bohn, M. Elimelech, J. G. Georgiadis, B. J. Marinas and A. M. Mayes, Nature, 2008, 452, 301–310.
2. A. Lima, S. Moore, M. Barboza, A. Soares, M. Schleupner, R. Newman, C. Sears, J. Nataro, D. Fedorko and T. Wuhib, J. Infect. Dis., 2000, 181, 1643–1651.
3. R. D. Ambashta and M. Sillanpää, J. Hazard. Mater., 2010, 180, 38–49.
4. A. Mills, R. H. Davies and D. Worsley, Chem. Soc. Rev., 1993, 22, 417–425.
5. R. K. Upadhyay, N. Soin and S. S. Roy, RSC Adv., 2014, 4, 3823–3851.
6. F. Macedonio and E. Drioli, Desalination, 2008, 223, 396–409.
7. G. M. Geise, H.-S. Lee, D. J. Miller, B. D. Freeman, J. E. McGrath and D. R. Paul, J. Polym. Sci., Part B: Polym. Phys., 2010, 48, 1685–1718.
8. A. G. Fane, R. Wang and M. X. Hu, Angew. Chem., Int. Ed., 2015, 54, 3368–3386.
9. S. Bolisetty and R. Mezzenga, Nat. Nanotechnol., 2016, 11, 365–371.
10. C.-C. Ho and A. L. Zydney, Ind. Eng. Chem. Res., 2001, 40, 1412–1421.
11. Y. Ma, T. Zhou and C. Zhao, Carbohydr. Res., 2008, 343, 230–237.
12. L. Song, J. Membr. Sci., 1998, 139, 183–200.
13. Z. Wang, J. Ma, C. Y. Tang, K. Kimura, Q. Wang and X. Han, J. Membr. Sci., 2014, 468, 276–307.
14. I. Banerjee, R. C. Pangule and R. S. Kane, Adv. Mater., 2011, 23, 690–718.
15. D. Rana and T. Matsuura, Chem. Rev., 2010, 110, 2448–2471.
16. H. Jenssen, P. Hamill and R. E. Hancock, Clin. Microbiol. Rev., 2006, 19, 491–511.
17. Q. Wu, Q. Qi, C. Zhao, C. Liu, L. Fan, W. Zhang, J. Shi and D. Guo, J. Mater. Chem. B, 2014, 2, 1681–1688.
18. G. Xu, D. Pranantyo, B. Zhang, L. Xu, K.-G. Neoh and E.-T. Kang, RSC Adv., 2016, 6, 14809–14818.
19. S. Fujimoto, R. Mizuno, Y. Saito and S. Nakamura, Heart Vessels, 2004, 19, 19–22.
20. J. P. Ruparelia, A. K. Chatterjee, S. P. Duttagupta and S. Mukherji, Acta Biomater., 2008, 4, 707–716.
21. A. V. Fuchs, S. Ritz, S. Pütz, V. Mailänder, K. Landfester and U. Ziener, Biomater. Sci., 2013, 1, 470–477.
22. Z. Jia and W. Xu, Carbohydr. Res., 2001, 333, 1–6.
23. P. Li, X. Li, R. Saravanan, C. M. Li and S. S. J. Leong, RSC Adv., 2012, 2, 4031–4044.
24. N. Beyth, I. Yudovin-Farber, R. Bahir, A. J. Domb and E. I. Weiss, Biomaterials, 2006, 27, 3995–4002.
25. S. Kittler, C. Greulich, J. Diendorf, M. Koller and M. Epple, Chem. Mater., 2010, 22, 4548–4554.
26. S. Buffet-Bataillon, P. Tattevin, M. Bonnaure-Mallet and A. Jolivet-Gougeon, Int. J. Antimicrob. Agents, 2012, 39, 381–389.
27. C. Yang, X. Ding, R. J. Ono, H. Lee, L. Y. Hsu, Y. W. Tong, J. Hedrick and Y. Y. Yang, Adv. Mater., 2014, 26, 7346–7351.
28. J. Deng, L. Ma, X. Y. Liu, C. Cheng, C. X. Nie and C. S. Zhao, Adv. Mater. Interfaces, 2016, 3, 1500473.
29. Y.-H. Zhao, B.-K. Zhu, L. Kong and Y.-Y. Xu, Langmuir, 2007, 23, 5779–5786.
30. C. Nie, L. Ma, Y. Xia, C. He, J. Deng, L. Wang, C. Cheng, S. Sun and C. Zhao, J. Membr. Sci., 2015, 475, 455–468.
31. C. Zhao, J. Xue, F. Ran and S. Sun, Prog. Mater. Sci., 2013, 58, 76–150.
32. M. S. Tijink, M. Wester, J. Sun, A. Saris, L. A. Bolhuis-Versteeg, S. Saiful, J. A. Joles, Z. Borneman, M. Wessling and D. F. Stamatialis, Acta Biomater., 2012, 8, 2279–2287.
33. M. S. Tijink, M. Wester, G. Glorieux, K. G. Gerritsen, J. Sun, P. C. Swart, Z. Borneman, M. Wessling, R. Vanholder and J. A. Joles, Biomaterials, 2013, 34, 7819–7828.
34. J. G. Zhang, Z. W. Xu, W. Mai, C. Y. Min, B. M. Zhou, M. J. Shan, Y. L. Li, C. Y. Yang, Z. Wang and X. M. Qian, J. Mater. Chem. A, 2013, 1, 3101–3111.
35. J. Zhang, Z. Xu, M. Shan, B. Zhou, Y. Li, B. Li, J. Niu and X. Qian, J. Membr. Sci., 2013, 448, 81–92.
36. C. Gao, C. D. Vo, Y. Z. Jin, W. Li and S. P. Armes, Macromolecules, 2005, 38, 8634–8648.
37. G. Capannelli, F. Vigo and S. Munari, J. Membr. Sci., 1983, 15, 289–313.
38. K. Yoon, K. Kim, X. Wang, D. Fang, B. S. Hsiao and B. Chu, Polymer, 2006, 47, 2434–2441.
39. X. Zhao and C. He, ACS Appl. Mater. Interfaces, 2015, 7, 17947–17953.
40. W. Zhao, C. He, H. Wang, B. Su, S. Sun and C. Zhao, Ind. Eng. Chem. Res., 2011, 50, 3295–3303.
41. X. Ding, C. Yang, T. P. Lim, L. Y. Hsu, A. C. Engler, J. L. Hedrick and Y.-Y. Yang, Biomaterials, 2012, 33, 6593–6603.
42. Y. Liu, Y. Su, Y. Li, X. Zhao and Z. Jiang, RSC Adv., 2015, 5, 21349–21359.
43. Y. Xia, C. Cheng, R. Wang, C. He, L. Ma and C. Zhao, Colloids Surf., B, 2016, 139, 199–210.
44. J. Huang, J. Xue, K. Xiang, X. Zhang, C. Cheng, S. Sun and C. Zhao, Colloids Surf., B, 2011, 88, 315–324.
45. X. Liu, Y. Xu, Z. Wu and H. Chen, Macromol. Biosci., 2013, 13, 147–154.
46. F. Ran, S. Nie, W. Zhao, J. Li, B. Su, S. Sun and C. Zhao, Acta Biomater., 2011, 7, 3370–3381.
47. F. Ran, S. Nie, W. Zhao, J. Li, B. Su, S. Sun and C. Zhao, Acta Biomater., 2011, 7, 3370–3381.
48. X. Zhao, Y. Su, Y. Li, R. Zhang, J. Zhao and Z. Jiang, J. Membr. Sci., 2014, 450, 111–123.
49. A. F. Ismail, P. S. Goh, S. M. Sanip and M. Aziz, Sep. Purif. Technol., 2009, 70, 12–26.
50. A. Rahimpour, M. Jahanshahi, S. Khalili, A. Mollahosseini, A. Zirepour and B. Rajaeian, Desalination, 2012, 286, 99–107.
51. E. Celik, H. Park, H. Choi and H. Choi, Water Res., 2011, 45, 274–282.
52. P. Roach, D. Farrar and C. C. Perry, J. Am. Chem. Soc., 2005, 127, 8168–8173.
53. K. Rechendorff, M. B. Hovgaard, M. Foss, V. Zhdanov and F. Besenbacher, Langmuir, 2006, 22, 10885–10888.
54. S. Jeon, J. Lee, J. Andrade and P. De Gennes, J. Colloid Interface Sci., 1991, 142, 149–158.
55. Q. Wei, T. Becherer, S. Angioletti-Uberti, J. Dzubiella, C. Wischke, A. T. Neffe, A. Lendlein, M. Ballauff and R. Haag, Angew. Chem., Int. Ed., 2014, 53, 8004–8031.
56. W. R. Gombotz, W. Guanghui, T. A. Horbett and A. S. Hoffman, J. Biomed. Mater. Res., 1991, 25, 1547–1562.
57. C. S. Gudipati, J. A. Finlay, J. A. Callow, M. E. Callow and K. L. Wooley, Langmuir, 2005, 21, 3044–3053.
58. W. T. Characklis and K. Cooksey, Adv. Appl. Microbiol., 1983, 29, 93–138.
59. W. Characklis, Biofilms, 1990, 523–584.
60. I.-S. Chang, P. Le Clech, B. Jefferson and S. Judd, J. Environ. Eng., 2002, 128, 1018–1029.
61. J. Ma, Y. Zhao, Z. Xu, C. Min, B. Zhou, Y. Li, B. Li and J. Niu, Desalination, 2013, 320, 1–9.
62. S. B. Liu, L. Wei, L. Hao, N. Fang, M. W. Chang, R. Xu, Y. H. Yang and Y. Chen, ACS Nano, 2009, 3, 3891–3902.
63. L. R. Arias and L. J. Yang, Langmuir, 2009, 25, 3003–3012.
64. P. Luo, Y. Zhao, B. Zhang, J. Liu, Y. Yang and J. Liu, Water Res., 2010, 44, 1489–1497.
65. J. Fan, C. Yu, F. Gao, J. Lei, B. Tian, L. Wang, Q. Luo, B. Tu, W. Zhou and D. Zhao, Angew. Chem., 2003, 115, 3254–3258.
66. C. Chen, J. Hu, D. Shao, J. Li and X. Wang, J. Hazard. Mater., 2009, 164, 923–928.
67. M. Kragulj, J. Trickovic, A. Kukovecz, B. Jovic, J. Molnar, S. Roncevic, Z. Konya and B. Dalmacija, RSC Adv., 2015, 5, 24920–24929.
68. R. Zhang, Y. Liu, M. He, Y. Su, X. Zhao, M. Elimelech and Z. Jiang, Chem. Soc. Rev., 2016 10.1039/c5cs00579e.
69. J.-H. Li, X.-S. Shao, Q. Zhou, M.-Z. Li and Q.-Q. Zhang, Appl. Surf. Sci., 2013, 265, 663–670.
70. Y. F. Zhao, L. P. Zhu, J. H. Jiang, Z. Yi, B. K. Zhu and Y. Y. Xu, Ind. Eng. Chem. Res., 2014, 53, 13952–13962.
71. C. Cheng, A. He, C. Nie, Y. Xia, C. He, L. Ma and C. Zhao, J. Mater. Chem. B, 2015, 3, 4170–4180.
72. Y. Xia, C. Cheng, R. Wang, C. Nie, J. Deng and C. Zhao, J. Mater. Chem. B, 2015, 3, 9295–9304.
73. X. L. Zhan, G. F. Zhang, X. Chen, R. He, Q. H. Zhang and F. Q. Chen, Ind. Eng. Chem. Res., 2015, 54, 11312–11318.
74. A. Rahimpour, M. Jahanshahi, B. Rajaeian and M. Rahimnejad, Desalination, 2011, 278, 343–353.
75. C. Zhao, X. Xu, J. Chen and F. Yang, J. Environ. Chem. Eng., 2013, 1, 349–354.
76. L. Yu, Y. Zhang, B. Zhang, J. Liu, H. Zhang and C. Song, J. Membr. Sci., 2013, 447, 452–462.
77. B. Kang, Y.-D. Li, J. Liang, X. Yan, J. Chen and W.-Z. Lang, RSC Adv., 2016, 6, 1710–1721.

---

Footnote

† Electronic supplementary information (ESI) available: Procedures for antibacterial experiments; TEM images for CNT, CNT-PMTAC and CNT-PEGMA; XPS C 1s spectra for the neat PES membrane; detailed cross-sectional SEM images of the membranes; changes of the water contact angle with prolonging the contact time with the membrane surface; detailed fluorescence microscopy images to reveal E. coli and S. aureus adhesion and survival on the membranes; bacterial killing percentages of the membranes; unmarked SEM images of the microbes adhered on the membranes. See DOI: 10.1039/c6ra18706d

---