Dual-mode antifouling ability of PVDF membrane with a surface-anchored amphiphilic polymer

Abstract

Polydimethylsiloxane-graft-polyethylene glycol (PDMS-graft-PEG), an amphiphilic polymer with four hydrophilic arms, was prepared and used to improve the antifouling property of PVDF membranes by a dip-coating method. The interface properties of the modified PVDF membranes were investigated and it was found that the increased molecular weight of the PEG segment exhibited a negative effect on the anchoring ability of the PDMS-graft-PEG on the membrane surface. However, the antifouling properties of the modified membranes were significantly improved due to the dual-mode antifouling ability of fouling repulsion (PEG) and fouling release (PDMS). The adsorption mass of BSA on the modified membrane surface decreased to a low level, and an enhanced water flux recovery ratio after washing (FRR-W) of the modified membranes was obtained for two typical pollutants of bovine serum albumin and sodium alginate. It was demonstrated that surface anchoring is an effective way to improve the interface characteristics of membrane materials with functional polymers.

---

1. Introduction

As one of the most important technologies used in water treatment, membrane technology has attracted a lot of attention.¹ However, membrane fouling problems exist in almost all membrane processes due to the adsorption and accumulation of various pollutants on the membrane surface and pore surface, especially in the ultrafiltration and microfiltration processes, which lead to reduced separation efficiency and increased cost of membrane application.^2–5 Therefore, the preparation and modification of non-fouling or low-fouling membranes are imperative and urgent.

In general, the hydrophilic modification of membrane materials is an effective way to improve their antifouling property. Water molecules can be preferentially adsorbed on a membrane interface due to the hydrogen bonds and ionic bonds of the hydrophilic components. The formed interface hydration layer is able to effectively inhibit non-specific adsorption between pollutants and the membrane materials, thus reducing irreversible membrane fouling.^6–9 PEG-based and zwitterion-based materials^10,11 are considered to be the best fouling-resistant materials due to their superior hydration ability. As the most widely used type of antifouling material, PEG-based polymers with linear C–O–C segments can form strong hydrogen bonds with water molecules and exhibit a superior mobility as a steric barrier for pollutant molecules.^12–15 On the other hand, some types of strong hydrophobic material with lower surface energy, such as PDMS and fluorinated copolymers, are also able to exhibit a strong antifouling capacity. According to the fouling release ability of PDMS that results from its shear resistance property,¹⁶ even the adsorbed pollutants can be converted to cause reversible fouling as purgeable pollutants.

Nowadays, PEG-based and PDMS-based polymers used for the improvement of antifouling properties of membrane materials are mainly applied through surface grafting and blending methods,^17–20 but there are some complex technical requirements for surface grafting and blended copolymers due to the harsh reaction conditions and circumscribed fouling resistance of the introduced materials. In recent years, a simple modification method for hydrophobic membranes has been used, where the functional amphiphilic copolymer can be fixed onto the hydrophobic membrane surface by hydrophobic interaction,^21,22 and the resulting modified membranes are able to show improved antifouling property. Nishigochi²³ reported that poly-(MPC-co-BMA) anchored onto a PVDF membrane surface with hydrophobic interaction significantly reduces the adsorption of BSA on the PVDF membrane surface. PPO-b-PSBMA copolymers were used to enhance the surface hydrophilicity and protein resistance of PVDF membrane by hydrophobic interaction,²⁴ and the modified membrane showed an excellent anti-adsorption property for single-protein solution and complex plasma feed.

In this study, functional polymers with dual-mode antifouling properties were used to construct a fouling resistance surface. PDMS-g-PEG with four arms was prepared though a directional reaction of the terminal functional groups, and was used to modify a PVDF membrane by a dip-coating method. The surface properties of the modified PVDF membranes were characterized by ATR-FTIR, CA, SEM and AFM, and the antifouling properties were investigated through adsorption fouling and filtration fouling. This study aims to provide a simple modification method for constructing a functional surface to improve the antifouling property of separation membranes using a multifunctional polymer.

2. Experimental

2.1 Chemicals and reagents

Polyvinylidene fluoride (PVDF, MG15) was purchased from Arkema. Polydimethylsiloxane with a terminal hydroxyl (HO-PDMS-OH, M_w = ∼4000) was purchased from Fangzhou Chemical (China). Trimesoyl chloride (TMC) and sodium alginate (SA) were purchased from Aladdin Chemical Co. (China). Bovine serum albumin (BSA, 67 000 Da) and polyethylene glycol (PEG, M_w = ∼400, 1000, 4000, 10 000) were purchased from Sinopharm Chemical Reagent Co. (China). Dimethylacetamide (DMAC) and dichloromethane (CH₂Cl₂) were both analytical grade. All the reagents were used as received without further purification.

2.2 Synthesis of PDMS-g-PEG

The four-arm copolymers of PDMS-g-PEG were prepared using a quantitative and progressive reaction. HO-PDMS-OH and TMC in a molar ratio of 1 : 2 were dissolved in CH₂Cl₂ and the reaction was carried out for 1 h under magnetic stirring. PEG with different molecular weights (M_w = ∼400, 1000, 4000 and 10 000) were added to the abovementioned solution for another 2 h with magnetic stirring to generate the final reaction products. The molar ratio of PEG to PDMS of 8 : 1 was used, which was greater than the theoretical value (4 : 1), to increase the probability of reaction of the end groups.

2.3 Modification of PVDF membrane

Neat PVDF membranes were prepared through a non-solvent induced phase separation (NIPS) method. PVDF/PEG20000/DMAC (5 g/3 g/25 g) was selected to prepare the casting solution. The freeze-dried PVDF membrane was immersed in the abovementioned PDMS-g-PEG solution for 2 h in a water bath oscillator, and a crosslinking process was introduced by immersing the membrane in TMC (0.1%) solution for 10 min. Finally, the membrane was placed in distilled water and rinsed by oscillations for another 5 h. The PVDF membranes immersed in PEG reaction solution with different molecular weights (M_w ∼ 400, 1000, 4000, and 10 000) for modification were labeled as M400, M1000, M4000, and M10000, respectively.

2.4 Characterization

The prepared amphiphilic PDMS-g-PEG were examined with infrared spectroscopy (IR, Nicolet 8700, USA). The surface composition of the membranes was investigated with attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR, Nicolet 8700, USA) and X-ray photoelectron spectroscopy (XPS, Shimadzu AXIS Ultra DLD, Japan) with a take-off angle of 90°. The stabilities of the anchored PDMS-g-PEG polymers were investigated according to the peak intensity of the Si–O bond in the ATR-FTIR spectra before and after immersing in an alkaline solution (pH 10 and pH 12) prepared by NaClO for 1 h. The water contact angle (CA) on the membrane surface was measured with an OCA40 Micro system (Dataphysics Co., Germany) at room temperature to evaluate the surface wetting ability, and the results were obtained using a drop shape image analysis system. The surface morphology of the PVDF and modified membranes were viewed with a field emitting scanning electron microscope (SEM, Hitachi SU8010, Japan); all the samples were coated with gold before observation. The surface roughness (Rms) of the membranes was characterized with R_q values using an Agilent 5500 atomic force microscope (AFM, Agilent Technologies Inc., USA) in the tapping mode.

2.5 Permeation experiments

The separation properties of the membranes were measured by a dead-end filtration unit with an effective filtration area of 12.5 cm²; the pressure of the filtration cell was supplied by a water pump, and all the filtration experiments were carried out at a pressure of 0.1 MPa. The volume of the permeated water was collected and the stable flux was calculated using eqn (1). The rejection ratio (R) of the model pollutant (BSA, pH 7.4, 1 g L⁻¹) was calculated from the feed and the permeate concentrations via a UV spectrophotometer (UV-1800, Shimadzu) using eqn (2):where J (L m⁻² h⁻¹) is the volume of permeated water, t (h) is the permeation time, A (m²) is the effective area for filtration, and C and C_P are the concentrations of BSA in the feed and permeate, respectively.

2.6 Fouling test

Adsorption fouling and filtration fouling were employed to investigate the antifouling properties of the modified PVDF membranes. For the adsorption fouling, protein adsorbed on the membrane surface under static and dynamic conditions (with or without stirring) were analyzed though quantitative analysis, with BSA as the model pollutant. All the tested membranes were cut into regular shapes and immersed into BSA (0.5 g L⁻¹, pH 7.4) in phosphate buffer solution. After the adsorption process with or without stirring for 12 h at room temperature to reach adsorption equilibrium, the concentrations of BSA solution before and after adsorption were measured with a UV-vis spectrophotometer, and the adsorption mass on the membrane surface was calculated using eqn (3):where M is the adsorption mass (μg cm⁻²), C₀ and C_d are the concentrations (g L⁻¹) of BSA before and after adsorption test, V (L) is the volume of the BSA solution and S (cm²) is the membrane area.

For filtration fouling, BSA and SA were used as two types of model pollutants. The fouled flux recovery ratio (FRR) and the fouled flux recovery ratio after washing (FRR-W) were employed to evaluate the antifouling ability; it should be noted that a membrane with the higher values of FRR and FRR-W exhibits better antifouling properties. The loop filtration process conducted consisted of three steps: first, a stable flux (J) was obtained by pure water filtration; second, the feed solution was changed to the pollutant solution (BSA or SA) and another stable flux (J_P) was obtained, followed by changing the pollutant solution to pure water again and obtaining a new stable flux (J₂); finally, a washing process was introduced and the membrane was withdrawn from the filter system and immersed in a phosphate buffer for 30 min. It was then rinsed with pure water for 10 min, and reinstalled back into the filtration unit, and the second water flux (J_2w) of pure water was recorded at the end. Another model pollutant solution of SA (1 g L⁻¹) was used to test the antifouling ability of the prepared membranes with the same mode. All the FRR and FRR-W values were calculated using eqn (4) and (5), respectively:

3. Results and discussion

As shown in Fig. 1, based on the directed reaction of the end groups and the feeding order, the esterification reaction could occur between the PDMS-OH and TMC, followed by the reaction between PEG and remaining functional groups of TMC; thus, the final reaction product should be an amphiphilic polymer with a four-arm structure. The IR spectra of PDMS-g-PEG1000 and PDMS-g-PEG4000 are shown in Fig. 2. Clearly, the two polymer spectra exhibited the same characteristic peaks but different peak intensities. The peaks at 2962 cm⁻¹ and 1260 cm⁻¹ were assigned to the absorption peak of methyl, and those at 1093 cm⁻¹ and 1024 cm⁻¹ were attributed to the stretching vibration peak of Si–O bond; these peaks were all assigned to PDMS. The strong peak at 1730 cm⁻¹ indicated that the esterification reaction of chloride and hydroxyl had occurred and that TMC was a composition of synthetic polymer. In addition, a broad peak that appeared around 3100 cm⁻¹ to 3400 cm⁻¹ was a characteristic absorption peak of OH and another at 2876 cm⁻¹ belonged to the stretching vibration of CH₂, which were attributed to the characteristic absorption peak of PEG. All these results demonstrated that PDMS, TMC and PEG exist in the synthesized copolymer of PDMS-g-PEG1000 and PDMS-g-PEG4000; thus, the hydrophilic PEG component and hydrophobic PDMS component constituted the amphiphilic polymer PDMS-g-PEG.

Fig. 1 Schematic diagram of amphiphilic polymer synthesis and membrane modification.

Fig. 2 IR spectra of the PDMS-g-PEG1000 and PDMS-g-PEG4000 copolymers.

The amphiphilic polymer could be assembled onto the hydrophobic membrane surface by hydrophobic interaction.²⁴ The composition of the modified membrane surface was measured by ATR and XPS. Fig. 3a shows the ATR spectra of PVDF membranes before and after modification with wavenumber ranging from 700 cm⁻¹ to 1800 cm⁻¹. Both the 1093 cm⁻¹ and 1024 cm⁻¹ peaks assigned to the Si–O bond characteristic peak were observed in all the spectra of the modified membranes; furthermore, the peak intensity of the Si–O bond varied with the molecular weight of PEG. M400 and M10000 showed the strongest and weakest peak intensity of the Si–O bond, respectively, indicating that low molecular weight PEG was conducive to anchor the copolymer on the membrane surface. The hydrophilic component of PEG affected the hydrophobic interaction between the amphiphilic polymer and PVDF membrane. XPS analysis of the membrane surface composition is shown in Table 1. According to the composition of the PVDF membrane and PDMS-g-PEG, Si element and F element belong to PDMS and PVDF, respectively, and the O element belongs to PEG and TMC. The Si/F ratio and Si/O ratio of the PVDF membrane were both 0, suggesting that there was no Si element in the PVDF membrane. For M400, M1000, M4000 and M10000, the Si/F ratio decreased from 0.95 to 0.04, indicating that PDMS-g-PEG covered the PVDF membrane surface, and that the adsorbed amount decreased with the increasing molecular weights of PEG. Furthermore, the Si/O ratio also decreased with the increasing molecular weight of PEG, which proved the increase of PEG component in the PDMS-g-PEG copolymer.

Fig. 3 (a) ATR spectra of PVDF membranes before and after modification; (b) normalized peak intensity of Si–O bond before and after alkali treatment.

Table 1 XPS analysis results of neat and modified PVDF membranes

Samples	Si/F atomic ratio	Si/O atomic ratio
PVDF	0	0
M400	0.95	0.62
M1000	0.65	0.37
M4000	0.19	0.12
M10000	0.04	0.05

---

Because the stability of a hydrophobic interaction is not as good as that of a covalent bond, and because PVDF membranes are typically used under harsh environments, the stability of the anchored PDMS-g-PEG was investigated by treatment in an alkaline environment, and its performance was evaluated by measuring the variation in the intensity of the detected peak for the Si–O bond in the ATR-FTIR spectra. As shown in Fig. 3b, the peak intensity decreased slightly after the treatment by an NaClO solution with a pH value of 10, and the peak intensity sharply decreased due to the harsh environment at pH 12, indicating polymers anchored on the membrane surface by hydrophobic interaction could stably exist in a weak alkaline environment.

Fig. 4 shows the surface wetting ability of the neat and modified PVDF membranes. There is no constant initial contact angle (CA) for the membrane surface in Fig. 4a, but the CA of the modified membranes showed a decreasing tendency as the time increased, and M4000 exhibited the minimum CA of 49°, indicating that the anchored PDMS-g-PEG was able to carry out a phase separation process, and then the opposite migration of hydrophilic and hydrophobic segments changed the wetting ability of the modified membrane surface. In addition, Fig. 4b shows the advancing angles and receding angles of all the membranes. The increased advancing angles of the modified membrane were attributed to the low surface energy segments of PDMS, while the decreased advancing angles of the modified membrane were attributed to the high surface energy segments of PEG. It could be inferred that the anchored PDMS-g-PEG changed the microenvironment of the membrane surface by surface reconstruction.

Fig. 4 Water contact angle of neat and modified PVDF membranes ((a): time dependent contact angle variation; (b): advancing and receding contact angles).

Surface morphologies of the PVDF membranes before and after modification are shown in Fig. 5. Lots of membrane pores were observed on the uniform PVDF membrane surface, and there was no difference between M10000 and the neat PVDF membrane. The surface morphologies of the modified M1000 and M4000 membranes differed significantly, with the dimension of membrane pore diminishing apparently, indicating the distinct adsorbed mass of PDMS-g-PEG1000 and PDMS-g-PEG4000 on the membrane surface; moreover, the adsorption amount of PDMS-g-PEG10000 was less than others. These results indicated that the amphiphilic polymer with an appropriate molecular weight could be adsorbed onto the membrane surface and could affect the microstructure of the membrane surface. Table 2 lists the Rms of neat and modified PVDF membranes, with the dimensions of 5 μm × 5 μm and 1 μm × 1 μm. Similar variations of Rms were observed for all the membranes. M4000 exhibited a changed Rms value of 30.231 nm, indicating that the PDMS-g-PEG was absorbed on the PVDF surface, and also showed that the larger the roughness, the greater the amount of adsorbed polymer. M10000 showed an approximate Rms with the PVDF membrane, indicating a little adsorbed amount of PDMS-g-PEG10000 on the membrane surface. All these AFM results were consistent with the SEM results. Furthermore, the increased molecular weight of PEG showed a negative effect on the anchoring effect of the PDMS-g-PEG copolymers on the membrane surface.

Fig. 5 SEM images of neat and modified PVDF membranes.

Table 2 AFM analysis results of neat and modified PVDF membranes

Samples	Rms (5 μm × 5 μm)	Rms (1 μm × 1 μm)
PVDF	12.534 nm	5.324 nm
M400	15.657 nm	7.767 nm
M1000	25.165 nm	9.141 nm
M4000	30.231 nm	10.977 nm
M10000	13.461 nm	6.322 nm

---

Separation performances of the modified PVDF membrane were investigated, and the flux and rejection results are shown in Fig. 6. There are no significant differences in the flux for all the membranes, but the rejection showed an increasing trend for the PVDF, M400, M1000 and M4000 membranes, and the M4000 membrane exhibited the highest R value of 81%. Considering this with the abovementioned XPS and SEM results, it could be concluded that the increased R values resulted from the reduced membrane pores and the hydrophilic PEG component, due to the adsorbed PDMS-g-PEG. The hydrophilic and hydrophobic segments could play different effects for water flux; thus, the water fluxes of the modified membranes were complex. The changed separation efficiency proved that the physical structure and chemical properties on the membrane surface were altered due to the fixed four-arm PDMS-g-PEG polymer coated onto the membrane surface by a dip-coating method.

Fig. 6 Separation performances of neat and modified PVDF membranes (0.1 MPa).

The antifouling properties of the modified membranes were investigated by adsorption fouling and filtration fouling. The results of static and dynamic adsorption fouling are shown in Fig. 7, with BSA as the adsorption pollutant. According to the comparison of the static adsorption capacity of each membrane, the BSA adsorption amount of the modified membranes decreased significantly; the adsorption mass of the M4000 membrane decreased to 22 μg cm⁻². As for the dynamic adsorption, the adsorption capacity of the M400, M1000 and M4000 membranes decreased more significantly with the minimum value of the adsorption mass being 10 μg cm⁻². An ultra-low adsorption capacity for BSA indicated that the modified membrane possessed the ability to suppress nonspecific adsorption between the pollutant and hydrophobic membrane surface due to the anchored four-arm PDMS-g-PEG. During the static adsorption fouling process, the reconstruction of the amphiphilic PDMS-g-PEG provided the basis for improving the anti-adsorption ability. PEG molecules with an extended state were capable of decreasing the probability of the approach of BSA molecules to the membrane surface and their aggregation due to the hydrophilicity and locomotor activity of the PEG chains. For the dynamic adsorption fouling, even adsorbed BSA molecules on the membrane surface could be released by the shear flow resulting from the shear resistance of the PDMS segments; with the dual-mode antifouling property of PEG and PDMS, the adsorption capacity of the PVDF membrane for pollutants was further weakened. The adsorption mass of M10000 was similar to that of the PVDF membrane, indicating a minimum amount of PDMS-g-PEG polymer was anchored on the M10000 membrane surface, which was consistent with the abovementioned characterization data.

Fig. 7 Adsorption fouling results of neat and modified PVDF membranes.

The test results of filtration fouling are shown in Fig. 8a using BSA as the pollutant feed. Due to the strong hydrophobic interaction between PVDF and the hydrophobic part of BSA, BSA was able to induce serious irreversible fouling to the PVDF membrane. Fig. 8b shows that the FRR values of the PVDF, M400, M1000 and M4000 membranes without washing were 57%, 61%, 66% and 71%, respectively. The increased FRR values of the modified membranes indicated that the introduced PDMS-g-PEG could exhibit a repulsive ability for pollutants during the filtration process. Furthermore, the FRR-W values after washing were 62%, 75%, 86% and 90%, and the enhanced FRR values showed a preferable ability to release pollutants during the flushing process. The reversible fouling composition of the membrane increased due to the anchored PDMS-g-PEG, leading to a reduction in the irreversible fouling ratio, which reduced the effects of pollution on the separation efficiency and thus prolonged the service life of the PVDF membrane. The surface phase separation of the anchored PDMS-g-PEG in water enhanced the functional performance of the amphiphilic polymer, the fouling repulsion ability of the PEG segments and the fouling release ability of the PDMS segments, thus providing these characteristic dual-mode antifouling properties for the PVDF membrane.

Fig. 8 Fouling test of neat and modified PVDF membranes with BSA as pollutant ((a) time dependent permeate flux variation: 0–50 min, water; 55–110 min, BSA solution; 115–150 min, water flux; 155–190 min, washing process; and 195–240 min, water. (b) FRR and FRR-W values).

Fig. 9a shows the filtration fouling results with SA as the pollutant feed. The hydrophilic SA with a characteristic cake layer formation during filtration could quickly decrease the separation effect of the separation membrane. However, the FRR values and FRR-W values were significantly increased after the surface modification with anchored PDMS-g-PEG on the PVDF membrane surface, and the FRR-W value of M4000 was as high as 95%. Obviously, due to the phase separation behavior and dual-mode antifouling function of the PDMS-g-PEG amphiphilic polymer, the adsorption and deposition of pollutants on the membrane surface were suppressed, leading to a significant reduction in irreversible fouling.

Fig. 9 Fouling test of neat and modified PVDF membranes with SA as pollutant ((a) time dependent permeate flux variation: 0–50 min, water; 55–110 min, SA solution; 115–150 min, water flux; 155–190 min, washing process; and 195–240 min, water. (b) FRR and FRR-W values).

4. Conclusions

An amphiphilic polymer of PDMS-graft-PEG with four arms was fabricated and anchored to alter the interface properties of a membrane surface by hydrophobic interaction for improving the antifouling property of a PVDF membrane. The modified PVDF membranes exhibited a dual-mode antifouling ability of fouling repulsion and fouling release, thus exhibiting an improved anti-adsorption ability and enhanced FRR-W values for the two typical pollutants feed BSA and SA, and the irreversible fouling was effectively reduced as well. It was demonstrated that surface anchoring is an effective way to improve the interface characteristics of hydrophobic material for antifouling modification.

Acknowledgements

This study is supported by grants from the National Science Foundation of China (No. 21174027), Program for New Century Excellent Talents in University (No. NCET-12-0827). Program of Introducing Talents of Discipline to Universities (No. 111-2-04) and Innovation Funds for the PhD Students of Donghua University (CUSF-DH-D-2015027).

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. M. Elimelech and W. A. Phillip, Science, 2011, 333, 712–717.
3. D. Rana and T. Matsuura, Chem. Rev., 2010, 110, 2448–2471.
4. W. Chen, Y. Su, J. Peng, X. Zhao, Z. Jiang, Y. Dong, Y. Zhang, Y. Liang and J. Liu, Environ. Sci. Technol., 2011, 45, 6545–6552.
5. A. M. Mayes and A. Asatekin, Environ. Sci. Technol., 2009, 43, 4487–4492.
6. H. Yamamura, K. Kimura, T. Okajima, H. Tokumoto and Y. Watanabe, Environ. Sci. Technol., 2008, 42, 5310–5315.
7. M. Kobayashi, Y. Terayama, H. Yamaguchi, M. Terada, D. Murakami, K. Ishihara and A. Takahara, Langmuir, 2012, 28, 7212–7222.
8. M. Kobayashi, Y. Terayama, M. Kikuchi and A. Takahara, Soft Mater., 2013, 9, 5138–5148.
9. A. W. Zularisam, A. F. Ismail and R. Salim, Desalination, 2006, 194, 211–231.
10. A. B. Lowe and C. L. McCormick, Chem. Rev., 2002, 102, 4177–4189.
11. Y. F. Yang, Y. Li, Q. L. Li, L. S. Wan and Z. K. Xu, J. Membr. Sci., 2010, 362, 255–264.
12. A. M. Mayes, A. Asatekin, K. Seoktae and M. Elimelech, J. Membr. Sci., 2007, 298, 136–146.
13. E. M. van Wagner, A. C. Sagle, M. M. Sharma, Y. H. La and B. D. Freeman, J. Membr. Sci., 2011, 367, 273–287.
14. Y. W. Chen, L. Ying, W. H. Yu, E. T. Kang and K. G. Neoh, Macromolecules, 2003, 36, 9451–9457.
15. Y. Chang, Y. J. Shih, R. C. Ruaan, A. Higuchi, W. Y. Chen and J. Y. Lai, J. Membr. Sci., 2008, 309, 165–174.
16. Y. Cho, H. S. Sundaram, C. J. Weinman, M. Y. Paik, M. D. Dimitriou, J. A. Finlay, M. E. Callow, J. A. Callow, E. J. Kramer and C. K. Ober, Macromolecules, 2011, 44, 4783–4792.
17. Y. Chang, C. Y. Ko, Y. J. Shih, D. Quemener, A. Deratani, T. C. Wei, D. M. Wang and J. Y. Lai, J. Membr. Sci., 2009, 345, 160–169.
18. M. Hashino, K. Hirami, T. Ishigami, Y. Ohmukai, T. Maruyama, N. Kubota and H. Matsuyama, J. Membr. Sci., 2011, 384, 157.
19. J. F. Hester, P. Banerjee, Y. Y. Won, A. Akthakul, M. H. Acar and A. M. Mayes, Macromolecules, 2002, 35, 7652–7659.
20. P. Wang, K. L. Tan, E. T. Kang and K. G. Neoh, J. Membr. Sci., 2002, 195, 103–110.
21. S. Park, K. B. Lee, I. S. Choi, R. Langer and S. Jon, Langmuir, 2007, 23, 10902–10905.
22. N. J. Lin, H. S. Yang, Y. Chang, K. L. Tung, W. H. Chen, H. W. Cheng, S. W. Hsiao, P. Aimar, K. Yamamoto and J. Y. Lai, Langmuir, 2013, 29, 10183–10193.
23. S. Nishigochi, T. Ishigami, T. Maruyama, Y. Hao, Y. Ohmukai, Y. Iwasaki and H. Matsuyama, Ind. Eng. Chem. Res., 2014, 53, 2491–2497.
24. A. Venault, Y. Chang, H. S. Yang, P. Y. Lin, Y. J. Shih and A. Higuchi, J. Membr. Sci., 2014, 454, 253–263.

---

Footnote

† These authors contributed equally.

---