Improved antifouling property of PVDF ultrafiltration membrane with plasma treated PVDF powder

Abstract

In this work, a novel modification method was introduced to improve the antifouling property of polyvinylidene fluoride (PVDF) ultrafiltration membranes, which were prepared with a one-pot process using plasma pretreatment and graft polymerization of PVDF powder. The amphiphilic copolymer of polyvinylidene fluoride-g-polyacrylic acid (PVDF-g-PAA) present in the modified PVDF membrane was demonstrated by infrared spectroscopy (ATR-FTIR) and X-ray photoelectron spectroscopy (XPS). The improved wetting ability was conducive to improve the water flux, bovine serum albumin rejection and antifouling ability of modified PVDF membranes. A totally novel pH quantitative method was employed to investigate the antifouling property based on the acid–base reaction of the amino acid residues of BSA (pollutant), according to the detected change of BSA adsorption amount on membrane surface and pore surface before and after modification, less irreversible fouling was detected in the modified PVDF membrane due to the existence of PVDF-g-PAA, and it was more convenient to remove the pollutants in the pores of modified PVDF membranes by water flushing.

---

1. Introduction

Polyvinylidene fluoride (PVDF) is known as one of the widely used separation membrane materials with excellent mechanical and physicochemical properties.^1–4 However, the PVDF membrane always exhibits strong hydrophobicity due to the symmetrical fluorine atoms in the molecular chain, leading to the serious membrane fouling problem. Membrane fouling, associated with nonspecific interactions between the hydrophobic pollutants and PVDF material, reduces the service life and increases the cost of PVDF membranes used in water treatment and other fields.^5–7

Numerous studies have demonstrated that the hydrophilicity of the membrane surface plays an important role for improving the antifouling properties of hydrophobic separation membranes,^8–10 it is generally accepted that the more hydrophilic the surface, the better the antifouling ability. As we all know, the hydrophobic membranes surface grafted with polyethylene glycol (PEG) based or zwitterionic based polymers^11–13 is able to exhibit excellent antifouling property due to the formed hydration layer, which can restrain hydrophobic effect between pollutants and hydrophobic membrane materials to decrease the irreversible fouling. However, the fouling problem of ultrafiltration (UF) membranes cannot be completely eliminated merely via surface hydrophilic modification.^14–16 Membrane fouling includes reversible and irreversible fouling and the fouled location includes membrane surface and membrane pores, the pollutants pushed into the pores and adhered to the pores surface will block the flow path of feed, leading to the great decline of flux,¹⁷ and these pollutants can't be washed away without chemical cleaning agent.⁸ Thus, the wetting ability of surface and pore surface of hydrophobic UF membranes are both ought to be improved though hydrophilic modification. The reported studies have shown that blending amphiphilic polymer is the most used method to improve the wetting properties of hydrophobic membrane materials,^18–20 hydrophilic interface of membrane surface and pore surface can be fabricated though blending amphiphilic polymer based on the assembly of hydrophilic and hydrophobic polymer chain segments. In order to improve the compatibility of amphiphilic polymers and PVDF material, attention have been paid to prepare the PVDF based amphiphilic polymer as additive to modified PVDF membrane, which is beneficial to exhibit better compatibility with PVDF matrix material. Kang²¹ obtains PVDF-g-PAA copolymer by ozone pretreatment PVDF powder and prepared a hydrophilic PVDF membrane. Li²² uses amphiphilic copolymer of PVDF-g-PEGMA to modify hydrophilic PVDF membrane, which exhibits perfect fouling recover ratio by water flushing. There is no doubt that the prepared PVDF membrane though blending PVDF based amphiphilic polymer will obtain excellent antifouling effect.

The objective of this study is to introduce a novel PVDF powder modification method to prepare PVDF based amphiphilic polymer via one-spot process for improving the antifouling ability of PVDF membranes. As a valid activation method, plasma is employed to pretreat PVDF powder to generate uniformly active site^23,24 and amphiphilic copolymer of PVDF-g-PAA are fabricated via followed polymerization, the modified PVDF membranes are prepared by one-pot process though non-solvent induced phase separation (NIPS) method. In order to investigate the antifouling performance of modified membrane, a totally new mean named pH quantitative method based on the acid–base reaction of the amino acid residues is proposed in this work for evaluating the improvement of antifouling property, cooperating flux recover ratio (FRR) to discuss the antifouling performance of the modified PVDF membrane.

2. Experimental

2.1. Materials

PVDF (FR 904, >99.5%, M_w ∼ 400 000) was purchased from 3F New Materials Co. (China), acrylic acid (AA, >99.8%) and bovine serum albumin (BSA, 67000 Da) were purchased from Sinopharm Chemical Reagent Co. (China). Dimethylacetamide (DMAC, AR) and hydrochloric acid (HCl) were purchased from local chemical company. All the reagents were used as received without further purification.

2.2. PVDF powder modification

Low-temperature plasma instrument (HD-1A, Zhongkechangtai, China) was employed to pretreat the PVDF powder. The PVDF powder was uniformly distributed on the cardboard and placed in the plasma chamber to treat for 100 s at 100 W with helium (He, 50 Pa) as discharge gas, after that, the treated PVDF powder was placed in the atmosphere for 1 h to oxidize sufficiently. A certain amount of treated PVDF powder was dissolved with DMAC in a three flask at room temperature, then AA monomer was added for the subsequent polymerization, a steady stream of nitrogen gas was bubbled through the mixtures for 30 min before the three flask being transferred into the oil bath to sustained reaction for 1 hour at 80 °C. The added mass of PVDF powder and AA monomer were shown in Table 1.

Table 1 The composition of casting solutions

Sample	Pretreatment PVDF powder (g)	PVDF powder (g)	AA (g)	DMAC (g)	PVDF powder of the second addition (g)	Coagulating bath (°C)
M1	0	1	4	30	5	20
M2	1	0	4	30	5
M3	2	0	4	30	4

---

2.3. Preparation of PVDF membranes

All the pure and modified PVDF membranes were prepared via non-solvent induced phase separation method by one-spot mode. When the polymerization reaction completed, another certain amount of pristine.

PVDF powder was added into the above mixture solution to obtain homogeneous casting solution, the casting solution was cast on a glass sheet with the casting knife of 200 μm and immediately immersed in water coagulation bath (20 °C). All the prepared membranes were stored in pure water for 24 h to remove the monomer and homopolymer of AA, the prepared membranes were stored in pure water for subsequent separation and antifouling test without drying process. All the casting solution compositions and casting conditions of PVDF membranes were showed in Table 1. In addition, different polymerization time of 1 h, 2 h and 4 h and coagulation bath temperature at 0 °C, 20 °C and 50 °C were applied to study the effect of prepared condition for the microstructures of modified membranes.

2.4. PVDF membrane characterization

The surface composition of PVDF membranes were examined by attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR), membranes were analyzed on a ZnSe crystal as the internal refection element with an aperture angle of 45°, the ATR-FTIR spectra were recorded with the accumulated average of 32 scans at 4 cm⁻¹ resolution on a Nicolet 8700 FTIR (Thermo Electron Co. USA). Membrane surface chemical composition was tested using X-ray photoelectron spectroscopy (XPS, Shimadzu AXIS UltraDLD, Japan), the take-off angle of photoelectron radiation was set at 90°. The water contact angles (CA) on the membranes were measured by OCA40Micro (Dataphysics Co., Germany) at 25 °C to evaluate the surface wetting ability, 3 μL of deionized water was dropped onto the membrane surface using a micro syringe and the results were obtained using the drop shape image analysis system. Surface and cross-section morphology of membranes were viewed with the field emitting scanning electron microscope (SEM, Hitachi SU8010, Japan) and all the samples were coated with gold before observation. The membranes were fractured by being immersed in liquid nitrogen for cross section observation.

2.5 Permeation measurement

The pure water flux of PVDF membranes were measured by a home-made filtration system (Fig. 1), the circulation filter system with an effective filtration area of 12.5 cm², the pressure of filtration cell was supplied by a water pump and all the filtration experiments were carried out at the pressure of 0.1 MPa, the volume of water permeation was collected for a certain time and flux (J) was calculated by the following eqn (1). The rejection (R) was tested with BSA solution (1 g L⁻¹, pH 7.4) as feed solution and the permeate concentrations were examined via UV-visible spectrophotometer (UV-1800) with the characteristic wavelength at 280 nm, R was calculated by the following eqn (2).where V (L) is the volume of permeated water, t (h) is the permeation time and A (m²) is the effective area of filtration. C and C_p are the concentrations of BSA in the feed and permeate, respectively.

Fig. 1 Schematic diagram of home-made filtration module.

2.6 Fouling test

The relative flux recovery ratio (FRR) and the irreversible flux ratio (IFR) were employed to evaluate the antifouling properties of modified PVDF membranes, which were calculated according to the eqn (3) and (4), it was noticeable that the higher FRR values (lower IFR value) of membranes indicate the better antifouling properties. The test process consisted of three steps, first of all, a stable flux (J) was obtained via pure water filtration, secondly, the feed solution was changed to BSA solution (1 g L⁻¹, pH 7.4) and another stable flux (J_B) was obtained, then the membrane was taken out from the filter system to immerse in the phosphate butter for 30 min, and flushed with pure water for 10 min, after the above washing process, the secondary pure water flux (J₂) was recorded.

Checked the methods listed in the literatures, the antifouling properties of membranes were mainly evaluated with surface adsorption mass and the flux recovery rate (FRR) of dynamic fouling process. However, unacceptable errors were obtained regularly because of protein denaturation and impact of different preloading time for membranes flux. Therefore, a more accurate method named pH quantitative method for the evaluation of antifouling properties was proposed in this paper, which was based on the amino acid residues of BSA having the consumption capacity of H⁺ and OH⁻ to effect the pH value of the solution contained BSA. Tuning the pH of BSA from 7.4 to 4.7 (isoelectric point of BSA), the consumed amount of acid was proportional to the molar amount of BSA. So the consumed acid amount of fouled membrane by BSA could exhibit the fouling situation. The specific details of pH quantitative method were described below, the tested membrane was placed in the dead-end filtration device (Fig. 1), and the BSA (1 g L⁻¹, pH 7.4) solution as pollutant feed was filtered for 2 hours. Then the membrane was taken out and placed in 100 mL ultrapure water to adjust the pH to 4.7 using HCl (0.01 mol L⁻¹), the consumed volume of HCl subtracting the consumed volume of membrane itself and water was assigned to the volume consumed by BSA. If the lower surface of the membrane was sealed with an impermeable support layer, the liquid was impermeable, so the membrane surface was the only fouled location, and the consumed volume by the adsorbed BSA on the membrane surface was labeled V₁, and the volume was labeled V_1W after flushing. When the feed solution could pass through the whole membrane with the applied pressure, BSA was able to contact the membrane surface and membrane pores, the consumed volume of HCl was labeled V, the consumed volume after cleaning was labeled V_W, all the details were shown in Table 2 and μL was the volume units.

Table 2 Consumption details of the HCl volume for pH quantitative method (the unit is μL)

	Whole membrane	Membrane surface	Membrane pores
Before flushing	V	V1	V − V1
After flushing	VW	V1W	VW − V1W
Flushing coefficient	VW/V	V1W/V1	VW − V1W/V − V1

---

Because the consumed amount of HCl was proportional to the molar amount of BSA, BSA adsorption amount could be calculated according to the eqn (5), which was determined in advance and the R² is 0.95. Where V (μL) was the consumed HCl volume, S (cm²) is the effective area of fouling. Y (μg cm⁻²) refers to the adsorption amount of BSA.

3. Results and discussion

As we all know, the plasma treatment could generate free radicals in the polymer chains, which would be oxidized by oxygen in air to gain hydroxyl group and peroxy group, and the peroxy group can generate free radicals by thermal initiation to take polymerization with the polymer monomer. PVDF powder was firstly treated by plasma in this paper and used to prepare PVDF-g-PAA copolymer to modify PVDF membrane, the mechanism for the preparation of modified PVDF membrane was shown in Fig. 2, AA monomers are used in the reaction of thermally initiated graft to fabricate PVDF based amphiphilic polymer. The PVDF-g-PAA existed in the modified PVDF membrane is characterized by ATR-FTIR and XPS.

Fig. 2 The proposed mechanism for the preparation of modified PVDF membrane.

Fig. 3a shows the ATR-FTIR spectra of the pure and modified PVDF membranes, the strong peak at 1715 cm⁻¹ appear in the spectra of M2 and M3 membranes, which indicate the existence of COOH, belonging to the PAA segment of amphiphilic PVDF-g-PAA. As a contrast, the characteristic peak of COOH cannot be viewed obviously in the spectrum of M1. To obtain further information about the composition of the modified membrane, the surface elemental composition percentages of pure and modified PVDF membranes are determined by XPS and the results are shown in Fig. 3b and c, the characteristic signal of oxygen in the XPS spectra are clearly observed, the surface the atomic percentages of oxygen increase from 1.05% to 16.49%, and the atomic percentages of fluorine element are significantly reduced, which means the oxygen-containing functional groups increase on the M2 and M3 membranes surface due to the added treated PVDF powder. In addition, the resolving results of M3 are shown in Fig. 3d and e, the C 1s core level spectra of M3 membrane consists of four peak components with binding energy at 285.7 eV, 287.0 eV, 288.5 eV and 291.3 eV, which are attributed to the CH₂, C–O, O–C=O and CF₂ species, the C–O and O–C=O are all assigned to the COOH group. As for the resolving results of O 1s core level spectra, a lot of oxygen element is detected on the M3 surface, and the peaks at 532.5 eV and 533.5 eV are agreement with the O–C=O and C–O binding energy, which are also attribute to the COOH group. It can be inferred that the PAA segments enriches on the modified PVDF membrane surface, fluorinated segments with low surface energy move toward the inside of the modified membrane based on the difference in free energy, and the oxygen-containing segment chains with high surface energy move to the membrane surface. The calculated molecular weight of PVDF-g-PAA in M3 by HNMR is about 450 000, the hydrophobic PVDF segments with high M_w are anchored in the membrane matrix and the hydrophilic PAA segments migrate to the membrane surface. These results are consistent with those reported in the literature,²¹ indicating the grafted copolymer PVDF-g-PAA is successfully synthesized and exists in the modified membranes. PAA homopolymer is easily washed away as porogen using pure water according to the ATR-FTIR and XPS spectra of M1.

Fig. 3 Surface composition of M1, M2 and M3 membranes ((a) ATR-FTIR spectra. (b) XPS spectra. (c) Surface atomic percentage. (d) C 1s resolving results of M3 membrane. (e) O 1s resolving results of M3 membrane).

Hydrophilic property is one of the most important factors that affect the antifouling property. The wetting ability of membranes is usually determined by the static water contact angle on the membrane surface. As shown in Fig. 4, the M1 membrane shows the highest contact angle of 91° and the modified M3 membrane exhibits the lowest water contact angle of 52°, it is clear that the increased PVDF powder in the membrane solution can decrease the water contact angle of membranes surface. It is commonly accepted that hydrophilic polymer chains in hydrophobic material are driven by interface free energy and enriched on the membranes surface, which are already demonstrated by ATR and XPS test. So the increased hydrophilicity of PVDF membrane is due to the enriched PAA chains on the membrane surface, and the amount of PVDF-g-PAA is related with the added PVDF powder.

Fig. 4 Water contact angle of pure and modified PVDF membranes.

In order to investigate the effect of grafted copolymer PVDF-g-PAA on PVDF membrane surface morphology, the surface SEM images of pure and modified PVDF membranes are shown in Fig. 5. Obviously, more pores appear on the surface of M2 and M3 compared to smooth M1, indicating amphiphilic copolymer is conducive to form the flow channel on the surface skin layer. As a general rule, surface morphology of PVDF membrane are affected by the hydrophilic polymer porogen or amphiphilic polymer additive, the interaction between hydrophilic PAA segments and water molecules increase the probability of forming membrane pores in double diffusion process of membrane formation, the phase aggregation process of amphiphilic polymer can effectively change the microstructure of membranes surface.^22,25 We also investigate the effect of formation condition for the whole structure of M3 membranes, viewed from the images of (a), (b) and (c) in Fig. 6, the cross section structure varies regularly depending on the coagulation bath temperature change from 0 °C to 50 °C, fewer finger like pores and thicker skin layer of the modified M3 membrane resulting from the high coagulation bath temperature are observed. In addition, Fig. 6b, d and e show a great number of spherulites formed in the matrix of the PVDF membrane with the increasing of the graft polymerization time, the generation of PVDF spherulites demonstrate that a large degree of micro-phase separation occurs due to the increased homopolymer of PAA in the graft polymerization process. Based on the different interfacial free energy of PAA and PVDF, PAA homopolymer with high molecular weight moving to the coagulation bath at low speed is conducive the to generate PVDF spherulites to change the bulk structure during NIPS process.

Fig. 5 Surface SEM images of PVDF membrane (M1) and modified membranes (M2 and M3).

Fig. 6 SEM images of the cross section morphology of M3 membrane prepared with different condition ((a–c) were prepared with different coagulating bath temperature of 0 °C, 20 °C and 50 °C. (b), (d) and (e) were prepared with different graft polymerization time of 1 h, 2 h and 4 h).

The permeation properties of all the membranes are tested by the self-made dead end filtration system, the results of pure water flux and the rejection of BSA solution are shown in Fig. 7, all the values are recorded when the flux of water or BSA solution are stable. Obviously, both the pure water flux and the BSA rejections of modified membrane (M2 and M3) are improved with the addition of treated PVDF powder, the pure water flux of M3 increases to 145 L m⁻² h⁻¹, and the BSA rejection increases by 18%. It may be attributed to the increased content of PVDF-g-PAA, the enriched hydrophilic PAA chains on the membrane surface are conducive to improve the wetting ability, so the hydrophilic surface can preferentially adsorb water molecules via hydrogen to enlarge the permeation flux and repel the hydrophobic BSA molecules to improve the rejection ratio,²⁶ which breaks the traditional trade-off phenomenon in membrane separation process.

Fig. 7 Pure water permeation flux and the BSA rejection of M1, M2 and M3 membranes.

The reversible fouling can be cleaned by water or buffer solution flushing, but the irreversible fouling as the main reason inducing membrane fouling cannot be washed away without chemical cleaning agent, which will lead to the degradation or oxidization of polymer membrane materials.⁸ Therefore, reducing the irreversible fouling is the most important factor in the preparation of antifouling separation membranes. To investigate the antifouling property under dynamic fouling process, the FRR values are employed. As shown in Fig. 8a, the water flux of all the membranes decrease dramatically at the initial fouling filtration, after the cleaning process, the secondary water flux are significantly different, high water flux of the M2 and M3 membranes are obtained compared to M1 membrane. All the FRR values of M1, M2 and M3 are listed in the Fig. 8b, the FRR of M3 is as high as 85% (the IFR is 15%), but the FRR of reference M1 is only 46%. The enhanced FRR values prove the antifouling property of modified PVDF membrane are improved though one-spot modification of pretreated PVDF powder, and the surface hydrophilization effect resulting from the contribution of hydrophilic PAA chains are beneficial to change the original hydrophobic interface of PVDF membrane surface and pore surface. PAA segments can reduce the adsorption probability of BSA on the membrane surface and pore surface though diminishing the hydrophobic interaction between hydrophobic membrane interfaces and hydrophobic pollutant, and the weakened anchoring effect is contribute to clean the BSA pollutants easily by water flushing, water molecules are selectively adsorbed on the hydrophilic segments to form a protective hydration layer to exclude BSA molecules for decreasing the irreversible fouling on membrane surface and pores.

Fig. 8 Time dependent flux variation (left) and FRR values (right) of prepared membranes at 0.1 MPa (0–60 minute: pure water, 60–120 minute: BSA solution, 160–200 minute, pure water).

In order to further verify the antifouling effect of the modified membrane, the pH quantitative method, a totally novel evaluation method based on the acid–base reaction of amino acid residues is proposed. As the instruction in experimental detail, the adsorption mount of BSA in different locations of membrane are calculated according to the consumed volumes of HCl solution, and the smaller the flushing coefficient, the better the antifouling ability.

As can be seen from Table 3, the adsorbed BSA amount of modified PVDF membrane after fouling process is reduced, the adsorption amount of BSA are 103 μg cm⁻², 68 μg cm⁻² and 57 μg cm⁻² for M1, M2 and M3 membranes, respectively. In addition, the flushing coefficient is also reduced, indicating the amount of irreversible pollutant in the modified membrane decrease effectively. According to the comparison of BSA adsorption amounts on membrane surface and membrane pores, it is found that the BSA amount in membrane pores are more than that on the membrane surface. According to the decreased flushing coefficient of modified membrane surface and pores, the pollutants in modified membrane pores and membrane surface are easily removed, indicating the adsorption capacity of BSA on the modified membrane surface and the membrane pores are weakened and the antifouling ability of the modified membrane was enhanced. All these results indicated that the introduction of hydrophilic PAA segments are beneficial to reduce the irreversible adsorption and accumulation of hydrophobic pollutants, PAA segment could be enriched on the membrane surface based on the anchoring effect of the hydrophobic PVDF segment to form the hydration layer structure near the surface of PVDF material to reduce the hydrophobic effect. Compared to the membrane surface, the pollutant in the membrane pores are more difficult to remove due to the physical barriers of structure and chemical effects of membrane pores, which is the important aggregation location of BSA pollutants. However, PAA segments can be employed to improve the hydrophilicity of membrane matrix including pore surface, so that the probability of pollutants being washed away from membrane pores is increased, and the irreversible fouling ratio is effectively reduced to enhance the overall antifouling ability of the PVDF membrane.

Table 3 The adsorbed BSA amount of all the fouled membranes (the unit is μg cm⁻²)

	M1			M2			M3
	Whole membrane	Membrane surface	Membrane pores	Whole membrane	Membrane surface	Membrane pores	Whole membrane	Membrane surface	Membrane pores
Before flushing	103	45.3	57.8	68	25.3	42.8	57	20.3	36.8
After flushing	71	28	43	38	15.3	22.8	30	10.5	19.5
Flushing coefficient	69%	62%	74%	56%	60%	53%	53%	52%	53%

---

4. Conclusions

The modified PVDF membranes with PVDF-g-PAA as antifouling additive were fabricated via one-spot method in this work. The novel plasma pretreatment and graft polymerization of PVDF powder were employed to generate PVDF-g-PAA, which is demonstrated existing in the membrane matrix by ATR-FTIR and XPS, the hydrophilic PAA chains enriched on the membrane surface lead to the water contact angle deceasing to 52°, and the water flux, BSA rejection and antifouling ability of modified membrane are all improved according to the improved hydrophilicity. The antifouling property is also tested by the FRR values of fouling filtration and pH quantitative methods, the fouling experiments confirm the added amphiphilic copolymer is conducive to reduce the irreversible fouling of the membrane pores and membrane surface to improve the overall antifouling performance of the PVDF membranes.

Acknowledgements

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

References

1. F. Liu, N. A. Hashim, Y. T. Liu, M. R. M. Abed and K. Li, J. Membr. Sci., 2011, 375, 1.
2. K. H. Oshima, T. T. Evans-Strickfaden, A. K. Highsmith and E. W. Ades, Biologicals, 1996, 24, 137.
3. X. Tan, S. P. Tan, W. K. Teo and K. Li, J. Membr. Sci., 2006, 271, 59.
4. G. Botelho, M. M. Silva, A. M. Goncalves, V. Sencadas, J. Serrado-Nunes and S. Lanceros-Mendez, Polym. Test., 2008, 27, 818.
5. R. Chan and V. Chen, J. Membr. Sci., 2004, 242, 169.
6. A. Saxena, B. P. Tripathi, M. Kumar and V. K. Shahi, Adv. Colloid Interface Sci., 2009, 145, 1.
7. F. Meng, S. R. Chae, A. Drews, M. Kraume, H. S. Shin and F. Yang, Water Res., 2009, 43, 1489.
8. Y. C. Woo, J. J. Lee, J. S. Oh, H. J. Jang and H. S. Kim, Desalin. Water Treat., 2013, 51, 5268.
9. R. Revanur, B. McCloskey, K. Breitenkamp, B. D. Freeman and T. Emrick, Macromolecules, 2007, 40, 3624.
10. J. R. Du, S. Peldszus, P. M. Huck and X. Feng, Water Res., 2009, 43, 4559.
11. P. Wang, K. L. Tan, E. T. Kang and K. G. Neoh, J. Membr. Sci., 2002, 195, 103.
12. Y. Chen, L. Ying, W. Yu, E. T. Kang and K. G. Neoh, Macromolecules, 2003, 36, 9451.
13. J. A. Callow and M. E. Callow, Nat. Commun., 2011, 2, 1.
14. A. Bhattacharya and B. N. Misra, Prog. Polym. Sci., 2004, 29, 767.
15. J. Wei, G. S. Helm, N. Corner-Walker and X. L. Hou, Desalination, 2006, 192, 252.
16. M. G. Buonomenna, L. C. Lopez, P. Favia, R. Agostino, A. Gordano and E. Drioli, Water Res., 2007, 41, 4309.
17. Z. Wang, X. Mei, J. Ma, A. Grasmick and Z. Wu, Sep. Sci. Technol., 2013, 48, 22.
18. J. F. Hester and A. M. Mayes, J. Membr. Sci., 2002, 202, 119.
19. F. Liu, Y. Y. Xu, B. K. Zhu, F. Zhang and L. P. Zhu, J. Membr. Sci., 2009, 345, 331.
20. J. F. Hester, P. Banerjee and A. M. Mayes, Macromolecules, 1999, 32, 1643.
21. L. Ying, P. Wang, E. T. Kang and K. G. Neoh, Macromolecules, 2002, 35, 673.
22. N. A. Hashim, F. Liu and K. Li, J. Membr. Sci., 2009, 345, 134.
23. S. Kaur, Z. Ma, R. Gopal, G. Singh, S. Ramakrishna and T. Matsuura, Langmuir, 2007, 23, 13085.
24. J. Chen, J. D. Li and C. X. Chen, Plasma Sci. Technol., 2009, 11, 42.
25. C. H. Loh and R. Wang, J. Membr. Sci., 2013, 446, 492.
26. A. Venault, Y. H. Liu, J. R. Wu, H. S. Yang, Y. Chang, J. Y. Lai and P. Aimar, J. Membr. Sci., 2014, 450, 340.

---

Footnote

† These authors contribute equally.

---