Composite nanofiltration membranes via the co-deposition and cross-linking of catechol/polyethylenimine

Abstract

High performance nanofiltration (NF) membranes are facilely fabricated via the co-deposition of catechol (CCh) and polyethylenimine (PEI) on the surface of a polysulfone (PSf) ultrafiltration membrane, with subsequent cross-linking by glutaraldehyde (GA). The surface properties of the studied membranes have been investigated in detail by Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, scanning election microscopy, atomic force microscopy, zeta potential, and water contact angle. The NF performance of the membranes are dependent on the CCh/PEI ratio, co-deposition time and cross-linking condition. Results reveal that the optimum membrane yields a rejection of 88% and a permeation flux of 25 L m⁻² h⁻¹ when filtrating the 1000 mg L⁻¹ MgCl₂ solution at 0.6 MPa. And the negatively charged membrane surface is related to the following salt rejection sequence: MgSO₄ > Na₂SO₄ > MgCl₂ > CaCl₂ > NaCl. Meanwhile, the membranes show excellent operation stability during a 240 h consistent filtration test.

---

Introduction

In recent years, nanofiltration (NF) has gained increasing popularity as an pressure-driven membrane process positioned between ultrafiltration and reverse osmosis, which allows for the high rejection of low molecular weight organic compounds (200–1000 Da) and multivalent ions with relatively high fluxes.^1–3 NF technology also has the features of low cost and energy consumption, facile operating conditions and broad applications in water softening,^4,5 wastewater reclamation^6,7 and industrial substance separation.⁸ To fulfill the as-mentioned performance, most NF membranes are thin-film composite (TFC) membranes and demand a skin layer with proper compactness. Traditional methods to prepare NF membranes include layer-by-layer assembly,⁹ phase inversion,¹⁰ surface grafting,¹¹ and interfacial polymerization.^12–14 Among them, interfacial polymerization has been developed as the dominating method, which guarantees both high permeation flux and effective separation performance. Nonetheless, the polymerization reaction is too intense and complicated to be precisely controlled.¹⁵ Furthermore, the prepared membranes may have poor adhesion strength between the skin layer and the ultrafiltration substrate.¹⁶ Consequently, it is worthwhile to research on novel and facile methods for high-performance TFC NF membranes.

Inspired by the remarkable and universal adhesive ability of marine mussel on various substrates, oxidative polymerization of dopamine (DA) into melanin-like polydopamine (PDA) coatings have become a versatile approach for surface modification.^17–19 Several attempts have been made to investigate the feasibility of adopting PDA coatings as the active layer for TFC NF membranes. Xu et al.²⁰ firstly developed a NF membrane by immersing PSf membrane into a fresh-made dopamine solution. The resultant membrane exhibited high flux while its salt rejection was mediocre (65% at 0.6 MPa, 2000 mg L⁻¹ MgCl₂). Gao et al.²¹ further reinforced the PDA coating by grafting polyethylenimine (PEI) followed with cross-linking. The denser PDA/PEI layer endowed the TFC membrane with excellent rejection (90% at 0.6 MPa, 2000 mg L⁻¹ MgCl₂). Our group proposed the co-deposition of PDA/PEI to construct the active layer for increasing the compactness of pure PDA coating and meanwhile simplifying the time-consuming preparation procedures.¹⁶ The addition of PEI has been reported to catalyze the polymerization of phenols and meanwhile to disintegrate large polyphenol aggregates for coating by Michael addition and Schiff base coupling.^16,22,23 The resultant dense and smooth skin layer renders high salt rejection of the TFC membranes along with enhanced structural stability. However, in term of industrial manufacture, dopamine is undesirable because its rapid aerobic deterioration requires strict storage condition and the high cost restricts large-scale applications. Tannin-like coatings have been derived from (poly)phenols, which also resemble the adhesive properties and compactness of PDA coatings.^24–27 Polyphenols are ubiquitous in plant tissues with huge reserves in nature. Moreover, compared with dopamine, the absence of amine group in their molecular structures decelerates the oxidation polymerization speed and remarkably enhances the controllability and storability.^28,29 Herein, we report a facile approach to construct the active layer on the surface of PSf ultrafiltration membranes via co-deposition of the simplest and low cost polyphenol catechol (CCh) and PEI, followed with further cross-linking by glutaraldehyde (GA). In this way, we successfully prepare a novel kind of high performance TFC NF membrane, which possesses high rejection for multivalent salt ions (ca. 90%) with desirable permeation flux (ca. 25 L m⁻² h⁻¹). Meanwhile, the membranes exhibit excellent operation stability towards long-term filtration test.

Experimental

1. Materials

PSf ultrafiltration membrane with a molecular weight cut off (MWCO) of 50 000 was purchased from Ande Membrane Separation Technology & Engineering Co., Ltd. (Beijing, China). The water permeation flux of this membrane is about 300 L m⁻² h⁻¹ under 0.6 MPa. All samples were cut into rounds with a diameter of 4 cm. Catechol and polyethylenimine (PEI, M_W = 600 Da) were purchased from Aladdin (China). Tris(hydroxymethyl)aminomethane (Tris), glutaraldehyde (GA) solution (50 wt%) and inorganic salts were acquired from Sinopharm Chemical Reagent Co. Ltd, China. Other reagents were of analytical grade and used as received.

2. Preparation of NF membranes

The preparation procedure is illustrated in Fig. 1 for the studied NF membranes. Specifically, catechol and PEI with designed mass ratios were dissolved in Tris buffer solution (pH = 8.5, 10 mM) and the concentration of catechol was fixed at 1.0 mg mL⁻¹. PSf ultrafiltration membranes were pre-wetted thoroughly by ethanol for 30 min and then immersed into the freshly prepared reaction solutions. After incubated at 25 °C for a certain time under mild vibration, the membranes were rinsed with ultrapure water for several times and dried at room temperature. Then, the CCh/PEI co-deposited membranes were further cross-linked by immersing in GA/ethanol solution (2 wt%) for 20 min at 50 °C, followed by post-treatment in an oven at 50 °C for another 20 min. The optimized cross-linking conditions were determined according to the NF performance of the corresponding TFC membranes (Fig. S1 and S2, ESI†). Finally, the prepared NF membranes were washed carefully and stored in DI water for future use.

Fig. 1 Schematic illustration of the preparation procedures and mechanism for the cross-linked TFC NF membranes.

3. Characterization

The CCh/PEI co-deposited membranes or GA cross-linked TFC membranes used for characterization refer to the samples prepared with a CCh/PEI ratio of 4 : 1 for 4 h unless specifically mentioned in the article. FT-IR/ATR spectra were measured with an infrared spectrophotometer (Nicolet 6700, USA) equipped with an ATR accessory (ZnSe crystal, 45°). The spectra were obtained in the region of 4000 to 400 cm⁻¹ with resolution of 4 cm⁻¹ for 64 scans. XPS analyses were performed on an RBD upgraded PHI-5000C ESCA system (Perkin Elmer, USA) with Al Kα radiation (hν = 1486.6 eV). Surface morphologies were observed by a field-emitting scanning electron microscope (FESEM, Hitachi S4800, Japan) after sputtered with a 10–20 nm gold layer on the membrane surface. Meanwhile, the surface topographies were also measured by atomic force microscopy (AFM, MultiMode, Vecco, USA) in the tipping mode. And the root mean square (RMS) roughness was calculated by three dimensional AFM images from each membrane surface. Water contact angles (WCA) were determined with a DropMeter A-200 contact angle system (MAIST Vision Inspection & Measurement Co. Ltd., China). Zeta potentials were measured by an electrokinetic analyzer using an integrated titration unit for solid surface analysis (Anton Paar, SurPASS, Austria) with KCl (1 mmol L⁻¹) solution as electrolyte solution. The size and distribution of CCh/PEI aggregates in the reaction solutions was measured using a Malvern Nano ZS90 laser particle size analyzer (Malvern, Nano ZS, U.K.). Temperature was kept at 25 °C during the whole measurements.

4. NF performance evaluation of the prepared membranes

NF performance was measured with a laboratory scale cross-flow flat membrane module under the constant pressure of 0.6 MPa at 25 ± 1 °C. The effective area for each membrane is about 7.07 cm². Inorganic salts dissolved in DI water (1000 mg L⁻¹) were used as the feed solutions with a similar pH of 6.0 ± 0.2. The membrane samples were firstly pre-compacted at 0.7 MPa for about 1 h to reach a stable flux. Then, the operation pressure was lowered to 0.6 MPa and data were recorded after their values reached a equilibrium state. Water flux (J_w, L m⁻² h⁻¹) and salt rejection (R, %) can be calculated by the following equations:where V, A and t represent the volume of filtrates, the effective membrane area and operation time, respectively.where C_f and C_p represent the concentrations of solute in the feeds and corresponding filtrates. And the concentrations of salt solutions were average values measured with an electrical conductivity meter (METTLER TOLEDO, FE30, China) for three times (Fig. 2).

Fig. 2 FT-IR/ATR spectra of nascent, CCh/PEI co-deposited and GA cross-linked membranes: CCh/PEI = 4 : 1, co-deposition time = 4 h.

Results and discussion

1. Chemical composition and surface morphology of the studied membranes

The bi-phenolic structure of CCh is liable to be oxidized into quinoid form and subsequently reacts with amine groups in aerobic and alkaline environment through Michael addition or Schiff base reaction. Previous work demonstrates that the addition of PEI in CCh/Tris solution will promotes the formation of CCh/PEI co-deposited coatings on various substrates,^23,24 which is also proven to be applicable for PSf ultrafiltration membranes. ATR/FT-IR spectra were used to characterize the chemical composition of the membrane surfaces and the results are shown in Fig. 1. Compared with the nascent PSf membrane, new peak at the region of 1724 cm⁻¹ is ascribed to the C=O stretching vibration of oxidized phenolic groups, while peak at 1660 cm⁻¹ is probably the overlap of C=C resonance vibration in aromatic ring and C=N stretching vibration. Another concerned peak at 1540 cm⁻¹ accounts for the N–H bending vibration, which becomes much weaker after the CCh/PEI co-deposited layer has been further cross-linked by GA because the amine groups have reacted with aldehyde groups to form Schiff bases.

The difference in chemical composition of these membrane surfaces can also be revealed from XPS spectra. Fig. 3 and Table 1 indicate that the pristine PSf membrane possesses a strong peak of O 1s and its atomic percentage is nearly 40%. However, after the co-deposition of CCh/PEI, the O 1s peak distinctly weakens and a new peak arises for N 1s. Meanwhile the N/O ratio is about 1.26, similar to that of CCh/PEI coating deposited on polypropylene membrane.²³ These results demonstrate that PSf membranes are successfully co-deposited with CCh/PEI layer and probably the layer thickness is beyond the detection depth limit of XPS (around 10 nm). Furthermore, the N/O ratio decreases to 1.16 after the CCh/PEI layer has been crosslinked by GA. And the high-resolution N 1s spectra suggest the enhancement of C=N bond, which is the evidence for the successful cross-linking of amine groups by GA through Schiff base mechanism.

Fig. 3 XPS spectra and corresponding high-resolution N 1s spectra of the nascent, CCh/PEI co-deposited and GA cross-linked membrane surfaces: CCh/PEI = 4 : 1, co-deposition time = 4 h.

Table 1 Chemical composition of the nascent, CCh/PEI co-deposited and GA cross-linked membrane surfaces

Samples	C 1s (%)	O 1s (%)	N 1s (%)	S 2p (%)
Nascent	58.57	40.33	15.44	0.18
CCh/PEI co-deposited	75.94	10.59	13.30	0.17
GA cross-linked	75.76	11.08	12.96	0.20

---

Surface morphologies were analyzed by FESEM and AFM for the studied membranes. Fig. 4a shows three-dimensional AFM images for the membrane surfaces. It can be seen that the nascent PSf membrane is relatively smooth with a root mean square (RMS) roughness of 10.332 nm. In comparison, the RMS roughness increases remarkably to 34.857 nm for the CCh/PEI co-deposited membrane, resulting from the adhesion and accumulation of nanosized CCh/PEI aggregates. However, it reduces to 25.204 nm after cross-linking with GA, probably because the cross-linking among the CCh/PEI nanoparticles makes the membrane surface relatively dense. Meanwhile, transformation of the surface morphologies can also be observed from the FESEM images in Fig. 4b. The nascent PSf membrane is dotted with tiny and irregular pores. In contrast, the membrane surfaces show much more compact structures and pores are not observable for those co-deposited by CCh/PEI and then cross-linked by GA.

Fig. 4 AFM (a–c) and SEM images (d–f) of the nascent (a and d), CCh/PEI co-deposited (b and e) and TFC NF (c and f) membrane surfaces: CCh/PEI = 4 : 1, co-deposition time = 4 h.

The cross-section was also characterized to further illuminate the structural changes from the membranes. As shown in Fig. 5, the images are covered with golden masking to discriminate the skin layers. It can be observed that the thickness of the dense skin layer increases remarkably with the co-deposition time, demonstrating the co-deposition of CCh/PEI aggregates is a time-depending accumulation process. It also increases with the CCh/PEI ratio from 1 : 2 to 8 : 1. These results are in accordance with our previous conclusion, indicating that PEI acts as a catalyst for the oxidative polymerization of CCh while excessive PEI will disintegrate the CCh/PEI aggregates and thus suppresses the co-deposition process.²³ In the meantime, nanostructured and papillose CCh/PEI aggregates can be observed on the membrane surfaces (Fig. S3 and S4, ESI†), which suggests the increased thickness of skin layer is indeed derived from CCh/PEI co-deposition.

Fig. 5 Cross-sectional SEM images of the TFC NF membranes with different co-deposition times (a–e, CCh/PEI mass ratio is 4 : 1) and different CCh/PEI mass ratios (f–j, co-deposition time is 4 h).

Previous studies have shown that surface hydrophilization is an effective approach to improve antifouling property for polymer membranes.^30–34 Hydrophilic membrane surface is prone to form a tightly bounded hydration layer, which impedes the adhesion or deposition of organic/biomacromolecular foulants.^35–37 PSf lacks hydrophilic groups and the membrane surface is relatively hydrophobic. Fig. 6a shows that the water contact angle (WCA) varies from 84.9° to 74.5° within 30 s. After CCh/PEI co-deposition, the surface hydrophilicity is significantly improved and WCA value is lower than 40° due to the multiple residual amine and phenolic hydroxyl groups in the co-deposited CCh/PEI aggregates. WCA is slightly increased to around 50° when the co-deposited CCh/PEI layer has been cross-linked by GA, probably because of the declined surface roughness and the decrease of the highly hydrophilic amine groups.

Fig. 6 Water contact angle (a) and zeta potential (b) of the nascent, CCh/PEI co-deposited and GA cross-linked membranes: CCh/PEI = 4 : 1, co-deposition time = 4 h.

Fig. 6b shows the zeta potentials of the membrane surfaces. The results are measured at pH = 6 to accord with the experimental conditions for performance evaluation. The nascent PSf membrane is negatively charged with a surface charge of −25.6 ± 0.2 mV. After CCh/PEI co-deposition, the value increases to −0.80 ± 0.4 mV due to the protonation of residual amine groups on the membrane surface. GA cross-linking further transfers the amine groups into Schiff bases, which slightly lowers the surface potential to −2.6 ± 0.1 mV. These surface charges are beneficial to improve the rejection of inorganic salt by Donnan exclusive effect for the prepared NF membranes.^38,39 Meanwhile, the charge capacity is relatively small so as to alleviate the adsorption of charged pollutants in the feed solutions.

2. NF performance of the TFC membranes

NF performance was evaluated by a cross-flow mode for the cross-linked TFC membranes. The CCh/PEI mass ratio and the co-deposition time are two factors that play significant roles in the surface/cross-section morphologies and so that on the NF performance of the cross-linked TFC membranes. The effect of CCh/PEI ratio was investigated with the co-deposition time fixed at 4 h. Fig. 7a shows that the flux decreases from 50.1 L m⁻² h⁻¹ to 18.2 L m⁻² h⁻¹ with the PEI concentration in the reaction solution. This result is in accordance with the thickness of the skin layers in corresponding membranes as shown in Fig. 4. And the rejection increases from 21.90% to 87.45% when the CCh/PEI ratios change from 1 : 2 to 4 : 1. However, the membrane co-deposited with a CCh/PEI ratio of 8 : 1 shows a rejection inversely decreased to 77.24%. This phenomenon can be rationalized by the fact that polyphenols is prone to form large aggregates with low content of PEI so that the structure of the stacked coating is relatively loose (Fig. S5, ESI†), which reduces the compactness of the active layer.

Fig. 7 Influence of CCh/PEI mass ratio (a) and co-deposition time (b) on the NF performance of the TFC NF membranes. Test conditions: MgCl₂ concentration = 1000 mg L⁻¹, pH = 6.0, 25 °C, 0.6 MPa, cross-flow rate = 30 L h⁻¹.

The influence of co-deposition time was also investigated and the CCh/PEI mass ratio was fixed at 4 : 1. As shown in Fig. 7b, water flux decreases from 42.23 L m⁻² h⁻¹ to 18.45 L m⁻² h⁻¹ with the time from 1 h to 5 h. This phenomenon is mainly resulted from the thickening of the active layer with time. Meanwhile, the rejection of MgCl₂ increases continuously from 37.44% to 87.05% before 4 h and seemingly reaches a stable state afterwards. The results demonstrate 4 h is adequate to form a dense layer and the MgCl₂ rejection of the CCh/PEI co-deposited membrane reaches approximately 75% even without cross-linking (Fig. S1, ESI†). Therefore, 4 h was selected as the co-deposition time to prepare TFC membranes with the optimum performance for the following experiments.

Effects of different ions were also investigated with five inorganic salts on the NF performance of the optimized TFC membranes. Fig. 8 shows the membrane possesses a better rejection of SO₄²⁻ than Cl⁻, the rejection of MgSO₄ and Na₂SO₄ is up to 91.3% and 88.4%, respectively. This result can be explained by Donnan exclusive effect that the negatively charged NF membranes exert stronger repulsive interaction with polyvalent anions than monovalent ones. On the other hand, as the zeta potential is not so high at the test pH of 6.0 ± 0.2, the steric hindrance also has a great impact on the rejection. The hydration ion radius of cations obeys the following order: Mg²⁺ (0.345 nm) > Ca²⁺ (0.307 nm) > Na⁺ (0.183 nm).⁴⁰ As a result, the rejection of MgSO₄ is higher than Na₂SO₄. Meanwhile, the rejection of MgCl₂ reaches 87.4%, higher than CaCl₂ of 79.9% and NaCl of 41.5%. And it should be mentioned that the rejections of multivalent ions for the CCh/PEI co-deposited membranes are similar to results of PDA/PEI prepared ones (around 90%), while the permeation flux is more that twice as much.¹⁶

Fig. 8 NF performance of the cross-linked TFC NF membranes (CCh/PEI = 4 : 1, co-deposition time = 4 h) with different inorganic salts. Test conditions: salt concentration = 1000 mg L⁻¹, pH = 6.0, 25 °C, 0.6 MPa, cross-flow rate = 30 L h⁻¹.

The stability of the TFC membranes was evaluated via a long-term operation experiment. The optimized cross-linked TFC membranes were employed to filtrate MgCl₂ solution (1000 mg L⁻¹) continuously for 240 h. As depicted in Fig. 9, during the filtration process, the membrane retains a rejection of around 87.0% and flux of around 26.0 L m⁻² h⁻¹ with minor fluctuation. The consistent NF performance reflects the stability of the co-deposited skin layer, which is probably originated from the molecular rigidity of stacked aromatic rings within polyphenol aggregates and the highly cross-linked structures. The operational stability of the membranes will prolong its service time toward practical applications. Meanwhile, the studied membranes also show excellent structural stability after alcohol treatment for 7 days, which may reflects the good adhesion between the skin layer and the PSf substrate.⁴¹

Fig. 9 Operational stability of the cross-linked TFC NF membrane (CCh/PEI = 4 : 1, co-deposition time = 4 h) with different operation time. Test conditions: MgCl₂ concentration = 1000 mg L⁻¹, pH = 6.0, 25 °C, 0.6 MPa, cross-flow rate = 30 L h⁻¹.

Conclusions

Novel TFC NF membranes are successfully fabricated via the co-deposition of catechol and PEI on the surface of PSf ultrafiltration membranes, followed by the cross-linking with GA. The constructed skin layer is dense and detect-free, endowing the optimized membranes with high rejection of multivalent ions (ca. 90%) with acceptable permeation flux (ca. 25 L m⁻² h⁻¹) under 0.6 MPa, which is suitable for desalination or water softening. NF performance of the membranes can also be facilely tuned by controlling the cross-linking condition, co-deposition time and the CCh/PEI ratio. Meanwhile, the membrane possesses excellent operational stability over 240 h filtration process, proving the structural stability of the prepared membrane.

Acknowledgements

Financial support is acknowledged to the National Natural Science Foundation of China (Grant no. 21534009).

Notes and references

1. S. Cheng, D. L. Oatley, P. M. Williams and C. J. Wright, Adv. Colloid Interface Sci., 2011, 164, 12–20.
2. K. Yoon, B. S. Hsiao and B. Chu, J. Membr. Sci., 2009, 326, 484–492.
3. J. Lv, K. Y. Wang and T.-S. Chung, J. Membr. Sci., 2008, 310, 557–566.
4. R. Rautenbach and A. Gröschl, Desalination, 1990, 77, 73–84.
5. A. Rahimpour, M. Jahanshahi, N. Mortazavian, S. S. Madaeni and Y. Mansourpanah, Appl. Surf. Sci., 2010, 256, 1657–1663.
6. P. Xu, C. Bellona and J. E. Drewes, J. Membr. Sci., 2010, 353, 111–121.
7. M. Alcaina-Miranda, S. Barredo-Damas, A. Bes-Pia, M. Iborra-Clar, A. Iborra-Clar and J. Mendoza-Roca, Desalination, 2009, 240, 290–297.
8. T. Gotoh, H. Iguchi and K.-I. Kikuchi, Biochem. Eng. J., 2004, 19, 165–170.
9. S. Baowei, T. Wang, Z. Wang, X. Gao and C. Gao, J. Membr. Sci., 2012, 423, 324–331.
10. I. C. Kim, H. G. Yoon and K. H. Lee, J. Appl. Polym. Sci., 2002, 84, 1300–1307.
11. C. Qiu, F. Xu, Q. T. Nguyen and Z. Ping, J. Membr. Sci., 2005, 255, 107–115.
12. Y. Mansourpanah, S. Madaeni and A. Rahimpour, J. Membr. Sci., 2009, 343, 219–228.
13. B. Tang, Z. Huo and P. Wu, J. Membr. Sci., 2008, 320, 198–205.
14. L. Lianchao, W. Baoguo, T. Huimin, C. Tianlu and X. Jiping, J. Membr. Sci., 2006, 269, 84–93.
15. V. Freger, Langmuir, 2003, 19, 4791–4797.
16. Y. Lv, H.-C. Yang, H.-Q. Liang, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2015, 476, 50–58.
17. H. Lee, S. M. Dellatore, W. M. Miller and P. B. Messersmith, Science, 2007, 318, 426–430.
18. Q. Ye, F. Zhou and W. Liu, Chem. Soc. Rev., 2011, 40, 4244–4258.
19. Y. Liu, K. Ai and L. Lu, Chem. Rev., 2014, 114, 5057–5115.
20. X.-L. Li, J.-H. Jiang, Z. Yi, B.-K. Zhu and Y.-Y. Xu, Chin. J. Polym. Sci., 2012, 30, 152–163.
21. M. Li, J. Xu, C.-Y. Chang, C. Feng, L. Zhang, Y. Tang and C. Gao, J. Membr. Sci., 2014, 459, 62–71.
22. H.-C. Yang, K.-J. Liao, H. Huang, Q.-Y. Wu, L.-S. Wan and Z.-K. Xu, J. Mater. Chem. A, 2014, 2, 10225–10230.
23. W.-Z. Qiu, H.-C. Yang, L.-S. Wan and Z.-K. Xu, J. Mater. Chem. A, 2015, 3, 14438–14444.
24. H. Wang, J. Wu, C. Cai, J. Guo, H. Fan, C. Zhu, H. Dong, N. Zhao and J. Xu, ACS Appl. Mater. Interfaces, 2014, 6, 5602–5608.
25. T. S. Sileika, D. G. Barrett, R. Zhang, K. H. A. Lau and P. B. Messersmith, Angew. Chem., Int. Ed., 2013, 52, 10766–10770.
26. D. G. Barrett, T. S. Sileika and P. B. Messersmith, Chem. Commun., 2014, 50, 7265–7268.
27. J.-R. Jeon, J.-H. Kim and Y.-S. Chang, J. Mater. Chem. B, 2013, 1, 6501–6509.
28. N. Aktaş and A. Tanyolaç, Bioresour. Technol., 2003, 87, 209–214.
29. N. Aktaş and A. Tanyolaç, J. Mol. Catal. B: Enzym., 2003, 22, 61–69.
30. Y.-F. Yang, Y. Li, Q.-L. Li, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2010, 362, 255–264.
31. Y.-F. Yang, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2009, 337, 70–80.
32. H.-Y. Yu, Y.-J. Xie, M.-X. Hu, J.-L. Wang, S.-Y. Wang and Z.-K. Xu, J. Membr. Sci., 2005, 254, 219–227.
33. Y.-F. Yang, L.-S. Wan and Z.-K. Xu, J. Adhes. Sci. Technol., 2011, 25, 245–260.
34. Y.-F. Yang, L.-S. Wan and Z.-K. Xu, J. Membr. Sci., 2009, 326, 372–381.
35. J. Xu, X. Feng, P. Chen and C. Gao, J. Membr. Sci., 2012, 413, 62–69.
36. J. Peng, Y. Su, W. Chen, X. Zhao, Z. Jiang, Y. Dong, Y. Zhang, J. Liu and X. Fan, Ind. Eng. Chem. Res., 2013, 52, 13137–13145.
37. S. Jiang and Z. Cao, Adv. Mater., 2010, 22, 920–932.
38. R. Levenstein, D. Hasson and R. Semiat, J. Membr. Sci., 1996, 116, 77–92.
39. J. Luo and Y. Wan, J. Membr. Sci., 2013, 438, 18–28.
40. D.-X. Wang, M. Su, Z.-Y. Yu, X.-L. Wang, M. Ando and T. Shintani, Desalination, 2005, 175, 219–225.
41. Y. Li, Y. Su, J. Li, X. Zhao, R. Zhang, X. Fan, J. Zhu, Y. Ma, Y. Liu and Z. Jiang, J. Membr. Sci., 2015, 476, 10–19.

---

Footnote

† Electronic supplementary information (ESI) available: Influence of cross-linking time on the NF performance of cross-linked composite membranes (Fig. S1); influence of and post-treatment time on the NF performance of cross-linked composite membranes (Fig. S2); surface SEM images of cross-linked composite membranes with different co-deposition time (CCh/PEI = 4 : 1) (Fig. S3); surface SEM images of cross-linked composite membranes with CCh/PEI mass ratios (deposition time = 4 h) (Fig. S4); the size distribution of CCh/PEI aggregates with time in various solutions of different CCh/PEI mass ratios (Fig. S5); effect of alcohol treatment on the performance of the TFC membranes (Fig. S6). See DOI: 10.1039/c6ra04074h

---