Preparation of a novel zwitterionic striped surface thin-film composite nanofiltration membrane with excellent salt separation performance and antifouling property

Thin-film composite (TFC) nanofiltration (NF) membranes were fabricated via the co-deposition of taurine, tannic acid (TA), and polyethyleneimine (PEI), followed by subsequent interfacial polymerization with trimesoyl chloride (TMC) on the surface of the polysulfone ultrafiltration substrates. The surface properties, including the roughness, hydrophilicity, surface potential, and NF performances were facilely tuned by varying the taurine content for the prepared TFC membranes. In addition, the as-prepared TFC NF membranes had an excellent antifouling property and flux recovery ratio (FRR) in humic acid (HA), bovine serum albumin (BSA) and sodium alginate (SA) filtration tests. These results also revealed that the taurine content controlled the formation of the striped surface. Thus, this work provided a viable strategy for fabricating TFC NF membranes with high selectivity and outstanding antifouling ability.


Introduction
Nanoltration (NF) is a pressure-driven membrane separation technology with a molecular weight cut-off (MWCO) ranging from 200 to 1000 Da and a pore size of around 0.5-2.0 nm, which was conrmed as an effective technique for desalination and could also be used in liquid purication, sewage, and industrial wastewater treatments because of its low-pressure requirement. [1][2][3][4] However, membrane fouling is now the bottleneck problem in membrane applications since the practical application conditions and separation systems of the membranes are far more complicated. 5 For wastewater treatment, the membrane foulants mainly arise from macromolecular organic matter, humic substances, hydrocarbons, bacteria, suspended sludge and inorganic colloidal or mineral scaling, which inevitably lead to the unfavorable change of membrane structures and signicant deterioration of the separation performance. 6,7 Therefore, the construction of antifouling nanoltration has a big signicance in membrane applications.
In an antifouling membrane, advanced materials (such as silica spheres, 8 nanoparticles, [9][10][11][12] hydrophilic polymers, [13][14][15] metallics 16,17 as well as some functional carbon materials 18,19 ) play an important role. Recently, polyphenol tannic acid (TA) has been widely used in membrane fabrication because of its abundance in plant tissue 20 and versatility for membrane synthesis. Tannic acid has anti-bacterial, anti-enzymatic and astringent properties. 21 With these properties, it is an ideal molecule for improving the antifouling performance. Kim et al. 22 coated TA-Fe III coordination complexes onto PES UF membrane surfaces via a facile one-pot assembly process to obtain desirable features, such as antifouling properties against proteins, oils, and microorganisms, as well as antimicrobial and heavy metal ion removal properties. Zhang et al. 23 developed a facile TA-assisted in situ assembly approach during nonsolvent induced phase separation (NIPS) to prepare antifouling NF membranes. Xu et al. 24 reported on the addition of TA to a PVDF membrane for oil/water emulsion ltration to avoid membrane over-fouling. Aside from tannic acid, zwitterions have attracted much attention in the study of antifouling membranes owing to their high intrinsic hydrophilicity, which forms a hydration layer through the electrostatic interaction between the zwitterions and water molecules. 25 This hydration layer provides a repulsive steric barrier to prevent the adsorption of organic molecules and bacteria on the substrate surface. 25 Wang et al. 26 introduced a zwitterionic copolymer carrying a positive charge into a negatively charged membrane via simple electrostatic adsorption in a one-step dip-coating process. During ltration with lysozyme for 1 h, the ux of this modied membrane was 93% of the initial ux, which was much higher than that of the original membrane. Davari et al. 27 tethered the zwitterionic polyelectrolyte poly[1-vinyl-3(2carboxyethyl)imidazolium betaine] (PVCIB) onto a commercial thin lm composite polyamide (TFC PA) membrane, which exhibited remarkable antifouling ability to resist non-specic protein adsorption at neutral and alkaline pH values. However, both PA-PVI and PA-PVCIB membranes exhibited high resistance to the positively charged lysozyme adhesion under acidic pH conditions. In addition, the surface zwitterionization can be constructed on the membrane surface rather than adding a zwitterionic copolymer. Researchers used 1,3-propanesultone to create a zwitterionic surface by ring-opening directly on the nanoltration membrane prepared from the interfacial polymerization of 3,3 0 -diamino-N-methyldipropylamine and 1,3,5benzenetricarboxylic chloride (TMC). 28 The zwitterionic N,Nbis(3-aminopropyl)methylamine was synthesized by Mi et al. to prepare an antifouling NF membrane with a 94.9% ux recovery ratio in BSA and SA solutions. 29 The zwitterionic structures improved the hydrophilicity of the nanoltration membrane, thus increasing the water permeability and antifouling performance to proteins and polysaccharides. Zwitterions and tannic acid have been widely applied in the preparation of antifouling membranes, while the effect of their combination still remains unknown.
Striped structures in the nanoltration membrane were rst illustrated by Tan et al. in 2018. 30 The nanoscale spotted and striped Turing structures were generated by controlling the concentration of the PVA addition, which changes the diffusion rate of piperazine from the aqueous phase to the organic phase by hydrogen bond. 30,31 In our study, taurine (3-amino-1-propanesulfonic acid) and polyethyleneimine were poured on the polysulfone (PSF) substrate. Taurine was introduced into the membrane to improve the antifouling performance, and tannic acid acted as a cross-linking agent to lock the zwitterion onto the membrane surface with catechol groups. 32 Then, this chemical-stable antibiofouling compound and polyethyleneimine (PEI) were used to build a hydrophilic zwitterionic striped structure membrane by interfacial polymerization with TMC. We rst used the strategy in which we attached the hydrophilic material by chemical bond to build the striped structure, which obviously improved the membrane hydrophilicity. The inuence of the molecule weight of PEI, the concentration of taurine, as well as the interfacial polymerization conditions on the membrane morphology, separation performance and anti-fouling property was explored. The as-prepared membranes were characterized by FTIR, SEM, EDX, XPS, AFM, contact angle goniometer and zeta potential measurement.

Fabrication of TFC NF membranes
TFC NF membranes were fabricated through a rapid codeposition of PEI/TA/taurine onto the top surface of a PSF ultraltration (UF) substrate, followed by interfacial polymerization (IP) with TMC ( Fig. S1 †). The detailed fabrication process is illustrated in the ESI (S1). † The synthesized membranes are denoted as "M a ", where "a" is the weight concentration of taurine (example: 0.1% wt is written as 1). M b was the membrane without crosslinking by TMC (Table 1).

Characterization of the TFC NF membranes
All samples were dried in a vacuum oven at 40 C for 12 h before characterization.
The functional structure of the TFC membrane surface was examined by attenuated total reection Fourier transform infrared spectroscopy (ATR-FTIR, Nicolet 6700). For ATR-FTIR analysis of the membrane samples, the resulting NF membranes were characterized by the total reection of a zinc selenide crystal. The chemical compositions of the NF membranes were measured by X-ray photoelectron spectroscopy (XPS, Kratos AXIS Ultra DLD). The morphologies of the NF membranes surface were observed by scanning electron microscopy (SEM, Hitachi S-4700). Then, the surface of the membranes was scanned aer coating with gold. The NF membrane surface roughness was obtained by atomic force microscopy (AFM, XE-100) with 5 mm Â 5 mm scanning range. RMS means the root mean square roughness, and R a means the arithmetic mean roughness. The water contact angle was measured on a dried membrane surface by a contact angle analyzer (OCA-20, Germany) at ambient temperature. Aer each drop, the image was captured and the reading was recorded. For each sample, at least 12 readings were obtained, and the average contact angle values were calculated from these readings. Zeta potentials of the membrane samples were measured by a SurPASS analyzer (Anton Paar, Austria). The test reagent used for our study was 0.001 mol L À1 KCl, and the pH value was varied to investigate its effect on the zeta potential. 33

Membrane performance evaluation
The membrane performance was tested using a laboratory-scale cross-ow apparatus with an effective area of 12.6 cm 2 at 0.5 MPa and 25 AE 1 C. Before the test, each membrane was pressurized with DI water at 0.75 MPa and 25 AE 1 C for 30 min to reach a steady state. Various salts (Na 2 SO 4 , NaCl, MgSO 4, and MgCl 2 ) in aqueous solutions with a concentration of 1 g L À1 were used as the feed solution to evaluate the separation performance of the membrane. The ux (J, L m À2 h À1 ) and rejection (R, %) were calculated according to eqn (1) and (2), respectively: where DV (L), A (m 2 ) and DT (h) correspondingly represent the volume of the permeate, the effective membrane area and the time for permeate collection.
where C p (mol L À1 ) and C f (mol L À1 ) refer to the concentration of permeate and feed solution, respectively.

Antifouling property of the membrane
The antifouling property of the membrane was investigated by using the organic foulant (BSA, SA, and HA) solution (500 mg L À1 ) in the same cross-ow apparatus at 25 AE 1 C. 34 The membranes were compacted with DI water at 0.5 MPa and 25 AE 1 C for 30 min to achieve a stable state, and the pure water ux of the original membrane (J 0 ) was measured for 3 h at 0.5 MPa. Then, the fouling experiment was carried out with 100 mg L À1 BSA, SA or HA aqueous solution as the feed solution for another 10 h, and the steady value of the ux was noted as J 1 .
The normalized ux J 1 /J 0 refers to the ux declining ratio during the fouling process. Aer that, the fouled membrane was thoroughly ushed by DI water for 30 min at 0.05 MPa. Finally, the pure water ux (J 2 ) of the cleaned membrane was measured for 3 h at 0.5 MPa. The above procedure was repeated 2 times to evaluate the ux recovery ratio (FRR), which was determined by eqn (3).
3 Results and discussion

Characterization of TFC NF membranes
3.1.1 Physical and chemical structures of the TFC membranes. As shown in Fig. 2, the chemical components of the membrane surfaces were characterized by ATR-FTIR. The specic absorption peak at 1039 cm À1 was denoted as the SO 3 À groups that were observed in M 2 because of the introduction of taurine. 35 The specic absorption peaks at 1658 cm À1 and 1530 cm À1 were attributed to the existence of the formation of -C]O (amide I and amide II). 36,37 With the increase in the concentration of taurine, the intensity of this peak was decreased. The band appearing in the M b and M 2 membrane spectra at 1645 cm À1 was assigned to the formation of the -C] Fig. 1 The mechanism of interface polymerization and zwitterionic striped structure formation. N bond. 38,39 The region from 2800 cm À1 to 2900 cm À1 in the blue circle was assigned to the specic absorption peak of the primary and secondary amines. 40 The absorption bands of -OH and N-H were assigned at around 3429 cm À1 in the green circle. 41 Due to the crosslinking by TMC, the broad peak became weaker in M 0 and M 2 . M 0 and M 2 showed a newly emerged broad peak at around 1725 cm À1 owing to the stretching vibrations of the ester groups (-COO) provided by the crosslinking of TA and TMC. 42 Furthermore, the XPS spectrum is shown in Fig. 3. When taurine was introduced, the intensity of the S 2s and S 2p peaks increased, which conrmed that the taurine was locked into the membrane surface. In addition, the XPS C 1s and N 1s spectra in Fig. 4 were further used to prove that the -C]N bond was formed during the membrane immersion in the aqueous phase, 39 and the C-S bond was conrmed to be introduced by comparing the species of M 0 and M 2 . Table 2 displays the elemental information of various TFC membranes, which was calculated from the corresponding Xray photoelectron spectroscopy (XPS) spectra in Fig. 3. The ratio of C/O and C/N increased from 3.50 to 3.74 as the amounts of taurine varied from 0 to 0.3%, which was caused by the introduction of the C and N atoms from taurine. The C/O and C/ N declined when the taurine content reached 0.4%, which was caused by the reaction of free taurine and TMC. The remaining taurine that did not attach to TA reacted with TMC directly to decrease the surface degree of crosslinking. The change of the C/O and C/N ratio in M 4 gave evidence for the reaction of taurine and TMC. Therefore, the results from ATR-FTIR and XPS demonstrated that the prepared membrane had a zwitterionic surface and the reaction process in IP.
3.1.2 Surface and cross-section morphologies of the membranes. The top surface and cross-section morphologies of each membrane were observed under SEM. From Fig. 5, the nanostrand hybrid morphology appeared in panels (e) and (g). The surface morphology was changed by the contents of the additional taurine. When 0.2% or 0.3% taurine was added, the roughness and surface area of the striped structure increased (as shown in Table 3). 30 The membrane surface property and hydrophilicity, in particular, were improved with increasing taurine content. Also, the reaction of PEI, TA, and taurine helped the hydrophilic groups attach to PEI, which hindered the further diffusion of the PEI monomer and efficiently reduced the polyamide thickness. 43 These hydrophilic groups also absorbed water to form a water layer to change the surface tension between the aqueous phase and organic phase in the IP reaction to create the striped structure 44 (Fig. 1). Fig. 1 illustrates that the sulfonic acid group was much more hydrophilic, which caused the PEI to remain at the aqueous phase for as long as possible, and increased the surface tension. So, the striped structure appeared. It is important to highlight that the striped structure disappeared when a larger amount of the taurine was added (0.4%). The IP reaction takes place in the organic phase, 45 and the micromolecules were easier to diffuse to the organic phase from the aqueous phase than the macromolecules. 30 Thus, the redundant taurine molecules (not combined with TA) were easily diffused to the organic phase to immediately react with TMC, and the reaction between PEI and TMC was further decreased. Moreover, the cross-section structure showed that  This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 16168-16178 | 16171 the degree of crosslinking was decreased by the increased taurine content. 46 Thus, a loose surface morphology could enhance the permeability of the water molecule. 47 The roughness of the membranes is depicted by AFM in Table 3. The R a and RMS of M 2 were 4.61 nm and 5.97 nm, respectively. Aer the taurine was introduced, much rougher surfaces with more nodular structures were observed for membranes M 1 through M 4 in comparison with M 0 . The R a values of membranes M 1 through M 4 were 3.54 nm, 4.61 nm, 5.64 nm, and 7.17 nm, respectively. These results proved the conclusions from the previously discussed SEM images that the formation of the striped structures increased the surface roughness. Therefore, the surface morphology and roughness, which are related to the permeability and anti-fouling property of the membrane, were dominated by the introduction of taurine.
3.1.3 Water contact angle and surface charge of the membranes. The water contact angles of the TFC membranes are shown in Fig. 6. It is intuitive that the water contact angle of the membrane was decreased by the increase of the taurine content. Amine groups and sulfonic acid groups have a strong affinity to water, which increased the surface hydrophilicity of the membrane. 48,49 Thus, the change of the water contact angle demonstrated that the introduction of taurine improved the surface hydrophilicity, corresponding to the result in Table 3. In addition, the surface charge of the membranes was evaluated by the zeta potential test. The result is displayed in Fig. 7. All membrane surfaces were found to be positively charged when the pH value was at 6.5 (test value) because the polymerization between PEI and TMC led to a polyamide thin-lm with a large number of unreacted positive primary and secondary amine groups. 50 When the concentration of taurine increased from 0 to 0.3%, the isoelectric point was shied to a higher pH value. One reason is that the increased taurine replaced some amino groups of PEI to react with TA and TMC to release more amino groups of PEI. Meanwhile, taurine itself brought in amino groups. Nevertheless, when the taurine content came to 0.4%, the isoelectric point was shied to the le. The explanation for this result was that the high concentration of the taurine strongly impeded the reaction between TA and PEI or TMC and PEI (Fig. 1). These unreacted PEIs were washed by the water, causing the amino groups in the membrane surface to decrease. Although the surface had become more positively charged with the introduction of taurine, the sulfonic acid groups still provided negative charges in the membrane to form a zwitterionic surface that could not be evaluated by the surface zeta potential test. 51

Membrane performance
The fabrication conditions were evaluated during the preparation process, and the inuence of these conditions on membrane performance are described below.
3.2.1 Concentration of taurine. The effect of the taurine content on the resulting membrane separation performance are presented in Fig. S2(a). † The reaction time in the aqueous phase and TMC organic phase was 3 min and 60 s, respectively, and the PEI molecular weight was 70 000, as listed in Table 1. When the taurine content varied from 0 to 0.4%, the rejection of Na 2 SO 4 and NaCl increased in the rst stage, and subsequently  decreased. Furthermore, when the taurine content was at 0.2%, the separation performance of Na 2 SO 4 and NaCl was superior and kept a relatively high MgCl 2 rejection. The membrane containing 0.2% taurine with a striped surface also had a larger water contact area, 30 which had a considerable water ux. The NF membrane is considered as a charged porous lm, and there are two main factors that affect the rejection of the NF membrane that coincides with the Donnan exclusion theory and steric hindrance effect: the surface charge plays a crucial role. 52 With the addition of taurine, the surface property of the membrane was changed. The high Na 2 SO 4 rejection and the reduction of MgCl 2 rejection can be rationalized by the Donnan effect and size exclusion effect, as there was a charge shielding effect in the charged membrane surface and the hydration radii of Na + and Cl À are too small to be rejected. 53 The sulfonic acid groups were taken into the membrane with the unreacted amino groups of PEI to construct a zwitterionic surface, which altered the surface potential. The taurine content increased, enhancing the negative potential, so the rejection of Na 2 SO 4 rst increased. On the other hand, the taurine content dominated the surface menology. When the taurine was introduced, the taurine that did not combine with TA easily reacted with TMC to decrease the degree of surface crosslinking to form a thinner PA layer. This is why the rejection of Na 2 SO 4 , MgCl 2 , and NaCl was reduced when the taurine content increased to more than 0.2%, and the ux of the membrane rose as a result of the increased taurine content. Thus, the membrane with 0.2% taurine exhibited relatively optimistic separation performance and water permeability. 3.2.2 PEI molecular weight. The inuence of the PEI molecular weight is presented in Fig. S2(b). † The reaction times in the aqueous phase and TMC organic phase were 3 min and 60 s, respectively, and the taurine content was 0.2%. When the molecular weight increased from 600 to 2500, the rejection of Na 2 SO 4 , MgCl 2 and NaCl increased. When the molecular weight was further enlarged, the rejection of NaCl was decreased. The water ux also declined with the increased molecular weight of PEI. The reason may be that the selective layer became more dense with the increased molecular weight of PEI. 39 In addition, a PEI molecule with a larger molecular weight has more amino groups to provide more positive charges for a high divalent cation rejection. 52 In addition, the larger molecular weight (10 000 and 70 000) PEI was harder to introduce to the organic phase to react with TMC, but provided more amino groups, which explained the stable Na 2 SO 4 rejection and rapid decline of the NaCl rejection.
3.2.3 Co-deposition time. The role of the co-deposition time in membrane performance is displayed in Fig. S2(c). † The reaction time in the TMC organic phase was 60 s. The taurine content was 0.2% and the PEI molecular weight was 70 000. When the time varied from 1 min to 3 min, the water permeability was improved and the rejection of Na 2 SO 4 , MgCl 2 , and NaCl was increased. That may be ascribed to the deposition of more hydrophilic monomers, especially the amino groups of PEI, in the aqueous phase. Aer extending the deposition time from 3 min to 9 min, the water permeability of the membrane dropped from 57.58 L m À2 h À1 to 43.82 L m À2 h À1 . The explanation for this phenomenon was that much more hydroxyl groups of TA reacted with the amino groups to form a crosslinking structure, which weakened the surface permeability of the membrane. The rejection of Na 2 SO 4 and NaCl showed no great changes, while the rejection of MgCl 2 was decreased since the rejection of salt in the TFC membranes was mainly dominated by the surface charge rather than the surface structure, 39 and the amino groups do not have much impact on the divalent rejection.
3.2.4 IP reaction time. The effect of the IP reaction time is illustrated in Fig. S2(d). † The reaction time in the aqueous phase was 60 s. The taurine content was 0.2% and the PEI molecular weight was 70 000. When the reaction time was 3 min, the Na 2 SO 4 rejection was >97%, as the reaction rate of the interface polymerization was very fast. 54 When the reaction  time rose from 30 s to 60 s, the rejection of Na 2 SO 4 , MgCl 2 and NaCl increased rapidly, while the water ux decreased rapidly. When the reaction time increased from 60 s to 150 s, the Na 2 SO 4 rejection became steady while that of MgCl 2 decreased, and the NaCl rejection still increased rapidly and the declination tendency of the water ux slowed down. That can be ascribed to the increase of the degree of crosslinking with the expanding reaction time. Therefore, the best condition for the membrane preparation was when the reaction times in the aqueous phase and TMC organic phase were 3 min and 60 s, respectively, the taurine content was 0.2% and the PEI molecular weight was 70 000, which were the preparation conditions of M 2 .
3.2.5 The membrane performance of M 0 and M 2 in four different salt solutions. Fig. 8 displays the membrane performance of M 0 and M 2 in four different salt solutions at 1000 ppm. For M 2 , the rejections of Na 2 SO 4 , MgSO 4 , MgCl 2 and NaCl were 97.55%, 94.72%, 90.40%, and 36.36%, respectively. For M 0 , the rejections of Na 2 SO 4 , MgSO 4 , MgCl 2, and NaCl were 94.29%, 92.00%, 93.32%, and 43.70%, respectively. The rejection mechanism of the nanoltration membranes was mainly determined by the molecular size (sieve effect) and the charge effect. 55 The M 2 membrane with the zwitterionic surface had a special charge distribution for divalent ion rejection. The sulfonic acid group helped to increase the rejection of SO 4 2À and maintain the relatively high rejection of Mg 2+ . In addition, the water ux of M 2 in these four salt solutions was 49.53 L m À2 h À1 , 52.97 L m À2 h À1 , 55.3 L m À2 h À1 , and 58.16 L m À2 h À1 , respectively. The striped surface increased the surface water contact area to result in a relatively higher water ux. This result showed that M 2 had a better salt separation performance and water permeability than M 0 .

Antifouling performance of the TFC membrane
Generally, nanoltration membranes easily suffer from membrane fouling in applications, which restricts the lifetime of the membrane. Factors like the electrostatic interaction, hydrogen bonding effect, hydrophobic impact, and van der Waals forces normally have a great impact on membrane fouling. 56 A comparison of the antifouling performances of M 2 and M 0 is shown in Fig. 9. Fig. 9(a)-(c) display the antifouling performances of M 2 and M 0 at pH ¼ 7 in HA, SA and BSA solutions of  This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 16168-16178 | 16175 500 ppm, respectively. When the feed solution was replaced by a model foulant solution, the water ux of both membranes declined due to the extra osmotic resistance caused by the deposition of foulants on the membrane surface. Fig. 9(a) shows that the ux of M 0 decreased rapidly and reached a quasi-steady stage of 85% to the initial water ux, while M 2 showed a lower ux reduction ratio (9%) in the HA solution. Aer being cleaned with DI water, the relative water FRRs of M 0 and M 2 were 91.5% and 97.8%, respectively. These ux variations during the antifouling test indicate that the M 2 membrane had an excellent humic acid antifouling performance. In addition, Fig. 9(b) shows the antifouling performances of M 0 and M 2 in a 500 ppm SA solution. The ux of M 0 decreased rapidly and reached a quasi-steady stage of 82% to the initial water ux and the ux reduction ratio was 10% aer being cleaned by DI water, while M 2 showed a relatively lower ux reduction ratio (7%).
Moreover, the antifouling performances of M 2 and M 0 in the BSA solution are shown in Fig. 9(c). The permeation ux of M 2 and M 0 experienced a marked decrease in 78% and 67%, respectively, demonstrating severe BSA membrane fouling. Aer being cleaned with DI water, the water permeabilities of these membranes recovered to 90% and 82%, which were much lower than the original pure water ux. The lower recoveries indicated that these membranes had a relatively low ux recovery and reversible fouling. These results were primarily ascribed to the introduction of the sulfonic acid groups on the membrane surface. As we all known, the BSA, HA and SA foulants used in this experiment were all negatively charged under the neutral ltration condition (pH ¼ 7). 12 The introduction of the sulfonic acid groups provided more surface negative charges, which hindered the foulant molecules from contacting the membrane surfaces via electrostatic interactions. Meanwhile, the sulfonic acid groups increased the surface hydrophilicity and helped absorb water molecules. These changes enabled the formation of a hydration layer that prevented the adsorption of organic foulants on the membrane surface due to the higher maximum Gibbs free energy than the unmodied surface. 57,58 Although the roughness of M 2 was greater than M 0 owing to it theoretically being embedded with more contaminants, the surface hydration layer was dense enough to cover the surface gap and help eliminate the inuence of the membrane surface roughness. In addition, the zwitterionic surface provides some electrostatic resistance to these charged foulants. Thus, the M 2 membrane showed better antifouling property than M 0 . Four parameters (FRR, R t , R r , and R ir ) were utilized to evaluate the antifouling performance of M 2 as summarized in Fig. 9(d). The FRR in the HA, BSA, and SA ltration test was 97.8%, 89.9%, and 93.1%, respectively. These results demonstrated that M 2 showed remarkable antifouling properties in the HA solution, while having relatively lower FRR in the BSA and SA solutions. M 2 also had a relatively better antifouling property in the HA solution. The explanation for this was that the BSA and SA solutions had relatively higher molecular weights and stronger viscidity values than HA. Therefore, the prepared membrane M 2 demonstrated excellent performance in the membrane fouling evaluation.

Conclusion
In summary, novel TFC NF membranes with zwitterionic striped surfaces were fabricated via the co-deposition of taurine, TA, and PEI onto the PSF ultraltration membrane, followed by IP with TMC. The skin layer of the membranes was dense and defect-free; this endowed the TFC membranes with a high rejection (over 90%) for multivalent ions. The pure water ux of the membrane was also more than 61.28 L m À2 h À1 because of the striped structure. We rst used hydrophilic micromolecules to change the diffusion rate of the macromolecules by chemical bonding rather hydrogen bond. In addition, the membrane showed a favorable antifouling property in HA, SA and BSA solutions (FRR > 89%). In conclusion, we propose a novel and effective strategy for promoting both the structural and chemical properties of TFC NF membranes to meet the antifouling demands of practical ltration processes, giving insight into the construction of a zwitterionic striped surface membrane by the introduction of micromolecules to macromolecules in the aqueous phase via chemical bonds.

Conflicts of interest
The authors declared that they have no conicts of interest to this work. We declare that we do not have any commercial or associative interest that represents a conict of interest in connection with the work submitted.