ABSTRACT
Nanofiltration (NF) membranes have already been widely used owing to the merits of high permeability, energy saving, and selective separation. Here, a convenient way to prepare a functional layer of a composite NF membrane was investigated, which used taurine to replace a part of piperazine (PIP) in the aqueous phase in interfacial polymerization to decrease the cross-linking density of the polyamide network. Although the molecular weight cutoff by the membrane decreased to 2,000 Da, the interception to sulfates and negatively charged dyes was strengthened. As the taurine ratio increased, the surface layer thinned, and pure water permeation improved, but MgSO4 rejection showed a tendency of ‘first-increase-then-decrease’ under the synergism of size exclusion and the Donnan effect. At the apex ratio of 4:1 (taurine: PIP), the T80 membrane showed >98% retention to sulfates, MgSO4, Na2SO4, and (NH4)2SO4, and >98.7% rejection to anionic dyes, methyl blue, congo red, and acid chrome blue K (ABK). The film also exhibited excellent fouling resistance to emulsified oil, humic acid, and bovine serum albumin, and fractionation of ABK and NaCl.
HIGHLIGHTS
The high permeability of loose nanofiltration membranes.
Taurine partially replaces piperazine.
SO3H covered surface.
Fractionation of dye and NaCl.
Reduced cross-linking density.
INTRODUCTION
Recently, the nanofiltration (NF) technique has also found special merits in the selective separation of charged solutes, such as dyes and inorganic salts, in the water treatment field (Ritt et al. 2019; Dai et al. 2020; Eslami et al. 2020). According to the various structures, some loose nanofiltration (LNF) membranes showed the capacity of fractionation of dyes and inorganic salts (Van der Bruggena et al. 2004) through the size exclusion effect, whereas others provided high retention to multivalent inorganic salts and high permeability to monovalent inorganic salts through Donnan effects (Ji et al. 2013; Dong et al. 2020). These make upgrading for recovery and cyclic utilization of high-value-added components possible at the same time.
Such commercially available NF membranes are mostly thin-film composite (TFC) polyamide (PA) membranes, such as Sepro NF2A and Sepro NF6 (Ultura) (Feng et al. 2022). Generally, TFC PA membranes comprise a compact, highly cross-linked PA selective layer and an underlying porous support layer. The PA selective layer is fabricated by interfacial polymerization (IP) reaction between piperazine (PIP) aqueous solution and trimesoyl chloride (TMC) organic phase. There are many strategies to adjust the properties of the selective layer to develop novel NF membranes, such as reduction of the concentration of reaction solution, employment of new monomers, and introduction of dissociable groups. Tian et al. (2021) used low-concentration PIP (0.05w/v%) to prepare an 8.5 nm defection-free PA film on a smooth, hydrophilic medium pore polyester ultrafiltration (UF) membrane. The as-produced TFC membrane achieved pure water permeability of 46.6 Lm−2h−1bar−1 and a 98.1% Na2SO4 removal rate. A loose skin layer might be obtained by replacing TMC or PIP with other multi-chlorides or multi-amines (Zhang et al. 2017; Zhu et al. 2020; Lu et al. 2023). Zhao et al. (2023) replaced PIP with arginine (Arg) to prepare an NF using polyethersulfone (PES) as a substrate. Pure water permeability of the membrane was up to 130.4 L m−2 h−1 bar−1 and all dye rejections were more than 95%, while salt rejections for NaCl and Na2SO4 were both less than 8%, exhibiting excellent dye/salt fractionation capacity. The addition of dissociable groups containing mono-functional compounds could not only decrease the concentration of the main monomer but also introduce charged groups into the membrane surface at the same time, which would endow the membrane with special selectivity for dissociable solutes in polluted water. Zhang et al. (2023) added a monofunctional amine, 4-hydroxybenzenesulfonic acid sodium salt (HBSA), as a modifier in the aqueous phase. The membrane HBSA-PA-2 exhibited water permeability of 34.4 L m−2 h−1 bar−1 as 99.1% Na2SO4 rejection, which was three times higher than that of commercial NF membranes at the same Na2SO4 retention rate.
Mono-functional monomers, such as mono-amine or mono-acyl chloride, can decrease the cross-linking density of the PA layer and, hence, the retention property of the membrane. However, charged groups can partially compensate for the retention deterioration to dissociable solutes through electrostatic repulsion when the monomer contains ionizable groups. The balance between the decrease of cross-linking density (size exclusion effect) and the increase of electrostatic repulsion (Donnan effect) is far from clear. The study of such a problem is profound for the development of NF membranes in various application fields.
Taurine is a metabolic waste of sulfur-containing proteins: it is abundant in the tissue fluid of mammals and serves as a facile osmolyte to adjust the concentration gradient and osmolarity across the plasma membrane accompanied by inorganic ions, such as K+ and Cl− (Huxtable 1992). Here, different ratios of taurine as a replacement for PIP were systematically investigated regarding their influence on the membrane structure and their corresponding performance in TFC membrane preparation and evaluation. It was found that a large proportion of taurine in the aqueous phase of the IP reaction could alleviate the decrease of cross-linking from dilute PIP concentration, which, accompanied by high SO3H density, brought high water permeation and effective retention for anionic solutes and high fouling resistance. The interaction between taurine and PIP was investigated in detail. The effect of taurine during the IP process was hence summarized. The mechanism of high permeation and rejection for sulfates and anionic dyes was deduced to access a facile way to prepare the functional NF membranes.
EXPERIMENTAL
Materials
Polysulfone ultrafiltration membrane (Psf-UF) was provided by Beijing OriginWater Technology Co., Ltd. Taurine was supplied by Shanghai Titan Scientific Co., Ltd Methyl blue (MB, AR). Congo red (CR, AR), acid chrome blue K (ABK, AR), NH4Cl, and sodium dodecyl sulfate were purchased from Tianjin Guangfu Fine Chemical Research Institute. Trimesoyl chloride (TMC) and bovine serum albumin (BSA) was supplied by Shanghai Aladdin Reagent Co. PIP, MgCl2, MgSO4, NaCl, Na2SO4, polyethylene glycol (PEG, 200-4000), and HCl (36.0 wt%) came from Tianjin Kemiou Chemical Reagent Technologies Co., Ltd. Humic acid (HA), (NH4)2SO4, Na2HPO4, KH2PO4, citric acid, and n-hexane were purchased from Tianjin Fengchuan Chemical Reagent Technologies Co., Ltd.Oil used is edible soybean oil.
Preparation of membranes
T membranes: Psf-UF membrane (50 mm × 50 mm) was dipped into a 2.0 wt% taurine/PIP blend aqueous solution for 10 min, and excess solution was blotted dry with filter paper. Sequentially, the membrane was immersed in a 0.1 wt% TMC n-hexane solution for 30 s to obtain a PA functional layer. The composite membrane was then washed with deionized (DI) water. The ratios of taurine/PIP are listed in Table 1.
Aqueous phase formula of TFC PA membranes
Membranes . | PIP in aqueous phase (wt%) . | Taurine in aqueous phase (wt%) . | Taurine/ (Taurine + PIP) × 100% (wt%) . | Membranes . | PIP in aqueous phase (wt%) . | Taurine in aqueous phase (wt%) . |
---|---|---|---|---|---|---|
T0/P0 | 2.00 | 0 | 0 | |||
T10 | 1.80 | 0.20 | 10.0% | P10 | 1.80 | 0 |
T20 | 1.60 | 0.40 | 20.0% | P20 | 1.60 | 0 |
T30 | 1.40 | 0.60 | 30.0% | P30 | 1.40 | 0 |
T40 | 1.20 | 0.80 | 40.0% | P40 | 1.20 | 0 |
T50 | 1.00 | 1.00 | 50.0% | P50 | 1.00 | 0 |
T60 | 0.80 | 1.20 | 60.0% | P60 | 0.80 | 0 |
T70 | 0.60 | 1.40 | 70.0% | P70 | 0.60 | 0 |
T80 | 0.40 | 1.60 | 80.0% | P80 | 0.40 | 0 |
T90 | 0.20 | 1.80 | 90.0% | P90 | 0.20 | 0 |
T95 | 0.10 | 1.90 | 95.0% | P95 | 0.10 | 0 |
T99 | 0.02 | 1.98 | 99.0% | P99 | 0.02 | 0 |
T99.5 | 0.01 | 1.99 | 99.5% | P99.5 | 0.01 | 0 |
Membranes . | PIP in aqueous phase (wt%) . | Taurine in aqueous phase (wt%) . | Taurine/ (Taurine + PIP) × 100% (wt%) . | Membranes . | PIP in aqueous phase (wt%) . | Taurine in aqueous phase (wt%) . |
---|---|---|---|---|---|---|
T0/P0 | 2.00 | 0 | 0 | |||
T10 | 1.80 | 0.20 | 10.0% | P10 | 1.80 | 0 |
T20 | 1.60 | 0.40 | 20.0% | P20 | 1.60 | 0 |
T30 | 1.40 | 0.60 | 30.0% | P30 | 1.40 | 0 |
T40 | 1.20 | 0.80 | 40.0% | P40 | 1.20 | 0 |
T50 | 1.00 | 1.00 | 50.0% | P50 | 1.00 | 0 |
T60 | 0.80 | 1.20 | 60.0% | P60 | 0.80 | 0 |
T70 | 0.60 | 1.40 | 70.0% | P70 | 0.60 | 0 |
T80 | 0.40 | 1.60 | 80.0% | P80 | 0.40 | 0 |
T90 | 0.20 | 1.80 | 90.0% | P90 | 0.20 | 0 |
T95 | 0.10 | 1.90 | 95.0% | P95 | 0.10 | 0 |
T99 | 0.02 | 1.98 | 99.0% | P99 | 0.02 | 0 |
T99.5 | 0.01 | 1.99 | 99.5% | P99.5 | 0.01 | 0 |
P membranes: A series of pure PIP membranes were prepared with the same PIP concentration corresponding to the T membranes as a control, and their formulation is also provided in Table 1.
P80-HCl and P90-HCl membranes were prepared using a 0.1 wt% TMC n-hexane solution and pH-adjusted PIP aqueous solution following the procedure of T membranes. The solution pH of PIP was titrated with a 10.0 wt% HCl solution according to the pH value of the T80 or T90 aqueous phases graphed in Figure S1, respectively.
Membrane characterization
The surface chemistry of the TFC membrane was analyzed by attenuated total reflectance infrared (ATR-IR) spectroscopy using a Nicolet 6700 infrared spectrometer (Thermo Fisher, USA). X-ray photoelectron spectrometry (XPS) analyses were investigated by a Thermofisher K-alpha spectrometer with a focused monochromatized Al Kα radiation. An S-4800 field emission scanning electron microscope (FESEM, Hitachi, Japan) was used to collect the surface and cross-section morphology of membranes. Membrane samples were freeze-dried for 8 h and broken in liquid nitrogen. The average roughness (Ra) of the membrane surface was analyzed by atomic force microscopy (AFM, Bruker, Icon, Germany) in tapping mode. The dynamic water contact angle (DWCA) of the membrane surface in 120 s was measured by a Drop Shape Analysis 100 contact angle meter (Kruss BmbH Co., Germany). The volume of the DI water droplet was set to 2.0 μL. Membrane surface charge properties were characterized by streaming zeta potential measurement with a SurPASS electrokinetic analyzer (Anton Paar, Austria) at pH 2–10. KCl aqueous solution (1 mM) was used as an electrolyte. A total organic carbon analyzer (TOC-VCPH, SHIMADZU, Japan) was used to analyze the concentration of PEG solutions. Liquid 1H nuclear magnetic resonance (1H NMR) spectra were tested by a 400 MHz nuclear magnetic resonance spectrometer (Bruker, Germany).
Filtration performance
The fractionation performance of the membrane was conducted using an ABK–NaCl (0.1 g·L−1/2.0 g·L−1) blend aqueous solution with a total recirculation mode. Water flux and solute rejection were recorded once per 10 h in 120 h.
Anti-fouling performance
RESULTS AND DISCUSSION
Relationship between membrane performance and taurine/PIP proportion
(a) Water permeation and (b) retention to MgSO4 of membranes in 2.0 g·L−1MgSO4 solution filtration test.
(a) Water permeation and (b) retention to MgSO4 of membranes in 2.0 g·L−1MgSO4 solution filtration test.
Membrane characterization
High resolution of C1s and molecular weight cutoff
Membranes . | Area of chemical compositions . | Molecular weight cutoff (Da) . | |||
---|---|---|---|---|---|
O–C = O/N–C = O . | C–N . | C–C/C–H . | C–S . | ||
T0 | 13,210.5 | 25,186.5 | 54,602.1 | 0 | 200–400 |
T80 | 12,827.1 | 21,884.8 | 53,711.0 | 5,139.0 | 2,000 |
P80 | 12,447.6 | 18,555.5 | 50,221.1 | 0 | 2,000 |
Membranes . | Area of chemical compositions . | Molecular weight cutoff (Da) . | |||
---|---|---|---|---|---|
O–C = O/N–C = O . | C–N . | C–C/C–H . | C–S . | ||
T0 | 13,210.5 | 25,186.5 | 54,602.1 | 0 | 200–400 |
T80 | 12,827.1 | 21,884.8 | 53,711.0 | 5,139.0 | 2,000 |
P80 | 12,447.6 | 18,555.5 | 50,221.1 | 0 | 2,000 |
Surface characteristics of membranes: (a) and (b): ATR-IR of membranes; (c)–(f): XPS spectra of membranes; (g) DWCA; and (h) zeta potential.
Surface characteristics of membranes: (a) and (b): ATR-IR of membranes; (c)–(f): XPS spectra of membranes; (g) DWCA; and (h) zeta potential.
The hydrophilicity of the membrane surface was evaluated by dynamic water contact angles (DWCAs) (Figure 2(g)). DWCAs of T0 and T80 rapidly descended in the first 10 s and were kept at fixed values of 53.18° and 32.88°, respectively. DWCAs of T80 were lower than those of P80 since taurine improved the hydrophilicity of the membrane. Figure 2(h) shows the surface zeta potential of the membranes. The isoelectric point of T0 was pH 5.13, which was a distinct characteristic of the TFC PA membrane. The isoelectric points of P80 were pH 4.29, while T80's was pH 4.06, which meant that T80 possessed the most surface negative charges among the three samples. SO3H in taurine further increased the electronegativity of the membrane surface.
In AFM images, it was observed that the granular structures were gradually indistinct and the average roughness (Ra) of the surface decreased as PIP diluted. T0 had a rough surface with visible nodular and Ra was 6.54 nm. T80 and P80 possessed granular-contained surfaces, which corresponded with the SEM images. Ra of T80 was 4.99 nm and greater than 4.79 nm of P80. It also demonstrated that taurine improved IP reaction and inhibited the decrease of cross-linking density of the PA layer. The order of surface roughness was T0 > T80 > P80, which was opposite to their water permeation.
Blend solution parameters of taurine and PIP: (a) 1H NMR spectra and (b) conductivity; (c) Filtration performance of membranes: permeance and (d) rejection for 2.0 g·L−1 MgSO4.
Blend solution parameters of taurine and PIP: (a) 1H NMR spectra and (b) conductivity; (c) Filtration performance of membranes: permeance and (d) rejection for 2.0 g·L−1 MgSO4.
The curves of pH change as PIP and taurine solution concentration are shown in Figure S1. According to the three curves, the pH values of blend solutions were all lower than pure PIP solutions of the same concentration. We were curious whether other acids could also alleviate PA layer thinning caused by PIP dilution. Therefore, an HCl solution was used to adjust the pH of the PIP solution and two membranes were prepared under PIP concentrations of 0.4 wt% and 0.2 wt%, which corresponded with 80 and 90% taurine in the blend solution (P80-HCl and P90-HCl), respectively. Their filtration performance for the 2.0 g·L−1 MgSO4 solution was compared with the corresponding T and P membranes (Figures 4(c) and 4(d). Low water permeance and high MgSO4 rejection of T membranes suggested that the PA layer of T membranes was the thickest. The highest water permeation and the lowest MgSO4 rejection of P membranes confirmed that their PA layer was thinner than that of the P-HCl membranes. This result illustrated that the reactivity of PIP was also slightly increased by other acids. However, a zwitterion, taurine, provided the best adjustment due to its moderate reduction in pH value. In previous literature (Li et al. 2009), an acid-binding agent was used to catch HCl, the byproduct of polycondensation between TMC and PIP. This is not contradictory to the results here. It could be found that only if PIP concentration was very low, the acidic additive could show an apparently positive effect. The more diluted the PIP concentration was, the more outstanding the acid effect was. An alkaline medium was necessary for this IP reaction. On the other hand, NH2 groups of taurine could also play a role in acid-binding agents due to their unique structure. PIP, accompanied by taurine, could grasp more COCl before their hydrolysis into COOH. As a result, the thinning of the PA layer was alleviated. The integrative effect of taurine is to increase PIP reactivity and cross-linking density of the surface layer.
Membranes performance in water treatment

Filtration performance of membranes: (a) water flux and (b) rejection for six salts; (c) Filtration performance of three anionic dye solutions (2.0 g·L−1); (d) fractionation of a blend solution of 0.1 g·L−1 ABK and 2.0 g·L−1 NaCl by T80; (e) anti-fouling performance of T80 in the filtration of emulsified oil and 1.0 g·L−1 HA solutions and (f) 1.0 g·L−1 BSA buffer solution of different pH (emulsified oil: 0.9 g soybean oil and 0.1 g sodium dodecyl sulfate in 1 L DI water).
Filtration performance of membranes: (a) water flux and (b) rejection for six salts; (c) Filtration performance of three anionic dye solutions (2.0 g·L−1); (d) fractionation of a blend solution of 0.1 g·L−1 ABK and 2.0 g·L−1 NaCl by T80; (e) anti-fouling performance of T80 in the filtration of emulsified oil and 1.0 g·L−1 HA solutions and (f) 1.0 g·L−1 BSA buffer solution of different pH (emulsified oil: 0.9 g soybean oil and 0.1 g sodium dodecyl sulfate in 1 L DI water).
Three anionic dyes, MB, CR, and ABK, were chosen to evaluate organic anions removal by T80. Their molecular weights are shown in Table S1. Within 5 days, water permeation of T80 gradually stabilized for all three dyes (Figures 5(c) and 5(d)). The rejections to CR, MB, and ABK were 99.4, 99.08, and 98.75%, respectively. High rejection of CR came from the self-association of CR molecules by hydrogen bonds, which increased particle size in the dye solution (Thong et al. 2018; Ren et al. 2019). A blend solution of ABK and NaCl was used to evaluate the fractionation of organic dissociable molecules and monovalent salt by T80. Figure 5(d) showed that water permeation and salt rejection of T80 both slightly declined at the first 50 h and then remained constant, with ABK at 99.5% and NaCl at 20%. T80 showed the capacity to fractionate anionic dye and NaCl with comparably high permeation, which is the typical application of the LNF membrane.
The anti-fouling performance of T80 was investigated through filtration tests of three simulated polluted solutions, including O/W emulsion (Oil), HA, and BSA. FRR of T80 was separately 98.8 and 98.5% for O/W emulsion and HA after a round of filtration and DI water rinse (Figure 5(e)). Sulfonic groups of taurine increased the hydrophilicity of the membrane surface and electronic repulsion to charged groups, which increased the anti-fouling ability of hydrophobic pollutants and anions. FRR of T80 for BSA buffer solution (pH = 7.3) was 98.7% (Figure 5(f)). When pH declined to 4.5, the FRR of T80 was up to 98.9%. The isoelectric point of BSA is pH 4.7. At pH 4.5, protein molecules were covered by positive charges, which were repelled by a negatively charged T80 surface and prevented from sticking to the membrane surface.
The molecular weight cutoff of T80 was 2,000 in PEG filtration tests (Table 1), which is common for LNF membranes and larger than 200–400 Da of T0. T80 showed different retention to neutral molecules and anionic organisms, which was suitable for the fractionation of anionic dyes and neutral small molecular solutes and chloride salts in a small complicated wastewater treatment process.
The rejection mechanism of T80 was the synergism of pore size exclusion and the Donnan effect due to the loose and SO3H-covered PA selective layer. Replacement of a large proportion of PIP by low-priced taurine not only decreased the cost of membrane fabrication but also improved the producing efficiency of the membrane process owing to high water permeation, effective retention for certain solutes, and high fouling resistance. T80 showed the potential of selectively removing multivalent anionic salts and anionic organic molecules accompanied by permeating small neutral organic molecules and chloride salts. This work provides a facile way to prepare an LNF membrane with zwitterionic monomer to implement special functions, such as the removal of sulfates, anionic dyes, and fractions of them and other solutes.
CONCLUSION
Taurine, the metabolite of sulfur-containing proteins in mammals, was used to regulate the IP reaction between PIP and TMC during the preparation of a thin film composite membrane. As a zwitterionic monomer, taurine strongly interacted with PIP to activate the NH group by SO3H and bind byproduct acid by NH2. As a result, the decrease in cross-linking density of the PA network was mitigated. The membrane with a 4:1 ratio of taurine to pip (T80) obtained a thicker and denser selective layer than the membrane with only 0.4 wt% PIP added (P80), which guaranteed effective rejection to sulfate salts and anionic dyes owing to strong electrostatic repulsion. It displayed high water permeation, a higher than 98% removal rate for three sulfates, more than 98.7% retention for three anionic dyes, efficient fractionation of NaCl and ABK, and good fouling resistance for emulsified oil, HA, and BSA. All these make T80 a suitable candidate for LNF membranes for complex wastewater treatment.
ACKNOWLEDGEMENT
The authors acknowledge the financial support from the National Natural Science Foundation of China (Grant No. 51373119) and Tianjin Key Projects of New Materials Science and Technology (17ZXCLGX00050). The authors would like to sincerely thank the following individuals for their contributions to this paper: BoWen Lv provided assistance in data graph beautification and modification. Shun Zhou offered support in the revision and polishing of some charts and figures. Qinxing Xie provided professional advice on academic norm supervision and submission.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.