Toward enhancing the chlorine resistance of forward osmosis membranes: an effective strategy via grafting cyclohexylamine Open Access

At present, the world's available fresh water resources are still in short supply and water quality problems are still severe. Forward osmosis (FO) is an effective way to obtain freshwater resources. The FO technology uses only the osmotic pressure of the solution on both sides of the membrane to provide power without applying external pressure to achieve the separation of pollutants and water. It has the advantages of low energy consumption, strong retention capacity, strong resistance to membrane fouling, and low membrane fouling. At present, it is mainly used for seawater desalination, sewage treatment and water purification.

Polyamide (PA) TFC membrane has been playing a leading role in the market for nearly 40 years. PA TFC membrane has a composite structure of interfacial polymerized (IP) PA film supported by a microporous supporting membrane. The porous support membrane and the relatively dense separation layer enable the membrane to obtain high water flux and salt rejection. However, one of the biggest drawbacks of PA TFC membrane is its sensitivity to chlorine (Ni et al. 2014). As the most common and effective disinfectant in the water treatment process, chlorine can make the membrane sharply lose its salt rejection function.

The presence of chlorine-sensitive sites in the PA film made the membrane vulnerable to chlorine. According to existing research, scholars have different theories on chlorination, which can be roughly divided into three types: chlorination of amides, chlorination of aromatic rings and Orton rearrangement. Chlorination of amides means that the N-H bond on the amide is changed into an N-Cl bond by the attack of active chlorine, and the degree of damage of this method to the membrane property is far less than the other two methods. Aromatic ring chlorination is due to the aromatic ring being rich in electron groups, with high electron cloud density and being very sensitive to chlorine. Electrophilic substitution occurs on the aromatic ring, which directly causes the chlorination reaction. In Orton rearrangement, the Cl⁻ attacks the N-H group (nitrogen-hydrogen bonding) of the amide bond and forms N-chloramine. The N-chloramine would undergo a rapid intermolecular rearrangement (Gohil & Suresh 2017).

Over the past few decades, extensive efforts have been made to overcome this limitation. Most of the current research is focused on preventing the chloride solution from contacting the PA selective layer via a sacrificial layer or constructing a surface protection (Yao et al. 2019). Surface coating and chemical grafting provided the possible solutions to the mentioned issues. Surface coating is a physical way to construct a protective layer on the membrane surface to cover sites sensitive to chlorine. Hydrophilic copolymers poly (N-isopropylacrylamide-co-acrylamide) (P (NIPAM-co-Am) (Liu et al. 2011), graphene oxides coated by tannic acid (GOT) (Kim et al. 2016), terpolymer poly (methylacryloxyethyldimethyl benzyl ammonium chloride-r-acrylamide-r-2-hydroxyethyl methacrylate) (P (MDBAC-r-Am-r-HEMA)) and so on have been proved to be effective in chlorine-resistant modification of PA membranes. Unfortunately, this kind of physical coating would gradually fall off in the long-term operation of a cross-flow cycle. On the other hand, the dense surface layer formed by coating could reduce the water flux and affect the separation efficiency of the membrane. For example, Gholami et al. (2018) studied a new coating way of polyethylene glycol diacrylate (PEGDA) as a chlorine resistance layer. Although the observed salt rejection increased by 9.4% in 90 wt% PEGDA-coated membrane, the excess PEGDA layer reduced the water flux, which limited the efficiency of forward osmosis to some extent.

Chemical grafting is the introduction of a new structure on the surface of a membrane by the combination of chemical bonds, which has been widely used in the construction of a chlorine-resistant surface for PA TFC membranes. Grafting by graphene oxide (GO) (Shao et al. 2021), nitrogen-doped graphene oxide quantum dots (N-GOQDs) (Yi et al. 2020), neutral hydrophilic polymer polyvinyl alcohol (PVA) (Liu et al. 2015) and so on have successfully enhanced the chlorine resistance of PA membranes. Wang successfully prepared nanocomposite membranes by interfacial polymerization of Ti₃C₂T_x embedded into the polyamide selective layer. The water flux of modified membrane reached 2.53 Lm⁻² h⁻¹bar⁻¹, the salt rejection was 98.5%. Simultaneously, the enhanced chlorine resistance of the membrane was also obtained (Wang et al. 2020). However, the flux reduction caused by grafting was still a problem that could not be ignored. Zhu improved the chlorine resistance of TFC membranes with in-situ grafted melamine while capturing the maximum water flux. Although the increase of hydrophilicity was beneficial to the increase of water flux, the increase of transport resistance of the graft layer was not conducive to the transfer of water. The maximum water permeability for the modified membranes decreased by 12.0% (Zhu et al. 2020).

As an aliphatic monoamine compound, cyclohexylamine was often used as an intermediate in the organic synthesis process and is very promising in membrane modification. Considering the chlorine resistance of the membrane, cyclohexylamine can be used to construct sacrificial layers, and the short chain structure formed by monoamine can reduce transport resistance. Thus high chlorine resistance can be obtained while obtaining the maximum flux. Nevertheless, the effect of cyclohexylamine on the chlorine resistance of FO membranes has not been explored.

In this study, cyclohexylamine was grafted to the FO membrane via an in-situ modification method to improve the chlorine resistance. The surface morphology, roughness, functional group and element content of the modified membrane were systematically characterized. The separation performance of the modified membranes was evaluated by water flux, salt reverse salt and salt rejection. Moreover, the chlorine resistance of the modified membrane was further analyzed through chlorine immersion experiments with different concentrations of NaClO solution at different pH.

Polyacrylonitrile (PAN, Mw:150,000, Macklin), N-Methyl pyrrolidone (DMP, purity > 99.5%, Macklin) and absolute ethyl alcohol (>99.7%, Macklin) were used to fabricate a support layer. 1,3,5-benzenetricarbonyl trichloride (TMC, purity > 99.0%, Macklin, stored in ice packs), m-phenylenediamine (MPD, purity > 99.5%, Macklin), n-hexane (95.0%, Damaol), Hydrochloric acid solution (HCl, 36% ∼ 38%), sodium hydroxide solution (NaOH, ≥96%), sodium (NaCl, ≥99.8%) chloride, sodium hypochlorite solution (NaClO, 8%) and cyclohexylamine (97%, stored in ice packs) were obtained from Damao Chemical. The membrane cell was made in the laboratory. The film applicator (stainless steel, Tianjin precision instrument, China) was used to scrape the membrane, and the control range of membrane thickness was 0–3500 nm.

The porous support membrane was fabricated by the nonsolvent-induced phase separation (NIPS) method. The PAN powder was dissolved in m-phenylenediamine at 65 °C and stirred for 12 h. The PAN solution (16.7 wt%) was stored in a sealed container after it was dissolved. Then the casting solution was put into the ultrasonic cleaner (F-008S, FuYang Technology, China) for degassing to remove bubbles. After 24 h, the solution was scraped on the glass plate with a casting height of ∼100 μm, and the membrane was immediately put into a DI water coagulation bath. The prepared support layers were stored in DI water and the DI water was replaced every 24 hours.

The PA selective layer was prepared on the PAN supports via the IP method. First, the prepared PAN support was soaked in MPD (water as solvent, 5 wt%) for 5 minutes, then removed excess MPD from the surface. Immediately, the membrane was immersed in the TMC (n-hexane as solvent, 0.5 wt%) and allowed to react for 2 min. In order to remove unreacted TMC, all finished membranes were rinsed with pure n-hexane (Kwon et al. 2019). Then the membrane was treated by water bath at 90 °C for 5 min. The fabricated membrane was named TFC. In addition, cyclohexylamine (0.5 wt%, 1.0 wt%, 1.5 wt%) was added to MPD, and the membranes prepared by the same IP method were named TFC-1, TFC-2, and TFC-3. All the finished membranes were stored in deionized water.

Changes in surface chemical properties of the membranes were characterized by the attenuated total reflectance Fourier transform infrared (ATR-FTIR, Thermo Fisher, USA) with a resolution of 1 cm⁻¹ and a range of 350–7,800 cm⁻¹. The membrane was ground and sampled by BrK tablet method. The surface morphologies of the support membrane and TFC membranes were observed by scanning electron microscopy (SEM, Zeiss EVO18, Germany). Before making the sample, all the membranes had to undergo a vacuum drying process of 60 °C for 24 hours. The membranes were all sprayed with an ultrathin and uniform gold membrane by a vacuum coating equipment (CCU-010LV, Shanghai, China). The surface morphological structure was observed and the surface roughness was measured quantitatively by the atomic force microscope (AFM, DIMENSION ICON, Bruker Nano, USA) imaging and analysis. Surface roughness was reported in terms of arithmetic average (Ra). To investigate the chemical bond and composition change in the surface of the FO membranes, X-ray photoelectron spectroscopy (XPS, Thermo Kalpha, USA) was performed by spectrometer, which applied an Al K alpha monochromatic X-ray radiation source.

The chlorine stability determination was carried out by a static test. Briefly, the single membrane sample was immersed in NaClO solution. The pH of the solution was adjusted to 7, 9 and 11. After soaking and stirring, all membranes needed to be cleaned repeatedly. Then the membranes were clipped into the FO membrane cells to measure their FO performance, including water flux, reverse salt flux and salt rejection. The total amount of exposure of the FO membrane to free Cl was expressed as CT_FC (ppm·h) = ∫C_FC dt (Powell et al. 2015). The concentration of the NaClO solution was adjusted to 2,000, 10,000, and 20,000 ppm along with the chlorination process.

Surface morphology of modified membranes was observed by SEM and AFM. As shown in Figure S2a, a large number of uniform micropores were observed on the surface of the PAN support. Plenty of cochlear structures appeared on the membrane surface, which was characteristic of polyamide structure, after the interfacial reaction (Figure S2b). The appearance of polyamide structure also confirmed the success of interfacial polymerization. As shown in Figure S2b–d, the surface structure of the membrane grafted with cyclohexylamine changed greatly. After grafting of cyclohexylamine at low concentration, the changes of peak-valley above the TFC-1 membrane surface were not obvious. With the increase of grafting amount, the surface of the membrane became denser and the peak-valley structure disappeared gradually. For the surface roughness, the pristine TFC membrane with an Ra of 22.0 nm had a relatively smooth surface (Figure S2g). The grafting of a high concentration of cyclohexylamine tends to form a much smoother surface. With the cyclohexylamine concentration increased to 1.0 wt% and 1.5 wt%, the Ra decreased to 18.7 nm and 17.2 nm, respectively. As an alicyclic monoamine compound, the amine groups of the cyclohexylamine could react with the acyl chloride groups, leading to the formation of a short-chain polyamide. Therefore, the composite structures composed of the long-chain PA were formed by TMC and MPD and the short-chain PA were formed by cyclohexylamine and TMC. With the increased cyclohexylamine concentrations, more PA would be formed by cyclohexylamine and unreacted acyl chloride groups. The surfaces of the modified membranes were denser. Thus, the amount of cyclohexylamine had different effects on the membrane surface roughness.

Figure S3 and Figure S4 show the FTIR and XPS spectrum of the modified membranes, which were conducted to further confirm the chemical elements and functional groups. Compared to the spectra of the PAN support, both the TFC membrane and TFC-1 membrane showed newly appeared peaks at 1,505 (aromatic ring skeleton stretching vibration), 1,556 (amide II, N-H stretching) and 1,648 (amide I, C = O stretching). The peak of 1,556 and 1,648 are the characteristic peaks of the PA (Kwon et al. 2008; Antony et al. 2010). The spectra indicated the polyamide structure has been grafted to the support layer successfully and the graft modification of cyclohexylamine does not cause any significant change in the peak. This might be because the modification process and the IP process were also the process of forming amides by amines and acyl chlorides. The XPS spectra could be used to give a further analysis of the element content and bond information of the grafted membrane surface. As shown in Figure S4, the high-resolution N1s spectrum of the modified membrane was used to analyze the bond information of the membrane surface. The N1s spectrum of the FO membrane was deconvoluted into three peaks at 398.6, 399.5 and 400.6 eV, which were attributed to the C ≡ N, N-C = O and N-H species, respectively (Tan et al. 2018; Zhu et al. 2020). The C ≡ N bond originated from the polyacrylonitrile-based support membrane, and the N-C = O and N-H bonds originated from the PA on the surface of the active layer. After cyclohexylamine grafting, the area ratio of N-C = O on the TFC-1 membrane surface decreased from 56.0 to 49.8%. High concentrations of cyclohexylamine made the area ratio of N-C = O on the membrane surface increase to 64.8%. The increase of N-C = O bond content also indicated that the addition of a relatively large amount of cyclohexylamine will lead to the formation of more amide bonds on the surface of the membrane.

Figure 1 shows the FO water Jw (Equation (1)) and Js (Equation (2)) of all the membranes evaluated in both FO and PRO modes. Through the determination of PA layer structure, monomer composition is one of the key parameters affecting the separation performance of TFC membrane. Hence, the FO performance of TFC membrane was investigated by adjusting the concentration of cyclohexylamine systematically to determine the best modified composition. As the cyclohexylamine concentration increased up to 0.5 wt%, Jw increased from 12.4 to 13.2 LMH, while Js increased from 7.1 to 9.3 gMH. Apparently, the increased roughness due to low grafting content favoured the improvement of Jw and Js. With further increase in the content of cyclohexylamine to 1.0 wt% and 1.5 wt%, Jw and Js began to show a downward trend. This was similar to other graft modification studies. The large number of graft structures would inevitably lead to greater steric hindrance, resulting in a decrease in flux. In particular, the Jw and Js of the TFC-3 are 7.45 LMH and 4.09 gMH, respectively. However, the flux performance of TFC-2 was relatively stable. The water flux of TFC-2 decreased by 13% due to chemical grafting, which was close to that of Zhu's study. In general, both Jw and Js are higher in PRO mode than those in FO mode due to more severe internal concentration polarization (ICP). This point has also been confirmed in this study, as shown in Figure 1(b).

In the case of rejections, Figure 2(e) shows the salt rejection (R) (Equation (3)) of the TFC, TFC-1, TFC-2 and TFC-3 membranes evaluated in both FO and PRO modes. The R of pristine TFC membrane is 98.5%, the addition of cyclopropylamine seemed to be able to further increase the R, except for TFC-1. The R of TFC-2 and TFC-3 is 99.3% and 99.7%, respectively, which demonstrated that the cyclohexylamine-grafted membranes possessed a higher NaCl selectivity.

This result was consistent with our previous SEM and AFM characterization of the membranes. The denser surface structure was beneficial to improve the salt retention performance of membranes. The reduced separation performance of TFC-1 may be attributed to a small amount of cyclohexylamine terminating the growth of polyamide chain, and the short chain amide structure had less steric hindrance effect on water and salt.

All the chlorination experiments were carried out by soaking in sodium hypochlorite solution, and the chlorination strength was 2,000, 10,000, 20,000 ppm·h, respectively. The Jw, Js and R were recorded to assess the effects of chlorine exposure on the separation performance. Figure 2(a)–2(d) show the Jw and Js of the membranes at different chlorination strengths at 9 pH in FO and PRO modes. The destruction of the amide bond by chlorination would increase the flux of Jw and Js. After 20,000 ppm·h Cl exposure, the Jw of the TFC membrane sharply increased from 12.4 to 34.7 LMH, the Js of the TFC membrane sharply increased from 7.18 to 31.7 gMH. This trend of the membrane properties did not change after modification. However, the sensitivity of the modified membrane to chlorine was reduced. The Jw of the TFC-2 and TFC-3 membrane just increased from 7.67 to 26.22 LMH and 4.34 LMH to 17.88 LMH, respectively. Even after 20,000 ppm·h Cl exposure, the Js of the TFC-2 and TFC-3 membrane is 19.60 gMH and 13.87 gMH, respectively, which is much lower than the TFC membrane. The denser surface and more amide bonds could explain the greater chlorine resistance. Similar to the case in FO mode, the FO performance of modified membranes was less affected by chlorination in PRO mode.

R, which was the most intuitive representation of the membrane retention performance, also demonstrated the excellent chlorine resistance of the TFC-2 and TFC-3 membranes. After 20,000 ppm·h Cl exposure, the R of the TFC-2 membrane is 15.5% higher than the TFC membrane. The R of the TFC-3 membrane is 19.7% higher than the TFC membrane. We think there are three possible reasons: (i) it might be attributed to the denser surface layer of cyclohexylamine production, which acts as a sacrificial layer during chlorination to protect the structural and functional integrity of the polyamide. (ii) Cyclohexylamine belongs to the alicyclic group and does not contain a benzene ring structure, so direct chlorination of the benzene ring and subsequent Orton rearrangement reaction will not occur, only chlorination of the amide would happen, the method that had the least degree of damage to the membrane, so it was also conducive to the chlorine resistance of the membrane. (iii) Alicyclic and aliphatic groups were electron-donor groups, which could increase the density of the electron cloud of amide bond. So it could improve the chlorine resistance of the PA membrane.

It was interesting to note that the cyclohexylamine modification with 0.5 wt% had completely opposite results, both FO performance and chlorine resistance were not as good as TFC. This phenomenon resulted from the low content of cyclohexylamine acting as chain terminators, resulting in short-chain amides that were more vulnerable to chlorine attack. Further evidence could be seen in the XPS, SEM and AFM diagrams of the TFC-1 membrane. To sum up, the chlorine resistant sacrificial layer could be constructed by adding the appropriate concentration of cyclohexylamine, and the original flux of the membrane can be kept as much as possible. Figure 3 shows the process of cyclohexylamine modifying the FO membrane.

However, the R of the TFC-2 and TFC-3 membranes were 33.1% higher than the TFC membrane, after 20,000 ppm·h Cl exposure. Direct chlorination of the benzene ring did not occur because the cyclopropylamine does not contain the structure of the benzene ring. So the destruction of the amide structure formed by the active chlorine was mainly through the mutual transformation of the N-H bond and the N-Cl bond. On the contrary, this difference became no longer significant with the increase of chlorination intensity. As pH modulates alkalinity, the R of the membrane was significantly higher than that of the acidic condition. That was because the amide bond could be dechlorinated reversibly by OH⁻, producing the unchlorinated amide and free Cl. Thus the chlorine resistance of the modified membrane was still affected by pH, and the alkaline environment was more conducive to the retention of salt.

A high FO performance and chlorine resistant TFC FO membrane was successfully fabricated with a PAN support and cyclohexylamine as grafting material via a traditional IP method. The low content of cyclohexylamine terminated the growth of the selective layer polyamide chain, and the sparse structure greatly reduced the separation and chlorine resistance of the membrane. Compared with the pristine membrane, the TFC-2 and TFC-3 membranes possessed a denser surface, higher salt rejection and chlorine resistance. Grafting of cyclohexylamine increased the chlorine resistance with a negligible decline in water flux. Strikingly, the TFC-2 and TFC-3 membranes exhibited 19.2% higher salt rejection than TFC after treating with 20,000 ppm·h NaClO solution at pH 11. The excellent chlorine resistance, separation performance and high flux of the modified membranes highlights the feasibility of constructing a chlorine-stable interface by in-situ grafting of cyclohexylamine for the desalination.

This research is supported by the National Key R&D Program of China (2019YFC1803800); the Fundamental Research Funds for the Central Universities (N2001016) and the National Natural Science Foundation of China (41571455).

All relevant data are included in the paper or its Supplementary Information.