Polysulfone (PSU; MW = 22 000) from Xian Lyphar Biotech Co., Ltd., Xian City, China, was used to fabricate the membrane substrates. N,N-dimethylformamide (DMF; 99.8 %) and 2,2,4-trimethylpentane (isooctane; 99 %) were purchased from Sigma-Aldrich Chemie GmbH, Buchs, Switzerland. MPD (> 99 %) and TMC (98 %) were ordered from Merck KGaA, Darmstadt, Germany. Triethylamine (TEA; ≥ 99 %) and (1S)-(+)-10-camphorsulfonic acid (CSA; 99 %) were purchased from Sigma-Aldrich, St. Louis, Missouri, USA. Sodium chloride (NaCl) was purchased from Thomas Baker (Chemicals) Pvt. Ltd., Mumbai, India. Deionized water (DI water) was used to prepare NaCl and MPD aqueous solutions and for other purposes such as cleaning.

The phase inversion method was used to prepare the PSU supporting sheets. The casting solution was prepared by dissolving 17 wt % dry PSU pellets in DMF. The mixture was stirred and heated to 60 ∘C for 6 h until a clear solution was formed, which was then degassed for more than 24 h at room temperature before use. Afterward, the solution was cast, using a homemade casting knife, by taking an aliquot from the clear solution with a pipette to spread on a clean glass plate to the desired thickness. The glass plate with the solution was then immersed into a water bath at room temperature, resulting in the immediate formation of the PSU support sheet that was separated from the glass plate in a moment. Then, all of the sheets were collected and stored in DI water for 24 h or more at 4 ∘C before use.

TFC FO membranes were fabricated on the top surface of the PSU sheet by the IP reaction between the MPD aqueous solution and TMC organic solution. The MPD aqueous solution was prepared by dissolving 2 % MPD in DI water, while the TMC solution was made by dissolving 0.15 % of TMC in isooctane. First, the MPD solution was poured onto the PSU sheet at different contact times. Then, the TMC solution was poured onto the PSU sheet that contains the MPD active sites, and the reaction time was also varied to study the effect of IP reaction time. At first, the MPD contact time was varied from 2 to 5 min, while keeping the contact time of the TMC at 1 min. Then, the best MPD contact time (i.e. 5 min) that gave the highest water flux was chosen, and the TMC solution contact time was studied in the range of 1 to 4 min. In order to study the effect of adding CSA-TEA at a weight ratio of 2 : 1, they were added to the MPD solution with a weight of 1 %. The IP reaction was conducted at room temperature. Finally, the obtained TFC membranes were dried in the oven at 60 ∘C for 10 min and then collected and stored in DI water for 24 h until testing.

A scanning electron microscope (Mira-3 FEG SEM, Tescan Orsay France S.A.R.L) and Atomic Force Microscope (Angstrom Advanced Inc., Stoughton, MA, USA) were used to determine the morphology of the prepared membrane. The hydrophilicity of the membranes was measured using contact angles (Theta Lite, TL-101; Ryts Instruments, Bangkok, Thailand).

The FO performance was tested in a bench-scale system, as shown in Fig. 1. This system consists of two tanks, with one being used for the feed solution and the other containing the draw solution. DI water was used as a feed solution, while 1 M NaCl was used as a draw solution based on the standard methodology that was described by Cath et al. (2013). These solutions were pumped to the membrane cell using two pumps from Pure Water Products, LLC, Denton, TX, USA (model – 75 GPD; volts – 24 VDC; workflow – 28 LPH). All experiments were conducted in FO mode (i.e., active layer faces the feed solution). The membrane was installed in a custom-made cell with the chamber's dimensions of 7.62 cm length, 2.54 cm width, and a depth of 0.3 cm. The water flux Jw can be estimated using the following equation (Al-Furaiji et al., 2018): 1Jw=ΔVAt, where Jw is the water flux (LMH is L m-2 h-1), ΔV is the change in feed solution volume (L), A is the active area of the membrane (m2), and t is the experiment's time (in hours).

Typically, salt flux is used in FO investigations to describe the selectivity of the membrane, while salt rejection is normally used in RO studies. The salt rejection equation can be used when there is a feed solution involved in the process, while in FO there is a feed solution and a draw solution. That is why the salt flux is used instead of the salt rejection. Salt flux through the membrane was estimated by monitoring the change in the conductivity of the feed solution and using the following equation (Al-Furaiji et al., 2020): 2Js=ΔCVAt, where Js is the salt flux (GMH is g m-2 h-1), ΔC is the change in the feed solution concentration (g L-1), V is the volume of feedwater flow (L), A is the active area of the membrane (m2), and t is the experiment's time (in hours).

The two dimensional (2D) and three dimensional (3D) AFM images of the top surface of the TFC membrane are presented in a scan area of 2500 × 2500 nm, as shown in Fig. 2. It can be clearly seen that the surface of the polyamide layer has a ridge-and-valley structure, with the average roughness of the FO membrane surface being 5.07 nm. The surface roughness of the prepared membrane was quite similar to what has been reported for typical FO (Mi and Elimelech, 2008) and nanofiltration membranes (Li et al., 2017, 2020). The rougher surface could be more beneficial for membrane performance as it gives a higher area for mass transfer, especially when dealing with a low-fouling feed solution.

SEM images were used to investigate the surface morphology of the TFC membranes, as shown in Fig. 3. It can be seen that the polyamide selective membrane was successfully formed on the PSU support sheet. This was confirmed by the leaf-like morphology, which is a typical structure for polyamide TFC membranes (Kadhom and Deng, 2019b; Zhou et al., 2014).

Figure 4 shows the contact angle measurement of the PSU support membrane and the polyamide selective layer. The contact angle of the PSU sheet was about 65∘, while the thin polyamide layer had a contact angle of 33∘. The contact angle of the membrane can be influenced by many parameters such as monomer concentration, reaction time, type of organic solution, posttreatment condition, etc., during the IP reaction process. However, the reported value of the contact angle in this research lies within the range of the previously reported contact angle of the TFC membrane (Kadhom et al., 2016; Lau et al., 2015). When the contact angle is small then it means that the membrane is hydrophilic and that the water penetrates easily into the pores of the membrane, which gives a better osmotic water flux performance. Nevertheless, when the contact angle is large, it means that the film is hydrophobic, so the internal surfaces of the pores become dry, and that leads to a membrane with lower water flux.

After the PSU support sheet has been prepared, the active layer is prepared by pouring the MPD solution onto the PSU layer after attaching it well on a glass plate. The effect of the MPD exposure time on the performance of the TFC FO membrane was studied by varying the contact time from 2 to 5 min while fixing the TMC reaction time at 1 min, as shown in Fig. 5.

The osmotic performance results revealed that water flux increases when the MPD exposure time increases. Also, increasing the MPD contact time leads to increasing salt flux. In fact, a higher MPD exposure time means more MPD molecules would react with the support layer and, accordingly, increase the IP reaction active sites. Moreover, well-formed cross-linking would be achieved at a higher MPD exposure time, which gives better IP reaction conditions when it reacts with the TMC later (Kadhom and Deng, 2019a).

In order to study the effect of IP reaction time, the TMC contact time was varied from 1 to 4 min by setting the MPD exposure time at 5 min, as shown in Fig. 6. The TMC organic solution was poured on the PSU substrate that contains the MPD active sites to conduct the IP reaction. It can be seen that the optimum condition that gave the highest water flux was recorded at a reaction time of 1 min. Interestingly, it can also be observed that the water flux and salt flux decreased sharply with the increase in TMC reaction time. This is mainly attributed to the fact that increasing the TMC contact time leads to generating a thicker polyamide layer and, consequently, a higher mass transfer resistance to permeation of water (Zhou et al., 2014). Moreover, the extent of the cross-linking is increased with the increase in the IP reaction time, and as a result, water flux and salt flux decreased (Wang et al., 2017).

The effect the addition of amine salt (i.e., CSA-TEA) on the performance of the FO process was studied, as shown in Fig. 7. It has been found that adding 1 % of the CSA-TEA to the aqueous MPD solution exhibited a moderate increase in both the water flux and salt flux. Similar behavior was reported for RO and nanofiltration processes (Khorshidi et al., 2017). It is known that polyamide formation during the IP reaction can result in the release of hydrogen chloride (Raaijmakers and Benes, 2016). The formation of hydrogen chloride can affect the reactivity of the monomer reactant in the aqueous phase (i.e., MPD). Therefore, the addition of a strong base such as TEA enhances the reactivity of the MPD and consumes the produced acid (i.e. hydrogen chloride). Also, it has been reported that the TEA acts as a catalyst by accelerating the MPD–TMC reaction and generating a thinner and more cross-linked polyamide layer (Vatanpour et al., 2017; Wang et al., 2017).

A comparison of the TFC-PSU membrane with some of the previously reported TFC membranes can be found in Table 1. It can be seen that the TFC-PSU membrane exhibited the highest water flux compared to the reported membranes. However, the reverse salt flux value lies within the range of the previously reported salt flux of the TFC membranes. If we look closely to the results of our previous work (Al-Furaiji et al., 2020) and compare it to the current work, it can be distinguished that the water flux of the current work is about twice that of the previous work, while the salt flux is a bit higher. There are two main differences between the previous work (Al-Furaiji et al., 2020) and the current work, as follows:

In the previous work, we used a polyacrylonitrile (PAN) polymer as a support for the TFC FO membrane, while in this work we used a polysulfone (PSU) polymer.