Deionized (DI) water was prepared using a water purification system (Synergy; Millipore, USA) with resistivity of 18.2 mΩ-cm at 25°C. An ultrasonicator (B8510-MT; Branson, USA) was used to remove invisible air bubbles in the polymer solutions.

PES (H-2000; Solvay, Korea) with a molecular weight (M_w) of 62,000–64,000 g/mol was used for PES membrane synthesis. The solvent was n-methyl-2-pyrrolidinone (NMP) (anhydrous 99.5%; Sigma-Aldrich, USA), and PVP (Mw: 10,000 g/mol; Sigma-Aldrich, USA) was used as a pore formation additive. Polyethylene terephthalate (PET) nonwoven fabric was used as substrate (AMFOR Inc., USA). NMP was also used as an adhesive to strongly bind the PES layer and PET substrate. A casting knife, made of laser precision-machined iron, was used to obtain membranes with a constant thickness. To evaluate the membrane permeability, a pure water permeation instrument was installed at the cross flow of assembled aryl cells (Millipore, USA). Divalent salt, CaSO₄·2H₂O, was dissolved in DI water to evaluate the membrane selectivity (BioXtra >98.0%; Sigma-Aldrich, USA). An atomic force microscope (AFM) was used in contact mode to evaluate the morphology and roughness of the prepared membranes (603–1001; Park systems, Korea).

In this study, PES membranes were synthesized via phase inversion, which is a widely used method in organic polymer membrane fabrication. They were synthesized under different synthesis conditions, such as the presence of substrate, adhesive, and pore former. The synthesis conditions of the PES membranes are summarized in Table 1. To make the PES membranes, PES powder was heated in an oven at 100° C for 1 hr to remove all moisture, which could influence the performance of the membrane during phase inversion. Then, 25 wt.% of PES powder in a ratio to total casting solution was dissolved in the NMP solvent for 3 hr. Following this step, the prepared solution underwent ultra-sonication to remove invisible air bubbles. The membrane casting step was followed by a casting knife with 200 μm thickness on a glass plate. Finally, casted membrane was immersed into a water bath to promote phase inversion between the NMP solvent and the DI water.

Different types of PES membranes (PES2, PES3, PES4, and PES5) were prepared under different synthesis conditions (Fig. 1). The prepared casting solution was poured on the PET substrate in the PES2 membrane without any treatment with additives. For the synthesis of PES3 membrane, the PET substrate was wetted with NMP before pouring the casting solution onto the PET substrate. NMP was used as an adhesive between the casting solution and the PET substrate. In PES4, 1 wt.% of PVP as a pore former was added into the casting solution. Both adhesive and pore former were applied for the synthesis of PES5 membrane. All the synthesized membranes were stored in DI water for one day to remove solvent remaining inside the membrane prior to further experiments.

Lab-scale cross-flow UF filtration was carried out to measure the water flux of the PES membranes. The effective membrane area of each cell was 18.56 cm², the temperature was kept at 20°C, and the circulation flow was maintained at 400 ccm. All the membranes were stabilized at a trans-membrane pressure (TMP) of 4.14 bar (60 psi) for 4 hr, and then the TMP was kept at 4.14 bar for 3 h to examine the water flux. The schematic of the lab-scale cross flow UF set-up is illustrated in Fig. 2. To evaluate the membrane selectivity with regard to rejection of divalent ions, the following equation was employed:

where R is the CaSO₄ rejection, and C_f and C_p are the concentrations of the feed water and permeate, respectively. The CaSO₄ rejection rate was calculated by measuring the conductivity and converting the values using a correlation curve between the conductivity and concentration of CaSO₄ (Fig. 3.). Fig. 3 shows that CaSO₄ was fully dissolved in DI water with concentration less than 200 ppm.

The membrane thickness and pore structure were measured by scanning electron microscope (SEM). The membrane was fully dried at 40°C for one day in an oven and immersed in liquid nitrogen. The frozen membrane was then broken, and its cross-section was measured with SEM. The surface morphology and roughness of the prepared membranes were measured with an AFM. The analyzed areas were 2 μm × 2 μm, and the membrane surface roughness was expressed as root mean square, R_q (Table 2). The AFM experiment was conducted in contact mode using contact cantilever tips.

Substrate effects were observed in the PES2 (with PET substrate) and PES membranes (without PET substrate). The thickness of PES2 membrane was thinner (109 μm) than that of the PES membrane (171 μm) due to the penetration of the casting solution into the PET substrate during casting step (Fig. 4. (a), (b)). The pure water permeation (PWP) of the PES2 membrane was 5.63 L/m²hr (LMH), whereas it was 3.24 LMH for the PES membrane (Fig. 5.). Therefore, the use of the PET substrate resulted in a thinner PES layer formation, which enhanced the PWP of PES2 membrane. In addition, only narrow finger-like pores were observed on the PET substrate which results in thin PES layer formation, since that macrovoids were formed in the substrate matrix due to the PES solution penetration into the PET substrate. This thin layer and enlarged inside pore structure contributed to enhance PWP simultaneously.

Adhesive was used in the PES3 but not the PES2 membrane. The PES3 membrane was significantly thinner (46 μm) than the PES2 membrane (109 μm) due to accelerated penetration of the casting solution into the PET substrate during casting step caused by pretreatment of PET substrate with adhesive (Fig. 4 (b), (c)). The casting solution readily penetrated the adhesive-wetted PET substrate. The PWP of the PES3 membrane was 7.55 LMH PWP, whereas that of the PES2 membrane was 5.63 LMH. The thinner membrane of the PES3 membrane with the adhesive enhanced the PWP (Fig. 5).

The effects of the pore former, PVP, on the PES membrane were determined by observing the difference in the performance of the PES2 and PES4 membranes. Previous researches reported that pore former, such as PVP, produced a thin top dense layer, straight finger-like pores, and a thick membrane due to increased hydrophilicity and casting solution viscosity [16–20]. In this experiment, 1 wt.% PVP was used because a previously reported study showed that 1 wt.% of PVP can decrease the thickness of the top dense layer without significantly changing the size of the surface pores [21]. The PES4 membrane, in which pore former was used in the casting solution, had more straight finger-like pores than the PES2 membrane (Fig. 4 (b), (d)). In addition, the thickness of the membrane increased from 109 μm to 122 μm due to the increased viscosity of the casting solution after adding the pore forming agent. Although the addition of the pore former only slightly increased the thickness of the PES4 membrane, the PWP of the PES4 membrane was 11.10 LMH, whereas that of the PES2 membrane was 5.63 LMH (Fig. 5). It can be explained that more straight finger like pore formation caused by addition of pore former increased PWP.

Fig. 6 shows the segmented rough surface of the PES4 membrane, which had pore former in the casting solution, compared to that of the PES2 membrane. The surface roughness of the PES4 membrane was 22.87 ± 1.20 nm, whereas that of the PES2 membrane was 17.27 ± 2.53 nm by AFM analysis (Table 2). As addition of pore former, surface roughness of membrane was increased which was result from the increased number of surface pores [21].

The PES5 membrane was formulated with adhesive, pore former, and substrate. PES5 membrane had the highest PWP (27.42 LMH), among all the prepared membranes (Fig. 5). The PWP of the membranes as well as membrane selectivity were measured using 0.001 M (200 ppm) CaSO₄ solution as feed water (Fig. 7). Although the PWP membranes containing the pore former was enhanced due to the increased numbers of straight finger-like pores, the addition of the pore former slightly decrease CaSO₄ rejection. Comparison of the PWP and CaSO₄ rejection of the PES4 and PES5 membranes showed that the use of adhesive did not influence the properties of active pores. Therefore, the synthesis conditions of the PES5 membrane resulted in dramatically enhanced PWP without sacrificing membrane selectivity, suggesting that these conditions could used to produce a suitable support layer for TFC membranes.