1. Introduction
In recent years, with the growth of population and industrial development, climate change and pollution have become increasingly serious, leading to a particularly prominent water shortage problem in developing countries [
1,
2,
3,
4]. Currently, approximately 30% of the global population is experiencing inadequate access to safe drinking water, and it is projected that this figure will increase to 66% by 2025 [
5,
6]. Water contamination has had a significant impact on the daily lives of individuals, as the majority of identified heavy metals present substantial health hazards [
7,
8]. Common water treatment technologies include chemical coagulation [
9], gravity separation [
10], and filtration [
11]. However, these methods are plagued by challenges such as intricate operational procedures, elevated operational expenditures, emissions of secondary pollutants, and suboptimal treatment efficacy [
12,
13]. Due to its environmentally friendly properties, high efficiency, cost-effectiveness, and ease of operation, NF has been widely utilized in the field of water treatment, encompassing seawater desalination, drinking water purification, softening water, wastewater treatment, and recycled water utilization [
14,
15,
16,
17,
18].
At present, the predominant type of NF membranes in commercial use is composite membrane. The composite membrane consists of a porous base membrane and an ultra-thin separation layer, wherein the support layer and separation layer are typically fabricated from distinct materials [
19,
20,
21]. Conventional NF membranes typically feature a separation layer composed of PA, which is prepared through interfacial polymerization (IP) using amine monomers and carbonyl chloride monomers on a porous support layer [
22,
23]. The flux of nanofiltration membrane prepared using the conventional IP method, however, is relatively low. For instance, the widely utilized NF90 NF membrane exhibits a flux rate of merely 7–8 LMH/bar [
24]. One of the primary factors is that the thickness of the PA separation layer, prepared by the conventional IP method, results in increased resistance to permeation and diminished permeate flux [
25]. In addition, the hydrophilicity of PA is insufficient, which also has an adverse effect on water penetration [
26].
To tackle this issue, researchers have endeavored to enhance permeability by selecting suitable monomers. Yuan et al. [
27] synthesized 1,2,3,4-cyclobutane tetracarboxylic acid chloride (BTC) as a novel oil-phase monomer. In comparison with the commonly utilized trimesoyl chloride (TMC), BTC exhibits a smaller molecular volume, enabling the molecules situated at the interface to react with more water-phase monomers and form a denser PA NF membrane. The resulting separation layer thickness measures only 15 nm. This ultra-thin separation layer structure has significantly enhanced the flux and rejection performance of the NF membrane. Li et al. [
28] chose 3,3′,5,5′-biphenyl tetra acyl chloride (mm-BTEC) as the organic phase monomer and piperazine (PIP) as the initiator for the fabrication of NF membranes. Due to the higher carboxyl content of mm-BTEC, the resulting separation layer exhibits enhanced hydrophilicity, leading to improved NF flux. However, the mm-BTE molecule possesses a larger volume, resulting in slightly increased pore sizes within the separation layer, thereby leading to a reduction in the separation efficiency of the NF membrane. In addition to selecting the appropriate reactive monomer, surface modification represents a straightforward approach for enhancing membrane permeability. Huang et al. [
29] modified the surface of the PA membrane with betaine, resulting in a distinctive leaf-like structure that increased the permeable area and enhanced hydrophilicity. This modification led to a 1.2-fold increase in permeability compared to the unmodified membrane while maintaining excellent rejection of Na
2SO
4. In addition, when the surface of the porous support layer contains large holes, it will lead to pore intrusion and the defects of the separation layer, which is also one of the reasons for the small permeate flux [
30,
31,
32]. Fang et al. [
33] introduced a novel covalent organic frameworks (COFs) material on the support layer to mitigate the permeation of the PA layer, resulting in a double-layered modified nanocomposite membrane with a pure water permeate flux of 27.3 LMH/bar, which is three times that of the unmodified membrane, and the Na
2SO
4 rejection rate of 98.40%. Bai et al. [
34] fabricated a cellulose nanocrystal (CNC) intermediate layer on the support layer using a two-step method, followed by polymerization of PA on the intermediate layer and subsequent surface modification with polydopamine (PDA). After undergoing this two-step modification process, the NF membrane exhibits a rejection rate of 98.2% for Na
2SO
4, with a permeate flux of 23.1 LMH/bar.
In recent years, PVA has been widely employed in the fabrication of NF membranes due to its abundant hydroxyl groups and excellent separation performance. Zhang et al. [
35] utilized PVA and modified graphene oxide (GO) as the intermediate layer, followed by cross-linking with glutaraldehyde (GA) for post-cross-linking. Subsequently, they fabricated a NF membrane with a PA layer thickness of merely 15 nm via IP, leading to a significant enhancement in the separation performance of the NF membrane. Under optimized conditions, the membrane achieved an impressive salt rejection rate of 99.70% for Na
2SO
4. Zhu et al. [
36] fabricated a NF membrane featuring a PA layer thickness of merely 9.6 nm, utilizing polyether sulfone (PES) as the support layer and PVA as the intermediate layer. The NF membrane incorporating a PVA intermediate layer exhibited a high permeate flux of 31.4 LMH/bar and an impressive rejection rate of 99.40% for Na
2SO
4. Januario et al. [
37] reported a remarkable retention rate exceeding 99.9% for dyes utilizing a microfiltration (MF)/PVA60+GO1+PVA30 membrane, which was constructed through layer-by-layer (LBL) self-assembly on the surface of the MF membrane with PVA, GO, and PVA applied sequentially. The flux recovery rate was approximately 80%, while the retention rate remained above 94% even after multiple cycles of use.
To enhance the permeate flux of the NF membrane, in this study, we developed a novel preparation method for NF membranes with PVA/PA/PVA sandwich structure separation layers. First, a PVA intermediate layer is applied to the surface of the polysulfone (PS) ultrafiltration membrane. This layer not only exhibits certain separation performance but also provides a dense support surface, which can greatly inhibit the defects of the PA separation layer. Subsequently, a PA separation layer was prepared on the surface of the PVA layer, and the separation performance of the composite membrane was significantly improved by the PA layer. Finally, a secondary PVA layer is applied to the PA surface, leveraging PVA’s excellent hydrophilicity and specific separation properties to further enhance membrane performance. Moreover, the PVA layer can bolster the membrane’s resistance to pollution and tolerance to residual chlorine. To further improve the permeability of the composite membrane, the PA layer was synthesized via IP using PIP and triethanolamine (TEOA) as co-solvents with TMC to augment the hydrophilicity of the PA layer [
38]. The effect of the concentration of PVA casting solution on the separation performance of the composite membrane was studied, and the membranes’ permeation performance, rejection performance, anti-pollution performance, and resistance to residual chlorine were comprehensively investigated.
2. Materials and Methods
2.1. Materials
The flat-sheet nonwoven-reinforced PS ultrafiltration membrane was procured from AnD Membrane Separation Technology Engineering Co., Ltd. (Beijing, China). PIP (99.5% purity) was procured from Beijing Bailingwei Technology Co., Ltd. (Beijing, China). N-hexane (98% purity) was procured from Tianjin Kangkede Technology Co., Ltd. (Tianjin, China). TMC (98% purity), PVA (99.5% purity), anhydrous ethanol (99.9% purity), sulfuric acid (H2SO4, 99.8% purity), sodium dodecyl sulfate (SDS, 99.7% purity), sodium hypochlorite (NaClO, 99.7% purity), and CR (98% purity) were procured from Shanghai McLin Biochemical Technology Co., Ltd. (Shanghai, China). Anhydrous sodium sulfate (Na2SO4, 99.7% purity), sodium chloride (NaCl, 99.7% purity), magnesium sulfate (MgSO4, 99.7% purity), and sodium alginate (SA, 99.5% purity) were procured from Guoyao Group Chemical Reagents Co., Ltd. (Shanghai, China). Magnesium chloride (MgCl2, 99.7% purity), trimellitic anhydride (TMA, 98% purity), and VB (80% purity) were procured from Aladdin Reagents Co., Ltd. (Shanghai, China). TEOA (97% purity) was procured from Xilong Chemical Industry Co., Ltd. (Shantou, China). Barium chloride (BaCl2, 97% purity) was procured from Beijing Mairaida Technology Co., Ltd. (Beijing, China). Neutral red (NR) was procured from Shanghai Yuanye Biological Technology Co., Ltd. (Shanghai, China). Rhodamine B (Rh B, 95% purity) was procured from Beijing Yinghai Fine Chemical Factory (Beijing, China). Eriochrome black T (EBT, 90% purity) was procured from Modern East Science and Technology Development Co., Ltd. (Beijing, China). Bovine serum albumin (BSA, 98% purity) was obtained from Beijing Huawi Ruike Chemical Technology Co., Ltd. (Beijing, China).
2.2. Membrane Preparation
2.2.1. Preparation of PVA Interlayer
The PVA interlayer was fabricated using a gradient cross-linking approach [
39]. Initially, the PS ultrafiltration membrane was securely positioned between a glass plate and a silicone rubber spacer. Subsequently, the cross-linking solution, comprising a blend of ethanol and deionized water in an equal volume ratio (1:1), containing 0.2%
w/
v TMA, 0.5%
w/
v SDS, and 0.05%
w/
v H
2SO
4, was applied onto the surface of the ultrafiltration membrane. After allowing it to stand for 10 min to remove excess solution, the sample was subjected to electric hot-air drying at 35 °C for 8 min. After undergoing heat treatment, the PVA solution at concentrations of 0.05%
w/
v, 0.10%
w/
v, 0.15%
w/
v, 0.20%
w/
v, and 0.25%
w/
v was uniformly applied onto the membrane surface, respectively, followed by removal of excess solution after a standing period of 5 min. Subsequently, membranes were returned to an electric hot-air drying oven at 80 °C for 10 min to yield a support membrane with a PVA intermediate layer. The membranes mentioned above were designated as PVA-0.05, PVA-0.10, PVA-0.15, PVA-0.20, and PVA-0.25 based on the varying concentrations of PVA.
2.2.2. Preparation of PA/PVA Membrane
Prepare a 0.05%w/v PIP solution and a 0.95%w/v TEOA solution in water, apply them onto the membrane surface with an intermediate layer of PVA, allow to stand for 2 min, remove excess liquid, and subject to heat treatment at 35 °C for 8 min. After the removal process, a 0.2%w/v TMC hexane solution is uniformly applied onto the membrane surface and allowed to stand for 30 s before promptly draining off the excess liquid and subsequently rinsing the membrane with hexane multiple times. The prepared membrane underwent heat treatment at 60 °C for 15 min to yield the PA/PVA NF membrane. The membranes mentioned above were designated as PA/PVA-0.05, PA/PVA-0.10, PA/PVA-0.15, PA/PVA-0.20, and PA/PVA-0.25.
2.2.3. Preparation of PVA/PA/PVA Membrane
In accordance with the steps detailed in
Section 2.2.1., a second PVA separation layer was fabricated on the surface of the PA/PVA membranes; the concentration of PVA solution used was the same as that used in the preparation of the intermediate PVA layer. The membranes prepared were designated as PVA-0.15/PA/PVA-0.15, PVA-0.20/PA/PVA-0.20, and PVA-0.25/PA/PVA-0.25 based on the concentration of PVA used in the preparation process.
Figure 1 illustrates the procedure for fabricating the PVA/PA/PVA sandwich structure NF membrane.
2.2.4. Preparation of Control TFC Membrane
A control TFC membrane was fabricated using conventional IP. Prepare a 1%w/v PIP aqueous solution, apply it to the surface of the PS substrate, and allow it to stand for 2 min. Subsequently, remove the excess liquid and uniformly apply the 0.2%w/v TMC hexane solution onto the membrane surface, allowing it to sit for 30 s before promptly removing any surplus liquid and repeatedly rinsing with n-hexane. The prepared membrane was subsequently subjected to a heat treatment at 60 °C for 15 min to yield the control TFC membrane.
2.3. Characterization of Membrane Physicochemical Properties
Attenuated total reflection Fourier-transform infrared (ATR-FTIR, Nicolet 6700, Thermo Fisher Scientific, Waltham, MA, USA) and X-ray photoelectron spectroscopy (XPS, Escalab 250Xi, Thermo Fisher Scientific) were employed for the analysis of the surface chemical structure and composition of the membrane. Membrane morphologies were characterized by scanning electron microscopy (SEM, SU8600, Hitachi, Tokyo, Japan). To determine the surface roughness of the membranes, atomic force microscopy (AFM, Dimension Icon, Bruker, Billerica, MA, USA) in tapping mode was used. The water contact angle of membrane surfaces was measured using a contact angle goniometer (WCA, OCAH 200, Data-physics, Filderstadt, Germany).
2.4. Membrane Performance Testing
2.4.1. Permeability and Rejection Performance of the Membranes
This study employed a crossflow apparatus to evaluate the separation and permeability characteristics of the membrane [
40]. In order to minimize experimental error, each membrane was measured three times, and the average was calculated. Prior to the test, the membrane was rinsed with ultrapure water and operated at 0.8 MPa for 30 min to stabilize its performance. Subsequently, under a pressure of 0.8 MPa, the permeate passing through the membrane per unit time was collected, and the filtration flux (
J) of the NF membrane was determined using Formula (1) as follows:
where
J denotes the filtration rate (LMH),
S denotes the effective membrane area (m
2), Δ
t is the duration for collecting the filtrate (h), Δ
m(g) is the increase in filtrate mass during Δ
t, and
ρ stands for the density of the filtrate (approximately 1 g/mL when the salt concentration is low).
The permeate flux refers to the volume of liquid that permeates through a unit area of membrane under unit pressure in a given time period. The relationship between permeate flux and filtration flux is depicted in Equation (2) as follows:
where ∆
P denotes the test pressure (bar), while the permeate flux is expressed in units of LMH/bar.
The electrical conductivity of the feed liquid and the filtrate was measured using a conductivity meter [
35]. As the electrical conductivity of a dilute, strong electrolyte solution is directly proportional to its concentration, the ratio of conductivities can be approximated by the concentration ratio. The feed liquid was maintained at a concentration of 0.5 g/L, and the test was conducted at an ambient temperature. The equation for calculating the rejection rate is presented in Equation (3) as follows:
where
R denotes the rejection rate (%),
CP denotes the concentration of the filtrate, and
CF denotes the concentration of the feed liquid.
Meanwhile, NR, Rh B, VB, EBT, and CR (0.1 g/L) were selected to evaluate the dye retention performance of the composite membrane. The operating pressure is 0.8 MPa, and the concentrations of dye in the feed liquid and permeate are individually measured, and the dye rejection rate of the composite membrane is calculated using Equation (3). The dye concentration is determined utilizing a spectrophotometric method [
41].
And static adsorption tests were carried out on the dye [
42]. For the static adsorption tests, the membrane samples (50.24 cm
2) were immersed in dye solutions (0.1 g/L,
Ci) for 2 h. Equilibrium concentrations of dye (
Ce) were measured by UV-vis spectrophotometry. The adsorbed mass of dye per unit area of membrane (
Q, mg/cm
2) was calculated using Equation (4) as follows:
where
A is the effective membrane area (cm
2),
V is the volume of dye solution (mL), and
Ci and
Ce are the initial and equilibrium dye concentrations (g/L), respectively.
2.4.2. Antifouling Performance
The anti-pollution performance test was carried out at 25 °C and 0.8 MPa using BSA and SA solution (0.5g/L). The test was conducted for 1.5 cycles with the following cycle process: deionized water solution was used as the feed solution to obtain the filtration flux J
0 during the initial 3 h. Then, the BSA/SA solution was applied for 6 h, during which the filtration flux
Jt was measured at 0.5 h intervals. And subsequently replaced with deionized water to rinse the membrane for 3 h, and the measured filtration flux was denoted as
Jr. The obtained time-dependent normalized flux (
Jt/
J0) profiles were used to investigate the fouling behaviors of the tested membranes. The flux recovery ratio (FRR, %) and flux decline ratio (FDR, %) were determined through the following Equations (5) and (6), respectively:
The higher FRR value indicates the higher cleaning efficiency, while the lower FDR value means the better antifouling property.
2.4.3. Chlorine Resistance Test
The membrane was dipped into a 1.0 g/L NaClO solution (pH = 4) for 10 h. The performance of the membrane was measured every 2 h. In order to ensure the concentration of NaClO in the solution remained unchanged, we replaced the soaked solution of NaClO every 5 h [
43,
44].
4. Conclusions
In this study, PVA/PA/PVA TFC NF membranes with sandwich structure separation layer were prepared, exhibiting exceptional separation performance. By employing PS ultrafiltration membrane as the support layer, a series of PVA/PA/PVA membranes were fabricated with sequential preparation of PVA, PA, and PVA separation layers. The influence of PVA concentration on separation performance was investigated, and the removal efficiency for dyes, anti-pollution capability, and residual chlorine resistance were also conducted. The experimental results demonstrate that the permeate flux and Na2SO4 rejection rate of the TFC NF membrane increased following preparation of a PVA layer on the surface of PA/PVA membrane. As the PVA concentration in the cast solution increased from 0.15% to 0.25%, the permeation flux of PVA/PA/PVA membranes decreased from 26.27 LMH/bar to 19.33 LMH/bar, while the rejection rate of Na2SO4 slightly increased, reaching a maximum of 96.62%. The separation performance of PVA-0.20/PA/PVA-0.20 is better compared with the control TFC membrane; pure water flux increased from 5.62 LMH/bar to 22.05 LMH/bar, demonstrating a significant enhancement, while the rejection rate of Na2SO4 was basically the same. The rejection rates of CR and VB were surpassed by 99.9%, while Rh B and EBT were exceeded by 97%. Additionally, NR exhibited a rejection rate of 84.03% due to its relatively low molecular weight. Moreover, the PVA-0.20/PA/PVA-0.20 membrane exhibited higher resistance to fouling and chlorine residual compared to the TFC reference membrane. Even though the current methodology for fabricating the PVA/PA/PVA sandwich structure membrane developed by our institute is relatively complex, posing challenges for its large-scale production. However, due to its superior comprehensive performance, it has great practical application prospects in the fields of water softening, reclaimed water reuse, and dyeing wastewater treatment.