Rapid fabrication of selective barriers through layer-by-layer self-assembly of non-stoichiometric polyelectrolyte complexes for salt and dye fractionation

Abstract

Layer-by-layer (LbL) self-assembly of oppositely charged polyelectrolytes (PEs) is a flexible and environment-friendly route for the construction of functional surfaces and membranes. However, LbL assembly is time-consuming and requires multiple layers to fabricate a separation membrane. In this context, we report the LbL self-assembly of non-stoichiometric PE complexes (PECs) for the rapid fabrication of molecular selective barrier layers. PECs containing strong charge groups and relatively large dimensions compared to those of uncomplexed PEs facilitate the rapid formation of barrier layers on base substrates. The Mem-PEC3 membrane prepared by the LbL self-assembly of the PECs (total three layers) shows 99.4% to >99.9% interception of several dyes in the presence or the absence of monovalent/divalent salts. Mem-PEC3 can fractionate salt (50 g L⁻¹) and DR-80 (1 g L⁻¹) with a salt-to-dye separation factor of ∼3300 at 1.38 bar of applied pressure. This membrane exhibits >99% flux recovery after the fractionation of salt and dye. The membrane exhibits stable performance at high salt concentrations and at a pH range of 3–12. PECs form strong ion pairs during LbL assembly starting from the first layer, which tightly hold the entire assembly on the base substrate. This work provides rapid electrostatic LbL assembly of premade PECs, which is hitherto unknown to create a molecular selective barrier.

---

1. Introduction

Molecular selective membranes have attracted widespread attention due to their applications in fractionation and resource recovery. Barrier layers with extremely low salt interception and very high organic interception ability have great potential for selective separation under low applied pressure. Selective separation is important not only for resource recovery but also for the smooth water recovery from a mixed feed. Tight ultrafiltration (TUF) and loose nanofiltration (LNF) membranes act as barriers for organics and dyes while preferentially permeating salts.^1–7 The selective separation of salts and organics by these membranes makes the separation system simple in terms of resource recovery under low applied pressure.^1,3 TUF membranes were prepared by modified non-solvent induced phase separation (NIPS). Both modified NIPS and interfacial polymerization (IP) processes were applied for the fabrication of LNF membranes.^1–7

This work focuses on the fabrication of selective barriers through layer-by-layer (LbL) self-assembly. LbL self-assembly of oppositely charged polyelectrolytes (PEs) is an environment-friendly and versatile process for the construction of thin functional films and membranes.^8–19 LbL self-assembly provides modulation of membrane charge and performance.^8,14 However, LbL assembly requires high concentrations of PEs and multiple layers to obtain the membrane. The process is time-consuming and often requires additional crosslinking for the fabrication of stable barrier layers.¹⁷ In this regard, concentration polarization-induced LbL self-assembly drastically lowered the required layer numbers for the fabrication of TUF (total three layers) and LNF (total five layers) membranes.¹² Conventional (dipping) LbL assembly could be beneficial for the rapid fabrication of molecular selective membranes on a large scale (roll-to-roll) if the required layer number could be reduced.

Earlier, a dope solution containing ion-exchanged poly(sodium-4-styrenesulfonate) (PSSNa) and a bore solution containing polycations were used to fabricate a hollow fiber NF membrane.¹⁹ This process required 4% w/w and 5–10% w/w of the PEs in the dope and bore solutions, respectively. The casting of oppositely charged PEs in the presence of excess salts followed by the removal of the salt gave an NF or a UF membrane.^20–22 The soluble PEC at certain pH values was casted onto a support followed by drying the membrane for pervaporation applications.^23,24 The PEs used in the casting solutions were either weak PEs or a combination of weak and strong PEs. The reported casting procedures needed high concentrations of PEs. Furthermore, the LbL assembly needed 6–12 or more layers to obtain NF or LNF membranes.^8,11–15

Herein, we propose the LbL self-assembly of oppositely charged premade non-stoichiometric PECs for the rapid fabrication of a molecular selective membrane. Unlike uncomplexed PEs, PECs disperse in water. Non-stoichiometric PECs were prepared by combining oppositely charged PEs with one PE in excess.^25–30 Non-stoichiometric PEC contains abundant free charge groups on the outer surface. These charge groups of a PEC form ionic bonds with the oppositely charged PEC or PE. The dimension of a PEC is greater than that of the uncomplexed PEs. The non-stoichiometric PECs carry a greater charge density.^27,28 As a result, the rapid LbL assembly of the oppositely charged PECs is expected to form a thicker layer than that of the uncomplexed PEs. The LbL assembly of the oppositely charged PECs was applied for the fabrication of functional films. However, to the best of our knowledge, there is no report on the fabrication of molecular selective membranes by the conventional LbL assembly of the oppositely charged PECs. The characteristic properties such as high charge density, dimension, and diverse structures (stoichiometry) of the non-stoichiometric PECs inspired us to apply specially designed PECs (containing strong charge groups) for the construction of membranes for the selective separation of salts and dyes.

In this work, oppositely charged PECs were prepared by combining quaternized poly(vinyl imidazole) (PVIm-Me) and poly(styrene sulfonic acid) sodium salts (PStSO₃Na) in non-stoichiometric proportions in the presence of salts. PVIm-Me and PStSO₃Na PEs were selected for preparing PECs, as these PEs form multilayers which are stable at a pH 3–12 and at a high salt concentration.^11,12 Herein, we demonstrate that the LbL assembly of the premade PECs on the cationic base membrane substrate rapidly (total two or three layers) affords a selective layer for high-efficiency salt and dye fractionation. The effects of the stoichiometric ratio (mol/mol of repeat unit) of the oppositely charged PEs and their concentration used for the PEC preparation on the membrane performance were evaluated. The mechanism of rapid formation of LbL-assembled membranes and the role of the PECs are proposed and experimentally validated. The salt, pH and pressure stability of the membrane were evaluated in detail. The mechanism of firm anchoring of the PEC layers on the charge base membrane substrate was probed by a leaching experiment with an organic solvent as well as by evaluating the stability of the membrane. This work provides a new insight for the quick fabrication of LbL-assembled membranes for the high-efficiency fractionation of dyes and salts.

2. Experimental section

2.1. Materials

The sources of materials, structures and molecular weights of the dyes (Table S1, ESI†), synthesis of PEs, and preparation of base substrates are included in the ESI.†

2.2. Preparation of non-stoichiometric PECs

PEs were synthesized and characterized (ESI†).^11,12 Stock solutions of PVIm-Me and PStSO₃Na were separately prepared in deionized water and then NaCl was added. Typically, PStSO₃Na (0.36 g, 7 mM of StSO₃Na) was solubilized in deionized water (250 mL) under continuous stirring at room temperature for 12 h. Then, NaCl (7.3 g, for 0.5 M) was added into it. Similarly, a PVIm-Me (0.42 g, 7 mM of VIm-Me group) solution was prepared (NaCl = 0.5 M). The PECs were prepared by varying the ratio (mol/mol equivalent to monomer unit) of the PEs. The major PE was taken in the flask and the minor PE was added slowly into the solution of the former. A typical example for the preparation of 5 : 1 (mol/mol) of PStSO₃Na/PVIm-Me PEC is as follows: a PStSO₃Na (50 mL, 7 mM) solution was taken in a flat-bottom flask and allowed to stir at 700 rpm at room temperature. Next, the PVIm-Me solution (10 mL, 7 mM) was added dropwise into the flask under continuous stirring. The addition of second PE is completed within 10 min. In a similar way, a PVIm-Me/PStSO₃Na PEC dispersion was prepared. The concentration of stock solutions and ratios of the PEs (mol/mol repeat unit) were varied to obtain different PECs. The PEC formed by the 5 : 1 (mol/mol of repeat unit) combination of PStSO₃Na (excess) and PVIm-Me is abbreviated as PStSO₃Na/PVIm-Me-5 : 1. The PEC of PVIm-Me (excess) and PStSO₃Na is abbreviated as PVIm-Me/PStSO₃Na-5 : 1 (Scheme 1A and B). The final repeat unit concentrations of PStSO₃Na and PVIm-Me for the preparation of PStSO₃Na/PVIm-Me-5 : 1 were 5.84 mM and 1.16 mM, respectively. PECs were similarly prepared by varying the ratio and concentration of PEs.

Scheme 1 Schematic representation of the preparation of non-stoichiometric (A) anionic and (B) cationic PECs. (C) Electrostatic LbL self-assembly of the PECs on a cationic base substrate and a cartoon showing the probable structure of the LbL-assembled separation layer.

2.3. LbL assembly of the PECs for the fabrication of molecular selective barrier layers

The base substrate containing quaternized amine groups was fabricated (ESI†).^11,12 The LbL assembly of PECs was performed as follows. The base substrate (20 cm × 20 cm) was fixed in a frame. First, an anionic PStSO₃Na/PVIm-Me-5 : 1 PEC dispersion (60 mL, 0.5 M NaCl) was poured on the support substrate and kept for 10 min. The PEC was then removed. The membrane was rinsed twice with a 0.5 M NaCl solution (50 mL) for 2 min each to remove the unbound PEC. This membrane was designated as Mem-PEC1. Next, PVIm-Me/PStSO₃Na-5 : 1 was applied on the Mem-PEC1 surface followed by washing with 0.5 M NaCl, resulting in Mem-PEC2. The third-layer assembly was performed similarly by applying the PStSO₃Na/PVIm-Me-5 : 1 dispersion on Mem-PEC2. The membrane containing total three layers of the PECs is abbreviated as Mem-PEC3. Scheme 1A–C show the preparation of PECs, LbL-assembled membrane and probable arrangement of the PECs on the membrane substrate.

2.4. Characterizations of the PECs

The PECs were characterized by dynamic light scattering (DLS, Malvern Zetasizer Nano ZS90) measurement for the determination of particle size and zeta potential. The PECs were also characterized by using a scanning electron microscope (SEM, JEOL, JSM-7100F) and an atomic force microscope (AFM, Ntegra Aura instrument, NT-MDT). After dialysis and removal of NaCl, a drop of five-fold diluted dispersion was placed on a silicon wafer. The dried sample was gold-coated (using a LEICA EM ACE 200 gold coater), and then SEM analysis was performed.

2.5. Characterizations of the membranes

The membrane pieces were washed with deionized water and air-dried for the characterizations. These membrane samples were coated with gold, and then the morphology of the membranes was visualized by SEM. Energy-dispersive X-ray (EDX, 20 kV, couple with SEM) analysis was performed. XPS analysis on the membrane surface was performed using a Thermo-Scientific NEXSA instrument (Al K-Alpha, 1486.6 eV). The membrane was separated from the fabric and then the membrane was broken under liquid nitrogen for the cross-sectional SEM examination. AFM analysis was performed to observe the surface topology and roughness. The apparent thickness of the barrier layer of the membrane was determined by gently detaching the membrane from the fabric support and submerging it in isopropanol (IPA) for 5 min. A silicon wafer wetted with IPA was placed on a glass slide, and then the membrane was carefully placed on a silicon wafer using tweezers. The top surface of the membrane faces the silicon wafer surface (facing upside down). The base substrate was leached out by the dropwise addition of DMF. After leaching, the crosslinked layer adhered to the silicon wafer. The isolated layer was scratched with a needle gently without damaging the silicon wafer substrate and subjected to AFM analysis. The height difference between the exposed silicon wafer and the film surface was determined using the Gwyddion software.

2.6. Determination of the permeate flux, MWCO, and pore size of the membranes and evaluation of the dye rejection efficacy and fractionation of the dye and salt

The filtration experiments were conducted in a crossflow system equipped with four cells (active area = 14.5 cm²) and a pump (flow rate = 40–45 L h⁻¹). The permeate flux, MWCO and pore size distribution of the membranes were determined as reported earlier (ESI†).^11,12 Various dye solutions were permeated through the membranes for the determination of the rejection efficacy of the membranes. The concentration of the dyes was varied. The selective separation of the dye and salt was performed and the separation factor was determined (ESI†). Diafiltration operation was performed to evaluate the dye recovery efficacy of the membrane (ESI†).^11,12

2.7. Pressure, salt and pH stability of the PEC self-assembled layer and dye antifouling property

The pressure stability of the membranes was evaluated by permeating the feed containing a mixture of dye (1 g L⁻¹) and salt (NaCl, 50 g L⁻¹) at different applied pressures. Furthermore, an aging test was performed by permeating saline water (50 g L⁻¹) at an applied pressure of 17.2 bar for 24 h. Next, the dye rejection and salt-to-dye separation factors were determined.

The pH stability was determined as follows: dye rejection at pH 7 was first measured. Next, water of pH values of 3, 10 and 12 was permeated through the membrane swatches independently for a total of 48 h at an applied pressure of 1.38 bar and a temperature of 27 °C. After that, the dye rejection and permeate flux were determined again at pH 7. The dye rejection, permeate flux and salt-to-dye separation factors before and after the pH treatment were evaluated. Furthermore, the solution containing the dye (0.2 g L⁻¹ and 1 g L⁻¹) and the salt (50 g L⁻¹) at pH 12 was permeated through the membrane swatches for a total of 30 h. The permeate flux and dye rejection performances were evaluated with the filtration time.

To investigate the potential of the membrane in textile wastewater treatment, simulated dye wastewater was prepared (CR 0.2 g L⁻¹, NaCl 1 g L⁻¹, CaCl₂ 0.01 g L⁻¹, K₂Cr₂O₇ 0.001 g L⁻¹, pH = 9).³¹ This simulated dye solution was permeated through Mem-PEC3 for 24 h at 1.38 bar to evaluate the separation efficacy. The dye antifouling property of the membrane was evaluated (ESI†).^11,12

2.8. Adsorption capacity

The representative Mem-PEC3 membrane swatches were dipped in 50 mL of a CR and DR-80 aqueous solution (0.5 g L⁻¹) separately and then put in an incubator shaker set to 25–27 °C for 24 h at a shaking speed of 200 rpm. The following formula was used to calculate the dye adsorption capacity (q_e):
where M, V, C₀ and C_e are the membrane mass (g), volume of dye solution (mL), starting dye concentration and equilibrium dye concentration (mg L⁻¹), respectively.

2.9. Statistical analysis

Average rejection and permeate flux with standard deviations were calculated from 12 membrane swatches of three independent batches of the samples.

3. Results and discussion

3.1. Preparation of stable non-stoichiometric PECs

First, PStSO₃Na/PVIm-Me and PVIm-Me/PStSO₃Na PECs containing strong anionic and cationic groups were prepared (Scheme 1A, Experimental section). The PECs prepared by the 4 : 1 to 6 : 1 mol/mol repeat units of the oppositely charged PEs were stable even after storage for 8 weeks in the presence of NaCl. However, the PECs of 2 : 1 and 3 : 1 stoichiometry underwent coagulation after 48 h of storage, while the 1 : 1 PEC formed flocculation and precipitation after 15 min of storage (Fig. 1A and B). The extent of ion pair formation increased, which led to subsequent phase separation and agglomeration of the PEC formed by 1 : 1 combination of the oppositely charged PEs. The primary PEC particles formed by 1 : 1 to 3 : 1 PE combination underwent agglomeration due to the decrease of free charge groups and inter-particle repulsion. The excess charge minimizes the agglomeration of the particles through the electrostatic repulsion for PECs of 4 : 1 to 6 : 1 PE combination. Our observation is consistent with the earlier reports for different PECs.^32,33 The effect of salt on the dispersion stability of the PECs was evaluated. The PEs attain flexible coil conformation at a certain range of ionic strength, which makes the conformation adaptation easier to minimize coagulation. However, at higher NaCl concentrations (>2 M), predominant PEC charge screening occurs. This leads to the lowering of electrostatic repulsion among the PEC particles and subsequent coagulation upon storage (Fig. S1A, ESI†).³³ However, solubilization of the PECs did not occur even in the presence of 4 M NaCl as observed visually. The salt stability of the PEC was further assessed by the conventional turbidimetry measurement by a UV-visible absorbance analysis (Fig. S1B, ESI†).³⁴ The critical salt concentration (CSC) was earlier used to assess the stability of the PECs. The absorbance of the system remained unchanged and the CSC was not observed up to 4 M NaCl. The solubilization of the PECs did not occur in the presence of Na₂SO₄. However, ion pair degradation occurs at 2.5 M KBr. The similar degradation of an ion pair of 1 : 1 PEC of PStSO₃Na and poly(diallyldimethylammonium chloride) was observed earlier in the presence of KBr.³⁴ The degradation of ion pair by KBr is well known. This experiment revealed that our PECs are stable at a high concentration of NaCl or Na₂SO₄, which are commonly present in the wastewater. Nevertheless, swelling of the self-assembled PEC layer in the presence of salt cannot be confirmed by this experiment.

Fig. 1 Digital photographs of representative PECs prepared using excess (A) anionic PE (PStSO₃Na/PVIm-Me) and (B) cationic PE (PVIm-Me/PStSO₃Na), excluding the 1 : 1 combination. The PECs were prepared in the presence of 0.5 M NaCl. The photographs were taken after eight weeks of settling. (C and D) SEM and AFM images of representative PVIm-Me/PStSO₃Na-5 : 1 PEC. (E and F) DLS particle size distribution profiles of the freshly prepared PECs. (G and H) Effect of stoichiometry and salt concentration on the zeta potential of the PECs.

The effect of salt and the PE concentration on the dispersion stability of the PECs was further evaluated.³⁰ The hydrodynamic size (DLS) of the freshly prepared representative PECs is highest in the absence of salts and had bimodal size distribution (Fig. S2, ESI†). The PECs prepared in the presence of 0.5 M NaCl remained dispersed even after storage for eight weeks as probed by DLS measurements (Fig. S3, ESI†). The SEM and AFM images of the representative PVIm-Me/PStSO₃Na-5 : 1 PEC show nearly spherical particles (Fig. 1C and D). The diameter of the PEC particles was measured to be ∼45 nm by the AFM analysis. The size of the PECs is different in the hydrated state (vide infra). The ion pair formation between cationic and anionic groups of the PEs produces a charge-neutralized core (hydrophobic), and the excess charge groups remain in the outer PEC surface (hydrophilic). It was reported that the PECs with one PE in excess exhibited a spherical structure, in which the charge shell contains excess components and is hydrophilic in nature.^30,33 The DLS measurements show monomodal size distribution of the PECs (0.5 M NaCl) (Fig. 1E and F). The hydrodynamic size centred at 250–360 nm for the PStSO₃Na/PVIm-Me PECs and 200–230 nm for the PVIm-Me/PStSO₃Na PECs depending on the PEC stoichiometry. The effect of the concentration of PEs on the size of the representative PECs (5 : 1) was also determined (Fig. S4, ESI†). The size of the PECs increases with the increase in the concentration of PEs during their preparation. Zeta potential measurements show an increase in the positive or negative zeta value with the increase in the non-stoichiometry of the PEs used for the PEC preparation (Fig. 1G). The increase in zeta values supports the enhancement of the PEC stability with the increase in non-stoichiometry. The effect of ionic strength on the zeta potential of the representative PECs was determined (Fig. 1H). The zeta potential decreases from −70 eV to −48 eV for anionic and +44 eV to +24 eV for cationic PECs from no salt to 0.1 M NaCl. The decrease in zeta potential was then marginal beyond >0.1 M NaCl for both the PECs. The extrinsic charge compensation in the presence of salt reduces the zeta potential of the PECs. At a salt concentration of >2 M, excessive charge screening occurs which leads to the coagulation of the PECs upon storage (Fig. S1A, ESI†).

The presence of free PEs in the PEC dispersion if any was evaluated. Earlier Sun and coworkers observed the exclusive formation of non-stoichiometric PECs from the oppositely charged PEs by the DLS analysis.³⁵ Herein, DLS measurements of the as-prepared PEC dispersions do not show any distribution below 60 nm. This indicates apparent absence of free PEs. The PEC dispersion was centrifuged to largely separate the particles. This helps to detect the free PE as the fraction of free PE if any will be increased in the supernatant. The DLS curve of the supernatant shows a decrease in PEC hydrodynamic size and narrowing down of the size distribution. It may be easily understood that the larger particles are removed upon centrifugation.³³ A very small fraction of uncomplexed PEs (∼1%) are seen in the DLS curves of the supernatant (Fig. S5, ESI†). The dispersion stability, zeta potential values and ion pair stability against the salt suggest that the oppositely charged PECs of 4 : 1 to 6 : 1 PE combination are useful for the LbL self-assembly.

3.2. Preparation of molecular selective barriers through the PEC self-assembly and morphology, thickness, MWCO and surface elemental analysis

The anionic and cationic PECs were alternatively applied on a cationic support substrate to obtain a densely packed layer (Scheme 1C, Experimental section). Standardization experiments revealed that PVIm-Me/PStSO₃Na-5 : 1 and PStSO₃Na/PVIm-Me-5 : 1 PECs formed by the 5 : 1 combination of PEs (7 mM each) upon the LbL self-assembly produced a selective barrier having 99.6% to >99.9% rejection of several dyes (Fig. S6A and B, ESI†). These PECs were stable and had a zeta potential of +20 mV and −36 mV in the presence of 0.5 M NaCl. The stability and zeta values indicate that these PECs are good candidates for the LbL assembly. Hence, PVIm-Me/PStSO₃Na-5 : 1 and PStSO₃Na/PVIm-Me-5 : 1 PECs were employed for the LbL self-assembly on the cationic base substrate to fabricate Mem-PEC1, Mem-PEC2 and Mem-PEC3. The SEM images show PEC particles on the membrane surfaces (Fig. 2A, S7A and B, ESI†). Tiny pores if any are not observed on the membranes even by high-magnification SEM analysis due to the limitation of the instrument (Fig. S8A–D, ESI†). Some overlapped particles are visualized in the SEM image of the air-dried Mem-PEC3. The PEC particles may be merged after drying/dehydration. Cryo-SEM analysis confirmed the densely packed PECs on the representative Mem-PEC3 (Fig. 2B). The AFM image also shows the assembled PEC particles on the representative Mem-PEC3 (Fig. 2C). The surface roughness (rms) increases from 34 nm to 65 nm from base to Mem-PEC3. The PECs after the LbL assembly create a relatively uneven surface. The increase in surface roughness with the increase in layer number was observed earlier when the LbL self-assembly of PEC and oppositely charged PE was performed on a silicon wafer.³⁵

Fig. 2 (A) Surface SEM (×30k magnification, dry), (B) cryo-SEM and (C) AFM images of representative Mem-PEC3. (D and E) Thickness of the isolated layer of Mem-PEC3 on a silicon wafer as determined by the AFM and cross-sectional SEM analysis. (F) Thickness of the model LbL assembled layer (three) on a silicon wafer. (G) Anchoring of the anionic PEC on the base membrane.

The base substrate contains quaternized moieties. The anionic PEC (excess PStSO₃Na) forms a strong ion pair with the base substrate to give Mem-PEC1. To verify the same, the layers of the membranes were isolated on the silicon wafers by leaching the base substrate using DMF. The apparent thickness of the isolated layers on the silicon wafers is in the range of 2–2.2 μm (Fig. 2D, S7C and D, ESI†). The cross-sectional SEM analysis of the representative isolated layers corroborates the AFM measurements (Fig. 2E and S7E, ESI†). However, the apparent thickness of the isolated layer seems to be quite high. Hence, the model LbL self-assembly of the PECs on the silicon wafer was performed to evaluate the thickness of the neat PEC-based layer. The determined thickness of the model PEC assembly is about 120 nm after total three layers of assembly (Fig. 2F). The cross-sectional SEM image of the as-prepared Mem-PEC3 shows a dense top layer of about 130 nm thickness (Fig. S7F, ESI†). However, the LbL self-assembly of the same uncomplexed PEs gave a layer thickness in the range of 30–38 nm.³⁶ Sun and coworker reported the greater thickness of the film formed by the LbL assembly of PEC and PE than that of the oppositely charged uncomplexed PEs.³⁵ The PECs have a larger dimension than that of the uncomplexed PEs. The larger dimension of the PECs is responsible to produce a relatively thick LbL assembled layer. Nevertheless, the height of the deposited PEC barrier layer on the support substrate is much smaller than the hydrodynamic diameter of the PECs. This is attributed to the deformation and shrinking of the PEC during the LbL assembly.³⁰

In contrast to the model LbL assembly on silicon wafer (Fig. 2F), the greater thickness of the isolated layers of the membranes is explained as follows. The isolated layers of the membranes contain ester linkages (1730 cm⁻¹ IR band) due to the copolymer of the base substrate. This observation is attributed to the ionic crosslinking of the top part of the base substrate (cationic) by the anionic PEC during first-layer formation. Hence, the determined thickness values are not true thickness of the self-assembled sole PECs layer. The size of most of the PEC particles is larger than the pore size of the base substrate. Therefore, ion pair formation between anionic PEC and the quaternized amine group of the base substrate occurs on the top surface and pore walls (first layer). The part of the copolymer chains present inside the base membrane (link to quaternized amine) surface is therefore part of the isolated layer, which ultimately gives high thickness (schematic representation, Fig. 2G). The apparent thickness of the isolated layers of the three membranes remained similar, as the contribution from the LbL-assembled part is significantly lower than that of the base substrate. This experiment is important as it confirms that the first layer is tightly held by the base substrate via electrostatic attraction, which provides very good stability to the whole LbL assembly system (vide infra). The base substrate and the PECs have strong ionic groups, which is crucial to hold the whole assembly tightly. The salt responsive nature of the LbL self-assembled membrane is reduced because of the strong ion pair formation starting from the base substrate. In contrast, weak PE-based PEC exhibits greater salt and acid/base sensitivity.

The MWCO and pore size distribution of the barrier layers were evaluated by the neutral solute sieving experiment. The MWCO values reduce and the pore size distribution curves become narrower from Mem-PEC1 to Mem-PEC3 (Fig. 3A and B). However, the base substrate had significantly greater MWCO and broader pore size distribution than that of the LbL assembled membranes (Fig. S9A and B, ESI†). The spaces among the PEC particles narrowed down with the increase in layer numbers. As a result, MWCO dramatically decreases from Mem-PEC1 (14 × 10³ g mol⁻¹) to Mem-PEC3 (1.4 × 10² g mol⁻¹) (Table 1). Probably, the defects of first layer are filled up by the oppositely charged PECs after the higher layer of self-assembly.

Fig. 3 (A) MWCO vs. PEG rejection, (B) pore size distribution curves, (C) XPS survey and (D–G) C 1s core-level spectra of representative membranes.

Table 1 Characteristic properties of the base substrate and the LbL-assembled membranes

Membrane	Pure water permeance (L m^−2 h^−1 bar^−1)	MWCO (g mol^−1)	Mean pore diameter (μp, nm)	Surface roughness (rms, nm)	Zeta potential (mV)
Base	550 ± 20	∼4 × 10^5	24	34	+4.3 ± 0.5
Mem-PEC1	75 ± 5	14 × 10^3	3.33	42	1 ± 0.5
Mem-PEC2	67 ± 5	8.5 × 10^3	1.71	47	3.4 ± 0.2
Mem-PEC3	52 ± 5	1.4 × 10^3	0.72	65	2 ± 0.2

---

The XPS survey spectra show a significant reduction in F content (∼0.7%, atomic) on the LbL assembled membranes from the base substrate (19.5%) (Fig. 3C). The drastic reduction in the F content on the membranes is due to the surface coverage by the PEC. The base substrate contains N and O due to the presence of a copolymer (a quaternized amine) and PVP (ESI†). The N and O contents are higher on the LbL-assembled membranes than those on the base membrane. The LbL-assembled membrane surfaces additionally contained S atoms. The PECs are composed of oppositely charged PEs (C, N, S and O atoms). The O content for the membranes follows the order of base < Mem-PEC2 < Mem-PEC3 < Mem-PEC1. The N content shows a trend for the membranes: base < Mem-PEC1 < Mem-PEC3 < Mem-PEC2. The top layer of Mem-PEC1 and Mem-PEC3 contain PStSO₃Na/PVIm-Me-5 : 1 (excess –SO₃Na groups), and hence, Mem-PEC1 and Mem-PEC3 had higher O contents than those of Mem-PEC2. Similarly, Mem-PEC2 contains more N atoms, as the last layer (top) of this membrane is made up of PVIm-Me/PStSO₃ Na-5 : 1. However, the O and N contents should be similar for Mem-PEC1 and Mem-PEC3 if only the top surface is considered. The slightly higher O content in Mem-PEC1 than that of Mem-PEC3 may be attributed to the contribution from the base substrate. Furthermore, the overall N content is higher in Mem-PEC3, which may be due to the second layer contribution. ATR-IR analysis indicates a nearly similar degree of PStSO₃Na/PVIm-Me-5 : 1 adsorption for Mem-PEC1 and Mem-PEC3 (Fig. S10A–C, ESI†).

The C 1s core-level XPS spectrum of the base substrate shows peaks at binding energies of 284.5 eV, 285.8 eV, 287.35 eV, 288.5 eV and 290.3 eV due to the C–C bonds of PVDF and the copolymer, CN/CN⁺ bond of PVP and the copolymer, C=O bond from PVP, –COO (ester) groups from the copolymer and C–F bonds of PVDF, respectively. The LbL-assembled membranes show no peak of –COO (ester) groups and a significant reduction in the peak area of the C–F bond (Fig. 3D–G). The N 1s core level spectra show significantly greater fractional area for the peak for the ⁺N–C bond than the base substrate (Fig. S10D–G, ESI†). Overall, PECs exclusively cover the membrane surfaces. Table 1 shows the pure water permeance, MWCO, mean pore diameter (μ_p), surface roughness (rms) and the zeta potential of the membranes. The pure water permeance of the membranes decreases with the increase in layer number, as the interstices between the complex particles decrease. The zeta potential of the membranes decreases or increases depending on the type of the last PEC. However, the zeta values did not show sign inversion after the LbL assembly. We suspect that the zeta potential of the membranes is not the true value for the top surfaces. The zeta potential of the base substrate (positive) substantially contributes to the values of the LbL-assembled membranes due to the existence of space. Earlier, Scheepers et al. observed no sign change in the zeta potential values from even to odd layers.³⁷ Wessling and co-workers reported the similar observation in the zeta potential values.¹⁴ They considered the contribution of the base substrate on the overall zeta potential of the LbL-assembled membranes.

3.3. Dye separation performance of the membrane

Mem-PEC1, Mem-PEC2 and Mem-PEC3 were employed for the evaluation of performance based on the standardization experiments (Fig. S6A and B, ESI†). Detailed separation performance of Mem-PEC3 (total 3 layers) was determined and compared with that of Mem-PEC1 and Mem-PEC2. The barrier layers of the membranes allow preferential permeation of monovalent and bivalent salts. The salt rejection decreases with the increase in feed concentration, which is due to the increasing membrane charge screening and lowering of Donnan exclusion (Fig. S11, ESI†).^38,39 Mem-PEC1 shows 99.4–99.5% CR (1 g L⁻¹) and DR-80 (1 g L⁻¹) rejection in the absence and ∼92–93% rejection in the presence of salt. Mem-PEC2 shows 99.8–99.9% CR and DR-80 rejection in the presence or absence of salt. However, Mem-PEC3 has ∼99.9% to >99.9% CR and DR-80 rejection, regardless of the presence or absence of salt (Fig. 4A). The base substrate fails to reject dyes, which is obviously due to its significantly greater MWCO and larger pore size (Fig. 4A and Table 1). Mem-PEC3 shows >99% rejection of RB, while Mem-PEC1 and Mem-PEC2 show <90% RB rejection. The separation of dye depends on the pore size, pore size distribution and charge of the membrane as well as size and nature of a dye molecule. The above-mentioned dye molecules contain aromatic rings and undergo aggregation in water, which increases the aggregate size.^9,12,39,40 The high CR and DR-80 rejection even by Mem-PEC2 is attributed to the larger dye aggregates than the membrane permeate space. The diffusion coefficients of these dyes are lower than the PEG of M_n 2000 g mol⁻¹.^1,38 However, the aggregate size of RB may be smaller than the CR and DR-80. The CR forms supramolecular aggregation in water by the hydrophobic interaction between the aromatic rings, which enhances the aggregate size and subsequent interception by the membranes.⁴¹ Van der Bruggen and coworkers observed the very good rejection of DR-80 by the membranes with MWCO values of 11 600 g mol⁻¹ and ∼17 500 g mol⁻¹, respectively, due to the aggregation of the dye molecules.³⁸ Earlier, our group and others probed the aggregation of dyes by the UV-visible analysis.^5,38 Narrower pore size distribution and lower MWCO (1400 g mol⁻¹) of Mem-PEC3 than those of Mem-PEC1 and Mem-PEC2 helped to intercept the dyes effectively by the former membrane. The salt and dye fractionation requires very high rejection of dye and low rejection of salt. Hence, further dye rejection and salt-to-dye fractionation efficacy of Mem-PEC3 were evaluated. The effect of dye concentration on the rejection behavior of Mem-PEC3 was assessed. The membrane shows very high CR and DR-80 rejection at a wide range of feed concentrations (Fig. 4B). The effect of concentration on rejection was more pronounced for RB, RB-5, OG and AO-7 dyes. As per previous report, CR and DR-80 rejection increased by the TUF membrane (MWCO = 4700 g mol⁻¹) with the increase in feed concentration.¹ The MWCO of Mem-PEC3 is about 1400 g mol⁻¹. It may be possible that the aggregate size of CR and DR-80 at their wide range of concentrations is larger than the pore size of Mem-PEC3. However, the aggregate size of the RB, RB-5, OG and AO-7 may be smaller than that of the CR and DR-80. Mem-PEC3 showed slightly improved dye rejection in the presence of salt (50 g L⁻¹) at different dye concentrations (Fig. S12, ESI†). The UV-visible spectra show increased dye rejection in the presence of salt (Fig. S13 and S14, ESI†). The slightly increased dye rejection by Mem-PEC3 is attributed to the formation of relatively large dye aggregates in the presence of salt than in its absence.^39,42 Salt is known to induce swelling of the LbL layer. The high concentration of salt may break the ionic bond and screen the charge sites of the membrane and the dye. The osmotic pressure difference between the bulk and the PEC network leads to diffusion of salt along with water from the bulk to the PEC network, which may lead to ion pair dissociation.⁴¹

Fig. 4 (A) Dye (1 g L⁻¹) rejection efficacy of the base and PEC-based membranes in the absence and presence of salt. (B) Effect of concentration on the dye rejection by Mem-PEC3. (C and D) Rejection and salt-to-different dye separation factors. (E) Digital photographs of the feed (salt + dye) and permeate solutions through Mem-PEC3. The salt concentration was 50 g L⁻¹, except for the CR feed, and the dye concentration was 1 g L⁻¹ (C and D). The applied pressure was 1.38 bar for the experiments.

Here, although dye rejection by Mem-PEC3 increased, the salt rejection showed a decreasing trend with the increase in salt concentration (Fig. S11, ESI†). The decrease in salt rejection efficacy of the membranes with the increase in salt concentration was observed earlier.^{1,12,36,38,40,42} This may be due to the enhanced membrane charge screening at higher salt concentrations. Our observation strongly indicates that the aggregate size of the dyes in saline water is larger than the pore size of Mem-PEC3 due to the formation of larger dye aggregates. Hence, the effect of salt on charge screening or the pore swelling if any was not reflected in the current work during dye separation by Mem-PEC3, as it has lower MWCO than that of Mem-PEC1. However, CR and DR-80 rejection by Mem-PEC1 decreases in the presence of salt as mentioned above (Fig. 4A). The negative effect of salt such as charge screening of both dye and membrane surface lowers the electrostatic repulsion between the membrane surface and dye molecules. Aggregation of the dye cannot compensate this effect for Mem-PEC1, as this membrane has larger MWCO and pore size than those of Mem-PEC2 and Mem-PEC3. Earlier reported membranes exhibited lower dye rejection in the presence of salt depending on the aggregate size of dyes and MWCO of the membranes.^1,12,38,40

The salt-to-dye separation factor of Mem-PEC3 was determined using a mixture of dyes and salts (Fig. 4C and D). The separation factor values indicate good salt-to-dye fractionation performance of the membrane. The salt concentration was 50 g L⁻¹ in the dye feed except CR (20 g L⁻¹) as the solubility of later dye decreases at >20 g L⁻¹ of salt.⁴² The permeates were almost colorless or slightly colored depending on the type of dye (Fig. 4E). Furthermore, the effects of salt concentration on the CR (1 g L⁻¹) rejection and salt-to-CR separation factors were evaluated (Fig. S15, ESI†). The separation factor showed an increasing trend with the increase in salt concentration due to the decreasing salt rejection.

3.4. Stability and salt/dye fractionation

Mem-PEC3 shows almost linear increase in pure water flux with the applied pressure of up to 6 bar (Fig. 5A). However, the permeate flux is not linear above 3 bar of applied pressure during the permeation of salt + dye (Fig. 5B). The RB rejection decreases from >99.9% to 99.6% as the applied pressure increases from 0.68 bar to 2.76 bar, and then the rejection reaches a constant value of about 99.2% (Fig. 5B). The slight decrease in RB rejection with the increase in applied pressure is due to the concentration polarization.^1,11,36,40 Van der Bruggen and coworkers reported a similar effect earlier.^1,40 The membrane was further subjected to aging experiment by permeating saline water at an applied pressure of 17.2 bar for 24 h, and then the DR-80 rejection was determined at an applied pressure of 4 bar for total 150 h (Fig. 5C). The membrane after high-pressure permeation experiments exhibited >99.9% DR-80 rejection. However, the permeate flux decreased as compared to before aging (Fig. 5B). This may be due to the effect of high pressure (17.2 bar) leading to pore compaction. These membranes are designed for the low pressure-driven application. The pH stability of the membrane was assessed by separately permeating water of pH values 3, 7, 10 and 12 for 12 h each (Fig. 5D). The dye rejection by the membrane remains almost unaltered before and after the pH treatment. The permeate flux of the membrane marginally increases after treatment at pH 12. We further studied the change in MWCO and pore size distribution of the membrane after the treatment at pH 12. Evidently, the MWCO and pore size distribution do not show much change after treatment at pH 12 for 12 h (Fig. S9C and D, ESI†). Furthermore, a simulated dye solution (pH = 9) was permeated through the membrane (Fig. 5E).³¹ The results indicate high performance of the membrane under relevant filtration conditions. It is seen that the salt rejection is lowered at a higher salt concentration (Fig. 5C and E). The higher salt concentration (Fig. 5C) lowers the salt rejection by the membrane.^1,12,38,40 The base membrane substrate tightly holds the PEC first layer by the strong electrostatic interaction. The addition of NaCl and Na₂SO₄ does not affect the stability of the PECs (Fig. S1, ESI†). The as-prepared PECs remain stable at pH 3 and 12 (Fig. 5F). These PECs were formed by the strong PEs. The PECs contain strong cations and anions for further ion pair formation during the LbL assembly, which provided good stability of the assembled layers in the pH range of 3–12 and a NaCl/Na₂SO₄ concentration of 50 g L⁻¹. However, swelling of the layer at high salt concentrations cannot be confirmed by this experiment.

Fig. 5 (A) Pure water flux and (B) permeate flux and rejection efficacy of Mem-PEC3 with the increasing applied pressure. (C) Dye rejection and permeate flux of the membrane determined at an applied pressure of 4 bar using DR-80 (1 g L⁻¹) and NaCl (50 g L⁻¹) feed after subjecting the membrane swatches to the permeation of water at 17.2 bar for 24 h. (D) Stability of Mem-PEC3 evaluated after permeation of water at different pH values for 12 h each separately. (E) Permeation results of simulated dye wastewater. Applied pressure is 1.38 bar for plots (D) and (E). (F) Stability of the PECs at pH 3 and 12.

3.5. Antifouling property

Membrane fouling occurs by the electrostatic attraction with the oppositely charged dye as well as by the hydrophobic interaction.⁴³ Membrane fouling may increase in the presence of salt as the dye aggregate size and hydrophobicity increase. The larger aggregates elevate fouling by the hydrophobic interaction.^38,39 The permeate flux decreases immediately as the water or salt solution is replaced by the dye solution. The permeate flux after initial reduction remained almost steady. Our membrane shows 22% and 27% flux reduction (FR) in the absence and presence of salt (Fig. 6A and B, ESI†). The initial permeate flux as well as flux during antifouling experiment decreases to a larger extent when water is replaced by the salt solution due to the increase in the osmotic pressure as well as membrane fouling. In addition to increase of hydrophobic interaction, electrostatic repulsion between the membrane surface and DR-80 decreases in the presence of salt, which are the reasons of higher FR in the presence of salt in the dye solution. The flux recovery ratio (FRR) of the membrane was determined after the permeation of dye or dye + salt mixture and after water washing. The FRR values of the membrane after water washing were ∼99%, indicating very good antifouling property of the membrane. The membrane surface is cleaned by the water due to the removal of the dye. The last layer of Mem-PEC3 is made up of negatively charged PEC, which contains free –SO₃Na groups, as confirmed by the zeta potential measurements (Fig. 1G and H). The overcompensation of the negatively charged PEC occurs during the top layer formation of Mem-PEC3. Hence, the top layer of Mem-PEC3 contains excess –SO₃Na moieties (Fig. 3C). Therefore, charge–charge repulsion with the membrane surface and dye molecules occurs. The charge shielding effect is minimized due to the removal of salt during water washing. The –SO₃Na groups facilitate the electrostatic repulsion with the negatively charged dyes during water washing. The DR-80 rejection remained >99.9% during the antifouling test, which further indicates the stability and antifouling property of the membrane.

Fig. 6 (A) Dye antifouling performance in terms of permeate flux of Mem-PEC3 during the separation of DR-80 (no salt) and DR-80 (1 g L⁻¹) + NaCl (50 g L⁻¹). (B) FR and FRR values during the antifouling test. The applied pressure was 1.38 bar.

Batch diafiltration was applied for the resource recovery and fractionation of the dye and salt. It involved 0.5 L of initial feed and then removal of 0.25 L of feed by the membrane and addition of an equal volume of water.³⁹ The NaCl and Na₂SO₄ concentrations reduce to 0.5 g L⁻¹ and ∼1 g L⁻¹, respectively, from 50 g L⁻¹ after 3.6 diavolume of operation (Fig. S16, ESI†). The dye concentration in the feed changes from 1 g L⁻¹ to ∼0.99 g L⁻¹. Little changes in the dye concentration indicate the potential application of this membrane for salt and dye fractionation. The permeate flux increases with the progress of the diafiltration and then remained almost unchanged due to the continuous decrease in the osmotic pressure of the system.

A critical comparison of the fabrication method and performance of the reported membranes and Mem-PEC2 and Mem-PEC3 was made to elucidate the novel features and benefits of our membranes (Fig. 7). Earlier, LbL self-assembly of the oppositely charged PEs was performed, which needed several layers to obtain NF or LNF membranes.^8,11–15 For example, the LbL-assembled membrane containing a total of 6–10 layers showed salt rejection of about 90% at a feed concentration of 2 g L⁻¹.⁸ Most of the LbL membranes contained a total of 8–12 layers to obtain good salt rejection performance.^13,16,17 Concentration polarization-induced LbL assembled membranes (5 layers) gave good salt-to-dye separation performance.¹² Solutions of the mixture of PEs or premade PECs in the presence of salt or acid/base upon casting and phase inversion afforded the membranes.^{20–22,43–45} These membranes were tested for desalination of water at low salt concentrations LbL membranes containing 11 layers gave good dye rejection at low dye concentrations.¹⁵ The commercial TUF and LNF membranes exhibited a lower permeate flux than that of Mem-PEC3 and almost similar or lower dye rejection.^1,38,40 Thin-film composite (TFC) LNF membranes were prepared by the IP and applied for the fractionation of the salt and dye.^{2,5,6,31,39,46–49} These TFC membranes and TUF membranes shown in Fig. 7 exhibit lower dye rejection than that of Mem-PEC3.⁵⁰ Mem-PEC3 (three layers) membrane exhibited higher dye rejection than that of the reported LbL-assembled membranes.^51–53 The potential of the current approach may be easily understood as even Mem-PEC2 shows ∼99.9% rejection of CR and DR-80 in the presence or absence of salt. However, Mem-PEC3 is our usual choice because of its high performance for a broad spectrum of dyes. Evidently, uncomplexed PEs under similar LbL conditions (three layers) afforded a membrane with significantly greater MWCO (10 000 g mol⁻¹) and lower dye separation performance than that of the current Mem-PEC3 (Fig. S17, ESI†).¹¹ Moreover, the required concentration of the uncomplexed PEs under conventional LbL conditions was 20 mM each for the fabrication of the membranes.¹¹ Herein, the total PE concentration for the preparation of PEC dispersion is 7 mM. However, the uncomplexed PEs after a total of seven layers of LbL assembly produced a membrane having an MWCO of 1800 g mol⁻¹, which is close to that of Mem-PEC3 (1400 g mol⁻¹). The thickness is 120 nm for the three-layer assembly of the PECs on a silicon wafer (Fig. 2F), while this value for the uncomplexed PEs is in the range of 30–38 nm.³⁶ The defects of base substrate are rapidly filled up by the PEC particles during the LbL assembly. The PEC-based LbL-assembled membranes containing merely a total of two to three layers show a dense surface morphology (Fig. 2A and S7B, ESI†). The membrane (Mem-PEC2) containing only two layers gave CR and DR-80 rejection as high as ∼99.9%. Clearly, PECs gave a membrane having lower MWCO and greater dye rejection than those of the uncomplexed parent PEs (Fig. S17, ESI†). These results strongly indicate the effect of PECs in the LbL assembly. PECs contain abundant free charge groups on the outer surface and are larger in dimension than that of the uncomplexed PEs.^27,28,30 These factors facilitate the rapid formation of selective barriers by the LbL assembly of the oppositely charged PECs. Each PEC is made up of two PEs. Hence, the deposition of each PEC layer usually gives a layer with a higher thickness. This explains the rapid membrane formation by the PECs than the uncomplexed PEs.

Fig. 7 Comparative performance of Mem-PEC2 and Mem-PEC3 with reported membranes.

3.6. Mechanism of dye separation

The dye rejection efficacy of the membranes increases with the increase in dye concentration (Fig. 4B). Mem-PEC3 exhibits steady dye rejection with time (Fig. 5C). The concentration of the dye in a certain volume of permeate and retentate streams is almost similar to the feed dye concentration as probed by the separate experiments. Indeed, Mem-PEC3 shows equilibrium CR and DR-80 adsorption capacities <1 mg g⁻¹. This value of adsorption capacity is negligible as compared to that of the feed dye concentration (1 L feed). Furthermore, about 99% dye was recovered after the diafiltration operation. Therefore, the dye rejection by the membrane is due to the size and charge-based separation.^3,5,36

4. Conclusion

We showed that non-stoichiometric polyelectrolyte complexes (PECs) of a strong oppositely charged poly(styrene sulfonic acid) sodium salt (PStSO₃Na) and quaternized poly(vinyl imidazole) (PVIm-Me) are potential candidates for the rapid fabrication of layer-by-layer (LbL) self-assembled molecular selective barriers. The LbL self-assembly of the PECs (1 : 5 or 5 : 1 mol/mol) of PVIm-Me and PStSO₃Na (total concentration = 7 mM) at a salt concentration of 0.5 M NaCl on the cationic base substrate provided Mem-PEC3 (total three layers) having high salt to dye (NaCl or Na₂SO₄) selectivity. The membrane exhibited >99% rejection of several dyes and good dye antifouling property with a flux recovery of ∼99%. The oppositely charged PECs contain strong cationic and anionic groups, respectively, in their outer surfaces. The larger dimension of the PECs than that of the uncomplexed PEs and free charge facilitate rapid LbL assembly to form molecular selective membranes after only two to three layers of the self-assembly. This was manifested by the fact that Mem-PEC2 containing merely two layers exhibited >99.5% CR and DR-80 rejection in the presence or absence of salt. The stability of the PEC-based LbL self-assembled membrane at pH = 3–12 and a NaCl or Na₂SO₄ concentration of ∼50 g L⁻¹ is attributed to the strong ion pair formation with the base substrate as well as between the PEC layers. However, the LbL assembly of the uncomplexed PEs (20 mM) needs a total of seven layers to obtain a membrane even having a lower separation performance than that of the PEC-based membrane (Mem-PEC3, a total of three layers). Conclusively, PECs offer facile LbL assembly at their lower concentration than that of the building block PEs. This work, thus, highlights a new approach for the rapid formation of molecular selective membranes for salt and dye fractionation at low applied pressures.

Data availability

The experimental procedures and analytical data are available within the manuscript and its ESI.†

Author contributions

This study was conceptualized by S. K. J. The experiments were carried out and the data were analysed by U. S. J. The investigation was supervised by S. K. J.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

A PRIS number CSIR-CSMCRI-04/2025 has been assigned for this manuscript. This work was supported by the Science and Engineering Research Board (SERB, file no. CRG/2023/001694), India. It was also supported by the in-house project of the Council of Scientific and Industrial Research (CSIR), Government of India. Urvashi S. Joshi thanks DST-WTI for a research fellowship. The authors thank the centralized analytical facility (CSIR-CSMCRI) for analytical support.

References

1. J. Lin, W. Ye, M. C. Baltaru, Y. P. Tang, N. J. Bernstein, P. Gao, S. Balta, M. Vlad, A. Volodin, A. Sotto, P. Luis, A. L. Zydney and B. Van der Bruggen, J. Membr. Sci., 2016, 514, 217–228.
2. H. F. M. Austria, J. O. Young, O. Setiawan, T. H. Huang, A. R. Caparanga, K. R. S. Pamintuan and W. S. Hung, Sep. Purif. Technol., 2025, 354, 129053.
3. Anuradha, U. S. Joshi and S. K. Jewrajka, ACS Appl. Mater. Interfaces, 2024, 16, 61344–61359.
4. S. Jabbarvand Behrouz, A. Khataee, V. Vatanpour and Y. Orooji, ACS Appl. Mater. Interfaces, 2024, 16, 10508–10521.
5. Anuradha, U. S. Joshi, S. Giri and S. K. Jewrajka, Chem. Eng. J., 2025, 509, 161392.
6. J. Li, J. L. Gong, S. Y. Fang, W. C. Cao, S. Q. Tang, M. Qin, Y. W. Wang, H. Y. Zhou and M. Long, ACS Appl. Mater. Interfaces, 2023, 15, 34272–34289.
7. J. S. Trivedi, D. V. Bhalani, G. R. Bhadu and S. K. Jewrajka, J. Mater. Chem. A, 2018, 6, 20242–20253.
8. S. U. Baowei, T. Wang, Z. Wang, X. Gao and C. Gao, J. Membr. Sci., 2012, 423, 324–331.
9. A. Toutianoush, J. Schnepf, A. El Hashani and B. Tieke, Adv. Funct. Mater., 2005, 15, 700–708.
10. I. J. Gresham, D. M. Reurink, S. W. Prescott, A. R. Nelson, W. M. De Vos and J. D. Willott, Macromolecules, 2020, 53, 10644–10654.
11. U. S. Joshi, D. V. Bhalani, A. Chaudhary and S. K. Jewrajka, J. Membr. Sci., 2022, 641, 119908.
12. U. S. Joshi, Anuradha and S. K. Jewrajka, J. Membr. Sci., 2023, 669, 121286.
13. N. Dizge, R. Epsztein, W. Cheng, C. J. Porter and M. Elimelech, J. Membr. Sci., 2018, 549, 357–365.
14. J. Kamp, S. Emonds, M. Seidenfaden, P. Papenheim, M. Kryschewski, J. Rubner and M. Wessling, J. Membr. Sci., 2021, 639, 119636.
15. A. Gopalakrishnan, M. L. Mathew, J. Chandran, J. Winglee, A. R. Badireddy, M. Wiesner, C. T. Aravindakumar and U. K. Aravind, ACS Appl. Mater. Interfaces, 2015, 7, 3699–3707.
16. R. M. DuChanois, R. Epsztein, J. A. Trivedi and M. Elimelech, J. Membr. Sci., 2019, 581, 413–420.
17. E. Virga, J. de Grooth, K. Žvab and W. M. de Vos, ACS Appl. Mater. Interfaces, 2019, 1, 2230–2239.
18. X. Li, C. Liu and B. Van der Bruggen, J. Mater. Chem. A, 2020, 8, 20870–20896.
19. C. V. Gherasim, T. Luelf, H. Roth and M. Wessling, ACS Appl. Mater. Interfaces, 2016, 8, 19145–19157.
20. K. Sadman, D. E. Delgado, Y. Won, Q. Wang, K. A. Gray and K. R. Shull, ACS Appl. Mater. Interfaces, 2019, 11, 16018–16026.
21. M. A. Restrepo, J. Kamp, L. Guericke, R. Schnichels, H. Roth and M. Wessling, J. Membr. Sci. Lett., 2023, 3, 100039.
22. E. N. Durmaz, J. D. Willott, M. M. H. Mizan and W. M. de Vos, Soft Matter, 2021, 17, 9420–9427.
23. Q. Zhao, J. Qian, Q. An, C. Gao, Z. Gui and H. Jin, J. Membr. Sci., 2009, 333, 68–78.
24. X. S. Wang, Y. L. Ji, P. Y. Zheng, Q. F. An, Q. Zhao, K. R. Lee, J. W. Qian and C. J. Gao, J. Mater. Chem. A, 2015, 3, 7296–7303.
25. F. Zhang, C. Song, X. Zhu, J. Gong and Q. Zhao, J. Mater. Chem. A, 2019, 7, 27450–27457.
26. X. Liu, B. Dai, L. Zhou and J. Sun, J. Mater. Chem. A, 2009, 19, 497–504.
27. L. Zhang and J. Sun, Chem. Commun., 2009, 26, 3901–3903.
28. M. Cui, W. S. Ng, X. Wang, P. Darmawan and P. S. Lee, Adv. Funct. Mater., 2015, 25, 401–408.
29. M. Wu, N. An, Y. Li and J. Sun, Langmuir, 2016, 32, 12361–12369.
30. L. Zhang, M. Zheng, X. Liu and J. Sun, Langmuir, 2011, 27, 1346–1352.
31. R. Zhao, P. Jin, J. Zhu, Y. Li, G. Li, A. Volodine, Y. Liu, J. Zheng and B. Van der Bruggen, J. Membr. Sci., 2023, 673, 121477.
32. M. Müller, B. Kessler and S. Richter, Langmuir, 2005, 21, 7044–7051.
33. H. Dautzenberg, Macromolecules, 1997, 30, 7810–7815.
34. Z. A. Digby, M. Yang, S. Lteif and J. B. Schlenoff, Macromolecules, 2022, 55, 978–988.
35. Y. Guo, W. Geng and J. Sun, Langmuir, 2009, 25, 1004–1010.
36. U. S. Joshi, S. Samanta and S. K. Jewrajka, ACS Appl. Mater. Interfaces, 2024, 16, 32748–32761.
37. D. Scheepers, A. Casimiro, Z. Borneman and K. Nijmeijer, ACS Appl. Polym. Mater., 2023, 5, 2032–2042.
38. M. Jiang, K. Ye, J. Deng, J. Lin, W. Ye, S. Zhao and B. Van der Bruggen, Environ. Sci. Technol., 2018, 52, 10698–10708.
39. R. Zhao, Y. Li, Y. Mao, G. Li, T. Croes, J. Zhu, X. You, A. Volodin, J. Zheng and B. Van der Bruggen, Environ. Sci. Technol., 2022, 56, 17998–18007.
40. J. Lin, W. Ye, H. Zeng, H. Yang, J. Shen, S. Darvishmanesh, P. Luis, A. Sotto and B. Van der Bruggen, J. Membr. Sci., 2015, 477, 183–193.
41. J. Chen, S. Xu, C. Y. Tang, B. Hu, B. Tokay and T. He, Desalination, 2023, 555, 116520.
42. J. Lin, C. Y. Tang, W. Ye, S. P. Sun, S. H. Hamdan, A. Volodin, C. V. Haesendonck, A. Sotto, P. Luis and B. Van der Bruggen, J. Membr. Sci., 2015, 493, 690–702.
43. J. Zhu, M. Tian, J. Hou, J. Wang, J. Lin, Y. Zhang and B. Van der Bruggen, J. Mater. Chem. A, 2016, 4, 1980–1990.
44. E. N. Durmaz, M. I. Baig, J. D. Willott and W. M. de Vos, ACS Appl. Polym. Mater., 2020, 2, 2612–2621.
45. M. I. Baig, E. N. Durmaz, J. D. Willott and W. M. de Vos, Adv. Funct. Mater., 2020, 30, 1907344.
46. M. Chen, Q. Song, Z. Li, W. Bai, M. Xu, X. Li and Y. Wang, Desalination, 2025, 593, 118206.
47. Y. Zhang, L. Wei, S. Pan, L. Zhang, Z. Yang, Y. Liu and Y. Fu, Chem. Eng. Sci., 2025, 304, 121065.
48. L. Wang, L. Zhao, X. Xiao, X. Wang, J. Cao, H. Ye and Y. Zhang, Sep. Purif. Technol., 2025, 359, 130551.
49. M. M. Tang, X. G. Jin, X. Tang, M. Yu, X. H. Ma and Z. L. Xu, Sep. Purif. Technol., 2025, 353, 128476.
50. L. Ding, D. Han, H. Zhang, S. Yang and Y. Zhang, Int. J. Biol. Macromol., 2025, 308, 142482.
51. H. Kang, J. Shi, L. Liu, M. Shan, Z. Xu, N. Li and L. Zhao, Appl. Surf. Sci., 2018, 428, 990–999.
52. S. Hao, L. Jiang, Y. Li, Z. Jia and B. Van der Bruggen, Chem. Commun., 2020, 56, 419–422.
53. Z. Lin, Q. Zhang, Y. Qu, M. Chen, F. Soyekwo, C. Lin and Q. Liu, J. Mater. Chem. A, 2017, 5, 14819–14827.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta02872h

---