High-flux nanofiltration membranes tailored by bio-inspired co-deposition of hydrophilic g-C₃N₄ nanosheets for enhanced selectivity towards organics and salts

Abstract

Surface modification with advanced nanomaterials (i.e., 2D nanosheets) can be used to strategically tailor membrane properties, providing improved solute permselectivity to targeted molecules. In particular, 2D graphite-like carbon nitride (g-C₃N₄) nanosheets are a promising alternative for membrane modification, due to their exceptional physicochemical properties and facile synthesis. Herein, high-flux nanofiltration (NF) membranes were designed using bio-inspired co-deposition of hydrophilic g-C₃N₄ nanosheets with a polydopamine (PDA)/polyethylenimine (PEI) layer onto porous ultrafiltration (UF) substrates. The g-C₃N₄ nanosheets created additional nanochannels in the PDA/PEI layer to facilitate water molecule transport, resulting in high permeability (28.4 ± 1.2 L m⁻² h⁻¹ bar⁻¹). Particularly, the bio-inspired layer structure was tailored from the UF to the NF (592 Da) scale by incorporating g-C₃N₄ nanosheets, thereby breaking through the permeability–selectivity trade-off effect. The tailored NF membrane enabled ultrahigh retention of three reactive dyes (610–630 Da, >99.3%) and low salt rejection (2.9% for NaCl; 7.6% for Na₂SO₄), significantly promoting the fractionation of dyes and salts for dye desalination. Additionally, the hydrophilic g-C₃N₄ nanosheets with oxygen plasma treatment further enhanced the wettability of the membrane surfaces, resulting in a superior antifouling performance. This study indicates the promise of g-C₃N₄ nanosheets to engineer high-flux NF membranes with desirable fractionation performance for sustainable treatment of highly saline wastewater.

---

Environmental significance

In sustainable management of highly saline wastewater, nanofiltration (NF) technology is increasingly applied in an emerging paradigm shift to separate and reuse organics/salt mixtures as resources rather than remove them completely as wastes. However, the commercial NF membranes with negative charges yield high rejections to both organics and inorganic salts, compromising the fractionation of organics/salt mixtures. This paper reports a strategy of fast bio-inspired co-deposition of hydrophilic g-C₃N₄ nanosheets with a polydopamine/polyethylenimine layer onto porous ultrafiltration substrates for fabrication of high-flux NF membranes. Specifically, these bio-inspired NF membranes show enhanced dye retention, salt transmission, and antifouling properties, demonstrating a competitive and practical solution for sustainable management of highly saline wastewater, e.g., textile wastewater.

1. Introduction

In the concept of advanced sustainability, wastewater treatment increasingly requires an emerging paradigm shift from removal-centered approaches to recovery processes.^1–3 By using novel strategies and practices, the pollutants in industrial wastewater streams can be potentially extracted and transformed into high value-added resources, i.e., nutrients, bioplastics, energy and even products, thereby expediting the water and material loop.⁴ Specifically, industrial waste streams, e.g., textile wastewater, glyphosate neutralization liquor, soy sauce wastewater, and liquors containing heterocyclic drug derivatives or iminodiacetic acid, are usually highly loaded with valuable organics with elevated salinity. To achieve the sustainable management of these liquors, the key challenge is to effectively separate the organic and inorganic compounds.

Nanofiltration (NF) has been emerging as a promising technology for fractionation of organics with molecular weights (MWs) of 200–1000 Da and inorganic salts through steric exclusion and the Donnan effect.^5,6 To date, many attempts have been made to tailor the physicochemical properties of the NF membrane surface, e.g., pore sizes, surface charges, wettability, and roughness, to improve the filtration performance. Nevertheless, restricted by the chain stiffness of the polymer matrix and the interchain spacing, polymer-based membranes suffer from an inherent limitation of the selectivity–permeability trade-off.^7–9 Recently, the incorporation of advanced nanomaterials into polymer matrices has attracted considerable interest as an alternative strategy to break through this trade-off effect.

Mixed matrix NF membranes blended with nanoparticles were developed through phase inversion to yield high water permeability and low salt rejection, showing potential for dye desalination and purification. Unfortunately, the poor miscibility of the polymer matrix with nanoparticles would result in nonselective interface voids and surface defects, deteriorating the solute permselectivity. Zhang et al. constructed loose NF membranes by blending nanoparticles, i.e., poly(ionic liquid) brush-grafted silica nanospheres and zwitterionic polymer-functionalized graphene oxide, with a polyethersulfone matrix which facilitated salt transport but yielded ca. 90% rejection of dyes.^10,11 The moderate dye rejection induced a high loss of dyes when utilizing the diafiltration process to desalinate dye/salt mixtures.

Alternatively, the use of thin film composite (TFC) NF membranes has emerged as a pathway to enhance the solute selectivity. Generally, TFC NF membranes are coated with a thin and selective polyamide layer by crosslinking of trimesoyl chloride (TMC) and m-phenylenediamine (MPD) or piperazine (PIP).¹² However, it is difficult to manipulate the crosslinking degree for polymerization between TMC and MPD or PIP monomers, inducing a tight surface structure of the resultant NF membranes to sufficiently retain both organic and inorganic solutes. This not only compromises the fractionation of organics and inorganic salts, but also significantly reduces the permeation flux of TFC NF membranes due to the enhanced osmotic pressure resulting from high salt rejection. Several approaches have been investigated to overcome this challenge. On the one hand, controllable interfacial polymerization kinetics are of great significance to construct a targeted selective layer for NF membranes. Tan et al. designed a polyamide NF membrane with a Turing structure by manipulating the reaction–diffusion kinetics within the aqueous–organic interface, thus breaking through the upper bound of the permeability–selectivity trade-off.¹³ In addition, the use of novel acyl chloride monomers (e.g., mm-biphenyl tetraacyl chloride,¹⁴ cyanuric chloride,¹⁵ and 5-chloroformyloxy-isophthaloyl chloride¹⁶) and amine monomers (e.g., dopamine,¹⁷ triethanolamine,¹⁸ sericin,¹⁹ and zwitterionic amine²⁰) diversifies and tailors the surface structures of the resultant NF membranes, potentially facilitating the solute selectivity. On the other hand, creating a thin film nanocomposite (TFN) NF membrane by incorporating nanomaterials into the polyamide layer would yield more pathways or channels to enhance the transport of water.²¹ Graphene-based nanosheets (i.e., GO and rGO) as representative two dimensional structured materials are emerging as candidates to design the surface structure of TFN NF membranes and enhance the filtration performance,^22,23 However, the complicated, high-cost synthesis procedure limits the practical applications of these graphene-based nanosheets, making it necessary to explore alternative two-dimensional (2D) nanosheets for membrane modification.

Recently, bulk carbon nitride (g-C₃N₄) with a graphite-like structure consisted of tri-s-triazine units bonded with tertiary amino groups, where the nanosheet layers interacted via delicate van der Waals forces.²⁴ This allows the bulk g-C₃N₄ to be facilely exfoliated into 2D g-C₃N₄ nanosheets. Due to their exceptional physicochemical properties (i.e., thermal and chemical stability, optical and photoelectrochemical characteristics) and facile synthesis (calcination of cheap nitrogen-rich precursors, e.g., dicyandiamide, melamine, carbamide and thiocarbamide), 2D g-C₃N₄ nanosheets have been increasingly applied for H₂ production, oxygen reduction, and photocatalysis.^25–28 Furthermore, with regularly distributed triangular nanopores (ca. 3.11 Å) throughout the entire laminar structure, the assembly of g-C₃N₄ nanosheets into 3D stacks potentially enables the creation of more nanochannels for the free passage of water molecules.⁸ Stacked g-C₃N₄ nanosheet-based membranes fabricated by simple vacuum filtration were found to have a high flux, which was confirmed by molecular dynamics simulation.²⁴ However, the unbound g-C₃N₄ nanosheet layers are too fragile to survive practical cross-flow testing conditions and cannot even tolerate water rinsing or a gentle finger touch, which is similar to the case of GO layers.²⁹ Therefore, new strategies are strongly required to address the weak binding of g-C₃N₄ nanosheets in membrane synthesis.

In this study, we proposed the first attempt to apply hydrophilic 2D g-C₃N₄ nanosheets to design high-flux TFN NF membranes for efficient dye/salt fractionation. First, oxygen plasma was applied to treat the 2D g-C₃N₄ nanosheets to enhance their hydrophilicity; subsequently, the homogeneous dispersion of hydrophilic 2D g-C₃N₄ nanosheets in dopamine and polyethylenimine (PEI) mixed solutions was coated on the top of porous hydrolyzed polyacrylonitrile (HPAN) ultrafiltration (UF) substrates using bio-inspired co-deposition. Finally, the resultant membranes were systematically characterized to evaluate their applicability for separation of organics (i.e., dyes) and inorganic salts, in view of sustainable treatment of highly saline wastewater, e.g., textile wastewater.

2. Experimental

2.1 Chemicals and membranes

Porous PAN substrates with a molecular weight cut-off (MWCO) of 200 kDa (AmFor Membrane Technology Engineering Co. Ltd., China) were hydrolyzed in a 2 M NaOH solution for 20 h (denoted as HPAN) and then completely rinsed with distilled water to remove the residual NaOH agent. Dopamine hydrochloride, PEI (M_n: 600 Da), tris(hydroxymethyl)aminomethane (Tris, purity of 99.8%), and ammonium persulfate (APS, purity of 98.0%) were purchased from Sigma Aldrich. Melamine powder (Sinopharm Chemical Reagent, China) was applied to synthesize the g-C₃N₄ nanosheets.

The model dyes, including reactive blue 19 (RB19), reactive orange 1 (RO1), and reactive orange 16 (RO16), were selected to test the separation performance of the as-prepared membranes. The molecule structures of these tested reactive dyes are shown in Fig. S1 of the ESI.†

2.2 Bio-inspired co-deposition of g-C₃N₄ nanosheets on HPAN membranes

2.2.1 Synthesis of hydrophilic g-C₃N₄ nanosheets. The bulk g-C₃N₄ was fabricated via calcination of melamine powder.³⁰ Typically, 5.0 g melamine powder as the precursor was heated at 600 °C for 2 h in an air atmosphere. Afterward, the resultant g-C₃N₄ bulk was yielded after being cooled down to room temperature and ground into powder before use.

Hydrophilic g-C₃N₄ nanosheets were obtained through an oxygen plasma treatment. Specifically, 0.5 g g-C₃N₄ bulk powder was put in a PC-6D microwave radical generator (Jiarun Wanfeng Technology Ltd., Beijing) with an operating power of 100 W and an O₂ flow rate of 36 L h⁻¹ for 0.5 h. Subsequently, the treated g-C₃N₄ powder was placed in 100 mL pure water for 12 h ultrasonication and then was centrifuged at 3000 rpm for 10 min to remove the unexfoliated layered g-C₃N₄.³¹ Ultimately, the hydrophilic 2D g-C₃N₄ nanosheet solution was prepared for further use.

2.2.2 Bio-inspired co-deposition of hydrophilic g-C₃N₄ nanosheets. The schematic of bio-inspired modification of HPAN substrates with hydrophilic g-C₃N₄ nanosheets is shown in Fig. 1. In detail, 84 mg APS and 40 mL Tris solution with pH 8.5 (50 mM) as the buffer were dissolved with hydrophilic g-C₃N₄ nanosheet suspensions of different concentrations (i.e., 0, 0.005, 0.01, 0.02, and 0.04%) using an ultrasonic treatment. Afterward, 2 g L⁻¹ dopamine and PEI were added and completely mixed in the buffer solution. The as-prepared solutions were poured into a customized round membrane holder to be in contact with the HPAN substrates for co-deposition of dopamine and PEI for 4 h. Subsequently, the modified membranes were taken out from the holder and washed with pure water five times to completely eliminate the unreacted agents. Finally, the cleaned membranes were exposed to air for 3 h for post-treatment and stored in pure water for further experiments. All modifications were performed at 25 °C. The modified membranes with different amounts of g-C₃N₄ nanosheets (i.e., 0, 0.005, 0.01, 0.02, and 0.04%) are denoted as M0, M1, M2, M3, and M4, respectively.

Fig. 1 Schematic of bio-inspired coating on HPAN UF substrates through co-deposition of dopamine/PEI and g-C₃N₄ nanosheets triggered by APS.

2.3 Characterization

2.3.1 Morphology and composition of treated g-C₃N₄ nanosheets. Transmission electron microscopy (TEM, FEI Tecnai G2 F20) was used to observe the morphology of the plasma-treated g-C₃N₄ nanosheets. Atomic force microscopy (AFM) was conducted by using a Bruker Dimension Icon in a tapping mode (Bruker, Billerica, MA). X-ray photoelectron spectroscopy (XPS) measurement was performed to explore the elemental composition of the plasma-treated C₃N₄ nanosheets.

2.3.2 Surface morphologies of the membranes. Scanning electron microscopy (SEM) analysis was employed to visualize the surface structure of the modified membranes using an FE-SEM (Hitachi SU8010, Japan) at a voltage of 5 kV. Prior to the SEM analysis, the membrane samples were completely dried at room temperature and then sputter-coated with gold nanoparticles.

The roughness of the membrane surface was determined by AFM under ambient conditions using a Bruker Dimension Icon with a FastScan-B tip with a scan area of 5 × 5 μm². The root-mean-squared roughness (R_rms) was used to evaluate the surface morphology.

2.3.3 Hydrophilicity and surface charges. The wettability of the prepared membranes was determined using a Krüss DSA 10 Mk2 system (Germany). Briefly, a water droplet was placed onto a dry surface of the modified membrane with dimensions of 1 × 10 cm². Afterward, the dynamic contact angle between the water and membrane surface was recorded within 60 s.

The surface charges on the modified membranes were measured in terms of zeta potential using a SurPASS electrokinetic analyzer (Anton Paar GmbH); all the measurements were performed in 1 mM KCl solution at pH 6.0 ± 0.2 and 25 °C.

2.3.4 Pore sizes. The pore sizes of the prepared membranes were measured through retention of 200 mg L⁻¹ polyethylene glycol polymers with average MWs of 200, 300, 400, 600, 1000, and 2000 Da. The separation of polyethylene glycol polymers for these prepared membranes was performed at 4 bar. The concentrations of the polyethylene glycol polymers were determined with a total organic carbon analyzer (Shimadzu, Japan). The rejection (R) of the polyethylene glycol polymers was evaluated from the concentration of the polyethylene glycol polymers in the feed (C₀) and permeate (C_i), which is described in the following equation:

Aiming to quantitatively determine the pore size of these as-prepared membranes, the log-normal density function between rejections of polyethylene glycol polymers and their Stokes diameters (d_p) was employed:³²

where μ_p denotes the mean effective pore size, defined as d_p at R of 50%. σ_p represents the geometric standard deviation, obtained as the ratio between d_p at R of 84.13% and μ_p.

Based on this log-normal density function, the MWCO of the bio-inspired membranes, defined as the molecular weights (MW) of solutes at R of 90%, can be determined through the relationship between the Stokes diameter (d_s) of solutes (i.e., polyethylene glycol) and their MW as shown below:³²

d_s = 33.46 × 10⁻¹² × MW^0.557 (3)

2.4 Filtration behaviors

The separation behaviors of the as-prepared membranes were assessed in a home-made cross-flow filtration cell. After compaction of the membranes through the filtration of pure water, a stable permeation flux was obtained at 25 °C. Subsequently, the filtration of the salt solutions (1.0 g L⁻¹ NaCl or Na₂SO₄) was conducted at 4 bar. Additionally, the rejection of three reactive dyes (i.e., RO16, RO1 and RB19) at 500 ppm was measured at different feed pH values ranging from 3 to 11 and salt concentrations (i.e., up to 10.0 g L⁻¹) to evaluate the applicability of the modified membranes for separating reactive dyes and inorganic salts. The permeability (J) and rejection (R_s) of dyes and salts for the as-prepared membranes are determined as follows:where V is the permeation volume at time interval Δt, ΔP is the applied operational pressure, and S is the effective area of the membrane coupons (i.e., 22.9 cm²). C_p and C_f are the concentrations of dyes or salts in the permeate and feed, respectively. Specifically, the concentrations of reactive dyes were determined with a double beam UV-vis spectrophotometer (Shimadzu UV-1601), and the concentrations of inorganic salts (NaCl and Na₂SO₄) were obtained using a Dionex ICS-2000 ion chromatography system.

2.5 Fouling propensity

The fouling propensity of the modified membranes was evaluated with a 100 ppm humic substance solution (mainly humic acid and fulvic acid) extracted from landfill leachate. In a typical process, pure water was filtered at 4 bar for 1 hour to render a steady initial permeation flux (J_w,0). Subsequently, the humic substance solution was filtered for another hour and the final permeation flux was recorded as J_s. Ultimately, the fouled membranes were thoroughly rinsed with pure water for 1 hour, which provided the recovered permeation flux (J_w,1). The antifouling behaviors of the as-prepared membranes were determined in 3 cycles. The flux recovery ratio (η) was determined to analyze the fouling propensity of the membranes, which is expressed as:

To describe the flux decline during the fouling experiments, the total fouling ratio (R_t) was used; it is defined as the degree of the flux decline caused by reversible and irreversible fouling. The reversible fouling (R_r) refers to the permeation flux reduction due to a gel layer deposited on the membrane surface and the irreversible fouling (R_ir) is ascribed to the blocking of the foulants into the membrane pores or adsorption onto the surface. These three fouling ratios are calculated as follows:

3. Results and discussion

3.1 Characterization of hydrophilic g-C₃N₄ nanosheets

The intrinsic hydrophobicity of the bulk g-C₃N₄ not only renders it difficult to well disperse in the aqueous solution but also does not improve the membrane wettability during membrane modification. Therefore, it is important to improve its hydrophilicity. After oxygen plasma treatment, the g-C₃N₄ can be well exfoliated into a layer structure with dimensions of ca. 200 nm × 200 nm (Fig. 2A and B). The AFM image (Fig. 2A) shows that the bulk g-C₃N₄ was successfully treated and exfoliated into one-layered nanosheets with a thickness of ca. 0.5 nm.³³

Fig. 2 (A) AFM and (B) TEM images of treated g-C₃N₄ nanosheets; (C) XPS spectra of g-C₃N₄ nanosheets before and after oxygen plasma treatment; (D) dispersion of bulk g-C₃N₄ in water after one week; (E) dispersion of g-C₃N₄ nanosheets in water after three months with occurrence of the Tyndall effect.

XPS analysis was conducted to investigate the elemental composition of pristine and treated g-C₃N₄ nanosheets (Fig. 2C). The characteristic peaks for C 1s, N 1s and O 1s species can be clearly observed in the XPS spectra of g-C₃N₄ nanosheets before or after oxygen plasma treatment. The O element in pristine g-C₃N₄ nanosheets is derived from intermediates of the melamine after thermal polymerization.³⁴ After oxygen plasma treatment, an increasing intensity of the O 1s peak but a decreasing intensity of both C 1s and N 1s peaks in g-C₃N₄ nanosheets were observed. Specifically, the atomic percentage of the O element increased from 11.3% to 25.7% (Table S1†). These results indicate that the oxygen plasma treatment resulted in successful incorporation of the O element into the g-C₃N₄ nanosheets, thereby potentially improving their wettability.³⁵

3.2 Surface morphologies of the membranes

Fig. S2† shows the digital images of the pristine and modified HPAN-based membranes treated with polydopamine (PDA)/PEI coating. As shown in Fig. S2A and B,† the surface of the HPAN substrate changed from white to homogeneously brown after the exposure to the PDA/PEI solution, suggesting that APS can trigger the coating of the PDA/PEI layer on the HPAN substrate through the Schiff base reaction and Michael addition. However, the color of bio-inspired coated membrane surfaces was slightly lighter when the g-C₃N₄ nanosheets were present in the PDA/PEI solutions during co-deposition (Fig. S2C–F†), suggesting the successful co-deposition of g-C₃N₄ nanosheets onto the membrane surfaces.

In order to study the effect of incorporation of hydrophilic g-C₃N₄ nanosheets on the surface morphology of the PDA/PEI layer, surface SEM and AFM images of the modified membranes were visualized (Fig. 3). The pristine HPAN membrane had a smooth surface. Specifically, the valleys in the AFM image (Fig. 3A), which denote the pore structure of the HPAN substrate, resulted in a surface roughness of 39.4 nm. After coating with the PDA/PEI layer, the pristine HPAN substrate with a porous structure was fully covered with a smooth and tight surface (Fig. 3B). However, some nanoparticles appeared and made the surface of the M0 membrane rougher, which was confirmed by the increase in the roughness (R_rms = 50.7 nm). This is attributed to the aggregation of PDA through covalent polymerization of dopamine. Furthermore, the surface morphologies of the membranes incorporated with g-C₃N₄ nanosheets were significantly altered. After the co-deposition of g-C₃N₄ nanosheets, the membrane surfaces became more compact (Fig. 3C–F). The g-C₃N₄ nanosheets present in the PDA/PEI solution may stimulate the generation of sulfate radicals from APS, allowing for fast polymerization of dopamine and PEI near the vicinity of g-C₃N₄ nanosheets. This allows the bio-inspired coating to firmly adhere to the surface of the g-C₃N₄ nanosheets, providing robust linkages between the g-C₃N₄ nanosheets and membrane surface and consequently enhancing the solute selectivity. Simultaneously, the covalent polymerization of dopamine was limited, minimizing the aggregation of PDA nanoparticles. On the other hand, the g-C₃N₄ nanosheets were stacked with more nodules observed on the membrane surfaces as the concentrations of nanosheets increased, exhibiting scale-like structures and rougher morphologies. Noticeably, the stacking of the g-C₃N₄ nanosheets would create more nanochannels, which is likely beneficial for enhanced permeation flux.

Fig. 3 Surface SEM and AFM images of the pristine HPAN substrate and as-prepared membranes. (A) Pristine HPAN; (B) M0; (C) M1; (D) M2; (E) M3; (F) M4.

3.3 Hydrophilicity and surface charge

Generally, the incorporation of nanoparticles into the polymer matrix is a key strategy to enhance the permeability and alleviate membrane fouling during filtration. Fig. 4A shows the dynamic contact angles of the pristine and bio-inspired HPAN-based membranes. Specifically, the initial contact angle of the HPAN-based membranes slightly dropped from 44.9° to 40.6° after coating with the PDA/PEI composite layer (i.e., M0 membrane), attributed to the presence of amino or hydroxyl groups derived from dopamine and PEI on the membrane surface for enhanced hydrophilicity. The wettability of the PDA/PEI coating was further boosted with co-deposition of hydrophilic g-C₃N₄ nanosheets. For example, the initial contact angle of the bio-inspired membranes was reduced from 40.6° to 37.1° as a result of the bio-inspired co-deposition of 0.005% hydrophilic g-C₃N₄ nanosheets. When more g-C₃N₄ nanosheets (i.e., 0.04%) were deposited onto the PDA/PEI layer, the initial contact angle of the membrane surface further dropped to 28.7°. This was mainly due to the greater stack of hydrophilic g-C₃N₄ nanosheets (SEM images in Fig. 3). Fig. 4B highlights the stable static contact angles of the membranes resulting from the bio-inspired co-deposition of g-C₃N₄ nanosheets. The static contact angles of the bio-inspired membranes exhibited a negative dependence on the amount of the co-deposited g-C₃N₄ nanosheets. However, the correlation between the membrane hydrophilicity and the amount of g-C₃N₄ nanosheets was slightly non-linear at higher concentrations of these nanosheets (i.e., 0.04%). This is mainly restricted by the intrinsic hydrophilicity of the g-C₃N₄ nanosheets after oxygen plasma treatment. In order to obtain membranes with higher wettability, a longer oxygen plasma treatment is recommended to further improve the hydrophilicity of the g-C₃N₄ nanosheets.

Fig. 4 (A) Time-dependent dynamic water contact angles, (B) stable static water contact angles, and (C) zeta potentials of the pristine HPAN substrate and as-prepared membranes.

In view of resource recovery from highly saline wastewater by membrane technology, a high salt permeation is favorable to obtain effective fractionation of organics and salts. Generally, the charge of the membrane surface critically dominates the separation behavior of the charged species through the Donnan effect, especially for multivalent ions. Therefore, the surface charges of the bio-inspired membranes were determined through zeta potential measurements (Fig. 4C). After the deposition of the PDA/PEI layer, the zeta potential of the coated HPAN substrate (i.e., M0 membrane) increased from −40.3 ± 2.4 to −6.7 ± 1.1 mV. On the one hand, the carboxyl groups with negative charges on the pristine HPAN substrate were completely covered by the bio-inspired coating; on the other hand, the binding of the positively charged PEI onto the membrane surface elevated the zeta potential to some extent. However, the result in this study deviated slightly from a previously reported positive zeta potential (4.3 mV) of a PDA/PEI coating with an H₂O₂/CuSO₄ trigger.³⁶ The plausible explanation for this difference is the aggregation of PDA nanoparticles with negative charges on the membrane surfaces, which resulted in a lower zeta potential in this study, as observed in the SEM images in Fig. 3B. Moreover, the zeta potential of the bio-inspired membranes positively increased with the concentration of the co-deposited g-C₃N₄ nanosheets. For example, the zeta potential of the M4 membrane surface was 7.6 ± 0.8 mV with the co-deposition of 0.04% g-C₃N₄ nanosheets onto the PDA/PEI layer. As previously mentioned, the presence of the g-C₃N₄ nanosheets triggered more sulfate radicals and thus facilitated the binding of more PEI molecules with dopamine, confining the self-polymerization of dopamine and thus giving rise to a highly positive charge density on the membrane surface. This elevated zeta potential may enable high salt permeation, especially for Na₂SO₄, thereby providing a higher organic/salt selectivity.

3.4 Membrane filtration performance

3.4.1 Pore size, pure water permeability and salt permeation. The pore size of the membranes is a significant factor in the fractionation of organics and salts. The pore size and MWCO of the modified membranes via bio-inspired co-deposition of g-C₃N₄ nanosheets were characterized through eqn (2); the results are summarized in Fig. 5A.

Fig. 5 (A) Pore size distribution, (B) pure water permeability, and (C) filtration performance of the bio-inspired membranes in 1.0 g L⁻¹ NaCl and Na₂SO₄ solutions.

As a substrate, the pristine hydrolyzed PAN membrane with a porous structure exhibited a pure water permeability of 143.9 ± 7.6 L m⁻² h⁻¹ bar⁻¹. After 4 h bio-inspired coating of the dopamine/PEI solution, the bio-inspired membrane (i.e., M0) yielded an effective mean pore size (μ_p) of 0.61 nm (i.e., 1355 Da MWCO), presenting the effective formation of the PDA/PEI layer. However, the pure water permeability decreased to 22.3 ± 1.0 L m⁻² h⁻¹ bar⁻¹, due to the enhanced hydraulic membrane resistance. After the incorporation of g-C₃N₄ nanosheets onto the PDA/PEI layer, the pore sizes of these membranes become more narrow and uniform, as reflected from the steeper slopes of the solute rejection curves and the decreasing geometric standard deviations (σ_p). In particular, when the concentration of the g-C₃N₄ nanosheets reached 0.01%, the modified membrane (i.e., M2) had an MWCO of less than 1000 Da (i.e., 985 Da). This suggests the successful tailoring of the pore sizes from the UF to the NF scale. When increasing the concentration of g-C₃N₄ nanosheets to 0.04%, the pore size (i.e., 0.52 nm for μ_p) of the modified membrane further decreased with 592 Da MWCO. This was mainly attributed to the fact that the increasing concentration of g-C₃N₄ nanosheets present in dopamine/PEI solution induced the generation of sulfate radicals originating from APS to rapidly trigger the polymerization of dopamine and PEI through the Michael addition and Schiff base reactions, yielding a denser and more compact structure.

Noticeably, the permeability of the membranes after the bio-inspired co-deposition of g-C₃N₄ nanosheets increased monotonically with the concentration of g-C₃N₄ nanosheets, even though the membrane surface became much denser (Fig. 5B). Specifically, the pure water permeability of the bio-inspired membrane (M4 with 592 Da MWCO) was significantly boosted to 28.4 ± 1.2 L m⁻² h⁻¹ bar⁻¹ as the concentration of incorporated g-C₃N₄ nanosheets reached 0.04%. Such improvement in permeation flux for these bio-inspired membranes with co-deposition of g-C₃N₄ nanosheets is mainly attributed to (1) the greater hydrophilicity of the membrane surfaces functionalized with hydrophilic g-C₃N₄ nanosheets and (2) the stacking of g-C₃N₄ nanosheets in the bio-inspired PDA/PEI layer creating more nanochannels for faster transport of water molecules. Compared to previously reported NF membranes, the co-deposition of g-C₃N₄ nanosheets into the PDA/PEI layer leads to the most advantageous water permeability and enhanced solute selectivity, which breaks through the permeability–selectivity trade-off effect (Fig. S3†).

The filtration performance of the as-prepared membranes through bio-inspired co-deposition of g-C₃N₄ nanosheets in the salt solutions was investigated (Fig. 5C). The PDA/PEI coating onto the HPAN-based membrane provided a fairly low rejection of Na₂SO₄ (i.e., 14.5 ± 1.6%) and NaCl (i.e., 4.7 ± 0.6%), although this bio-inspired coating was negatively charged. However, the reduced steric effect caused by the large pore size dominated the salt retention behavior, which facilitated the salt permeation. With the co-deposition of hydrophilic g-C₃N₄ nanosheets, the membranes had a decreasing pore size, but an increasing permeation of NaCl and Na₂SO₄ was observed. Specifically, the rejection of Na₂SO₄ for the modified membranes decreased from 14.5 ± 1.6% to 7.6 ± 0.8% as the concentration of the g-C₃N₄ nanosheets increased from 0 to 0.04%. This was mainly due to the increasing number of nanochannels created by stacking of g-C₃N₄ nanosheets onto the membrane surface. Additionally, the increased zeta potential on the membrane surfaces partially contributed to enhanced salt transport (Fig. 4C).

To recover the renewable resources from highly saline wastewater, conventional dense NF membranes typically have a moderate or high salt rejection, compromising the permeation flux due to the enhanced osmotic pressure; this limits the sufficient fractionation of organics and salts. In this study, the higher salt permeation of the bio-inspired membranes functionalized with g-C₃N₄ nanosheets would suppress the reduction in permeation flux.

3.4.2 Fractionation performance for high-salinity textile wastewater. Currently, the advanced treatment of highly saline textile wastewater encourages sustainable resource recovery through fractionation of dye species and inorganic salts. Therefore, large amounts of inorganic salts should be eliminated from textile wastewater to recover high-purity dyes with improved stability for reuse. To explore the applicability of the modified membranes in the sustainable treatment of textile wastewater, the membrane filtration performance under different conditions (i.e., pH and salt additives) is shown in Fig. 6.

Fig. 6 (A) Filtration performance of the modified membranes in three dye solutions (RO16, RO1, and RB19); (B) effect of feed pH on RO16 rejection by the M0 and M4 membranes; (C) effect of the NaCl concentration on RO16 rejection by the M0 and M4 membranes; (D) effect of Na₂SO₄ concentration on RO16 rejection by the M0 and M4 membranes.

The PDA/PEI layer (i.e., M0 with 1355 Da MWCO) provided a moderate retention to three dyes (i.e., >96%) on the HPAN membrane, due to synergistic effects of steric exclusion and Donnan repulsion (Fig. 6A). Unfortunately, this moderate dye rejection may be insufficient to effectively extract these reactive dyes during diafiltration for dye desalination, since these reactive dyes with small molecular sizes would penetrate through the membrane. Furthermore, the bio-inspired coating functionalized with the hydrophilic g-C₃N₄ nanosheets enabled an increasing tendency in the retention of these reactive dyes because of the reduced pore size for enhancing steric exclusion. In particular, an increase in the concentration of the g-C₃N₄ nanosheets to 0.04% achieved nearly complete retention (i.e., >99.3%) of the reactive dyes. Simultaneously, the permeation flux of the dye solutions for the modified membranes was also elevated with incorporation of the g-C₃N₄ nanosheets. Since the fouling of dyes occurred on the membrane surface, the permeation flux of these bio-inspired membranes in all dye solutions was lower than that in pure water (Fig. 5B). Nevertheless, a sufficiently high permeation flux can be successfully achieved for practical applications. For instance, the permeability of the bio-inspired membrane (i.e., M4) functionalized with 0.04% hydrophilic g-C₃N₄ nanosheets in dye solutions was observed to be around 23.7 L m⁻² h⁻¹ bar⁻¹.

Fig. 6B shows the effect of the feed pH on the dye rejection of the bio-inspired membranes. The dye rejection of the M0 and M4 membranes was boosted at higher pH, due to the enhanced Donnan effect. Although the electrostatic attraction between the dye molecules and membrane surface reduced the rejection of the charged dyes at low pH, the M4 membrane with a smaller pore size still had a considerably higher dye rejection (i.e., 98.7 ± 0.3% for RO16 at pH 3) than the M0 membrane (94.8 ± 0.6% for RO16) at the same pH.

In textile wastewater derived from dye production or dyeing processes, inorganic salts (i.e., NaCl or Na₂SO₄) are either additives or by-products, which have a tremendous impact on the filtration performance of the membranes. Fig. 6C and D show the effect of NaCl or Na₂SO₄ on the fractionation of dyes and salts for these bio-inspired membranes. The rejection of the RO16 dye for the M0 membrane decreased from ca. 96.3% to ca. 92.8% with increasing concentration of inorganic salts from 0 to 10.0 g L⁻¹. This is due to the fact that the presence of inorganic salts in the dye solutions can generally intensify the “shielding effect” and thus weaken the electrostatic repulsion of the charges in the membrane pores, which facilitates the transport of dyes through the membranes.^37–40 Nevertheless, the size exclusion mechanism dominated the dye retention behavior over the “shielding effect” for the M4 membrane with a smaller pore size. Therefore, only a slight reduction (i.e., 0.3%) in the dye rejection was observed in the considered salt concentration range. To highlight the advantage of these bio-inspired membranes, Table S2† summarizes the filtration performance of the synthesized NF membranes previously reported in terms of water permeability and salt/dye retention. Combining the extremely low rejection of inorganic salts (i.e., NaCl or Na₂SO₄) and high retention of dyes, the NF membranes functionalized with the g-C₃N₄ nanosheets through bio-inspired co-deposition exhibited a superior performance for dye and salt fractionation in the dye/salt binary solutions.

3.5 Fouling propensity

NF membranes may suffer from the fouling caused by organic foulants, e.g., humic substances, which are commonly present in the aqueous environment. In this case, humic substances (mainly humic acid and fulvic acid) extracted from landfill leachate were employed as model foulants. Fig. 7 shows the fouling propensity of the bio-inspired membranes functionalized with g-C₃N₄ nanosheets for the humic substance foulant solutions.

Fig. 7 (A) Normalized flux of the bio-inspired membranes before and after the filtration of humic substance solution; (B) flux recovery and fouling resistance ratio of the membranes.

Initially, a stable flux was obtained during the filtration of pure water, indicating that no membrane compaction occurred at 4 bar. When the pure water was replaced with the humic substance solution for filtration, the permeation flux of the bio-inspired membranes was markedly decreased, due to the membrane pore blocking and foulant cake layer. The permeation flux of these membranes was recovered to varying degrees after rinsing with pure water for foulant removal during the 3-cycle filtration (Fig. 7A).

The flux recovery rate and fouling resistance were calculated for a quantitative evaluation of the antifouling performance of the bio-inspired membranes (Fig. 7B). The pristine PDA/PEI coating (i.e., M0) had the lowest flux recovery rate (78.8%) but the highest R_t value (33.7%), presenting a high fouling propensity. When the hydrophilic g-C₃N₄ nanosheets were incorporated into the PDA/PEI layer, all bio-inspired membranes exhibited higher flux recovery (>83.0%) while maintaining a low R_t value. Specifically, the flux recovery of the M4 membrane functionalized with the highest concentration of the g-C₃N₄ nanosheets (i.e., 0.04%) reached 93.7%, and the R_t value was only 20.8%. This is due to the increased hydrophilicity of the membrane surfaces functionalized with hydrophilic g-C₃N₄ nanosheets, which significantly improved the antifouling performance. On the other hand, the increasing roughness of the membrane surfaces functionalized with the g-C₃N₄ nanosheets potentially facilitated the deposition of organic foulants, resulting in a slight increase in the reversible fouling resistance (R_r); these accumulated foulants can be eventually removed by water rinsing. However, the M0 membrane with the pristine PDA/PEI coating suffered from the most severe irreversible fouling, likely attributed to the pore blocking. With an increase in the concentration of g-C₃N₄ nanosheets, a decrease in the irreversible fouling ratio was yielded, as the reduction in pore sizes of these bio-inspired membranes prevented the organic foulants with larger sizes from entering the pore structure. Therefore, the g-C₃N₄ nanosheets can tailor the PDA/PEI layer from the UF to the NF scale, providing an excellent antifouling and filtration performance to the modified membranes.

Conclusions

In this study, a bio-inspired co-deposition of PDA/PEI with hydrophilic g-C₃N₄ nanosheets triggered by APS was proposed to construct high-flux NF membranes through bridging UF membrane surface cavities. SEM and AFM measurements demonstrated the incorporation of g-C₃N₄ nanosheets onto the PDA/PEI layer, which provided an excellent antifouling and filtration performance to the bio-inspired NF membranes. This is plausibly related to the fact that g-C₃N₄ nanosheets can act as catalysts in inducing the generation of sulfate radicals from APS, which accelerates the polymerization behavior of dopamine and PEI, and thus alters the structure of the membrane surface. In particular, the incorporation of g-C₃N₄ nanosheets can tailor the PDA/PEI layer from the UF to the NF scale in order to boost the retention of various reactive dyes (610–630 Da, >99.3%) with enhanced pure water permeability (28.4 ± 1.2 L m⁻² h⁻¹ bar⁻¹). Low rejections of inorganic salts (i.e., 2.9% for NaCl; 7.6% for Na₂SO₄) can still be achieved, establishing these bio-inspired membranes as competitive and practical alternatives in dye desalination and purification of highly saline textile wastewater. Given the outstanding photocatalytic properties of the incorporated g-C₃N₄ nanosheets under visible light exposure, the photodegradation performance of these bio-inspired membranes will be investigated in future work.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors would like to acknowledge the funding support from the National Natural Science Foundation of China (Grant No: 21707018 and 21706035), the Natural Science Foundation of Fujian Province (Grant No: 2019Y0006 and 2017J01413) and the Fujian Agriculture and Forestry University Program for Distinguished Young Scholar (Grant No.: xjq201704) for this work.

References

1. J. Lin, C. Y. Tang, W. Ye, S. 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.
2. J. Lin, W. Ye, J. Huang, B. Ricard, M. Baltaru, B. Greydanus, S. Balta, J. Shen, M. Vlad, A. Sotto, P. Luis and B. Van der Bruggen, ACS Sustainable Chem. Eng., 2015, 3, 1993–2001.
3. S. Wu, S. Zou, G. Liang, G. Qian and Z. He, Sci. Total Environ., 2018, 610–611, 137–146.
4. P. L. McCarty, J. Bae and J. Kim, Environ. Sci. Technol., 2011, 45, 7100–7106.
5. Z. F. Gao, Y. Feng, D. Ma and T. S. Chung, J. Membr. Sci., 2019, 574, 124–135.
6. Y. Zhang, S. Japip and T. S. Chung, Carbon, 2017, 123, 193–204.
7. G. M. Geise, H. B. Park, A. C. Sagle, B. D. Freeman and J. E. McGrath, J. Membr. Sci., 2011, 369, 130–138.
8. K. Cao, Z. Jiang, X. Zhang, Y. Zhang, J. Zhao, R. Xing, S. Yang, C. Gao and F. Pan, J. Membr. Sci., 2015, 490, 72–83.
9. B. D. Freeman, Macromolecules, 1999, 32, 375–380.
10. L. Yu, Y. Zhang, H. Zhang and J. Liu, Desalination, 2015, 359, 176–185.
11. J. Zhu, M. Tian, J. Hou, J. Wang, J. Lin, Y. Zhang, J. Liu and B. Van der Bruggen, J. Mater. Chem. A, 2016, 4, 1980–1990.
12. D. Hu, Z. Xu and C. Chen, Desalination, 2012, 301, 75–81.
13. Z. Tan, S. Chen, X. Peng, L. Zhang and C. Gao, Science, 2018, 360, 518.
14. L. Li, S. Zhang and X. Zhang, J. Membr. Sci., 2009, 335, 133–139.
15. K. P. Lee, G. Bargeman, R. de Rooij, A. J. B. Kemperman and N. E. Benes, J. Membr. Sci., 2017, 523, 487–496.
16. M. Liu, D. Wu, S. Yu and C. Gao, J. Membr. Sci., 2009, 326, 205–214.
17. J. Zhao, Y. Su, X. He, X. Zhao, Y. Li, R. Zhang and Z. Jiang, J. Membr. Sci., 2014, 465, 41–48.
18. B. Tang, Z. Huo and P. Wu, J. Membr. Sci., 2008, 320, 198–205.
19. C. Zhou, Y. Shi, C. Sun, S. Yu, M. Liu and C. Gao, J. Membr. Sci., 2014, 471, 381–391.
20. Q. An, W. Sun, Q. Zhao, Y. Ji and C. Gao, J. Membr. Sci., 2013, 431, 171–179.
21. W. J. Lau, S. Gray, T. Matsuura, D. Emadzadeh, J. Paul Chen and A. F. Ismail, Water Res., 2015, 80, 306–324.
22. J. Yin, G. Zhu and B. Deng, Desalination, 2016, 379, 93–101.
23. A. Anand, B. Unnikrishnan, J. Mao, H. Lin and C. Huang, Desalination, 2018, 429, 119–133.
24. Y. Wang, L. Li, Y. Wei, J. Xue, H. Chen, L. Ding, J. Caro and H. Wang, Angew. Chem., Int. Ed., 2017, 56, 8974–8980.
25. J. Zhang, M. Zhang, R. Sun and X. Wang, Angew. Chem., Int. Ed., 2012, 51, 10145–10149.
26. Z. Lin and X. Wang, Angew. Chem., Int. Ed., 2013, 52, 1735–1738.
27. X. Wang, X. Chen, A. Thomas, X. Fu and M. Antonietti, Adv. Mater., 2009, 21, 1609–1612.
28. Q. Han, B. Wang, J. Gao, Z. Cheng, Y. Zhao, Z. Zhang and L. Qu, ACS Nano, 2016, 10, 2745–2751.
29. M. Hu and B. Mi, Environ. Sci. Technol., 2013, 47, 3715–3723.
30. X. Wang, S. Blechert and M. Antonietti, ACS Catal., 2012, 2, 1596–1606.
31. Q. Lin, L. Li, S. Liang, M. Liu, J. Bi and L. Wu, Appl. Catal., B, 2015, 163, 135–142.
32. M. Jiang, K. Ye, J. Deng, J. Lin, W. Ye, S. Zhao and B. Van der Bruggen, Environ. Sci. Technol., 2018, 52, 10698–10708.
33. J. Xu, L. Zhang, R. Shi and Y. Zhu, J. Mater. Chem. A, 2013, 1, 14766–14772.
34. Z. Huang, J. Song, L. Pan, Z. Wang, X. Zhang, J. Zou, W. Mi, X. Zhang and L. Wang, Nano Energy, 2015, 12, 646–656.
35. X. Bu, J. Li, S. Yang, J. Sun, Y. Deng, Y. Yang, G. Wang, Z. Peng, P. He, X. Wang, G. Ding, J. Yang and X. Xie, ACS Appl. Mater. Interfaces, 2016, 8, 31419–31425.
36. J. Wang, J. Zhu, M. T. Tsehaye, J. Li, G. Dong, S. Yuan, X. Li, Y. Zhang, J. Liu and B. Van der Bruggen, J. Mater. Chem. A, 2017, 5, 14847–14857.
37. 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.
38. J. Lin, W. Ye, M. 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.
39. W. Ye, J. Lin, R. Borrego, D. Chen, A. Sotto, P. Luis, M. Liu, S. Zhao, C. Y. Tang and B. Van der Bruggen, Sep. Purif. Technol., 2018, 197, 27–35.
40. J. Qin, X. Dai, Y. Zhou, L. Zhang, H. Chen and C. Gao, J. Membr. Sci., 2014, 468, 242–249.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9en00692c

---