1. Introduction
Nowadays, water pollution is one of the most concerned and worldwide issues, raising global concern. There are many kinds of wastewater with different compositions and properties, which bring great obstacles by conventional treatment methods. For example, with the development of textile industry, amounts of textile wastewater were produced and approximately 3600 different types of dyes and 8000 types of chemicals were leached [1]. 1.84 billion tons of textile wastewater were generated every year in China, which contained high colority, salinity, biochemical oxygen demand (BOD), chemical oxygen demand (COD) and toxicity [2], posing great threats to environment and living organics [1]. Also, amounts of emerging contaminants (e.g., pharmaceuticals, perfluorooctanoic acid) are discharged into the pharmaceutical and municipal wastewater, which are difficult to be removed by the conventional wastewater treatment process (e.g., adsorption, coagulation, biological treatment) in wastewater treatment plants owing to their non-biodegradable and highly persistent nature [3, 4]. In addition, wastewater reuse attracted intensive attention to face its challenges of large discharge volume and high treatment difficulty. China stipulates wastewater reuse rate should be over 60%. Moreover, the rich pollutants in wastewater can be recovered to high values chemicals, such as salt, dye, which reduced the water treatment cost and enabled sustainable and cycling wastewater treatment [5]. Unfortunately, the resource recovery was still insufficient, which was restricted by the technology and engineering factors. For example, the varied pollutants in wastewater are difficult to be separated and low-purity products cannot compensate the high input of treatment.
Benefiting from small footprint, efficient permeate quality and selective separation of target pollutants, nanofiltration (NF) technology attracted intensive attention in wastewater treatment and resource recovery [6–8]. Moreover, molecular weight cut-off (MWCO) of NF ranges from 200 to 1000 Da, which falls in the range of majority of small molecules in wastewater (e.g., dye, antibiotics), which is feasible to separate organic molecules and multivalent salts [9]. Nowadays, mainstream NF membranes are based on a thin film composite (TFC) polyamide (PA) active layer that synthesized by interfacial polymerization (IP) on top of a porous support layer, which became a gold standard for desalination in the past decades [10, 11]. PA active layer was quite thin (<100 nm), thus high crosslinking was employed to avoid defect formation by forming dense active layer, causing low permeability and suffering from the permeability-selectivity trade-off [12]. Moreover, the low MWCO of conventional NF membrane cause its insufficient selective separation of pollutants (e.g., dye and antibiotics) and salt, which is likely not feasible for resource recovery from wastewater. More importantly, conventional NF membranes are prone to occur fouling in complex wastewater since the pollutants tend to accumulate and adhere on membrane surface or inside pore, causing severe flux loss, shorten lifetime and deteriorated permeate quality [13, 14]. Therefore, it is impressive to develop next-generation NF membrane toward wastewater reuse and resource recovery with low membrane filtration resistance, high antifouling ability and selective recovery of resources in wastewater. In this regard, many strategies were developed in recent years, such as modifying PA membrane with nanomaterials [15], developing alternative fabrication methods for highly porous membrane matrix generation [16] or designing non-PA NF membranes [17]. NF membrane can be divided into tight NF (TNF) and loose NF (LNF). Among them, LNF membrane holds great potential to separate organic pollutant/salt owing to the merits of low operating pressure, high permeability and selectivity. Generally, LNF membranes have wider and higher MWCO (500–2000 Da) than the traditional TNF membrane (MWCO<500 Da) [18], which allows the permeation of monovalent and multivalent salts through the membrane. Moreover, compared with traditional NF membranes, LNF membranes have a looser selective layer, causing low filtration resistance and energy consumption, and high water permeability [12]. Above characterisitics can be recognized as the specific criteria of LNF to differentiate with common NF. Tight UF can also be categorized as LNF. The key performance differences between loose NF and tight NF were provided in Table S1. Normally, LNF membrane presented higher permeability, rejection of organic pollutants, lower rejection of multivalent and monovalent salts, higher organic pollutants (e.g., dye and pharmaceuticals) recovery efficiency and less membrane fouling compared with TNF membrane. Besides, they can also combine with biological and catalytic treatment for advanced treatment of contaminants and membrane fouling control [19]. Therefore, they are promising to apply in wastewater treatment, resource recovery and pharmaceuticals purifications, which address the limitations of tight NF membranes [20]. We searched the published literature from 2015 to 2024 from web of science using “loose nanofiltration”, “tight ultrafiltration” and “water treatment” as the keywords. As depicted in Fig. 1, the literature numbers sharply increased from 2015 to 2022 and then remained constant at approximately 100 articles per year from 2022 to 2024, suggesting this field received more and more attention.
Nevertheless, compared to the plenty of reviews that reported the NF membrane for water treatment [6, 21], there are rare reviews that concluded the fabrication of LNF membranes and their application in wastewater treatment. Thus this review systematically concluded the conventional and state-of-the-art fabrication methods of LNF membranes. Then the application of LNF membranes in wastewater treatment and resource recovery was discussed in the case of typical industrial wastewater and municipal wastewater. In addition, the challenges, gaps in the research, key trends, trade-offs and future trends of LNF technology in wastewater treatment were proposed to guide their development on fundamental research and engineering applications. By connecting these aspects, we believe this review will fill the gap of LNF membrane structure design and wastewater treatment.
2. Materials Fabrication Methods
LNF membrane can be fabricated by many methods, such as interfacial polymerization, phase inversion, deposition and spinning methods. Notably, these methods were similar with these of common NF membranes, while it has specificity in pore structure and selective layer porosity.
2.1. Interfacial Polymerization
IP process is leading and well-developed technology to produce PA TFC NF membrane by crosslinking between monomers (e.g., trimesoyl chloride (TMC) and piperazine (PIP)) on a substrate [22]. However, it is difficult to precisely control on conventional polymeric UF/MF membranes owing to the limited diffusion and the dense selective layer causes low water permeability; moreover, the poor chlorine resistance limited the biofouling control [23]. To address these issues, intensive efforts were devoted to optimizing IP processes (e.g., adding additive, changing synthesis temperature and pH), developing new monomers (e.g., biobased monomers), or incorporating nanomaterials interlayers (e.g., metal organic frameworks (MOFs), covalent organic frameworks (COFs), graphene oxide (GO), carbon nanotubes (CNTs), MXene (Ti3C2Tx)).
2.1.1. Optimizing IP processes
Optimizing IP processes is important to modulate membrane morphology and performance in industrial applications. Similar to common NF fabrication, modulating the diffusion of monomers and reaction rate can optimize the structure and performance of the active layer. However, the functional layer of LNF membrane should be loose with lower crosslinking degree to modulate the MWCO of membranes to 500–2000 Da. The addition of polyvinyl alcohol (PVA) modulates the diffusion coefficient of PIP, forming Turing-like morphology and thin film according to scanning electron microscope (SEM) images, thus achieving 3.5 times higher than that of pristine NF membrane [24]. Zhang et al. [25] developed a lignin alkali (LA) mediated IP process by promoting the diffusion of PIP molecules into organic phase. The ultrathin and highly cross-linked PA layer breaks the permeability-selectivity trade-off. You et al. [26] utilized surfactant sodium dodecyl benzene sulfonate as additives to accelerate the diffusion rate of amine monomers at the organic phase interface for precise control of roughness and thickness of functional layer. Organophosphorus can retard piperazine diffusion rate via hydrogen bonding, thereby lowering the crosslinking degree of PA selective layer and enhancing permeability to 18.3 L m−2 h−1 bar−1 [27].
In addition, similar to common NF membrane, modulating the IP processes conditions can also adjust the structure and performance of PA TFC LNF membrane, such as monomers concentration [28], synthesis time [29] and pH [30]. Low-temperature (−17°C) IP processes reduced the diffusion of amine monomer, which achieved a relatively loose structure and higher water permeance than the membranes that synthesized at room temperature [31]. Stepwise increasing TMC concentration (SITC) can modulate the compactness and reduce the thickness of selective layer, showing synchronous improvement of permeability and selectivity (Fig. 2a and 2b) [32]. The operating condition should also consider the industrial feasibility since many operating and conditions are difficult to realize a large-scale application. Moreover, more advanced techniques can also employ to adjust the IP processes for more facile and scalable processes.
2.1.2. Developing new monomers
Similar to common NF membrane, although PIP and TMC monomers have demonstrated the robust and high-performance in PA NF membrane fabrication, the rational and molecular design of new monomers is still imperative to enhance the chlorine resistance and separation performance of LNF membrane. For example, PEA TFC membranes with tunable looseness functional layer can be fabricated by IP process between Tris(hydroxymethyl) aminomethane (THAM) and TMC [28] (Fig. 2c). Hu et al. [33] used piperazine-2-carboxylic (PIP-COOH) as monomer to reduce IP reaction rate, which fabricated membrane with higher thickness (up to 200 nm) and negative charge density for superior rejection of negatively charged pollutants. IP reaction between 3,5-diaminobenzamide (3,5-DABA) and TMC can increase the membrane hydrophilicity (Fig. 3a), causing looser and smoother active layer owing to the function of electron-withdrawing and hydrophilic formamide group on 3,5-DABA [34]. IP process between 5-(1-pyrrolidinyl)-1,3-benzenedicarbonyl dichloride (PIPC) and 3-bromopropionic acid (3-BPA) can form loose and zwitterionic layer on the PA membrane surface, thus enhancing permeability and antifouling property [35]. To increase the chlorine resistance, polyester membrane was developed recently by using new monomer (e.g., 3,5-dihydroxy-4-methylbenzoic acid [36], ribitol (Fig. 3b) [37], maltitol [38], tris (4-hydroxyphenyl) phosphine oxide (Fig. 3c) [39], poly(vinyl alcohol) [40], N-(5-methyl acrylamide-2,3,4 hydroxy benzyl) acrylamide [41]), while it was sensitive to hydrolyze at higher pH values and many attempts were devoted to increasing their pH stability.
Recently, biobased monomers attracted increasing attention in green and sustainable fabrication of PA membrane. For example, tannic acid [42], cyclodextrins [43], α-amino acid [44], sugars [45], algal biomass [46], 2,5-furandicarboxaldehyde (FDA) [47]. Biomimetic co-deposition has the advantages of sample control of layer thickness with the aid of self-oxidation by dopamine. Wang et al. [12] prepared LNF membrane by co-deposition of dopamine and polyamide-amine, which presented high water permeability (126.1 L m−2 h−1 bar−1) and methyl blue (MB) rejection (98.9%), but low methyl orange (MO) rejection (13.2%). Except for molecular design of monomer, modulating the monomer distribution, diffusion, and reactivity activity can also optimize IP process, ultimately forming high-performance membranes [48]. Although many new monomers were designed in many previous works, their potential in scalable membrane fabrication and the feasibility of the fabricated LNF membrane in industrial application still remain unclear. Further researches should be devoted to unveiling the fundamental mechanism of controlled reaction and testifying their performance in the presence of different real water matrices containing coexisting pollutants.
2.1.3. Incorporating nanomaterials interlayers
With the fast development of emerging nanomaterials, incorporating nanomaterials interlayers in PA layer became a hotspot research and attracted intensive attention in past decades [49]. Benefiting from the rich pore channels and tunable aperture, the variety nanomaterials can modify the LNF membrane structure (pore structure and morphology) and physiochemical properties (e.g., charge and hydrophobic properties), therefore imparting the membrane with additional water channel and unique properties while maintaining the sieving function. Therefore, incorporating nanomaterials interlayers can influence the separation and anti-fouling performance of LNF membranes, which is quite appealing to break the permeability-selectivity trade-off and enable the selective separation of target pollutants from wastewater [50].
Carbon based functional nanomaterials have the advantages of high stability and tunable fabrication process, and rich porosity, and low production cost, which are widely used to modify PA membrane [51]. Graphene oxide (GO) is a two-dimensional nanomaterial to construct LNF membrane with high pollutants rejection and chemical stability owing to its high specific surface area and rich functional groups [52]. GO intermediate layer between substrate and PA layer increased 10 times higher permeability than the pristine membrane, attributing to its intrinsic water channels for the facilitated water filtration through membrane [53]. PA@GO LNF membranes were constructed by a confined IP method (Fig. 4a) and the reaction was illustrated in Fig. 4b, which created hierarchical channels for water transport and realized high permeance of 75.5 L m−2 h−1 bar−1 [54]. GO can also impart PA membrane with stripe-like and loose structure, breaking the permeability-selectivity trade-off by optimizing the transport channels and charges [55]. GO quantum dots/amino graphene quantum dots interlayer can effectively retard the diffusion of amine monomer by forming hydrogen bond, which slowed down IP process reaction rate and generation of ultra-thin (9.4 nm), smooth and dense PA layer with high permeability (27.2 L m−2 h−1 bar−1) [56]. MXene (Ti3C2Tx) has the merits of rich chemistry property, which can be functionalized by hydroxyl, nitrogen, sulfur, fluorine, and oxygen based groups, which imparted membrane with high selectivity and antifouling abilities [57]; moreover, it imparted membrane with superior chemical and thermal stability than polymer membrane [58]. For example, the rich oxygen-containing functional groups on MXene adsorbed PIP monomer and then inhibited the IP reaction rate, which helped to formation of loss PA membrane with thin active layer (<100 nm) and low filtration resistance [59]. Introducing amino group-functionalized MXene in PA layer can enhance the hydrophilicity and positive charge density, causing superior permeability and rejection toward positively charged pollutants [60]. Wen et al. [61] incorporated MXene nanosheets at the IP interface to regulate diffusion of monomers for fabrication of PA membrane with thin thickness and high hydrophilicity. The incorporation of multi-layer MXenes in PA layer can improve the surface area and provide additional water channels, thereby increasing water permeability by 3 times compared to the pristine membrane [62].
Porous frameworks are emerging nanomaterials in recent years, which attract increasing attention in membrane fabrication owing to their rich porosity and tunable properties, such as MOFs, COFs and hydrogen-bonded organic frameworks (HOFs). MOFs restrict PIP diffusion and help its distribution via hydrogen bonds, which reduce the defect of membrane, and enhance the surface negative charge density and hydrophilicity, thus optimizing water nano-transport path and realizing higher water permeability [22, 63]. A confined IP strategy was developed using the stacking of the PIP-grafted MOFs nanosheets on PA layer, which healed the defect regions in PA layer and acted as separation function, hence improving the separation performance [64]. MOFs nanoflakes can also accelerate the transport of PIP across the interface during IP process, which produces a crumpled and ultrathin PA nanofilm (5 nm) by retaining the gas bubbles and reaction heat in the interfacial reaction zone [15]. Wang et al. [9] developed a rough and irregular nanostructure-mediated IP process to form thin PA layer with crumpled structures by polydopamine (PD) decorated MOFs ZIF-8 as sacrificial templating materials on a porous substrate. Therefore, high water permeability was achieved at 53.2 L m−2 h−1 bar−1 because of the high available area without sacrificing the rejection of NaSO4 (95.2%). Similarly, COFs have good compatibility with PA layer and serve as a co-monomer in IP reaction (Fig. 4c), which can modify the pore size and morphology of PA layer for rejection of target contaminants [65–67]. For example, the unreacted amine groups on defect COFs can interact with acyl chloride groups during IP process, forming PIP-COFs NF membrane with rough surface [68]. Sulfonic acid groups (–SO3H) functionalized COFs nonosheet interlayer reduced the diffusion of PIP by electrostatic interaction, causing the lower thickness of active layer (17 nm) and three times higher permeability than the pristine PA membrane [69]. COFs can also provide abundant active binding sites with pollutants and enhance pollutants removal [66].
New types of functional nanomaterials with low-cost, environmental-friendly or unique functional groups and pore structure, which will endow membrane with new function and avoid the inherent drawbacks of current nanomaterials. Wang et al. selected HOF-21 as interlayer to construct PA layer with highly ordered porous and wrinkled structure, making high water permeability (36.44 L m−2 h−1 bar−1) [70]. Moreover, different dimensions of materials (one to three) can impart membrane with unique structure by designing different transport channels, thus influencing separation performance. For example, 2D materials enabled short water transport channels and favored water permeability [71]. Besides, many emerging IP processes were developed recently to resolve the inherent drawbacks of conventional ones, such as filtration, reverse and electrospray-assisted IP strategies [13]. Although Non-PA NF is non-mainstream, it is an important composition of NF membrane since they have different structure and properties, which may impart membrane with high chlorine-resistance, stability and separation performance and more feasibility in some circumstances [17]. The stability of nanofillers interlayer on PA membrane is still challenging and further studies should resolve these problems. Moreover, nanoparticles with rich porosity adsorb pollutants, which can contribute to pollutant removal during membrane filtration. Therefore, prior to filtration, nanofillers modified membrane should immerse into pollutants solution for certain time until adsorption reached equilibrium to eliminate the effect of adsorption on pollution rejection. If one conducted short-term membranes filtration without adsorption saturation, the impact of adsorption on pollutants removal cannot be neglected.
2.2. Phase Inversion
Phase inversion is a conventional membrane fabrication method by precipitating dissolved polymer into porous membrane structure, including non-solvent-induced phase separation (NIPS), controlled solvent evaporation and thermally induced phase separation. Membrane structure and performance are influenced by the types of solvent, non-solvent, polymer and nanofillers, synthesis temperature and phase inversion conditions [72]. Among them, NIPS is the most widely applied method to fabricate water treatment membrane since 1960’s owing to the rich porosity and controllable fabrication conditions [73]. Benefiting the advantages of facile and scalable fabrication, they are widely applied to fabricate low-pressure membrane (microfiltration and ultrafiltration) [72]. However, compared to IP strategies, conventional phase inversion is difficult to fabricate common NF membrane with low MWCO since it has stochastic nature and achieved polydisperse pore size distributions. Hence, it is difficult to realize the defect-free and ultrathin functional layer owing to the limitation of the fabrication process [74]. By comparison, phase inversion, especially for NIPS by incorporating nanomaterials was feasible for fabrication of LNF membrane. Under this background, modulating chemicals properties, optimizing the processes and introducing nanofillers into membrane matrix have been attempted to obtain mixed-matrix membranes (MMMs) with high porosity and loose selective layer, which combined the merits of good processability of polymers and transport properties of porous nanofillers [75].
2.2.1. Polymers, porogens, solvent and phase inversion conditions
Polymer, additives and phase inversion conditions significantly affected the NF membrane structure, surface wettability and performance. Many polymers are developed until now, including poly(ether sulfone) (PES), poly(vinylidene fluoride) (PVDF), polyvinyl chloride (PVC), polyacrylonitrile (PAN). PES-grafted PAN LNF membrane was synthesized by NIPS method (Fig. 5a) and the pure water permeance was up to 77.9 L m−2 h−1 bar−1 [76]. PVDF hollow fiber LNF membrane was also developed to possess high chemical and solvent resistance [77, 78]. Hyperbranched polyester amide (PEA) and PES can also construct LNF membrane and the porosity of active layer can be modulated by changing the ratio of PEA to PES [79]. In addition to synthetic polymer, biomass based NF membrane attracted intensive attraction in membrane application, such as cellulose (Fig. 5b) [80], chitosan [81]. Especially, cellulose membrane Besides, ceramic membrane is promising to face the challenges of harsh conditions in many conditions by phase inversion, such as γ-Al2O3 hollow fiber membrane [82], while the high cost limited its application in water treatment.
Porogens are widely applied to create pore in membrane during NIPS, such as polyvinylpyrrolidones (PVP) and polyethylene glycols (PEG) [83]. Besides, green nanobubbles can act as pore-forming agents for fabrication of loose PES NF membrane by NIPS, which has pore size of 3.81 nm and high flux (200.2 L m−2 h−1) [84]. Solvents (e.g., N,N-Dimethylformamide) are used to dissolve polymer and disperse the nanofillers for formation of casting solution, which are important to construct LNF membrane by NIPS method. Recently, low-cost and green solvents were developed to enable LNF membrane fabrication [85]. Moreover, modulating the baths coagulation to HCl can improve the membrane flux without compromising selectivity (Fig. 5c) [84]. Adding dicarboxylic acids and aliphatic alcohols additives to solvent can modulate pore size and morphology of PES NF loose membrane by NIPS [86]. Hydrophilic deep eutectic solvent additives can promote the finger-like pores generation throughout membrane and produce hydrophilic channels [87]. Wang et al. [78] found that changing the co-solvent ratio of N,N-dimethylacetamide and tetrahydrofuran can reduce the selective layer thickness and membrane filtration resistance. Operating conditions also influenced the phase inversion process for LNF membrane synthesis, such as phase inversion time, and coagulation bath temperature, solvent and pH [88]. Increasing the coagulation temperature facilitates the formation of more porous morphology, making higher permeability [89]. Furthermore, increasing ethanol concentration in coagulation bath enabled the higher hydrophobicity of NF membrane [90].
2.2.2. Incorporating porous nanofillers
Nanoparticles can be loaded on membrane surface through in-situ growth or incorporated into membrane matrix by phase inversion. For the former, nanoparticles can form a selective layer with well-aligned pore for separation and a substrate layer acts as supporting function. The pore and physicochemical properties, and arrangement of nanoparticles will influence the membrane properties. For the latter, nanoparticles can provide more water channel and change the physicochemical properties of membranes. Incorporating porous nanofillers is effective to reduce the defects in MMMs that synthesized by NIPS and impart membrane with different sieving, charge and hydrophobicity properties, thereby becoming a promising method for fabrication of NF membrane from flat membrane to hollow fiber membrane [91]. Moreover, nanofillers may act as porogens owing to their porosity, which avoided the addition of chemical porogens. Many nanofillers had been used for fabrication of LNF membrane, such as silica [92], zwitterionic functionalized MoS2 (Fig. 6a) [93], GO [94], carbon nanotubes [95], COFs [96], MOFs [16, 97] and graphitic carbon nitride [98].
Similar to nanomaterials interlayer in PA TFC membrane, the comparability between nanofillers and membrane matrix is important to maintain the stability and performance of LNF membrane, which may cause the possible leaching of nanofillers [91]; moreover, the agglomeration of nanofillers also blocked membrane pore and produced voids and defects within membrane matrix, which significantly deteriorated the membrane performance [99]. Above issues became more serious in the presence of high loading of nanofillers. Therefore, plenty of attempts were conducted in many strategies, especially for MOFs based nanofillers. Deng et al. [100] developed a water-based mixing strategy to construct MMMs with high loadings of ZIF-8 (39 wt%) to avoid the aggregation of ZIF-8 nanoparticles within PVA matrix (Fig. 6b). Chen et al. proposed a solid-solvent processing strategy to fabricate ultrathin MMM membrane (<100 nm) with high MOFs loading of 80% while avoiding the nanofiller agglomeration, the penetration of casting solution into substrate pores and interfacial defects formation [101]. In this strategy, polymer matrix acted as solid solvent to immobilize metal, thereby realizing high comparability between metal salt and polymer matrix. Luo et al. [102] designed a MOFs membrane with wrinkled configuration to resolve the bottleneck issues of low mechanical properties of MOFs membrane owing to the rigid MOFs and soft substrate. Furthermore, Ji et al. [103] overcame the challenges of large-area fabrication of MOFs membrane using nanoreactor-confined crystallization method. However, until now, the large scale fabrications of NF based on MMMs and applications in water treatment are still limiting and they are promising to realize industrial application when some fundamental and engineering issues were resolved.
2.3. Other Fabrications Strategies
Pressure-assisted filtration (PAF) is widely used to fabricate advanced membrane in lab sale by filtrating of nanomaterials through a substrate under vacuum. This sample method is more feasible for 2D membrane fabrication, such as GO [104], CNT [105], MXene (Fig. 7a) [106], layered double hydroxide (LDH) [107] and 2D-COFs (Fig. 7b) [108]. For example, Zhang et al. [104] synthesized GO/MoS2 membrane with loose structure by pressure-assisted assembly and the performance can be tuned by changing the ratio of GO and MoS2. Poly(ethyleneimine) (PEI) and sodium alginate (SA) are also assembled on membrane to form loose layer by PAF [109]. Layer spacing of GO membrane can be finely tuned during PAF by modulating the interlayer valence salts [110]. However, the restacking, agglomeration and low stability of nanometerials largely limited this method for industrial fabrication since there were insufficient interaction between nanomaterial layers and substrate. To resolve these issues, PAF can combine with IP method to fabricate nanomaterials confined PA membrane, which can be anchored within PA layer; moreover, nanomaterials favored adsorption of amine monomer and facilitated IP reaction for PA membrane formation, such as iron oxohydroxides nanorods [111] and MoS2-incorporated nano-striped [112].
Layer-by-layer (LBL) method enables precise, simple and mild control of the thickness, morphology, and functionality of membrane by alternate deposition of oppositely charged polyelectrolytes on a substrate [20], which is considered as a green and sustainable method because of the use of water based solvent [113]. Membrane structure can be modulated by polyelectrolytes properties and concentration, temperature, ionic strength and pH [114]. Many polyelectrolytes based LNF membranes have been constructed by LbL assembly, such as charged 2D crown-ether framework [20], anionic poly(styrene sodium) sulfonic acid (PStSO3Na)/cationic quaternized polyvinyl imidazole (PVIm-Me) [114], anionic polyacrylic acid (PAA)/cationic polyethyleneimine (PEI) [115], cationic poly(diallyldimethylammonium chloride) (PDADMAC)/anionic poly (sodium 4-styrenesulfonate) (PSS) (Fig. 7c) [116], biodegradable lignin polyelectrolyte [18, 113]. Niu et al. found the valences and sizes of cations in LBL-based polyelectrolytes membrane influenced the electronegativity and packing density of polyelectrolytes layer, which helped to optimize membrane performance [117]. Sun et al. [18] selected alkaline lignin (AL) and sodium ligninsulfonate (SL) to obtain LNF by LBL assembly. The hydrogen bonds and π-π stacking between AL and SL ensured the high stability of LNF and dense layer for pollutants rejection. However, conventional LBL based membranes are restricted by the complicated preparation procedures (requiring multiple steps) and unsatisfying separation performance and defects were difficult to be avoided [117]. Moreover, the electrolytes layers are unstable and experience fouling in complex water conditions. For example, salt ions can penetrate and screen the charges of polyelectrolytes [118]. An extra step like cross-linking is usually required, which limited their application in large scale NF membrane based water treatment [119].
In addition to above preparation methods, there were many attempts for fabrication of LNF (e.g., deposition and spinning methods). Obviously, these methods have some merits and limitations. We should avoid considering one single parameter and systematically optimized different parameters (e.g., nanomaterials structure and loading amount, thickness of active layer, interaction between membrane and nanomaterials) during membrane fabrication [120]. Moreover, water/salt selectivity dose not meant water/organic pollutants selectivity [120]. A combination of different methods is an alternative option to optimize LNF performance, for instance phase inversion and IP process [121]. Moreover, nanomaterials attracted increasing attention for enhancing membrane stability, separation and antifouling performance, while their cost, possible leaching and stability during long-term operation in real water matrices (e.g., pH, saline concentration, pollutants types and concentrations) cannot be overlooked [122]. This is a dominant and key task before industrial applications. Unfortunately, until now, the large scale fabrications of NF based on these new fabrications are still missing. Their feasibility in practical applications in water treatment still remained unclear for long-term operations in complex water matrices. Besides, characterizations played an important role to understand the crystal structure, functional groups, pore information, thermal stability and physicochemical properties of membranes, such as X-ray diffractometer (XRD), BET surface area, Fourier transform infrared spectroscopy (FT-IR) and thermogravimetry analysis (TGA) [123]. Emerging characterizations also emerged (e.g., time of flight secondary ion mass spectrometry, low-field nuclear magnetic resonance) to in-depth understand the water transport and rejection mechanism, which is promising to guide the design of high-performance LNF membranes for advanced water treatment and resource recovery. We believed that more LNF membrane will be designed and applied in water treatment in pilot plant and large-scale applications.
3. Application For Wastewater Treatment And Resource Recovery
3.1. Textile Wastewater
Conventional tight NF membranes have a higher rejection for inorganic salts, which are difficult to separate dyes and inorganic salts in high-salinity textile wastewater. In comparison, LNF can efficiently remove dye while providing free passages for NaCl by concisely modulating the pore size and charge properties, therefore realizing resource recovery of dye and salts. Many LNF membranes were designed for textile wastewater treatment, such as NH2-MIL-101(Fe) MOFs [124], GO [125], quercetin [126], tannic acid-polyethyleneimine-modified hollow fiber [127], polydopamine/β-cyclodextrin [128], dopamine and diamine-zwitterion (Z-DNMA) co-deposited [129] based LNF.
Table S2 illustrated the key LNF membranes properties for textile wastewater and resource recovery. Gao et al. [128] synthesized polydopamine/β-cyclodextrin based LNF membrane with efficient and stable dye/salt separation in the presence of different salt concentrations (Fig. 8a), attributing to the unique transport channels in β-cyclodextrin and loose selective layer (Fig. 8b). Ji et al. [130] constructed PSF/GO LNF hollow fiber by phase inversion, which realized high Congo red rejection (99.9%) and low NaCl rejection (<5%) during long-term filtration (Fig. 8c) via size exclusion and Donnan effect (Fig. 8d). In addition, Xiong et al. [131] reported that PEI and PEI-PO3Na based LNF by LBL method presented high rejection of Victoria blue BO (99.5%), BO/NaCl selectivity, and permeability (24.2 L m−2 h−1 bar−1). The hydrophilic and chelation ability of polyelectrolytes also imparted membrane with high antifouling and stability. Moreover, trehalose/TMC based LNF membrane was designed for zero-discharge textile wastewater treatment, which presented high NaCl/Direct red 23 separation factor (275) with a 295% reduction in energy consumption [132]. GO composite LNF was also fabricated to separate the binary dyes mixtures with good antifouling property [133]. Cerium oxide/PES LNF membrane presented high dye recovery and COD removal (>79%) from real textile wastewater, while negligibly rejected salts (Na2SO4 (3.3%), NaCl (1.2%)) [134]. A PEI assembled LNF was reported with high rejection to Congo red (99.37%) and flux (36.89 L m−2 h−1 bar−1) while allowing NaCl pass (rejection=2.47%) [135]. Zhao et al. used renewable quercetin as monomer to fabricate loose polyester layer by reacting with TMC by IP process, which presented high water permeability (198 ± 2.1 L m−2 h−1 bar−1) and Congo red/NaCl selectivity (123), owing to the moderate pore size, loose, hydrophilic and ultrathin structure [126]. These results demonstrated that incorporating nanomaterials or functionalization of PA membrane can adjust membrane structure to enhance the dye selectivity and permeability via different synthesis methods. Besides, the combination of LNF and other processes enabled the synchronous water reclamation and resource recovery. Lin et al. [136] developed LNF membrane (MWCO=800 Da) filtration-electrodialysis integrated processes to fractionate dye and salt in textile wastewater by enrichment and purification, realizing excellent recovery of dye (98.4% purity) and NaCl (99.96%), and pure water reclamation. In this integrated process, LNF membrane can act as anion conductive membrane for recovery of dye [137].
Nevertheless, majority researches selected dye/salt simulated wastewater, the treatment of real textile wastewater was rare by new fabricated LNF membranes. Although many studies proposed their new LNF membranes had excellent dye and salt separation and stability, their practicability in real textile wastewater treatment required further investigations for long-term filtration. In addition to dye and salt, there were also other chemicals in textile wastewater, which may affect the water treatment efficiency and cause membrane fouling [138]. Unlikely, the attempts are usually devoted to designing advanced membranes, while their performance in combined treatment process is usually overlooked. Moreover, the underlying separation mechanism should be unveiled at microscopic level and in-situ analysis in order to guide the development of LNF membrane for textile dyeing wastewater treatment.
3.2. Pharmaceutical Wastewater
Pharmaceutical wastewater is another important industrial wastewater and pharmaceuticals at mg/L level can be recycled by LNF membrane, while common tight NF membrane is insufficient to selectively separate and recover pharmaceuticals and salts. For example, antibiotics (e.g., tetracycline hydrochloride (TC) and bacitracin (BAC)) and inorganic salts are normally coexisted. Separation antibiotics from salts, also called antibiotics desalination is important to achieve high-purity antibiotics [139]. The molecular weight range of pharmaceuticals is similar with that of dye, hence many LNF membranes have been constructed to remove pharmaceuticals while providing the channel for water and salt transport. Table S3 illustrated the key LNF membranes properties for pharmaceutical wastewater and resource recovery. Matin et al. [140] compared the pharmaceuticals removal performance of different commercial NF membranes. They found the pharmaceuticals removal was impacted by the feed characteristics (pH and salts) and membrane characteristics. For example, pH influenced the charge of some pharmaceuticals and LNF membrane, thereby affecting Donnan effect and membrane performance [141]. Interestingly, although commercial NF270 and DL membranes have lower pore size than many antibiotics, the rejection was not efficient (e.g., 25%–60% by NF270) [142], especially for the neutral ones since they can penetrate through membrane by solution-diffusion mechanism [143]. Anike et al. [144] explained the mechanism of antibiotics transport through membrane by Spiegler–Kedem–Katchalsky model via irreversible thermodynamics. Maryam et al. [145] found commercial HYDRACoRe NF50 LNF membrane with lower WMCO (1000 Da) rejected ibuprofen (MW=206 Da), diclofenac (MW=318 Da) and paracetamol (MW= 151 Da) by 80.54%, 99.74% and 36.16%, respectively. Above results indicated that commercial LNF membranes are not effective to retain pharmaceuticals with low MW and strong adhesion with membrane and their separation performance should be improved to meet the requirement of the resource utilization of pharmaceutical wastewater.
Recently, many novel LNF membranes were developed to reject target pollutants by modulating membrane morphology or introducing nanomaterials for providing diverse rejection functions. PES/SPSf LNF membranes were fabricated to remove ionic pharmaceuticals mainly via electrostatic repulsion, but removed non-ionizable ones by adsorption [146]. 1,4,7,10-tetraazacyclododecane-TMC LNF membrane with IP process could reject tetracycline by 92.7%, while allow NaCl (rejection=4.5%) and Na2SO4 (18.9%) pass through membrane, attributing to the loose, smooth and highly negative charged layer [147]. Given that membrane favors to reject pollutants with the same charge property, dually-charged LNF membrane is expected to reject both positively and negatively charged pollutants. Liu et al. [148] found that dually-charged LNF membranes had better rejection performance toward charged pollutants compared with single charged ones. Surface modification was also employed to improve membrane performance, especially for the amine based materials. Quaternary triethanolamine modification provided a loosened active layer on PA LNF membrane and decreased negative charge density, which weakened Donnan effect and then improved the antibiotics desalination performance [149]. In a similar manner, quaternary diethanolamine [150] and spirocyclic quaternary ammonium diamine [151] modified LNF membranes were also fabricated to enable outstanding antibiotics desalination. However, this will inevitably sacrifice the rejection toward pollutants. To address these drawbacks, N-Oxide connected zwitterionic N,N-bis(3-aminopropyl)methylamine can impart the membrane with high salt/antibiotics separation factor (41.1) via its strong hydration effect [152]. Above results indicate that modulating the pore structure and physicochemical properties of membranes is beneficial for the improvement of pharmaceuticals separation.
Additionally, introduction of nanomaterials can provide additional channels for solvent transport and pharmaceuticals rejection, such as β-cyclodextrin [153], polydopamine [154], MOFs [155, 157], γ-AlOOH@Naringin [158] and ignocellulosic nanofibrils [159]. Fang et al. [160] developed a nature-inspired MOFs incorporated TFC membrane to break the permeability-selectivity trade-off effect, which presented stable and efficient antibiotics rejection over 90% when treating pharmaceutical wastewater secondary effluent. Notably, the mainstream NF membranes possess negative charges, which is not effective to retain positively charged pollutants. Incorporating β-cyclodextrin into PA-TMC layer can reduce the “outer space” distance [153] and energy barrier for pollutants passage through cyclodextrin [161], thereby realizing high-performance antibiotics desalination (tetracycline rejection= 94.8%, NaCl rejection=8.5%) and ultrahigh permeability (82.9 L m−2 h−1 bar−1). TA/transition metal interlayers in PA TFC membrane achieved high water permeability and antibiotics desalination performance by modulating the moderate pore size, roughness, and negative charge density [162]. Yin et al. [163] incorporated metal-organic frameworks/carbon nanotubes into MMMs, causing higher permeability by 3–100 folds than most state-of-the-art NF membranes and high rejection (71.8%–99.3%) toward typical antibiotics because of the water pump like function and shelving effect of nanomaterials. Moreover, they proposed that the membrane did not always require ultrathin active layer, instead hundreds micron scale selective layer with highly porous structure can also enable ultrafast removal of pollutants, which provided a diagram shift for design of LNF membrane. The mechanism can be attributed to the loose selective layer of membrane matrix, and pore sieving and charge properties of UiO-66/CNTs.
Membrane performance in pharmaceutical wastewater treatment is influenced by membrane structure and property, water matrices properties (e.g., pH, pharmaceuticals concentration, MW, size and geometry, pKa and LogKow), operating conditions (e.g., flow rate and pressure), and physicochemical characteristics [164]. For example, pH changed the charged and non-charged interactions between membrane and pharmaceuticals via variation of pollutants polarization and membrane charges, thereby influencing electrostatic interaction for pharmaceuticals rejection [145]. The hydrophobicity of pharmaceuticals also strongly affected the LNF membrane performance and the hydrophilic compounds rejection was better when the size exclusion governed the separation mechanism [165]. LNF can also combine with other techniques to enhance the pharmaceuticals retention. A recent study developed a biofilm reactor-nanofiltration-membrane bioreactor to treat roxithromycin-containing wastewater, which can efficiently remove COD (98.4%) and roxithromycin (74.1%) via adsorption and biodegradation effect [166]. Notably, although adsorption by LNF membranes can contribute to pharmaceuticals removal, the low specific surface area of membranes and the adsorption saturation limited the application of adsorption membranes in water treatment [167, 168]. Moreover, there were many organic solvents in pharmaceutical wastewater, which caused the swelling of PA TFC NF membrane, and deteriorated its stability and separation performance. Thus organic solvent resistant LNF membranes are imperative to enable stable and efficient removal of pharmaceuticals in wastewater [169]. Moreover, the hydrophobic, complexation and electrostatic interactions between membrane and pharmaceuticals also contributed to the separation mechanism. However, the real pharmaceutical water treatment and resource recovery by LNF in pilot plant or large scale experiments are quite rare, which should be overcame in the future research from the aspects of practical application.
3.3. Municipal Wastewater
The total volume of municipal wastewater was up to 680–960 million m3/d [170]. Municipal wastewater reclamation attracted increasingly interests as an alternative water resource, which had diverse purposes in many applications (e.g., irrigation, industrial and domestic purpose) [171]. However, many emerging organic contaminants (EOCs) were also detected in municipal wastewater, such as perfluoroalkyl sulfonic acid compounds, pharmaceutical and personal care products, persistent organic pollutants and microplastics. They have high toxicity and pose great potential to the environment despite of low concentration (from ng/L to μg/L level), which was facilitated by the co-existence effects [172]. Nevertheless, conventional municipal wastewater treatment plants (WWTPs) failed to remove EOCs, causing the leaching into the receiving environment [173]. Conventional NF membrane was designed for desalination, which was not effective to reject toxic and harmful emerging contaminants, especially for the small, non-charged and high hydrophobicity ones [3, 174]. For example, rejection rate of N-nitrosodimethylamine (NDMA) with small MW and high polar was lower than 50% and the presence of defects in ultrathin NF membrane further deteriorated the separation performance [174]. Notably, although pharmaceuticals (e.g., antibiotics) were also detected in municipal wastewater, their concentrations were much lower (from ng/L to μg/L) than that in pharmaceutical wastewater, which bring more obstacles to the treatment efficiency; moreover, it is not economic and efficient to recover the pharmaceuticals from municipal wastewater compared to pharmaceutical wastewater. Despite that, it is also quite meaningful to remove pharmaceuticals and other trace emerging organic contaminants from municipal wastewater to protect the safeguard of reclaimed water. In this regard, intensive efforts were devoted to resolving this issue by designing novel LNF membranes with tailored structure or combining LNF with other processes. Table S4 illustrated the key LNF membranes properties for municipal wastewater and resource recovery.
The separation performance depends on the pollutants-membrane interactions, which is further influenced by the physicochemical properties of pollutants (e.g., size, charge and hydrophobic properties) and membrane (e.g., MWCO, flow channel, charge and hydrophobicity), and water matrices conditions (e.g., pH, saline concentration, coexisting inorganics and organic matter) and operating conditions (e.g., pressure, flow rate and mode) [175]. For example, natural organic matter (NOM) affected the EOCs rejection and membrane fouling performance. Commercial TS80 membranes rejected ulfamethoxazole and diclofenac by size exclusion at pH<4 and electrostatic repulsion also involved when pH was above 7 [176]. Similar to textile and pharmaceutical wastewater, constructing novel LNF membranes by modifying the membrane structure or incorporating nanomaterials can also enable superior municipal wastewater treatment efficiency. For example, Wang et al. [177] constructed a polyelectrolyte multilayer membrane, which enabled high retention toward emerging organic contaminants (90%), which was promising to municipal wastewater reuse. Shukla et al. [178] incorporated Zn-MOFs into PA TFC membrane for high rejection toward antibiotics in municipal wastewater (>93%). Meanwhile, MOFs played a dual function in permeability enhancement by facilitating the water flow via hydrogen bonding and enhancing surface hydrophilicity. Fang et al. [179] found molybdenum disulfide/MOFs increased rejection rates of antibiotics by 27–55% via size exclusion, Donnan effect, and hydrophilic-hydrophobic interaction. Similarly, Hu et al. [16, 157] also confirmed MOFs (Cu-HHTP, CoFe-BDC) could impart MMMs with ultrahigh permeability, exceptional antibiotics rejection and self-cleaning performance. Recently, A NF pilot system using Desal 5DK membrane was developed to remove trace anticancer drugs (e.g., capecitabine, etoposide and tamoxifen) with ng/L concentration level in actual domestic wastewater effluent and the rejection rate was higher than 96%, which avoided the release of EOCs into receiving water body [180]. A pilot scale NF system enabled high and stable rejections (>80%) of 49 micropollutants from real municipal wastewater [181]. Meanwhile, rejection increased at lower temperature since the membrane pore lowered. Besides, NF can combine with other techniques to enhance contaminants removal and mitigating membrane fouling, such as MBR and catalytic membrane [3, 182–184]. MBRMF + NF processes can remove most pollutants and markedly reduce membrane fouling owing to the low applied pressure [185].
In addition, it is meaningful to recovery of nutrients (e.g., N and P) in municipal wastewater treatment. Nieminen et al. [186] found total P retention was above 95% in municipal secondary effluent by LNF membrane, while the residual nitrogen was poorly rejected since it typically presented as dissolved nitrate; meanwhile, high rejections of citalopram, diclofenac, venlafaxine and candesartan were obtained at 95%, 82%, 84% and 97%, respectively. Notably, membrane fouling is a major bottleneck of NF membrane application in municipal wastewater remediation. Owing to the abundant foulants in wastewater, the membrane fouling is more severe in wastewater treatment compared to seawater desalination or surface water treatment. Abada et al. [187] proposed that anthropogenic silicon compounds were the major foulants during municipal wastewater reclamation in pilot plant tests. However, the studies about resources recovery from municipal wastewater by LNF were still few, which were worth to investigate in future researches in large-scale application.
Notably, there were also other important industrial wastewater applications for LNF membranes, such as leather wastewater, electroplating wastewater, while the practical industrial investigations were still insufficient. Therefore, the development of LNF membrane will resolve the inherent drawbacks of conventional tight NF membrane, which has the advantages of higher permeability, superior resources selectivity and separation efficiency, better anti-fouling ability and lower energy consumption, which had wide range of applications in future. Noteworthily, its priority over conventional tight NF still required the verification in real wastewater treatment from the aspects of separation efficiency, stability and purposes. Additionally, some challenges should be addressed before large scale and industrial uses, which would be elucidated hereinafter.
4. Challenges and Prospects
4.1. Severe Membrane Fouling During Wastewater Treatment
Membrane fouling is the “Achilles’ Heel” of membrane technology by the accumulation and deposition of foulants on membrane surface or inside pore [188]. There are different kinds of foulants (e.g., inorganic, organic, biological and combined foulants) in wastewater, causing inorganic fouling, organic fouling, biofouling and combined fouling by accumulating corresponding foulants on membrane [189, 190]. For example, membrane fouling by inorganic foulants (e.g., SiO2) called inorganic fouling. This will markedly reduce the water permeability and membrane lifetime, which enhanced the operating cost and deteriorated membrane performance [191, 192]. The multivalent inorganic ions (Ca2+ and Mg2+) can form complexation with organic substances in wastewater, which caused more severe inorganic-organic combined fouling than the individual one [193].
Generally, when materials composition and structure were similar, LNF membrane presented superior anti-fouling properties compared to common TNF membrane since low salts rejection and high permeability decreased concentration polarization and operating pressure. Moreover, the fouling mitigation strategies of LNF membrane are similar to these of TNF membrane. To mitigate membrane fouling, pretreatments (e.g., adsorption, coagulation, oxidation and ion exchange) before NF membrane unit was a common and industrial method by directly removing foulants or converting them into substances with low fouling potential [188]. Nevertheless, the additional chemical consumption and long processes of pre-treatment-membrane methods limited its application in some circumstances. Chemical cleaning using acid, base or oxidants is also a widely applied method to restore membrane performance. However, this method is also not sustainable due to the discharge of cleaning effluent and the cost of chemical cleaning reagents. To address these problems, constructing catalytic membrane with self-cleaning ability is an alternative option by degrading membrane foulants using green and renewable energies (solar and electricity) [192]. In addition, design of antifouling LNF membrane with low fouling propensity, high hydrophilicity and negative charge density is also an important trend [188]. Anyway, the membrane fouling depends on the membrane and wastewater property, and operating conditions. Notably, choosing the fouling mitigation strategies should consider the cost and environmental factors. Currently, conventional pretreatment methods consumed chemical reagents and generated secondary pollutants (e.g., cleaning wastewater and sludge), which required high chemical consumption and caused secondary pollution. It is imperative to optimize fouling control processes in order to avoid the excessive addition of cleaning reagents and reduce the cleaning frequency, while ensured the high and stable membrane performance. Moreover, developing emerging and environmental friendly methods are more promising to reduce the cost and pollution, such as clean and renewable energy (light and electricity) driven strategies. Consequently, further studies should be devoted to designing anti-fouling LNF membrane and developing more sustainable, low-cost and high-performance fouling control strategies.
4.2. Low-cost And Scalable Fabrication And Commercial Feasibility Of LNF Membrane
Although numerous LNF membranes were fabricated in past decades and the researchers proposed their membranes had excellent separation performance and antifouling ability, the low-cost, scalable and commercial fabrication are still challenging. Some LNF membranes have been commercialized, while the poor separation and anti-fouling performance (low permeability and weak anti-fouling ability) still limited their applications compared to conventional NF membranes [194]. For example, water permeabilities of commercial LNF membranes CK (GE Osmonics, composed by cellulose acetate, MWCO=~2000 Da), GK (GE Osmonics, composed by polyamide, MWCO=2000 Da) and NP010 (Microdyn Nadir, composed by polyether sulfone, MWCO=~1000 Da) were only 3.45, 10.0 and 5.0 L m−2 h−1 bar−1, respectively. Moreover, most commercial LNF membranes presented relatively lower dye removal than common NF membranes. Thus, it is imperative to improve the comprehensive performance of LNF membranes to meet the increasing demands for advanced water treatment and resource recovery.
Recently, ultrathin selective layer membrane (even sub 10 nm) has also been reported in many works [56, 195], while they are difficult to enable scalable manufacturing currently for forming ultrathin selective layer while avoiding defect formation. Moreover, many materials cannot manufacture at scale or roll into modules [11], centimeters or even millimeters scale membrane were usually found in many studies. Large-scale fabrication and long-term filtration of these new membranes in industrial applications are still limited [91]. Thus their reproducibility and operating cost still restricted the commercial application during long-term operation over a broad range of temperatures and pH; moreover, they also required high tolerance and robustness under harsh conditions during acid and alkaline cleaning. Incorporating nanomaterials into membrane attracted intensive attention for improvement of the membrane performance, while their possible leaching into permeate is a non-neglected issue and nanomaterials should be firmly anchored on LNF membranes by proper strategies (e.g., cross-linking, blending). Besides, the membrane module design was also very important to ensure the high-performance water treatment. In addition to roll-to-roll module, other types of membrane modules should also be considered as good option in many circumstances, such as hollow fiber membranes [196]. However, the full-scale applications of membrane modules using new LNF membrane are still insufficient [197].
The poor chlorine resistance of PA TFC membrane restricted its application in wastewater treatment and the strict limitation of free chlorine brought additional operating costs [198]. Designing more resistant LNF membrane became an important issue. Recently, changing the PIP monomer or surface modification of TFC membrane by IP processes attracted intensive researches. However, these membranes are still limited in lab-scale fabrication and application, the stability, antifouling and lifetime during long-term filtration of actual wastewater should be evaluated. Moreover, these membranes may also have inherent drawbacks. For example, polyester membrane is prone to hydrolysis, which is facilitated at elevated pH values. Yao et al. [36] selected a new monomer (3,5-dihydroxy-4-methylbenzoic acid) to react with TMC, which formed a polyester membrane that stabilized at pH<9. Polymeric membrane with high chemical stability, such as PVDF and PES, can also enable high chlorine resistance [199]. However, these polymers are usually selected to fabricate low-pressure membranes and the commercial membranes are still rare in this field. As discussed before, NIPS based polymer membrane may play more important role in LNF membrane fabrication owing to their facile and scalable features, and porous matrix. Besides, it is challenging to optimize different membrane performances (ant-fouling, anti-chlorine, permeability, selectivity and stability) since there are many trade-off effects (e.g., permeability-selectivity). Thus membrane design should consider from the aspect of system-level contextualization in full cycle. Special attentions should also focus on the regeneration of used membrane after scrapping. In this regard, biomass based membrane (e.g., cellulose) is more promising to meet this requirement, while their stability is also challenging in long-term water treatment.
4.3. Membrane Separation Mechanism
The unclear membrane separation mechanism is another challenge for LNF membrane based water treatment. Generally, membrane separation mechanism included size exclusion, Donnan effect or hydrophobic interaction. Among them, membrane can reject pollutants by size exclusion if its pore size lower than pollutants size. Therefore, size exclusion is a dominant mechanism in many circumstances. However, when membrane pore size was higher than pollutants size, pollutants may also be rejected by other mechanisms. For instance, most LNF membranes presented negative charge, which repelled negatively charged pollutants by Donnan effect. IN addition, membranes with hydrophilic surface were also effective to retain hydrophobic pollutants by hydrophobic interaction. Nevertheless, although one can attribute rejection to size exclusion, Donnan effect or hydrophobic interaction, the underlying mechanisms in nanoconfined channel are still not clear [49]. An in-depth understanding of pollutants and water transport through membranes at nanometer or angstrom scale is still challenging. Such a knowledge gap largely impedes the optimization of membrane design for fit-for-purpose application. Moreover, the difference separation mechanism between LNF and common TNF membranes should also unveiled since the transport and rejection of solvents and solutes may differ in different sizes channels.
A quantitative structure–activity relationship (QSAR) model was useful for predicting the separation performance of LNF membranes [200], which helps to develop advanced membranes for rejection of target contaminants with high antifouling and antioxidation ability. The extended Nernst-Planck equation Donnan-steric pore model with dielectric exclusion also predicts salt rejection performance [201]. Although theoretical models can gain some understanding on water transport behavior, the complexity of unknown parameters caused they were not accurate and universal when facing complex water matrices (e.g., pH, saline concentration, pollutants types and concentrations) [202]. The precise prediction of separation performance of LNF membrane is still a challenge. Machine learning (ML) algorithms can unveil the complex relationships among diverse variables, thus can predict membrane performance [202] and guide the design of NF membrane for advanced water treatment [203]. However, data from literature are usually related to membrane performance in different conditions and some parameters and experimental details are missing [204]. Moreover, ML learning required large number of experimental data and it still has some limits until now. These problems will be overcome with the fast development of artificial intelligence.
Except for above challenges, NF concentrates presented high concentrations of inorganic salts, emerging organic contaminants and general organic contaminants, which were high-salinity, toxicity and non-biodegradability. The chemical treatment method usually enhanced the complexity of pollutants, which posed nonegligible environmental challenges [205]. The preparation of calibration curves and stock solution is also important to ensure the accuracy of the experiments [206]. In addition, the conclusions from lab-scale tests may not be applicable to the practical industrial performance because of the different membrane scales and operational settings [197]. Combining the experiments and modelling is expected to guide the engineering scale up. Liu et al. [207] treated surface water by LNF in pilot plant, which enabled high rejection of TOC (>90%) and pesticides (54%–82%), but has insufficient total dissolved solids removal (approximately 27%). More similar work should be reported to discover and solve the problems of LNF membranes in wastewater treatment. Moreover, resource recovery from wastewater still remained huge challenges from the aspect of engineering and economy.
4.4. Prospects
To address above challenges, we proposed the future trends of LNF technology in wastewater treatment and resource recovery. Firstly, facile, low-cost, scalable and sustainable fabrication methods are promising to develop the advanced LNF membranes with exceptional separation, anti-fouling and stability. For example, replacing conventional monomer and solvents into biodegradable and environmental friendly ones are important to meet the increasingly stringent environmental requirements. Incorporating nanomaterials is a good option to break the trade-off effect between permeability and rejection of target pollutants and enhance the anti-fouling ability of membranes. The structural designability, easy functionalization and diverse physicochemical properties of nanomaterials provide powerful tool to design custom-tailored LNF membranes in specific water treatment applications. Further researches should enhance the stability of membranes by firmly and uniformly anchoring nanomaterials in membrane during long-term filtration of practical wastewater. More importantly, the large-scale application of LNF should also conduct to demonstrate its superiority over common NF membrane. The gap of fabrication and application between lab scale and large scale largely limited the application of LNF technology in wastewater treatment and resource recovery. Additionally, the performance of commercial LNF membranes should also improve. We believe more LNF membranes with outstanding performance will appear in near future, which will be a competitive and good supplement of NF technology.
Secondly, the variation in separation and fouling mechanism between LNF and common TNF should also be unlocked to guide the advanced separation of wastewater by LNF membranes. In-depth separation and fouling mechanism of LNF membranes should be revealed in virtue of the microscopic and operando aspects. Emerging analyses and characterization methods can be employed to unveil the transport behavior of solvents and solutes through membrane pore and identify the dominant transport nanochannel. Dynamic fouling processes can be monitored to understand the fouling mechanism and guide the membrane fouling Moreover, the separation and fouling behavior can be predicted through machine learning, which helps to understand the underlying mechanism and guidance of advanced LNF membrane fabrications. Considering the complexity of real water matrices (varied and fluctuated water quality, and complex pollutants), separation mechanism was explored in simulated wastewater containing specific foulants in current stage. The gap between simulated wastewater and real wastewater may lead to the misunderstanding of separation mechanism. Future researches should attempt to unveil the separation mechanism in complex water matrices and mitigate the interference of backgrounds.
Thirdly, compared to direct wastewater treatment, recovery of resources from wastewater is more appealing and promising to relieve the problems of resource scarcity. Compared to common TNF, LNF technology is more feasible to recover resources (e.g., dye, pharmaceuticals and salts). However, the recovery ratio of target resources should improve to ensure the product quality and economic value. Specially, recovering resources from practical wastewater in large scale are still challenging, which should be overcome in future. Moreover, the commercial feasibility of LNF membranes in wastewater resources is another important index, which should comprehensively consider all aspects (e.g., membrane cost and lifetime, processing cost, recycled products value and energy consumption). Besides, the gap between fundamental research and practical application restricted the development of LNF in resource recovery. Future researches should balance above factors and enable the industrialization.
5. Conclusion
Above all, LNF membranes have significant advantages over common NF membranes that encourage further investigations about the membrane design and application in wastewater treatment and resource recovery. Fortunately, more novel LNF membranes were developed to construct loose active layer for fast water transport without sacrificing selectivity by modifying PA TFC membrane or developing alterative materials by interfacial polymerization, phase inversion and other fabrication strategies. Although LNF membrane had achieved significant improvement, challenges of few large scale application, severe membrane fouling, economic feasibility and insufficient in-depth mechanism still limited its application. LNF membranes should enable low-cost, scalable and sustainable fabrication and their performance should also improve to meet the practical requirement. Moreover, considering the complexity of wastewater, the LNF membranes processes should be optimized from the aspects of membrane feature, wastewater composition and properties, and operating conditions. Deeply understanding the separation and anti-fouling mechanism will help to guide the membrane design and applications. We believe LNF membrane technology will play a more important role in wastewater resource recovery and near-zero discharge to meet the requirement of low-carbon water treatment concept.
PDF Links
PubReader
Full text via DOI
Download Citation 
Supplement
Print 
