1. Introduction

Solution separation technologies are in high demand as contaminants continue to pollute environmental resources in a consumer-driven era. Air and water purification methods are widely studied and are proceeding towards tackling micro/nano pollutants, which are both health-hazardous and difficult to remove. Filtration membranes, such as thin film composite (TFC) and thin-film nanocomposite (TFN) membranes, have been synthesized and applied for some time as they are known to be more economically efficient and simple compared to other purification methods, like multi-step chemical treatments. Their main drawbacks, however, include chemical and pressure instability, low solute selectivity, slow solvent permeation, and membrane fouling. To overcome these challenges, metal-organic framework (MOF) nanoparticles (NPs) have been implemented into TFN membranes and have gained much attention in the past decade[1-14].

Since gases do not pose as many drawbacks as liquids, many studies have tested MOF-based membranes for gas separation. Nonetheless, in recent years, significant advances have been made by MOF-doped TFN membranes in liquid separations, including organic solvent nanofiltration (OSN) and water desalination. Through membrane-based liquid separation processes, such as nanofiltration (NF), forward osmosis (FO), and reverse osmosis (RO), solute-solvent separation can be achieved with few equipment and materials while also being energy-efficient and waste-free. Moreover, MOF modifications have shown to compensate for many inherent flaws in TFC and TFN membrane performance usually by increasing solvent flux, enhancing elemental stability, strengthening pressure integrity, and maintaining reliable defect/fouling tolerance[15-22].

This article reviewed recent studies that focus on four prominent MOFs-MIL-101(Cr), UiO-66, ZIF-8, and HKUST-1 [Cu₃(BCT₂)], respectively-and their first applications in NF membranes. For each MOF, we discuss, the synthesis of the MOF, the synthesis of the membrane, characterizations, and the membrane performance. The membrane fabrication and separation process are explained in Fig. 1 and articles are summarized in Table 1.

2. Metal Organic Framework

2.1. MIL-101(Cr) (MIL, materials of institute lavoisier)

MIL-101(Cr) was synthesized by mixing chromium (III) chloride hexahydrate (CrCl₃⋅6H₂O) and terephthalic acid at 180°C for 30 min. Polyacrylonitrile (PAN) was used as the support layer[23]. The TFC membrane was prepared by in situ interfacial polymerization (IP) of piperazine (PIP) aqueous phase and trimesoyl chloride (TMC) organic phase [combined with either n-hexane (HX) or nitrobenzene (NB)] on the PAN membrane. The TFN membrane was prepared by mixing MIL-101 (Cr) NPs into the organic phase followed by similar polymerization technique used for TFC membrane fabrication. Scanning electron microscopy (SEM) characterization evaluated the TFN membrane morphology, confirming the formation of NP aggregates in the membranes synthesized with HX. Atomic force microscopy (AFM) characterization evaluated the rms roughness of the membrane surface, confirming the synthesis of two distinctive textures (i.e., smooth and rough). Although TFN membranes are generally known to demonstrate better OSN performance than TFC membranes, fabricating TFN membranes using HX organic phase showed inconsistent permeance and lower rejection than the TFC membrane control. For MIL-101- based TFN membranes, solution-phase measurements confirmed considerable MOF particle aggregation in HX, hindering the overall performance. Aggregates lead to non-selective defects, but further experimentation suggests that aggregates also lower selectivity in TFN membranes. Indeed, a well-dispersed loading of NPs in the membrane is required for high performance. The novelty of using NB as an organic phase during IP re sulted in an aggregation-free poly(piperazine trimesamide) membrane with a MIL-101 suspension that increased by at least two orders of magnitude (~0.14 w/v%). Under NB fabrication, two types of TFN membranes were created-by changing IP conditions-and compared: smooth and rough. For smooth membranes, the effect of MIL-101 concentration was negligible; on the other hand, compared to the TFC control, rough membranes showed at least 1.9 and up to 3.6 times greater permeation of organic solvents and similar levels of rejection (~90%). In the case of rough membranes, NBbased TFN membranes has the potential to show even higher performance with increased concentration of MOF fillers, the geometry allowing up to five times more filler.

NH₂-MIL-101(Al) was synthesized by combining AlCl₃⋅6H₂O and terephthalic acid at 130°C for 72 h. NH₂-MIL-101(Cr) was synthesized by combining Cr(NO₃)₃⋅9H₂O and 2-aminoterephthalic acid (ATA) at 130°C for 24 h. Polysulfone (PSF) was used for the support layer[24]. The MOF/chitosan NF membrane was fabricated by coating the chitosan layer-containing either MOFs-onto the PSF membrane. X-ray diffraction analysis (XRD) characterized the crystalline structures of the MOFs, showing almost identical images with past reports and confirming their successful synthesis. Zeta potential qualitatively measured the surface charge of the NF membranes, which confirmed the increased positivity of MOF-doped membranes. Each of two MOFs, NH₂-MIL-101(Al) and NH₂-MIL-101(Cr), are used in the fabrication of NF membranes. Structurally, Al-MOF possessed rod-like structures in comparison to the grainy structure of the Cr-MOF. As a result, the MOFs attributed to different morphologies, significantly affecting the permeability of the membrane. The innovation, however, lies in the facile preparation method, in which each MOF is homogenously incorporated in the chitosan polymeric matrix to finalize into a positively charged NF membrane. Several characterization tests confirmed the even distribution of MOF nanoparticles (loading of 15 wt%) and reported that the NH₂-MIL-101(Al) membrane had double the water flux of the NH₂-MIL-101(Cr) membrane due to structural differences. Further tests were performed for NH₂-MIL- 101(Al), such as measuring the effects of MOF loadings for the NF of multivalent cations. With emphasis on the divalent Ca²⁺ and Mg²⁺ cations, test results indicated increased rejections up to 93.0% (MgCl₂, CaCl₂, NaCl, and Na₂SO₄, in decreasing rejections), most likely a result of both size exclusion and electrostatic interaction. To conclude, apart from other beneficial properties of the MOFs, the highly positively charged character of the novel NF membranes contributed immensely to the repulsion of multivalent several cations, highlighting an electrostatic factor that is to be considered in future membrane technologies.

At 453 K (180°C) for 30 min, CrCl₃⋅6H₂O was mixed with terephthalic acid to produce MIL-101(Cr). Cross-linked asymmetric polyimide (P84^®) was used as the support layer[25] (Figs. 2~5). The LS film was prepared by suspending MIL-101(Cr) NPs in a solvent mixture of chloroform, methanol, and behenic acid, which was directly pressed onto the polyimide (P84^®) substrate (LS-P84). The TFC membrane or LS-TFN membrane was completed when a polyamide (PA) top layer was created by IP of m-phenylenediamine (MPD) and TMC on plain polyimide or LS-P84, respectively. Energy dispersive X-ray spectroscopy (EDX) confirmed the presence of MIL-101(Cr) MOFs in the synthesized membrane. AFM characterization measured the rms roughness, which verified the smooth and even MOF layer in the LS-TFN membranes compared to the conventional ones. The flaw in TFN membrane preparation is the tendency for the MOF nanoparticles to agglomerate unevenly inside the thin polyamide due to the heterogeneous nature of the layers, increasing the formation of unselective defects. An innovative solution is to implement a Langmuir-Schaefer (LS) film, which is a controlled MOF monolayer. The experiment created an LS film with the MOF MIL-101(Cr) and successfully synthesized the PA top layer, middle LS-MIL-101(Cr) film, and bottom support layer [i.e., cross-linked asymmetric polyimide (P84^®)]. Not only is the LS-TFN membrane characterized by homogeneous and defect-free layers, but it also requires a record minimum amount of MOF NPs (3.8 μg cm^-2) with no filler loss during OSN application. The seamless mechanism was proven via multiple characterization procedures; furthermore, the membrane showed optimal performance with minimal decrease in selectivity. Two methanol solutions of Sunset Yellow (SY) dye and Rose Bengal (RB) dye were filtered through the novel LS-MIL-101(Cr)-TFN membrane, and results showed one of the best solvent permeances (10.1 ± 0.5 L m^-2 h^-1 bar^-1 for SY, 9.5 ± 2.1 L m^-2 h^-1 bar^-1 for RB) and rejection rates higher than 90%. This is an outstanding improvement from conventional TFC and TFN membranes, their permeance averaging between 7.0 and 8.0 L m^-2 h^-1 bar^-1. Overall, the LS-TFN membrane promotes both OSN efficiency and low-cost preparation of the membrane due to the decrease in MOF usage.

NH₂_MIL-88B was formed by mixing a solution of FeCl₃⋅6H₂O and ATA. PSF was used as the support layer, upon which the NH₂_MIL-88B in situ growth procedure fabricated a modified support layer (NH₂MIL- 88B@PSF)[26]. The TFC membrane was synthesized by coating PA on the PSF support via interfacial polymerization of MPD and TMC; the PA@NH₂_MIL-88B membrane followed the same procedure, but with the NH2MIL-88B@PSF support. Fourier transform infrared (FTIR) analysis visualized the presence of amino groups in the MOF. Contact angle measurements characterized the membrane hydrophilicity, verifying the hydrophilic nature of the MOF which contributes to the membrane. In order to improve the NF and RO performance for highly selective TFN membranes, incorporating MOFs in the PA layer is widely studied. In this study, the novelty of the fabricated TFN membrane is its use of the amino-functionalized MOF, NH₂_MIL-88B. After growing a layer of NH₂_MIL-88B NPs on the surface of the PSF support membrane (NH₂MIL-88B@PSF), a PA top layer coated the nanoparticle interlayer to finalize the novel PA@NH₂_MIL-88B TFN membrane. Several characterizations confirmed the successful fabrication of this membrane as well as its new structure, which was less cross-linked, smoother, thinner, and more hydrophilic than a pure PA membrane. Moreover, the NH₂_MIL-88B NPs provided water-preferential pathways, and the synthesized PA@ NH₂_MIL-88B membrane prevented defects. As a result, membrane permeance improved. This is supported by the NF of phenol from water using PA@NH₂_MIL-88B membrane: TFN2 membrane showed notable water flux that increased from 23.5 to 35.8 kg m^-2 h^-1 and at least 90% rejection, a significant improvement from a TFN membrane with a pure PA layer. Since the PA@NH₂_MIL-88B membrane is durable and exhibits excellent removal of small organic solutes from water, further study on surface modifications will be beneficial for high performance, MOF-based nanocomposite membranes.

MIL-101(Cr) was prepared as previously reported. Polyether sulfone (PES) was used as the ultrafiltration (UF) support layer[27]. The novel TFN membrane was synthesizing by pouring a mixture of MIL-101(Cr) NPs and TMC organic phase on the PES UF support before the IP process. Other control membranes were prepared: TFC membranes without MOF modifications, and TFN membranes with MIL-101(Cr) in MPD aqueous phase. SEM characterization was used to evaluate the membrane cross-section, showing loose adhesion between the PA and PES for membranes synthesized in aqueous phase; in contrast, membranes in organic phase showed tight adhesion. Contact angle measurements indicated the hydrophilic properties of the membranes and MOF. Modern water-treatment technologies pursue RO and NF methods, in which both require the use of membranes under pressure. In TFN membranes, inorganic porous materials, such as zeolite and silica, are often the fillers in the organic PA layer, which inevitably creates instability due to the clashing properties between the filler and PA layer. Such instability is a growing issue in pressurized processes like RO and NF. This research studies the effects of the MOF, MIL-101(Cr) (~200 nm), and is the first to apply it in manufacturing TFN membranes for water treatment. Because the MIL-101(Cr) NPs establish microporous water channels and multiple separation properties in the TFN membrane, test results showed an increased water permeance of 3.91 L m^-2 h^-1 bar^-1 in the 0.2 w/v% MIL-101(Cr) membrane compared to undoped membranes (i.e., 158% increase). Also, the membrane rejection is controlled and adjustable in relation to the amount of MOF nanoparticles. All MIL-101(Cr)-TFN membranes showed high water fluxes and rejections in the NF of several dye solutions, such as methyl orange, methyl violet, congo red, and methyl blue. In addition to its high performance, this novel membrane exhibited strong affinity between the organic NPs and organic PA layer, increasing overall membrane stability. This study unfolds the potential use of MIL-101(Cr), and possibly other MOFs, for advanced TFN membranes for various water purification fields.

MIL-101(Cr) was synthesized by blending Cr(NO₃)₃ ⋅9H₂O and terephthalic acid, which was stored at 218 °C for 16 h. PSF UF membrane was used as the support substrate[28] (Figs. 6, 7). The in situ preparation of TFC and TFN-MIL-101(Cr) membranes required directly pouring TMC-doped with MIL-101(Cr) for TFN membranes-onto the MPD-soaked PSF UF support and synthesizing by interfacial polymerization. Other TFN membranes added the MOFs into the aqueous MPD before IP for comparison. Attenuated total reflection FTIR (ATR-FTIR) verified the suitability of the MIL-101(Cr) NPs and the organic phase as they incorporated well into a dense PA layer. X-ray photoelectron spectroscopy (XPS) characterization showed the elemental composition of the membrane surface, suggesting that, since increasing MOF loadings decreased crosslinking extent, there is a maximum to the loadings before performance declines. After the successful doping of the dense selective PA layer with MIL-101(Cr) NPs at 0.05 w/v%, synthesized with a PSF ultrafiltration support, the fabricated TFN membrane was tested for water desalination. The novel TFN RO membrane was further characterized by several tests, such as SEM, AFM, XPS, wettability measurement, and RO test. The strong affinity between the organic particles and layers assured tight integration in the membrane, ensuring membrane stability for easier application. Unlike other water stable MOFs, like ZIF-8 and UiO-66, the structure of the MIL-101(Cr) MOF possesses larger pore size and surface area, creating broader direct water channels for easier water permeation; the hydrophilicity of MIL-101(Cr) also encourages water attraction to through the TFN membrane. Moreover, the doped MIL-101(Cr) NPs provided other membrane modifications (e.g., morphologies, roughness, crosslinking extent, and wettability), significantly enhancing the performance. In the desalination of an NaCl solution, the novel TFN-MIL-101(Cr) membrane increased permeance to 2.2 L m^-2 h^-1 bar^-1 (up to 44%)- which improves with more MIL-101(Cr)-while salt rejections remained higher than 99%. Altogether, the high permeation levels, outstanding salt rejection, and verified stability of the TFN-MIL-101(Cr) membrane is a significant advancement in NF and RO processes using membrane technology, especially in the water purification field.

Both MIL-101(Cr) and ZIF-11 MOFs were prepared as previously reported. The support layer used was a cross-linked asymmetric polyimide (P84^®)[29]. TFC and TFN membranes were synthesize by creating a PA top layer on the support membrane by IP of MPD aqueous phase and TMC organic phase. The novel TFN membrane preparation simultaneously added the two MOFs in the TMC before IP, but the TFN control only added one kind of MOF to the organic phase. XRD analyzed the crystallinity of the MOFs, which notably showed that nano ZIF-11 was not crystalline like typical micro ZIF-11 NPs. Energy dispersive spectroscopy (EDS) analysis detected Cr and Zn atoms simultaneously in the novel TFN membrane, indicating the successful incorporation of both MOFs. MIL-101(Cr) and ZIF-11 NPs differ in chemical and textural properties, each offering different results when synthesized with a TFN membrane and undergone the OSN process. Individually, MIL-101(Cr) is hydrophilic and shows high rejection whereas ZIF-11 is hydrophobic and shows faster permeation. The combination of these two MOFs, however, provide the resultant membrane additive properties that improve the overall membrane performance. The novelty of this experiment is, therefore, the development of a unique combined-MOF membrane, but also the focus on temperature as an important factor in the membrane performance. During the NF of different dye-methanol solutions using single-MOF membranes, the ZIF-11 membrane showed better permeance of 4.9 and 3.2 L m^-2 h^-1 bar^-1 for SY and acridine orange (AO) dyes, respectively, whereas MIL-101(Cr) showed better rejections of more than 90%. The novel TFN membrane combining the two MOFs resulted in an intermediate performance, suggesting the simultaneous improvement in ZIF-11 rejection and in MIL-101(Cr) permeance. DMF post-treatment applications further improved the performance of the combined-MOF membrane. In addition, increasing temperature reduced the TFN membrane’s adsorption for enhanced permeance; activation energies of pure methanol and SY-methanol for OSN measured 13.2 ± 2.1 and 8.3 ± 1.1 kJ mol^-1, respectively. It is notable that the effect of the temperature on the performance was irreversible because the permeance when returning to the initial temperature was shown to be lower than the permeance before the temperature was raised; in contrast, solute rejection was unaffected in this manner.

MIL-101(Cr) and Mil-68(Al) were synthesized in tetrahydrofuran solvent, and, along with ZIF-11, were prepared based on previous reports. The cross-linked asymmetric polyimide (P84^®) support layer was fabricated and treated with several chemicals[30]. The PA top layer was formed on the polyimide substrate by interfacial polymerization of MPD and TMC, which creates the TFC membrane. The same IP process was used in the TFN membrane preparation, except the TMC organic phase was pre-doped with MOF NPs. Thermogravimetric analysis (TGA) on MOFs studied their thermal stability and confirmed their well-activated status. XRD characterization was useful to confirm that IP did not affect the crystallinity of the MOFs. TFN membranes were prepared by synthesizing membranes with ultrathin PA layers that were topped with MOFs of either MIL-68(Al), MIL-101(Cr), or ZIF-11 (70, 103, and 79 nm respectively). Each TFN membrane was tested in OSN and observed in close attention to several important variables: non-solvent bath, chemical post-treatment, concentration of precursors for interfacial polymerization, and polymerization time. Combinations of various solutes and solvents were also taken into consideration, such as the innovative use of ketones (acetone) with MOF-based TFN membranes. In general, ZIF-11 and DMF were the only factors that showed significant increase in the permeance of solutions, exhibiting a maximum permeance of 6.2 L m^-2 h^-1 bar^-1 and more than 90% rejection when both were applied. The exception was the AO-methanol solution, in which permeance decreased with the presence of ZIF-11. The negative effect was most likely due to unwanted interactions between the solute and membrane along with the organic linker in ZIF-11, but fouling phenomena may also cause the decrease in observed permeance. Indeed, the interactions between the solvent-membrane and solute-membrane as well as the hydrophilicity of the membrane itself are the main parameters determining the performance level of the resultant membranes. This is verified when comparing the permeances of four solvents: both TFN and TFC membranes concluded that THF had the lowest permeance, then water, methanol, and acetone (highest).

2.2. UiO-66

UiO-66 was synthesized by combining zirconium(IV) chloride (ZrCl₄), 1,4-benzene-dicarboxylate (BDC), and acetic acid at varied temperatures and time for different sizes. PES substrate was used as the support layer [31]. Both TFC and TFN membranes required the interfacial polymerization of piperazine aqueous phase and TMC organic phase to form the PA layer on the PES support, but UiO-66 NPs were first added to the organic phase for the TFN membranes. Contact angle goniometer measured the hydrophilicity of the membranes, which showed that hydrophilicity increased with the addition of MOFs, but began decreasing due to higher mass transport resistance when the selective layer was too thick. A SurPASS electrokinetic analyzer measured the membrane surface charge, which indicated enhanced electrostatic repulsion of anions at higher pH. In this study, the challenge of removing harmful selenium (Se) and arsenic (As) from aqueous environments is addressed with OSN, using TFN membranes containing the MOF, UiO-66. This experiment compared the pure water permeability and pollutant rejections of TFN membranes that were synthesized with each of 30, 100, and 500 nm sizes of UiO-66 NPs. The smaller pores and hydrophilic character of TFN membranes contributes better performance compared to TFC membranes. Accordingly, the TFN membrane consisting of the 30 nm UiO-66 NP showed the best performance. Further experimentation for the influence of particle loading was made for the MOF of this size: the prepared TFN membrane with a UiO-66 layer of 0.15 wt% showed the highest PWP (11.5 L M H bar^-1), and rejections of varying pollutants (of Se and As) were all above 95%. Generally, the key factors of selectively designing these TFN membranes were found to be particle size and loading. Results suggest that UiO-66-TFN membranes improved overall performance due to higher hydrophilicity, unique aperture, and large porosity. In addition, TFN membranes synthesized with UiO-66 NPs exhibited robust long-term stability, highlighting its effectiveness in larger and harsher environments. As this is the first reported study on UiO-66 incorporation in TFN membranes for Se and As nanofiltration, this simple synthesis has much potential in future membrane technologies.

UiO-66 was synthesized by mixing ZrCl₄ and BDC. Alumina hollow fiber substrates were used as the support membranes[32] (Figs. 8, 9). To fabricate the final membrane, in situ solvothermal synthesis was applied at 120°C for 3 days to grow a layer of UiO-66 NPs on the substrates. SEM characterization evidence no alterations in morphology nor ion exchange reactions of the UiO-66 NPs after they were tested in water solutions; this proves the MOF’s water stability. EDX analysis clearly showed the sharp transition between the MOF layer and substrate, confirming smooth and tight adhesion. Focusing on zirconium(IV)-based MOFs, the pure-phase Zr-MOF, UiO-66, was the basis of the membranes that were fabricated and tested in this study. Due to their strong internal bonds, the Zr-MOfs possess exceptional chemical and thermal stabilities. UiO-66 in particular is a Zr(IV)-carboxylate MOF, which has fcu topology, ~6.0 Å aperture size, and a hydrophilic surface. Its properties, as a result, is highly selective of small water molecules over larger hydrated ions (e.g., 6.6 < Å), and water stable. The application of such UiO-66 NPs in the formation of continuous Zr-MOF polycrystalline membranes-via in situ solvothermal synthesis-for single-gas separation is unique to this study. Results of the novel membrane showed excellent multivalent ion rejection, different metal ions ranging from about 86% to over 99%. Moderate permeance of 0.14 L m^-2 h^-1 bar^-1 and good permeability of 0.28 L m^-2 h^-1 bar^-1 μm was also observed. Furthermore, the MOF’s stability prevented any degradation of the membrane performance, exhibiting integrity and functionality in various tests up to 170 hours and through a large range of salt solutions. In conclusion, the UiO-66 is one of many similar Zr-MOF materials with varying pore sizes and functional groups, and this study opens the possibilities of Zr-MOF application in pressure-driven desalination and other industrial separation processes.

The UiO-66 synthesis used solvothermal reaction of ZrCl₄, BDC, and acetic acid at 120°C for 40 h. Udel polysulfone (PSU) was used as the support substrate [33] (Figs. 10~12). The synthesis of TFC control membranes required the interfacial polymerization of pure TMC and MPD-coated substrate to create a PA top layer. MOF-modified TMC was used in the same IP procedure to synthesize TFN membranes. FE-SEM characterization of UiO-66 NPs showed the MOF crystal to have the largest reported Zr-cluster at the center, which explains the MOF’s high chemical stability (high degree of connectivity). Contact angle goniometer char- acterized the surface hydrophilicities of the membranes, showing an increase in hydrophilicity with the increase in NP loading. In contrast with RO, FO processes move solvents by chemical potential difference, so they operate under lower pressures. The advantages of the FO conditions are considered in this study. The incorporation of the MOF, UiO-66, in TFN technology and its performance in salt solution (NaCl in DI water) NF was observed. PA-based TFN membranes are gaining more attention, and the UiO-66 MOF is well known for its naturally hydrophilic structure. Not only does the UiO-66 NP add hydrophilicity to the overall TFN membrane, but its molecular sieving significantly alters the morphology and chemistry of the PA selective layer. The synthesis of the PA-based TFN-UiO-66 membrane is a new combination for FO OSN. The TFN-U2 membrane, with 0.1 wt% particle loading, showed the best results of 52% increase in permeability and ~95% rejection. In 1 M NaCl, compared to the TFC control membrane, the TFN-U2 membrane showed 40% and 25% increase in water flux under the PRO and FO tests, respectively. Additionally, as salt concentrations increased up to 2.0 M, TFN-U2 showed steady increase in FO water flux and limited solute reverse diffusion, which suggests a slight reduction in the internal concentration polarization effect. The results of this novel TFN membrane will be useful in creating other fabrications of high-performance membranes, especially for Zr-MOFs.

UiO-66-NH₂ was synthesized via the solvothermal process, which mixed ZrCl₄, ATA, and acetic acid at 100°C for 24 h. The support layer used was PSF membrane, which some were sprayed with a homogenous layer of UiO-66-NH₂ NPs[34]. The TFC and TFN membrane was synthesized by interfacial polymerization of MPD and TMC on the plain PSF and sprayed PSF, respectively. AFM characterization found that the membrane surface roughness increased with a small loading of MOFs, decreased when additional loadings created a monolayer, and then increased again with even more additions. XPS verified no NP protrusion or leakage across layers except within its designated PA coating layer. Although RO processes are the leading technology for desalination, their trade-offs, (e.g., stunting permeability for greater selectivity) is a current challenge. To tackle this problem, TFN membranes with MOFs are widely studied. The MOF, UiO-66-NH₂, has an average size of ~100nm; and its influence in PA-based TFN membranes is the main focus of this particular study. Naturally hydrophilic and porous, the UiO-66-NH₂ NP has triangular apertures, its structure inherently stable in water and chemically affable when bonding with the PA membrane. As a result, the addition of UiO-66- NH₂ NPs to the membrane allows greater surface hy- drophilicity, less cross-linking, and preferential pathways across the selective layers for water permeation. The novelty of this research lies in the method in which these MOF particles are incorporated into the TFN membrane: spray-assisted pre-disposition before the IP step. Test results indicated that the 3-minute spray disposition of the nanoparticles (0.02 wt%) on the PSF support layer correlated with about 50% increase in water permeance, showing superior RO separation performance. The rejection levels remained similar to TFC membranes in brackish water desalination tests. In fact, TFN-UiO-66-NH₂ membranes had higher chlorine (Cl) resistance than general TFC membranes. To sum, the UiO-66-NH₂ NP is beneficial to next-generation TFN membranes studies; furthermore, the spray technique is highly economically efficient as the process requires minimal use of nanoparticles while displaying even deposition in quick procession.

2.3. ZIF-8

ZIF-8 seeds, or ZHNs, were synthesized by mixing ethanol, water, and Zn(NO₃)₂⋅6H₂O for 30 min. polyvinylidenefluoride (PVDF) hollow fiber was used for the support layer[35]. To synthesize the ZIF-8/gelatin/ PVDF membrane, the PVDF substrate was coated in the prepared ZHN solution and was immersed in several solutions for growing the final ZIF-8 layer (on the inside and outside surface of the PVDF hollow fiber). XRD characterization was performed, which confirmed the formation of the ZIF-8 layer after the secondary growth. SEM morphology was used the test different ZIF-8 seeds; ones grown without gelatin were found to be discontinuous and uneven in the final membrane. Many biological and physicochemical processes have unsuccessfully removed dye from wastewater, so separation by membrane technology, which is seen to be clean, convenient, and energy-efficient, is crucial for dye removal. This work features the MOF, ZIF-8, and a macro-porous PVDF support layer in the fabrication of a TFN membrane. Using gelatin-assisted growth process in room temperature, a ~2 μm thick membrane of ZIF-8 NPs are strongly adhered onto the surface of the PVDF hollow fiber, completing the TFN membrane. Note that the gelatin simultaneously acts as a buffer layer, seed-growth layer, and adhesion layer. Proceeding the second stage of growth, the ZIF-8/gelatin/PVDF membrane exhibited excellent NF for the small dye molecule-water solution. In the separation of Rhodamine B molecules from water, results showed a permeance of 137 L m^-2 h^-1 bar^-1 and dye rejection up to 97.5%. Because of the challenge due to water instability and lack of technology to produce continuous and strongly bonded membranes, much of MOF research focuses less on liquid separation compared to gas separation. The ZIF-8/gelatin/PVDF membrane is a unique composite membrane that attempts to overcome these challenges and perhaps present potential related research topics (e.g., precoating ZIF-8 seeds on the inner and outer surface of the PVDF for quicker preparation). Overall, the ZIF-8/gelatin/PVDF membrane presented successful liquid separation and reveals the promising future of MOF-based membrane technology for dye-wasted water purification.

ZIF-8 MOFs were synthesized by mixing a solution of Zn(NO₃)₂⋅6H₂O, methanol, and 2-methylimidazol at room temperature for 1 h. Polytetrafluoroethylene (PTFE) was the support membrane used[36]. The fabrication of the PTFE-ZIF-8 membrane involved solvent evaporation, where an aqueous ZIF-8 solution was coated onto the PTFE support layer. FTIR characterization showed the existence of the imidazole ring, verifying the presence of ZIF-8 NPs in the membrane. EDX evaluated zinc concentration, also verifying an even distribution of MOF NPs in the membrane. In the formation of a microfiltration (MF) membrane, the zeolite imidazolate MOF, or ZIF-8 was utilized for potential micropollutant removal. More specifically, the membrane is comprised of a double-layer matrix of polytetrafluoroethylene (PTFE) that is modified with the ZIF-8 NPs. The synergy provided by the incorporation of ZIF-8 consequently increases the membrane adsorption capacity, water permeability, and fouling tolerance. Aside from MF performance, the new combination of polymer and MOF is already advantageous because both promote durability, stability, economic efficiency, and simplicity in preparation. Using highly-concentrated progesterone, the MF results for the PTFE-ZIF-8 membrane were positive as well: rejection remained about 95% even after three regeneration cycles, and high water flux was observed. The excellent rejection was attributed to stronger hydrogen bonding between the ZIF-8 and hormone molecules and increased surface area. Likewise, the water permeation was improved because of the formation of better water channels and enhanced water-membrane hydrogen bonding with the presence of the hydrophilic ZIF-8 NPs. In an attempt to bring attention to MF membranes, which easily overcome challenges posed in the NF field (e.g., chemical stability and cost), the novel PTFE-ZIF-8 membrane presents an approach that is highly advantageous to efficient micropollutant removal from real water sources.

The ZIF-8 and ZIF-67 NPs were synthesized the same way: combining methanol, 2-methylimidazole, and metal particles [Zn(NO₃)₂⋅6H₂O for ZIF-8, Co(NO₃)₂⋅ 6H₂O for ZIF-67] for 10 min to produce an MOF dipping solution. Polyimide (P84^®) asymmetric membranes were used as the substrates[37]. The synthesis of the MOF-TFN membranes involved dip-coating the substrate into either of the prepared MOF solutions for 10 min before drying. EDX characterization confirmed homogenous distributions of Zn and Co in their respective prepared TFN membranes. SEM imaging was used to measure coating thickness and show dense membrane surfaces without defects or agglomerations. The conventional preparation of TFN membranes attempts a heterogeneous deposit of MOF NPs inside the top PA layer, allowing agglomerations and uneven NF throughout the membrane surface. Dip-coating polyimide P84^® asymmetric supports with MOFs-in this case, ZIF-8 (70 ± 10 nm) and ZIF-67 (240 ± 40 nm) NPs-is a new preparation method that results in a controlled monolayer of particles and a homogeneous coating to the support layer. The advantages of dip-coating are fewer agglomerations and required NPs, which improve OSN and reduce material costs, respectively, although the process was better with the smaller ZIF-8 particles. Moreover, this technique seems highly suitable for the ZIF group as these MOFs maintained their crystalline character after the IP process while no excess MOFs were lost during IP. In terms of OSN performance, the dip-coated ZIF-8-TFN membrane showed high methanol permeance of 8.7 L m^-2 h^-1 bar^-1 (150% increase from TFC membranes) and maintained high SY dye rejection of 90%. Further modifications have been tested, such as switching to ZIF-67 or double dip-coating, but neither changes to the membrane enhanced OSN performance. The apparent limit of the dip-coating method is most likely due to the presence of defects and a compact PA layer. Nonetheless, the dip-coating technique is proven to be simpler, cheaper, environmentally friendlier, and more effective than the average TFN membrane.

The synthesis of 75 nm ZIF-8 NPs was done by mixing Zn(NO₃)₂⋅6H₂O, 2-methylimidazole, and methanol at room temperature for 24 h; the 150 nm ZIF-8 followed the same procedure with DMF instead of methanol and stirred for 4 h. PSF was used as the support membrane[38]. Conventional TFN (CONV-TFN) membranes were fabricated by the interfacial polymerization of TMC organic phase and MPD aqueous phase. Similar IP procedures were used to fabricate the EFP membrane with the exception dispersing the MOFs on the PSF substrate via evaporation before proceeding with IP. SEM and TEM characterizations suggested that filler particles influence the top layer morphology of membranes (i.e., thicker MOF layer showed lower permeance). XPS was the most appropriate method to assess the Zn concentration in the top layer membrane. Conventional TFN preparation synthesizes MOF filler particles during the IP process, preventing most particles from effectively incorporating into the thin PA top layer and therefore greatly limiting the membrane performance. A new and advanced methodology for MOF-based TFN preparation is Evaporation-controlled Filler Positioning (EFP), where MOF NPs are prepositioned exactly at the water-solvent interface and before the IP process. This experiment compared the performance of EFP-prepared TFN membranes and the conventionally prepared ones; utilizing MOF ZIF-8 as the filler, all membranes were sufficiently characterized (via SEM, TEM, ATR-FTIR, and XPS), and the effect of filler size was also considered. When putting NaCl solution through OSN, the EFP-TFN membrane showed no salt rejection loss as well as 220% permeance increase compared to the membranes without the filler, being the highest reported permeance for MOF-based TFN membranes. For these improvements, it is crucial to preposition the filler at the organic-water interface and before the IP process. Otherwise, compared to other reported ZIF-8-TFN membranes, the ZIF-8 membranes prepared by EFP requires 80 times less filler (0.005 w/v)-regardless of the particle size-significantly reducing the cost of materials. In addition, EFP membranes showed better reproducible performance. Overall, EFP is a significant step in preparing TFN membranes that are high-performance, cost-efficient, and industrially applicable.

2.4. HKUST

HKUST-1 [Cu₃(BCT₂)] MOFs were synthesized by mixing HKUST-1 with sulfuric acid and DMF for 24 h, creating a MOF-PSF membrane. A separate porous PSF was used as the support layer[39]. The TFC RO membrane was synthesized by combining the prepared MOF-PSF membrane and PSF support via interfacial polymerization of TMC and MPD. EDX dot-mapping verified even dispersion and good stability of MOFs in the polymer. TGA characterization showed water-weight loss of synthesized MOFs, in which acid-MOFs lost more water than normal MOF particles and are, therefore, more hydrophilic. Polyamide-based TFC RO membranes have high salt rejection levels and are cheap, making them common in water treatment technologies; however, their inefficient permeation is an issue that seeks a solution. The effects of the support layer on the performance of the TFC RO membrane is studied. Hydrophilicity and porosity were the main parameters when measuring the effectiveness of the support layer. This experiment synthesized a novel TFC RO membrane that consisted of a PSF membrane overlaying a support layer containing an acidic MOF, HKUST-1 [Cu₃(BCT₂)]. In the creation of such TFC RO membrane, m-phenylene diamine (MPD) in aqueous solution and trimesoyl chloride (TMC) in organic solution were put through the IP process, forming a thin and smooth film on the support layer, verified by XPS. Also, compared to the commercial RO membrane thickness of 0.02 μm, the FTIR analysis confirmed the MOF-modified membrane to be 0.029 μm thick. The performance of this MOF/PSF membrane is then compared to a pure PSF membrane: the presence of acidic HKUST-1 [Cu₃(BCT₂)] enhanced both the hydrophilicity and porosity of the support layer, consequently increasing water flux of the membrane by 33% (21 to 28 GDF) without rejection loss. The addition of hydrophilic functional groups and the strong copper-cluster bonding was most likely the reason for these improvements. Furthermore, additional BSA and MFI testing showed fouling resistance on the TFC RO membrane with MOF modifications.

3. Conclusions

Membrane filtration is one the most promising and needed processes in chemical purification industries. With the help of MOF nanoparticles, TFN membranes have exhibited one the best permeances and rejections for nanofiltration. Whereas earlier studies first utilized certain MOFs in TFN membranes, later studies approached MOF modifications differently, such as simultaneously dispersing different MOFs into the same membrane or applying unique methodologies that synthesize membranes either with or without interfacial polymerization. Of course, a wide range of characterization procedures-from imaging with SEM to analyzing chemical compositions with XPS-aided in identifying detailed information on MOF and membrane properties, confirming successful fabrications. Other tests including chemical post-treatments have also shown to enhance membrane hydrophilicity and stability. In any case, with little to no reduction in salt rejection, solvent flux was shown to significantly improve with these MOF modifications due to better chemical affinity, structural integrity, particle dispersion, and defect prevention. Additionally, most, if not all, studies in this review apply methods that are cost-effective, economically friendly, and simple, showing promising reproducibility for mass applications in large purification applications.