Metal–Organic Frameworks (MOFs)-boosted filtration membrane technology for water sustainability Open Access

Clean and sustainable freshwater is a basic requisite for daily life and industrial activities, while its reserve on earth is rather limited. In the past decades, the rapid growth in the global population and the ever-increasing industrialization have not only increased the consumption of clean water, thereby directly producing a huge volume of wastewater, but also involuntarily caused severe pollution to the water resource. The degree of overall water pollution largely exceeded the self-purification ability of the natural water ecosystem. As a result, the water shortage has become a worldwide threat. As has been predicted, more than one billion people in arid regions will suffer from water scarcity in 2025, and thereby, ensuring available access to clean water has been listed among the main goals of the Sustainable Development Agenda of the United Nations.¹ 

Polluted water refreshing is considered as the most sustainable way to address the above-mentioned issues. Different technologies such as adsorption, chemical precipitation, coagulation sedimentation, membrane separation, biological process, and advanced oxidation processes have been developed for water and wastewater treatment.² Among them, membrane separation is regarded as the most cost-effective and technically feasible technique to replace conventional water treatment systems due to the relatively simple operation and no/little chemical addition.³ Currently, polymeric membranes have been widely adopted in the industry because of their easy fabrication and low cost. However, in some harsh environments, such as high temperatures and corrosive conditions, their applications are still limited. In this regard, inorganic ceramic membranes have gained much attention and been increasingly used in some industrial fields with special requirements on the mechanical, chemical, and temperature stability.⁴ Although the fabrication cost of ceramic membranes is generally higher than the polymeric counterparts due to the relatively expensive raw materials and high-temperature process required, the excellent stability and long lifetime are expected to potentially reduce the overall cost.⁵ 

Selectivity and permeability are the two main indicators to evaluate the separation performance of filtration membranes. The separation process based on the porous membrane is achieved fundamentally based on the size exclusion mechanism. Thus, there exists a trade-off between selectivity and permeability,⁶ in which the level of high porosity and uniform pores in the filtration membrane will certainly result in a high permeability, while their selectivity will accordingly be reduced. To achieve both good permeability and selectivity simultaneously, other separation mechanisms, such as adsorption, catalytic degradation, charge repel, and hydrophilic interaction, have been further introduced into the membrane separation process. In addition, another common issue of both polymeric and ceramic membranes for water purification is membrane fouling,⁷ which refers to the deposition of foulants on the external surface and/or in the internal pores. Membrane fouling will not only deteriorate the filtration efficiency, thereby requiring higher transmembrane pressure (TMP) to maintain the constant flux or reducing the flux dramatically at the same TMP, but also shorten the lifetime of the membrane and increase the overall cost with more frequent maintenance and membrane replacement. These above-mentioned issues of membrane technology can be addressed by taking advantage of the recent new advances in membrane materials. Novel materials can be incorporated into or even used to substitute the conventional membrane materials, thereby integrating their attractive properties (such as catalytic, adsorptive, and antimicrobial) with the high efficiency of membrane separation process.

As a class of uniquely porous materials, metal–organic frameworks (MOFs), which are constituted of metal ions and organic ligands by coordinate bonds, have gained increasing attention during the past decades. Their high and tunable level of porosity, adjustable and uniform pore size, abundant active sites, and large specific surface area enable them to serve as absorbents,⁸ catalysts,⁹ and membranes in water purification.¹⁰ However, most MOFs are of insufficient water stability, which severely impedes their applications in the water environment. Encouragingly, several water-stable MOFs have been demonstrated in recent years,¹¹ typically including MIL family,¹² UiO-66 series, zirconium- and pyrazole-based MOFs,¹³ which have been increasingly explored for water treatment. These known water-stable MOFs usually have strong coordination bonds or significant steric hindrance, which can prevent the potential destruction of metal–ligand bonds in the water/moisture containing environment.¹⁴  Figure 1 shows the crystal structure of a representative and well-studied water-stable MOF, UiO-66, which is constructed by octahedral Zr₆O₄(OH)₄ units [Fig. 1(a)] and terephthalate [(1,4-benzene dicarboxylate (BDC)] linkers. As shown in Fig. 1(b), each octahedral unit is connected to 12 adjacent units via BDC linkers, forming an expanded face-centered-cubic (FCC) structural unit. The high level of topological connectivity and strong coordination bonds between zirconium and oxygen collectively ensure the outstanding water stability of UiO-66, even in acidic or weak alkaline conditions. Note that MOFs are generally presented as crystalline powders, which would inevitably cause second contamination in wastewater treatment, making them hard to be recovered/recycled. Therefore, a rational integration of MOFs into mechanically stable structures can potentially make use of their advantages and at the same time overcome their own shortcomings for water and wastewater treatment.¹⁵ 

Application of MOFs in membrane technology involves the combination of MOF-based materials and membrane separation process, which can effectively improve the separation performances. First, the microporous MOFs with abundant active sites toward small ions/molecules will enhance the separation efficiency and extend the applied range of conventional membranes. Second, compared with traditional solid materials, the highly porous MOFs in filtration membranes can facilitate the transport of water molecules and improve the water permeability. Third, the pores and surface properties of MOFs can be purposely manipulated, which can be used to modify the filtration membranes for high selectivity/permeability and antifouling properties. Finally, the stable immobilization of MOFs on the bulk membranes will retard their potential aggregation and maximize the effective surface area and accessible pores/channels. Indeed, MOF materials have drawn intensive attention for water treatment as adsorbents and photocatalysts recently. For example, Li et al.¹⁷ have summarized the recent advances of MOFs for environmental pollutant elimination, mainly focusing on adsorption, advanced oxidation process, heterogeneous Fenton-like reactions, and MOF-based filtration membranes. Mon et al.¹⁸ have reviewed the latest developments of MOFs in the adsorptive removal of different contaminants. As we can see, MOF-boosted membrane technology is a multidiscipline of material science, chemical engineering, and environmental engineering. However, there still lacks qualified review papers in this area,¹⁰ especially the ones to exclusively clarify the role of MOFs in filtration membranes for water and wastewater treatment. This article will focus on the applications of MOFs in filtration membrane technology for water reclamation, with a general consideration of permeability-selectivity trade-off and membrane fouling. First, we will propose four configurations of the MOF-based filtration membranes. Together with that, the preparation strategies for these MOF-based filtration membranes are summarized. Then, the effects of MOFs on the performance of filtration membranes as well as the underlying mechanisms involved will be emphasized, which are identified as the MOF-boosted highly permeable membranes, adsorptive membranes, antifouling membranes, and catalytic membranes successively. Finally, we will conclude the main features of these membrane technologies in water and wastewater treatment, followed by the challenges and perspectives. A scheme showing the main content is provided in Fig. 2.

The availability of membranes with high quality is the key to high-efficient membrane separation, which shall be able to strictly reject the bigger molecules/particles while guarantee free access to smaller molecules. Till now, membranes with different configurations have been established, including sheets/disks/plates,¹⁹ tubes,²⁰ hollow fibers,²¹ and fibrous membranes.²² Plate membranes are among the simplest ones, which have been widely used. Hollow fiber membranes are advantageous for their low footprint, large surface area, and easy package. Hollow flat-sheet membranes are characterized by their ease of cleaning and have been widely used for membrane bioreactors.²³ Fibrous membranes are made of fibrous materials, which have been widely studied because of the large surface area-to-volume ratio, high level of porosity, high adsorption efficiency, and low cost.

Given the diversity of membrane configurations, herein, we divide the MOF-boosted membrane technology into four categories according to the physical relationship between the MOF materials and the selective layer, including MOF-plus membranes, MOF-as membranes, MOF-on membranes, and MOF-in membranes (see Fig. 3). As a subfamily of the MOF-as membranes and MOF-in membranes, free-standing membranes are also briefly introduced. Besides, the preparation strategy for each configuration and its applications in water and wastewater treatment will be summarized.

Filtration membranes have been widely used to reject contaminants in water and wastewater by the size exclusion. Based on the pore size on top surface, filtration membranes can be divided into microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO). For the small-sized molecules and ions, NF²⁴ and RO²⁵ with the pores in nanoscale and angstrom scale are required, respectively, while their fabrication is relatively challenging and of high cost and the filtration process is relatively energy intensive and of low efficiency due to the high TMP involved. On the other hand, the traditional MF/UF membranes have been well-established and commercialized,²⁶ being widely used in the separation of the bacterial and macro-molecules from water/wastewater. However, MF/UF membranes can hardly be directly used to effectively remove the dissolved heavy metal ions or small-sized complexes, since their pore size is much larger than the dimension of the dissolved metal ions and dye molecules.

The highly porous characteristic of MOF materials enables them to be promising adsorbents for metal ions and small molecules in wastewater. At present, MOFs have been demonstrated as a highly efficient adsorbent to remove organic dyes and heavy metal ions, such as Cd(II),²⁷ Pb(II),²⁸ Cr(VI),²⁹ Hg(II),³⁰ and Cu(II).³¹ However, most studies on adsorption have been conducted discontinuously at small scale, and the direct usage of MOF powders as absorbents still suffers from the unsatisfied sorption capacity.³² Based on the above consideration, highly adsorptive MOFs combined with membrane ultrafiltration (also known as MOF adsorption-membrane filtration process) have great potential in wastewater treatment. In this concept, heavy metal ions and small molecules are first adsorbed by MOFs, and the MOFs with the metal ions can be rejected by the UF membrane due to the relatively large size of MOFs (tens of nanometers), as shown in Fig. 4(a).

A pioneer work has been reported by Yin et al.³³ regarding the combination of MOF adsorption and membrane filtration for Pb(II) removal, where the effects of TMP, cross-flow velocity (CFV), and temperature were systematically investigated. It has been found that a higher TMP enables the rapid transfer of both solvent and solute through the membrane, while it causes more serious membrane fouling due to the concentration polarization. The CFV would induce the cross-flow disturbance, which can affect the hydrodynamic stability and promote the adsorption of heavy metals. At the same time, a higher CFV will enhance the diffusion induced by the shear on the membrane surface, which can effectively counter the fouling mitigation and permeate drag. The effect of temperature on water flux and removal efficiency of the MOF-plus filtration process is contradictory. On the one hand, the decreased feed viscosity with the increasing temperature would increase the water permeance and removal efficiency. On the other hand, the osmotic pressure will increase with the raising temperature,³⁴ which thus negatively affects both permeation flux and removal efficiency.

The combination of MOF adsorbents and membrane filtration can achieve an enhanced adsorption efficiency and continuous separation of heavy metal ions. Promisingly, the permeability of the MOF-plus membrane separation process is significantly improved, showing a steady-state permeate flux up to 1400 l m⁻² h⁻¹ at a CFV of 3.0 m s⁻¹ and a TMP of 0.23 MPa,³³ which is considerably higher than that of defect-free MOF membranes.¹³ Moreover, the deposition of MOF absorbents on the membrane surface just reduced the flux insignificantly because of the loosen structure. For instance, the two ceramic membranes with a pore size of 50 nm and 200 nm, respectively, showed a flux decline of less than 20% of the initial flux values [Figs. 4(c) and 4(d)], indicating that the presence of MOFs did aggravate the degree of membrane fouling.³¹ Besides, the loosen fouling layer will greatly ease the cleaning process. As shown in Fig. 4(b), the fouled membranes after cleaning by using the mixture of 0.5% HNO₃ and 5% ammonium citrate showed the comparable pure water flux to the virgin ones (1200 l m⁻² h⁻¹), with a regeneration rate of ∼100%.

The efficient adsorption of small ions/molecules by MOFs acts as a pretreatment of wastewater. Such a physical combination of MOF adsorption and membrane separation can be facilely realized after the separate preparation of MOF materials and filtration membranes. Therefore, the adsorptive MOF-plus membrane separation process can be easily scaled up. However, there are also some drawbacks to such an adsorption–filtration combination technique. For example, in the initial period, MOF nanoparticles tend to aggregate together and the big MOF aggregation will be easily swept away from the membrane surface, and therefore, a higher shear stress (i.e., a hydrodynamic force) from the cross-flow is required to re-disperse them.

MOF-as membranes refer to those MOF layers with high level of porosity and uniform pore size, which serves as the filter. They can be divided into two categories, namely, free-standing MOF membranes and supported MOF membranes [Fig. 3(c)]. Several strategies have been developed to fabricate free-standing MOF membranes, including the interfacial synthesis,³⁵ the reactive template method,³⁶ and solution casting.³⁷ These ultrathin free-standing MOF membranes show excellent permeability and selectivity, which have drawn considerable interests. However, the free-standing MOF membranes can hardly be adopted for practical application due to their poor mechanical strength, which will not be involved in this article.

To improve the mechanical strength of the MOF membranes, macroporous substrates are generally selected as the mechanical support, thereby forming the supported MOF-as membranes.³⁸ Basically, there are two strategies being developed to prepare the supported MOF membranes, including post-assembly³⁹ and in situ growth.⁴⁰ Post-assembly of the supported MOF membranes involves two steps, where the MOF particles are first synthesized and then deposited on the ceramic/polymeric supports. Although the post-assembly method is relatively facile, the supported MOF membranes thus prepared still suffer from poor stability, originating from the weak adhesion between the MOF membranes and supports, as well as the insufficient interaction between the MOF particles. To upgrade the stability of supported MOF membranes, some additives are often adopted to modify the MOF particles⁴¹ or introduced during the assembly process as binders.⁴² 

In addition to the post-assembly strategy, MOF crystals can be directly grown on the supports, which enhances interfacial adhesion. Several methods, such as solvothermal growth,^13,40 roll-to-roll hot pressing,⁴³ and electrophoretic deposition,⁴⁴ have been selected to grow the MOF membranes on supports. Among them, the solvothermal growth has been widely adopted, where the substrate is immersed into the solutions containing both organic ligand and metal salt, and the MOFs will be formed on the substrate during the heating process. The perfect MOF membranes on the supports should be featured by the strong interfacial adhesion, well intergrowth, defect free, and high controllability in thickness and orientation.

The successful growth of MOFs on the supports strongly depends on the chemical compatibility between the MOFs and the supports. The well-established example is the growth of MIL-53(Al) MOFs on the alumina supports,⁴⁰ where the metal ions (i.e., Al³⁺) in the supports can participate in the formation of MOFs [i.e., MIL-53(Al)], thereby promoting the heterogeneous nucleation, as illustrated in Fig. 5(a). However, the heterogeneous growth of other MOFs with different ions is rather challenging because of the insufficient nucleation sites. Therefore, deliberate regulation of the nucleation and growth process is essential⁴⁵ to obtain the defect-free and intergrown MOF membranes on the supports. To this end, pretreatment or modification of pristine substrates has been widely adopted.⁴⁶ This strategy could enrich the nucleation sites for heterogeneous growth, thereby benefiting the formation of continuous MOF membranes, as illustrated in Fig. 5(b). The supports can be modified by both inorganic and organic additives. The inorganic additives containing the same metal ions as the target MOFs are first decorated on the supports, which would then interact with the organic ligand to form the MOFs. Till now, ZnO nanoparticles⁴⁷ and ZnAl–CO₃ layered double hydroxides⁴⁸ have been deposited on the supports to ensure the subsequent growth of MOFs. As a result, various MOF membranes, such as ZIF-8 and UiO-66, can be successfully grown on the alumina supports. Since MOFs contain the inorganic metal ions and organic ligand, a suitable organic modification of the supports could also promote the interaction between MOFs and supports, thereby promoting the heterogeneous growth of MOFs on the supports. Among them, the most widely used modifiers include amino-terminal compounds (e.g., 3-aminopropyltriethoxysilane⁴⁹) and ionic liquids [e.g., 1-(3-aminopropyl)-3-methylimidazolium bromide]. Huang et al.⁵⁰ demonstrated a general strategy that enables the formation of compact ZIF-8 shells on various substrates, where the substrates were successively modified by (3-aminopropyl) trimethoxysilane and poly(sodium 4-styrenesulfonate). Very recently, Wu et al.⁵¹ pointed out that the organic surface modification of supports for MOF membrane preparation actually can be understood based on the “like dissolves like” principle, and as an example, ZIF-8 membranes were successfully grown on the 1H, 1H, 2H, 2H-perfluoroalkyltriethoxysilanes modified α-Al₂O₃ substrates.

The surface modification provides an effective way to promote the heterogeneous nucleation of MOFs on substrates and thus enhance the interfacial adhesion, but it cannot guarantee the formation of continuous MOF layers. As can be noted, most of the direct grown MOF membranes show imperfect intergrowth of polycrystals; namely, there exist some defects including cracks or voids in the MOF membranes. To this end, a seeded secondary growth process^40,52,53 has been developed. The MOF seeds can be first grown or deposited on the supports, and then, the seeded supports are immersed in the precursor for secondary growth, as shown in Fig. 5(c). This strategy makes it possible to control the nucleation and growth process separately. As a result, the continuous membranes with well-intergrown grains can be obtained. As a matter of fact, an MOF membrane without pinhole or crack is not equal to defect free. The crystal defects in MOFs resulting from the missed linkers, which cannot be observed using SEM, would also negatively affect the selectivity of the membranes.⁵⁴ Therefore, an even higher standard of defect-free MOF membranes is required for their application in nanofiltration and gas separation. Another advantage of using MOF as the selective layer is that it provides additional space for the surface functionalization because of the abundant chemical ligands. In this way, their binding affinity to target compounds in wastewater/water and the steric-hinderance effect of the MOF membranes can be purposely modified. For example, amino-functionalization linear molecules (such as PEG-NH₂) can be chemically grafted on MOF membranes. Since a part of the alkyl chain is accessible to the interior of the solids and the rest exposes on the surface, there can form a long superficial PEG “brush” on the resultant MOF membranes.⁵⁵ 

According to the H–P relationship,⁵⁶ the permeance of MOF membranes depends closely on their thickness. Generally, the thinner the selective layer is, the higher the permeance of the membranes would be. The supported MOF membranes are usually in the thickness of tens of micrometers since the formation relies on the intergrowth of MOF crystals. Therefore, some advanced fabrication strategies, such as layer-by-layer growth,⁴⁹ contra-diffusion,⁵⁷ microfluidic secondary growth,^58,59 and gel vapor deposition,⁶⁰ have been developed (Fig. 6), aiming at the ultrathin MOF membranes on the porous supports. During the layer-by-layer method, the substrate is exposed to each MOF component (i.e., ligands or metal ions) repeatedly to generate MOF layers. Such a process generally requires pretreatment aiming at a stronger interfacial adhesion. Although the layer-by-layer method enables the preparation of MOF membranes with high quality, it is more suitable for laboratory-scale research because of the extremely low production efficiency. Alternatively, the two MOF components, namely, metal and ligand solutions can be separately located at the two opposite faces of the substrate, respectively. With the diffusion of two precursor solutions, an MOF membrane will form on the substrate once they meet at the interface. Since the diffusion of species at the voids is faster than that in the areas covered with MOF crystals, these defects will be self-repaired automatically. Therefore, the diffusion method is advantageous for the formation of defect-free MOF membranes. Although some encouraging results have been achieved in the preparation of thin MOF membranes,³⁸ the feasibility of these thin MOF membranes for practical application is problematic. As the pore size of MOF membranes is quite small, a much higher TMP is necessary to drive the filtration process, consequently requiring the thin MOF membranes to be mechanically strong enough. Thus, there is a trade-off between the thickness and the mechanical strength of supported MOF membranes for water filtration.

When the continuous MOF layer is used as the selective layer, the inner pores will serve as the transport channels. Therefore, it is meaningful to deliberately regulate the orientation of MOF membranes, especially for the ones with an asymmetric crystalline structure, to shorten the distance of transport channels and enhance the permeability. There are a few works that reported the oriented MOF membranes. For instance, Bux and co-workers⁶² prepared a continuous and well-intergrown ZIF-8 membrane on the porous alumina support by seeded secondary growth. Although the seeds were randomly deposited, the ZIF-8 membranes were demonstrated to be highly oriented. Besides, Friebe et al.⁶³ achieved the successful preparation of oriented UiO-66 layer on alumina supports. The orientation degree of the UiO-66 membranes was determined to be 85% fraction along (002) direction. In both the above works, the preferred crystal orientation was understood based on the evolutionary selection model and the coordination modulation mechanism,⁶⁴ which has been well-demonstrated in oriented growth of zeolite membranes. Although both above works reported the successful preparation of oriented MOF membranes, we can see that the current understanding remains at a stage of phenomenological explanation, and a more in-depth scientific analysis is thus required to develop the reproducible technique.

The high porosity and uniform pore size of MOFs enable the defect-free MOF membranes among the best candidates for metal ions removal in wastewater and seawater. As the preparation of defect-free MOF membranes is rather challenging, until now only a few works reported their application in water and wastewater treatment (Table I). Due to the extremely small pores in defect-free MOF membranes, the rejection toward most metal ions can achieve a record level of above 80%, while 45% rejection was achieved for single valent ions such as Na⁺¹³.

Different from the MOF-as membranes where the MOFs are assembled/grown on the supports to form a continuous and defect-free layer, MOF particles can be alternatively decorated on the membrane surface to enhance the separation performance. The preparation process is similar to the above-mentioned ones to fabricate MOF-as membranes. In this configuration, the MOF materials serve as the functional additives, rather than the independent selective layer. Therefore, the adhesion between the MOFs and the membrane surface becomes the most important consideration. With the addition of MOF materials on the membranes, the adsorptive properties, antibacterial, selectivity, and permeability of the MOF-on membranes can be improved. For example, Ag-MOFs were in situ grown on the surface of a thin-film composite (TFC) membrane by the deposition of silver ions and subsequent reaction with NH₂-BDC.⁶⁵ Once Ag-MOFs were decorated on the surface of the TFC membrane, the color obviously changed from white to yellow [Figs. 7(a)–7(c)]. As a result, the water permeability was shown to decrease slightly from 1.1 l m⁻² h⁻¹ bars⁻¹ to 0.94 l m⁻² h⁻¹ bars⁻¹. Notably, Ag-MOF nanoparticles introduced excellent antibacterial property to the membrane, showing a nearly 100% reduction of live bacteria, as confirmed by using confocal microscopy. The additional decoration of MOFs on the membranes inevitably decreases the pure water flux because of the blocking effect. However, the unique chemical properties can also benefit the water permeance in some cases. For example, Abdullah et al.⁶⁶ reported that when the UiO-66 particles were decorated on the alumina membranes, the pure water flux was greatly dropped to 9.36 l m⁻² h⁻¹ compared with the pristine membranes of 231.24 l m⁻² h⁻¹. Surprisingly, the solute flux of the membranes for humic acid (1 gL⁻¹) was increased to 68.36 l m⁻² h⁻¹. Such an abnormal phenomenon was explained based on the adsorption–desorption analysis, and it is the similar surface charge between UiO-66 particles and humic acid at high pH values (≥9) that results in a repulsion effect during the filtration process. The interaction between the humic acid and UiO-66 can explain the improved separation efficiency, but the abnormal increment in solute flux still needs more evidence to be understood. The interesting results clearly suggest the great potential of MOFs in the advanced membrane filtration process.

When MOFs are decorated on the membrane layers, the interface adhesion between the MOFs and membrane surface is important. Those surface modification strategies developed for supported MOF membranes can be adopted to enhance the adhesion strength. Additionally, special attention can be paid to the geometry design of the substrates. Wang and co-workers⁶⁷ have designed a novel ceramic hollow fiber with two different layers. As shown in Fig. 7(d), the thin barrier layer at the lumen shows a 3D-pore network structure with a thickness of about 20 mm and a mean pore size of ∼450 nm, and the other layer with a plurality of conical microchannels toward the shell side is 500 mm and 25 mm in length and opening diameter, respectively. These micro-channels can, on the one hand, ease the mass transfer and ensure the improvement in permeation flux; on the other hand, the thus-formed pockets would benefit for the steady deposition of MOF powders during vacuum filtration. The combination of metal–organic framework and porous alumina in this unique way results in the efficient adsorption within radial microchannels without losing adsorbents during the filtration process, which is able to effectively purify the arsenic-contaminated wastewater.

Currently, three methods are mainly adopted to load MOFs into a membrane, including mechanical blending, binder-assisted binding, and in situ growth. Mechanical blending represents a facile and direct way to load various MOFs into membranes. However, the obtained hybrid membranes generally suffer from aggregation, random distribution, and low mass loading of the MOFs due to the lack of nucleation sites. The addition of binders will improve the mass loading and mechanical strength of the hybrid membranes, while it can cause the blocking of the pores and active sites in MOFs. In contrast, the in situ growth provides an effective way to loading MOFs and reserving their pristine structure/activity, which shows significant potentials in industrial applications. Nevertheless, the lack of continuous nucleation sites leads to the low mass loading and poor uniformity of MOFs in the prepared hybrid membranes. In this regard, surface modification has been widely adopted. Even so, the loading rate of MOFs on the membrane surface is still quite low (∼10%). Although the repeated growth can increase the mass loading of MOFs to some degree, it would complicate the fabrication process and reduce the product efficiency.

MOFs are widely incorporated into the polymeric membranes to form mixed matrix membranes (MMM) and thin-film nanocomposite (TFN) membranes. Since most MOFs are in the form of powders, they have been incorporated into the polymeric matrix to prepare MOF-based MMMs. In the process, the MOF powders are first added into the polymeric solution to form the precursor and then the solution is cast into membranes. Whereas the thin film membrane is formed via the interfacial polymerization (IP) between MPD aqueous solution and TMC organic solution, as shown in Fig. 8(a). As the thickness of the produced polyamide (PA) layer is several hundred nanometers, the IP reaction usually takes place on a support layer to ensure enough mechanical strength. For example, the metal ions (Zn²⁺) were first settled on a ceramic substrate. Then, the modified substrate was immersed into the solution containing organic ligands (2-methylimidazole, Hmim) and polymer [poly(sodium 4-styrenesulfonate) (PSS)]. Finally, the MOF hybrid membrane was formed as Zn²⁺ ions can bridge the organic ligands and polymer through the coordination bonds.⁴⁹ 

Compared with the traditional membrane configuration, the incorporation of MOFs into inorganic and organic nanofibrous membranes has been widely studied, taking advantage of the rapid development of the electro-spinning technique.⁷¹ The in situ self-assembly is a reliable strategy to fabricate MOF-based hybrid membranes since MOF particles can strongly connect with the polymer through the coordination bonds between the organic groups in the polymer and the metal ions of MOFs.⁷² However, the loading of MOF into the fibrous substrates is quite limited, thereby resulting in the insufficient accessibility and adsorbing capacity. To address this issue, some engineering strategies have been explored. For example, zirconium-based MOFs were grown into polyurethane foam (Zr-MOFs-PUF), followed by the hot-pressing process [Fig. 8(b)]. Due to the complete deposition of Zr-MOFs in the whole internal structure, the mass loading of MOFs has been greatly improved.⁶⁹ Inspired by biomineralization, Li et al.⁷³ proposed the preparation of electrospun-silk-nanofiber on MOF (ZIF-8 or ZIF-67) hybrid membranes with the compact loading and fully coverage of pristine MOFs. The mass loading of ZIF-8 and ZIF-67 reached 36% and 34%, respectively. The high mass loading and density is attributed to the continuous nucleation sites on the ESFs benefiting from the biomineralization effect. Due to the ultra-high loading of porous MOFs and the good accessibility, the hybrid membranes showed high removal rates of nearly 100% toward both heavy metal ions and organic dyes.

The incorporation of MOFs into inorganic membranes is relatively hard, as the interaction between the inorganic component and MOFs is generally quite poor. Only a few works have reported on the in situ growth of MOFs into the macroporous ceramic membranes.⁷⁴ For example, Bao et al.⁷⁵ proposed a surface-nucleated method to grow ZIF-67 into the alumina ceramic membranes, which were demonstrated as a unique precursor toward catalytic ceramic membranes for wastewater treatment. Besides, MOF-based nanofibrous membranes can be prepared based on the in situ reaction process using all-inorganic nanofibrous membranes as the templates [Fig. 8(c)]. Several types of flexible MOF nanofibrous membranes such as MIL-53(Al), ZIF-67, ZIF-8, HKUST-1, MIL-88B(Fe), and UiO-66 have been prepared by the complete phase transformation of metal oxide fibrous membranes.⁷⁰ The resulting hierarchical porous structure with extremely high surface area and excellent pore accessibility endows the membranes a highly efficient mass transport. The complete phase transformation process is much effective for the preparation of MOF membranes in inorganic membranes, which is substantially similar to the in situ growth of MIL-53(Al) membranes on alumina supports.

When MOFs are incorporated into filtration membranes, their unique features will collectively or separately contribute to the separation performance depending on the configuration. The possible contributions of MOFs to the performance of filtration membranes including permeability, selectivity, and antifouling are summarized in Table II. The effect of MOFs on the selectivity of filtration membranes has been mainly attributed to their adsorption ability to small ions/molecules; therefore, it is specified as the MOF-boosted adsorptive membranes. The development of catalytic filtration membranes is aiming at solving the membrane fouling issues. However, different from the traditional antifouling membranes to retard the fouling process, catalytic filtration membranes can achieve the self-cleaning. So, we will introduce the catalytic membranes in an individual section.

The highly porous MOFs can be incorporated into the membrane layers to increase the porosity of the membranes. Since the pore sizes in MOFs are usually in the range of several nanometers or even smaller, their contribution to the porosity of nanofiltration membranes would be significant, while its effect on microfiltration or ultrafiltration can be ignored. Nanofiltration has been widely utilized in the fields of drinking water, pharmaceuticals, textile, food, etc., due to its low energy consumption and maintenance cost compared with the distillation and reverse osmosis.

Thin-film composite (TFC) membranes composed of a support layer and a polyamide (PA) active layer are widely used in nanofiltration (NF). However, it is quite challenging to achieve both high water permeation and high rejection of multivalent ions and antifouling resistance simultaneously. The incorporation of nanoparticles into the active layer during the interfacial polymerization of polyamide has been demonstrated to be an effective way to increase the water permeability. The synthesized membranes are the well-known TFN membranes.⁷⁶ When the porous MOFs are introduced to fabricate the TFN membranes, the polymer chain packing will be disrupted due to the existence of MOFs. As a result, the fractional free-volume in the active layer will increase, thereby improving their water permeability. To further increase the porosity of the membranes, hollow MOFs with both micropores and big pores have been purposely designed and incorporated in the NF membranes.⁷⁷ If the MOFs introduced into the NF membranes are hydrophilic, their contribution to water permeability would be further enhanced.^{68,72,78–80} The hydrophilic MOFs would increase the surface hydrophilicity of NFs, which would promote the formation of a hydration layer on the membrane surface, thereby increasing their access to water molecules. As shown in Figs. 9(a)–9(d), with the aid of tannic acid (TA), hydrophilic hollow zeolitic imidazolate framework-8 (hZIF-8) has been synthesized via a surface functionalization-assisted etching process. The water permeation of hybrid UF membranes with hZIF-8 was highly improved (597 l m⁻² h⁻¹) compared with the pristine PSf membrane (210 l m⁻² h⁻¹), while the rejection rate was well-maintained [Fig. 9(e)]. Unfortunately, water-stable MOFs are usually poor in hydrophilicity, and there thus requires certain surface modification by hydrophilic polymers. In this way, these polymer chains on the surface of MOF nanoparticles also effectively prevent the aggregation and improve their dispersibility.⁸¹ In addition, these porous MOFs have also been incorporated into the lamellar membranes, so as to widen the internal channels and improve the water permeability.⁸² 

Recently, nanovoids as a novel secondary phase have been introduced to increase the free volume of TFC NF membranes. For instance, Tan et al.⁸⁴ successfully generated nanovoids in the polyamide membranes from the Turing-type nanobubbles and channels. However, one can note that the preparation process is rather complicated, and it is also hard to control the nanovoids generated. In this regard, the water unstable MOFs can be introduced into the membrane layers first and then dissolved by water. In this way, some nanovoids are produced in the membranes, and their permeability would be improved due to the enhanced porosity. For example, Wang et al.⁸³ prepared TFC NF membranes with ZIF-8 nanoparticles loaded between polyethersulfone composite support and polyamide (PA) layer. After interfacial polymerization, the water unstable ZIF-8 nanoparticles inside can be removed by water dissolution, and a rough PA active layer with crumpled nanostructure was formed [Figs. 9(f)–9(l)]. Due to the increase in surface roughness and free volume, the NF membranes showed a greatly enhanced permeance of 53.5 l m⁻² h⁻¹ bars⁻¹ and a high Na₂SO₄ rejection of above 95%.

When MOFs are used to prepare the thin-film nanocomposite (TFN) membranes, the structure–performance relationship is closely related to the intrinsic characteristics of MOFs and membrane fabrication process. Zhao et al.⁸⁵ reported a comparable study of permselectivity in TFN membranes with the incorporation of different water-stable MOFs, namely, MIL-53(Al), NH₂-UiO-66, and ZIF-8 as the fillers. They found that the addition of MOFs increased the thickness, surface roughness, and surface negative charge of the polyamide active layer, accompanied with the decreased cross-linking degree. Furthermore, the interaction between MIL-53(Al) and polyamide is much stronger than that between NH₂-UiO-66 or ZIF-8. The differences in interfacial strength and the hydrophilicity of MOF particles collectively determined the perm selectivity. The introduction of MOFs into the polymer matrix is the most common way to prepare the thin-film nanocomposite (TFN) membranes. Recently, Paseta et al.⁸⁶ proposed a bilayered thin-film composite membranes, where a continuous HKUST-1 MOF layer was first prepared onto polyimide supports through interfacial synthesis and a PA layer was then coated on the surface by interfacial polymerization. When the membranes were applied to remove diclofenac and naproxen from water, the bilayered thin-film composite membranes showed a water permeance of 33.1 l m⁻² h⁻¹ bars⁻¹ and 24.9 l m⁻² h⁻¹ bars⁻¹, respectively, along with a rejection of ≥98%. The increase in water permeability is correlated with the porosity of MOFs used, the improved hydrophilicity, and surface roughness of the membranes. The results demonstrate that highly porous MOFs can improve the permeability of TFN membranes regardless of their position and state (continuous or discontinuous).

The strong chelating between small-sized contaminants in water (such as metal ions and dye molecules) and functional groups of organic linkers (e.g., hydroxyl and amide) in MOFs enables the high adsorption of MOFs toward these molecules/ions.⁸⁷ The adsorption mechanisms between MOFs and small molecules/ions include electrostatic interactions, hydrogen bonding, acid–base interactions, π-π stacking/interactions, and cation exchange process. Because of the highly chemical tunability of MOFs, different groups can be purposely introduced into the crystal structure to regulate these interactions,⁸⁸ thereby improving the adsorption efficiency. For example, as demonstrated by Zhou and co-workers,⁸⁹ the removal efficiency of MOFs toward Hg(II) can be greatly improved when methylthio groups (−SCH₃) were introduced. In addition, the MIL MOFs have been demonstrated as excellent absorbents to remove arsenic ions. Specifically, due to the Fe–O–As bond, MIL-100(Fe) showed an extremely high adsorption capacity for arsenic (As⁵⁺) removal from wastewater, which is 6-fold higher than that of commercial Fe₂O₃ nanoparticle powders (50 nm).⁹⁰ 

Due to the scalable preparation of MOFs (for example, in kilogram) is still challenging, their practical utilization as adsorbents for heavy metal removal has not been achieved. As mentioned in Sec. II A, MOFs can be used as the adsorbents in the feed solution, followed by membrane filtration [Fig. 3(a)]. Alternatively, immobilization of the MOF particles onto the external surface or the internal pores of the membranes will promote their application and improve the efficiency [Fig. 3(b)]. As it is known, the traditional membrane filtration system for separation is based on the size exclusion mechanism. When highly porous MOFs are introduced, the separation efficiency of the membranes toward small molecules and metal ions would be improved with the additional adsorptive separation mechanism. The MOF-boosted adsorptive membranes are generally evaluated by the adsorption capacity, selectivity, regeneration ability, and cost-efficiency. Table III presents the adsorption capacity, water flux, and removal efficiency reported in some of the recent studies on the MOF-boosted adsorptive membranes.

MOFs show high adsorption capacity toward metal ions and small molecules due to their high specific surface area and accessible porosity. Intrinsically, the adsorption capacity depends on the number of active sites in MOFs, which have an affinity to the target components. For example, UiO-66 has been recorded an adsorption capacity of 303.4 mg g⁻¹ for As⁵⁺ removal, which is higher than that of any other reported MOFs and currently available adsorbents (less than 280 mg g⁻¹).¹⁶ The highly porous crystalline structure of UiO-66 together with surface functional groups such as hydroxyl group and benzene dicarboxylate ligand delivers a large number of available active sites. In particular, the formation of the Zr–O–As coordination bond can accelerate the adsorption of As⁵⁺ in UiO-66. During the adsorption process, the highly porous microstructure of MOFs ensures the abundant transfer pathways for metal ions, while the active sites on the MOFs provide the position for the adsorption of specific ions/molecules. As such, an effective activation process is significantly vital to ensure a high pore structure available. These activation routes include conventional vacuum drying at the appropriate temperature and solvent exchange, followed by vacuum drying for porous water-stable MOF materials.⁹³ 

The integration of MOFs onto/into filtration membranes can overcome the problems of the direct utilization in powder form, including aggregation, recycling ability, and secondary contamination, while some other factors should be considered for the MOF-boosted adsorptive membranes. Basically, the adsorption capacity is positively proportional to the mass loading of MOFs on the filtration membranes. However, the MOF particles tend to aggregate at high concentrations with the sacrifice of surface area and active sites, thereby reducing the adsorption capacity of MOF-boosted membranes. As has been reported,⁷¹ when the MOF content is increased to 15%, the MOFs in the nanofibers matrix could aggregate together, which would decrease the surface area, block the diffusion of metal ions, and sacrifice the available active sites. At the same time, when MOFs are loaded into the PAN/chitosan nanofibers, the resultant membranes show decreased water flux, compared with the pristine PAN/chitosan nanofibers. This is mainly related to the reduced open pores and thereby raising resistance for water transfer.

An effective way to boost the adsorption capacity of MOF-boosted adsorptive membranes is to engineer the location of MOFs on the membrane. Compared with the attachment on the membrane surface, a uniform distribution of MOFs in the whole membrane would increase their mass loading, and at the same time avoid the undesirable aggregation. With the intensive interests in the natural materials, some special attention has been paid to the natural woods as the supports for MOF-boosted adsorptive membranes. As we know, wood cell walls contain a large content of cellulose, where the hydroxyl groups can easily coordinate with metal ions. When the metal ions modified wood channels are exposed to organic ligands, the 3D MOF incorporated wood membranes can be formed. The open and aligned channel in woods would ensure the MOF nanoparticles inside contact entirely with the organic pollutants, thereby improving the adsorption efficiency, as shown in Figs. 10(a)–10(c). Moreover, the abundant wood resource in nature would potentially reduce the cost and enable their scale-up for practical applications. As has been demonstrated,⁹² the UiO-66 MOF incorporated natural wood as the adsorptive membranes show a dramatic increase in the removal efficiency. Significantly, the all-in-one filter with the combination of three UiO-66/wood membranes maintained a removal efficiency of more than 97% even at a flux up to 1.0 × 10³ l m⁻² h⁻¹ [Figs. 10(d)–10(g)]. First, the uniform and full coverage of UiO-66 MOFs in the whole aligned wood lumens enhanced the utilization efficiency of UiO-66 MOFs with the high surface area. Second, the UiO-66 MOF nanoparticles have sufficient contact with organic contaminants in wastewater, taking advantage of the elongated and irregular channels in the wood matrix. Finally, the aligned microchannels with low tortuosity and hydrophilicity would benefit the rapid transport of water molecules inside. Importantly, the three-layer filter can be easily generated when being washed with methanol (3 × 5 ml) and de-ionized (DI) water (10 ml). Even after six times of regeneration, the three-layer filter remained at a removal efficiency of more than 95.0% [Fig. 10(h)], which is superior to that of activated carbon, UV-treated carbon, and natural wood [Fig. 10(i)].

In a typical adsorption process, the adsorption capacity is also related to pH and temperature. The pH value of the feed solution can be adjusted to increase the adsorption capacity with the maximization of the electrostatic attraction between MOFs and pollutes. For example, UiO-66 MOFs have a positive surface charge in the pH range of 1–6 and a negative one in the range of 7–10. The surface charged UiO-66 MOFs helps capture organic molecules through electrostatic attraction. Cationic dyes such as Rh6G can thus be easily absorbed by the negatively charged UiO-66 MOFs through electrostatic attraction at a high pH value.⁹² Besides, the removal efficiency would be dramatically enhanced when the pH is above 4, demonstrating that both weak acid and alkaline environment would promote the adsorption of Rh6G on the UiO-66.

The contaminants in real wastewater are quite complex and diversiform; thereby, studying on the adsorption capacity of MOFs toward more than one kind of contaminants is of significance. As an example, zirconium-based MOFs were loaded on polyurethane foam (Zr-MOFs-PUF), and their removal efficiency toward wastewater containing neutral, cationic, and anionic dyes was conducted.⁶⁹ The results showed that the Zr-MOFs-PUF membrane could simultaneously remove rhodamine B (RB), methylene blue (MB), and Congo red (CR) in their binary system (i.e., RB/MB, RB/CR and MB/CR mixtures) and ternary system [Figs. 11(a)–11(d)]. The effective adsorption is a collective result of hydrogen bond interaction, electrostatic interaction, and Lewis acid–base interaction between the membrane and dye molecules, as shown in Fig. 11(e). For example, the adsorption of MB on Zr-MOFs-PUF membranes is attributed to electrostatic attraction, since the surface charge of MB is positive while that of Zr-MOFs-PUF membranes is negative due to the free carboxyl acid of organic linkers. Whereas for the neutral dye RB, the electrostatic interaction also exists, because of the zwitterion feature of RB, which contains both cationic and anionic functional groups. Regarding the adsorption of CR on Zr-MOFs-PUF membranes, both the electrostatic interaction and Lewis acid–base interaction are involved. When the Zr-MOFs-PUF membranes were introduced, the CR dye solution becomes weak acidic, as confirmed by the color change of the solution from red to blue. The partially protonated NH₂ in CR molecules would be electrostatically adsorbed on the Zr-MOFs-PUF membranes. At the same time, the sulfonate groups in CR would form Zr-(SO₃⁻) with the open active sites of zirconium ions through the Lewis acid–base interaction.

As both the configuration and surface chemistry of heavy metal ions depend on the pH value of the solution, their adsorptive behavior on MOFs is rather complex than that of organic dyes. Therefore, it is necessary to execute these adsorption experiments at different pH conditions. Since the Pb(II) solution is usually acidic and the isoelectric point (IEP) of Zn/Co-ZIF is observed in alkali solution, a strong electrostatic attraction has been demonstrated between Zn/Co-ZIF microcrystals and Pb(II) ions in acidic solution. The strength of electrostatic attraction is fundamentally determined by the number of negative charges in solution, which serves as the active sites for metal ion adsorption. When the pH value is increased above 6, the adsorption of metal ions starts to reduce due to the formation of Pb(II) hydroxylated complexes, which will also consume the active site of adsorbents. Although the effect of pH on the adsorption of both organic dyes and metal ions by MOFs shares the same principle, the adsorption process between MOFs and metal ions is more complex due to the possible hydration and hydroxylation process.

Another factor that affects the adsorption process between MOFs and small molecules such as metal ions and organic dyes is temperature. On the one hand, the adsorption is an exothermic process; thus, an increase in temperature would reduce the adsorption capacity. On the other hand, the increase in temperature would accelerate the mass transfer and benefit for the adsorption process. Generally, the adsorption process can be described by the thermodynamic parameters including standard Gibbs free energy change (ΔG₀), standard enthalpy change (ΔH₀), and standard entropy change (ΔS₀).⁹¹ Namely, the overall negativity and the increase in ΔG₀ value with rising temperatures suggest a spontaneous adsorption process. While a positive ΔH₀ indicates the increased randomness at the solid–solution interface during the adsorption process. Besides, the positive enthalpy suggests an endothermic adsorption interaction.

Other aspects such as adsorption mechanism, adsorption efficiency, preference, stability, and recycling ability should also be considered for the potential practical application. The adsorption mechanism involved in the interaction between adsorbents and adsorbates can be identified by Fourier transforms infrared (FT-IR) and photoelectron spectroscopy (XPS). Adsorption experiments toward metal ions at different concentrations should be conducted to understand the adsorption efficiency. Note that the adsorption efficiency will not change significantly with the increase in concentration once the active sites on the adsorbents are saturated. The preference of the adsorptive membrane toward different ions can be evaluated by the batch adsorption experiments in different solutions, reflecting by the competitive capture ability in mixed ions. For example, batch experiments showed a good selectivity of carboxymethylcellulose sodium (CMC)-MOF/cloth composite membranes for Pb(II).⁹¹ A continuous adsorption–desorption process was further performed with the use of methyl alcohol and HCl solution to desorb the Pb(II) ions on MOFs. The removal efficiency of CMC-MOF/cloth composite membranes for Pb(II) remained 79.12% after five cycles, demonstrating the good reusability of the CMC-MOF/cloth composite membranes. Both the intrinsic chemical stability of MOFs and their adhesion to substrates would determine the stability of the whole membranes. Therefore, the selection of water-stable MOFs is the primary step to prepare the corresponding adsorptive membranes for water treatment. Usually, some structural characterizations, such as X-ray diffraction (XRD), Brunauer–Emmett–Teller (BET) surface areas, scanning electron microscopy (SEM), and inductively coupled plasma mass spectrometry (ICP-MS), are used to evaluate the stability of MOFs.¹³ Regarding the interfacial adhesion between MOFs and the substrates, a common way is to examine the chemical composition of the permeate solution. Due to the high compatibility with polymers and organic linkers in MOFs, the MOF-boosted polymeric membranes usually show good stability. For example, when the Zr/Fe-MOFs are incorporated in the PVDF nanofibrous membranes, there was no MOF detected in the permeates after four cycles with the successive filtration and desorption process.⁹⁵ 

Membrane fouling refers to the deposition of organic, inorganic, and/or biological components on the external surface and/or in the internal pore walls of membranes. As the major issue for the membrane-based separation process, membrane fouling reduces the permeation flux, deteriorate the produced water quality, and increase energy consumption. Various methods have been developed to minimize/mitigate membrane fouling in terms of pre-treatment, operation optimization, and membrane engineering. As the filtration membrane is the core for the separation process, great attention has been focused on membrane engineering, including surface modification⁹⁶ and microstructure design/optimization.⁹⁷ Surface modification means the regulation of surface roughness, charge, and hydrophilicity, which has been widely adopted to develop antifouling membranes,⁹⁸ while the microstructure design mainly involves the optimization of the pore structure and the membrane configuration.

Depending on the filtration types, membrane fouling can be dictated by the flux decline ratio (FDR) under constant transmembrane pressure condition (dead-end) or the TMP increase ratio under a constant flux condition. As has recently been demonstrated by Akamatsu et al.⁹⁹ the initial adsorption of foulants on the membrane surface is crucial for the formation of the fouling layer. Therefore, the development of the membrane surface with high anti-adhesive capacity against the foulants is significantly necessary for antifouling membranes. The anti-adhesive ability can be determined directly by the adsorption capacity of foulants on the membrane surface or indirectly by comparing the pure water flux before and after the static adsorptive experiments. The normalized flux decline with the filtration time has been widely adopted to evaluate the antifouling performance. A slower decrease in flux suggests a higher antifouling capacity of the membrane. For the fouled membranes, one can use different methods to wash them (hydraulic backwashing and chemical cleaning) and compare the pure water permeate before and after the washing. In this way, one can determine the contribution of irreversible and reversible fouling according to the flux recovery ratio (FRR). Some representative studies on the MOF-boosted antifouling TFC membranes are summarized in Table IV.

Given that the membrane fouling is primarily correlating to the interaction between the membrane surface and foulants, the surface characteristics such as roughness, charge, wettability, and morphology can be regulated to minimize the formation of fouling layers. Due to the diversity of MOFs and their tenability in surface charge and wettability, the introduction of MOFs can open a broad avenue to develop antifouling membranes for water and wastewater treatment. Moreover, the open channels with the expose of metal ions in MOFs would enable the corresponding membranes showing improved antibacterial properties, which would be very efficient in retarding the biofouling process.

The hydrophilicity of the filtration membranes is one of the most important surface properties for water and wastewater treatment, as a hydrophilic surface benefits the formation of a hydration layer on the membrane surface that would improve the water permeability and reject the hydrophobic foulants, thereby enhancing the fouling resistance at the same time. However, water-stable MOFs usually show poor hydrophilicity.¹⁰⁴ Therefore, several strategies have been formulated to increase the hydrophilicity of these MOFs prior to their incorporation into the filtration membranes. For example, MOFs Cu (tpa) and graphene oxide were added together into the PES ultrafiltration membranes,¹⁰¹ where the hydrophilic graphene oxides can improve both the hydrophilicity [Fig. 12(a)] and antifouling properties [Figs. 12(b)–12(d)] of the PES ultrafiltration membranes. Besides, a surface engineering strategy involving the etching by tannic acid has been developed to obtain the hydrophilic ZIF-8 MOFs.⁸⁰ The hydrophilic ZIF-8 showed a hollow structure, and their incorporation into the polysulfone (PSf) UF membranes greatly enhanced the surface hydrophilicity and negative charge, both of which are highly desirable for high-permeable and antifouling membranes for water treatment.

Surface charge of filtration membranes is another important factor to regulate the antifouling properties. When the surface change of the filtration membrane is the same as that of the foulants in the wastewater, the attachment of the foulants onto the membrane surface could be weakened due to the electrostatic repelling effect. The surface charge depends on the pH of the solution; thus, during the membrane filtration process, the pH value of the feed solution can be regulated so as to maximize the electrostatic repelling effect. Note that the contribution of electrostatic interaction to antifouling properties is relatively weak in UF and MF, compared with that of surface hydrophilicity. It has been demonstrated that the electrostatic interaction can contribute to the anti-adhesion at low transmembrane pressure¹⁰⁵ or under static adsorptive condition.⁹⁹ However, the contribution of surface charge to the antifouling potential would be significant in NF and RO, mainly because the nanosized channels would enhance the electrostatic interaction between the membrane surface and foulants.

The unique antibacterial feature of MOFs enables their incorporated membranes showing enhanced anti-bacterial and anti-biofouling properties. Moreover, MOFs are superior to other metal-based antimicrobial inorganic nanomaterials, such as Ag nanoparticles, Cu nanoparticles, and ZnO nanoparticles, regarding the aggregation and poor compatibility with polymeric matrix.¹⁰⁶ However, the possible release of metal ions from MOFs would cause some defects to the membranes especially in long-term service. The inherent disadvantage triggers great efforts to seek water-stable MOFs to fabricate the anti-biofouling membrane. For example, water-stable Cu-BTTri MOFs were incorporated into the TNF membranes, and the antibacterial properties were investigated using Pseudomonas aeruginosa as an example.¹⁰⁰ The obvious reduction in flux decline ratio (FDR, ∼36%) compared with the controlled samples (∼70%) under cross-flow condition confirmed the excellent antifouling performance of Cu-BTTri MOFs incorporated TNF membranes. The enhanced antibacterial properties were demonstrated by counting the bacteria on the membranes after 24 h of contact. Given that the Cu-BTTri MOFs are stable in water, the release of Cu ions is not the dominant factor in the antibacterial properties. Alternatively, the direct contact of bacteria with membrane surface would cause their damage, as a result of the depolarization of bacteria membrane and oxidation of functional groups in the bacteria. Similarly, some other MOFs such as silver-based MOFs¹⁰² have been incorporated into the TFN membrane to further improve their permeability, fouling resistance, and antibacterial performance.

The deposition/decoration of MOF particles on the membrane surface is an alternative way to deliver the antibacterial properties. For example, Ag-MOFs were decorated on the surface of polyamide TFC membranes by a facile in situ assembly via covalent binding.⁶⁵ Due to the existence of Ag-MOFs on the membrane surface, the antibacterial property was highly improved with complete elimination of live bacteria, while the water permeability was shown to decrease slightly from 1.1 l m⁻² h⁻¹ bars⁻¹ to 0.94 l m⁻² h⁻¹ bars^-1. Moreover, after immersion in water and depletion of silver for several months, there observed the regeneration of Ag-MOFs, suggesting the potential for long-term application. Besides, there existed plenty of carboxylic and amine-containing groups on the Ag-MOF functionalized membrane surface, and the resultant hydrophilic surface would prevent the adhesion of hydrophobic foulants and improve the fouling resistance. In addition, 2D MOF nanosheets have been incorporated into the TFN membranes during the interfacial polymerization.¹⁰³ Because of the improved hydrophilicity and biocidal ability, the CuBDC incorporated TNF membranes show enhanced water flux and anti-organic/bacterial behavior.

Among the huge family of MOFs, some of them show catalytic degradation of organic dyes or/and reduction of heavy metal ions.¹⁰⁷ When these MOFs are incorporated into filtration membranes, catalytic membranes will be developed. Compared with traditional catalysts (such as metal oxides), MOFs with high surface area and high level of porosity are extremely advantageous in maximizing the accessible active sites, since the catalytic process is closely related to the active sites on the surface. First, the large surface area and the highly porous structure can accelerate the adsorption process and increase the adsorption capacity of pollutants, which are fundamentally essential for catalytic degradation. Second, the catalytic active sites in MOFs would decompose the pollutant molecules into smaller ones, and their adsorption by the MOFs are more readily. Third, the successive adsorption and catalytic degradation enable the regeneration and recycling of MOFs. However, MOFs in the powder form can hardly be separated from the treated water for recycling usage, which greatly impedes their practical applications. With the introduction of catalytic MOFs into membranes, self-clean catalytic membranes will be generated, which can improve the selectivity and effectively address the membrane fouling due to the degradation of contaminants in solution.

Till now, only a few investigations have been reported on the integration of photocatalytic MOFs and filtration membranes for water and wastewater treatment. Du et al.¹⁰⁸ prepared the UiO-66-NH₂ (Zr) membranes on the α-Al₂O₃ substrate via a reactive seeding approach, and the obtained MOF membranes were then used to reduce Cr(VI) ions in wastewater. Under the irradiation of sunlight, UiO-66-NH₂(Zr) membranes exhibited a photocatalytic reduction efficiency of ∼93.1% from Cr(VI) to Cr(III) within 120 min, while that of the α-Al₂O₃ substrate was negligible [Fig. 13(a)]. Moreover, the reduction efficiency of Cr(VI) remained above 94% after 20 cycles [Fig. 13(b)]. The excellent cycling stability is attributed to the relatively low reaction rate resulting from the decreased contact area with Cr (VI) compared with that of MOF powders. Furthermore, the UiO-66-NH₂(Zr) in wastewater was quite stable, as shown in Fig. 13(c), MOF particles well remained on the surface. When the photocatalytic reduction was conducted in the real lake water, the MOF membranes showed a reduction efficiency of 97% within 120 min [Fig. 13(d)], demonstrating their potential for real wastewater treatment.

The rapid recombination of photogenerated electrons and holes often gives rise to the unsatisfactory efficiencies of the photocatalytic MOFs. Some researchers have proposed to modify MOFs with metals or metal oxides to promote the separation of photogenerated electrons and holes. However, they would increase the overall cost for water and wastewater treatment especially when noble metals are incorporated. Alternatively, the oxidants such as ammonium persulfate [(NH₄)₂S₂O₈], hydrogen peroxide (H₂O₂), persulfate (PS), and peroxymonosulfate (PMS) can be added as electron acceptors into the photocatalytic process to promote the separation of photogenerated electrons and holes. Actually, the coupling of oxidation and membrane filtration is known as a traditional pathway to enhance the pollutant treatment efficiency and prevent membrane fouling through the generation of reactive oxygen species. The immobilization of photocatalysts onto the membrane surface can decrease the aggregation of nanoparticles and increase the surface area. MOFs are known as the unique precursors for well-dispersed metal/metal oxide nanoparticles. Recently, Bao et al.⁷⁵ prepared the Co-ZIF on alumina ceramic membranes and then converted to the Co₃O₄ nanocatalysts with a honeycomb structure [Figs. 14(a) and 14(b)]. After the deposition of Co₃O₄ nanoparticles on the alumina membranes, there did not show a detectable change in weight and the pure water permeability slightly reduced from 3323 l m⁻² h⁻¹ bars⁻¹ to 3020 l m⁻² h⁻¹ bars⁻¹. Both results suggested the small amount of Co₃O₄ nanolayer that did not increase the filtration resistance too much. During the filtration process, sulfamethoxazole (SMX) as one kind of widely used antibiotic in the wastewater can be retained on the membrane surface. When the oxone was introduced, the Co₃O₄ nanoparticles on the membrane surface would act as the catalyst to activate the oxone and thereby produce the reactive oxygen species (ROS) that can degrade the SMX. As shown in Figs. 14(c)–14(e), the removal efficiency toward SMX was gradually increased with the increasing concentration. In particular, the removal efficiency of M450 is significantly higher than that of the pristine ceramic membranes (M0).

The catalytic membranes (M450) show a slight decrease of <10% in the SMX removal after three cycles [Fig. 14(f)]. The authors attributed the decline to the accumulation of SMX and the degraded products on the membrane surface, which reduced the density of accessible active sites. After cleaning with the oxone solution (0.1 g l⁻¹), the removal efficiency toward SMX returned to more than 95%, suggesting that the durability of Co₃O₄ membranes can be realized by oxone solution cleaning. However, accompanying with the degradation of SMX, the generated smaller molecules would become the secondary contaminants in the permeate water, which have not been examined in the work. In addition to the removal of targeted pollutants, more attention should be paid to the quality of permeate with the consideration of other possible secondary contaminants.

MOFs with high and tunable porosity, large surface area, and great structural diversity have drawn increasing attention in water and environmental applications in recent years. Instead of the direct use of MOF powders/monoliths, a proper integration of MOFs with filtration membranes is a promising engineering pathway for widening their practical applications. In the field of filtration membranes, there has been rapid progress in the controlling of permeability-selectivity trade-off and significantly minimized membrane fouling. The same applies to MOF-mediated membranes, such as high-permeable membranes, adsorptive membranes, antifouling membranes, and catalytic membranes. Indeed, great efforts have been devoted to them for water sustainability, and some encouraging results have been demonstrated that have stimulated the global interest from academic research, prototype development, and industrial applications.

Despite the very encouraging process, there are still many challenges to be addressed. First, poor water stability is the biggest challenge. Except for a few types of MOFs such as MIL, UiO-66, and ZIF series, most of the MOFs are unstable in the presence of water, which would cause the difficulty for recovery and second pollution. The well-adapted strategy to improve the water stability of MOFs is the hydrophobic modification; however, it is not suitable for their applications in filtration membranes for water treatment. Therefore, continuous efforts should be devoted to exploring new strategies to enhance the intrinsic water stability of MOFs, especially in the wastewater environment, and at the same time, without sacrificing the hydrophilicity. Second, the reported methods for MOF preparation have been largely achieved in the laboratory. Therefore, novel fabrication methods and strategies are urgently needed to prepare the designed MOFs in a large scale (at least kilogram quantities) for practical applications. In addition, the interfacial adhesion between MOFs and substrates (e.g., those ceramic filtration membranes) should also been further enhanced. Third, multifunctionalities will undoubtedly be among the new developments. One example is the introduction of photocatalytic MOFs into filtration membranes, which is quite promising for the development of antifouling and self-clean membranes in water and wastewater treatment. In this regard, engineering strategies based on the recent progress in traditional photocatalysts can be adopted to improve the efficiency. Besides, MOFs can be further modified by metal-oxoclusters and organic linkers to achieve a larger surface area, higher stability, stronger light response, etc., and eventually a boosted photocatalytic performance. The contaminants in wastewater after photocatalytic degradation may cause secondary contaminants. Therefore, the photocatalytic MOFs with the ability to degrade the contaminants into eco-friendly substances are the best choice. Alternatively, the products after the photocatalytic process can get away from the liquid system automatically such as water-insoluble gas. Finally, there are some new emerging applications of MOF-boosted filtration membranes for water sustainability such as water capture¹⁰⁹ and photo-thermal desalination,¹¹⁰ which are also worthy to be investigated thoroughly.

This work was supported by the National Research Foundation Singapore (Grant No. NRF-CRP17-2017-01), conducted at the National University of Singapore.

BDC

benzene dicarboxylate

BET

Brunauer–Emmett–Teller

CFV

cross-flow velocity

CMC

Carboxymethylcellulose sodium

CR

Congo red

DMF

N, N-dimethylformamide

FDR

flux decline ratio

FM

filtration membrane

FRR

flux recovery ratio

FT-IR

Fourier transforms infrared (FT-IR)

GO

graphene oxide

HAc

acetic acid

HKUST

Hong Kong University of Science and Technology

Hmim

2-methylimidazole

HPAN

hydrolyzed polyacrylonitrile

hZIF-8

hollow zeolitic imidazolate framework-8

ICP-MS

Inductively coupled plasma mass spectrometry

IEP

isoelectric point

IP

interfacial polymerization

MB

methylene blue

MF

microfiltration

MIL

Materials Institut Lavoisier

MMM

mixed matrix membrane

MOF

metal—organic framework

MPD

m-phenylenediamine

NF

nanofiltration

PA

polyamide

PAN

polyacrylonitrile

PD

polydopamine

PES

polyethersulfone

PHF

polymer hollow fiber

PIP

piperazine

PMS

peroxymonosulfate

PS

persulfate

PSf

polysulfone

PSS

poly(sodium 4-styrenesulfonate)

PTFE

polytetrafluoroethylene

PUF

polyurethane foam

PVDF

polyvinylidene fluoride

RB

rhodamine B

RO

reverse osmosis

ROS

reactive oxygen species

SEM

scanning electron microscopy

SMX

sulfamethoxazole

SWCNTs

single-wall carbon nanotubes

TA

tannic acid

TEM

transmission electron microscopy

TFC

thin-film composite

TFN

thin-film nanocomposite

TMC

trimesoyl chloride

TMP

transmembrane pressure

TPA

terephthalic acid

UF

ultrafiltration

UiO

University of Oslo

UV-vis

ultraviolet-visible

XPS

X-ray photoelectron spectroscopy

XRD

X-ray diffraction

ZIF

zeolitic imidazolate framework

BDC

benzene dicarboxylate

BET

Brunauer–Emmett–Teller

CFV

cross-flow velocity

CMC

Carboxymethylcellulose sodium

CR

Congo red

DMF

N, N-dimethylformamide

FDR

flux decline ratio

FM

filtration membrane

FRR

flux recovery ratio

FT-IR

Fourier transforms infrared (FT-IR)

GO

graphene oxide

HAc

acetic acid

HKUST

Hong Kong University of Science and Technology

Hmim

2-methylimidazole

HPAN

hydrolyzed polyacrylonitrile

hZIF-8

hollow zeolitic imidazolate framework-8

ICP-MS

Inductively coupled plasma mass spectrometry

IEP

isoelectric point

IP

interfacial polymerization

MB

methylene blue

MF

microfiltration

MIL

Materials Institut Lavoisier

MMM

mixed matrix membrane

MOF

metal—organic framework

MPD

m-phenylenediamine

NF

nanofiltration

PA

polyamide

PAN

polyacrylonitrile

PD

polydopamine

PES

polyethersulfone

PHF

polymer hollow fiber

PIP

piperazine

PMS

peroxymonosulfate

PS

persulfate

PSf

polysulfone

PSS

poly(sodium 4-styrenesulfonate)

PTFE

polytetrafluoroethylene

PUF

polyurethane foam

PVDF

polyvinylidene fluoride

RB

rhodamine B

RO

reverse osmosis

ROS

reactive oxygen species

SEM

scanning electron microscopy

SMX

sulfamethoxazole

SWCNTs

single-wall carbon nanotubes

TA

tannic acid

TEM

transmission electron microscopy

TFC

thin-film composite

TFN

thin-film nanocomposite

TMC

trimesoyl chloride

TMP

transmembrane pressure

TPA

terephthalic acid

UF

ultrafiltration

UiO

University of Oslo

UV-vis

ultraviolet-visible

XPS

X-ray photoelectron spectroscopy

XRD

X-ray diffraction

ZIF

zeolitic imidazolate framework