Composite materials for organic solvent filtration

The composite material with a polycrystalline MOF layer, formed via amorphous-to-crystalline transformation, addresses scalability and stability issues in nanofiltration and ultrafiltration by tuning pore sizes for efficient organic solvent filtration and selective solute removal, enhancing separation performance and stability.

WO2024243557A1PCT designated stage expired Publication Date: 2024-11-28NATIONAL UNIVERSITY OF SINGAPORE +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
PCT/US2024/031107
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-24
Filing Date
2024-05-24
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

Current nanomaterials for organic solvent filtration face challenges such as lack of scalable processing strategies, poor stability, and high energy consumption, limiting their practical application due to random intercrystalline defects and inefficiencies in separation performance as manufacturing scale increases.

Method used

A composite material comprising a porous substrate with a polycrystalline metal-organic framework (MOF) layer, formed through an amorphous-to-crystalline transformation process, which allows for controlled pore size tuning from 0.5 nm to 100 nm, enabling efficient filtration and selective removal of solutes based on molecular weight.

Benefits of technology

The composite material achieves high-efficiency organic solvent filtration with tailored molecular weight selectivity and permeance, demonstrating improved stability and scalability by controlling intercrystalline channel sizes, thereby overcoming previous limitations in nanofiltration and ultrafiltration processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US2024031107_28112024_PF_FP_ABST
    Figure US2024031107_28112024_PF_FP_ABST
Patent Text Reader

Abstract

The present invention provides a composite material for organic solvent filtration, the composite material comprising a porous substrate and a polycrystalline material layer disposed on a surface of the porous substrate, wherein the polycrystalline material layer is formed from an metal-organic framework (MOF) according to formula (I) as disclosed herein, and wherein the polycrystalline material layer comprises intercrystalline channels, the intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm. The present invention also provides a method of preparing the composite material, and a method of removing a solute from an organic solvent by passing the solvent through the composite material.
Need to check novelty before this filing date? Find Prior Art

Description

[0001]COMPOSITE MATERIALS FOR ORGANIC SOLVENT FILTRATION FIELD OF INVENTION The present invention provides a composite material for organic solvent filtration and a method of preparing the composite material. The present invention also provides a method of removing a solute from an organic solvent by passing the solvent through the composite material. BACKGROUND The listing or discussion of a prior-published document in this specification should not necessarily be taken as an acknowledgement that the document is part of the state of the art or is common general knowledge. Mass transport at the sub-nanometer scale that selectively transports small molecules, such as gases, solvents, and ions, is desirable for high-efficiency separation, catalysis, chemical sensing, and energy storage and conversion. Nanoporous materials are considered the key components to construct nanochannels with specific properties that can discriminate molecules of similar sizes. Examples of those materials include one-dimensional (1D) nanotubes (e.g., carbon nanotubes and organic nanotubes), two-dimensional (2D) nanosheets (e.g., graphene and graphene oxide), and three-dimensional (3D) porous frameworks (e.g., zeolites, metal-organic frameworks (MOFs), and covalent organic frameworks (COFs)). Novel membranes synthesized by such materials exhibit ultrafast mass transport efficiency and precise separation performance, owing to their tunable pore sizes, shapes, and chemical environments. Nanochannels constructed by 1D carbon nanotubes have been shown to form tunable pores in the range of 0.8 nm to 2 nm. Experimental and theoretical studies demonstrate that 1D carbon nanotubes with diameters of 0.8 nm exhibit ultra-high water permeability and ion selectivity. Graphene oxide (GO) can form nanochannels through layer-by-layer stacking of the 2D nanosheets. Typically, the interlayer spacing between layered GO sheets in the dry state is around 0.9 nm, while such spacing increases to around 1.3 nm in the wet state. For 2D GO membranes, the interlayer spacing serves as the mass-transport pathways, which have been widely applied in molecular separations and shown precise sieving properties. Despite the research progress of these advanced nanochannels, their practical applications still face challenges concerning the lack of scalable processing strategies, poor stability, and high energy consumption. For example, the preparation of 2D nanosheets generally involves 1      complicated procedures, including laborious etching and exfoliation. Thus, these 2D nanochannels are typically limited to proof-of-concept studies of millimeter-scale flakes. Interconnected MOFs, where nanocavities are linked by sub-nanometer-sized windows, form transport pathways in three spatial dimensions. The MOF platform provides the ability to fine- tune the nanopore systems via isoreticulation by deploying metal clusters and organic ligands with a suitable length or bulkiness. To date, numerous efforts have been made to construct sub-0.5 nm gas channels, as well as sub-1 nm liquid or ion channels within polycrystalline MOF membranes. Typical examples include MIL-160 membrane fabricated by solvothermal synthesis, ZIF-8 membrane synthesized by semi-solid processing, and isoreticular face- centered cubic (fcu)-MOFs membrane prepared by electrochemical synthesis. Although polycrystalline MOF membranes exhibit an improvement in both permeability and selectivity that has the potential to overcome the trade-off limitation, developing scalable fabrication methods remains a challenging step toward their evolvement from laboratory curiosity to industrial implementation. Moreover, the random and inevitable formation of intercrystalline defects significantly reduces the overall separation performance, and this phenomenon deteriorates along with the membrane manufacturing scale. Therefore, there is a need for improved and / or alternative materials for organic solvent filtration, and methods for making such materials. SUMMARY The present invention provides a composite material for organic solvent filtration, the composite material comprising: a porous substrate; and a polycrystalline material layer disposed on a surface of the porous substrate; wherein the polycrystalline material layer is formed from a metal-organic framework (MOF) according to formula (I): X1-xYx(HCOO)3(I) wherein, X is a metal in the +3 oxidation state; Y is one or more metals in the +3 oxidation state, provided that each of the metals that represent X are different to each of the one or metals that represent Y; x is from 0 to 0.9999; and 2      wherein the polycrystalline material layer comprises intercrystalline channels, the intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm. The present invention also provides a method of preparing the composite material of the present invention, the method comprising the steps of: (a) providing a porous substrate and a colloidal solution comprising an organic solvent, water, a metal hydroxide, and a chelating agent; (b) coating a surface of the porous substrate with the colloidal solution and heating the coated porous substrate at a temperature sufficient to form an amorphous metal hydroxide gel layer or amorphous metal oxide gel layer on the surface of the porous substrate; and (c) contacting the amorphous metal hydroxide gel layer or amorphous metal oxide gel layer with a formic acid vapor at a temperature and for a period of time sufficient to form a polycrystalline material layer formed from a MOF according to formula (I), thereby providing the composite material. The present invention further provides a method of removing a solute from an organic solvent, the method comprising passing the organic solvent through the composite material of the present invention. Preferred but optional features are set out in the dependent claims. Additional aspects and embodiments of the composite material and methods of the present invention will be apparent from the following description, drawings and claims. As can be appreciated from the foregoing and following description, each and every feature described herein, and each and every combination of two or more such features, is included within the scope of the present disclosure provided that the features included in such a combination are not mutually inconsistent. In addition, any feature or combination of features may be specifically excluded from any embodiment. BRIEF DESCRIPTION OF DRAWINGS FIG.1 depicts an illustration of the amorphous-to-crystalline (ACT) process used to prepare the composite material of the present invention, and the derived sub-nm intercrystalline channels present in the composite material. (a) The sol-gel dip-coating method to form an amorphous alumina gel layer on mesoporous support and the following crystallization process triggered by formic acid vapor. (b) Intercrystalline spacing constructed by 3D aluminium formate bulk crystals to serve as sub-nm transport channels. 3      FIG.2 depicts (a-c) Membrane pore size evaluation by the NKP method. Top-view and cross- sectional SEM images of (d-f) ACT-11h membrane with sub-2 nm pores, (g-i) ACT-14h membrane with sub-1 nm pores, and (j-l) ACT-21h membrane with sub-0.6 nm pores. FIG.3 depicts (a) a sketch of the size tunability of the intercrystalline channels. (b) Separation performance of different dyes and ethanol permeance with prolonged reaction time. (c) Rejection behavior of different Al(HCOO)3membranes versus the molecular weight of various dyes: methyl orange (MO, 304 Da), crystal violet (CV, 372 Da), acid fuchsin (AF, 542 Da), Congo red (CR, 653 Da), and methyl blue (MB, 756 Da). (d) Continuous filtration performance of ACT-11h (sub-2 nm). (e) Continuous filtration performance of ACT-14h (sub-1 nm). (f) Continuous filtration performance of ACT-21h (sub-0.6 nm). (g-h) Ultraviolet-visible (UV) absorption spectra of the permeate, retentate, and feed solutions of CR and AF, respectively. Inserts are the photos of the feed (left) and permeate (right). FIG.4 depicts (a-b) an Illustration showing the separation of mixed dyes with similar sizes by a pressure-driven cell. (c-d) UV absorption spectra of the permeate, pure dye solutions, and mixed feed solutions. (e) Ethanol permeance and dye-mixture selectivity of ACT-21h (sub-0.6 nm) and ACT-14h (sub-1 nm). FIG.5 depicts (a) Analysis of pure solvent permeance through membranes with 100-nm-scale pores. (b) Analysis of solvent transport through membranes with 2-nm-scale pores. (c) Analysis of solvent transport through sub-1 nm channel. (d) Comparison of separation performance of Al(HCOO)3membranes to other materials in the literature. (e) Summary of ACT-21h’s contact angle for nonpolar, polar aprotic, and polar protic solvents. In this figure, ^^ represents solvent viscosity, ^^0indicates the total solubility parameter, and ^^^^is solvent molar diameter. FIG. 6 depicts an illustration of hydrolysis-dehydration-gelation process according to the present invention. (1)-(2) The illustration of partial hydrolysis and complete hydrolysis. (3) Dimer formed by condensation reaction. (4) Chain formed by condensation reaction. FIG. 7 depicts the production of unit membrane area per volume of precursor solution (productivity cm2mL-1). The result indicates that the ACT method can fully utilize the membrane precursor compared to other methods shown in Table 1. 4      FIG.8 depicts (a) Illustration of the crystal structure of Al(HCOO)3, viewed along the z-axis, (b) viewed along the y-axis, and (c) viewed along the x-axis. FIG.9 depicts experimental and Pawley refined PXRD patterns of Al(HCOO)3synthesized by the vapor-phase reaction. FIG.10 depicts Fourier-transform infrared (FTIR) analysis of Al(HCOO)3obtained by vapor- phase reaction under different reaction time. FIG.11 depicts X-ray photoelectron spectroscopy (XPS) data of the bare ceramic support and Al(HCOO)3thin films with different ACT reaction time, as shown by: (a) Al 2p, (b) O 1s, and (c) C 1s photoelectron lines. The chemical shifts in binding energy observed in the Al 2p and O 1s photoelectron lines can be associated with the changes in the local electrostatic energy, which is due to potential shifts caused by the dipole movements of adsorbed formic acid. C 1s photoelectron line exhibits a new peak corresponding to formic acid. FIG.12 depicts (a) Experimental and simulated PXRD patterns of Al(HCOO)3, as well as the PXRD pattern of amorphous aluminium gel. (b) SEM images of amorphous aluminium gel (left) and Al(HCOO)3powder (right). FIG. 13 depicts cross-sectional SEM images of Al(HCOO)3membranes prepared with amorphous-to-crystalline transformation method under different reaction time. FIG.14 depicts top-view SEM images of Al(HCOO)3membranes prepared by amorphous-to- crystalline transformation method under different reaction time. FIG. 15 depicts PXRD analysis of Al(HCOO)3membranes prepared by amorphous-to- crystalline transformation method under different reaction time. FIG. 16 depicts CO2uptake of amorphous aluminium dry gel and Al(HCOO)3powder synthesized with prolonged reaction time (solid, adsorption; open, desorption). FIG. 17 depicts AFM images and surface roughness analyses of Al(HCOO)3membranes prepared by amorphous-to-crystalline transformation method under different reaction time. FIG.18 depicts Energy Dispersive Spectrometry (EDS) mapping of Al(HCOO)3membranes. 5      FIG.19 depicts the effect of reaction temperatures, as indicated by (a) PXRD and (b) SEM analyses of Al(HCOO)3membranes prepared by different temperatures after comparable reaction time. FIG. 20 depicts the total energy of the amorphous aluminium gel (assuming complete hydrolysis) and the crystalline Al(HCOO)3calculated by DFT calculations. FIG.21 depicts N2, CO2, and CH4adsorption isotherms of Al(HCOO)3at 298 K. Al(HCOO)3has a high uptake for CO2while extremely low uptakes for N2and CH4. FIG.22 depicts N2sorption isotherm of Al(HCOO)3at 77 K. FIG.23 depicts methanol vapor sorption isotherm of Al(HCOO)3at 298 K. The low adsorption amount indicates that the uptake for methanol vapor is mainly due to surface adsorption. FIG.24 depicts Arrhenius temperature dependence of H2, CO2, CH4, and C3H6permeance for (a) ACT-11h membrane (sub-2 nm), (b) ACT-14h membrane (sub-1 nm), and (c) ACT-21h membrane (sub-0.6 nm). FIG. 25 depicts N2permeation tests for (a) ACT-11h membrane (sub-2 nm), (b) ACT-14h membrane (sub-1 nm), and (c) ACT-21h membrane (sub-0.6 nm) at different transmembrane pressures. FIG.26 depicts gas separation test using binary equimolar mixtures for ACT-21h membranes. The results imply that the membrane channel size approaches the size of permeating molecules. FIG.27 depicts R parameters of ACT-11h membrane (sub-2 nm), ACT-14h membrane (sub- 1 nm), and ACT-21h membrane (sub-0.6 nm) as a function of positron incident energy. FIG. 28 depicts (a) Fluorescence confocal microscopy (FCOM) analysis of Al(HCOO)3membranes. The lighter areas represent the fluorescence signal of pyrene adsorbed on the channel walls (for example, the area within the dot lines represents the grain boundary). (b) Top-view SEM images of the above membrane samples. 6      FIG.29 depicts an illustration of the fluorescent signal of the adsorbed fluorescent molecules on channel walls, indicating the presence of intercrystalline transport pathways. FIG.30 depicts dye rejection performance (50 ppm MB in ethanol) of Al(HCOO)3membranes after exposing the membranes to different organic solvents for 21 days. FIG.31 depicts a calibration line of ICP-OES for Al3+. After filtration tests with methanol and DMF for 7 days, the concentration of Al3+in the permeate was extremely low. FIG.32 depicts PXRD patterns of Al(HCOO)3membranes after exposing the membranes to different organic solvents for 21 days. FIG. 33 depicts (a) Ethanol permeance versus filtration time. (b) Ethanol flux and ethanol permeance versus applied pressure. (c) Dye rejection performance (50 ppm CR in ethanol) and ethanol flux versus applied pressure. FIG.34 depicts thermogravimetric test of Al(HCOO)3powder in air and N2up to 800 ℃. FIG. 35 depicts water stability tests of Al(HCOO)3membranes. (a) PXRD patterns of Al(HCOO)3membranes soaked in ethanol solutions with different volume percentages of water for three weeks. (b) Long-term filtration test using 50 ppm MB ethanol solution with 10% water. FIG.36 depicts chemical structures and sizes of dyes optimized by Gaussian-DFT. FIG.37 depicts zeta potential of Al(HCOO)3membranes under different pH values. FIG. 38 depicts UV absorption spectra of the feed and permeate for ACT-11h membrane. Level 1 refers to the separation capacity with MWCO of around 650 Da. The inserted figures represent the feed (left) and the permeate (right). FIG. 39 depicts UV absorption spectra of the feed and permeate for ACT-14h membrane. Level 2 refers to the separation capacity with MWCO of around 500 Da. The inserted figures represent the feed (left) and the permeate (right). 7      FIG. 40 depicts UV absorption spectra of the feed and permeate for ACT-21h membrane. Level 3 refers to the separation capacity with MWCO lower than 400 Da. The inserted figures represent the feed (left) and the permeate (right). FIG. 41 depicts long-term filtration test with 500 ppm MB ethanol solution. The test was conducted with a dead-end permeation cell, with an applied pressure of 5 bar and a stirring speed of 500 rpm. FIG.42 depicts the accumulation of rejected dyes on membrane surface. FIG. 43 depicts UV-Vis spectrum of the initial dye solutions and the dye solutions after adsorption tests (14 days). The adsorption tests were conducted by soaking 20 mg Al(HCOO)3crystals in the corresponding dye solutions (20 mL) for 14 days. FIG. 44 depicts (a) The illustration of the tubular membrane module at lab scale and the crossflow operation mode, and performance summary of the tubular membrane operated under 10 bars. (b) The illustration of dead-end cells. FIG. 45 depicts (a) Photos of the ceramic tube substrates. (b) The top-view (top two) and cross-sectional (bottom two) SEM images of the tube membrane substrates. FIG.46 depicts permeance versus viscosity. As for Al(HCOO)3membranes with 100-nm-scale large pores, liquid flow follows viscous flow behavior. By contrast, deviations from this viscous flow behavior were observed for Al(HCOO)3membranes with 2-nm-scale and sub-1 nm pores. FIG. 47 depicts analysis of solvent transport through membranes with 100-nm-scale large pores. (a) Plot and fitting of solvent permeance against solvent viscosity. (b) Plot and fitting of solvent permeance against the combined solvent property (solubility parameter due to dipole forces and viscosity). (c) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter and viscosity). (d) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter, viscosity, and molar diameter). According to (a), solvent flow through 100-nm-scale large pores follows the equation: ^^^^^ି^^^LMH^ ∝ ^^ு^^^ି^∆ ^^ (Hagen-Poiseuille equation - Rogers et al., ACS nano 2013, 7, 725-731), (Eq. S20) 8      ^^ is solvent flux, ^^ு^is constant, which may be related to pore size, thickness, tortuosity, and porosity, ∆ ^^ is transmembrane pressure. FIG. 48 depicts permeance divided by total solubility parameter versus viscosity. As for Al(HCOO)3membranes with 2-nm-scale pores, the above deviation (FIG.46) is mainly due to the interaction between solvent molecules and membrane channel walls. By contrast, as for membranes with sub-1 nm pores, extra factors may contribute to the deviation. FIG.49 depicts analysis of solvent transport through membranes with 2-nm-scale pores. (a) Plot and fitting of solvent permeance against solvent viscosity. (b) Plot and fitting of solvent permeance against the combined solvent property (solubility parameter due to dipole forces and viscosity). (c) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter and viscosity). (d) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter, viscosity, and molar diameter). According to (c), solvent flow through 2-nm-pore follows the equation: ^^ଶି^^^LMH^∝ ^^ு^^^^^^ି^∆ ^^ (Eq. S21) FIG.50 depicts permeance divided by total solubility parameter and molar diameter versus viscosity. As for Al(HCOO)3membranes with sub-1 nm pores, the above deviation (FIG.48) is mainly because of the interaction between solvent molecules and membrane channel walls, as well as the effect of molecular size of solvents due to the enhanced confinement. FIG.51 depicts analysis of solvent transport through membranes with sub-1 nm pore. (a) Plot and fitting of solvent permeance against solvent viscosity. (b) Plot and fitting of solvent permeance against the combined solvent property (solubility parameter due to dipole forces and viscosity). (c) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter and viscosity). (d) Plot and fitting of solvent permeance against the combined solvent property (total solubility parameter, viscosity, and molar diameter). According to FIG.51d, solvent flow through sub-1-nm-pore follows the equation: ^^ௌ௨^ି^^^^^^ ^^ ^^^∝ ^^ு^^^^^^ି^^^^ଶ∆ ^^ (Eq. S22) FIG.52 depicts the optimized intercrystalline channels with (a) 2-nm size and (b) 1-nm size for simulating the solvent transport in host channel during the MD simulation. FIG.53 depicts illustration of hydrogen bonds. (a) The donor-acceptor interactions between DMF and Al(HCOO)3channel. (b) The donor-acceptor interactions between isopropanol and Al(HCOO)3channel. 9      FIG.54 depicts simulated radial distribution function (RDF) of solvents in 2-nm host channel (FIG.52a). RDF means the ratio of the local atomic density of atom ^^, which is r away from target atom ^^ and the average density of ^^ in the system. By analyzing the partial radial distribution function g(r), the inter-atomic distance and other important information are available (Zahn et al., Physical Chemistry Chemical Physics, 2011, 13, 15083-15093). Here, the solvent RDF represents the average distance of solvent molecules to membrane channel walls, which indicates the binding strength between solvent and membrane. Isopropanol, with the largest molecular size and polarity, showed the shortest distance due to the hydrogen bonding and increased van der Waals forces, while o-xylene showed a shorter distance compared to heptane and hexane. FIG.55 depicts simulated solvent trajectory in 1-nm host channel (FIG.52(b)) under 298 K with fixed loading mode. The simulated trajectory of (a) hexane, (b) isopropanol, (c) o-xylene, and (d) heptane. The high density of solvent molecules located near the channel walls indicates the strong interaction between solvent molecules and the channel wall under enhanced confinement. Such strong interactive forces between solvent molecules and membrane lead to the correction factor ^^^^in the modified Hagen-Poiseuille model. FIG. 56 depicts simulated mean square displacement (MSD) plots of diffusion of hexane, heptane, isopropanol, and o-xylene in 1-nm host channel (FIG.52(b)). FIG.57 depicts contact angle of different solvents. DESCRIPTION The present invention provides composite materials that are surprisingly effective for the filtration of organic solvents, for example ultrafiltration and nanofiltration of organic solvents. As described herein, the composite material of the present invention comprises: a porous substrate; and a polycrystalline material layer disposed on a surface of the porous substrate; wherein the polycrystalline material layer is formed from a metal-organic framework (MOF) according to formula (I): X1-xYx(HCOO)3(I) 10      wherein, X is a metal in the +3 oxidation state; Y is one or more metals in the +3 oxidation state, provided that each of the metals that represent X are different to each of the one or metals that represent Y; x is from 0 to 0.9999; and wherein the polycrystalline material layer comprises intercrystalline channels, the intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm. The present inventors have surprisingly found that the intercrystalline channels of the polycrystalline material layer act as mass transport pathways through the polycrystalline material layer. The intercrystalline channels allow for organic solvents to pass through the composite materials of the present invention, whilst selectively removing solutes present in the organic solvent based on molecular weight. The present inventors have also demonstrated that composite materials of the present invention can be prepared by an amorphous-to- crystalline transformation (ACT) process which allows control over the pore size of the intercrystalline channels present in the composite material. The ACT process allows for the pore size of the intercrystalline channels to be tuned from about 100 nm to below 1 nm such that the composite material displays a tailored molecular weight selectivity and organic solvent permanence. Thus, the present invention also provides a method for providing the composite material of the present invention comprising a polycrystalline material layer that comprises intercrystalline channels having specific pore sizes, in particular pore sizes in the nanometer and subnanometer range. In embodiments herein, the word “comprising” may be interpreted as requiring the features mentioned, but not limiting the presence of other features. Alternatively, the word “comprising” may also relate to the situation where only the components / features listed are intended to be present (e.g. the word “comprising” may be replaced by the phrases “consists of” or “consists essentially of”). It is explicitly contemplated that both the broader and narrower interpretations can be applied to all aspects and embodiments of the present invention. In other words, the word “comprising” and synonyms thereof may be replaced by the phrase “consisting of” or the phrase “consists essentially of” or synonyms thereof and vice versa. The phrase, “consists essentially of” and its pseudonyms may be interpreted herein to refer to a material where minor impurities may be present. For example, the material may be greater than or equal to 90% pure, such as greater than 95% pure, such as greater than 97% pure, 11      such as greater than 99% pure, such as greater than 99.9% pure, such as greater than 99.99% pure, such as greater than 99.999% pure, such as 100% pure. For the avoidance of doubt, it is explicitly contemplated that where a number of numerical ranges related to the same feature are cited herein, that the end points for each range are intended to be combined in any order to provide further contemplated (and implicitly disclosed) ranges. The term “about” as used herein can allow for a degree of variability in a value or range, for example, within 10%, within 5%, within 1%, within 0.5%, within 0.1%, within 0.05%, within 0.01%, within 0.005%, or within 0.001% of a stated value or of a stated limit of a range, and includes the exact stated value or range. As used herein, the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. In formula (I), X may be any suitable metal in the +3 oxidation state. Examples of suitable metals include, but are not limited to, Al, Fe, Mn, Y, Sc, Ga, In, Ti, V, Cr, Co, Tl, Bi, Ru, Os, Re, Ir, Rh, Ce, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu. In formula (I), the value of x may range from 0 to 0.9999. For example, x may be from 0 to 0.99, such as, from 0.25 to 0.85. Particular point values of x that may be mentioned herein include 0.25, 0.50, 0.75 and 0.85. As will be appreciated, this approach essentially results in X being doped by Y (or vice versa, depending on the relative amounts of X and Y). For the avoidance of doubt, it is explicitly contemplated that where several numerical ranges related to the same feature are cited herein, that the end points for each range are intended to be combined in any order to provide further contemplated (and implicitly disclosed) ranges. Thus, for the values of x mentioned herein, there is disclosed the following ranges for x: from 0 to 0.25, from 0 to 0.50, from 0 to 0.75, from 0 to 0.85, from 0 to 0.99, from 0 to 0.9999; from 0.25 to 0.50, from 0.25 to 0.75, from 0.25 to 0.85, from 0.25 to 0.99, from 0.25 to 0.9999; from 0.50 to 0.75, from 0.50 to 0.85, from 0.50 to 0.99, from 0.50 to 0.9999; from 0.75 to 0.85, from 0.75 to 0.99, from 0.75 to 0.9999; from 0.85 to 0.99, from 0.85 to 0.9999; and from 0.99 to 0.9999. As noted herein, X and Y may be selected from the same set of metals. However, it is explicitly contemplated that X is not the same metal as Y. 12      As noted herein, Y may be one or more metals (e.g., two metals). When Y is one or more metals, any suitable proportion of said metals may be used. For example, when there are two metals, Y may be represented by Y’1-yY’’y, where the values of y may be the same as the values of x discussed above. In certain embodiments, Y may be a single metal (that differs from the metal of X). In certain embodiments, X may be selected from Al, Fe, Mn, Y, Sc, Ga, In, Ti, V, Cr, Co, Tl, Bi, Ru, Os, Re, Ir, Rh, Ce, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu. For example, X may be Al, Fe or V. In certain embodiments, X is Al. In certain other embodiments, Y may be selected from Al, Fe, Mn, Y, Sc, Ga, In, Ti, V, Cr, Co, Tl, Bi, Ru, Os, Re, Ir, Rh, Ce, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu, provided that Y is not the same metal as X. For example, when present, Y may be selected from: (a) one or more of the group consisting of Al, Tl, Fe, Ga, In, Mn, V and Cr; (b) one or more of the group consisting of Fe, Ga, In, Mn, V and Cr; (c) from the group consisting of Al, Tl, Fe, Ga, In, Mn, V and Cr; or (d) from the group consisting of Fe, Ga, In, Mn, V and Cr. As will be appreciated, (a) and (b) above allow for the presence of 2 or more (e.g., 2, 3 or 4 metals as Y), while (c) and (d) only allow for the presence of one metal as Y. In certain embodiments, Y, when present, is Fe or V. In certain embodiments, formula (I) may be one or more (e.g., one) selected from: (a) Al(HCOO)3; (b)3; (c) (d)3; (e)3; (f) V (g) V0.3Al0.7(HCOO)3. For example, formula (I) may be one or more (e.g., one) selected from: (a) Al(HCOO)3; (b) Al0.75Fe0.25(HCOO)3; (c) Al0.5Fe0.5(HCOO)3; (d) Al0.25Fe0.75(HCOO)3; and (e) Al0.15Fe0.85(HCOO)3. 13      In certain embodiments, formula (I) is Al(HCOO)3(i.e. aluminium formate). Accordingly, the present invention provides a composite material comprising: a porous substrate; and a polycrystalline material layer disposed on a surface of the porous substrate; wherein the polycrystalline material layer is formed from an Al(HCOO)3metal- organic framework (MOF); and wherein the polycrystalline material layer comprises intercrystalline channels, the intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm. In the present invention, the polycrystalline material layer of the composite material of the invention comprises intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm. In certain embodiments, the polycrystalline material layer comprises intercrystalline channels with an average pore size of from about 10 nm to about 100 nm. When the polycrystalline material layer of the composite material of the invention comprises intercrystalline channels with an average pore size of about 10 nm to about 100 nm, the composite material finds particular use in the ultrafiltration of organic solvents. For example, the composite material may find use in the removal of small particles, including viruses, large proteins, and certain polymers from a solvent. In certain other embodiments, the intercrystalline channels may have an average pore size of from about 0.5 to about 10 nm. For example, from about 0.5 to about 5 nm, about 0.5 to about 2 nm. When the polycrystalline material layer of the composite material of the invention comprises intercrystalline channels with an average pore size of about 0.5 to about 10 nm, the composite material finds particular use in the nanofiltration of organic solvents. For example, the composite material may find use in the removal of small organic molecules, metals and other molecules that may be present in a solvent. In certain exemplary embodiments, the polycrystalline material layer of the composite material comprises intercrystalline channels with an average pore size of less than or equal to about 2 nm, for example, from about 1 nm to about 2 nm, for example about 1.5 nm. Or, for example, the polycrystalline material layer may comprise intercrystalline channels having an average pore size of less than or equal to about 1 nm, for example, from about 0.6 nm to about 1 nm, for example about 0.9 nm. Or, for example, the polycrystalline material layer may comprise intercrystalline channels having an average pore size of less than or equal to about 0.6 nm, for example, from about 0.5 nm to about 0.6 nm, for example about 0.55 nm. In embodiments wherein the polycrystalline material layer comprises intercrystalline channels with an average 14      pore size of less than or equal to about 2 nm, it has been found that the composite material displays a molecular weight cutoff (MWCOs) ranging from about 300 Da to about 650 Da. When used herein, the terms “molecular weight cut off” and “MWCO” refer to the lowest molecular weight (in Daltons) at which greater than about 90% of a solute is excluded by (i.e. prevented from passing through) the composite material. For example, when a composite material has a molecular weight cut off (MWCO) of 300 Da, it is capable of excluding greater than 90% of the solutes in a solvent that have molecular weight of 300 Da or more. As demonstrated in the Examples disclosed herein, the composite membranes comprising intercrystalline channels with an average pore size of less than or equal to about 2 nm display MWCOs of from about 300 Da to about 800 Da (e.g., about 650 Da, about 500 Da, about 400 Da, or less). As used herein, the term “pore size” refers to the size of a channel present in the composite material. For example, in the composite material of the invention the average pore size of the intercrystalline channels present in the polycrystalline material layer may be less than or equal to about 2 nm. The pore size of the channels may be determined by methods known in the art such as those based on a modified gas-translation model. For example, the pore size of the channels may be determined using the normalized Knudsen-based permeance (NKP) method, as described in Kanezashi et al., (J Phys Chem C.2014;118:20326–20331) and Yoshioka et al., (AIChE J.2013;59:2179–2194). The terms “permeance”, “organic solvent permanence”, and similar terms used herein, refer to the volume of a liquid (e.g. an organic solvent) that can flow through the composite material of the present invention per unit area of composite material, per unit time, and per unit of pressure. Typically, the permeance of the composite material of the present invention is expressed in units of L m-2h-1bar-1, unless otherwise stated. The composite materials of the present invention typically display an ethanol permeance of from about 0.5 L m-2h-1bar-1to about 25 m-2h-1bar-1. For the avoidance of doubt, the composite material of the present invention may be referred to herein generally as a “membrane”, a “MOF membrane”, or in embodiments wherein the MOF is Al(HCOO)3, a “Al(HCOO)3membrane”. As demonstrated in the Examples disclosed herein, when the polycrystalline material layer of the composite material of the invention comprises intercrystalline channels with an average pore size of from about 1 nm to about 2 nm (e.g. about 1.5 nm), the composite material displays a MWCO of about 650 Da and an ethanol permeance of about 12 L m-2h-1bar-1to 15      about 17 L m-2h-1bar-1(e.g. about 14.6 L m-2h-1bar-1). As further demonstrated, when the polycrystalline material layer comprises the intercrystalline channels with an average pore size of from about 0.6 nm to about 1 nm (e.g. about 0.9 nm), the composite material displays a MWCO of about 540 Da and an ethanol permeance of about 6 L m-2h-1bar-1to about 6 L m-2h-1bar-1(e.g. about 4.1 L m-2h-1bar-1). As further demonstrated, when the polycrystalline material layer comprises intercrystalline channels of the polycrystalline material layer have an average pore size of from about 0.6 nm to about 1 nm (e.g. about 0.9 nm), the composite material displays a MWCO of about 540 Da and an ethanol permeance of about 6 L m-2h-1bar-1to about 6 L m-2h-1bar-1(e.g. about 4.1 L m-2h-1bar-1). In certain embodiments, when the polycrystalline material layer is formed from Al(HCOO)3, the polycrystalline material layer may have a surface area of from about from about 10 m2.g-1to about 50 m2.g-1. For example, the polycrystalline material layer may have a surface area of about 15 m2.g-1to about 18 m2.g-1(e.g., about 16.8 m2.g-1). In certain embodiments, the polycrystalline material layer has an average thickness of from about 2 µm to about 10 µm as determined by scanning electron microscope (SEM). For example, the polycrystalline material layer may have a thickness of about 3 µm to about 6 µm (e.g., about 3.1, 5.1, and 5.9 µm). In certain embodiments, when the polycrystalline material layer is formed from Al(HCOO)3, the polycrystalline material layer has a surface average roughness of from about 150 nm to about 350 nm (e.g., about 179 nm and 343 nm). As demonstrated in the Examples disclosed herein, the surface average roughness of the the polycrystalline material layer increases with increasing transformation time primarily due to the progressive formation of large crystal faces. The surface area of materials disclosed herein is determined from nitrogen adsorption measurements at 77 K using the Brunauer-Emmett-Teller (BET) method. See J. Am. Chem. Soc.1938, 60, 309-331, which is incorporated herein by reference. The pore volume and pore size of the material may also be calculated from N2adsorption isotherms measured at 77 K, as determined using BET. As used herein, the term “surface average roughness” refers to the average of surface heights and depths across a given surface. Methods for measuring surface average roughness of a surface are known in the art, and include for example, Atomic Force Microscopy (AFM). As mentioned hereinabove, in certain embodiments, formula (I) is Al(HCOO)3(i.e. aluminium formate). Aluminium formate, Al(HCOO)3, an earth-abundant, inexpensive and highly porous metal-organic framework (MOF), can be obtained from a non-porous precursor. Al(HCOO)3is 16      thermally stable to around 250 °C and is also stable in the presence of water, acids, alkalis and organic solvents. Al(HCOO)3has excellent and highly selective adsorption capacity for carbon dioxide gas in the presence of nitrogen gas and water vapour, at and above room temperature. Al(HCOO)3is very easy to make from readily available starting materials such as aluminium hydroxide,  aluminium oxide, aluminium alkoxides, and formic acid. Al(HCOO)3is very inexpensive compared to other metal-organic frameworks or solid sorbents. In certain embodiments, the polycrystalline material layer is formed from Al(HCOO)3crystals. In such embodiments, the Al(HCOO)3crystals may be characterized by powder X-ray diffraction (XRD). Thus, in certain embodiments wherein the polycrystalline material layer is formed from Al(HCOO)3MOF crystals, the Al(HCOO)3crystals may have a powder X-ray diffractogram with characteristic peaks at 2θ of about 15.6° and about 21.9 °, for example 15.6 ° ± 0.2° and 21.9° ± 0.2°. In certain embodiments, the powder X-ray diffractogram of the Al(HCOO)3crystals may also include characteristic peaks at 2θ of about 24.5°, about 26.9°, about 29.1°, about 31.2°, about 33.1°, about 35.0°, about 36.7°, about 38.4°, about 40.1°, for example 24.5° ± 0.2°, 26.9° ± 0.2°, 29.1° ± 0.2°, 31.2° ± 0.2°, 33.1° ± 0.2°, 35.0° ± 0.2°, 36.7° ± 0.2°, 38.4° ± 0.2°, and 40.1° ± 0.2°. The porous substrate may be selected from the group consisting of a ceramic (e.g. alumina), a carbon cloth, a metal, a metal oxide, and a polymer. Examples of suitable porous polymer substrates include, but not limited to, a polyacrylonitrile (PAN) porous polymer substrate and polyethersulfone (PES) porous polymer substrate. The porous substrate can be in any suitable form, which includes, but is not limited to, tubes, meshes, sheets and hollow fibers (e.g. porous alumina ceramic hollow fibers), plus other forms that can be obtained by the folding of these primary forms. Examples of porous substrate in the form of sheets include, but are not limited to, carbon film / cloth. The flexibility of carbon films / cloths as porous substrates can provide good mechanical properties to the resultant composite material. Meshes may include, but are not limited to, metal meshes and metal oxide meshes. Hollow fiber structures that may be mentioned herein include, but are not limited to ceramics (e.g. alumina) and zirconia hollow fiber. The porous substrate may be an Al2O3substrate, for example an α-Al2O3substrate. An example of a suitable porous α-Al2O3substrate is porous asymmetric α-Al2O3plates with a maximum pore size of about 100 nm, a diameter (for the substrate) of about 18 mm, and a 17      thickness of about 1 mm. Suitable porous Al2O3substrates may be prepared using techniques known in the art, or purchased from commercial supplies, such as the Fraunhofer Institut für Keramische Technologien and Systeme (IKTS), Germany. As will be appreciated, materials with different pore sizes may also be used, for example the maximum pore size may be from 30 nm to 200 nm. In certain embodiments, the composite material of the invention comprises a porous substrate having a first surface and second surface, wherein the polycrystalline material layer is disposed on the first surface, and the second surface is substantially free from the polycrystalline material layer. It has been found that mass transfer resistance through the composite material can be reduced by depositing the polycrystalline material layer on only the first surface of the porous substrate. Methods of manufacture: As disclosed herein, the present inventors have demonstrated that composite materials of the present invention can be prepared by an amorphous-to-crystalline transformation (ACT) process which allows control over the pore size of the intercrystalline channels formed in the polycrystalline material layer by controlling the transformation time. This allows for the pore size to be tuned such that the composite material displays a tailored molecular weight selectivity and organic solvent permanence. Accordingly, the present invention provides a method of preparing the composite material of the present invention. The method of preparing the composite material of the present invention comprises the steps of: (a) providing a porous substrate and a colloidal solution comprising an organic solvent, water, a metal hydroxide, and a chelating agent; (b) coating a surface of the porous substrate with the colloidal solution and heating the coated porous substrate at a temperature sufficient to form an amorphous metal hydroxide gel layer or amorphous metal oxide gel layer on the surface of the porous substrate; and (c) contacting the amorphous metal hydroxide gel layer or amorphous metal oxide gel layer with formic acid vapor at a temperature and for a period of time sufficient to form a polycrystalline material layer formed from a MOF according to formula (I), thereby providing the composite material. The organic solvent may be any organic solvent that is suitable for forming the colloidal solution according to the present invention. For example, the organic solvent may be an 18      alcohol, such as methanol, ethanol, or isopropanol. In certain embodiments, the organic solvent is isopropanol. The chelating agent may be any chelating agent that is suitable for peptizing the metal hydroxide to form the colloidal solution according to the present invention. For example, the chelating agent may be ethyl acetoacetate or acetyl acetone. In certain embodiments, the chelating agent is ethyl acetoacetate. The metal hydroxide may be any suitable metal hydroxide selected based on the identity of X (and Y when present) in the MOF according to formula (I). That is to say that the metal of the metal hydroxide corresponds to X in formula (I), and Y in formula (I) when present. For example, when X in formula (I) is Al, the metal hydroxide may be aluminum hydroxide. Or, when X in formula (I) is Fe, the metal hydroxide may be iron hydroxide. Or, when X in formula (I) is V, the metal hydroxide may be vanadium hydroxide. In certain embodiments, step a) of the present invention may further comprise hydrolyzing a metal hydroxide precursor in the presence of water to form the metal hydroxide, followed by peptizing the metal hydroxide in the presence of a chelating agent to form the colloidal solution. The metal hydroxide precursor may be any suitable precursor that is capable of being hydrolyzed in the presence of water to form the metal hydroxide. For example, the metal hydroxide precursor may be a metal alkoxide. For example, when the metal hydroxide precursor is an aluminum hydroxide precursor, the metal hydroxide precursor may be an aluminium alkoxide, such as aluminum-tri-sec-butoxide (Al(O-sec-Bu)₃), aluminum isopropoxide (Al(O-i-Pr)₃), aluminum ethoxide (Al(OEt)₃), and aluminum methoxide (Al(OMe)₃). In certain embodiments, the metal hydroxide precursor is aluminum-tri-sec- butoxide (Al(O-sec-Bu)₃). When the metal hydroxide precursor is an iron hydroxide precursor, the metal hydroxide precursor may be an iron alkoxide, for example, iron butoxide. When the metal hydroxide precursor is a vanadium hydroxide precursor, the metal hydroxide precursor may be a vanadium alkoxide, for example, vanadyl isopropoxide. Typically, step (b) of the method of the invention comprises coating a surface of the porous substrate with the colloidal solution, and heating the porous substrate at temperature of from about 160 °C to about 200 °C for about 5 hour to about 7 hours, and step (c) of the method of the invention comprises contacting the porous substrate following step (b) with formic acid vapor at temperature of from about 160 °C to 180 °C for about 10 hour to about 21 hours. As mentioned above, the amorphous-to-crystalline (ACT) transformation process can be controlled to tune the pore size of the intercrystalline channels formed in the polycrystalline 19      material layer. This can be achieved by controlling the transformation time of the ACT process. The “transformation time” is to be understood to refer to step c) of the method in which the porous substrate is contacted with formic acid vapor. In certain embodiments, the method of preparing the composite material of the present invention comprises the steps of: (a) providing a porous substrate and a colloidal solution comprising an organic solvent, water, an aluminum hydroxide, and a chelating agent; (b) coating a surface of the porous substrate with the colloidal solution and heating the coated porous substrate at temperature sufficient to form an amorphous aluminium hydroxide gel layer or amorphous aluminium oxide gel layer on the surface of the porous substrate; and (c) contacting the amorphous aluminium hydroxide gel layer or amorphous aluminium oxide gel layer with formic acid vapor at a temperature of from about 160 °C to about 180 °C for a period of time sufficient to form a polycrystalline material layer formed from Al(HCOO)3, thereby providing the composite material. For the avoidance of doubt, the terms “amorphous aluminium hydroxide gel layer” and “amorphous aluminium oxide gel layer” may also be referred to herein generally as “aluminium gel”, “alumina gel”, or similar terms. Similarly, the terms “metal hydroxide precursor” and “aluminium hydroxide precursor” may be referred to herein generally as “metal precursor” or “aluminium precursor”. As used herein, the term “aluminum hydroxide” includes aluminum monohydroxide and aluminum trihydroxide. In embodiments wherein the polycrystalline material layer is to be formed from Al(HCOO)3, the organic solvent, chelating agent and water are reacted with the aluminium hydroxide precursor (e.g. aluminum-tri-sec-butoxide) at a molar ratio of from about 18:1:2 to about 22:1:5 relative to the aluminium hydroxide precursor (e.g. about 20:1:4 relative to the aluminium hydroxide precursor). For the avoidance of doubt, when the organic solvent, chelating agent and water are reacted with the aluminium hydroxide precursor at a molar ratio of from about 18:1:2 to about 22:1:5 relative to the aluminium hydroxide precursor, this is to be understood to mean that the organic solvent to aluminium hydroxide precursor molar ratio is 18:1 to 22:1, the chelating agent to aluminium hydroxide precursor molar ratio is 1:1, and water to aluminium hydroxide precursor molar ratio is 2:1 to 5:1. For example, the organic solvent to aluminium hydroxide precursor molar ratio may be 20:1, the chelating agent to aluminium 20      hydroxide precursor molar ratio may be 1:1, and water to aluminium hydroxide precursor molar ratio may be 4:1. In embodiments wherein the polycrystalline material layer is to be formed from Al(HCOO)3, step (b) of the method of the invention may comprise coating a surface of the porous substrate with the colloidal solution, and heating the porous substrate at temperature of from about 160 °C to about 200 °C for about 5 hour to about 7 hours. Step (b) of the method of manufacture results in the coating of a surface of the porous substrate with film or layer of an amorphous aluminium hydroxide gel or amorphous aluminium oxide gel on a surface of the porous substrate by a sol-gel process. In embodiments wherein the polycrystalline material layer is to be formed from Al(HCOO)3, step (c) of the method of the invention may comprise contacting the porous substrate following step (b) with formic acid vapor at temperature of from about 160 °C to 180 °C for about 10 hour to about 21 hours. This process results in the amorphous aluminium hydroxide gel or amorphous aluminium oxide gel on the surface of the porous substrate being transformed into a polycrystalline material layer of Al(HCOO)3. Step (c) of the method of the invention can be controlled such that intercrystalline channels form between the crystals of the MOF. Surprisingly, the present inventors have found that the pore size of the intercrystalline channels can be easily controlled to have an average pore size of from about 100 nm to below 1 nm by altering the transformation time of the process (i.e. the period of time in step (c) of the method). In particular, the present inventors have found that the pore size of the intercrystalline channels decreases as the transformation time is increased. As demonstrated in the Examples disclosed herein, the average pore size of the intercrystalline channels can be tuned to ≥2 nm (e.g. from about 1 nm to about 2 nm), ≥1 nm (e.g. from about 0.6 nm to about 1 nm), and ≥0.6 nm (e.g. from about 0.5 nm to about 0.6 nm), simply by extending the transformation time from about 11 h to about 21 h. As mentioned above, the intercrystalline channels present in the polycrystalline material layer of the composite material allow for organic solvents to pass through the composite material whilst selectively excluding, and therefore removing, solutes present in the organic solvent based on their molecular weight. Therefore, the present invention also provides a method of removing a solute from an organic solvent, the method comprising passing the solvent through the composite material as disclosed herein. It follows that the present invention also provides the use of the composite material of the present invention for the filtration of an organic solvent. 21      In certain embodiments, when the intercrystalline channels of the polycrystalline material layer have an average pore size of from about 1 nm to about 2 nm, the method of removing a solute from an organic solvent comprises passing an organic solvent comprising a solute with a molecular weight of greater than about 650 Da through the composite material. In other embodiments, when the intercrystalline channels of the polycrystalline material layer have an average pore size of from about 0.6 nm to about 1 nm, the method comprises passing an organic solvent comprising a solute with a molecular weight of greater than about 540 Da through the composite material. In further embodiments, when the intercrystalline channels of the polycrystalline material layer have an average pore size of from about 0.5 nm to about 0.6 nm, the method comprises passing an organic solvent comprising a solute with a molecular weight of greater than about 300 Da. The organic solvent comprising the solute may be selected from the group consisting of a nonpolar solvent (e.g. hexane, toluene, decane, heptane, o-xylene), an aprotic solvent (e.g. acetone, dimethylformamide (DMF), and N-methyl-2-pyrrolidone (NMP)), and a protic solvents (e.g. ethanol, methanol, isopropyl alcohol). For example, the organic solvent comprising the solute may be selected from the group consisting of hexane, toluene, decane, heptane, o- xylene, acetone, DMF, NMP, ethanol, methanol, isopropyl alcohol, and a mixture of two or more thereof. Preferably, the organic solvent comprising the solute comprises less than 20 vol% water. The following statements describe further aspects and embodiments of the disclosure provided herein: 1. A process for producing a crystalline ceramic nanoporous membrane, comprising the steps of: i) hydrolyzing an aluminium precursor in the presence of water to form a metal hydroxide compound; ii) subsequently peptizing the metal hydroxide compound in the presence of a chelating agent to form a clear sol; iii) providing a ceramic substrate derived from alpha-aluminium oxide; iv) forming an amorphous aluminium gel layer via dehydration-gelation process on a surface of the alpha-aluminium oxide ceramic substrate by dip coating thereof in the same clear sol; v) heating the coated alpha-aluminium oxide ceramic substrate; and vi) subsequently treating the coated alpha-aluminium oxide ceramic substrate by exposing thereof to a nucleation agent of formic acid vapour to form the crystalline 22      ceramic nanoporous membrane. 2. The process according to item 1, wherein the aluminium precursor is aluminium tri-sec butoxide. 3. The process according to item 1, wherein the metal hydroxide compound is aluminium monohydroxide or aluminium trihydroxide. 4. The process according to item 1, wherein the organic solvent is isopropanol. 5. The process according to item 1, wherein the chelating agent is ethyl acetoacetate. 6. The process according any one of the preceding items, wherein the organic solvent, chelating agent and water are reacted with the aluminium precursor at a molar ratio in the range of 18:1:2 to 22:1:5 relative to the aluminium precursor. 7. The process according to any one of the preceding items, wherein the step of heating the coated ceramic substrate is performed at a temperature ranging from 160 ℃ to 200 ℃ for 5 to 7 hours. 8. The process according to any one of the preceding items, wherein the step of treating the coated ceramic substrate is performed by exposing thereof to a nucleation agent of formic acid vapour at a temperature ranging from 160 ℃ to 180 ℃ for 10 to 21 hours. 9. A crystalline ceramic nanoporous membrane comprising: an alpha-aluminium oxide ceramic substrate formed from a clear sol comprising a hydrolysis and condensation reaction product of an aluminium precursor and an organic solvent in the presence of water and a chelating agent; and a crystalline aluminium formate film deposited on a surface of the alpha-aluminium oxide ceramic substrate, wherein the crystalline aluminium formate ceramic film is formed from a reaction product of an amorphous aluminum gel derived from the same clear sol with a nucleation agent. 10. The membrane according to item 9, wherein the aluminium precursor is aluminium tri-sec butoxide. 11. The membrane according to item 9, wherein the organic solvent is isopropanol. 23      12. The membrane according to item 9, wherein the chelating agent is ethyl acetoacetate. 13. The membrane according to item 9 to 12, wherein the organic solvent, chelating agent and water are present in a molar ratio in the range of 18:1:2 to 22:1:5 relative to the aluminium precursor. 14. The membrane according to item 9, wherein the nucleation agent is formic acid vapour. 15. The membrane according to any one of items 9 to 14 has a pore size in the range of 0.5 nm to 100 nm. 16. The membrane according to any one of items 9 to 14 has a molecular weight cut-off (MWCO) in the range of 300 Da to 800 Da. 17. The membrane according to any one of items 9 to 14 exhibits an ethanol permeance in the range of 0.5 L m-2h-1bar-1to 25 L m-2h-1bar-1. 18. A crystalline ceramic nanoporous membrane comprising: an alpha-aluminium oxide ceramic substrate derived from an aluminium precursor; and a crystalline aluminium formate film deposited on the ceramic substrate. 19. The membrane according to item 18, wherein the aluminium precursor is aluminium tri- sec butoxide. The contents of the articles, patents, and patent applications, and all other documents and electronically available information mentioned or cited herein, are hereby incorporated by reference in their entirety to the same extent as if each individual publication was specifically and individually indicated to be incorporated by reference. The applicant reserves the right physically to incorporate into this application any and all materials and information from any such articles, patents, patent applications, or other physical and electronic documents. The following Examples illustrate the invention. EXAMPLES Materials and Equipment   All the reagents were obtained from commercial suppliers and used without further purification: aluminum sec-butoxide (97%, Alfa Aesar), formic acid (97%, Alfa Aesar), ethanol (99%, Aik Moh Singapore), isopropanol (99%, Sigma-Aldrich), ethyl acetoacetate (99%, Sigma-Aldrich). The deionized water produced by Millipore-Q System (Millipore, Billerica, MA, USA) was used in all experiments. Different testing solvents: acetone (99%), acetonitrile (99%), hexane (99%), heptane (99%), o-xylene (99%), dimethylformamide (DMF, 99%), N-methyl-2-pyrrolidone (NMP, 99%) were obtained from Tee Hai Chem. Porous asymmetric ^^-Al2O3plates with a top pore size of 100 nm, a diameter of 18 mm, and a thickness of 1 mm were purchased from Fraunhofer Institute for Ceramic Technologies and Systems (IKTS). Porous asymmetric ^^- Al2O3tubes with O.D. of 10 mm, a thickness of 2 mm, and a length of 10 cm were purchased from Foshan YIRUN fine ceramic new material Co. Ltd. Powder X-ray diffraction (PXRD) results were obtained from a Rigaku Miniflex 600 X-ray powder diffractometer equipped with a Cu sealed tube ( ^^ = 1.540598 Å) at a scan rate of 3° min-1. Fourier-transform infrared spectroscopy (FTIR) was performed on a Bio-Rad FTS-3500 ARX FTIR spectrometer. The chemical properties of the as-fabricated membranes were further confirmed by X-ray photoelectron spectroscopy (XPS, Kratos AXIS Ultra DLD, England) with a monochromatic Al K ^^ radiation (1486.71 eV) at 15 kV as the excitation source. Gas / vapor sorption measurements were conducted on a Micromeritics ASAP 2020 instrument. Before sorption tests, each sample (80 mg) was degassed at 393 K for 12 h. The sorption isotherms were collected at a pressure range of 0-0.99 bar. Thermogravimetric analyses (TGA) were conducted on a TA instruments STD-600 system with a heating rate of 10 ℃ min-1up to 800 ℃, and the N2flow rate was set at 50 mL min-1. Membrane morphology and thickness were observed by field emission scanning electron microscopy (FESEM, JSM-6700, JEOL) under 15 kV. Before characterization, all the samples were sputtered with Pt by a sputter coater (Cressington 208 HR) under a current of 20 mA for 70 s. The membrane surface was investigated by atomic force microscopy (AFM, Bruker Dimension ICON), and the surface roughness was analyzed by a tapping mode with a rate of 1 Hz. The surface charge of membranes was determined by measuring zeta potentials using a SurPASS electro-kinetic analyzer (Anton Paar, GmbH). To evaluate the surface wetting property of various organic solvents, in-air sessile drop contact angle (CA) with different organic solvents was measured using an optical instrument (DataPhysics, OCA25). A small droplet of solvents (around 3 ^^L) was dropped on the membrane surface, and the CA value was obtained by analyzing the shape of the solvent droplet. The concentration of the aluminum ion was analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES, Optima 7300DV, Perkin 25      Elmer). Before analysis, the solvent was fully evaporated by the hot plate, and the same amount of water was injected into the vial. Example 1: Membrane Design and Characterization combining sol-gel film formation and the into crystalline phases. The sol-gel process was adopted from a previous study with minor modifications (Tadanaga et al., Journal of the American Ceramic Society 1997, 80, 3213-3216). Briefly, aluminum tri-sec-butoxide and isopropanol were mixed and stirred at room temperature for around 1 h. After that, ethyl acetoacetate was added to the mixed solution as the chelating agent, and the mixture was stirred at room temperature for around 3 h. Pure water diluted by isopropanol was then added dropwise to the above solution for hydrolysis, and the resultant solution was kept aging at room temperature overnight. The molar ratios of isopropanol, ethyl acetoacetate, and water to aluminum tri-sec-butoxide are 20, 1, and 4, respectively. Finally, the sol was diluted by isopropanol, and the concentration of aluminum tri-sec-butoxide in the diluted sol was 0.00023 mmol L-1. In the dip-coating process, the bare ceramic substrate was loaded on a Teflon holder, and the loaded substrate was further dipped into the diluted sol for 30 s. Afterward, the excess sol coated on the backside of the substrate was wiped using filter paper before it was transferred to the conventional oven at 180 ℃ for 6 h. Finally, the coated substrate was loaded on the Teflon holder and vertically placed into the autoclave, and then 300 µL formic acid was added to generate formic acid vapor. The autoclave was sealed and put into a conventional oven at 180 °C for 10 h to 21 h. The as-prepared Al(HCOO)3membrane was washed by immersing it in ethanol overnight and then dried at room temperature. As for the synthesis of Al(HCOO)3powder, the original sol was directly heated in the conventional oven at 180 °C for 12 h. The obtained powders were grounded into fine particles and put into an autoclave. Afterward, formic acid (1 mL) was added to the autoclave. Notably, the formic acid liquid should not come into direct contact with the powder. Finally, the autoclave was sealed and put into a conventional oven at 160 °C for 12 h. 1.2: DFT calculations to obtain the total energy of crystalline and amorphous Al-network First-principle density functional theory (DFT) calculations were conducted by the Dmol3 module in BIOVIA Materials Studio (Zhang et al., Angewandte Chemie International Edition 2017, 56, 16282-16287) Vanderbilt-type ultra-soft pseudopotentials and generalized gradient 26      approximation (GGA) with Perdew-Burke-Ernzerhof (PBE) functional were used to approximate the exchange correlation energy (Perdew et al., Physical review letters 1996, 77, 3865). In addition, the double-numeric-quality basic set with polarization functions (DNP) was applied for valence electrons, and the self-consistent field (SCF) method was applied to solve the Kohn-Sham equations. For the calculation of the crystalline Al-network, a 2 × 2 × 2 k-point mesh (generated with the Monkhorst-Pack grid) was used. For the calculation of the amorphous Al-network, an amorphous aluminum hydroxide model with the same number of aluminum atoms was used, and the unsaturated bonds at the boundaries were saturated by protons. The tolerance of SCF was set at 1 × 10-6Ha, while the tolerances of energy, gradient, and displacement convergence for the geometry optimization were set at 1 × 10-5Ha, 2 × 10-3Ha / Å, and 5 × 10-3Å, respectively. The total energy is the combination of the sum of atomic, kinetic, electrostatic, exchange-correlation, and spin polarization energies. Results and Discussion: FIG.1a shows the fabrication process of the Al(HCOO)3membranes made by the ACT method (referred to as ACT-Al(HCOO)3), in which the sol-gel coated amorphous alumina layer was transformed into a crystalline structure by thermal treatment with formic acid vapor. Al(HCOO)3was selected because it has one of the simplest structures of all MOFs and can be synthesized from cheap and abundant reagents such as Al(OH)3and formic acid (FIG.7 and Table 1). In addition, such crystals are easy to grow and assemble, suggesting that they are suitable for constructing membranes in a facile and reproducible manner. Specifically, aluminum sec-butoxide was used as the starting material, which was hydrolyzed by water to form mono- or tri-hydroxides and further peptized to a clear sol by chelating agents. After dip coating, the sol was concentrated, and the colloidal hydroxide particles condensed on the mesoporous support through a dehydration-gelation process (FIG. 6). Finally, the obtained amorphous alumina gel served as the pre-cursor for further crystallization. FIG.1b and FIG.8-9 reveal the crystal structure of ACT-Al(HCOO)3obtained from refined powder X-ray diffraction (PXRD) data. The unit cell of ACT-Al(HCOO)3slightly shrunk compared to the solvothermally prepared samples (Table 2). The compound crystallizes cubically in the Im3 ̅ space group, and each aluminum atom is six-fold coordinated by six formate linkers in an anti-anti bridging mode. Table 1: Comparison of productivity (cm2mL-1) for different membranes and methods. FIG.7 and Table 1 were prepared according to Ma et al., Angewandte Chemie 2020, 132, 22093- 22098. 27      Amount of precursor used for unit membrane area (mL m-2or mg m-2) IlAluminum-tri-sec-F i i p3y p -p and solvothermal synthesis. Al(HCOO)3obtained by vapor-phase 28      ^^ (°) 90 ^^ (°) 90 the aluminum gel was contacted with formic acid vapor, a considerable number of Fourier-trans- form infrared (FTIR) peaks, ranging from 1800 to 800 cm-1, were detected in the product (FIG. 10). As indicated by the appearance of symmetric and asymmetric stretching of carboxylate bands at 1760 and 1620-1120 cm-1and bending vibration of C-H bonds at 1380 and 860 cm-1, the reaction between the aluminum gel network and formic acid vapor would start in a very short time. This conclusion agrees with X-ray photoelectron spectroscopy (XPS) data, confirming the decisive role of formic acid in the ACT process (FIG.11). However, at this stage, the membrane was still amorphous (FIG.12), and its thickness remained virtually unchanged (FIG.13). The membrane thickness indicates the degree of ACT process, and it is expected to increase drastically when the dense aluminum gel transforms into a porous crystalline structure. After the reaction proceeded for 3 h, scanning electron microscope (SEM) images (FIG.13-14) show that the surface of the aluminum gel gradually changed from smooth to rough, and a considerable expansion in film thickness was observed. PXRD analysis indicates a clear signal arising from (200) and (220) reflection planes with 2θ values of 15.6° and 21.9° (FIG.15). As the reaction extended from 3 to 18 h, the following observations of 1) the increase in PXRD (200) reflection signal intensity, 2) the narrowing of PXRD peaks, 3) the appearance of new PXRD peaks (e.g., (110), (222), and (400) reflections), 4) the increase in particle size 29      (FIG. 14), and 5) the increase in surface area (e.g., higher CO2uptake upon prolonging reaction, FIG.16) all indicate that longer transformation time would benefit the formation of ordered crystalline domains. Meanwhile, the obtained crystalline membranes exhibit much higher surface roughness, as evidenced by atomic force microscopy (AFM). The root-mean- square roughness (Rq) increased from 179 nm at 3 h to 343 nm at 18 h (FIG.17), primarily due to the progressive formation of large crystal faces. The energy-dispersive X-ray spectroscopy (EDS) mapping further validates the homogeneous distribution of carbon, oxygen, and aluminum elements in the ACT-Al(HCOO)3membranes (FIG. 18). Notably, temperature has a significant influence on the ACT process. An increase in transformation temperature led to a higher crystallization rate after a comparable transformation time (FIG. 19), which can be attributed to the faster diffusion of formic acid vapor and higher reaction kinetics. Accordingly, the transformation reaction front would progress deeper into the aluminum gel, thus mobilizing more Al3+for crystallization. Density functional theory (DFT) calculations were applied to further understand the ACT process. Briefly, the total energy gradually decreases as the aluminum gel network transforms into a crystalline structure, suggesting that this reaction is thermodynamically favorable (FIG. 20). The transformation is largely driven by the stabilization of the aluminum coordination sphere, replacing disordered Al-O with ordered Al-HCOO bonds. Based on the above results, it is proposed that during the ACT process, the formic acid is firstly adsorbed by the aluminum gel, simultaneously breaking the Al-O bonds. With the extension of time, the disordered Al- HCOO fragments start to re-arrange into partially ordered structures driven by the minimization of total energy, leading to products with the coexisted crystalline and amorphous phases. Finally, the process will achieve the thermodynamic minimum configuration, namely, highly crystalline ACT-Al(HCOO)3membranes. Example 2: Pore Size Evaluation 2.1: Membrane pore size evaluation based on modified gas-translation model Xiao and Wei first developed the gas-translation (GT) or activated Knudsen diffusion model to evaluate the diffusion of hydrocarbons in zeolites (Xiao et al., Theory. Chem. Eng. Sci.1992, 47, 1123-1141). After that, Shelekhin extended the model and applied it to study the gas permeance in microporous molecular sieve membranes (Shelekhin et al., AlChE J.1995, 41, 58-67). This model covers several permeation mechanisms such as Knudsen diffusion, activated diffusion, surface diffusion, and molecular sieving, where permeating molecules are restricted in their movements because of the interaction from the potential field of the nm-sized 30      channel wall, and therefore, they must overcome certain energy barriers. Detailed expressions are provided in Table 3 (Nagasawa et al., AlChE J.2014, 60, 4199-4210). Table S3. Expressions of the modified GT model. ^ For microporous membranes, permeance (mol m-2s-1Pa-1) for the ithgas component in the GT model, ^^^^ ^^, ^^, can be described by: ^^, ^^ൌ ^^^^^^ ^^ ^^^^^^^^ ^^ ^^ ^ ^^ ^^ ^^ ^^ ^^(Eq. S1) of ^^ ^^component, ^^ is the gas constant, and ^^ represents the absolute temperature. ^ Porosity can be defined as the ratio of the effective pore area and the total membrane area: ^^ ^ ^^^^^^^^ ^^^^^^ / ^^ ^^ ൌ^ ^^^^ൌ (Eq. S2) ^^ ^^ ^^of pores, ^^^^is the physical effective area of the pores, ^^^^ diameter, and ^^^^is the total membrane area. ^ The probability of diffusion ^^^^can be explained by the combination of geometrical probability and the activation energy of permeation: ^^^^ൌ ^^^^^^ ^^ ^^ ^െ^^ ^^, ^^^^ ^^^ (Eq. S3) molecules and membrane pore walls. ^ By substituting Eq. S3 to Eq. S1, and taking the pore diameter ( ^^^^) as the effective diffusion length ( ^^^^), ^^^^ ^^, ^^, can be further described by: ^^^^ ^^ ^^^^ ^^, ^^ൌ ^^^^^^, ^^^^ ^^^^^^^^^ ^^ ^^ ^^ ^^^^ ^^ ^^^െ^^ ^^^ (Eq. S4) the if the pore size becomes larger, ^^^^, ^^tends to be 0, and therefore, the permeation shifts to the Knudsen type. ^ In the modified GT model, the geometrical probability is modified by the size of the permeating molecules and the random factor of 1 / 3 in a 3D pore space, as shown below: ^^^^^^, ^^ൌ^^ ^^^^ ^^(Eq. S5) the gas^^cross- area pore. ^ To simplify the equation, the effective cross-sectional area for permeation ( ^^^^) is assumed to be a circle with the diameter equal to the effective diffusion length ( ^^^^, ^^): 31      ^^^^, ^^ൌ ^^^^െ ^^^^(Eq. S6) Where ^^^^refers to the kinetic diameter of the ithgas component. ^ By substituting Eq. S6 into Eq. S5, the geometrical probability can be expressed as: ^^^ ^^ ି ^^ ^ ^^ / ^^ ^ ^^ ି ^^^^^^ ^^^^, ^^ൌ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^^ ^^ ^^ൌ^^ ^^ ^^ ^^^^ / ^^ൌ^^ ^^ ^^^^^ the effective diffusion length ( ^^^^, ^^), can be expressed as: ^^^^ ^^ ^ ^^ ି ^^ ^ ^^ ^^^^ି ^^ ^^, ^^ൌ^^ ^^ ^^൫ ^^^^െ ^^^^൯^^ ^^^^ ^^^^ ^^^^^ ^^ ^^ ^^ ^^^^ ^^ ^^^െ^^, ^^^^ ^^^ (Eq. S8) ^ ^^ି ^^ ^^, ^^ൌ^ ^^ ^^ ^^ ^^ ^^ ^^(Eq. S10) of membrane configuration factors and the kinetic ^ By analyzing ^^^^, ^^with temperature-dependent single gas permeance data, the mean effective pore size of the membrane can be approximately determined: ^^^^, ^^ൌ ^^^ ^^^^െ ^^^^^^^(Eq. S11) The linear relationship between ^^^^and ^^^^ / ^^^^, ^^is shown in Eq. S13, and therefore, from the linear plot of ^^^^ / ^^^^, ^^versus ^^^^, the mean effective pore size ( ^^^^) and the structural parameter ( ^^) can be obtained. Notably, the activation energy of permeation ( ^^^^,) and the pre-exponential factor ( ^^0,) can be determined by linear regression using Eq. S9 with the temperature-dependent single gas permeance data. For gas permeation through membranes with ultra-micropores, the diffusion barrier is the dominant factor, and therefore, ^^^^,>0. This permeation regime is defined as the activated diffusion. For gas permeation through membranes with larger micropores, the interactive force between the gas molecules and membrane pore wall is more dominant than the diffusive barrier, and therefore, ^^^^,<0. This permeation regime is defined as surface diffusion. 32      2.2: Gas separation tests Single gas and binary-gas mixture separation performance was tested using a homemade Wicke-Kallenbach gas permeation setup described in the previous studies (Ying et al., J. Am. Chem. Soc. 2020, 142, 4472-4480, and Ying et al., Adv. Mater. 2021, 2104946). The membrane was loaded and sealed in the permeation cell, which was placed in a furnace to control the testing temperature. The feed-side pressure was controlled by a back-pressure regulator, while the permeate-side was operated under atmospheric pressure. The gas flow rate was controlled by the mass flow controllers (MFC, D07-26C, SevenStar China). Argon served as the sweep gas, and its volumetric flow rate was set at 25 mL min-1. The molar concentrations of the permeate gas were tested by gas chromatography (Agilent 7890B) with two TCD detectors, and the gas permeation data were recorded at a steady state. Every recorded data point was tested at least three times to minimize the random error. Before the permeation test, the membrane was heated at 120 ℃ for 6 h to remove the adsorbed moisture. For the single gas permeation test, the permeance of gas component i ( ^^^^, GPU) was calculated with the equation: ^^ே^ ൌ^∆^ൈ^ൈଷ.ଷସ଼ൈ^^షభబ(Eq. S14) Where ^^^^represents the molar flow rate of component i (STP, mol s-1), ^^ represents the tested membrane area (m2), and Δ ^^ is the pressure gradient (Pa). For the binary-gas mixture test, the volumetric flow rates at the feed side for both gases were set at 20 mL min-1. The separation factor ( ^^) of gas pair i, j was calculated by the equation: ^^^,^ൌ௫^ / ௫ೕ௬^ / ௬ೕ(Eq. S15) and ^^ represent the mole fractions of component i and j in the permeate and feed gas, respectively. 2.3: N2permeation tests For gas flow through a porous medium under Knudsen and Poiseuille flow regions, the gas permeance can be described by the combination of a diffusive term and a viscous term, as expressed by Carman (Carman et al., Ind. Eng. Chem. Res.2001, 40, 5710-5718, and Khayet et al., Ind. Eng. Chem. Res.2001, 40, 5710-5718. ^^ଶ^ ఌ ^.ହ ^మఌ^ ൌ^^଼ோ்^ ^^^^^ൌ ^^^^ ^^^^^^   Where ^^^^is gas permeance (mol m-2s-1Pa-1), ^^ ^^ (m) is pore radius, ^^^^(m) is the effective pore length, ^^ is the surface porosity, ^^ (8.314 J mol-1K-1) is the universal gas constant, ^^ (kg m-1s-1) is the gas viscosity, ^^ (kg mol-1) indicates the molecular weight, and ^^^^(Pa) represents the mean pressure. Notably, ^^0represents the diffusive term while ^^0^^^^represents the viscous term. If the pore sizes of the porous medium continuously decrease, the contribution of the viscous term becomes negligible, and the gas permeance tends to become constant at varying mean pressures. 2.4: Positron annihilation spectroscopy (PALS) test PALS coupled with Doppler broadening energy spectroscopy (DBES) was used to analyze the microstructure properties of the crystalline membrane, which was conducted according to a previous study (Chen et al., Macromolecules 2007, 40, 7542-7557). In brief, a variable mono-energy slow positron beam (0.1-27 keV) was generated by the radioisotope22Na, and the DBES spectra were collected by the HP Ge detector. In the PALS tests, aluminum-formate membrane sample disks (diameter = 18 mm) were tightly attached to an aluminum testing plate, with top surfaces facing the positron beam. R parameter is related to the relative amount of cavities (nm- ^^m), and it is used to explain the DBES spectra. 2.5: Fluorescence confocal optical microscopy (FCOM) test FCOM test was conducted on a Nikon A1 confocal laser scanning microscope and analyzed with the NIS Elements Viewer. Before the imaging process, the membrane top surface of each sample was contacted with the fluorescent molecular probe solution (1 mM pyrene acetone solution) for 6 h using a dead-end diffusion cell. Results and Discussion: The aim to create sub-nanometer transport pathways with intercrystalline spacing raises an intriguing question of whether organic liquids can transport through the intrinsic pores of Al(HCOO)3. To exclude this possibility, adsorption tests with four different gases were conducted. The CO2, CH4, and N2adsorption isotherms in FIG. 21 show that although Al(HCOO)3has a high uptake capacity for CO2, the uptakes for CH4and N2are extremely low, indicating that molecules larger than CO2would be excluded from entering the framework pores. The Brunauer-Emmett-Teller (BET) surface area of Al(HCOO)3was determined to be 16.8 m2g-1, confirming that its structure is too dense to be probed by N2(FIG.22). Furthermore, the uptake of methanol vapor at 298 K is also very low, indicating the inaccessibility of methanol molecules to the framework pores (FIG.23). By evaluating these physisorption data, it is evidentially concluded that the transport of organic liquids through the intrinsic crystal 34      pores is blocked by the ultramicroporous nature of Al(HCOO)3. Therefore, intercrystalline spacing is the only transport pathway in ACT-Al(HCOO)3membranes. Besides physisorption tests, single-gas permeation analysis was conducted on membrane samples to directly evaluate the channel properties. The sizes of the intercrystalline spacing were quantitatively determined by the NKP method (Nagasawa et al., AIChE Journal 2014, 60, 4199-4210), which is based on the modified gas-translation (GT) model, using experimentally obtained permeance data of H2(0.289 nm), CO2(0.33 nm), CH4(0.38 nm), and C3H6(0.425 nm). By analyzing ^^^^,(activation energy of permeation) and ^^0,(the pre- exponential factor) in Equation S9, the effective pore size ( ^^^^) can be approximately estimated. The linear plot of ^^^ / ଷ^,^against the kinetic diameter of ith gas molecule ( ^^^^) is shown in FIG.2a- c, and ^^^^could be determined at the intercept of the x-axis. FIG.2d-l shows the average thicknesses of ACT-Al(HCOO)3membranes of 3.1, 5.1, and 5.9 ^^m after 11, 14, and 21 h of reaction, respectively, and the resulting membranes were denoted as ACT-11h, ACT-14h, and ACT-21h. According to the fitting results, the sizes of the intercrystalline channels could be controlled from sub-2 nm to sub-1 nm and further to sub-0.6 nm, simply by extending the reaction time from 11 h to 21 h. Notably, for all the tested membranes, the permeance of different gases slightly decreased with increasing temperature. The calculated ^^^^,values are negative, indicating that the permeation processes are dominated by surface diffusion (FIG. 24). In other words, the interaction affinity of permeating gases with the channel wall is more dominant than diffusion, suggesting that large mesopores were eliminated. N2permeation tests at different transmembrane pressures confirm that large-pore-induced viscous flow played no part in permeation (FIG.25). Gas separation tests using binary equimolar mixtures were also conducted on the ACT-21h membranes. As shown in FIG. 26, the tested membranes exhibit selectivities of 3.1 for CO2 / CH4and 3.5 for CO2 / N2, which are higher than the corresponding ideal Knudsen selectivities, implying that the channel sizes approach the sizes of permeating molecules. Meanwhile, the selectivities for H2 / CO2, CO2 / C3H8, and CO2 / C3H6were calculated to be 0.69, 0.64, and 0.62, respectively, which are lower than the corresponding ideal Knudsen selectivities, indicating that CO2has higher adsorption affinity than H2. In contrast, C3H6and C3H8have stronger interactions with the channel wall than CO2. This observation is consistent with the temperature-dependent single-gas permeance results (FIG.24). Furthermore, positron annihilation spectroscopy (PALS) analysis was conducted to provide more direct information on porosity. The R parameter of PALS analysis represents the intensity of nm-to- ^^m-sized voids within the membrane; therefore, a larger R parameter 35      indicates a larger ^^^^. Additionally, in FIG.27, the initial decline and the subsequent ascent in the R parameter curves with increasing positron-incident-energy can be interpreted as the transition from the top Al(HCOO)3layer to the mesoporous ceramic support. A comparison of R parameters at the low-incident-energy region shows that R values decreased as the ACT process time increased, indicating a decreasing trend in the size of intercrystalline voids. A smaller void volume implies a more selective pathway for transporting small molecules. Finally, fluorescence confocal microscopy (FCOM) analysis was conducted to visualize these channels. FIG.28a-b reveals the existence of intercrystalline voids in ACT-11h by FCOM. The images were taken after the membranes were brought into contact with pyrene solution by pressure-driven filtration. Pyrene is a fluorescent molecule with a molecular size larger than the pore size of Al(HCOO)3(i.e. larger than 0.35 nm); therefore, it cannot enter the intrinsic framework pores. Thus, the detected fluorescent signal, highlighting the grain boundaries across the membrane’s top layer, indicates the presence of intercrystalline transport pathways (FIG.29). Example 3: Ultrafast Organic Solvent Nanofiltration 3.1: Geometry optimization of the dye molecules Geometry optimization of the dye molecules was performed on the basis of Gaussian-DFT calculations, as implemented in the Gaussian 09 quantum chemistry package (Frisch et al., J. Gaussian 16 Rev. C.01, Wallingford, CT, 2016). Spin-unrestricted calculation was performed using B3LYP hybrid exchange-correlation functional, and cc-pVDZ was chosen as the basis set. All possible configurations of spin for each dye molecule were considered to designate the most representative geometry. 3.2: Organic solvent nanofiltration (OSN) test The pressure-driven OSN tests were performed by homemade dead-end permeation cells and tubular membrane modules. For tests with dead-end permeation cells, the applied pressure was set at 2 bar, and the stirring speed was set at 500 rpm. For tests with tubular membrane modules, the applied pressure was set at 10 bar (FIG. 44). Prior to the OSN test, the membrane was stabilized by ethanol with an applied pressure of 2 bar for 12 h. Afterwards, the permeance of pure ethanol was measured. For the OSN test, the concentration of ethanol solution with various dye solutes was set at 50 ppm, and it was used to measure the permeance and rejection. The feed volume was 100 mL, and the permeance and rejection were tested after steady permeance was achieved (12 h). The dye concentrations in the feed 36      and permeate solutions were measured by UV-Vis spectrometer (Agilent, Cary 60), and the solvent permeance ( ^^) and dye rejection ( ^^) were calculated by: ^^ ൌ∆^ௌൈ∆^ൈ∆௧(Eq. S17) ൈ 100% (Eq. S18) Where Δ ^^, ^^, Δ ^^, and Δ ^^ represent the volume of the collected permeate, the effective membrane area, the transmembrane pressure, and the testing time, respectively. ^^^^, and ^^^^are the dye concentrations of the permeate and feed solutions, respectively. 3.3: Binary dye mixture separation test Two ethanol solutions containing two different dye solutes (50 mL, 50 ppm) were mixed to form the mixed feed solution (100 mL). The binary dye mixture separation tests were performed with the dead-end filtration cells. The operation pressure was set at 4 bar, the stirring speed was set at 500 rpm, and the testing time was 12 h. The concentrations of the mixed dye solutions were calculated by Beer's law to mixtures (Skoog et al., Principles of instrumental analysis. Cengage learning: 2017). ^^௧^௧^^ൌ ^^^ ^^௫^^௫^ ^^௬^^௬^ ⋯^ (Eq. S19) Where ^^ is the path length (cm), ^^ is the molar extinction coefficient (cm-1^^M-1), and ^^ is concentration ( ^^M). Results and Discussion: Before performing the organic solvent nanofiltration (OSN) experiments, the stability of Al(HCOO)3membranes on exposure to organic liquids was accessed. ACT-14h exhibited a rejection of over 99% for methyl blue in ethanol and remained stable after exposure to nonpolar (hexane and toluene), aprotic (acetone, dimethylformamide (DMF), and N-methyl-2- pyrrolidone (NMP)), and protic (ethanol and methanol) solvents for 21 days (FIG. 30). Inductively coupled plasma-optical emission spectrometry (ICP-OES) results show an extremely low concentration of Al3+(0.3 ppm) after continuously monitoring the permeate of pure methanol and DMF through the Al(HCOO)3membranes for 7 days (FIG.31). Compared to the pristine membrane samples, almost the same PXRD patterns were observed in the membrane samples after solvent treatment (FIG. 32). These results indicate that the membranes retained their intact nanostructure in the presence of various organic solvents. Ethanol permeance was plotted as a function of time in FIG.33a. For ACT-14h, permeance decreased by 45% over the first 5 h, which is presumably the result of physical compaction of 37      the Al(HCOO)3membrane with ethanol filtration. To explore the physical stability of the selective layer, ACT-11h and ACT-14h were tested under a pressure range of up to 10 bar. The ethanol flux increased linearly with increasing applied pressure (FIG. 33b) without affecting the dye rejection performance (FIG.33c), indicating that the intercrystalline channels did not collapse / fail under high working pressures. Finally, the thermal stability was analyzed by thermogravimetric tests (FIG.34), showing high thermal stability of Al(HCOO)3membranes at 300 °C. However, when exposed to water solution, Al atoms in the crystal framework tend to be hydrolyzed and demetallized from the lattice, with the breaking of the Al-O bond. As demonstrated by PXRD and long-term filtration tests over 200 h (FIG.35 a-b), Al(HCOO)3membranes can only be used in organic solvent systems containing less than 20 vol% of water, limiting their applications in aqueous systems. Next, the potential of Al(HCOO)3membranes with intercrystalline channels for OSN was evaluated in a pressure-driven dead-end permeation cell. Ethanol was chosen as the solvent because it is among the top 10 solvents used in the pharmaceutical industry. The transmembrane pressure was set constantly at 2 bars for each test, and dyes with different sizes and chemical functionalities (FIG. 36, Table 4) were used as the markers at a concentration of 50 ppm. The permeate was collected after the steady state was reached. With ACT process for 10 h, a relatively high pure ethanol permeance was achieved at 22.0 L m-2h-1bar-1. The value was further decreased to 19.4, 13.0, 6.2, 2.0, and 0.8 L m-2 h-1bar-1at 11, 13, 14, 20, and 21 h, respectively. It is also observed from FIG.3a-b that for ACT-10h, the rejection of CR (653 Da), AF (542 Da), and CV (372 Da) reached 88.1%, 74.3%, and 17.2%, respectively. When the ACT process was prolonged to 21 h, these values were boosted to 99.8%, 99.7%, and 98.7%, respectively. To explore the separation properties of Al(HCOO)3membranes, dyes with different molecular weights, ranging from 304 to 756 Da, were tested. FIG.3c shows an increasing rejection trend versus molecular weight, suggesting that ACT-11h with sub-2 nm channels has a larger MWCO of approximately 650 Da, ACT-14h with sub-1 nm channels has a smaller MWCO of around 540 Da, while ACT-21h with sub-0.6 nm channels has the smallest MWCO lower than 300 Da. This range of molecular weight corresponds to 2.2 × 1.8 × 0.4 nm to 1.2 × 1.1 × 0.3 nm in terms of molecular size (geometry optimized by Gaussian-DFT, FIG. 36). Dyes can be rejected by size exclusion or Donnan effect / electrostatic repulsion. The zeta potential of the Al(HCOO)3membranes at pH 7 was measured to be 38.2 mV (FIG.37), indicating that, in principle, these membranes could reject positively charged dyes more effectively. However, considering the fact that ionic dyes should not dissociate much in organic liquids and the tested dyes contain all types of charges, the 38      contribution of the Donnan effect can be principally excluded. Thus, the rejection should be largely based on physical size exclusion. Table 4. The abbreviations, molecular weights, charges, and UV absorption peaks of the dyes used in this study. Name MW (g mol-1) ChargeUV-Vis absorption peak(nm)On of the three mentioned membranes were continuously monitored for over 10 h. As shown in FIG.3d- f, ACT-11h (sub-2 nm) showed the highest permeance of 14.6 L m-2h-1bar-1with 94.8% CR rejection, ACT-14h (sub-1 nm) revealed moderate permeance of 4.1 L m-2h-1bar-1with 94.5% AF rejection, and ACT-21h (sub-0.6 nm) exhibited low permeance of 0.8 L m-2h-1bar-1with 99.9% CV rejection. FIG.3g-h and FIG.38-40 present the UV absorption spectra of various dyes before and after filtration tests, in which the in-set photographs are the feed and permeate solutions. These results, along with the temperature-dependent single-gas permeance data, demonstrate the tunability of the intercrystalline channel sizes. To confirm that dye adsorption did not contribute much to the separation performance, a long-term filtration test was continuously operated with 500 ppm MB solution for up to 280 h. During this period, the permeance decreased by 82%, while the membrane retained a high MB rejection of 98.5% (FIG.41). The decreased permeance can be attributed to the increased osmotic pressure and the extra resistance due to the accumulation of rejected dyes sticking on the membrane surface (FIG. 42). Additionally, soaking Al(HCOO)3crystals (20 mg) in dye solutions (20 ml) would not lead to dye-adsorption on the crystal surface after 14 days (FIG. 43), confirming that dyes were rejected by the membrane. Notably, the separation capacity is generally determined by the MWCO, where the rejection reaches 90%. The separation lower limit is defined by molecular weight retention (MWRO), where the rejection decreases to 10%.Thus, the steepness of size exclusion is determined by the difference between MWCO 39      and MWRO. For ACT-14h, the difference between MWCO and MWRO is around 300 Da (FIG. 3c). This value is similar to that of 2D COF membranes, indicating that the intercrystalline channels have sharp molecular sieving properties. To further explore the monodispersity of these channels, sieving experiments were performed to separate dyes with similar sizes from a binary equimolar mixture (FIG.4a-b). Initially, two kinds of mixtures were prepared in ethanol, i.e., the mixture of AF and CV, and the mixture of AF and CR. Since ACT-21h has narrower channels, only the smallest molecule, CV, could diffuse through the membrane. In contrast, the larger molecule, AF, was rejected, leading to a high selectivity of 58.2 and moderate permeance of 1.9 L m-2h-1bar-1(FIG.4c-e). As for ACT-14h, the channel size is relatively larger. Therefore, a small amount of AF could pass through the membrane, while the largest molecule, CR, was rejected, leading to a moderate selectivity of 6.8 with higher permeance of 5.7 L m-2h-1bar-1. The precise molecular-level selectivity demonstrates the controllable nanostructure of the intercrystalline channels. Finally, a lab-scale tubular membrane module with an effective membrane area of around 23 cm2was fabricated to explore the scalability of this method. The tubular membrane was made by a simple dip-coating method and tested in crossflow operation mode under 10 bar (FIG. 44). As expected, the resultant module exhibited ethanol permeance (5.8 L m-2h-1bar-1) identical to the single flat-sheet membranes, with a slight decrease in rejection rate (selectivity of methyl blue: 95.8%), which could be ascribed to the rough substrate surface (FIG.45). In conclusion, the generality and simplicity of the ACT process afford high potential for large- scale applications. Example 4: Transport Behavior of Organic Liquids 4.1: Transport behavior of organic liquids The Hagen-Poiseuille model describing solvent flow is derived from integrating the Navier- Stokes differential equation assuming the no-slip boundary conditions. Therefore, it is generally agreed with the ultrafiltration process, in which the only considered solvent property is the viscosity ( ^^): ^^ ൌఢగௗ^ర ∆ ^^^ಹ^ு^ൌ ∆ ^^ (Eq. S23) and ^^^^refers to membrane effective pore size. 40      In nanofiltration processes, where interactions between solvent and membrane pore wall and the steric hindrance are expected to play a much more important role, the Navier-Stokes differential equation should be integrated with proper boundary conditions considering a slip- length ( ^^^^), and this leads to the modified Hagen-Poiseuille model with a correction factor ^^^^: ^^ ൌ ^^ு^^1 ^ସ ^ೞௗ^^ ൌ ^^ு^^1 ^ ^^^^ (Eq. S24) ^^^ൌ∆^ൌ ^^^,ு^^1 ^ ^^^^ൌఎ^1 ^ ^^^^ (Eq. S25) ^^^^^ ൌ Where ^^^^ ^^is a constant which represents the intrinsic properties of the membrane. Thus, for solvent flow through 2-nm-pore (Eq. S16), the correction factor ( ^^^^) is proportional to the interactive forces between solvent molecules and membrane, which is expressed by the total solubility parameter ( ^^0): ^^^,ଶି^^∝ ^^^(Eq. S27) Furthermore, for solvent flow through sub-1-nm-pore (Eq. S17), the steric effect indicated by the molar diameter of the solvent molecule ( ^^^^) is introduced to the correction factor ( ^^^^): ^^^,^௨^ି^^^∝ ^^^^^^ଶ(Eq. S28) 4.2: Molecular dynamic (MD) simulations The MD simulations were conducted by the Forcite module in BIOVIA Materials Studio. The original configurations for MD simulations were taken from the grand canonical Monte Carlo (GCMC) simulations (Kang et al., Nature Communications 2022, 13, 1-9). All the GCMC simulations were conducted by the Sorption module in Material Studio software. The atom- based summation method was used to describe the van der Waals interactions, and the Ewald summation method was applied to present the electrostatic interactions between the loaded guest molecules and the overall framework. In addition, the van der Waals interactions were described in the Lennard-Jones (L-J) terms, with a cut-off distance of 12.5 Å (Zheng et al., Cell Rep. Phys. Sci.2022, 100903). The intercrystalline model was rebuilt from the aluminum- formate crystal structure, and Forcite module was used to geometrically optimize the structure. For MD simulations, the L-J parameters of atoms and the electrostatic parameters were taken the same as the GCMC simulations, where Ewald and atom-based summation methods were 41      applied for the electrostatic and van der Waals interactions, respectively. In addition, simulations were performed in the constant-volume-and-temperature (NVT) ensemble at 298 K, which was controlled by the Nose thermostat. The total simulation time and the time step were set as 3000 ps and 1 fs, respectively. The self-diffusion coefficient ( ^^^^ ^^ ^^ ^^) was calculated from the mean square displacement (MSD, FIG.56) by fitting the MSD plots to time in the region between 100 and 3000 ps: ^^ ^^ ^^^ ^^^ ൌ 6 ^^^^^^^^ ^ ^^ (Eq. S29) Where ^^ is time interval and ^^ is the offset at time zero. Results and Discussion: It is well known that the continuum flow of fluids will break down as the pore size becomes comparable to that of the permeating molecules. Here, using 10 organic solvents (Table 5), pressure-driven permeation tests were conducted across Al(HCOO)3membranes with decreased pore sizes to analyze the dependence of permeance on solvent viscosity and the deviation from viscous continuum flow. Results show that solvent flow across loose Al(HCOO)3membranes with 100-nm-scale pores was inversely proportional to the solvent viscosity, which can be described by the Hagen-Poiseuille model (FIG.5a, FIG.46-47). Such a model is derived from integrating the Navier-Stokes differential equation with a no-slip boundary condition, assuming zero velocity at the interface. In contrast, deviations from this viscous flow were observed from Al(HCOO)3membranes with 2-nm and sub-1 nm pores (FIG. 46). This transition in flow behavior can be explained by the induced interactive forces as mesopores shrink to the nm-range. Previous studies with the Hagen-Poiseuille model suggested that the laws of classical fluid dynamics are applicable to describe liquid flow in nanoscale dimensions. However, the basic properties of liquids at nano-regions are different from those on larger scales. When approaching the nm-range, the no-slip boundary condition, which is typical of classical fluid dynamics, should be corrected. Thus, for liquid flow in confined channels where liquid-membrane interactive forces become significant, the Navier- Stokes differential equation should be integrated with proper boundary conditions considering a slip-length, leading to the modified Hagen-Poiseuille model with a correction factor ^^^^(Equation S19-20). Notably, the slip length indicates several types of interactive forces between membrane and solvent molecules. It provides the theoretical foundation for the correction factor ^^^^in the modified Hagen-Poiseuille model. For Al(HCOO)3membranes with 2 nm scale pores, liquid permeance scaled linearly with the combined solvent parameter ^^0^^−1, where ^^0is the total solubility parameter, and ^^ represents solvent viscosity (FIG.5b, FIG. 48-49). The ^^0factor is a function of several interactive forces, i.e., the dispersion forces, dipole forces, and donor-acceptor forces (hydrogen bonding), which as a whole contribute to 42      the correction factor ^^^^for solvent transport through 2-nm-scale pores (Equation S25). The enhanced solvent-membrane interactions in confined geometries were further confirmed by molecular dynamics (MD) simulations, as evidenced by the simulated density distribution of solvent trajectory along the channel walls (FIG.52-55). Besides, the solvent radial distribution function (RDF) g(r) was used to analyze the average distance between solvent molecules and membrane channel walls, which is indicative of the interaction strength (FIG.54). As expected, isopropanol showed the shortest distance to the channel wall, revealing that polar solvent molecules with larger molecular sizes will exhibit stronger interactions due to the hydrogen bonding and increased van der Waals forces. Table 5. Hansen solubility parameter and other physical properties of organic solvents used in this study. Hansen solubility parameter Molar (M1 / 2 d)ViscositPa ) d volume ^ଶ rk:   c) Molar volume is taken from Webster et al., (J. Am. Chem. Soc.1998, 120, 5509-5516) d) ^^ ^^= solubility parameter due to dispersion forces, ^^ ^^= solubility parameter due to dipole was studied by analyzing the contact angle values ^^ (FIG.5e, FIG.57, Table 7). According to the Lucas-Washburn equation, the negative cosine of ^^ implies the lack of affinity, while the positive value represents high affinity. For this reason, the Al(HCOO)3membranes formed wettable surfaces with high affinity to nonpolar, polar aprotic, and polar protic solvents, leading to a positive capillary force inside the pores, facilitating liquid transport. Thus, the membranes exhibited high permeance for both polar and non-polar solvents. Table 7. Summary of contact angle for different solvents. Solvent Contact angle (°) Standard deviation (°)     As membran of the pore becomes comparable with that of the solvent molecules, and therefore, steric hindrance is possible. Consequently, the solvent permeance scaled linearly with another solvent parameter ^^0^^−1^^^^−2, where ^^^^is the solvent molar diameter (FIG.5c, FIG. 50-51). The ^^^^term is to the correction factor ^^^^for solvent transport through sub-1 nm pores since it indicates the molecular steric effect in enhanced confinement (Equation S26). This 44      observation is consistent with the gas separation results using binary equimolar mixtures, where gas molecules with different kinetic diameters exhibited different diffusivities (FIG.26). MD simulation further demonstrated the presence of steric hindrance. It is well known that a larger slope of the mean square displacement (MSD) illustrates a higher diffusivity. On the basis of the MSD of hexane, heptane, and o-xylene in 1 nm channels, it was demonstrated that hexane with a smaller effective molecular dimension could easily diffuse through the intercrystalline channels. In contrast, o-xylene with a bulkier structure diffused much slower (FIG. 56, Table 6). Interestingly, the above deviations in the transport behavior of organic liquids are similar to the transition of gas transport mechanisms in 2D nanoporous membranes. Briefly, as pore sizes continuously decrease, the interactions between gas molecules and membranes, as well as the steric effect, are taken into account, and the gas transport behavior changes from viscous flow to Knudsen diffusion, surface diffusion, activated diffusion, and finally to molecular sieving. Table 6. Summary of self-diffusion coefficient ( ^^^^ ^^ ^^ ^^, cm2s-1) of solvent molecules obtained from MD simulation (Wang et al., Inorg. Chem. Front.2022). NameChannel sizeTemperature (nm) (K)^^ (cm2s-1) ^^^^ / ^^ି ^^ ^^ ^^ ^^ ^^ ^^^^ ^^ ^^ ^^^^ ^^ ^^ ^^^^^^ ^^ ^^ ^^ na y, comparsons wt varous types o membranes n t e terature are gven n G.5d and Table 8. Clearly, our membranes show higher permeance than the state-of-the-art materials with comparable selectivity, demonstrating the potential for technological translation to practical liquid-based molecular separation applications. Table 8. Nanofiltration performance comparison of membranes from the literature and this work. Membrane Ethanol Dye M D k ti Ref 7 45     PIM-1 / cellophane 0.06Remazol Brilliant 38Blue (626) 98Remazol Brilliant 8890091234566789501234567 Annealed MWCNTS-NH2- 2.3 Eosin Y (648) 9958P84 899012133456- - - - 12, 423, 65-72; (35) Fritsch et al., J. Membr. Sci.2012, 401, 222-231; (36) Ali et al., Adv. Mater. 2020, 32, 2001132; (37) Li et al., J. Membr. Sci. 2020, 601, 117951; (38) Geens et al., J. Membr. Sci.2005, 255, 255-264; (39) Wang et al., J. Membr. Sci.2009, 326, 222-233; (40) Gao et al., J. Membr. Sci.2018, 565, 169-178; (41) Vanherck et al., J. Mater. Chem.2011, 21, 6079-6087; (42) Xing et al., Green Chem.2014, 16, 1383-1392; (43) Zhao et al., Sep. Purif. Technol. 2021, 255, 117702; (44) Mertens et al., J. Membr. Sci. 2018, 566, 223-230; (45) Tham et al., J. Membr. Sci.2017, 542, 289-299; (46) Vanherck et al., J. Phys. Chem. C.2012, 116, 115-125; (47) Aburabie et al., J. Membr. Sci.2017, 523, 264-272; (48) Zhu et al., RSC Adv.2015, 5, 73068-73076; (49) Wang et al., Sep. Purif. Technol.2019, 227, 115687; (50) Li et al., Sep. Purif. Technol.2016, 165, 60-70; (51) Vanherck et al., J. Membr. Sci.2011, 373, 5-13; (52) Xu et al., J. Membr. Sci.2017, 526, 32-42; (53) Farahani et al., Sep. Purif. Technol. 2017, 186, 243-254; (54) Farahani et al., J. Membr. Sci.2018, 548, 319-331; (55) Abadikhah 47      et al., Chem. Eng. Sci.2019, 204, 99-109; (56) Gao et al., Carbon 2017, 124, 263-270; (57) Li et al., Carbon 2018, 130, 503-514; (58) Hua et al., Carbon 2017, 122, 604-613; (59) Gao et al., Chem. Eng. J.2018, 352, 10-19; (60) Li et al., Commun. Chem.2015, 51, 918-920; (61) Ma et al., ACS Appl. Mater. Interfaces.2019, 11, 45290-45300; (62) Yu et al., J. Membr. Sci. 2022, 644, 120130. Conclusion: In summary, Al(HCOO)3membranes with tunable intercrystalline channels at the nanometer scale were constructed by combining ACT transformation with vapor-phase processing. Such nm-scale channels were formed by assembling Al(HCOO)3crystals, and the intercrystalline spacing serves as the mass transport pathways. By controlling the ACT process time, the size of the intercrystalline channels could be tuned from macroscopic scale to sub-2 nm, sub-1 nm, and further to sub-0.6 nm, which was measured by the normalized Knudsen-based NKP permeance method and verified by gas / vapor sorption, PALS, and organic solvent nanofiltration. As the channel size decreased from macroscale to nanoscale, comparable to the size of solvent molecules, a transition from viscosity-dominated liquid flow to sub- continuum flow, which is dependent on solvent molecular size and solvent-membrane interactions, was observed. These channels were further visualised by FCOM. In addition, the obtained membranes exhibited precise molecular-level selectivity and fast solvent transport efficiency due to their tunable pore sizes and good wettability for various organic solvents. Finally, the scalability of this membrane fabrication method was demonstrated by fabricating a lab-scale tubular membrane module. This work facilitates the development of next- generation sustainable separation technologies with new energy-efficient membrane devices. 48

Claims

CLAIMS 1. A composite material for organic solvent filtration, the composite material comprising: a porous substrate; and a polycrystalline material layer disposed on a surface of the porous substrate; wherein the polycrystalline material layer is formed from a metal-organic framework (MOF) according to formula (I): X1-xYx(HCOO)3(I) wherein, X is a metal in the +3 oxidation state; Y is one or more metals in the +3 oxidation state, provided that each of the metals that represent X are different to each of the one or metals that represent Y; x is from 0 to 0.9999; and wherein the polycrystalline material layer comprises intercrystalline channels, the intercrystalline channels having an average pore size of from about 0.5 nm to about 100 nm.

2. The composite material of claim 1, wherein X is selected from Al, Fe, Mn, Y, Sc, Ga, In, Ti, Fe, Mn, V, Cr, Co, Tl, Bi, Ru, Os, Re, Ir, Rh, Ce, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu.

3. The composite material of any one of the preceding claims, wherein X is Al, Fe or V.

4. The composite material of any one of the preceding claims, wherein Y, when present, is selected from one or more of the group consisting of Al, Fe, Mn, Y, Sc, Ga, In, Ti, V, Cr, Co, Tl, Bi, Ru, Os, Re, Ir, Rh, Ce, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu.

5. The composite material of any one of the preceding claims, wherein Y, when present, is Fe or V.

6. The composite material of any one of the preceding claims, wherein x is from 0 to 0.99, for example, from 0.25 to 0.

85. 49       7. The composite material of any one of the preceding claims, wherein the MOF according to formula (I) is selected from the group consisting of Al(HCOO)3, Al0.75Fe0.25(HCOO)3, Al0.5Fe0.5(HCOO)3, Al0.25Fe0.75(HCOO)3, Al0.15Fe0.85(HCOO)3, V(HCOO)3, and V0.3Al0.7(HCOO)3.

8. The composite material of claim 7, wherein the MOF according to formula (I) is Al(HCOO)3< / sub>.

9. The composite material of claim 8, wherein the polycrystalline material layer is characterized by a powder X-ray diffractogram with characteristic peaks at 2θ = about 15.6° and about 21.9°, for example 15.6 ° ± 0.2° and 21.9° ± 0.2°.

10. The composite material of claim 9, wherein the powder X-ray diffractogram has further characteristic peaks at 2θ = about 24.5°, about 26.9°, about 29.1°, about 31.2°, about 33.1°, about 35.0°, about 36.7°, about 38.4°, about 40.1°, for example 24.5° ± 0.2°, 26.9° ± 0.2°, 29.1° ± 0.2°, 31.2° ± 0.2°, 33.1° ± 0.2°, 35.0° ± 0.2°, 36.7° ± 0.2°, 38.4° ± 0.2°, and 40.1° ± 0.2°.

11. The composite material of any one of claims 8 to 10, wherein the polycrystalline material layer has a surface area of from about from about 10 m2.g-1to about 50 m2.g-1, for example about 15 m2.g-1to about 18 m2.g-1< / sup>.

12. The composite material of any one of claims 8 to 11, wherein the polycrystalline material layer has a surface average roughness of from about 150 nm to about 350 nm.

13. The composite material of any one of claims 8 to 12, wherein the composite material displays an ethanol permeance of from about 0.5 L m-2h-1bar-1to about 25 m-2h-1bar-1< / sup>.

14. The composite material of any one of the preceding claims, wherein the polycrystalline material layer has an average thickness of from about 2 µm to about 10 µm, for example from about 3 µm to about 6 µm.

15. The composite material of any one of claims 1 to 14, wherein the intercrystalline channels have a pore size of less than or equal to about 2 nm, for example, from about 1 nm to about 2 nm, for example about 1.5 nm. 50       16. The composite material of one of claims 1 to 14, wherein the intercrystalline channels have an average pore size of less than or equal to about 1 nm, for example, from about 0.6 nm to about 1 nm, for example about 0.9 nm.

17. The composite material of any one of claims 1 to 14, wherein the intercrystalline channels have an average pore size of less than or equal to about 0.6 nm, for example, about 0.5 nm to about 0.6 nm, for example about 0.55 nm.

18. The composite material of any one of claims 15 to 17, wherein the composite material has a molecular weight cut off (MWCO) of from about 300 Da to about 800 Da.

19. The composite material of any one of the preceding claims, wherein the porous substrate is selected from the group consisting of a ceramic (e.g. alumina), a carbon cloth, a metal, a polymer, and a metal oxide.

20. The composite material of any one of the preceding claims, wherein the porous substrate is provided in the form of a tube, a mesh, a sheet, or hollow fibers (e.g. porous alumina hollow fibers), or other arrangements that are obtainable by the folding of a tube, a mesh, a sheet, or hollow fibers.

21. The composite material of any one of the preceding claims, wherein the porous substrate is a ceramic, for example an α-Al2O3substrate.

22. A method of preparing the composite material of any one of claims 1 to 21, the method comprising the steps of: (a) providing a porous substrate and a colloidal solution comprising an organic solvent, water, a metal hydroxide, and a chelating agent; (b) coating a surface of the porous substrate with the colloidal solution and heating the coated porous substrate at a temperature sufficient to form an amorphous metal hydroxide gel layer or amorphous metal oxide gel layer on the surface of the porous substrate; and (c) contacting the amorphous metal hydroxide gel layer or amorphous metal oxide gel layer with formic acid vapor at a temperature and for a period of time sufficient to form a polycrystalline material layer formed from the MOF, thereby providing the composite material. 51       23. The method according to claims 22, wherein step a) comprises hydrolyzing a metal hydroxide precursor in the presence of water to form the metal hydroxide, followed by peptizing the metal hydroxide in the presence of the chelating agent to form the colloidal solution.

24. The method according of claim 22 or 23, wherein the organic solvent is methanol, ethanol, isopropanol, or a mixture thereof.

25. The method according of any one of claims 22 to 24, wherein the chelating agent is ethyl acetoacetate.

26. The method according to any one of claims 22 to 25, wherein the metal hydroxide in step (a) is an aluminum hydroxide, and the amorphous metal hydroxide gel layer or amorphous metal oxide gel layer in step (b) and (c) is an amorphous aluminum hydroxide gel layer or amorphous aluminum oxide gel layer.

27. The method according claims 26, wherein the MOF is Al(HCOO)3< / sub>.

28. The method according claims 26 or 27, wherein the metal hydroxide precursor is aluminum-tri-sec-butoxide.

29. The method according to claim 26 to 28, wherein step b) comprises coating a surface of the porous substrate with the colloidal solution, and heating the porous substrate at temperature of from about 160 °C to about 200 °C for about 5 hour to about 7 hours.

30. The method according to claim 26 to 29, wherein the step c) comprises of contacting the amorphous aluminium hydroxide gel layer with formic acid vapor at temperature of from about 160 °C to 180 °C for about 10 hour to about 21 hours.

31. A method of removing a solute from an organic solvent, the method comprising passing the organic solvent through the composite material of any one of claims 1 to 21. 52

Citation Information

Patent Citations

  • A simple metal-organic framework for the selective adsorption of carbon dioxide from FLUE gas

    WO2022260592A2

  • Multivariate polycrystalline metal-organic framework membranes for gas separation

    WO2023018372A2