Fabrication of Oriented Mixed-Matrix Metal-Organic Framework Membranes for Molecular Separation

Chemical separations are highly energy-intensive and consume around 50% of global industrial energy ^1,2. Membrane-based separation processes are highly energy-efficient compared to conventional cryogenic distillation and/or adsorptive separation. Pure polymer membranes commonly suffer from a trade-off behavior (a relationship between permeability and selectivity) known as Robeson's upper bound^3,4. Mixed-matrix membranes (MMMs), combining the benefits of selective adsorbents (molecular sieving effects and enhanced gas diffusion properties) and polymers (easy solution processability), are regarded as prospect-separating agents towards energy-efficient and environmentally sustainable technologies^5,6,7. Nevertheless, effective translation of adsorbent distinct separation properties into the processable matrix as a form of mixed-matrix membranes remains a significant challenge because of severe agglomeration and sedimentation of fillers inside the polymer matrix and the encountered incompatibility issue between filler and polymer interface. As a result of these challenges, the attainment of highly selective membranes is hampered as well as the membrane's mechanical properties are lessened⁸.

To address the compatibility issue, a porous adsorbent comprising organic moieties is suited to promote adhesion with the polymer matrix. Metal-organic frameworks (MOFs), a tunable class of hybrid solid-state materials consisting of metal nodes or metal clusters coordinated to multifunctional organic linkers^9,10,11, are expected to offer better compatibility with the polymer matrix in contrast to pure inorganic zeolites¹². It is to be noted that MOFs have received considerable attention as excellent adsorbents for selective gas separation^13,14,15.

Various MMMs have been reported using isotropic or near-isotropic fillers (e.g., nanoparticles). The fillers used in addition to MOFs include zeolites, mesoporous silica, and covalent-organic frameworks (COFs)^5,6,7. Nevertheless, only a limited MMMs exhibit a concurrent improvement in selectivity and permeability^16,17,18.

Evidently, introducing fillers with non-isotropic morphologies like nanosheets with a high-aspect-ratio is essential for the construction of high-performance MMMs¹⁹. The relatively high external surface area and reduced micropore diffusion length of nanosheets directionally promote gas diffusion with preserved molecular discrimination, affording a considerable increase in both permeability and selectivity²⁰. Importantly, the high external surface areas of nanosheets would proffer a significant enhancement of the nanosheet-polymer interface compatibility compared to nanoparticles. It is to be noted that Cu-BDC^21,22,23 and NH₂-MIL-53(Al)²⁴ MOF nanosheets have been used as fillers in MMMs, atypically these membranes displayed a moderate selectivity improvement at the expense of permeability, pinpointing the vital importance of a proper alignment of nanosheets within the polymer matrix. Along with the production of high-aspect-ratio nanosheets of high-performing MOFs, it is essential to develop a suitable methodology that can offer the key in-plane alignment of nanosheets inside the polymer matrix. Particularly, the quest for a new synthetic protocol to render selected MOF structures as nanosheets is of prime importance, as many contracted pore MOF structures have unique adsorption and diffusion properties^25,26, but they are not compatible with the conventional exfoliation methods²⁷.

In addition to the MOF's nanosheet morphology and proper in-plane alignment of MOF nanosheets inside the polymer matrix, the judicious selection of MOF filler and polymer pair is also an extremely important parameter to construct a high-performing MMM. It should be noted that these two constituents can have very different intrinsic permeability and selectivity. Therefore if a MMM is fabricated using a relatively low permeable MOF filler with a high permeable polymer, then the MOF will have a minimal impact on gas separation because of gas molecules will more likely diffuse/permeate through the polymer phase. In contrast, a MOF with high permeability, however, the fabricated MMM with very low permeable polymer will lead to the polymer-dominated gas separation.

This manuscript describes a concept and construction of an oriented mixed-matrix metal-organic framework (MMMOF) membrane based on three essential criteria: (i) AlFFIVE-1-Ni [NiAlF₅(H₂O)(pyr)₂] as an advanced molecular sieve filler with a suitable pore structure that facilitates the transport/diffusion of targeted gases (CO₂ and H₂S) while impeding the diffusion of other gases (CH₄ or bigger molecule); (ii) tailoring the AlFFIVE-1-Ni MOF crystal morphology along a defined crystallographic direction to maximize the exposure of pore/channel system (high-aspect-ratio nanosheets); (iii) in-plane alignment of nanosheets inside the polymer matrix and translate the remarkable molecular separation/sieving properties of discrete nanosheets into a macroscopic continuous membrane²⁸.

1. Preparation of (001) oriented AlFFIVE-1-Ni MOF nanosheet (Figure 1)

NOTE: (001) oriented AlFFIVE-1-Ni nanosheet can be synthesized by combining two different precursor-containing solutions (i) metal and ligand solution and (ii) pillar-containing solution under solvothermal reaction²⁸.

1. Prepare metal and ligand solution by dissolving 1.80 g of pyrazine (22.03 mmol) and 2.20 g of nickel acetate tetrahydrate (8.84 mmol) in 9 mL of 2:1 (v/v) mixture ethanol-H₂O by sonication for 3 min.
2. Prepare [AlF₅(H₂O)]^2- pillar solution by dissolving 0.32 g of aluminum hydroxide (4.10 mmol) in 4 mL of 3:1 (v/v) mixture HF-H₂O by sonication for 2 min at room temperature (RT; 25 °C).
3. Combine metal, ligand, and pillars containing solutions and sonicate the mixture for 3 min.
   NOTE: The sonication frequency for steps 1.1-1.3 is 20 kHz.
4. Perform solvothermal reaction at 50 °C for 3 days under a static condition.
   NOTE: Solvothermal reaction can be performed in a 50 mL conical tube in a preheated oven.
5. Collect AlFFIVE-1-Ni MOF nanosheet with a centrifugal force of 1789 x g for 5 min.
6. Wash the MOF nanosheet with 300 mL of water, and wash it with 60 mL of ethanol.
7. Prepare AlFFIVE-1-Ni nanosheet stock solution by dispersing the nanosheet in the required amount of chloroform and adjusting the nanosheet concentration to about 100 mg of MOF/mL solution.
   NOTE: To prevent agglomeration, do not dry the nanosheet before membrane preparation.
8. Prepare AlFFIVE-1-Ni nanoparticles by using 8.85 mmol of [AlF₅(H₂O)]^2- pillar solution instead of 4.10 mmol [AlF₅(H₂O)]^2- pillar solution²⁸.

2. Fabrication of [001]-oriented mixed-matrix metal-organic framework (MMMOF) membrane

1. Use chloroform as a dispersant medium for in-plane alignment of nanosheets inside the polymer matrix (Figure 2).
   NOTE: Dispersant mediums have a critical role in fabricating a defect-free mixed-matrix membrane.
2. Perform solution-casting method to fabricate [001]-oriented MMMOF membranes with different thicknesses.
3. Prepare polymer solution by dissolving 150 mg of dried polyimides (6FDA-DAM) in 2 mL of chloroform by shaking on a mechanical shaker for 3 h at RT.
4. Take the required amount of aliquot from AlFFIVE-1-Ni nanosheets stock solution (step 1.7.) (e.g., 1.02 mL for 40 wt%, 1.50 mL for 50 wt%, and 2.25 mL for 60 wt% MMMOF membranes) in a 25 mL glass vial and add a required amount of chloroform to make a total volume of 2.5 mL.
5. Add polymer solution dropwise into the MOF suspension and stir the mixture at 150 rpm for at least 150 min at 35 °C.
   NOTE: This MOF and polymer mixture is called a dope solution.
6. Pour the dope solution into a glass Petri dish or polytetrafluoroethylene (PTFE) dish on a leveled surface.
   NOTE: The glass Petri dish or PTFE dish must be placed in a glove bag pre-saturated with chloroform vapor for at least 10 min.
7. Keep the dope solution-containing dish in the glove bag overnight at RT (25 °C).
8. Allow slow evaporation of chloroform solvent to self-assemble MOF nanosheets inside the polymer matrix (Figure 3).
   NOTE: MOF nanosheets' in-plane alignment is performed by the slow evaporation of solvent during the membrane preparation²⁸.
9. Peel off the freestanding [001]-oriented MMMOF membrane from the Petri dish using a spatula (Figure 4).
10. Dry the membrane in a vacuum oven at 200 °C for 20 h to remove any residual chloroform.
11. Fabricate AlFFIVE-1-Ni nanoparticles containing mixed-matrix membrane using the same procedure (steps 2.1-2.10) but use AlFFIVE-1-Ni nanoparticles instead of AlFFIVE-1-Ni nanosheets.

3. Characterization of [001]-oriented MMMOF membranes

1. Obtain the cross-section scanning electron microscopy (SEM) image of the membrane to corroborate that the (001) nanosheets are uniformly in-plane aligned (Figure 5A).
   NOTE: Before SEM image measurement, the membrane should be cryogenically fractured in liquid nitrogen to preserve its microstructures. SEM images were obtained using Magellan 400-FEG operating at an acceleration voltage of 1.5 kV and an emission current of 13 pA.
2. Evaluate X-ray diffraction (XRD) of the membrane to ascertain the crystallographic orientation of (001) nanosheet in the polymer matrix (Figure 5B).
   NOTE: [001]-oriented membrane is supposed to display only two major Bragg diffractions from the two crystallographic planes ([002], [004]) of the AlFFIVE-1-Ni structure. The XRD data collection was performed on a diffractometer (Cu K_α = 1.54056 Å) in Bragg-Brentano θ-θ geometry with the following settings: operating power: 40 kV/40 mA; automatic divergence slit (irradiated length = 0.6 mm), sample holder: flat plate sample holder; temperature: RT; method: continuous-count method (3° min^-1); 2 range: 4°-40°.
3. Perform focused ion beam SEM (FIB-SEM) image analyses of the oriented membrane to verify the highly aligned nature of nanosheets and the membrane's internal microstructure on an extensive area (Figure 6).
   NOTE: FIB-SEM is an excellent characterization technique to identify MOF-polymer interface compatibility.
   1. Create about 35 µm wide trench on the top surface of the MMMOF membrane, and mask the remaining area by deposited platinum (Figure 6A).
      1. Sputter coat the membrane with ⁓10 nm of Iridium. Before milling, deposit a protective layer of Pt (0.3-0.4 µm thickness) on the areas of interest of the specimen.
      2. Then, use the gallium (Ga⁺) ion beam FIB working at 5 kV and 75-100 pA to create a trench on the top surface of the membrane.
   2. Slice and mill away the membrane with a minimum thickness of 50 nm by the FIB, operating at about 5 kV and 25 pA (Figure 6B).
   3. Collect between 100 and 300 individual SEM micrographs of the consecutive cross-sections subjected to milling at a magnification range of 12,000x-45,000x, with a secondary electron detector at ~1 kV (Figure 6C).
   4. Align the stacking of images to an external feature on the membrane surface using a cross-correlation algorithm.
4. Analyze the relative viscosity changes of nanosheets-polymer suspension (dope solution) with respect to the stirring time (Figure 7).
   NOTE: Rheological characterization is rather simple yet effective in elucidating the enhanced interaction between MOF nanosheets and polymers.
5. Collect stress-strain curves of membranes at different MOF loadings and calculate Young's modulus and elongation strain from stress-strain curves.
   NOTE: Mechanical properties of the membrane can be assessed using stress-strain analysis.
6. Assess the MOF nanosheet loading in the oriented membrane by thermogravimetric analysis method.
   1. First, activate about 10 mg of nanosheet powder under N₂ purging at 200 °C for 135 min to ensure activation.
   2. Increase the temperature up to 800 °C at a heating rate of 10 °C·min⁻¹ under O₂ flow and hold the temperature at 800 °C for 2 h. Compare the residual metal oxide mass to the initial activated mass of the MOF²⁸.
   3. Use the same protocol (steps 3.6.1 and 3.6.2) for MOF nanosheet-containing membranes to determine the MOF loading.
      NOTE: From the residual metal oxide mass after heating at 800 °C for 2 h under O₂ environment, the content of MOF nanosheets in the membrane can be determined.

4. Assess the single and mixed-gas separation properties of oriented membranes

1. Evaluate single/pure gas permeation measurements using a homemade variable-pressure constant-volume method²⁸.
   NOTE: Single gas permeation condition typically at 35 °C and 2 bar feed pressure for He, H₂, CO₂, O₂, N₂, CH₄, C₂H₄, C₂H₆, C₃H_6, and C₃H₈. A homemade gas separation instrument was used. Refer to our previously published work²⁸ for the details of the setup.
   1. Cut the membrane using a mechanical punch into a circle/coin shape about 2 cm² in size.
   2. Mount the membrane on the permeation cell, mask it with adhesive aluminum tape, and seal it with epoxy resins.
      NOTE: The membrane sealing using epoxy resin should be carefully performed to avoid any undesirable pinholes.
   3. Keep a selected area of the membrane exposed for gas permeation.
   4. Measure the accessible area of the membrane for gas permeation using a scanner and measure the membrane thickness using a depth gauge.
   5. Evacuate the permeation system at 35 °C for 12 h before performing any permeation test.
      NOTE: This step helps to remove any adsorbed moisture during membrane sealing.
   6. Feed the gas at 2 bar (upstream pressure) and wait for 20 min to equilibrate the permeation system temperature.
   7. Calculate the single gas permeability using the slope of downstream pressure vs. time (dp/dt) at the steady state condition by the following equation:
           (1)
      NOTE: P refers to the gas permeability of a membrane in barrer (1 barrer = 10^-10 cm³(STP)·cm^-2·s^-1·cmHg^-1), V is the volume of the downstream chamber (cm³), l is the membrane thickness (cm), p₀ is the upstream pressure (cmHg), A refers to the effective area of membrane (cm²), R is the gas constant (0.278 cm³ cmHg cm^-3(STP) K^-1), T is the operating temperature (K).
   8. Obtain the ideal selectivity (S_A/B) of gas A to gas B by using the following equation:
          (2)
      NOTE: P_A and P_B denote the permeability of gases A and B, respectively.
2. Evaluate mixed-gas permeation measurements.
   1. Use the binary gas mixture (CO₂/CH₄:10/90; 20/80 and 50/50) and/or ternary gas mixture (H₂S/CO₂/CH₄:1/9/90; 2/18/80 and 5/5/90) for mixed-gas separation tests.
   2. Use modified single-gas permeation equipment for mixed-gas permeation measurement, which must be integrated with micro-gas chromatography (Agilent 490 Micro-GC)²⁸.
      NOTE: Mixed-gas feed pressure can be varied from 2-30 bar and temperature 25-100 °C.
   3. Maintain the stage cut (the flow rate ratio of permeate to feed) below 1%.
      NOTE: The gas concentration polarization on the upstream side of the membrane can be circumvented in this way, and it will also maintain the constant driving force across the membrane during the gas permeation study.
   4. Allow the gas mixture to permeate through the membrane until a steady-state permeation rate is reached, typically above 5-time lags.
   5. Evacuate the permeate mixture and then collect the gas mixture under steady-state conditions.
   6. Dilute the permeate gas with He gas and analyze the gas mixture with a Micro-GC.
   7. Calculate the mixed gas permeability of component i (P_i) using the following equation.
          (3)
      NOTE: y and x are the mole fractions in permeate and feed, respectively, and the p₀ is the feed pressure of component i.
   8. Analyze the mixed-gas selectivity (S_i/j) using the ratio of the permeability of two components, i and j.
          (4)

Oriented nanosheet fillers (Figure 1) intrinsically possess several advantages in MMMs compared to nanoparticles. The relatively large external surface area and reduced micropore diffusion distance of nanosheets directionally promote gas diffusion with preserved molecular discrimination (Figure 2). The relatively large external surface area of nanosheets compared to nanoparticles proffer a greater enhancement of nanosheet-polymer interface compatibility, enabling high nanosheets/filler loading. We noticed that very high filler loading is possible (50-60 wt%) using nanosheet morphology with excellent physical stability of the membranes (Figure 3 and Figure 4).

In contrast, the membranes constructed using nanoparticles maintain relatively good mechanical properties at relatively low filler loadings; however, further increasing the filler loading to about 35 wt% produces very fragile and/or defective membranes that instantly break into pieces upon mechanical punching18. This is a significant drawback in nanoparticles containing MMMs for their successful deployment without special precautions. The ability to increase the nanosheet loading ratio to polymers offered a great opportunity to closely mimic the allied pure MOF membrane. The outstanding separation performance of the membranes can be achieved if the agglomeration, sedimentation, and random orientation of nanosheets within the polymer matrix are avoided at high filler loadings (Figure 5 and Figure 6). Using nanosheets morphology, no such adverse effects are observed. The relative viscosity changes of MOF/polymer suspension and polymer solution with respect to time were measured (Figure 7). The higher viscosity in MOF nanosheets/polymer suspension implies an enhanced MOF nanosheets/polymer interaction.

Filler's crystallographic orientation control is another key issue in MMM for enhanced gas separation (Figure 3). Nevertheless, nanoparticle orientation control is not a critical issue for MOFs containing three-dimensional (3D) channel/pore systems17; however, it does severely influence MOFs having two-dimensional (2D) or one-dimensional (1D) channel/pore28. We noticed that the incorporation of similar amounts of nanoparticles or nanosheets (>35 wt%) with 1D channel-type MOF into the polymer matrix resulted in a significant difference in gas permeability. Specifically, nanoparticles containing membranes exhibited about half of lowering gas permeability than oriented nanosheet membranes. It can be attributed to the abundant non-permeable facets of nanoparticles oriented perpendicular to the gas diffusion direction28. These results confirm the importance of MOF morphology, channel orientation, and their desirable alignment in the polymer matrix (Figure 2).

Despite a large number of MOFs and other filler-containing MMMs reported, nevertheless, only a few MMMs demonstrated concurrent enhancement of selectivity and permeability. The enhancement of separation performance of those membranes is moderate and often below the trade-off curve, except few examples16,17,18,29. Distinctly, the gas separation performance of oriented MMMOF membranes is far better than nanoparticle-containing MMMs28. Oriented MMMOF membranes often demonstrate 3 to 4-fold enhancement of both selectivity and permeability simultaneously, as compared to associated pure polymers.

Figure 1: SEM and scanning transmission electron microscopy (STEM) image of nanosheet. (A) SEM image of nanosheet. Inset: schematic illustrations of nanosheet showing various percentages of 001, 110, and 1-10 facets. (B) Low-resolution scanning transmission electron microscopy (STEM) image of nanosheet. Inset: pictures showing the Tyndall effect on nanosheets using a green laser. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 2: In-plane aligned (001) AlFFIVE nanosheets embedded in the polymer matrix. Schematic illustrations of in-plane aligned (001) AlFFIVE nanosheets embedded in the polymer matrix and an efficient H2S/CO2/CH4 separation process through 1D channels of nanosheets. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 3: Schematic illustration of 'slow evaporation-induced in-plane alignment' of (001) nanosheets in the polymer matrix. Slow evaporation of solvent permit nanosheets' self-assembly to the minimum energy configuration. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 4: Photographs of [001]-oriented membranes. The pictures show the flexible nature of the oriented membrane. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 5: SEM and XRD. (A) Cross-section SEM image of [001]-oriented membrane. (B) XRD patterns of [001]-oriented membrane, nanosheet crystallite, and polymer. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 6: FIB-SEM analyses of [001]-oriented MMMOF membrane. (A) Top view SEM image of the membrane selected for FIB analyses. The white line frame indicates the selected area for further analysis. (B,C) Cross-sections SEM images of the oriented membrane (001)-nanosheets (bright motifs) are embedded inside the polymer matrix (dark grey). (D) Schematic illustration of nanosheet arrangement in the polymer matrix. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

Figure 7: Rheological characterization. Relative viscosity changes of MOF nanosheets/polymer, MOF nanoparticle/polymer, and polymer suspension with respect to time. This figure has been reprinted with permission from Datta et al.28. Please click here to view a larger version of this figure.

The synthesis of oriented MOF nanosheets with a high-aspect-ratio is remarkably challenging. Unlike other applications, the MOF nanosheets for gas separation must be synthesized without surfactant and/or template to avoid any undesired substance on the surface of MOF nanosheets. In addition, the channel/pores of nanosheets should be oriented out of the plane direction to directionally promote gas diffusion. Each MOF structure has a different crystal growth mechanism, therefore same synthesis procedure can not apply to all MOF nanosheet synthesis. For a given MOF, the nanosheet synthesis procedure should be developed/optimized. Interestingly, we have attained several fluorinated MOFs that can be synthesized as high-aspect-ratio nanosheets and fabricated oriented MMMOF membranes using this protocol.

The enhanced performances attained using oriented MMMOF membranes can be reorganized by identifying the importance of the three essential standards described earlier. The attainment of extremely high nanosheet loading (>60 wt%) is exclusively responsible for the outstanding separation performance. Nanosheets here act as molecular carriers, selectively enhancing the diffusion of gases based on their kinetic diameter and/or molecular properties through the oriented membranes. In reality, this centimeter-scale bendable [001]-oriented membrane can be considered as a piece of a bendable crystal where thousands of (001) nanosheets are uniformly in-plane aligned, and the gaps between adjacent nanosheets are filled with polymer.

This research finding confirms the great potential of tuning MOF crystal morphology into high-aspect ratio nanosheets, which permits the looked-for orientation control of the 1D channels inside the polymer matrix and offers unique opportunities to remarkably enhance the performance of the oriented membrane²⁸. The construction of a predesigned nanosheet MMM by implementing the newly designed strategy resulted in a MMM that fulfills all the predefined criteria and offers an outstanding selectivity-permeability trade-off. We believe this newly conceived and tested approach offers a paradigm shift in our understanding of MMMs and can certainly promote the development of next-generation high-performing MMMs for highly challenging gas separations. These unprecedented findings, noted for the nanosheets/6FDA-polyimides matrices, pave the way for a comprehensive deployment of the presented approach to cover a broader range of other matrices where the concurrent enhancement of selectivity and permeability is desired.