Ultrapermeable 2D-channeled graphene-wrapped zeolite molecular sieving membranes for hydrogen separation

The efficient separation of hydrogen from methane and light hydrocarbons for clean energy applications remains a technical challenge in membrane science. To address this issue, we prepared a graphene-wrapped MFI (G-MFI) molecular-sieving membrane for the ultrafast separation of hydrogen from methane at a permeability reaching 5.8 × 106 barrers at a single gas selectivity of 245 and a mixed gas selectivity of 50. Our results set an upper bound for hydrogen separation. Efficient molecular sieving comes from the subnanoscale interfacial space between graphene and zeolite crystal faces according to molecular dynamic simulations. The hierarchical pore structure of the G-MFI membrane enabled rapid permeability, indicating a promising route for the ultrafast separation of hydrogen/methane and carbon dioxide/methane in view of energy-efficient industrial gas separation.

Microtrac MRB (BELSORP MAX) apparatus (Fig. 2C). The sample was degassed at 523 K for 3 h prior to the adsorption measurements.

#7 Gas permeance
Gas permeability measurements were carried out using a custom-made apparatus. Permeation tests for graphene-wrapped MFI membranes were carried out by using a pressure-driven method at 298 K. A membrane was fixed into a membrane module using O-rings. The pressure on the feed and permeate sides were set at 200 kPa and 100 kPa, respectively, and the permeate flow rate was measured using a bubble flow meter. In the case of separation tests for equimolar mixtures of H2/CH4 and CO2/CH4 the permeate flow of the mixture was measured by a bubble flow meter, and the composition of the permeate gas was determined using gas chromatography with a thermal conductivity detector (GC-TCD). The measurements were conducted after 1h of purging the equimolar H2/CH4 and CO2/CH4 mixtures at a feed pressure of 200 kPa, assuming that the H2, CO2, and CH4 were pre-adsorbed on the G-MFI membrane. The steady permeabilities and selectivities were measured.

#8Diffusion coefficient estimation by solution diffusion transport
The diffusion coefficient is related to the permeance by the well-known permeation equation (45): where J is the molar flux of the permeate gas (mol m -2 s -1 ), D is the diffusion coefficient (m 2 s -1 ), S is the solubility coefficient (mol m -3 Pa -1 ), ∆p is the transmembrane pressure (Pa), and δ is the thickness of the membrane (m).
The J to ∆p ratio is equal to the permeance P (mol m -2 s -1 Pa -1 ), and the diffusion coefficient is given by the following equation: The permeance of H2 through the G-MFI membrane was 1.3 × 10 -5 mol m -2 s -1 Pa -1 and the thickness was 150 µm. The solubility coefficient of H2 in the G-MFI membrane was determined using a molecular dynamics simulation model based on Henry's law: where c is the concentration of H2 in the G-MFI membrane (mol m -3 ) and p is the pressure of H2 over the membrane of 8.5 MPa. The concentration of H2 was calculated from the number of H2 molecules in the upper part of the G-MFI membrane model using MD simulations. The average number of H2 molecules in the upper part of the membrane model was 4.0 after 3 ns of simulation time. The volume of the upper part of the G-MFI membrane model was 3.8 × 10 -27 m 3 , and a solubility coefficient of 2.0 × 10 -4 mol m -3 Pa -1 was obtained using Equation (3). A diffusion coefficient of 9.5 × 10 -2 cm 2 s -1 was calculated using Equation (2).
The diffusion coefficient for CH4 obtained using the aformentioned method was 1.1 × 10 -4 cm 2 s -1 . The ratio of the diffusion coefficients of H2 and CH4 was calculated as the diffusivity selectivity. The ratio of the solubility coefficients of H2 and CH4 was calculated as the solubility selectivity, as described in the main text.     (Table S4).         Table S3. Porosity parameters of G-MFI and MFI membranes and their powders. We obtained the specific surface area (SBET) from the BET plot (66) and the specific surface area (Sαs), micropore volume (VMicro.) and external surface area (Sex.) by subtracting pore effect method from high resolution αS-plot (67).