Predicting Gas Separation through Graphene Nanopore Ensembles with Realistic Pore Size Distributions.

The development of nanoporous single-layer graphene membranes for gas separation has prompted increasing theoretical investigations of gas transport through graphene nanopores. However, computer simulations and theories that predict gas permeances through individual graphene nanopores are not suitable to describe experimental results, because a realistic graphene membrane contains a large number of nanopores of diverse sizes and shapes. With this need in mind, here, we generate nanopore ensembles in silico by etching carbon atoms away from pristine graphene with different etching times, using a kinetic Monte Carlo algorithm developed by our group for the isomer cataloging problem of graphene nanopores. The permeances of H2, CO2, and CH4 through each nanopore in the ensembles are predicted using transition state theory based on classical all-atomistic force fields. Our findings show that the total gas permeance through a nanopore ensemble is dominated by a small fraction of large nanopores with low energy barriers of pore crossing. We also quantitatively predict the increase of the gas permeances and the decrease of the selectivities between the gases as functions of the etching time of graphene. Furthermore, by fitting the theoretically predicted selectivities to the experimental ones reported in the literature, we show that nanopores in graphene effectively expand as the temperature of permeation measurement increases. We propose that this nanopore "expansion" is due to the desorption of contaminants that partially clog the graphene nanopores. In general, our study highlights the effects of the pore size and shape distributions of a graphene nanopore ensemble on its gas separation properties and calls into attention the potential effect of pore-clogging contamination in experiments.