(435c) Accelerating Nanoporous Materials Discovery for Gas Storage and Separations via Advanced Simulation and Materials Characterization

Porosity is ubiquitous, from naturally formed systems (e.g., shale rocks, wood, and human bones) to synthetic materials (e.g., metal-organic frameworks and molecular cages). The ability of nanoporous materials to adsorb various molecules of interest makes them strong candidates for applications in energy storage, chemical separation, sensing, catalysis, and nano-manufacturing. Finding the top-performing candidates, however, is a daunting task due to the exponentially increasing chemical and structural variables in the materials design space. Physics-based simulations and machine learning (ML) have been proven highly efficient in navigating this virtually infinite space.

In this talk, I will first show how the ML force field significantly improves the accuracy of the grand canonical Monte Carlo (GCMC) simulations for modeling challenging gas adsorption in metal-organic frameworks (MOFs), compared to the classical force field. Then I will present a novel concept of “pore graphs”, a compact yet comprehensive way to characterize and represent nanoporous materials. The pore graph representation is flexible and informative to retain any cavity information in a porous structure. We envision this new computational characterization method, coupled with ML, will facilitate the discovery and design of advanced nanoporous materials to address global challenges in energy and sustainability.