Multi-level computational screening of anion-pillared metal-organic frameworks for propane and propene separation

Accelerated Discovery of Metal–Organic Frameworks for CO2 Capture by Artificial Intelligence

“…Most of the ML studies in this field primarily center around removal of CO 2 from flue gas (CO 2 /N 2 ), natural gas (CO 2 /CH 4 ), and precombustion gas (CO 2 /H 2 ) mixtures since many computational research groups generated very large amounts of data for the adsorption of these three gas mixtures by molecular simulations which were then used in training ML models. Separation of CO 2 from other gases such as acetylene (C 2 H 2 ), ethylene (C 2 H 4 ), ethane (C 2 H 6 ), propane (C 3 H 8 ), and hydrogen sulfide (H 2 S) is also critical for industrial and environmental reasons, but the literature data are limited for them. − This can be attributed to several reasons such as (i) the challenge in computational modeling of large, polar gas molecules in addition to the difficulty of accurately defining the interactions in multicomponent gas mixtures and (ii) the experimental challenges posed by toxic or flammable gases like hydrogen sulfide and acetylene when testing CO 2 capture. The available literature on processes like pervaporation indicates that performing such molecular simulations for these systems, especially for separating very large molecules, is also computationally expensive. , Thus, the usage of low-data effective AI-based models can be useful to study MOFs as adsorbents and membranes in this regard.…”