MOFs with the Stability\nfor Practical Gas Adsorption\nApplications Require New Design Rules

Metal–organic frameworks (MOFs) have been widely\nstudied\nfor their ability to capture and store greenhouse gases. However,\nmost computational discovery efforts study hypothetical MOFs without\nconsideration of their stability, limiting the practical application\nof novel materials. We overcome this limitation by screening hypothetical\nultrastable MOFs that have predicted high thermal and activation stability,\nas judged by machine learning (ML) models trained on experimental\nmeasures of stability. We enhance this set by computing the bulk modulus\nas a measure of mechanical stability and filter 1102 mechanically\nrobust hypothetical MOFs from a database of ultrastable MOFs (USMOF\nDB). Grand Canonical Monte Carlo simulations are then employed to\npredict the gas adsorption properties of these hypothetical MOFs,\nalongside a database of experimental MOFs. We identify privileged\nbuilding blocks that lead MOFs in USMOF DB to show exceptional working\ncapacities compared to the experimental MOFs. We interpret these differences\nby training ML models on CO<sub>2</sub> and CH<sub>4</sub> adsorption\nin these databases, showing how poor model transferability between\ndata sets indicates that novel design rules can be derived from USMOF\nDB that would not have been gathered through assessment of structurally\ncharacterized MOFs. We identify geometric features and node chemistry\nthat will enable the rational design of MOFs with enhanced gas adsorption\nproperties in synthetically realizable MOFs.