Abstract

Because of their high surface areas, crystallinity, and tunable properties, metal–organic frameworks (MOFs) have attracted intense interest as next-generation materials for gas capture and storage. While much effort has been devoted to the discovery of new MOFs, a vast catalog of existing MOFs resides within the Cambridge Structural Database (CSD), many of whose gas uptake properties have not been assessed. Here we employ data mining and automated structure analysis to identify, "cleanup," and rapidly predict the hydrogen storage properties of these compounds. Approximately 20 000 candidate compounds were generated from the CSD using an algorithm that removes solvent/guest molecules. These compounds were then characterized with respect to their surface area and porosity. Employing the empirical relationship between excess H2 uptake and surface area, we predict the theoretical total hydrogen storage capacity for the subset of ∼4000 compounds exhibiting nontrivial internal porosity. Our screening identifies several overlooked compounds having high theoretical capacities; these compounds are suggested as targets of opportunity for additional experimental characterization. More importantly, screening reveals that the relationship between gravimetric and volumetric H2 density is concave downward, with maximal volumetric performance occurring for surface areas of 3100–4800 m2/g. We conclude that H2 storage in MOFs will not benefit from further improvements in surface area alone. Rather, discovery efforts should aim to achieve moderate mass densities and surface areas simultaneously, while ensuring framework stability upon solvent removal.