Porous materials such as zeolites and metal-organic frameworks (MOFs) have applications as materials for separations, catalysis and storage. In particular, they are promising candidates for use as selective adsorbents for carbon dioxide capture, or in vehicular storage of natural gas. In recent years, thousands of new porous materials have been synthesized, and databases comprising millions of hypothetical materials have been developed. In principle, promising candidate materials can be identified by screening large material databases. However, while computational techniques such as molecular simulation enable accurate prediction of individual material properties, their computational expense prohibit application to very large datasets.
In this presentation, I will discuss our recent efforts in the field of material informatics, introducing tools for accelerating the computational discovery and design of materials. We utilize the Voronoi decomposition to efficiently compute geometric pore space descriptors which enable the representation, searching and screening of materials. Promising carbon dioxide adsorbents are identified two orders of magnitude faster than through molecular simulation, by screening for local binding sites, and new record high surface area MOFs are automatically designed by combining high-throughput geometric analysis and mathematical optimization.