Metal-Organic Frameworks

The MOF design space is astronomically large — combinations of metal nodes, organic linkers, and framework topologies create millions of theoretically possible structures, yet only a tiny fraction have been synthesized and characterized. Current approaches rely on database screening of known or computationally enumerated MOFs, limiting discovery to variations on established reticular chemistry themes while novel topologies with potentially superior properties remain unexplored.

High-throughput computational screening evaluates gas adsorption properties for large databases of hypothetical MOFs but is limited to structures that can be enumerated from known building blocks and topologies. Machine learning models predict adsorption isotherms for MOFs similar to training data but cannot reliably generate novel frameworks in unexplored regions of topology space where the most impactful materials may reside.