Automatic Chemical Reaction Discovery

Recent advances in computational chemistry have opened the door toward detailed understanding of challenging chemical problems, such as analyzing complex reaction networks. Significant progress toward this end has been achieved using ab initio methods in conjugation with advanced potential energy surface (PES) walking algorithms (e.g., freezing string method, growing string method, and nudged elastic band method), which automatically locate transition states and/or intermediates of reactions of interest. However, because most of such algorithms require prior knowledge of the reaction, i.e., the products formed or the chemical bonds altered during the reaction, exploring unknown reactions using computer simulations without any human intervention is still very challenging.

To address this issue, we are developing a fully automated scheme for new reaction discovery using ab initio calculations. This effort involves assessing the efficiency and reliability of commonly used transition state search algorithms and developing a scheme to systemically generate all viable reactions of a given reactant.

For more details of this project, please see our previous publication: Suleimanov, Y. V.; Green, W. H. J. Chem. Theory Comput. 2015, 11 (9), 4248–4259.