Prediction of Transition State Structures of General Chemical Reactions via Machine Learning

Abstract The elucidation of transition state (TS) structures is ssential for understanding the mechanisms of chemical reactions and exploring reaction networks. Despite advances in computational approaches, TS searches remain still a challenging problem due to the difficulty of constructing an initial structure and heavy computational costs. Herein, a novel machine learning (ML) model for predicting TS structures of general organic reactions is proposed. The proposed model derives interatomic distances of a TS structure from atomic pair features reflecting reactant, product, and linearly interpolated structures. The model shows excellent accuracy, particularly for the atomic pairs where bond formation or breakage occurs. The predicted TS structures result in a high success ratio (93.8%) of quantum chemical saddle-point optimizations, and 88.8% of the optimization results have energy errors of less than 0.1 kcal/mol. Additionally, as a proof-of-concept, exploring multiple reaction paths of an organic reaction is demonstrated with the ML inferences. I envision that the proposed approach will aid the construction of initial geometry for TS optimization and reaction path explorations.