Automated Predictive Chemical Reaction Modelling applied to Gold(I) Catalysis

Computational modelling is a powerful tool to study chemical reactions. Currently, human guidance is nearly always required to avoid the untractable complexity of all a priori possible reaction steps, which consequently greatly limits automated predictive applications. Despite recent advances in the field, predictive reaction modelling without human guidance remains limited. In this work, we present a theoretical framework based on atomic reactivity as well as a "neophile" kinetic model, demonstrating how they enable unbiased automated reaction modelling with molecules of size typically encountered in experimental methodologies. Our framework allows the identification of unlikely or redundant reaction steps based on first principles and previous analyses, while the neophile kinetic model separates crucial reaction intermediates from inconsequential ones. These advances greatly improved modelling efficiency and allowed us to automatically model 17 unimolecular gold(I)-catalyzed reactions of increasing complexity starting only from the reactant and catalyst. In 11 reactions, the experimental product distribution is closely reproduced, with an additional 4 being essentially correct. Our results demonstrate that it is possible to predictively model catalytic reactions without human guidance through a convenient reformulation of the problem. We anticipate that this work will enable the rapid generation of unbiased reaction data. In addition to providing chemical insight, this data could train machine-learning models to manifest mechanism-based chemical reasoning. These models could eventually be combined with self-driving laboratories to form powerful self-teaching, self-correcting autonomous research agents.