Machine learning (ML) became a premier tool for modeling chemical processes and materials properties. ML interatomic potentials have become an efficient alternative to computationally expensive quantum chemistry simulations. In the case of reactive chemistry designing high-quality training data sets is crucial to overall model accuracy. To address this challenge, we develop a general reactive ML interatomic potential through unbiased active learning with an atomic configuration sampler inspired by nanoreactor molecular dynamics. The resulting model is then applied to study five distinct condensed-phase reactive chemistry systems: carbon solid-phase nucleation, graphene ring formation from acetylene, biofuel additives, combustion of methane and the spontaneous formation of glycine from early-earth small molecules. In all cases, the results closely match experiment and/or previous studies using traditional model chemistry methods. Importantly, the model does not need to be refit for each application, enabling high throughput in silico reactive chemistry experimentation.
Read full abstract