Systematic Construction of Chemical Reaction Networks: An Automated Nanoreactor Approach Powered by Integrated Tempering Sampling

Abstract

The systematic construction of complex reaction networks from given reactants remains a fundamental challenge in computational chemistry. To address this, we introduce a fully automated and generally applicable workflow centered on integrated tempering sampling (ITS) within a nanoreactor framework. The protocol integrates ITS (a collective-variable-free enhanced sampling method) at the semiempirical GFN2-xTB level for reaction-space exploration, a hidden Markov model (HMM) for reaction-event identification, and high-level quantum-chemical calculations for mechanistic refinement, enabling autonomous discovery and validation of reaction pathways without any prior mechanistic input. Using formaldehyde–ammonia and HCN–water mixtures as case studies, the method successfully constructs extensive reaction networks. Notably, in the HCN–water system, the generated network includes prebiotically relevant heterocycles and peptide-like species. This approach provides an efficient and broadly applicable route for systematic reaction-network discovery, shifting the paradigm from the hypothesis-driven verification of reaction mechanism to the automated exploration of unknown chemical spaces.