Predicting the whole process of a chemical reaction while solving kinetic equations presents an opportunity to realize an on-the-fly kinetic simulation that directly discovers chemical reactions with their product yields. Such a simulation avoids the combinatorial explosion of reaction patterns to be examined by narrowing the search space based on the kinetic analysis of the reaction path network, and would open a new paradigm beyond the conventional two-step approach, which requires a reaction path network prior to performing a kinetic simulation. The authors addressed this issue and developed a practical method by combining the artificial force induced reaction method with the rate constant matrix contraction method. Two algorithms are available for this purpose: a forward mode with reactants as the input and a backward mode with products as the input. This article first numerically verifies these modes for known reactions and then demonstrates their application to the actual reaction discovery. Finally, the challenges of this method and the prospects for ab initio reaction discovery are discussed.
Keywords: kinetic simulation; organic synthesis; potential energy surface; quantum chemical calculation; reaction path network.