Analysis of Branching Mechanisms for Post‐Transition State Bifurcation Reactions Using Molecular Dynamics Simulations

ABSTRACT Many organic reactions exhibit post‐transition state bifurcation (PTSB) behavior, in which a single transition state leads to multiple product channels on a high‐dimensional potential energy surface (PES). Product selectivity in these systems is governed by nuclear dynamics in the bifurcation region, which is structurally characterized by valley‐ridge transition (VRT) and valley‐ridge inflection (VRI) points. To identify the physical descriptors that quantitatively predict product branching fractions—and to assess the predictive performance of existing simplified models—we perform extensive classical molecular dynamics (MD) simulations on 18 PTSB cycloaddition systems obtained by systematically varying the tether connecting two reacting fragments. A key finding is that the 18 systems fall into two distinct dynamical regimes: an early‐VRT regime , in which the VRT occurs immediately after the ambimodal transition state on the steep energy descent, and a late‐VRT regime , in which the VRT lies in the flat, low‐energy region of the intrinsic reaction coordinate (IRC). We show that the lateral position of the VRT in the reaction coordinate space, combined with the IRC curvature at the peak after the VRT, accurately predicts the MD branching fractions through a simple two‐parameter linear model. Benchmarking against the VRA, dBond, and RMCF selectivity models reveals that dBond performs well in the late‐VRT regime while RMCF performs well in the early‐VRT regime—a complementary pattern that follows directly from the two‐regime dynamics. These results establish VRT‐based structural descriptors, efficiently accessible via machine‐learning potential energy surfaces, as a unified framework for understanding and predicting PTSB selectivity.