Accelerated Entropic Path Sampling Elucidates Entropic Effects in Mediating the Ambimodal Selectivity of NgnD-Catalyzed Diels–Alder Reaction

Reaction dynamics trajectory simulations have been conducted to predict the product ratio of reactions with post-transition state bifurcation. However, it remains unknown how the entropy of reactive species along the reaction path mediates ambimodal selectivity. Here, by leveraging deep generative model, we developed an accelerated entropic path sampling approach that evaluates the change of entropy along the post-transition-state reaction path for each product using merely a few hundred reaction dynamic trajectories. The new method, called bidirectional generative adversarial network - entropic path sampling (BGAN-EPS), can enhance the estimation of probability density functions of molecular configurations by generating pseudo-molecular configurations that are statistically indistinguishable from the true data. The method was tested using cyclopentadiene dimerization as a model reaction, in which we reproduced the reference entropic profiles (derived from 2,480 trajectories) using merely 124 trajectories. We further applied BGAN-EPS method to NgnD-catalyzed Diels–Alder reaction to investigate the entropic origin behind its ambimodal selectivity. The results show that the ambimodal preference towards the formation of the [6+4]-adduct over the [4+2]-adduct is contributed by both energetic and entropic forces.