Reaction Discovery in Porous Materials Using Periodic Nanoreactor Molecular Dynamics

Understanding catalytic processes is essential for advancing energy-efficient molecular transformations. In heterogeneous catalysis, porous materials such as zeolites play a central role due to their structural complexity and large surface area. Here, we present a periodic ab initio nanoreactor molecular dynamics (NMD) approach to investigate the reaction network of selective catalytic reduction (SCR) of NO over copper-exchanged chabazite zeolites. This method enables autonomous discovery of both established and previously unreported pathways, including a water-assisted tautomerization mechanism that facilitates $N_2$ formation and a novel radical-driven route to $N_{2}O$ . Notably, NMD simulations also capture reactivity involving Brønsted acid sites of the zeolite framework. By using automated reaction detection, a comprehensive reaction network was constructed, which elucidates the formation of both desired and undesired products. Refinement of the reaction path including free energy corrections by computing the phonon spectrum allows to make quantitative statements about the discovered reactions. The results of this work provide insight into the side-reactions of SCR, and also demonstrate the versatility of NMD for agnostic reaction discovery of complex systems such as heterogeneous catalysis.