Interfacial Cooperation-Driven Volcano Shifts in Ammonia Decomposition Revealed by Reaction-Aware Machine Learning Simulations

Ammonia decomposition is governed by an intrinsic trade-off between NH3 dehydrogenation and N-N coupling, giving rise to the well-known volcano relationship that has long positioned ruthenium (Ru) near the activity maximum. However, whether this relationship can be reshaped through catalyst design remains an open question. Here we show that metal-oxide interfacial cooperation qualitatively alters the reaction pathway of ammonia decomposition. Using reaction-aware machine learning simulations, we find that on MgO(111)-supported metal nanoparticles, NH3 dehydrogenation, one of the most kinetically demanding steps, occurs predominantly at the metal-MgO(111) interface with substantially reduced activation barriers, while N-N coupling remains governed by the metal surface. This spatial separation of reaction bottlenecks relaxes the classical single-site trade-off and drives a shift of the activity volcano toward weaker nitrogen-binding metals. As a consequence, Ni/MgO(111) and Cu/MgO(111) emerge near the new activity optimum and are predicted to rival or surpass Ru/MgO(111), as supported by density functional theory calculations and microkinetic modeling. Our results establish interfacial cooperation as a general design principle for tuning volcano relationships in heterogeneous catalysis and for activating earth-abundant metals beyond conventional single-site paradigms.