Machine Learning Design of Single-Atom Catalysts for Nitrogen Fixation

First-principlescalculations have been combined with machine learningin the design of transition-metal single-atom catalysts. Readily availabledescriptors are selected to describe the nitrogen activation capabilityof metals and coordinating atoms. Thus, a series of V/Nb/Ta-N- x single-atom catalysts are screened out aspromising structures upon considering the stability, activity, andselectivity investigated computationally. Furthermore, by using thegradient boosting regression algorithm, an accurate prediction ofthe hydrogenation barriers for the nitrogen reduction reaction (NRR)is achievable, with a root-mean-squared error of 0.07 eV. The integrationof high-throughput computation and machine learning constitutes apowerful strategy for the acceleration of catalyst design. This approachfacilitates the rapid and accurate prediction of the NRR performanceof more than 1000 single-atom catalyst structures. Moreover, the currentwork provides further insights by elaborately correlating the structureand performance, which may be instructive for both the design andapplication of vanadium-group catalysts.