(509cw) Accelerating Ammonia Electrooxidation Catalyst Discovery through Interpretable Machine Learning

The electrooxidation of ammonia to dinitrogen plays a crucial role within the nitrogen transformation cycle. It has many applications including electrochemical sensing of ammonia, wastewater remediation, and direct ammonia fuel cells. While platinum (Pt) based catalysts, specifically terminated with (100) facets, have shown promising activity. They still suffer from a large overpotential (~0.5 V) and surface deactivation. One approach that has been utilized to tackle these issues is the use of platinum based alloys1. While these alloys have shown some improvements compared to pure platinum catalysts, the lack of a clear mechanistic understanding of ammonia electrooxidation at the electrode/electrolyte interface has hindered significant improvements. To tackle this problem we have investigated the reaction mechanism for AOR on various transition metals through Density Functional Theory (DFT). Insights from this mechanistic study are combined with descriptor based kinetic modelling and machine learning tools to screen a large material space to find active and stable electrocatalysts.

Specifically, the thermodynamics and kinetics for the oxidation of NH3 to N2 on various transition metals were mapped out through DFT. These results are then used to rationalize the experimentally observed activity trends for transition metals, as well as explain what makes platinum a unique catalyst for this reaction. Furthermore these calculations are incorporated into a kinetic model to generate a volcano plot. Such an activity map allows us to evaluate catalyst activity based on the nitrogen binding energies. Using our recently developed machine learning framework 2 to predict nitrogen binding energies, we are able to screen ~9000 doped Platinum based alloys and identify new promising candidates. These catalysts are further evaluated for stability and activity through additional DFT calculations. Key mechanistic insights will be highlighted which can then be exploited in new strategies to design more active, selective and robust electrocatalysts for ammonia oxidation.

(1) Li, Y.; Li, X.; Pillai, H. S.; Lattimer, J.; Mohd Adli, N. Ternary PtIrNi Catalysts for Efficient Electrochemical Ammonia Oxidation. ACS 2020.

(2) Wang, S.-H.; Pillai, H. S.; Wang, S.; Achenie, L. E. K.; Xin, H. Infusing Theory into Machine Learning for Interpretable Reactivity Prediction. arXiv [physics.chem-ph], 2021.