Robust Active Site Design of Single Atom Catalysts for Electrochemical Ammonia Synthesis

TL;DR

This paper introduces a computational framework for designing single-atom catalysts for electrochemical ammonia synthesis, emphasizing robustness against parameter variations and providing insights into activity and selectivity trade-offs.

Contribution

The study develops a Bayesian error estimation-based method to predict catalytic activity and analyze selectivity, advancing the computational design of single-atom catalysts.

Findings

Ensemble of limiting potentials predicts activity robustly.

NNH* identified as key descriptor for reaction intermediates.

Catalysts on the strong binding side of the volcano are promising.

Abstract

In this work, we provide a computational methodological framework using the single-atom systems as an example material class for ammonia synthesis that is robust towards parameter selection. Applying this to Pt$_1$/g-C$_3$N$_4$, Ru$_1$/g-C$_3$N$_4$, and Fe$_1$/g-C$_3$N$_4$, we generate ensembles of limiting potentials, using the ensemble of functionals collected via Bayesian Error Estimation Functionals (BEEF), to robustly predict catalytic activity. We then extend this to study the scaling between NRR reaction intermediates and use it to identify that NNH* as the best descriptor for these relations. In addition, a procedure to investigate selectivity is outlined, and a more robust way to analyze the selectivity-activity trade-off is presented. For this single-atom material class, we find choosing catalysts that lie on the strong binding leg of the activity volcano are worth further exploration. Given the ease of integration of the proposed method with minimal additional computational cost, we believe this should become a routine part of analysis workflow for multi-electron electrochemical reactions.