(389g) Combining Physics Driven and Graph Theory-Based Methodologies for Modeling Complex Heterogeneous Electro-Catalytic Surfaces

Electrochemical reactions, which use electrons as energy carriers to either store or harness energy, are an important component in building a sustainable energy future.[1,2] However, unlike their thermal counter parts, heterogeneous electrocatalytic reactions are not well understood with only a few pertaining to oxygen, hydrogen and chlorine electrochemistry currently commercialized.[1] The disparity foreshadows the need to develop an in-depth fundamental understanding of these reactions, which take place at a solid-liquid interface under an applied potential. To gain such an understanding of the atomic scale phenomenon, ab-initio Density Functional Theory (DFT) is a widely used tool, and its application has led to successful modeling of many catalytic systems of varying degrees of complexity.[2,3] Recently, growing computational power has enabled the extension of DFT analyses to understand increasingly complex electrocatalytic systems, which include a combination of different phenomenon: high adsorbate coverages, multidentate adsorbates, explicit solvent effects, multi-elemental alloys, and defected catalyst morphologies. Taken together, these complexities result in (i) a large number of possible atomic configurations (O>10⁶), creating a need for the development of algorithmic approaches to sample large phase spaces, and (ii) a need to develop systematic approaches to understand the complex physics of the electrochemical interface formed between solvent and the catalytic surface.

To account for these phenomena, we present a generalized workflow. We first utilize in-house python-based graph theory algorithms to systematically sample and identify relevant catalytic models, and then use them to understand the underlying reaction mechanism at the electrochemical interface under an explicit solvent environment.[4,5,6,7] We demonstrate the utility of such an approach to analyze the reaction mechanism of NO* electroreduction on Pt₃Sn(111) surface, which is an important reaction for water remediation.[4,7] Utilizing the graph-theory based algorithms, initially, all possible unique catalytic configurations of NO* on Pt₃Sn(111) surface are enumerated. The enumerated configurations are then used to devise a machine learning based surrogate model to successfully sample the large number of atomic configurations and identify the most relevant ones.[5,6] Using such an algorithmic approach, we show that the most relevant configurations can be identified using only ~10% of the total possible configurations (350 configurations vs. 3500 possible unique configurations). More importantly, the method requires minimal human input to draw the underlying insights.

A thorough analysis of the reaction mechanism under explicit solvent conditions, utilizing the identified configuration, then shows that the high coverages of NO* and the presence of Sn in Pt₃Sn(111) surface plays a crucial role in promoting the experimentally observed selectivity to hydroxyl amine (NH₃)⁺OH in the low potential region.[4] In summary, we demonstrate the utility of a combined approach, utilizing Python-based workflows[5,6] and detailed physics driven mechanistic analysis of the reactions taking place at the electro-chemical interface, to gain an in-depth atomistic understanding into complex heterogeneous electro-catalytic systems. Such approaches now form the basis to study even more complex electro-catalytic reactions containing complex feedstock, such as biomass-based reactants, and complex morphologies such as three phase boundaries.

References

[1] Seh, Z. W. et al. Science 355, eaad4998 (2017).

[2] Nørskov, J. K., Bligaard, T., Rossmeisl, J. & Christensen, C. H. Nature Chemistry 1, (2009), 37–46.

[3] J. Greeley, Annual Review of Chemical and Biomolecular Engineering 7 (2016) 605–635.

[4] S. Deshpande*, J. Greeley, ACS Catalysis 10 (2020) 9320-9327.

[5] S. Deshpande*, T. Maxson*, J. Greeley, NPJ Computational Materials 6 (2020) 79.

[6] P. Ghanekar*, S. Deshpande*†, J. Greeley†, Accepted, Nature Communications.

[7] Yang, J., Duca, M., Schouten, K. J. P. & Koper, M. T. M. Journal of Electroanalytical Chemistry 662, 87–92 (2011).