Hydrogen Adsorption on CoCuFeMnNi High Entropy Alloy Surface: A Combined Density Functional Theory and Machine Learning Study

抄録

In the past few years, there is increasing interest about high entropy alloys (HEA) as promising materials for various applications. High entropy alloys, a new class of alloys, are composed of five or more component atoms with almost similar compositions. These alloys form single phase solid solution due to the increased configurational entropy of such a multicomponent system [1]. With this background, HEAs offer the opportunity to discover materials with appropriate functionality for a specific application. It is possible to combine cheap elements and tune their properties. However, it also entails that investigating these alloys would be a challenging task due to the numerous elements in the periodic table and the possible combinations of such elements to form HEA.

In this study, we aimed to evaluate the reactivity of HEA which is composed of transition metals that are abundant in Southeast and East Asia regions for its potential application as a catalyst. Specifically, we investigated the adsorption of H atom on CoCuFeMnNi alloy by performing density functional theory (DFT) and machine learning (ML) methods. Previous experimental works have synthesized CoCuFeMnNi and its corrosion behavior and mechanical properties have been studied [2-3]. We first evaluated the stability of the alloy by applying the Hume-Rothery rules and calculating thermodynamic parameters relevant to HEAs. We have verified that CoCuFeMnNi will tend to form a solid solution and will be stable as face-centered cubic crystal. We calculated, via DFT, the adsorption energies of H atom on the hollow site of a random subset of the CoCuFeMnNi surfaces. From the results, we observed that the presence of Cu in the vicinity (nearest neighbor) of H reduces the adsorption strength while Mn and Fe as nearest neighbors enhance the H adsorption. We employed ML algorithm to predict the remaining adsorption energies. In the implementation, we defined microstructure data based on the local environment of the adsorbed H on the surface as features [4]. We have verified the validity of the algorithm by establishing its accuracy and obtaining good agreement between the DFT calculated and ML predicted values. The analyses of the results will be presented at the conference.

References

[1] D.B. Miracle, O.N. Senkov, Acta Materialia 122, 448 (2017).

[2] S. Ozturk, A. Furkan, S. Onal, S.E. Sunbul, O. Sahin, K. Icin, Journal of Alloys and Compounds, 903, 163867 (2022).

[3] R. Sonkusare, P.D. Janani, N.P. Gurao, S. Sarkar, S. Sen, K.G. Pradeep, K. Biswas, Materials Chemistry and Physics, 210, 269 (2018).

[4] D. Roy, S.C. Mandal, B. Pathak, ACS Applied Materials & Interfaces 13, 56151 (2021).