O- and OH-induced dopant segregation in single atom alloy surfaces: A combined density functional theory and machine learning study
Anne Nicole P. Hipolito, Marianne A. Palmero, Viejay Z. Ordillo, Koji Shimizu, Darwin B. Putungan, Alexandra B. Santos-Putungan, Joey D. Ocon, Satoshi Watanabe, Karl Ezra S. Pilario, Allan Abraham B. Padama
The authors studied the different factors that influence the segregation in single-atom alloy (SAA) surfaces in the presence of adsorbates. They performed computational investigations and employed density functional theory calculations and machine learning methods to determine these factors. SAA surface is a binary alloy that is composed of a host system and a small concentration of dopant atom. Previous studies found that there is a tendency for atoms of alloys to segregate from the topmost layer to the subsurface both in the clean surface and in surface with adsorbate. To examine the segregation in SAAs, we constructed SAA surfaces using different transition metals (Ag, Au, Co, Cu, Ir, Ni, Pd, Pt, and Rh) and used O and OH as adsorbates. These adsorbates are key intermediates in a reaction in a fuel cell.
The segregation energies of clean SAAs and SAAs with adsorbates were computed. The segregation energies are used to predict if segregation will occur. The authors then used various machine learning algorithms which include 44 features (variables) based on elemental, energetics, and electronic properties. The number of features was reduced by employing a two-stage feature selection method. They found that the segregation phenomenon in these different SAAs is influenced by the atomic radii of the elements, stability of the dopant in SAA, preference of the elements to form surfaces, and the surface atoms electronic properties. It was identified that support vector machine regression model exhibited the best performance for the segregation prediction.
Lastly, it was identified that Rh-Au(111) SAA is a potential catalyst for fuel cells. Rh is stable in the topmost layer when O and OH interact with the surface. The promising characteristic of Rh-Au(111) is due to the atom-like property of the Rh atom which is a consequence of the weak interaction between the Rh and the Au atoms in its vicinity. These findings are supported by the calculated adsorption energies and electronic properties.
See full paper here: https://doi.org/10.1016/j.commatsci.2023.112607
Complete Citation: Hipolito, A.N.P., Palmero, M.A., Ordillo, V.Z., Shimizu, K., Putungan, D.B., Santos-Putungan, A.B., Ocon, J.D., Watanabe, S., Pilario, K.E.S., Padama, A.A.B. 2024. O-and OH-induced dopant segregation in single atom alloy surfaces: A combined density
functional theory and machine learning study, Computational Materials Science 232, 112607.
