Interaction trends between single metal atoms and oxide supports identified with density functional theory and statistical learning

Nolan J. O’Connor,A. S. M. Jonayat,Michael J. Janik,Thomas P. Senftle

DOI: https://doi.org/10.1038/s41929-018-0094-5

IF: 37.8

2018-07-01

Nature Catalysis

Abstract:Single-atom catalysts offer high reactivity and selectivity while maximizing utilization of the expensive active metal component. However, they are susceptible to sintering, where single metal atoms agglomerate into thermodynamically stable clusters. Tuning the binding strength between single metal atoms and oxide supports is essential to prevent sintering. We apply density functional theory, together with a statistical learning approach based on least absolute shrinkage and selection operator regression, to identify property descriptors that predict interaction strengths between single metal atoms and oxide supports. Here, we show that interfacial binding is correlated with readily available physical properties of both the supported metal, such as oxophilicity measured by oxide formation energy, and the support, such as reducibility measured by oxygen vacancy formation energy. These properties can be used to empirically screen interaction strengths between metal–support pairs, thus aiding the design of single-atom catalysts that are robust against sintering.

chemistry, physical