Abstract
A central challenge in heterogeneous catalyst informatics is developing descriptors that capture catalyst behavior while remaining physically interpretable. Catalyst–reaction-species interactions provide a basis for representing heterogeneous catalytic reactions. Here, a high-throughput density functional theory adsorption-energy library (Ad-Lib) is constructed, and designed descriptors are applied to oxidative coupling of methane (OCM) and CO2 methanation, with OCM as the principal case study. Mapping catalysts in an adsorption-energy-defined interaction space provides a landscape of catalytic performance. The utility of this representation for describing catalytic behavior is further demonstrated through machine learning and SHAP analysis. This work establishes the multidimensional adsorption-energy space as a reaction-informed, physically grounded framework for heterogeneous catalyst informatics.
Funding Information
  • Japan Society for the Promotion of Science (24K01241)
  • Japan Science and Technology Agency (JPMJBS2426)
  • Precursory Research for Embryonic Science and Technology (JP-MJPR24T5)
  • JST-Mirai Program (JPMJMI25G1)

This publication has 41 references indexed in Scilit: