Universal Predictive Power: Introducing the Electronic Structure Decomposition Approach for CO Adsorption and...

Accurate prediction of catalyst performance is crucial for designing materials with specific catalytic functions. While the density functional theory (DFT) method is widely used for its accuracy, modeling heterogeneous systems, especially supported transition metals, poses significant computational challenges. To address these challenges, we introduce the Electronic Structure Decomposition Approach (ESDA), a novel method that identifies specific density of states (DOS) areas responsible for adsorbate interaction and activation on the catalyst. As a case study, we investigate the influence of α-Al₂O₃(0001) as a support material on CO adsorption energy and the stretching frequency of the C-O bond on Ru nanoparticles (NPs). Using multiple linear regression analysis, ESDA models were trained with data from isolated Ru NPs and adjusted using supported NP sample data. The ESDA models accurately predict the CO adsorption energies and C-O vibrational frequencies, demonstrating strong linear correlations between predicted and DFT-calculated values with low errors across various adsorption sites for both isolated and supported Ru NPs. Beyond pinpointing the DOS areas responsible for CO adsorption and C-O bond activation, this study provides insights into manipulating these DOS areas to control CO activation, hence facilitating CO dissociation. Additionally, ESDA significantly accelerates the characterization and prediction of CO adsorption and activation on both isolated and supported Ru NPs compared to DFT calculations, expediting the design of new catalytic materials and advancing catalysis research. Furthermore, ESDA's reliance on the electronic structure as a descriptor suggests its potential for predicting various properties beyond catalysis, broadening its applicability across diverse scientific domains.