A Universal Quasi-Atom Interaction Model and Application in Designing Single-Atom Catalysts for Alkali-Metal...

A deep understanding of reactant adsorption behaviors is crucial for unraveling structure-activity relationships and identifying simple descriptors for the kinetics of multi-electron redox reactions, such as the sulfur redox process in Li-S batteries (LSBs). Recently, the "10-electron rule" for evaluating the adsorption strength of single atoms (E_ads) has been established for free-atom-like single-atom alloys (SAAs). However, our density-functional theory calculations show that this rule is unfeasible for single-atom catalysts (SACs) with metal centers covalently coordinated by nonmetal atoms. Herein, we proposed a new 18-electron rule for covalent SACs by establishing the quasi-atom interaction model (Q-AIM), which interprets the U-type behavior of E_ads(S) dominated by antibonding-orbital filling. Additional calculations demonstrate that E_ads(S) can serve as an efficient descriptor of the kinetics of sulfur redox involving multiatom polysulfides. Then, guided by the efficient predictor and Q-AIM, we employed machine learning to screen ∼800,000 candidate SACs. Ultimately, we obtained over 400 and 8000 promising SACs for sulfur evolution and sulfur reduction, respectively. Furthermore, E_ads(S) can be extended to predict the catalytic performance of SACs in Li/Na/K-S/Se batteries. This work not only simplifies complex multi-electron processes to single-atom adsorption energy but also establishes a link between occupied electronic states and the catalytic performance of SACs.