High-Throughput Screening of Ru-Based MOF-Supported Single-Atom Catalysts for Hydrogen Evolution Reaction Via Machine Learning Interatomic Potential

Single-atom catalysts supported on metal–organic frameworks (SAC-MOFs) offer high metal utilization and tunable coordination environments. However, identifying optimal metal-MOF combinations from the vast chemical space remains a formidable challenge. Here, we address this by employing a machine learning interatomic potential (MLIP) to construct a comprehensive database of 4101 Ru-SAC-MOF structures. We validate the structural fidelity of this approach, showing that the MLIP correctly predicts the Ru atom’s coordination environment with 96% accuracy. This high-throughput screening, based on stability and hydrogen adsorption criteria, then served as a powerful initial filter, narrowing the pool to 2488 promising candidates. Subsequently, we performed an in-depth density functional theory analysis on these candidates to elucidate a set of rational design principles governing their catalytic activity for the hydrogen evolution reaction. Our findings reveal that top-performing catalysts are defined by a confluence of factors. Electronically, the most active sites consistently feature a Ru d-band center optimally positioned near −1.48 eV. Geometrically, this is achieved through specific coordination motifs, such as Ru atoms sandwiched between aromatic rings, which promote strong p-d orbital hybridization. Furthermore, we confirm the kinetic stability of the single-atom state. While Ru aggregation is thermodynamically favorable, significant charge transfer to the MOF and steric hindrance provide a substantial kinetic barrier against clustering. Ultimately, this work transcends a mere screening report by establishing clear electronic and geometric design principles, offering a robust framework for the rational design of future, high-performance single-atom catalysts.