Catalytic Tango in Diatomic Catalysts: From Precision-Guided Pair Construction to Machine-Learning-Driven Identification and Design

Abstract

Diatomic catalysts (DACs) are composed of two neighboring metal centers, each stabilized by its own coordination environment, that extend single-atom precision into a cooperative two-site manifold. By programmably tuning composition, geometry, and local microenvironment, DACs bridge the gap between isolated single-atom sites and extended clusters, and can enable cross-site electronic coupling, dual-site adsorption motifs, and short-range relay pathways that are difficult to realize on mononuclear sites, effectively creating a catalytic “tango” in which two adjacent centers cooperate through shared electronic structure and intermediates. This Review first highlights strategies for precision-guided pair construction, including probability-controlled density generators that increase the likelihood of forming adjacent sites, ordered coordination networks (MOFs, COFs, and related 2D frameworks) that pre-organize metal nodes at defined separations, and multinuclear metal–organic complexes that offer molecular blueprints for well-defined dimers and higher oligomers. These platforms are then used to discuss representative modes of intersite cooperativity: charge redistribution and orbital hybridization between neighboring metals, distance- and orientation-dependent co-adsorption and transition states on dual sites, and sequential reaction pathways in which different elementary steps are preferentially accommodated by different atoms in the pair. Finally, the Review surveys how atomic-resolution electron microscopy and related correlative imaging are beginning to be combined with machine-learning workflows for automatic identification, classification, and screening of neighboring metal sites, sketching an emerging and promising route toward machine-learning-assisted identification and design of diatomic motifs.