Structure-activity relation of single-atom electrocatalysts for CO2 reduction to CH4

Atomically dispersed single-atom catalysts (SACs) supported on graphene provide a ladder of opportunity for achieving the electrocatalytic CO₂ reduction to value-added chemicals, where the efficient combinations of active sites and coordination environments are a powerful approach to optimize both activity and selectivity. Nevertheless, the understanding of the underlying mechanism about how the catalytic performance varies via active sites and coordination environments is very limited. Herein, we successfully constructed the activity trend of SACs with different coordination environments (MX_nY_4−n X, Y = N, S, P, and M = 19 transition metals) for CO₂ reduction to CH₄, using easily obtainable parameters. Based on the entire reaction free energy over 110 stable SACs, the binding strength of *OCHO intermediate (ΔE*_OCHO) is identified as an initial activity indicator towards CO₂ reduction to CH₄. With the help of multi-task symbolic regression, a simple active descriptor consisting of intrinsic properties (valence-electron number and electronegativity of metal atoms and coordination atoms) is further constructed, which can well describe the variation of ΔE_*OCHO and the onset potential for CH₄. Importantly, this active descriptor enables rapid evaluation on the activity of SACs without DFT computations. This work is instructive for the rational design of other CO₂ electrocatalysts and establishing more multiplex descriptors through decoupling the factors and handling them separately.