Machine Learning‐Driven Selection of Two‐Dimensional Carbon‐Based Supports for Dual‐Atom Catalysts in CO2 Electroreduction

The electrocatalytic reduction of carbon dioxide by metal catalysts featuring dual‐atomic active sites, supported on two‐dimensional carbon‐nitrogen materials, holds promise for enhanced efficiency. The potential synergy between various support materials and transition metal compositions in influencing reaction performance has been recognized. However, systematic studies on the selection of optimal support materials remain limited, primarily due to the intricate structure of dual‐atom catalysts generating a variety of potential adsorption sites. Incorporating the influence of support materials further amplifies computational challenges, doubling the already substantial calculation requirements. This study addresses this challenge by introducing a machine learning approach to expedite the identification of the most stable intermediate adsorption sites and simultaneous prediction of adsorption energy. This innovative method significantly reduces computational costs, enabling the simultaneous consideration of active sites and support materials. We explore the use of both graphene‐like (g−)C2N and g‐C9N4 materials, revealing their main distinction in the adsorption capacity for the intermediate *CHO. This variation is attributed to the different C : N ratios influencing support for the active site through distinct charge transfer conditions. Our findings offer valuable insights for the design and optimization of dual‐atom catalysts. The graphical illustration presented herein delineates the inclusion of dual‐atom catalysts (DACs) supported on two different carbon‐based materials, g‐C2N and g‐C9N4, within the scope of this study. A high‐throughput machine learning‐assisted computational approach was employed to accelerate the screening for optimal DAC candidates, aiming to enhance carbon dioxide electroreduction (CO2RR) efficiency.