▎ 摘　　要

Developing high-performance cathode catalysts by a natural extension of renowned single-atom catalysts (SACs) for electrochemical CO2 reduction (CRR) into useful chemical feedstocks has recently received great attention, but challenges remain due to the lack of rational design as well as poor understanding of their high selectivity and activity. Here we use density functional theory (DFT) calculations to systematically investigate the supremacy of extended metallic biatom catalysts (BACs), namely Cu-based non-noble transition metals anchored on two (four or six) nitrogen-doped defected graphene substrates, denoted as MCu@2(4,6)NG (M: Mn-Zn), for CRR catalytic performance. All MCu dimers are embedded in the double vacancy of both 2(4)NG surfaces via the most stable "anchor-antenna" pattern but in the quarter vacancy of the 6NG substrate via "in-plane" configuration. Most BACs could effectively suppress hydrogen evolution and enhance the CRR activity and selectivity toward HCOOH among various C-1 products. Intriguingly, the MCu@2NG class not only possesses relatively high geometric stability but also outperforms its counterparts with smaller limiting potentials (vertical bar U-L vertical bar) for catalyzing the CRR Remarkably, NiCu@2NG, Cu-2@2NG, and CoCu@2NG are identified as the best catalysts due to the lowest vertical bar U-L vertical bar values of 0.20, 0.29, and 0.30 V, respectively. Furthermore, we clarify the underlying origin of catalytic CRR enhancement and the activity trend of our metallic binuclears in the 2NG substrate family via atomic ensembles and magnetic-electronic properties of metallic dimers MCu in the key *OCHO intermediate. Our results also unveil the intriguing linear correlation of CRR performance (related to vertical bar U-L vertical bar) with d-band centers and magnetic moments for the synergistic effect of MCu binuclears in BACs. Therefore, these findings might offer valuable perceptions and promote feasible design strategies for the new frontier of heterogeneous BACs for CO2 conversion.