Theoretical Design of Dual-Site Metallo-Covalent Organic Frameworks for Efficient Co2 Photoreduction into C2h4

Abstract

Designing and modulating active sites in photocatalysts to achieve efficient CO2 conversion into high value-added C2 products is crucial but challenging for CO2 resource utilization. A promising strategy is the construction of asymmetric dual-metal active sites to generate asymmetric charge distribution and synergistic effect between adjacent metal centers, which can promote the key C–C coupling process in generating C2 products. In particular, metallo-covalent organic frameworks (M-COFs) provide an ideal platform for strategic design of active sites, owing to their naturally large pores to integrate active metal sites into extended frameworks and the well-defined coordination environments of these metal atoms for understanding the structure-activity relationships. In this work, a series of homonuclear and heteronuclear dual-site M-COFs (M2/MM'-S-COFs) were constructed, and their CO2 reduction (CO2RR) performance and photocatalytic mechanism were systematically investigated using density functional theory (DFT). Among these, MgFe-S-COF and MgZn-S-COF were identified as the most promising catalysts for CO2RR into C2H4 production, exhibiting ultra-low energy barriers in the C–C coupling process of 0.04 and 0.03 eV, respectively. The superior CO2RR performance of heteronuclear MgFe-S-COF and MgZn-S-COF compared to their homonuclear counterparts is attributed to the asymmetric charge distribution and a “donation-back donation-resonation” mechanism, which facilitates efficient C–C coupling. Furthermore, machine learning (ML) further predicted 69 structures with the potential for generating C2 products out of 136 M2/MM'-S-COF candidates. This work highlights the potential of dual-site M-COFs with charge-polarized active sites to address the challenges in C2 production, offering theoretical insights for future experimental synthesis.

Keywords: Photocatalytic CO2 reduction reaction, metallo-covalent organic frameworks, C2H4, Density functional theory, machine learning