Machine learning and DFT-assisted design of A2B2X6 2D materials for photocatalytic CO2 reduction

Photocatalytic CO2 reduction reaction (CO2RR) is an important technology to address the energy and environmental crisis, and finding high-performance photocatalysts is the core task in this field. Two-dimensional (2D) materials have shown great potential in photocatalytic CO2RR due to their excellent stability, photoelectric properties and high surface catalytic activity. This study is the first to explore the large-scale application of two-dimensional A(2)B(2)X(6) compounds in CO2RR. Considering that iterative experiments and computational methods cannot quickly identify ideal candidates, we screened all suitable candidate materials from the periodic table by combining machine learning (ML) and first-principles calculations, which are similar to 10(7) faster than ab initio calculations. Finally, two suitable CO2RR photocatalysts were identified: Bi2ZnPtO6 and Bi2CrZnO6. Density functional theory (DFT) calculations show that the HSE bandgaps of Bi2ZnPtO6 and Bi2CrZnO6 were 2.30 eV and 3.10 eV, respectively, and their VBM and CBM positions were suitable for the reduction of CO2 to CO and CHOH. Bi2CrZnO6 exhibited strong ultraviolet absorption (100-400 nm, peak at 200 nm), while Bi2ZnPtO6 demonstrated a broad spectral response, maintaining absorption in the 600-800 nm range. Both materials have good CO2 reduction and light absorption properties, and are expected to be efficient CO2RR photocatalysts. This study provides theoretical guidance for the development of high-performance photocatalytic materials and lays the foundation for the innovation of sustainable energy technologies.