Transition metal-anchored BN tubes as single-atom catalysts for NO reduction reaction: A study of DFT and deep learning

The reduction of nitric oxide (NO) to ammonia has attracted wide attention because this method can not only reduce active nitrogen pollutants to generate economic benefits, but also greatly promote the nitrogen cycle of the ecosystem. Boron nitrogen (BN) nanotube is often selected as a matrix for catalyst research due to excellent electronic properties. In order to develop high energy conversion efficiency and durable catalysts for NO reduction reaction (NORR), a series of BN tube-based single-atom catalysts (SACs) with different tube's diameters made of hexagonal boron nitride and 28 transition metals were designed. The effects of the tube's diameter (the curvature of active sites at transition metals) and transition metal active sites on the NORR process were explored by generalized gradient approximation level (GGA-PBE) and deep learning algorithms. The results show that the theoretical overpotentials corresponding to NORR and hydrogen evolution reaction (HER) on Mn@BN(9,0) are-0.23 and-0.67 eV, respectively. This reflects the high activity and selectivity of the catalyst for NORR. The feature importance analysis of deep learning model shows that the curvature of the transition metal active site has a greater impact on the NORR catalytic performance. Its proportion is much greater than the atomic properties such as the number of d-orbital electrons from the transition metal. Therefore, the catalytic performance of the catalyst can be better improved by changing the curvature of the transition metal sites than by replacing the metal. This may provide certain theoretical guidance for the design of efficient catalysts for NORR.