Strain engineering enhances electrocatalytic hydrogen evolution of B-N pair substituted biphenylene networks

Abstract

The exploration of novel metal-free electrocatalysts with excellent activity and stability is important for scaling up hydrogen production via water splitting. Herein, we report a series of novel two-dimensional networks obtained by doping biphenylene (BPN) with B-N pairs, referred to as BCN-BPNs. Density functional theory calculations indicate that the designed BCN-BPNs are electronically and dynamically stable with metallic or semiconducting characteristics. It is further revealed that uniaxial lattice strain engineering is an effective approach to activating inert BPN and BCN-BPNs toward the hydrogen evolution reaction (HER). The best overpotential for HER is predicted to be 0.011 V, achieved by finely tuning strain. A machine learning model with an extreme gradient boosting regressor and a polynomial feature approach was successfully built to predict HER activities. The model achieved mean absolute errors and root mean squared errors of 0.068 and 0.261, respectively. Feature importance analysis highlights the role of both active sites and their microenvironment in determining HER activities. Our study suggests a series of novel metal-free BCN-BPNs with considerable potential for HER. Additionally, a predictive strategy is proposed for fast screening, which would benefit the development of HER catalysts that exhibit both cost-effectiveness and superior performance.