Machine-Learning-Assisted Design and Optimization of Single-Atom Transition Metal-Incorporated Carbon Quantum Dot Catalysts for Electrocatalytic Hydrogen Evolution Reaction

Hydrogen evolution reaction (HER) in acidic media has been spotlighted for hydrogen production since it is a favourable kinetics with the supplied protons from a counterpart compared to that within alkaline environment. However, there is no choice but to use a platinum-based catalyst yet. As for a noble metal-free electrocatalyst, incorporation of earth-abundant transition metal (TM) atoms into nanocarbon platforms has been extensively adopted. Although a data-driven methodology facilitates the rational design of TM-anchored carbon catalysts, its practical application suffers from either a simplified theoretical model or the prohibitive cost and complexity of experimental data generation. Herein, an effective and facile catalyst design strategy is proposed based on machine learning (ML) and its model verification using electrochemical methods accompanied by density functional theory simulations. Based on a Bayesian genetic algorithm ML model, the Ni-incorporated carbon quantum dots (Ni@CQD) loaded on a three-dimensional reduced graphene oxide conductor are proposed as the best HER catalyst amongst the various TM-incorporated CQDs under the optimal conditions of catalyst loading, electrode type, and temperature and pH of electrolyte. The ML results are validated with electrochemical experiments, where the Ni@CQD catalyst exhibited superior HER activity, requiring an overpotential of 151 mV to achieve 10 mA cm⁻² with a Tafel slope of 52 mV dec⁻¹ and impressive durability in acidic media up to 100 h. This methodology can provide an effective route for the rational design of highly active electrocatalysts for commercial applications.