Machine learning-assisted the Ag/Ni(OH)2 heterostructure design for boosting electrocatalytic hydrogen evolution through charge redistribution

Abstract

Searching for highly efficient electrocatalysts for the hydrogen evolution reaction (HER) is principal to the development electrolytic water production industry. Experimental screening of highly active electrocatalysts is time-consuming and complicated. In this work, an Artificial Neural Network model is proposed to accelerate the screening for Ag(M) catalysts (M = Al, Si, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, Ge, Mo, Ru, Rh, Pd, In, Sn, Sb, W, Re, Os, Ir, Pt, Au and Hg), which is used to predict the Gibbs free energy of hydrogen. The Ag(Ni) catalyst is identified as a potential electrocatalyst with the nearly ideal (ΔG_H), which affords the relatively low overpotentials of 159 mV for HER at 10 mA cm⁻². According to the prediction of our ANN model, we synthesized Ag(Mn), Ag(Co), and Ag(Cu) catalysts. The Ag(Ni) catalyst exhibits the best HER activity, which is 120 mV smaller than Ag(Mn), Ag(Co), and Ag(Cu) catalysts. The incorporation of Ni effectively optimizes the electronic environment of the materials, which drives the upshift of d-band center and drastically reduces the Gibbs free energy of hydrogen (ΔG_H). Our method is significantly more efficient, running 1184 times faster than the traditional DFT method. Our work paves an efficient way to design and develop potential HER catalysts.