Identifying Metallic Transition-Metal Dichalcogenides for Hydrogen Evolution through Multilevel High-Throughput Calculations and Machine Learning.

High-performance electrocatalysts not only exhibit high catalytic activity but also have sufficient thermodynamic stability and electronic conductivity. Although metallic 1T-phase MoS2 and WS2 have been successfully identified to have high activity for hydrogen evolution reaction, designing more extensive metallic transition-metal dichalcogenides (TMDs) faces a large challenge because of the lack of a full understanding of electronic and composition attributes related to catalytic activity. In this work, we carried out systematic high-throughput calculation screening for all possible existing two-dimensional TMD (2D-TMD) materials to obtain high-performance hydrogen evolution reaction (HER) electrocatalysts by using a few important criteria, such as zero band gap, highest thermodynamic stability among available phases, low vacancy formation energy, and approximately zero hydrogen adsorption energy. A series of materials-perfect monolayer VS2 and NiS2, transition-metal ion vacancy (TM-vacancy) ZrTe2 and PdTe2, chalcogenide ion vacancy (X-vacancy) MnS2, CrSe2, TiTe2, and VSe2-have been identified to have catalytic activity comparable with that of Pt(111). More importantly, electronic structural analysis indicates active electrons induced by defects are mostly delocalized in the nearest-neighbor and next-nearest neighbor range, rather than a single-atom active site. Combined with the machine learning method, the HER-catalytic activity of metallic phase 2D-TMD materials can be described quantitatively with local electronegativity (0.195·LEf + 0.205·LEs) and valence electron number (Vtmx), where the descriptor is ΔGH* = 0.093 - (0.195·LEf + 0.205·LEs) - 0.15·Vtmx.