Rational design of transition metal-intercalated Ti-doped WS2 bilayers for enhanced hydrogen evolution catalysis: A synergistic DFT-machine learning approach

The rational design of cost-effective hydrogen evolution reaction (HER) electrocatalysts remains a critical challenge in advancing sustainable energy technologies. This study employs an integrated computational approach combining density functional theory (DFT) with machine learning (ML) algorithms to systematically investigate transition metal (TM; Sc-Zn) intercalation effects in Ti-doped WS₂ bilayers. Our results reveal that intercalation of specific TM atoms (Ti, V, Cr, Mn, Fe) significantly enhances HER performance, with eight configurations exhibiting ultralow hydrogen adsorption Gibbs free energy ($\Delta G_{H^{*}}$, 0.003–0.083 eV), surpassing commercial Pt catalysts ($\Delta G_{H^{*}}$ = 0.09 eV). Stability analyses confirm the thermodynamic robustness of these systems under operational conditions. The d-band center of intercalated atoms is found to be the dominant factor governing HER activity through ML models, with TM–S bond lengths as a secondary contributor. This is particularly evident in gradient boosting regression (R² = 0.954, RMSE = 0.30 eV). By proposing a combined “intercalation-surface doping” strategy, we have established a dual optimization framework for tailoring electronic structures and active sites. This work not only provides fundamental insights into TM chalcogenide catalysis but also delivers a computational protocol for accelerating the discovery of high-performance, non-precious electrocatalysts, offering transformative potential for next-generation energy conversion systems.