DFT and hybrid classical–quantum machine learning integration for photocatalyst discovery and hydrogen production

Abstract

Photocatalytic hydrogen production is a key pathway toward sustainable energy, driven by semiconductors that utilize sunlight for water splitting. This review highlights recent advances in material design, theoretical modeling, and data-driven discovery. Focus is given to visible-light-active semiconductors with optimal band gaps (1.8–2.4 eV), such as BiVO4, g-C3N4, and CdS, which enable efficient redox reactions. Hybrid architectures, including Pt-loaded TiO2 and CdS/ZnS core–shell systems, demonstrate hydrogen evolution rates exceeding 105 mol m−2 s−1. Upconversion nanomaterials based on rare-earth-doped fluorides extend light harvesting into the NIR, enhancing quantum yields when combined with quantum dots. Engineered heterojunctions and carbon-based 2D interfaces improve charge separation and suppress recombination. Thermodynamic parameters such as low overpotentials (<0.3 V) and high absorption coefficients (>105 cm−1) correlate with high catalytic efficiency. Time-dependent simulations and density functional theory (DFT) offer insights into structure–property relationships. Additionally, machine learning models expedite discovery by navigating complex compositional and structural spaces. While integrating theoretical, experimental, and AI-driven approaches, this review presents a framework for the rational design of scalable photocatalysts that meet future energy demands.