Decision-aware multi-objective discovery of Cu2O-based photovoltaic materials integrating supply risk, stability, and bandgap engineering

The development of photovoltaic materials that simultaneously satisfy performance requirements and practical constraints is essential for large-scale energy applications. In this study, we propose a decision-aware, multi-objective screening framework for Cu2O-based materials targeting top-cell applications in Si-based multi-junction solar cells. The framework integrates three key criteria: bandgap suitability, supply risk, and thermodynamic stability. Bandgap properties are predicted using machine learning models with uncertainty-aware scoring, while supply risk is quantified using the Herfindahl–Hirschman Index, and thermodynamic stability is evaluated via formation enthalpy predictions. These metrics are combined into a unified scoring function to systematically rank candidate materials. Top-ranked candidates are subsequently validated using first-principles calculations. The screening results identify Se-, P-, and S-substituted systems as high-scoring candidates. However, first-principles validation reveals that many top-ranked compounds exhibit undesirable electronic properties, such as metallic behavior or indirect bandgaps. In contrast, Cu2O0.80Se0.20 retains semiconducting character with a direct bandgap and shows a reduction in bandgap relative to pristine Cu2O, while maintaining thermodynamic stability. These results demonstrate that multi-objective screening alone is insufficient to guarantee physically viable materials and highlight the importance of integrating data-driven approaches with physics-based validation. The proposed framework provides a robust approach for identifying practically viable photovoltaic materials and can be extended to other material systems under real-world constraints.