Integrating Density Functional Theory with Machine Learning for Enhanced Band Gap Prediction in Metal Oxides

Chidozie Ezeakunne, Bipin Lamichhane, and Shyam Kattel*

Florida A&M University

In this comprehensive study, we used a combination of density functional theory with Hubbard U correction (DFT+U) and machine learning (ML) to accurately predict the band gaps and lattice parameters of metal oxides: TiO₂ (rutile and anatase), ZnO, ZnO₂, CeO₂, and ZrO₂. The results show that including U_p values for oxygen 2p orbitals alongside U_d/f for metal 3d or 4f orbitals significantly improves the band gap and lattice parameters predictions. Our extensive DFT+U calculations identified optimal (U_p, U_d/f) pairs, which closely predicted the experimental value of band gap and lattice parameters: (8 eV, 8 eV) for rutile TiO₂, (3 eV, 6 eV) for anatase TiO₂, (6 eV, 12 eV) for ZnO, (10 eV, 10 eV) for ZnO₂, (9 eV, 5 eV) for ZrO₂ and (7 eV, 12 eV) for CeO₂. The ML analysis shows that of all ML models included in the current study, polynomial regression reliably produces results with accuracy comparable to DFT calculations. Therefore, this study not only identifies the best (U_p, U_d/f) pairs to predict experimentally measured band gap and lattice parameters but also highlights the effectiveness of a simple regression ML model in predicting band gaps of metal oxides.