Unveiling the Synergy of Photocatalytic, Electrochemical, and Machine Learning Studies in CO2 Conversion to Renewable Fuels by the Z-Schemed MIL-53(Cr)/Ag2O/Bi2O3 Nanocomposite

This study presents the development of a ternary composite based on metal–organic framework (MOF) MIL-53(Cr) co-doped with Ag2O and Bi2O3 to enhance photocatalytic CO2 reduction efficiency. The co-doping strategy improves light absorption, charge-carrier separation, and redox performance. Structural and morphological characterizations using SEM-EDS and PXRD confirm a uniform dopant distribution and successful composite formation. FTIR spectroscopy verifies the retention of functional groups, while UV-vis absorption and Tauc’s plot analysis indicate a band gap reduction from 3.66 to 3.29 eV. BET analysis reveals increased porosity and surface area, facilitating higher CO2 adsorption. Electrochemical studies, including LSV, CV, capacitance, ECSA, EIS, and Mott-Schottky analysis, demonstrate enhanced charge mobility and reduced interfacial resistance in the co-doped systems. The MIL-53(Cr)/Ag2O/Bi2O3 composite achieved a methanol production rate of 0.86 ± 0.03 mmol gcat-1; three replicates are studied under UV irradiation. Improved photocatalytic activity is attributed to a Z-scheme mechanism that promotes efficient electron-hole separation. In parallel, machine learning (ML) models developed using experimental data predict methanol yield, with Bayesian Ridge Regression showing the highest predictive accuracy (R2 = 0.800; RMSE = 1.27 × 10-7). SHAP analysis identifies the preparation temperature, band gap, CO2 inlet pressure, and dopant loading as key factors influencing yield. Statistical validation through boxplot and autocorrelation analyses confirms data consistency and independence. This integrated experimental and ML based approach offers a comprehensive understanding of structure-activity relationships and demonstrates the potential for AI-driven design of next-generation photocatalysts for sustainable CO2 capture and conversion.