Efficient catalytic CO2 cycloaddition reaction via zinc-immobilized amine-functionalized MCM-41 and SBA-15 mesoporous heterogeneous catalysts enhanced by machine learning insights
摘要
Advanced nanomaterials, particularly mesoporous materials with high surface area and tunable properties, play a crucial role in efficient CO2 capture and catalytic conversion. Their application supports the transition toward CO2 utilization as a carbon feedstock, aligning with SDGs and net-zero emission goals. This study presents the design, synthesis, and catalytic evaluation of mononuclear zinc complex immobilized on amine-functionalized mesoporous silicas MCM-41 and SBA-15, demonstrating their potential in heterogeneous catalysis for the cycloaddition of CO2 with epoxide. The zinc complex, containing 2,9-dimethyl-1,10-phenanthroline coordinated to zinc chloride, was anchored onto the supports. Functionalized mesoporous silicas were utilized as hosts to immobilize Zn1 complex, producing organic–inorganic hybrid materials, also termed as “Immobilized Heterogeneous Catalyst (IHC),” Structural and physicochemical characterizations by powder XRD, FT-IR, N2 adsorption/desorption, SEM, NMR, and thermal analysis confirmed the integrity and functionality of the hybrid materials. Catalytic testing revealed high efficiency with MCM-41 IHC achieving a 94% conversion, 93% selectivity (30 mg, 130 °C, 5 h, 10 bar), and SBA-15 IHC delivering 96% conversion, 94% selectivity (20 mg, 140 °C, 4 h, 10 bar). The novelty of this work lies designing of a zinc complex-immobilized mesoporous structure, providing the unique structural and electronic synergy between the Zn1 complex and the 2-D ordered mesoporous supports. This approach showed high catalytic efficiency for a solvent-free, co-catalyst-free hybrid system in CO2 utilization. Also, machine learning analysis successfully showed its capability to predict and to interpret CO2 conversion and selectivity with great confidence. Of the regression models tested, the ARD regression and Huber regressor provided the best level of fit (i.e., R2 > 0.95, and low RMSE and MAE errors), which proved their robustness and ability to generalize. These combined experimental and machine learning insights enhanced the development of efficient CO2 utilization systems, contributing to the goals of sustainable green chemistry.
Graphical abstractZn complex immobilized on MCM-41 and SBA-15 catalysts achieved high conversion and selectivity in solvent-free CO2 cycloaddition with epichlorohydrin proved strong Zn–amine synergy enhanced activity with ML predicted trends and enabled reaction optimization.