Hydrotalcite-derived Cu-Al mixed oxides for dimethyl carbonate synthesis via transesterification: Kinetic analysis and GA-BP neural network modeling

Cu-Al mixed oxide catalysts (CxA-LDO), prepared via calcination of layered double hydroxides (LDHs) precursors, were investigated for the transesterification of propylene carbonate (PC) with methanol to dimethyl carbonate (DMC). Characterization demonstrated that calcination temperature and the Cu/Al molar ratio critically influence the surface basicity properties and structural stability. Under optimized conditions (80 °C, 2 h, 2.0 wt% loading, MeOH/PC = 10:1), the C 0.5 A-400-LDO catalyst achieved 57.93% PC conversion and 57.29% DMC yield (TON = 28.94). Surface basicity was identified as the key governing factor for the reaction. This conclusion is supported by a strong correlation observed between the reaction rate and the number of total basic sites. However, the catalyst deactivated significantly over six cycles (PC conversion dropped from 57.93% to 47.2%), which ICP-AES analysis unambiguously attributed to significant copper leaching (8.1 wt.% net loss) rather than sintering. Kinetic analysis revealed a complex reaction mechanism that could be simplified to pseudo-first-order reaction behavior. The activation energies were determined to be Ea( k + ) = 51.61 kJ·mol -1 for the forward reaction and Ea(k - ) = 55.34 kJ·mol -1 for the reverse reaction. Concurrently, a hybrid genetic algorithm-optimized backpropagation (GA-BP) neural network was developed. Compared to conventional BP, the GA-BP model demonstrated superior predictive accuracy (R 2 = 0.997 for training set and 0.977 for testing set) and exhibited significantly reduced error metrics. This study not only contributes to the fundamental understanding of base-catalyzed transesterification but also introduces a synergistic experimental-computational strategy applicable to rational catalyst design.