Ultrasound-assisted synthesis of TiO2-La2O3-CuO@zeolite for visible-light photocatalysis of crystal violet: Process optimization via decision trees and dragonfly algorithm

International audience The release of synthetic dyes into aquatic environments has recently become a significant environmental and public health concern. Owing to their high stability and resistance to biodegradation, these dyes can cause substantial ecological harm. Consequently, developing efficient and sustainable strategies for their removal from wastewater represents a critical challenge in the field of environmental engineering. In this work, a novel TiO2-La2O3-CuO heterojunction supported on raw zeolite extracted from the Tinebdar region (Bejaia, Algeria) was synthesized using an ultrasound-assisted impregnation method as a sustainable solution to the pollution problem. Its performance was evaluated for the efficient removal of Crystal Violet (CV) dye under visible light irradiation. The structural, morphological, optical, and textural properties of the prepared photocatalysts were examined through XRD, FE-SEM/EDS,FTIR UV-Vis spectroscopy, and PL. The results confirmed a synergistic effect between adsorption and photocatalysis. The zeolite support enhanced the surface area, improved the dispersion of active nanoparticles, and promoted electron-hole separation. A parametric study investigated the influence of catalyst dosage, pH, initial dye concentration, and the addition of H2O2on the degradation efficiency. Among the tested materials, the 70Z@30Het composition exhibited the highest photocatalytic activity, achieving 100 % CV removal within 90 min under visible light, and in only 15 min when combined with 43 mu L of H2O2.To optimize operational parameters, a hybrid decision tree-bootstrap (DT_Bootstrap) model was developed and trained on experimental data. The model achieved an R2 value above 0.999 and RMSE below 0.1. It was then integrated with the Dragonfly Algorithm (DA) for optimization. The optimal conditions identified were: reaction time of 15 min, initial dye concentration of 29.4 mg/L, catalyst dosage of 0.35 g/L, solution pH of 5.5, with a catalyst composition of 70 % zeolite and 30 % heterojunction, and the presence of 43 mu L of H2O2.Under these optimized conditions, the predicted and experimental degradation efficiencies were 100 % and 99.98 %, respectively. This study demonstrates a robust and sustainable approach for the removal of persistent organic pollutants from wastewater, using a locally sourced, high-performance photocatalyst. The system offers notable economic and environmental advantages due to the use of abundant natural zeolite, cost-effective dopants, and energy-efficient operation.