Adsorption/photocatalytic degradation of Cefixime by the green Bi2WO6/g-C3N4/ZIF-67 dual S-scheme heterojunction: Artificial neural network, genetic algorithm, density functional theory, and toxicity assessments

This study investigated the photocatalytic adsorption, degradation, and mineralization of Cefixime (CFX) through hydrothermally synthesized Bi2WO6(36 %)/g-C3N4(54 %)/ZIF-67(10 %) dual S-scheme heterojunction (BCZ). Full characterization analysis for the fresh and reused photocatalyst was explored to study the formation of heterojunction, stability, and reusability of the BCZ. VB-XPS, Mott-Schottky plots, and UV–Vis DRS defined the band structure and electron transfer mechanism of BCZ. The effects of ten operating condition factors, including reaction time, initial concentration of CFX, the dosage of photocatalyst, reaction temperature, initial pH of the reaction, visible and UV intensity, and concentration of Na2SO4, NaOH, and NaCl in the reaction solution were experimentally investigated. These factors were then modeled through artificial neural networks (ANN). The number of neurons, training algorithm, and the type of activation functions in the ANN were optimized by mean squared error metric followed by the visualizations of the ANN predictions. Two cost functions (the ratio of instantaneous CFX concentration to initial CFX concentration (C/C0) and the ratio of photocatalyst dosage to the amount of CFX removed) were employed to optimize the value of these operating condition factors separately and simultaneously through single and multi-objective genetic algorithms. Coupling LC-MS results and DFT calculations, a degradation pathway for Cefixime was proposed and then analyzed by QSAR. The toxicity of the treated solution was investigated using the wheat seed culture, along with MIC and MBC testing conducted by E. coli and S. aureus, and its eco-friendliness was confirmed by TOC and COD. Furthermore, ICP-OES confirmed that BCZ is a green photocatalyst.