Artificial intelligence-enabled optimization of Fe/Zn@biochar photocatalyst for 2,6-dichlorophenol removal from petrochemical wastewater: a techno-economic perspective

While numerous studies have addressed the photocatalytic degradation of 2,6-dichlorophenol (2,6-DCP) in wastewater, an existing research gap pertains to operational factors’ optimization by non-linear prediction models to ensure a cost-effective and sustainable process. Herein, we focus on optimizing the photocatalytic degradation of 2,6-DCP using artificial intelligence modeling, aiming at minimizing initial capital outlay and ongoing operational expenses. Hence, Fe/Zn@biochar, a novel material, was synthesized, characterized, and applied to harness the dual capabilities of 2,6-DCP adsorption and degradation. Fe/Zn@biochar exhibited an adsorption energy of −21.858 kJ/mol, effectively capturing the 2,6-DCP molecules. This catalyst accumulated photo-excited electrons, which, upon interaction with adsorbed oxygen and/or dissolved oxygen generated •O₂−. The •OH radicals could also be produced from h⁺ in the Fe/Zn@biochar valence band, cleaving the C–Cl bonds to Cl⁻ ions, dechlorinated byproducts, and phenols. An artificial neural network (ANN) model, with a 4-10-1 topology, “trainlm” training function, and feed-forward back-propagation algorithm, was developed to predict the 2,6-DCP removal efficiency. The ANN prediction accuracy was expressed as R² = 0.967 and mean squared error = 5.56e–22. The ANN-based optimized condition depicted that over 90% of 2,6-DCP could be eliminated under C₀ = 130 mg/L, pH = 2.74, and catalyst dosage = 168 mg/L within ∼4 h. This optimum condition corresponded to a total cost of $7.70/m³, which was cheaper than the price estimated from the unoptimized photocatalytic system by 16%. Hence, the proposed ANN could be employed to enhance the 2,6-DCP photocatalytic degradation process with reduced operational expenses, providing practical and cost-effective solutions for petrochemical wastewater treatment.