Green-synthesized ZrO₂ Nanoparticles for Congo Red Adsorption: Machine Learning Prediction and Multi-criteria Optimization
Abstract
Zirconium dioxide (ZrO₂) nanoparticles were synthesized from zirconyl chloride octahydrate using Toddalia asiatica leaf extract through an environmentally sustainable route and evaluated for Congo Red (CR) adsorption. XRD, FTIR, SEM–EDX, TGA, and XRF analyses confirmed the crystalline structure, morphology, surface functional groups, elemental composition, and thermal stability of the synthesized nanoparticles. Adsorption performance was investigated using an integrated Taguchi–ANOVA–PSI–PIV–CFANN framework. ANOVA revealed that ZrO₂ dosage predominantly influenced CR removal, whereas initial concentration and dosage governed adsorption capacity. PSI–PIV identified Run 27 at 75 mg L⁻¹ CR, 90 min, 40 °C and 0.25 g L⁻¹ ZrO₂ as the best experimental compromise, achieving 63.56% removal efficiency and the experimentally calculated adsorption capacity. CFANN–PSI–PIV predicted an optimum at 75 mg L⁻¹, 90 min, 50 °C, and 0.25 g L⁻¹, with 67.2169% removal efficiency and the corresponding predicted adsorption capacity. The latter condition represents a model-predicted optimum. These findings demonstrate that green-synthesized ZrO₂ nanoparticles, combined with data-driven optimization, offer a promising approach for treating CR-contaminated wastewater.