Machine Learning‐Based Process Optimization for Adsorption Removal of Emerging Contaminant (Tetracycline) From Aqueous Solution Onto Zinc Chloride Activated Biomass (<i>Cassava Peels</i>)

Abstract

ABSTRACT The treatment of tetracycline (TC) in conventional water treatment facilities is difficult because the compound is generally considered nonbiodegradable. In this study, the performance of a synthesized adsorbent, activated carbon produced from cassava peels using the sol‐gel method with the surface area modified by the activating agent zinc chloride (ZnCl₂), was evaluated using a combined statistical experimental design and machine learning techniques. The effects of operational parameters, including solution pH, adsorption duration, adsorbent dose, and initial TC concentration, were examined through a single‐factor analysis of adsorption removal efficiency. The ranges tested for these parameters were pH 3–9, adsorption time of 30–90 min, adsorbent dosage of 0.025–0.075 gm/L, and initial contaminant concentration of 2.5–7.5 mg/L. Combined response surface methodology and artificial neural network techniques were employed to optimize TC adsorption removal efficiency. Under the ideal operating conditions of a solution pH of 7.49, an adsorption time of 74.15 min, and an adsorbent dosage of 0.059 gm/L, an optimal TC adsorption removal effectiveness of 97.2% was achieved, with a desirability score of 1.000 for the desired initial pollutant concentration (TC) of 2.5 mg/L. The pseudo‐second‐order kinetic model and Langmuir isotherm were found to best fit the adsorption and chemical kinetic studies, respectively.