Modeling of tetracycline removal from water using plastic waste-carbon nanomaterial: a study based on machine learning and mathematical models

Abstract This study synthesized plastic waste-carbon nanomaterial (PW-CN) and evaluated its effectiveness in removing tetracycline (TC) from water. Characterization of the material was carried out with FTIR, BET, UV-Vis, PXRD, FESEM, and HRTEM. Machine learning models of Box Behnken design (BBD) and artificial neural networks (ANNs) coupled with non-linear regression analysis of kinetics and isotherms were employed to optimize and understand the adsorption process. Results indicate a BBD optimized time of 24.73 min, at a 3.84 pH, 0.052 g dosage, and concentration of 29.79 mg/L with a removal efficiency of 93.96% through the desirability function. The ANN approach predicted the optimal conditions for TC removal with removal efficiencies of 93.67%. The pseudo-second order kinetic model with a correlation coefficient (R 2 = 0.9948) described the mechanism to occur by chemisorption. The Langmuir model gave a maximum adsorption capacity (q max ) of 123.93 mg/g. The data also show that the Freundlich model, with a high correlation coefficient (R 2 = 0.9771), described the mode of adsorption of TC by PW-AC to occur on multilayer energetically stable heterogeneous surfaces. The study presents a theoretical approach of using machine learning and mathematical models to describe the adsorption of TC onto PW-CN.