Application of machine learning and statistical approaches for optimization of heavy metals (Cd2+, Pb2+, Cu2+, and Zn2+) adsorption onto carbonized char prepared from PET plastic bottle waste

Abstract

This study focuses on the probable use of graphene prepared from discarded polyethylene terephthalate plastic bottles for heavy metal (HM) adsorption. The prepared graphene is characterized by FE-SEM, EDX, and FTIR. Batch adsorption experiments were conducted with the influence of different operational conditions, namely, the time of contact (1–180 min), adsorbate concentration (25–300 mg/L), adsorbent dose (0.5–6 g/L), pH (3–7), and temperature (25–60 °C). High coefficient values (Cd (R2 = 0.99), Pb (R2 = 0.97), Cu (R2 = 0.94), and Zn (R2 = 0.98)) of the process optimization model suggested that this model was significant, where pH and adsorbent dose expressively showed stimulus removal efficiency of 86.68, 73.66, 67.10, and 57.04% for Cd, Pb, Cu, and Zn at pH (7). Furthermore, the machine learning approaches (artificial neural networks and BB-response surface methodology) revealed a good association between the tested and projected value. The maximum monolayer adsorption capacity of Cd, Pb, Cu, and Zn was 263.157, 78.740, 196.078, and 84.745 mg/g, respectively. Pseudo-second-order was the well-suited kinetics, where Langmuir and Freundlich isotherms could explain better the equilibrium adsorption data. A thermodynamic study shows that HM adsorption is favorable, exothermic, and spontaneous. Finally, this study indicates that graphene could be a potential candidate for the adsorption of HMs from wastewater.