Removal of heavy metals using lichen-derived activated carbons: adsorption studies, machine learning, and response surface methodology approaches

Abstract

Biomass-based activated carbons are promising as they are effective and low-cost for wastewater remediation. In this study, the removal of lead, copper, and zinc was investigated using activated carbons obtained from two different lichens. The performance of the 5th-order Response Surface methodology (RSM), Machine Learning (ML), and Artificial Neural Network (ANN) model based on Face-Centered Central Composite Design (FCCCD) was evaluated considering initial concentration, temperature, and time effects. The effectiveness of using ANN for accurate prediction in lead and copper removal and the superior performance of ML-based 5th-order RSM for zinc removal were demonstrated. Among the Langmuir, Freundlich, and Temkin isotherm models, the Freundlich model best described the adsorption processes, and the Langmuir maximum adsorption capacities were found to be 105.26 mg/g (Pb/AC-1), 59.52 mg/g (Cu/AC-1), and 53.19 mg/g (Cu/AC-2). Additionally, the pseudo-first-order, pseudo-second-order, and intra-particle diffusion models were examined, and it was found that the adsorption processes followed the pseudo-second-order kinetics and intra-particle diffusion played a significant role. The activation energies and Delta H0 values less than 40 kJ/mol and Delta G0 values below - 20 kJ/mol showed that the metals were adsorbed by physical mechanisms. The novelty of this study is that the 5th-order RSM model is applied to adsorption processes for the first time, and a multi-faceted approach is used to analyse adsorption processes, including machine learning and ANN, isotherm modeling, thermodynamic evaluation, kinetics analysis, and activation energy calculations.