Novel Sesbania wood-based activated carbon for paracetamol removal: integrating batch adsorption, fixed-bed studies, and machine learning

Abstract

Pharmaceutical pollution in water is a critical environmental issue. This study investigates the removal of paracetamol (PCM) from water using activated carbon derived from Sesbania wood, a fast-spreading plant with promising structural properties for activated carbon. The batch adsorption results demonstrated the effectiveness of Sesbania-derived activated carbon (SDAC) in removing PCM solution, achieving a removal efficiency of 89%. In fixed-bed adsorption, a removal efficiency of 87.6% was attained within 210 min while treating 1050 ml of solution. The Redlich-Peterson model was employed as the best adsorption isotherm, with a maximum adsorption capacity (q_max) of 70.68 mg/g. Kinetics analysis favours the pseudo-second-order model. Thermodynamic results suggest an exothermic and spontaneous adsorption mechanism. The decision tree machine learning (ML) model outperformed the gradient boosting (R² = 0.88), random forest models (R² = 0.88), and the artificial neural network model (R² = 0.75) in predicting PCM removal using the adsorbent. Sensitivity analysis using Shapley additive (SHAP) revealed that adsorbent mass is the most influential parameter in PCM removal. This study presented a novel application of activated carbon derived from the Sesbania plant, highlighting its high efficiency in PCM removal through experimental analysis and ML-based optimization.