Experimental, Machine-Learning, and Computational Studies of the Sequestration of Pharmaceutical Mixtures Using Lignin-Derived Magnetic Activated Carbon

Abstract

Pharmaceutical pollutants pose significant risks to human health and aquatic ecosystems. This study investigates lignin-derived magnetic carbon composite (L-MAC) for removing atenolol (ATN), carbamazepine (CBZ), diclofenac (DCF), and sulfamethoxazole (SMZ) from aqueous media. Characterization of L-MAC’s physicochemical properties, along with isotherm and kinetic studies, revealed that the Langmuir and pseudo-second-order models best describe sorbent–sorbate interactions, with maximum adsorption capacities ranging from 11.30 to 27.97 mg/g. The adsorption efficiency followed the order ATN < SMZ < CBZ < DCF, achieving over 99% removal under optimal conditions of 1–4 h contact time and pH 2–7. Strong π–π interactions, hydrogen bonding, and chemisorption contributed to sorption irreversibility. Artificial intelligence models predicted a material performance with high accuracy. The adaptive neuro-fuzzy inference system model outperformed others, achieving error coefficients of 5.745, 3.125, and 11.085 during training and 6.123, 4.974, and 12.456 during testing. Density functional theory analysis examined reactivity and binding strength using descriptors like HOMO–LUMO energy gaps. DCF showed the highest electron-donor capacity, followed by CBZ, ATN, and SMZ, confirming L-MAC’s high efficacy in removing pharmaceuticals. This study demonstrates L-MAC’s robustness for the adsorptive removal of contaminant mixtures.