Efficient sulfadiazine removal from wastewater using lotus-based biochars: Characterization, adsorption, and machine learning prediction

Antibiotic pollutants are persistent in aquatic environments and pose potential risks to human health through various exposure pathways. Biochar adsorption has emerged as an efficient and cost-effective method for removing antibiotics from environmental matrices.In this study, biochars were prepared via pyrolysis from four different parts of lotus plant, among which the biochar from the lotus stem (LSt-BC) exhibited the highest specific surface area (624.7 m 2 ·g −1 ). LSt-BC showed superior adsorption performance for sulfadiazine (SDZ), with optimal efficiency observed at a solution pH of 2.5. The adsorption behavior was well described by the Langmuir isotherm and pseudo-second-order kinetic models, yielding a maximum adsorption capacity of 38.63 mg·g −1 . Characterization analyses indicated that the dominant adsorption mechanisms include π–π stacking interactions, pore-filling effects, and interactions involving C C and C-O functional groups. To further elucidate the adsorption mechanism, three machine learning models—Extreme Gradient Boosting (XGBoost), Random Forest, and Artificial Neural Network—were employed. Among them, XGBoost exhibited the best predictive performance, with a correlation coefficient of 0.965. Feature importance analysis based on the machine learning models revealed that, in addition to biochar dosage and initial pollutant concentration, solution pH significantly influenced SDZ adsorption. These findings highlight the potential of machine learning-assisted optimization for enhancing adsorption conditions. Overall, LSt-BC presents a promising adsorbent for the remediation of SDZ-contaminated water.