Sugarcane bagasse-MnFe2O4 magnetic biochar composite for ammonium removal: isotherms, kinetics, RSM optimization and machine learning prediction.

Ammonium (NH4 +) pollution in wastewater poses severe ecological and health risks. This study introduces a magnetic biochar composite (SB@MFO) synthesized from sugarcane bagasse and MnFe2O4 for efficient NH4 + removal. The composite exhibits a mesoporous structure (BET surface area: 56.58 m2 g-1; average pore diameter: 11.26 nm) and contains abundant oxygen-containing functional groups that facilitate adsorption. Response surface methodology based on the Box-Behnken design was employed to optimize the adsorption conditions. Under the optimum conditions (MFO : SB = 1 : 3, pH = 7, contact time = 20 min, and adsorbent dosage = 0.03 g·50 mL-1), the experimentally determined adsorption capacity reached 46.66 mg g-1. Equilibrium adsorption was satisfactorily described by both the Langmuir and Temkin isotherm models. The Langmuir model yielded the highest coefficient of determination (R 2 = 0.9812) and a maximum adsorption capacity of 47.80 ± 1.72 mg g-1, while the Temkin model exhibited the lowest RMSE, χ 2, SSE, and AIC values among the evaluated models. The adsorption kinetics were adequately described by the pseudo-second-order (R 2 = 0.9564) and Elovich (R 2 = 0.9052) models, suggesting the involvement of surface-controlled adsorption processes. Combined characterization and adsorption-modelling results suggested that NH4 + uptake by SB@MFO may involve the contribution of electrostatic attraction, ion exchange, surface-mediated interactions involving oxygen-containing functional groups, and physical adsorption within the mesoporous structure. Three machine-learning models were developed to explore the adsorption capacity, with the support vector machine (SVM) model showing the best performance within the investigated dataset (R 2 = 0.994 and RMSE = 1.558). Feature-importance analysis identified the initial NH4 + concentration and adsorbent dosage as the most influential predictive variables. The SB@MFO composite retained 75.8% of its initial adsorption capacity after five adsorption-desorption cycles, demonstrating acceptable reusability. Owing to its adsorption performance, magnetic recoverability, and agricultural waste origin, SB@MFO shows promising potential for ammonium removal from aqueous systems. Further studies using real wastewater matrices and continuous-flow conditions are recommended to evaluate its practical applicability.