Mechanistic and machine-learning insights into microplastic adsorption on modified magnetic biochar for circular-economy applications

• The study presents a sustainable adsorption technology for MPs removal in water. • Stearic acid-modified magnetic biochar (SMBC) was used for MPs and exhibited ∼94% removal efficiency. • Machine learning models (Gradient Boosting and Decision Tree) enhanced the interpretation of microplastic adsorption behaviour. • SHAP analysis identified contact time, pH, and adsorbent type as the key predictors of removal efficiency. • Microplastic-laden adsorbent was upcycled and used in the decontamination of dye, with ∼91% removal efficiency attained. Microplastic (MP) contamination in aquatic systems poses a significant threat to water quality and ecosystem health, necessitating the development of advanced treatment solutions. This study investigates stearic-acid-modified magnetic biochar (SMBC) for removing polystyrene MPs under varied conditions. This work integrates a greener hydrophobic modification with mechanistic interpretation (wettability–charge–porosity coupling), interpretable machine learning, and an end-of-life upcycling pathway for MP-loaded sorbents. Characterization confirmed that stearic-acid grafting increased surface hydrophobicity and introduced aliphatic –CH domains that favor PS affinity. SMBC achieved up to 94% MP removal, with adsorption kinetics following the pseudo-first-order model (R 2 = 0.987). The Sips model (R 2 = 0.970) best represented the equilibrium data, with n ≈ 1 indicating weak heterogeneity and adsorption behavior approaching Langmuir characteristics. In real water matrices, SMBC retained high performance over five reuse cycles, with a gradual decline after the third cycle. SMBC removed 84% of MPs from rainwater and maintained strong stability over five reuse cycles. Machine-learning models enhanced predictive accuracy and operational optimization; Gradient Boosting and Decision Tree performed best. Shapley Additive Explanations (SHAP) analysis provided interpretable insights, identifying contact time, pH, and adsorbent type as dominant factors, aligning with mechanisms such as hydrophobic interaction, electrostatic attraction, pore filling, and π–π stacking. Spent SMBC loaded with MPs was successfully upcycled via pyrolysis and reused for methylene blue removal (∼91%), demonstrating circular-economy potential. Overall, SMBC offers a robust, sustainable adsorbent for advanced water treatment targeting MP contamination, integrating mechanistic understanding with data-driven insights.