Microwave-assisted magnetic biochar crosslinked with β-cyclodextrin for tetracycline adsorption: a multidimensional evaluation by site-energy distribution, machine learning, density functional theory insights, and life cycle assessment

A Fe-modified biochar crosslinked with β-cyclodextrin (Fe@RSBC-β-CD) was prepared from rice straw and waste siderite via a microwave-assisted crosslinking route and applied to the adsorption of tetracycline (TC) from water. Characterizations revealed a well-developed hierarchical pore network, abundant oxygen-containing functionalities, and a favorable magnetic response. Under the standard batch condition (pH 7.0, 25 ± 1 °C), Fe@RSBC-β-CD exhibited an experimental equilibrium uptake of 219.47 mg·g -1 . Isotherm analysis further yielded a Sips-fitted maximum adsorption capacity of 292.46 mg·g -1 at 45 °C. Laboratory fixed-bed experiments demonstrated continuous-flow feasibility under the tested conditions. Machine learning was used as a comparative, in-domain prediction tool. Across 20 group-aware random train–test splits, Extra Trees (ET) provided the strongest overall test performance among six ensembles, while the wide dispersion quantified the uncertainty associated with the small dataset. SHAP analysis identified initial concentration as the largest contributor to the ET model. Isotherm fitting combined with site energy distribution analysis evidenced pronounced surface energetic heterogeneity. Integrating spectroscopic evidence with density functional theory calculations, the results suggest that TC adsorption is synergistically driven by π–π interactions, hydrogen bonding, β-CD-associated microenvironmental interactions, and inner-sphere complexation at Fe–O sites. Life cycle assessment further indicates that the upstream supply chains for β-cyclodextrin and NaOH account for the majority of the cradle-to-gate environmental burdens, providing clear targets for greener process optimization. This work provides a laboratory-scale evaluation of a magnetically responsive adsorbent by integrating materials design, mechanistic analysis, in-domain prediction, and screening-level environmental assessment; validation in real wastewater and at larger continuous-flow scales remains necessary.