Unveiling modified montmorillonite enhancing the remediation ability of biochar for phthalate ester and Cd co-contaminated soil using an explainable machine learning approach.

The integration of sulfhydryl-modified montmorillonite (MMT-SH) with biochar (BC) shows significant promise for remediating phthalate esters (PAEs) and Cd co-contaminated soil. However, the synergistic remediation mechanism of composites in contaminated soil has not been thoroughly investigated. In this work, the EconML model together with the SHAP method was employed to evaluate the remediation ability and underlying mechanism of BC and MMT-SH composites. The results indicated that the composites enhanced the synergistic removal of ΣPAEs and exchangeable Cd, up to 45.89 % and 60.36 %, respectively, while also facilitating the conversion of pollutants from free to bound forms. Among them, composites with a lower proportion of MMT-SH increased the stability of the soil bacterial community, alleviated the negative effect of biochar on soil bacterial community richness, and increased the relative abundance of bacteria related to soil element cycling and pollutant remediation. In addition, a higher C/N ratio (> 54.18) exerted a significant positive effect on the synergistic remediation ability of composites, serving as the primary predictive feature for remediation ability. Clay content and sobs were identified as auxiliary and supplementary features in actual environmental treatment. This study represents a novel attempt to employ machine learning methods in explaining and guiding the preparation and application of biochar-based composites for remediating organic and inorganic co-contaminated soil, which holds great significance for environmental protection.