Machine learning-aided interpretation of interfacial functional-group regulation and kinetic transition in bioleaching by Acidithiobacillus thiooxidans

Extracting rare earth elements (REEs) from secondary sources helps relieve REE supply shortages and mitigate environmental risks from tailings stockpiling. Bioleaching, an environmentally promising strategy for REEs recovery, faces bottlenecks including slow microbial adaptation and variable leaching efficiency. This study explores the interfacial characteristics and underlying regulatory mechanisms of Acidithiobacillus thiooxidans ( At.t ) during bioleaching of rare earth tailings, where biological, chemical and physical factors jointly govern the overall reaction. Firstly, Batch leaching trials, combined with SEM‑EDS, 3D‑EEM, FTIR, and XPS analyses, were performed to investigate the surface interactions between microbial metabolites and the mineral surface during bioleaching. Additionally, the CatBoost model (R 2 0.9905) was utilized to analyze 4968 leaching datasets. The optimal operating ranges were identified as sulfur dosage of 5–10 g/L, liquid‑solid ratio (L‑S Ratio) of 200–600 mL/g, and bioleaching pH of 2.0–4.0. Subsequently, shrinking core model analysis showed that diffusion control predominated during leaching across pH 2.0–6.0, whereas a chemically controlled mechanism dominated across all tested L-S ratios. Sulfur addition markedly changed the reaction‑controlling pathway: increasing sulfur dosage from 0 to 10 g/L shifted the leaching process from chemical‑reaction‑limited to diffusion‑limited. Finally, Scaled‑up verification was performed in a 2 L batch stirred-tank reactor under optimized conditions (pH 2.0, sulfur addition of 5 g/L, L-S Ratio of 200 mL/g). The ultimate leaching efficiencies of lanthanum (La), dysprosium (Dy) and yttrium (Y) were 90.65%, 93.88% and 94.54%, respectively. The favorable leaching outcomes validate the technical strengths of this bioleaching strategy, providing theoretical support for further scale‑up tests.