Machine-learning-assisted optimization of glutamate leaching followed by sulfide precipitation for base metal recovery from waste printed circuit boards

Glutamic acid (or glutamate) is one of the most promising amino acids shown to recover base metals from e-waste in terms of cost, selectivity, and bulk availability. This study integrates machine learning (ML) to optimize glutamate leaching of waste printed circuit boards (PCBs) and the subsequent sulfide precipitation for metal recovery. Among the evaluated ML models, the tree-based random forest (training R 2 = 0.99, testing R 2 = 0.77 for Cu and training R 2 = 0.98, testing R 2 = 0.88 for Zn) and boosted tree (training R 2 = 0.99, testing R 2 = 0.93 for Pb) models exhibited the best prediction accuracy for Cu, Zn, and Pb extraction as a function of experimental variables. Parameter contribution and SHapley Additive exPlanations analyses revealed that Cu and Zn extraction were mainly affected by leaching time, initial pH, and glutamate dosage, whereas Pb extraction was affected by multiple variables. Sulfide precipitation efficiently recovered 96.9% of Cu, 100% of Co, 96.0% of Pb, 46.2% of Sn, and 10%–17% of Zn, Al, and Ni as mixed sulfides from the leachates, with CuS purity reaching 85.2%. Moreover, the regenerated glutamate ligand released during precipitation enabled faster leaching kinetics and higher Cu extraction in a subsequent cycle. Thus, our results demonstrate a sustainable data-driven pathway for base metal recovery from e-waste based on amino acids. Moreover, the framework developed here can also be applied to other complex resources.