Rare earth elements recovery from end-of-life NdFeB magnets towards new applications using design of experiments for leaching process

This study investigates the recovery of rare earth elements (REE) from end-of-life (EoL) NdFeB permanent magnets through acid leaching, combining Design of Experiments (DoE) and exploratory machine learning (ML) tools to optimize process conditions. Acid leaching experiments were designed using a Box–Behnken approach, in which acid concentration, temperature, and solid-to-liquid (S/L) ratio were used as experimental factors. Regression modeling and response surface methodology were applied, while ML tools − Principal Component Analysis (PCA) and k-means clustering − were used to provide complementary insights into the multivariate patterns. Hydrochloric acid (HCl) and nitric acid (HNO 3 ) were evaluated, with HCl demonstrating superior leaching efficiency, higher magnet mass degradation, and improved metal recovery. The models indicated that neodymium (Nd) leaching was primarily influenced by the interaction between acid concentration and S/L ratio, while praseodymium (Pr) recovery was mainly controlled by acid concentration. Response surface plots (RSPs) confirmed that higher acid concentrations and temperatures enhanced leaching performance, whereas higher S/L ratios reduced efficiency. Exploratory ML analysis corroborated the trends identified by DoE, supporting the interpretation of optimal operating conditions. Under optimized parameters, near-complete magnet degradation (98.99 %) and high REE recovery were achieved, with effective extraction occurring within 0.5 h. These findings demonstrate the feasibility of efficient REE recovery from NdFeB magnets and provide a robust framework for process optimization and future scale-up.