An applied machine learning framework for waste lithium-ion battery leaching with integrated preliminary environmental and economic assessment

Abstract

Technological innovation has led to the widespread adoption of lithium-ion batteries (LIBs) for portable energy storage. Correspondingly, sustainable solutions to end-of-life battery disposal are crucial to manage their growing volume. Beyond their potentially hazardous nature, waste LIBs contain several economically relevant critical raw materials such as lithium, manganese and cobalt. However, their recovery by hydrometallurgical approaches often relies on the excessive use of corrosive solutions during the leaching steps, negatively impacting the atomic efficiency, effluent treatment, and cost of the process. Despite numerous studies on the topic, identifying optimal leaching conditions is challenging given the variety of available battery chemistries and leaching agents, compounded by economic and environmental concerns. This work presents a methodical, data-driven approach to model the leaching of key metals from oxide-based LIB cathode active materials using machine learning algorithms, implementing pairwise difference algorithms for data augmentation. The developed model underwent thorough evaluation and screening, and its output is integrated to compute a simplified economic and environmental assessment, accounting for key performance indicators such as heating requirements, solvent cost, and environmental impact, thereby enabling an agile screening of potential preliminary leaching conditions. The methodology described herein is an important step in integrating emerging computational tools in the development of novel, greener metal recycling processes.