The increasing popularity of lithium-ion batteries (LIBs) as both portable and large-scale energy storage devices necessitates sustainable solutions for managing their eventual disposal. LIBs contain critical raw materials, such as lithium, cobalt, and graphite, whose supplies are considered strategically, socially, and economically important by the European Union. Therefore, waste LIBs represent a significant urban ore, a recovery opportunity for ensuring a more sustainable and circular economy of such materials.

Efficient recovery and separation are of paramount importance. However, identifying optimal leaching conditions can be complex due to the diversity of battery chemistries, leaching agents, and the need to balance economic and environmental considerations. While some attempts have been made to apply machine learning techniques to tackle this interesting optimization problem, these efforts have often overlooked the practical aspects of implementation and real-world applications.

In this work, a methodical data-driven approach to modelling the leaching of key metals from LIB cathodes is presented. The existing literature is leveraged to construct a leaching model using machine learning algorithms, allowing for efficient and agile screening of leaching conditions. Our model considers key performance indicators such as yield, selectivity towards a particular metal, heating requirements, solvent costs, CO₂ emissions and other environmental impact indicators to provide a preliminary economic and environmental assessment of different leaching strategies. This approach enables researchers to identify promising conditions that maximize metal recovery while minimizing environmental harm. To showcase the practical application of this methodology, a user-friendly graphical interface was developed to leverage machine learning packages currently available for Python for this purpose. This makes these powerful tools accessible to a wider audience, including researchers and industry professionals who may not have a strong computational background.

The data-driven methodology presented here represents a significant advance in integrating computational tools into the development of novel, greener metal recycling processes. By providing a more comprehensive approach to evaluating leaching conditions, these tools help advise the development of more sustainable and economically viable LIB recycling practices. Furthermore, this approach can be extended to address other waste recycling challenges, offering a versatile framework for optimizing resource recovery across various industries.