Quantum-assisted machine learning screening for sustainable anode discovery in lithium-ion batteries
Marco Fronzi, Catherine Stampfl, Amanda Ellis, Eirini Goudeli
Abstract
A comprehensive analysis of 9835 crystal structures, 211 of which are calculated to be thermodynamically stable, is presented, assessing their potential as anode materials for lithium-ion batteries. Density functional theory (DFT) calculations and advanced machine learning techniques are employed to explore the stability, lithium diffusion, bulk modulus and shear stress, along with the relationships between atomic orbital overlap, energy density, and ion mobility, which is a crucial factors for rapid charging capabilities. The study also examines the combined effects of elemental composition and crystallographic space groups to identify the key drivers of structural toughness. A number of crystal structures are identified as promising anode materials, with some standing out for their exceptional stability and efficient lithium-ion mobility. These materials demonstrate significant potential for high-capacity, durable battery anodes, highlighting the importance of a multidimensional approach in battery material development. These insights provide a novel perspective on the interplay between physical, chemical, and electronic properties in optimising anode materials. This work offers valuable guidance for the future design and development of high-performance lithium-ion batteries, contributing to a more sustainable economy. • Quantum-assisted machine learning screens 9835 crystal structures for LIB anodes. • Identifies 211 thermodynamically stable materials with high lithium-ion mobility. • Novel RE*(recyclable element)-based anodes support a sustainable circular economy. • RE*-based materials outperform silicon anodes with enhanced stability and capacity. • Combines DFT and ML for scalable, sustainable anode material discovery.