In the relentless march of technological innovation, the use of lithium-ion batteries has become ubiquitous. While many researchers are striving to find materials suitable for the batteries of the future, it might just be that an artificial intelligence (AI) has now achieved this feat, and in a record time.
A remarkable breakthrough has indeed been made by Microsoft in collaboration with the Pacific Northwest National Laboratory (PNNL), making use of AI to identify innovative solid electrolytes. This development could signify a turning point in battery technology, with a potential reduction in combustion risk and a decrease in lithium usage.
Samples of the new solid electrolyte discovered by AI and Microsoft's HPC tools. Solid electrolytes are safer than liquid ones.
Photo by Dan DeLong for Microsoft. The project began with the examination of 32 million potential materials. Using AI and high-performance computing (HPC) in the cloud, this list was narrowed down to 23 promising candidates, of which 18 were previously unknown. This swift and accurate method contrasts with traditional, lengthy, and expensive methods. Indeed, the AI was able to achieve this result in just 80 hours. Ultimately, the entire process of creating this material took just a few weeks, whereas it would have taken several years using traditional techniques!
The AI efficiently screened materials based on various criteria such as reactivity and conductivity. After thorough analysis, six materials, specifically suited for use in batteries, were finalized. This advancement is noteworthy since lithium-ion batteries, despite their widespread use, have environmental repercussions. The extraction of lithium, essential for these batteries, involves evaporating brine and extracting hard rock minerals. Not only is this operation costly, but it also has detrimental environmental impacts, raising questions about the sustainability of these technologies. Moreover, the limited geographical distribution of lithium deposits and the time required for the evaporation process can lead to supply fluctuations.
The research conducted by Microsoft and PNNL is therefore a significant step towards more sustainable alternatives. Their discovery could reduce the use of lithium by up to 70%, partly replacing it with sodium—an element that is more abundant and easier to mine. However, the conductivity of batteries made with these new materials remains lower than that of batteries with higher lithium content.
These discoveries are still in the early stages. Large-scale tests have not yet been conducted, and numerous challenges must be overcome before commercial implementation. Indeed, transitioning from a laboratory discovery to a marketable product is a complex and uncertain journey. Nonetheless, the use of AI in research is speeding up the discovery process, thereby increasing the likelihood of significant innovations.
Article author: Cédric DEPOND