In a major development, researchers at Microsoft have managed to slash the time taken to discover new materials for advanced batteries from months or years to less than 80 hours. This breakthrough was achieved using a combination of artificial intelligence (AI) and cloud computing power.
Accelerating the Pace of Sustainable Energy Breakthroughs
By deploying around 1,000 virtual machines in the cloud, the scientists were able to screen over 32 million candidates in record time*. This allowed them to predict approximately half a million potentially stable compounds that could be used as electrolytes in solid-state batteries.
Microsoft released ‘Azure Quantum Elements’ AI systems in 2023. They were keen to research and prove that their technology could identify new materials that could be synthesised for use in batteries. To explore the reduction of lithium use, they developed a cloud-based computing system which slashed the manual time taken to research. Microsoft aimed to improve daily life with something that was widely used globally, hence the battery research.
Cornell University released the paper on 8th January 2024, based on the description in the paper, it seems they developed custom AI models and workflows to do the high-throughput computational screening and prediction of promising solid electrolyte materials.
Such batteries promise much higher energy densities and safety compared to conventional lithium-ion batteries. However, identifying suitable solid electrolytes has been a major roadblock.
AI’s Efficiency in Material Science: Rediscovering Decades of Knowledge
The AI approach proved incredibly efficient at navigating the colossal search space and zeroing in on the most promising materials. It was able to rediscover decades of accumulated knowledge in the field as a side effect.
The researchers then experimentally tested the best predictions, including a series of sodium-lithium-yttrium chloride compounds. Early results indicate these materials demonstrate excellent conductivity and stability – making them prime candidates for next-gen batteries.
This incredibly rapid process demonstrates the power of AI-assisted materials discovery. By taking a brute force approach leveraging cloud computing, the scientists were able to accomplish in days what would have previously taken years. Their methodology could massively accelerate innovation and lead to breakthroughs in sustainable energy and other key technologies.
The Brute Force of AI and Cloud Computing in Scientific Innovation
Battery technology has been a key part of research, especially in recent years, given they offer the potential for more sustainable forms of energy as well as being key to technologies such as self-driving cars and more.
“The development of novel batteries is an incredibly important global challenge,” said Brian Abrahamson, the chief digital officer at PNNL, in a statement. “It has been a labor-intensive process. Synthesizing and testing materials at a human scale is fundamentally limiting.”
Microsoft then screened more than 32 million potential materials and found more than 500,000 stable candidates. It reported those results last August but noted that screening materials in that way is usually only the start of scientific breakthroughs.
This time around, Microsoft used the AI system and worked with the Department of Energy’s Pacific Northwest National Laboratory to identify a material that was both unknown and not present in nature. It showed potential to make resource-efficient batteries, it said.
Scientists from the laboratory then synthesised the material and made it into a working prototype. that demonstrated that it really worked in practice, and that it has a potential to be a new way of storing energy.
Chi Chen, Dan Thien Nguyen, Shannon J. Lee, Nathan A. Baker, Ajay S. Karakoti, Linda Lauw, Craig Owen, Karl T. Mueller, Brian A. Bilodeau, Vijayakumar Murugesan, Matthias Troyer
Subjects: Materials Science (cond-mat.mtrl-sci); Computational Physics (physics.comp-ph)
*Cite as: arXiv:2401.04070 [cond-mat.mtrl-sci]