How Microsoft found a potential new battery material using AI

Artificial intelligence (AI) and large-scale cloud computing are accelerating the search for new battery materials. An AI-enhanced collaboration between Microsoft and the Pacific Northwest National Laboratory (PNNL) has already produced promising new material, which the two are publicly sharing today.

They discovered a new type of solid-state electrolyte, the kind of material that could lead to a battery that is less likely to catch fire than current lithium-ion batteries. Less lithium is also being used, which is becoming increasingly difficult to source as demand for rechargeable EV batteries soars.

There is still a long way to go to see how viable this material is as an alternative to traditional lithium-ion batteries. What scientists are most excited about is the potential of generative AI to accelerate their work. This discovery is just the first of many materials they will test in search of a better battery.

“If we can see such an acceleration, I bet this is the way of the future to find these types of materials.”

“The big point we have to make is the speed with which we came up with a new idea, a new material. If we can see such an acceleration, I bet this is the way of the future to find these types of materials,” said Karl Mueller, physical chemist and director of program development at PNNL.

Microsoft reached out to PNNL researchers last year to offer its Azure Quantum Elements (AQE), a platform that brings together high-performance computing and AI – and ultimately quantum computing, Microsoft said. The company launched it last year as a tailored tool for discoveries in chemistry and materials science.

The researchers asked AQE about battery materials that use less lithium, and soon 32 million different candidates emerged. From there, the AI ​​system had to determine which of these materials would be stable enough to use – which ended up being around 500,000. They used more filters to infer how well each material could conduct energy, simulate how atoms and molecules move within each material, and figure out how practical each candidate would be when it comes to cost and availability.

Ultimately, only 23 candidates remained, five of which were already known materials. All the teardown took just 80 hours – a feat so fast it would have been virtually impossible without AI and AQE.

“Thirty-two million is something we could never do…Imagine a human going through 32 million materials and choosing one or two. It’s just not going to happen,” said Vijay Murugesan, staff scientist and materials science group leader at PNNL.

a:hover]:text-gray-63 [&>a:hover]:shadow-underline-black dark:[&>a:hover]:text-gray-bd dark:[&>a:hover]:shadow-underline-gray [&>a]:shadow-underline-gray-63 dark:[&>a]:text-gray-bd dark:[&>a]:shadow-underline-gray”>Photo by Dan DeLong for Microsoft.

PNNL synthesized one promising candidate from that search for testing. They were able to turn it into a working battery and use it to power a light bulb and a clock. Hundreds of prototype batteries will need to be tested and modified before this new material can prove itself. So don’t expect it to hit store shelves anytime soon; A lot of research has been done into promising new materials that never reach the market.

What’s exciting about this particular candidate is that it uses a combination of lithium and sodium, an abundant element and the main component of salt. Microsoft says the new material can reduce the amount of lithium used in a battery by as much as 70 percent.

Additionally, it could be used to create a solid-state battery that is safer than current lithium-ion batteries, made with liquid electrolytes that are more susceptible to overheating. The tricky part is that solid electrolytes generally aren’t as good at conducting energy as their liquid counterparts. That’s a challenge researchers are still trying to overcome with this new material, as it showed lower conductivity in laboratory tests than initially predicted.

Fortunately, there are other promising candidates that researchers can create and test as they try to create the next generation of batteries needed to power the world with renewable energy. Keep in mind that generative AI itself is having an increasing impact on the environment, especially the greenhouse gas emissions associated with all the energy burned with computers. That makes it important to simultaneously increase the energy efficiency of computers and run data centers on clean energy – which requires better batteries.

“We really need to condense the next 250 years of chemical materials science into the next two decades, right? And that’s because we want to save our planet,” said Krysta Svore, who leads the Microsoft Quantum – Redmond (QuArC) group at Microsoft Research. “As you can see from these results, AI and high-performance computing together can accelerate that scientific discovery.”

Correction January 10, 11:00 am ET: An earlier version of this story said the new material could reduce the amount of sodium used in a battery by as much as 70 percent. It has been corrected to say that the material can reduce the amount of lithium by as much as 70 percent. We regret the error.

Leave a Reply

Your email address will not be published. Required fields are marked *