Microsoft AI accelerates new solid-state battery discovery

One of Artificial Intelligence’s (AI) greatest promises is the faster development of new scientific breakthroughs. This is no longer merely a promise: it is already becoming reality.

In June 2023, Microsoft launched Azure Quantum Elements to “accelerate scientific discovery with the power of Artificial Intelligence (AI)” alongside high-performance computing (HPC), and its initial results are already apparent.

“Demonstrating what can be developed with AI is not the same as proving that this technology can identify something new.”

Dr Nathan Baker, product lead, Azure Quantum Elements, Microsoft

Microsoft chose to apply the technology to the research and development of new battery chemicals, particularly for solid-state batteries, and the early findings are encouraging.

First steps with Azure Quantum Elements

The Azure Quantum team joined forces with the Department of Energy’s Pacific Northwest National Laboratory (PNNL), presenting its first findings in August 2023.

Following several months of work and tests involving 32 million materials that could potentially be used in battery manufacturing, the artificial intelligence system determined that more than 500,000 were “stable”. This marked the first stage of the process.

Those 500,000 materials were then narrowed down to a single candidate, based on several factors and functional properties required to produce a solid-state battery. Find out more about this type of battery:

After characterising the selected material’s structure, assessing its connectivity and exposing it to tests at various temperatures, the team evaluated its technical feasibility using an experimental battery.

New solid-state battery

After nine months of research, PNNL identified a material - one not found in nature - that uses 70% less lithium than current batteries, partly addressing the shortage of this raw material. Although there is still a long way to go, AI made it possible to “reduce this research process from years to weeks and from weeks to days”.

Tasks that previously relied on lengthy high-performance computing (HPC) calculations and expensive, time-consuming laboratory research were completed in just nine months with the help of AI.

“PNNL has demonstrated that the potential of new HPC and AI approaches significantly accelerates the innovation cycle.”

Dr Nathan Baker, product lead, Azure Quantum Elements, Microsoft

As well as helping to synthesise millions of material tests, AI enabled researchers to identify a material that uses less lithium by replacing it with sodium.

This new material combination, uncovered through the research, addresses some of the key drawbacks associated with lithium-ion batteries: safety and scarcity. At the same time, it promises greater energy density.

A cheaper, more sustainable battery chemistry

In other words, the research concerns a new battery chemistry that is cheaper, more sustainable and offers higher energy density.

“By using less lithium, this battery will have less impact on the planet, will be safer and will also deliver greater economic benefits.”

Dr Nathan Baker, product lead, Azure Quantum Elements, Microsoft

However, this is only the beginning of a development process that could be shortened further with AI’s assistance.

Either way, there is still a considerable distance to cover before this new battery chemistry is used in cars, computers, smartphones and other devices that require an internal energy-storage source.

Source: Microsoft