The combination of advanced artificial intelligence and next-generation cloud computing is accelerating scientific discoveries at unimaginable speeds just a few years ago. Microsoft and the Pacific Northwest National Laboratory (PNNL) in Richland, Washington, are collaborating to demonstrate how this acceleration can benefit chemistry and materials science, two crucial fields for finding the energy solutions the world needs.
Scientists at PNNL are testing a new battery material that was discovered in a matter of weeks, not years, as part of the collaboration with Microsoft, using advanced AI and high-performance computing (HPC). HPC is a type of cloud-based computing that combines large amounts of computers to solve complex scientific and mathematical tasks.
As part of this effort, Microsoft’s Quantum team used AI to identify around 500,000 stable materials in just a few days. The new battery material emerged from a collaboration that used Microsoft’s Azure Quantum Elements to reduce 32 million potential inorganic materials to 18 promising candidates in just 80 hours. This work lays the groundwork for a new way to accelerate solutions to urgent challenges in sustainability, pharmaceuticals, and other fields, while offering a glimpse of the advancements that will be possible with quantum computing.
“We believe there is an opportunity to do this in various scientific fields,” says Brian Abrahamson, digital director at PNNL. “Recent technological advancements have opened up the opportunity to accelerate scientific discovery.”
A New Approach to Materials Discovery
PNNL is a United States Department of Energy laboratory conducting research in various areas, including chemistry and materials science, with goals of safety and energy sustainability. This makes it the ideal collaborator for Microsoft, leveraging advanced AI models to discover new battery material candidates.
“The development of new batteries is an incredibly important global challenge,” says Abrahamson. “It has been a labor-intensive process. Synthesizing and testing materials on a human scale is fundamentally limiting.”
From Hypothesis to Practice
Traditionally, the first step in material synthesis is to review all published studies on other materials and formulate hypotheses about how different approaches might work. “One of the main challenges is that people publish their success stories, not their failure stories,” says Vijay Murugesan, leader of the materials science group at PNNL. This means that scientists rarely learn from the failures of others.
The next traditional step is to test the hypotheses, a long and iterative process. “If it’s a failure, we go back to the drawing board,” says Murugesan. In a previous project at PNNL, a vanadium redox flow battery technology took several years to solve a problem and design a new material.
AI to the Rescue
Microsoft trained different AI systems to sophisticatedly evaluate all usable elements and suggest combinations. The algorithm proposed 32 million candidates. Then, the AI filtered stable materials, eliminating candidates based on their reactivity and potential to conduct energy.
“In every step of the simulation where I had to run a quantum chemistry calculation, I instead call the machine learning model. That way, I get the perception and detailed observations that come from running the simulation, but the simulation can be up to 500,000 times faster,” says Nathan Baker, product leader for Azure Quantum Elements.
Thanks to this combination of AI and HPC, discovering the most promising candidates took only 80 hours.
Broad Applications and Accessibility
Microsoft scientists used AI to perform the majority of candidate reduction, accounting for around 90% of the computational time. PNNL materials scientists then evaluated the shortlist of materials. Because Microsoft’s AI tools are trained for chemistry, not just battery systems, they can be used for any type of materials research, and the cloud is always accessible.
“We believe the cloud is a tremendous resource for improving accessibility to research communities,” says Abrahamson.
Surprising Results
The newly discovered material that PNNL scientists are testing uses both lithium and sodium, as well as other elements, significantly reducing the lithium content. Despite the process being in its early stages, significant advances have already been made in rapidly identifying viable battery chemistry.
With this collaboration between Microsoft and PNNL, a new era of acceleration in scientific discovery is being ushered in, with AI and HPC transforming the research landscape and offering fast solutions to critical global issues.
References: Microsoft news and Microsoft Quantum Team. Photo by Microsoft.