According to news on January 10, Microsoft’s quantum computing team collaborated with the Pacific Northwest National Laboratory (PNNL) under the U.S. Department of Energy, using Azure Quantum Elements high-performance computing resources to conduct simulations and artificial intelligence models, and discovered a new battery material that can replace about 70% of the lithium in the battery with sodium. This new material is expected to be widely used in modern life, from smartphones to electric vehicles to power grids.

The following is the full text of the translation:

Recently, Microsoft's quantum computing team conducted an innovative experiment by modifying a clock to run an experimental battery they developed in collaboration with PNNL researchers. They also added the logo of Azure Quantum Elements, Microsoft's platform for artificial intelligence (AI)-enhanced scientific discovery, which has played a major role in developing new battery technology, on the battery's casing.

Brian Bilodeau, director of partnerships, strategy and operations at AzureQuantumElements, explained that the purpose of this DIY project was to demonstrate the potential of batteries. "We wanted to create a wow moment," he said, and the moment was designed to impress Microsoft CEO Satya Nadella.

Catching Nadella's attention won't be easy. However, dedicating vast amounts of Azure high-performance computing (HPC) resources to solving tough technical challenges is exactly what Nadella is interested in. “I was very, very excited to see this project succeed,” Nadella recalled.

The battery that powers this clock, code-named CR2032, is small and looks like the coin cells we commonly see in pocket calculators or garage door openers. However, inside the battery, Microsoft researchers used a solid electrolyte that replaced about 70% of the battery's lithium with sodium. This design has the potential to solve a range of problems faced by lithium batteries, such as limited charging life, capacity degradation over time, poor performance in extreme temperatures, and risks such as possible fires and even explosions. In addition, reducing reliance on lithium and switching to abundant and cheap sodium may help relieve pressure on the battery supply chain.

As technology develops further, this new material is expected to be used in all aspects of modern life, from smartphones to electric vehicles to power grids. However, Microsoft sees this breakthrough as a strong validation of the Azure Quantum Elements platform, which is designed to serve customers. Launched in June last year, AzureQuantumElements is still in a "private preview" stage and has been tested by institutions such as Johnson Matthey in the UK. The agencies are using the platform to help design catalytic converters and hydrogen fuel cells.

Nathan Baker, Microsoft's senior director of chemistry and materials partnerships, who joined Microsoft in 2022 after 12 years at PNNL, sees the battery research endeavor as an example of Microsoft's customer-first philosophy. "When developing a new product, the first thing we say is: 'Okay, let's prove to ourselves that this works, let's try it ourselves,'" he said.

Nadella also emphasized that the experience gained from the battery experiment will be used to improve the Azure Quantum Elements platform. "Ultimately, what matters is being able to make powerful technology accessible to ordinary people so that others can develop more of it," he said. "I try to avoid talking about 'our great scientific achievements.' That's not our style."

Beyond trial and error

Even scientists at the 59-year-old PNNL knew that working with Microsoft was part of a larger goal. How to make better batteries has long been a key area of ​​research at PNNL. The agency is headquartered in Richland, Washington, about 320 kilometers from Microsoft's headquarters in Redmond. Scientists here understand that troubled laboratory research can lead to dead ends.

More than 100 years ago, Thomas Edison tested thousands of filaments before discovering a material that made the light bulb a reality. This willingness to try anything became known as the Edison method and is still used in scientific discovery.

"It's always a process of trial and error," Vijay Murugesan, head of PNNL's materials science group, said of traditional research methods. "Even if you think of something in a dream or in the shower, it might take two years to test its effectiveness. Then, repeat the cycle for another ten years. To be honest, the success rate is not that high."

The impact of this traditional research method is far-reaching. PNNL's research areas are broad, ranging from coastal science to national security. However, Brian Abrahamson, deputy director and chief digital officer of the laboratory, said: "The time scale on which we are making discoveries today is not fast enough to meet some of the global challenges we are trying to solve. If we can make significant progress on this, that will be a big deal."

Microsoft's intervention brought a turn for the better. By virtualizing some of that upfront work and handing it over to Microsoft's newest, fastest computers, that could change everything in the coming years. As Nadella said: "We must compress 250 years of progress in chemistry into 25 years, and the only way to achieve this goal is to have powerful computational simulation capabilities."

Taking new battery technology as an example, Microsoft started with a database containing 32.6 million materials and used Azure Quantum Elements high-performance computing resources to conduct simulations and artificial intelligence model runs, significantly reducing the number of materials to 500,000, then 500, 150, and finally only 18. The company consulted PNNL materials science experts to further shortlist the list and ultimately determined that the lithium-sodium material was worthy of being synthesized and tested in the laboratory.

"It only took us two weeks to sift through so many possibilities and eliminate most of the non-conforming materials," said Jason Zander, Microsoft's vice president of strategic mission and technology execution and a 31-year veteran of the company. "This is much faster than traditional methods, which can sometimes take years." Overall, Microsoft's battery project took just one year from inception to manufacturing about 40 test cells. In addition to powering the clock, these batteries are also used to light up the Microsoft logo lights.

Although high-performance computing plays a key role in the project, testing new materials still requires physical verification. At PNNL, materials scientists need to perform complex steps such as manual grinding, pressure testing, and melting at high temperatures of up to 600 degrees Celsius to verify the relevant materials. Sometimes it’s necessary to be bold and try new technologies. For example, PNNL researchers found that some materials identified through the algorithm were not easy to manipulate for practical applications. "After being left to dry in the open air for 30 minutes, some materials will absorb moisture from the air and turn into a sticky substance," said PNNL materials scientist Shannon Lee.

While traditional laboratory experiments may be outdated, these researchers remain convinced that commercial collaborations in computing are the way forward. PNNL has established flexible partnerships with cloud service providers such as Microsoft to meet its on-demand resource needs. This partnership not only helps accelerate scientific discovery, but also provides government agencies with a new model for working with commercial companies.

"We're particularly leaning toward commercial partnerships in the computing space," said Tony Peurrung, PNNL's deputy director of science and technology. He added that being "in Microsoft's backyard" has not hurt the lab's relationship with the software giant, and PNNL also has agreements with other cloud computing providers.

Inseparable from quantum computing

Speaking of the battery projects of Microsoft and PNNL, there is one thing that must be mentioned. Although the Azure Quantum Elements platform played a key role in the development process, quantum computing was not directly involved. However, employees from Microsoft's quantum computing division did participate in the project, making the achievement inextricably linked to quantum technology.

All of this can be explained by the current state of quantum computers. Quantum computing utilizes the principles of quantum physics. Unlike the binary bits of traditional computers, qubits can represent 0 and 1 at the same time. This amazing technology promises to make computers incredibly powerful at processing complex calculations, beyond the capabilities of existing computers. Quantum computers will play a huge role in tasks that require large amounts of computing resources, such as scientific discovery.

Like other tech giants such as IBM and Google, as well as numerous startups, Microsoft has high hopes for the potential of quantum technology. However, this technology is still in the development stage and has not yet become a commercial reality. Krista Svore, senior vice president of quantum development at Microsoft, said: "We believe that the development of quantum technology is measured in years, not years."

At the same time, traditional computers have not been eliminated. Computers equipped with powerful GPU chips are enabling more transformative artificial intelligence capabilities. This advancement enables high-performance computing in the cloud, which is increasingly becoming a service, such as that provided by Azure resources for pinpointing battery materials.

The development of events has even exceeded the expectations of supporters of the quantum field. Nadella and others originally thought quantum computing would be the next big breakthrough, but it turns out that GPUs are.

No matter how quantum computing technology develops, it is unlikely to completely replace high-performance computing used to find battery materials. Swore predicts that in the future, scientists will be able to build a system that integrates high-performance computing, artificial intelligence, and quantum computing to provide more accurate solutions to currently unsolvable problems.

When that era truly arrives, the boundaries of exploration will be infinitely expanded. Jason Zander, executive vice president of strategic missions and technology at Microsoft, felt deeply: "We think we know a lot, but in fact, our understanding of many things is still very superficial. For example, we can't even explain how photosynthesis works."

Microsoft's collaborative project with PNNL provides a valuable case study for the Azure QuantumElements discovery pipeline. The potential value of battery technology is a key factor driving Microsoft's research. Brian Bilodeau, director of partnerships, strategy and operations at AzureQuantumElements, said frankly: "This kind of project is easier for people to recognize and understand."

Although Microsoft and PNNL researchers did not disclose how the battery technology in the laboratory can be scaled up, Vijay Murugesan, head of PNNL's materials science group, admitted: "There is still a long way to go before these technologies can be truly applied to the consumer market." In the short term, they plan to manufacture and test larger batteries, such as "pouch" batteries that are widely used in electronic products, electric vehicles and other fields.

Microsoft seems to see a huge business opportunity lurking in this, although the specific details have not yet been determined. Bilodeau said: "Microsoft will not get involved in the battery manufacturing business, and similarly, PNNL will not get involved in this area. Large-scale production will be a cooperative project with other companies."

Regardless of how this technology will be used in the future, Microsoft has achieved its original goal. Nadella proudly said: "The reason why I like this project is that it not only achieves a theoretical breakthrough, but also proves the feasibility of a trial-and-error cycle in practice. Through this method, we can create truly efficient new materials, perhaps even at a speed that was unimaginable just a few years ago." (Xiao Xiao)