Researchers at UC San Diego have used a supercomputer and artificial intelligence to improve sodium-ion battery materials, a step that could lower the cost of large-scale energy storage for power grids.

The team used the Expanse supercomputer at the San Diego Supercomputer Center to study how small material changes inside a battery cathode could boost energy storage and extend battery life.

Sodium-ion batteries are seen as a lower-cost alternative to lithium-ion cells because sodium is abundant and widely available. That makes them attractive for storing renewable energy from solar and wind farms, where cost is a major factor.

But sodium batteries have struggled with lower performance and faster degradation than lithium-based batteries, especially under high-voltage use.

Tiny tweaks, big gains

To address that, scientists modified an existing sodium-based cathode material by adding small amounts of lithium and titanium.

“These subtle changes turned out to matter a lot: the modified material could store more energy and remained stable even when the battery was pushed to higher voltages, a key requirement for getting more energy out of each charge,” explained Professor Shirley Meng of UC San Diego.

“In lab tests, the improved cathode held significantly more charge and kept most of its capacity after many cycles, even under demanding high-voltage conditions that usually cause sodium materials to break down more quickly.”

The challenge for researchers was understanding exactly why those minor chemical changes had such a large impact on performance.

That is where Expanse came in. Using computing allocations through the U.S. National Science Foundation ACCESS program, the team ran large-scale simulations of sodium-ion movement through the material’s crystal structure during charging and discharging.

AI speeds discovery

The simulations relied on AI models known as foundation potentials, which can perform atom-level calculations faster and at lower cost than traditional computational methods.

Researchers said the digital modeling showed that lithium and titanium helped sodium ions move more freely while preventing the crystal framework from collapsing during repeated use.

“By narrowing down promising designs on Expanse before heading into the lab, we were able to move much faster than if we had relied on trial and error alone,” said Shyue Ping Ong, a UC San Diego professor and collaborator on the project.

“Our results point to a practical pathway for improving sodium-ion batteries, making it more feasible to build large battery farms that store renewable energy and release it when the sun isn’t shining or the wind isn’t blowing.”

The work also highlights how supercomputers are becoming key tools in battery development. Instead of relying only on lab experiments, scientists can now simulate thousands of possible material combinations before building prototypes.

That could shorten development timelines for next-generation batteries used in grid backup systems, renewable power storage, and future electric vehicles.

The findings were published in Advanced Energy Materials.