In the fight against climate change, scientists are working hard to develop better and cleaner energy solutions.
One of the most promising technologies is the solid-state battery—a type of battery that could one day power electric cars, smartphones, and even homes more efficiently and safely than today’s lithium-ion batteries.
But progress has been slow, with researchers testing materials one by one in a long and uncertain trial-and-error process.
Now, a team from Tohoku University in Japan has come up with a smarter way to move forward. They’ve created a powerful AI-based tool that can predict which materials are most likely to work well in solid-state batteries.
Even better, it can explain how and why these materials might work, saving scientists time and effort in the lab. Their findings were published on April 17, 2025, in Angewandte Chemie International Edition.
This new AI system pulls information from a huge database of past experiments and research studies.
It uses advanced methods—like large language models (LLMs), machine learning, and simulations—to scan thousands of possible materials and pick out the most promising ones for solid-state electrolytes. These are key parts of a battery that allow ions to move between the battery’s positive and negative ends.
According to Professor Hao Li, who led the study, this tool is like having an assistant that does all the time-consuming lab work virtually.
It not only narrows down the best candidates but also helps scientists understand what makes a material effective in the first place. The AI even predicts how ions will move through different materials, a process that is critical for battery performance.
The research team made some exciting discoveries, including a unique “two-step” movement of ions in certain materials. This kind of insight could help scientists design batteries that charge faster and last longer. Their tool also showed that a type of computer simulation, called ab initio MetaD, gives results that closely match what scientists have observed in real-world experiments.
All of this data and analysis is now available in a massive new database created by the team, called the Dynamic Database of Solid-State Electrolytes (DDSE).
It’s the largest collection of its kind and could help researchers all over the world make faster progress toward the next generation of high-performance, eco-friendly batteries.
The team hopes to expand the use of this AI approach to other kinds of battery materials and even use generative AI to explore how reactions unfold inside batteries.
