Artificial intelligence is helping scientists discover new medicines, create advanced materials, and now even design better batteries.
Researchers at the University of Chicago have developed a new AI system called “ElectrolyteGPT” that can create complete battery electrolyte formulas, potentially speeding up the search for longer-lasting and higher-performing batteries.
The research, published in JACS Au, focuses on one of the most important yet complicated parts of a battery: the electrolyte.
This liquid or gel allows charged particles to move between the battery’s electrodes, making it possible for the battery to store and deliver energy.
Designing an effective electrolyte is far more difficult than choosing a single chemical. A battery electrolyte is usually a complex mixture of salts, solvents, and additives.
Scientists must carefully balance many properties, including conductivity, stability, viscosity, and efficiency. Improving one property can often make another worse, creating a difficult optimization problem.
Traditionally, researchers have relied on experiments and experience to develop new formulations. However, the number of possible combinations is so enormous that testing them all is impossible.
According to the researchers, the number of potential molecules that could be used in battery electrolytes is estimated to be around 10 to the power of 60—far more than the number of stars in the universe. When all the possible ways of mixing these molecules are included, the possibilities become almost limitless.
ElectrolyteGPT was created to help explore this vast chemical landscape. Unlike many AI tools that focus on finding individual molecules, this system generates complete electrolyte formulations.
It determines not only which ingredients to use, but also their concentrations, mixture ratios, and other important characteristics.
The researchers trained the AI using a carefully selected database of battery-relevant chemicals. This was necessary because many existing AI models were originally designed for drug discovery and tend to generate molecules useful for medicines rather than batteries.
The team also taught the AI to target specific battery performance goals. Instead of simply creating random formulas, ElectrolyteGPT generates candidates designed to meet important requirements such as high conductivity, strong stability, low viscosity, and efficient battery operation.
One of the key innovations behind the project is a new chemical language developed by the researchers called “fLine.” Traditional chemical languages describe individual molecules, but fLine goes much further. It can describe an entire electrolyte formulation, including chemical structures, concentrations, mixing ratios, temperature conditions, and other factors that influence performance.
This allows the AI to understand battery electrolytes as complete systems rather than isolated ingredients.
To test the technology, the researchers synthesized several electrolyte formulations proposed by ElectrolyteGPT and evaluated them in lithium-metal batteries. Some of the AI-designed formulas performed as well as today’s leading electrolyte technologies.
While the system has not yet surpassed the best existing electrolytes, the results demonstrate that AI can generate realistic and effective formulations that rival those developed by expert scientists.
The researchers believe this is an important step toward a future where AI can rapidly design next-generation battery materials. As the models become larger and more sophisticated, they may help discover entirely new electrolyte formulations that improve battery performance, lower costs, and support the growing demand for electric vehicles and renewable energy storage.
Source: University of Chicago.
