Article

A team from IKST and RISE in India has developed a deep-learning tool that predicts the operating voltage of sodium-ion battery materials, potentially advancing their commercial use. While lithium-ion batteries dominate, sodium is abundant and cheaper.

The researchers trained their AI model on over 4,300 materials, achieving a mean absolute error of 0.24 volts on new compounds. The model was used to propose new cathode materials, with voltages closely matching theoretical and experimental data. The study highlights a workflow that could reduce the time and costs of identifying viable sodium-based cathodes, promoting their use in grid storage and electric mobility.