Researchers from the Indo‑Korea Science and Technology Center (IKST), Bangalore, and the Research Institute for Sustainable Energy (RISE) under TCG‑CREST, Calcutta, have developed a deep‑learning model that accurately predicts the operating voltage of battery materials. The work, led by Prof. G.P. Das and Prof. S. Bhattacharjee, has been published in the journal Small Methods.
While lithium‑ion batteries dominate current technology, lithium is scarce and costly. Sodium, by contrast, is abundant and inexpensive, but identifying stable, high‑voltage sodium‑based cathode materials has been slow and expensive. To address this, the team trained a deep neural network on data from over 4,300 battery materials, enabling rapid voltage predictions without time‑consuming quantum‑mechanical calculations.
The model achieved a mean absolute error of 0.24 volts and was further used to design new layered oxide sodium cathode materials (NaMO₂) with promising performance and stability. For selected candidates, AI predictions closely matched both density‑functional‑theory calculations and existing experimental data.
By combining machine learning with selective theoretical validation, the study offers a fast and cost‑effective workflow for discovering new cathode materials. The researchers believe this approach will accelerate the commercialization of sodium‑ion batteries, supporting large‑scale applications in grid storage, electric mobility, and renewable energy integration.
You may read the full article here:
https://www.tcgcrest.org/wp-content/uploads/2026/03/Eurasia-Review-Newsitem-11Mar26.pdf