Argonne National Laboratory

Researchers at the U.S. Argonne National Laboratory have applied a combination of machine learning and artificial intelligence to help narrow down a list of 166 billion molecules that could be used to form the basis of a battery electrolyte. The technique, say the researchers, offers a way to greatly reduce the cost of narrowing down such an enormous data set, while still providing a precise understanding of each molecule and its likely suitability.

Scientists at the United States’ Argonne National Laboratory worked with a class of electrolyte materials they say could greatly improve the performance of lithium-sulfur batteries. The group has devised a selection rule which it says will help researchers select the most suitable electrolyte materials for different battery systems.

Researchers from the U.S. Department of Energy’s Argonne National Laboratory, working with Cambridge University, programmed a ‘supercomputer’ to narrow down a list of almost 10,000 materials with the potential to be used in dye-sensitized solar cells to just five that fit their parameters for high performance, low cost and low environmental impact.

Researchers were able to effect electronic transitions at different frequencies, and through a different physical process, by altering the angle at which nanolayers of gold were positioned. Being able to predict and fine-tune the nature of such transitions could have a fundamental affect on solar PV cell efficiency.