AI targets 21 new solid electrolytes for less flammable batteries

December 19, 2016 // By Nick Flaherty
Researchers at Stanford University have used artificial intelligence to identify over 20 solid electrolytes that could replace the volatile liquids currently used in batteries.

"Electrolytes shuttle lithium ions back and forth between the battery's positive and negative electrodes," said Austin Sendek, a doctoral candidate in applied physics. "Liquid electrolytes are cheap and conduct ions really well, but they can catch fire if the battery overheats or is short-circuited by puncturing. The main advantage of solid electrolytes is stability. Solids are far less likely to blow up or vaporize than organic solvents. They're also much more rigid and would make the battery structurally stronger."

Despite years of laboratory trial and error, researchers have yet to find an inexpensive solid material that performs as well as liquid electrolytes at room temperature. So the team used AI and machine learning to build predictive models from experimental data. They trained a computer algorithm to learn how to identify good and bad compounds based on existing data, much like a facial-recognition algorithm learns to identify faces after seeing several examples.

"The number of known lithium-containing compounds is in the tens of thousands, the vast majority of which are untested," Sendek said. "Some of them may be excellent conductors. We developed a computational model that learns from the limited data we already have, and then allows us to screen potential candidates from a massive database of materials about a million times faster than current screening methods."

To build the model, Sendek spent more than two years gathering all known scientific data about solid compounds containing lithium.

"Austin collected all of humanity's wisdom about these materials, and many of the measurements and experimental data going back decades," said Evan Reed, an assistant professor of materials science and engineering. "He used that knowledge to create a model that can predict whether a material will be a good electrolyte. This approach enables screening of the full spectrum of candidate materials to identify the most promising materials for further study."

The model used several criteria to screen promising