Machine learning approach opens insights into an entire class of materials being pursued for solid-state batteries

The atomic processes that make a family of chemicals known as argyrodites appealing candidates for both solid-state battery electrolytes and thermoelectric energy converters have been discovered by a research team at Duke University and its partners.

The findings, along with the machine learning strategy that led to them, may herald in a new era of energy storage for uses like home battery walls and quick-charging electric vehicles.

The findings were published online in the journal Nature Materials on May 18.

Given the size and complexity of each component of the material, Olivier Delaire, an associate professor of mechanical engineering and materials science at Duke, remarked, "This is a puzzle that has not been solved before." We've uncovered the atomic-level mechanics behind this class of materials' current popularity in the field of solid-state battery advancement.

Researchers must create new technologies for energy storage and distribution to households and electric cars as the world moves toward an energy future based on renewable sources. The lithium-ion battery with liquid electrolytes has been the industry standard up to this point, although it is far from the best option due to its low efficiency and propensity for spontaneous combustion and explosion.

These restrictions are mostly caused by the chemically reactive liquid electrolytes used within Li-ion batteries, which allow lithium ions to travel between electrodes relatively unhindered. The liquid component makes them vulnerable to high temperatures, which can lead to degeneration and, finally, a runaway thermal disaster, despite the fact that they are excellent for transporting electric charges.

A lot of effort and money is being invested by both public and commercial research institutes in the development of alternative solid-state batteries made of various materials. In principle, this strategy produces a device that is considerably safer, more stable, and has a larger energy density if it is properly constructed.

While no economically feasible method for solid-state batteries has yet been found, one of the top candidates is based on a group of substances known as argyrodites, which take their name from a mineral that contains silver. These compounds are composed of two distinct, stable crystalline frameworks with room for a third element to move freely inside the chemical structure. While some ingredients, like silver, germanium, and sulfur, are found in nature, the overall structure is open-ended enough for scientists to come up with a broad range of combinations.

Every manufacturer of electric vehicles is attempting to switch to new solid-state battery designs, but none of them have revealed which compositions they are placing their bets on, according to Delaire. "Winning that race would be a game-changer because cars could charge faster, last longer, and be safer all at once."

In the current article, Delaire and his coworkers examine a promising possibility called Ag8SnSe6, which is formed of silver, tin, and selenium. The molecular behavior of samples of Ag8SnSe6 was revealed in real-time by the researchers using a combination of neutrons and X-rays to bounce these incredibly fast-moving particles off atoms within the samples. Using first-principles quantum mechanical simulations, team member Mayanak Gupta, a former postdoc in Delaire's lab who is now a researcher at the Bhabha Atomic Research Center in India, developed a computational model to match the observations. She also developed a machine learning strategy to make sense of the data.

The outcomes demonstrated that although the scaffolding made by the tin and selenium atoms was rather stable, it wasn't static. In order to allow the charged silver ions to freely travel through the substance, the crystalline structure continually bends to form windows and channels. Delaire described the system as having tin and selenium lattices that are still solid and silver that is practically liquid-like.

Delaire described the movement of the silver atoms as resembling marbles rattling about the bottom of a very shallow well, as if the crystalline scaffold weren't a solid structure. What really surprised me was that a substance could exist in both a liquid and a solid form.

The findings and, perhaps more crucially, the method of replacing lithium-ion batteries in many essential applications by combining superior experimental spectroscopy with machine learning, should speed up research in this area. Delaire claims that this study is only one in a series of initiatives aiming at various potential argyrodite compounds made up of various recipes. Due to its potential for use in EV batteries, one combination that uses lithium instead of silver is of special interest to the group.

We are carefully examining the complete family of compounds since many of these materials provide extremely quick conduction for batteries while also serving as effective heat insulators for thermoelectric converters, according to Delaire. "This work serves as a benchmark for our machine learning methodology, which has greatly improved our capacity to mimic these materials in just a few years. As a result, I think we'll be able to swiftly recreate new chemicals realistically and discover the greatest recipes they have to offer.

Provided by Duke University