Machine learning approach opens insight into entire class of materials being investigated for solid-state batteries

A team of researchers at Duke University and their collaborators have elucidated the atomic mechanisms that make a group of compounds called algyridites attractive candidates for both solid-state battery electrolytes and thermoelectric energy converters.

This discovery, and the machine learning approach used in it, could usher in a new era of energy storage for applications such as home batteries and fast charging of electric vehicles.

The results were published in an online journal on May 18 Natural materials.

“This is a previously unsolvable puzzle because each component of the material is so large and complex,” said Olivier Delaire, an associate professor of mechanical engineering and materials science at Duke University. “We have elucidated at the atomic level the mechanisms that make this whole class of materials a hot topic in the field of all-solid-state battery innovation.”

As the world moves toward a renewable energy-based future, researchers must develop new technologies for storing and distributing energy to homes and electric vehicles. The standard bearer to date has been lithium-ion batteries containing liquid electrolytes, but given their relatively low efficiency and the potential for liquid electrolytes to ignite and explode in some cases, they are an ideal solution. far from

These limitations are primarily due to the chemically reactive liquid electrolyte in lithium-ion batteries that allows lithium ions to move relatively freely between the electrodes. Although they are excellent at transporting electrical charges, their liquid components make them sensitive to high temperatures, which can lead to degradation and ultimately runaway thermal hazards.

Many public and private research institutes are devoting significant time and money to developing alternative solid-state batteries from a variety of materials. If this approach is designed correctly, at least in theory, it will provide devices with higher energy densities, safer and more stable.

Although no one has yet found a commercially viable approach to all-solid-state batteries, one of the leading candidates relies on a class of compounds called argyrodites, named after minerals containing silver. I’m here. These compounds are composed of two elements and are built from specific stable crystalline frameworks that allow the third element to move freely within the chemical structure. Some recipes for silver, germanium, and sulfur occur naturally, but the general framework is flexible enough to allow researchers to create a wide range of combinations.

“Even though all electric vehicle makers are moving to new solid-state battery designs, none have disclosed what composition they’re betting on,” Delea said. “Winning that race would make all the difference as cars charge faster, last longer and are safer at the same time.”

In a new paper, Delaire and his colleagues consider one potential candidate made of silver, tin, and selenium (Ag).₈SNSe₆). The researchers combined neutrons with her X-rays to bounce these extremely fast-moving particles off the atoms within the silver sample.₈SNSe₆ Reveal molecular behavior in real time. Team member Mayanak Gupta, a former postdoctoral fellow in Delea’s lab and now a researcher at the Baba Atomic Research Center in India, has also developed machine learning approaches to understand data and used first principles. Then, we created a computational model that agrees with the observed results. Quantum mechanical simulation.

The results showed that the tin and selenium atoms created a relatively stable scaffold, but it was far from static. The crystal structure is constantly flexing, creating windows and channels that allow charged silver ions to move freely through the material. Delea said the system is such that the silver is in an almost liquid-like state, while the tin and selenium lattice remains solid.

“It’s like the silver atoms are like rocking marbles at the bottom of a very shallow well, moving around like the crystal doesn’t have a solid footing,” Derail said. “I found the duality of matter, which exists between both liquid and solid states, the most amazing thing.”

This result, and perhaps more importantly, an approach that combines advanced experimental spectroscopy and machine learning will help researchers make faster progress towards replacing lithium-ion batteries in many important applications. must. According to Delaire, the study is just one in a series of projects targeting a variety of promising argyrodite compounds of different recipes. One combination of replacing silver with lithium is of particular interest to this group given its potential as an EV battery.

“Many of these materials provide very fast conductivity for batteries, while also being excellent insulators for thermoelectric converters. said Deller. “This work serves as a benchmark for our machine learning approach, which has enabled us to make significant progress in our ability to simulate these materials in just a few years, allowing us to rapidly simulate new compounds virtually. , we believe that we will be able to find the optimal recipes for these compounds’ needs.” ”