If the limits of lithium-ion battery performance are indeed being reached, one way forward is to extend those limits with new materials. Mark Shwartz at Stanford University describes their research into solid electrolytes, which promise to be more energy dense than the standard liquid form. To identify the best compounds, artificial intelligence and machine learning were used rather than the usual and much lengthier trial-and-error experimental method. They settled on four little-known compounds made of lithium, boron and sulphur. Atomic level simulations suggest that, in theory, they could double the driving range of an EV. But simulations are a long way from the real thing. The next step is to synthesise the lithium-boron-sulphur materials and test them in a working battery. This will be a challenge too, but the entire battery storage industry will be watching the results of this kind of research.
Stanford University scientists have identified a new class of solid materials that could replace flammable liquid electrolytes in lithium-ion batteries.
The low-cost materials â made of lithium, boron and sulphur â could improve the safety and performance of electric cars, laptops and other battery-powered devices, according to the scientists. Their findings are published in a study in the journal ACS Applied Materials & Interfaces.
âA typical lithium-ion battery has two solid electrodes with a highly flammable liquid electrolyte in between,â said study lead author Austin Sendek, a visiting scholar in Stanfordâs Department of Materials Science & Engineering. âOur goal is to design stable, low-cost solid electrolytes that also increase the power and energy output of the battery.â
Stability, combustibility
Battery electrolytes shuttle lithium ions between the positive and negative electrode during charging and discharging. Most lithium-ion batteries use a liquid electrolyte that can combust if the battery is punctured or short-circuited. Solid electrolytes, on the other hand, rarely catch fire and are potentially more efficient.
âSolid electrolytes hold promise as safer, longer-lasting and more energy-dense alternatives to liquid electrolytes,â said senior author Evan Reed, an associate professor of materials science and engineering. âHowever, the discovery of suitable materials for use in solid electrolytes remains a significant engineering challenge.â
Most solid electrolytes in use today are too unstable, inefficient and expensive to be commercially viable, the authors said.
âConventional solid electrolytes canât conduct as much ionic current as liquid electrolytes,â Sendek said. âThe few that can usually degrade once they come in contact with the battery electrodes.â
Machine learning
To find reliable solid electrolytes, Sendek and his colleagues in 2016 trained a computer algorithm to screen more than 12,000 lithium-containing compounds in a materials database. Within minutes the algorithm identified approximately 20 promising materials, including four little-known compounds made of lithium, boron and sulphur.
âAs we were looking at the candidates, we noticed that four lithium-boron-sulphur compounds kept popping up,â Sendek said. âUnfortunately, there wasnât much about these materials in the existing scientific literature.â
In the current study, the researchers took a closer look at the four compounds using a technique called density functional theory, which simulates how the materials would behave at the atomic level.

SOURCE: âCombining Superionic Conduction and Favorable Decomposition Products in the Crystalline LithiumâBoronâSulphur System: A New Mechanism for Stabilizing Solid Li-Ion Electrolytesâ / Austin D. Sendek et al.
EV driving range
Lithium-boron-sulphur electrolytes could be about twice as stable as the leading solid electrolytes, the current study shows. Stability can impact the amount of energy per unit weight a battery can store. In electric vehicles, that can mean a longer driving range.
âTeslas and other electric cars can go 250 to 300 miles on a single charge.â Sendek said. âBut with a solid electrolyte you could potentially double the energy density of lithium-ion batteries and get that range above 500 miles â and maybe even start thinking about electric flight.â
When a typical solid electrolyte breaks down, it chemically transforms from a good conductor into a bad conductor, causing the battery to stop working. The study predicted that when mixed together, the four lithium-boron-sulphur compounds would continue functioning even as they decompose.
âAll four compounds are chemically similar,â Sendek said. âSo when the mixture breaks down, each compound will likely transform from a good conductor to another good conductor to another. That means the materials can withstand several cycles of breaking down before they decompose into a bad conductor that ultimately kills your battery.â
Better conductivity, more power
The study also predicted that certain phases of the lithium-boron-sulphur materials could be three times better at conducting lithium ions than state-of-the-art solid electrolytes made with costly germanium.
âIf you get good ionic conductivity you can get more current flow out of your battery,â Sendek said. âMore current means more power to accelerate your car.â

SOURCE: âCombining Superionic Conduction and Favorable Decomposition Products in the Crystalline LithiumâBoronâSulphur System: A New Mechanism for Stabilizing Solid Li-Ion Electrolytesâ / Austin D. Sendek et al.
Some of the best solid electrolytes available today are made with rare elements like germanium, a kilogram of which costs about $500. Lithium, boron and sulphur are abundant chemicals with a price tag of $26 per kilogram.
âOur computer algorithm was searching for new materials based on their physical properties,â Sendek said. âBut it just so happened the four compounds were also much cheaper than the alternatives.â
Ideal for lithium metal batteries
Finding a viable solid electrolyte could also lead to the development of lithium metal batteries â energy-dense, lightweight batteries that are ideal candidates for electric cars.
Most lithium-ion batteries have a negatively charged electrode made of graphite. In lithium metal batteries, graphite is replaced with metallic lithium, which can store significantly more charge per kilogram.
âLithium metal is really the holy grail of battery research,â Sendek said. âBut lithium metal electrodes have a tendency to internally short during operation, which liquid electrolytes do nothing to prevent. Solid electrolytes seem to be our best chance of overcoming that problem, and lithium-boron-sulphur electrolytes are promising candidates.â
Live demonstration still needed
The Stanford study provides a theoretical roadmap for future research. The next step is to synthesise all four lithium-boron-sulphur materials and test them in a battery.
âFrom what my experimentalist friends tell me, making these materials in the lab may be quite difficult,â Sendek said. âOur job as theorists is to point the experimentalists to promising materials and let them see how the materials perform in real devices.â
The ability to identify these promising materials from thousands of candidates was made possible through artificial intelligence and machine learning, Reed added.
âThe discovery of most new materials to date has been accomplished by inefficient trial-and-error searches,â he said. âOur results represent an inspiring success for the machine-learning approach to materials chemistry.â
Other Stanford co-authors of the study are Yi Cui, associate professor of materials science & engineering and of photon science at SLAC National Accelerator Laboratory; and PhD students Evan Antoniuk in the School of Humanities & Sciences and Brandi Ransom in the School of Engineering. Other co-authors are Ekin Cubuk at Google Brain, and Brian Francisco and Josh Buettner-Garret at Solid Power, Inc.
Funding for the study was provided by the Stanford TomKat Center for Sustainable Energy and the Toyota Research Institute Accelerated Materials Design & Discovery program.
***
Mark Shwartz is a Communications Associate, Precourt Institute for Energy at Stanford University
This article is published with permission