AI Rapidly Created a Low-Lithium Battery with a Novel Material

In 80 hours, Microsoft’s AI tool narrowed 32 million theoretical materials to 18, leading scientists to synthesize one that can decrease lithium usage in batteries by 70%.

An AI program has pinpointed a synthetic material, absent in nature, that can potentially decrease lithium usage in batteries by as much as 70%.

Out of 32 million candidates, the optimal choice was identified as a mixed metal chloride consisting of sodium, lithium, yttrium, and chloride ions.

However, rechargeable batteries primarily rely on lithium as the vital component, and the demand for this metal has surged in recent years. Unfortunately, the energy-intensive mining process for lithium often results in enduring water and land pollution. Consequently, numerous companies are actively seeking alternative materials for battery construction.

To accomplish this, Microsoft and the Pacific Northwest National Laboratory (PNNL) collaborated to develop a screening tool for potential new materials that could be used in low-lithium batteries using Microsoft’s Azure Quantum Elements tool. The researchers published their results on January 8 via the pre-print server arXiv.

Building a new type of battery

Batteries transport charged particles between positive and negative terminals, called electrodes. When wires connect lithium ions from the negative electrode to the electrolyte, reaching the positive electrode. Simultaneously, electrons move through the wires in the same direction, facilitating energy extraction from the battery.

In this investigation, the scientists concentrated on solid electrolyte materials, aiming to create a safer and more effective substitute for the existing liquid electrolytes. The crucial requirement is that the electrolyte material should align with the electrodes, permitting smooth passage of lithium ions while entirely hindering the movement of electrons within the battery.

Narrowing Down 32 Million Candidates for Novel Battery Material

The researchers initiated the process with over 32 million potential candidates, generated by swapping various elements into existing electrolyte structures. They employed a combination of AI techniques to filter the materials, considering their properties in the selection process.

In fact, Kandler Smith, a mechanical engineer from the National Renewable Energy Laboratory, explained to Live Science that a substantial portion of the candidate materials, generated through theoretical computer calculations, is typically not stable enough for practical laboratory synthesis. He mentioned that their initial emphasis was on filtering for stability, and this preliminary screening swiftly reduced the pool from 32 million to half a million materials within hours.

AI-driven Criteria Refine Material Choices for Accelerated Research

The team then selected nine additional standards. It employed AI to systematically apply them, arranging the candidates based on their electronic characteristics, cost, and strength. This process narrowed the selection to 18 finalists. Smith expressed his admiration, stating, “I was very impressed that they accomplished all this in just 80 computer hours — experimenting with all those materials would have taken 20 years. Their machine learning approach, paired with physics-based molecular dynamics models, is a significant advancement and will greatly accelerate research.“

The scientists produced a set of final materials comprising lithium, sodium, the rare earth element yttrium, and chloride ions in different ratios. Notably, the combination of lithium and sodium in this material enables the conduction of both ion types, a previously considered impossible phenomenon. This material could potentially function in sodium-ion batteries as well. Noteworthy is a high-sodium variant that contained 70% less lithium than a standard battery, suggesting a significant potential for reducing the cost and environmental impact of these batteries in the future.

A starting point for AI-powered material discovery

The team subsequently assessed the electronic properties of the candidates. Smith explained that the key property of an electrolyte is ionic conductivity, which is how fast the lithium ions can move. He noted that this factor determines the charging speed of the battery, and it is crucial for electric vehicles.

Regular lithium-ion batteries utilize a liquid organic solvent electrolyte, enabling fast ion movement and charging. However, these solvents are flammable, and interactions with the electrodes cause battery degradation over time. According to Smith, solid-state electrolytes have the advantage of being more chemically stable and much less flammable. The downside is that they move the lithium ions slowly, resulting in slower charging times.

Breakthrough Discovery

The AI pinpointed the top-performing candidate, an order of magnitude less conductive than current liquid electrolytes. This translates to a significant difference in charge time, from 30 minutes to five hours. Therefore, the material’s electronic performance improvements are necessary before it becomes practical for applications. Nevertheless, Microsoft representatives mentioned in an email to Live Science that the researchers successfully constructed a working prototype from the final material and used it to power a lightbulb.

Smith views it as a solid initial step. He emphasized that the work’s most significant accomplishment was streamlining material discovery through AI. He explained that the same machine learning pipeline could be beneficial in supporting research across hundreds of other related areas.

According to Brian Abrahamson, PNNL’s chief digital officer, both Microsoft and PNNL show a keen interest in exploring this further in the future. Abrahamson stated that the new battery results are just one example — a proof point. He mentioned that they recognized early on that the key lies in the speed of AI assisting in identifying promising materials and their ability to implement those ideas in the laboratory promptly. He added that they plan to push the boundaries of what’s possible by fusing cutting-edge technology and scientific expertise.

Read the orinal article on: livescience

Read more on Scitke: Mars-Bound Astronauts Will Undergo Time Dilation Effects