Will AI discover our next-gen battery technology?
In 1956 Ford showcased the Nucleon concept car. It could theoretically drive 8,000km without refueling because the energy was produced onboard via a nuclear reactor. In the 1985 cult movie Back To The Future, Doc Brown developed the flux capacitor, which absorbs waste and produces the energy needed for time travel.
The Nucleon never went into a concept phase, it was only a model on the scale of 1:2, and nuclear-powered mobility has only ever been reserved for rare applications like military submarines and aircraft carriers. But the ambition of Nucleon’s 8,000km range and Doc Brown’s low-emission travel remains. Today’s challenge is to find the battery technologies able to deliver this feat of electric mobility.
Battery-powered electrification is impacting most areas of mobility. The race to discover and produce more energy-dense, lighter, faster charging, and sustainable battery technology, which is economically viable to manufacture at scale, is red hot. The economic factor is critical to widespread adoption. While a micro e-scooter and an electric aircraft use battery technology, they have very different energy demands, technical constraints, and price points.
Newer generation premium electric vehicles are already using batteries capable of accepting higher charging rates. This cuts the charging time significantly but comes at a price. Batteries capable of handling megawatt charging are already in production, although they are aimed at trucks rather than cars. In addition, solutions are emerging to recycle the existing generation of batteries.
Research on batteries continues. Solid-state batteries have been discussed and researched for over a decade, but manufacturing processes for solid-state batteries remain expensive, and real-life performance and production quality remain challenging. Some solid-state batteries will appear around 2025, but a real game-changer will be a Li-metal solid-state battery, which will take an additional decade.
One of the reasons for the slow pace of battery research is that many potential chemicals can be used in various ratios, and optimizing them is labour-intensive and time-consuming. Scaling and optimizing industrial production techniques also takes a long time.
Recently, researchers have started applying AI and robotics to accelerate this process. They have been able to run experiments on possible battery chemical ratios more rapidly using automated labs. They have then fed the results into an AI system to analyze the data and look for patterns to propose alternative ratios that might perform better. The researchers then ran the experiments suggested by the AI, allowing the AI system to optimize the chemical recipes further. The best test cell containing one of the robot-AI-developed electrolytes boasted a 13% improvement compared to the best-performing test cell using the commercially available electrolyte.