Around a billion lithium-ion batteries are produced each year, so any improvements in their efficiency are very welcome.  New research from Imperial College London advocates the use of artificial intelligence to produce more effective designs of the microstructure of fuel cells and lithium-ion batteries.

It’s a process that the team believe could result in faster smartphone charging, as well as longer life for batteries in electric vehicles and even better performance of data centers.

The researchers explain that the performance of both fuel cells and lithium-ion batteries is strongly linked to their microstructure, and especially how the pores inside their electrodes are shaped and arranged.  This affects not only how much power that can be generated, but also how quickly the batteries charge and discharge.

Fiddly business

The micrometer scale of the structures can make studying them at a high enough resolution extremely difficult, much less to successfully test different shapes and sizes for potential performance gains.

The Imperial team believe that machine learning can come to the rescue, as it allows the pores to be examined virtually, with three dimensional simulations undertaken to test potential cell structures.  The team’s algorithm was trained using nano-scale imaging.

“Our technique is helping us zoom right in on batteries and cells to see which properties affect overall performance,” the researchers say. “Developing image-based machine learning techniques like this could unlock new ways of analysing images at this scale.”

Training days

When running the kind of simulations used in the study, it’s important that enough training data is available to statistically represent the whole cell.  Traditionally, this has been a challenge as obtaining the data at the required resolution is very difficult.

By using AI, the team found they could generate large datasets, either with the same properties or with unique structures that might result in better performance of the batteries.

“Our team’s findings will help researchers from the energy community to design and manufacture optimised electrodes for improved cell performance,” the researchers say. “It’s an exciting time for both the energy storage and machine learning communities, so we’re delighted to be exploring the interface of these two disciplines.”