Less scrap for the energy transition

Jonas Witt | Siemens Inventors of the Year | Newcomer

A look at battery manufacturing

A man in a black shirt leans over a laptop with

Lithium-ion batteries have an aluminum cathode (- pole) and a copper anode (+ pole). During manufacturing, thin layers of active materials – e.g. lithium cobalt oxide or lithium iron phosphate – are applied to long copper or aluminum foil strips. These long electrode strips are cut into small segments, combined, and wound around or stacked to individual battery cells. “The coating of the metal foils is a critical process step. The active layers need to be of uniform thickness across the entire width of the foil,” Jonas says. “Errors here are the main cause of bad battery cells. The two inventions that are being recognized are helping us better understand and improve this production step.”

A pattern like a fingerprint

The more precisely the cause of an error can be traced, the easier it is to develop solutions – and this principle also applies to battery manufacturing. Jonas and his team looked for a way to establish a link between a specific battery cell and the sections of copper / aluminum foil strips from which they were produced. “In the past, production data was assigned to long strips of material,” Jonas explains. “Our new procedure is more precise and non-invasive. We’ve discovered that the surface pattern of the material is unique for individual coating segments, similar to a fingerprint. Any differences in the patterns of the different segments are so clear-cut that for a specific battery cell, we can subsequently track the exact foil segments from which they were manufactured, when they were manufactured, and the settings that were used. With this knowledge, we’ve not only made the old markings obsolete, we’re also able to analyze when and why so much or so little waste was produced more precisely than ever before.”

Hands use scissors to cut a dark, flexible material with a red laser line on it, for battery component fabrication.

Learning from the experts

A man with a serious expression is standing in front of a white wall, wearing a black shirt and a black jacket.

The precise calibration of production machines is a key factor. “Machine operators generally have to undergo several months of training to develop a proper feel for optimally setting the process parameters,” Jonas says. “Large amounts of scrap are often produced because they’ve overlooked something.” That’s why Jonas and his team have trained an AI model with the inspection behavior of an highly skilled machine operator in order to develop a system that provides recommendations for optimal configuration during ongoing operation. “Based on the operational data, a well-trained AI model is much faster at recognizing when we deviate too far from the optimum,” Jonas affirms. “Our goal for the near future is to be able to control the coating process fully automatically so that we can reduce the scrap rate for battery cells by around ten percent. We’ve already obtained commercial licenses.”