Title: Scanning Acoustic Microscopy-based Battery Analysis

Inhomogeneities and other defects have proven to decrease the lifespan of battery cells over time. Currently, there is a lack of detection capabilities of such defects, both during end-of-line inspections and during entrance inspections. This not only has the potential to result in reduced performance when defects go unnoticed but could also lead to catastrophic failures during operation in a worst-case scenario.
As a result, it is crucial to develop tools that help to ascertain quality in battery cells with ease, in vast numbers, and at low cost.

Project SAMBA (Scanning Acoustic Microscopy-based Battery Analysis) therefore focuses on the early-on detection of defects and inhomogeneities introduced in production using a scanning acoustic microscopy (SAM) technique. This allows to create image data from the scans of batteries with ultrasound. The image date then serves as training data for an artificial intelligence (AI) model that is meant to detect defects fast and with high certainty. State-of-the-art pouch cells serve as the source for imaging data and are provided with defects in advance.

The AI algorithm which will act as the heart of the detection system takes image data and localizes potential defects in the images. The identified defects or inhomogeneities are then further classified in order to determine their impact on battery performance. The methodology will be highly tailored towards automation.
Furthermore, the cells will undergo a post-mortem analysis to both compare them with the AI results and characterize identified defects for ongoing enhancement of the AI algorithm.

To incorporate industry expertise, both a battery producer and a SAM specialist are involved in the project. This will provide expert views on the batteries and achieve necessary improvements on the SAM technique.