Volume 60, Issue 3

In Lithium-ion batteries, the graphite anode is known to undergo a noticeable chromatic change during lithiation and de-lithiation by forming graphite intercalation compounds. Additionally, the graphite anode primarily contributes to the volume change of the battery. Using a novel in-situ optical microscopy setup for imaging cross-sections of Li-ion full cells, both effects can be studied simultaneously during charging and discharging. In this work, we describe feature extraction methods to quantify these effects in the image data (3730 images in total) captured during the lithiation and de-lithiation process. Automated and manual evaluations are compared. The images show graphite anodes and NMC 622 cathodes. For colorfulness, we evaluate different methods based on classical image processing. The metrics calculated with these approaches are compared to the results of ColorNet, which is a trainable colorfulness estimator based on deep convolutional neural networks. We propose a supervised semantic segmentation approach using U-Net for the layer thickness measurement and the anode dilation derived from it.

Modern day rotating X-ray anodes utilize a conversion layer comprised of a tungstenrhenium alloy. The thermomechanical loading of this layer during computed tomography operation introduces various signs of fatigue like cracking, roughening, melting, or microstructural changes. Previous work on preparing tungsten samples primarily considered intact thin films or bulk material. This work focuses on the metallographic preparation of the conversion layer surface, which represents the sample edge in polished cross-sectional cuts. The main goals were minimizing preparation artefacts and maximizing obtainable image quality during electron backscatter diffraction. Twelve preparation methods were compared with regard to edge rounding, chipping, and obtainable image quality. Coating the samples with a thin layer of molybdenum and adding a tungsten sheet for edge stabilization led to vastly improved results. Chemical-mechanical polishing of such a sample gave the most balanced set of considered benchmarks.

The damage case dealt with in the present work relates to a generally less common manufacturing defect that was discovered during its installation. The part was sourced from a supplier. An installation of the defective part in a large gas turbine engine used for energy generation could thus be prevented. The defect, a solidification crack, could have been repaired by welding and the component could have been used for operation. Instead, it was subjected to destructive examinations aiming at identifying what caused the crack. As the component could not possibly perform its function prior to being repaired by welding we are, in any event, talking about a damage case. The solidification crack initiated at a site of oxide infiltration. It was thus clearly a casting defect. The respective supplier was advised to reduce the content of non-metallic inclusions by the use of ceramic filters when the material is poured into the shell mold and a better control of the vacuum in the casting furnace.