Materials Data Science and Informatics (IAS-9)

Detecting and analyzing defects in semiconductor materials is essential for understanding underlying mechanisms and optimizing production workflows. Traditionally, interpreting microscopy images—requiring complex tasks like segmentation and object detection—has been slow, labor-intensive, and limited by human capacity.

Our solution leverages advanced Machine Learning and Deep Learning models to transform this process. By automating the analysis of high-resolution microscopy data, we enable fast, consistent, and scalable detection of every defect across KOH-etched 4H-SiC wafers.

Through an intelligent pipeline that unites state-of-the-art image analysis, data mining, and AI-driven pattern recognition, we deliver robust, high-accuracy identification of defect types and positions—empowering researchers and manufacturers to unlock deeper material insights and accelerate innovation.

Related Publications:

• B. D. Nguyen, J. Steiner, P. Wellmann, S. Sandfeld, Combining unsupervised and supervised learning in microscopy enables defect analysis of a full 4H-SiC wafer, MRS communications 14, 612-627 (2024) [10.1557/s43579-024-00563-2]