New architecture and systems for wafer inspection are developed.



The demonstration presents a fully automated measurement toolchain. Excellent measurement quality and stability are required to be to analyse the images. Research focuses on computer vision tasks and additionally on methods for automated analysis techniques of semiconductor front-end technologies. The demonstrator provides automation that is necessary for image measurement and technology identification. The segmentation will enable a fully automated image measurement and technology verification.

Beyond state-of-the-art developments and impacts

The high degree of automation and the ability of data-intensive approaches open new possibilities that were not available before. The developments within this supply chain open new innovative ways for semiconductor front-end process improvement and a new method for counterfeit detection. The deep learning (DL) component aims to provide fast, reliable, and accurate segmentation based on grey-scale SEM images. The lack of data to be used for training can find similarities between the given task and other DL applications in medical image segmentation, computed tomography images, magnetic resonance images, etc. The necessity of reliable and resilient segmentations draws further parallels to applications in the field of autonomous driving.