Optimized quality and process control through AI-based image classification at Fraunhofer IPMS

Inhalt:
Optical defect inspection and classification is an important part of semiconductor manufacturing. They provide relevant information to correct process problems and thus improve product yield and quality. The highly automated inspection and microscopy systems generate an enormous amount of data. AI-based methods are an appropriate way to ensure 100% classification coverage, to improve monitoring accuracy and to support anomaly detection. Initial experience with the use of an AI-based system for optical image classification was gained in the cleanroom environment of Fraunhofer IPMS in Dresden.