The early in-line detection of defects is a fundamental step in semiconductor manufacturing to ensure the device quality. Inspection techniques currently available can effectively detect large epitaxial defects causing morphological surface variations like stacking faults, while dislocations go undetected. Herein we introduce a new technology with enhanced machine learning analysis, based on contactless and non-destructive room temperature micro-photoluminescence imaging (micro-PL), for the detection and classification of defects in silicon epitaxial layers. With laboratory microscopy techniques we investigate the correspondence between different defect morphologies in micro-PL images and extended crystallographic defects. A good matching in terms of defect density is found between automatic micro-PL analysis and the standard laboratory analysis in an interval spanning from few defects/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> up to 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> defects/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> .
Read full abstract