Improvement of Multi-lines bridge Defect Classification by Hierarchical Architecture in Artificial Intelligence Automatic Defect Classification

Abstract

Defect classifications are the very important steps as the in-line defect inspection of the semiconductor manufacturing procedure. The precisely identify the defect morphology based on scanning electron microscopy (SEM) images can provide crucial information to find out the root causes of those defects. The conventional defect inspection steps are usually through visual judgement by engineer or technical assistant. However, it's time-consuming and laborious. In our recent study, the Artificial Intelligence Automatic Defect Classification (AI-ADC) performs promising good accuracy and purity of the auto defect classification by deep learning method. Nevertheless, some kind of tiny defects are still difficult to classify by this method, such as multi-lines bridge defect. In this paper, we propose the novel method, called “Hi-erarchical structure AI-ADC”, which join a second binning classifier for more precise defect classification. As a result, the proposed hierarchical AI-ADC method not only can improve the multi-lines bridge defect binning purity from 56% to 88%, but also be applied to classify the similar defect types. Indeed this approach achieves high defect classification performance.