Scholarlink

<p>As semiconductor manufacturing technology has been rapidly advanced, conventional approaches cannot classify new wafer defect patterns without retraining. To overcome this, we propose an image matching-based pattern search method to analyze the similarity between the wafer defect patterns using new pre-processing and similarity metrics. The proposed search method finds the correlation of wafer defect patterns to determine the similarity value between wafer bin maps&#xa0;(WBMs). The pre-processing performs the denoising to reduce the effects of non-significant defect patterns. Besides, we propose two metrics for performing robust pattern searches based on the shape, location, and area of defect patterns. Experimental results show that the proposed method is effective on industrial-driven datasets WM-811K and MixedWM38, having significant benefits of the pre-processing and similarity metrics in the search for various defect patterns.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Recognition and ranking using similarity on defective wafer bin maps

  • 期刊论文

  • 发表日期：

    2025年5月10日

查看全文
  • YoungWook Kwon,
  • SuMin Oh,
  • HyunJin Kim

摘要

As semiconductor manufacturing technology has been rapidly advanced, conventional approaches cannot classify new wafer defect patterns without retraining. To overcome this, we propose an image matching-based pattern search method to analyze the similarity between the wafer defect patterns using new pre-processing and similarity metrics. The proposed search method finds the correlation of wafer defect patterns to determine the similarity value between wafer bin maps (WBMs). The pre-processing performs the denoising to reduce the effects of non-significant defect patterns. Besides, we propose two metrics for performing robust pattern searches based on the shape, location, and area of defect patterns. Experimental results show that the proposed method is effective on industrial-driven datasets WM-811K and MixedWM38, having significant benefits of the pre-processing and similarity metrics in the search for various defect patterns.