Scholarlink

<p>The growing demand for electronic devices and the accelerated development of the semiconductor industry have imposed stricter requirements for high-quality integrated circuits. As the fundamental substrate in chip manufacturing, wafers play a critical role in determining circuit quality, making defect detection essential. As a result, vision-based wafer defect inspection has received significant research interest in recent years. Although several reviews have summarized developments and trends in this field, most overlook critical aspects, including defect acquisition equipment, systematic inspection methodologies, and recent detection advances. To address these gaps, this survey provides a systematic review centered on three core aspects: data acquisition and processing, vision-based defect detection algorithms, and evaluation metrics and loss functions. It begins by introducing essential background on wafer and integrated circuit fabrication, defect inspection systems, related datasets, and preprocessing methods. The paper then comprehensively examines the evolution of detection methodologies over the past two decades, spanning traditional image processing, conventional machine learning, and modern deep learning approaches. Finally, the study discusses current limitations and future trends across data, models, and applications, offering insights to enhance the efficiency, accuracy, and applicability of wafer defect inspection. This work aims to facilitate advancements in wafer inspection technology and support quality improvement in semiconductor manufacturing.</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

Review of wafer defect detection in semiconductor manufacturing: Algorithms, systems, and data

  • 期刊论文

  • 发表日期：

    2026年4月21日

查看全文
  • Jian Li,
  • Xiangyan Zhang,
  • Xuexiu Liang,
  • Gefan Yin,
  • Lutao Yan,
  • Shimin Wei

摘要

The growing demand for electronic devices and the accelerated development of the semiconductor industry have imposed stricter requirements for high-quality integrated circuits. As the fundamental substrate in chip manufacturing, wafers play a critical role in determining circuit quality, making defect detection essential. As a result, vision-based wafer defect inspection has received significant research interest in recent years. Although several reviews have summarized developments and trends in this field, most overlook critical aspects, including defect acquisition equipment, systematic inspection methodologies, and recent detection advances. To address these gaps, this survey provides a systematic review centered on three core aspects: data acquisition and processing, vision-based defect detection algorithms, and evaluation metrics and loss functions. It begins by introducing essential background on wafer and integrated circuit fabrication, defect inspection systems, related datasets, and preprocessing methods. The paper then comprehensively examines the evolution of detection methodologies over the past two decades, spanning traditional image processing, conventional machine learning, and modern deep learning approaches. Finally, the study discusses current limitations and future trends across data, models, and applications, offering insights to enhance the efficiency, accuracy, and applicability of wafer defect inspection. This work aims to facilitate advancements in wafer inspection technology and support quality improvement in semiconductor manufacturing.