基于机器视觉的晶圆表面缺陷检测方法研究

张金宇1; 朱立军1; 2; 林紫强1

基于机器视觉的晶圆表面缺陷检测方法研究

引用本文：张金宇1,朱立军1,2,林紫强1.基于机器视觉的晶圆表面缺陷检测方法研究[J].计算技术与自动化,2025,(3):94-99

摘要点击次数: 444

全文下载次数: 216

作者	单位
张金宇1,朱立军1,2,林紫强1	（1. 沈阳化工大学计算机科学与技术学院,辽宁沈阳 110142；2. 辽宁省化工过程工业智能化技术重点实验室,辽宁沈阳 110142）

中文摘要:晶圆作为半导体制造业中重要的原材料，其表面缺陷会影响其制成芯片的质量。为了确保芯片的质量和提高生产效率，对晶圆表面的缺陷进行高效、准确、实时的检测具有必要性。本文针对有图案的晶圆表面提出了一种缺陷检测方案，首先，获取晶圆图像并对其做预处理操作，对图像进行光照均衡使其灰度均匀性得到提升；其次，根据晶圆图像纹理的规律性和周期性，设计了图像熵信息与灰度模板匹配相结合的方法提高匹配的准确性，划分出每个小单元格；然后，将划分的单元与模板单元进行差分，使用傅里叶变换设计滤波器去除图像中光学系统畸变影响；最后，通过计算图像离散程度，根据不同区域离散度与全图离散度对比进行缺陷增强。最终实验表明，对于缺陷检测准确率可达93.5%，该方法可以有效地应用于有图案晶圆表面缺陷的检测。

中文关键词:自动光学检测熵灰度模板匹配傅里叶变换伽马变换

Research on Wafer Surface Defect Detection Method Based on Machine Vision

Abstract:Wafer, as an important raw material in the semiconductor manufacturing industry, its surface defects can affect the quality of chips produced. In order to ensure the quality of chips and improve production efficiency, it is necessary to conduct efficient, accurate, and real-time detection of surface defects on wafers. This article proposes a surface defect detection scheme for patterned wafers. Firstly, the wafer image is obtained and preprocessed, and the image is illuminated to improve its grayscale uniformity; Secondly, based on the regularity and periodicity of wafer image texture, a method combining image entropy information with grayscale template matching was designed to improve the accuracy of matching, and each small cell was divided; Then, the divided units are differentiated from the template units, and Fourier transform is used to design filters to remove the optical system distortion effects in the image; Finally, by calculating the degree of image dispersion, defect enhancement is performed by comparing the dispersion of different regions with that of the entire image. The final experiment shows that the accuracy of defect detection can reach 93.5%, and this method can be effectively applied to the detection of surface defects on patterned wafers.

keywords:automatic optical inspection entropy grayscale template matching Fourier transform Gamma transform

查看全文 查看/发表评论 下载pdf阅读器