重复图案晶片自动检测新方法

Novel automatic inspection method for repetitive patterned wafer

针对具有重复图案阵列的IC晶片结构特征,提出了一种基于旋转不变子空间技术(ESPRIT)算法的自比较模板匹配缺陷检测方法。应用ESPRIT算法精确计算出重复图案中结构块的大小,然后将图像中所有结构块对应位置的像素值平均,计算出标准结构块。再根据显微镜视野大小扩展标准结构块形成标准模板,通过比对实现缺陷检测。实验结果表明:采用ESPRIT算法求取结构块大小,具有较高的速度和精度,能够满足IC检测实时性要求。算法的计算复杂度为O(N3/2),精度可达到0.04 pixel。

A defect detection method of self-compared template matching based on Estimating Signal Parameter via Rotational Invariance Techniques（ESPRIT） was presented mainly for inspecting the wafer with the same pattern. The ESPRIT arithmetic was used to calculate precisely the periods of two one-dimensional images and get the size of the building block. The building block could be reconstructed through shifting a proper size window throughout the image and adding the corresponding pixel values. By averaging over all of the blocks in the image, the amount of noise and the effect of defects were reduced considerably. By comparing the original image with the standard model formed by extending standard building block, the point could be identified as a possible defect if the difference was larger than a threshold. The experiment shows that this method has a high-speed and a high-pre- cision to meet the requirement of real-time detection. The complexity of this algorithm is O（N^3/2） and the precision is 0.04 pixel.