硅通孔键合硅片预对准边缘信息采集与处理

Edge Imformation Collecting and Processing of TSV Wafer Pre-alignment

摘要:

研究硅通孔即TSV(through-silicon vias)键合硅片的预对准边缘信息采集与处理方法。TSV硅片与标准硅片相比,有减薄、键合不同心、边缘毛刺多、存在崩边；缺口被填充、内有鼓胶、镀铜等工艺特点,使得传统基于线阵CCD一维图像采集与处理预对准方法失败。针对TSV硅片的特点,把线阵CCD配合扫描运动采集的一维原始图像集拼接获得二维图像,应用二维图像处理技术提取边缘信息,硅片整周边缘数据用最小二乘圆拟合算法识别出圆心位置,缺口边缘数据用Hough直线变换识别出缺口两条斜边,其交点定位为缺口位置,从而实现TSV硅片的自动预对准。实际测量表明,该方法预对准重复性定位精度<20um、预对准时间<40s,满足指标需求,为光刻机能够曝光TSV硅片提供有力支持。

Abstract:

This study was focused on edge imformation collecting and processing of TSV wafer pre-alignment. Compared to standard wafers TSV wafers have features of being thinned or deflected when bonding, with burrs and flaws around the wafer edge, notch with remnant glues and copper in it being covered, which results in failure in traditional pre-alignment based on 1-dimensional image collecting and processing sampled by line CCD. Aiming to TSV wafers, the 1-demensional images sampled by line CCD were put together to form a 2-demensional image, working in with scanning motion. The wafer edge information was collected via 2-demensional image processing technology, the center position was recognized via least square round fitting method, two notch hypotenuses whose jointing point was regarded as notch position were recognized via Hough line transform, thus an automatic TSV wafer pre-aligning method was achieved. The method has been proved its precision and efficiency are up to the standard by practical test, with the repeatability less than 20u m and consuming time less than 40s, providing strong support for lithography exposing TSV wafers.