摘要
针对90 nm工艺、300 mm直径大尺寸硅片对制造设备的新要求,提出一种硅片边缘信息采集、处理方法,并在此基础上建立硅片预对准机器人视觉系统。该方法在小波阈值折衷去噪的基础上,利用最小二乘回归拟合和坐标加权平均的数值方法,通过对硅片边缘特征点的标定、过滤、综合和提取,有效地实现CCD采集信号向特征位置数据的转换,解决大尺寸硅片位置信息的获取和分析问题。重复定位精度测量试验结果表明,硅片预对准机器人视觉系统完成了硅片中心、缺口的检测定位,硅片预对准机器人对准精度达到了技术指标的要求,提高了整套硅片预对准系统的工作性能。
In view of the new requirements of manufacturing equipment for large-size wafer of 300 mm diameter and 90 nm process, a method of edge information collecting and processing is proposed, based on which the vision system of pre-alignment robot is established. By the help of de-noising using compromise of wavelet coefficient, the method which synthesizes least square regression and weighted means, can successfully realize the conversion from CCD signal to position data through calibration, filtering, analysis and synthesis of light information, and the problem of wafer edge detection and location is resolved. The result measurement test of repetitive positioning accuracy shows that the robot can carry out the detection and alignment of wafer pinpoint, attain the precision requirement and promote the working performance of whole pre-alignrnent system.
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2008年第10期210-215,共6页
Journal of Mechanical Engineering
基金
国家自然科学基金(50675027)
大连市科技计划(2004A1GX086)资助项目。
关键词
预对准
视觉系统
特征点标定
数值处理
位置修正
Pre-alignment Vision system Calibration of pinpoints Data processing Position correction
