中国半导体行业协会封装分会会刊

中国电子学会电子制造与封装技术分会会刊

导航

电子与封装 ›› 2017, Vol. 17 ›› Issue (3): 40 -44. doi: 10.16257/j.cnki.1681-1070.2017.0038

• 微电子制造与可靠性 • 上一篇    下一篇

Overlay chart管控方法的研究

姜颖洁1,2   

  1. 1.上海交通大学,上海 200240;2.中芯国际集成电路制造(上海)有限公司,上海 201203
  • 收稿日期:2017-01-26 出版日期:2017-03-20 发布日期:2017-03-20
  • 作者简介:姜颖洁(1983—),女,上海人,本科,现就职于中芯国际集成电路制造(上海)有限公司,于质量部从事统计过程控制的相关工作。

Study on Inline Overlay Control Method

JIANG Yingjie1,2   

  1. 1.Shanghai Jiaotong University,Shanghai 200240,China; 2.SMIC,Shanghai 201203,China
  • Received:2017-01-26 Online:2017-03-20 Published:2017-03-20

摘要: 在半导体制造光刻工艺中,OVL(Overlay,层对准) 是测量前层与本层之间曝光准确度的参数。现阶段对于OVL的管控,一般根据工程师经验判断设定管控上下限,并没有合理地运用统计方法进行管控。这项研究的目的就是为了寻找合理的统计方法对OVL的管控上下限进行科学有效的设置,从而实现对OVL参数的科学管控。基于长期以来对于OVL数据的特性分析可知,这些数据并不符合最常见的正态分布,因此用3Sigma作为管控上下限的方法并不适合对于OVL的管控。建议可以利用T分布及卡方分布对OVL进行管控,使用这种新的管控方法设置上下限,得到了合理的 OCAP ratio,从而实现对OVL更有效的管控。

关键词: Overlay(OVL), T 分布, 卡方分布

Abstract: In semiconductor manufacture industry, Lithography (Litho) OVL is defined as the positional accuracy, a criteria parameter and to detect possible inline nonconformance from recipe and equipment shifts[1]. Gener-ally, most control limit settings are based on engineering judgment with a tolerance of about 1/3 critical di-mension, and no statistical control assurance is implemented. The goal of this study is to find a statistic method for inline OVL chart control limit setting and thereby set up a reasonable control of OVL chart. It is based on a study of the statistic method “t-distribution & chi-squared distribution”. By studying the Inline OVL data and standard deviation data, the data are offside data which does not follow normal distribution. Control charts of these data are not proper to be controlled by n*sigma, using t-distribution and chi-squared distribution to control OVL chart is strongly recommended here. New control limits are calculated using this method and reasonable OCAP ratio is obtained.

Key words: Overlay (OVL), t-distribution, chi-squared distribution

中图分类号: