摘要
针对半导体晶圆生产中存在着位置间变异、晶圆间变异及批次间变异,提出了基于多变异分析(MVA)的工序质量分析方法.该方法通过建立基于方差分析晶圆生产工序质量的多变异分析模型,研究了晶圆生产过程中3种变异与总变异之间的定量关系.由定量关系推导出晶圆生产中的抽样规则,并提出了3种常用过程能力指数Cp,Cpk和Cpm的改进计算方法.通过在晶圆生产中的实际应用,证明了该方法中的抽样规则能捕获加工过程中主要随机变异,该方法计算获得的过程能力指数可较为真实地反映生产过程质量状况.
The process quality of semiconductor manufacturing is influenced by variant factors, such as positional variation, cyclical variation and temporal variation. A new process quality analysis method based on multi-vari analysis is presented. In this method, a model based on multi-vari analysis is set up to analyze the relationship between the total variation and the three kinds of variation. How to make a reliable sampling plan is discussed. And the improved expression of process capability index base on multi-vari relationship is addressed. The process quality analysis method has been tested in a semiconductor manufacturing process. The result shows that the sampling plan can capture the main random variation effectively, and the process quality can be described precisely by the improved process capability indices.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第3期351-355,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(70272046)
关键词
多变异分析
方差分析
过程能力指数
multi-vari analysis
analysis of variance
process capability index
