Data Mining is emerging as a new tool to support semiconductor yield improvement efforts. This paper presents a brief survey of data mining methods in use in other industries and discusses the relative advantages of decision tree algorithms for the current context. Enhancements to some of the basic algorithms to adapt them to problems typically found in yield enhancement are discussed. These methods are also compared to more traditional methods based on the automation of regressions and Kruskal-Wallis analyses to find sources of process variation. Analysis of data related to some real world problems are presented. Lastly, we discuss limitations of the current methods and plans for future work in this area.
展开▼
