INFORMATION ENGINEERING LABTORY

•Performance Chamber Matching By Fault Detection and Classification

Proposed method

•Fifth step of the process of seal system is the plating. Because during the different tools has the difference, we hoped that causes tools' different reduction because of the adjusting key variables. So , we need to find the key variables to control the height after plating.

• Run to run control for mixed run process and Fault Detection and Classification in semiconductor industry
• mixed run process in semiconductor industry

System and problem

• Mix run process in the semiconductor industry

– 70% of products were ran in less than 10 times

– These infrequent products account for ~25% of runs

• How to improve the performance of products in a mix run process?
• How to improve the performance of low frequency products?

Proposed method

• An ANOVA model with ARIMA disturbance approach

• EWMA control base on adaptive k-means cluster

Result

• Cluster result

Result

•By linearfit P value vs. IQR P value , we f ind key variables (E1 , E2 , E3 , E4) .

•By m ultiple compare , we can find that raising the key variable will enhance y data. The setting value of key variable must be very close. We can reduce the difference between various tools by adjusting the key variable.

• Control result