Combined product and tool disturbance estimator for the mix-product process and its application to the removal rate estimation in CMP process

An-Chen Lee*, Tzu Wei Kuo, Chung Ting Ma

*Corresponding author for this work

Research output: Contribution to journalArticle

8 Scopus citations

Abstract

In this paper, we proposed a control strategy: Combined Product and Tool Disturbance Estimator (CPTDE) which combines threaded double EWMA with the drift compensation scheme, to adaptively estimate the disturbance for a mix-product situation in semiconductor processes. This approach considers the disturbances are related to the combination of the specific product and the tool, and further separates the error into an intercept term and a drift term, where the former is related to the variation of products, whereas the latter is related to the interaction between the tools and the products. The proposed method continuously updates the intercept and drift terms to obtain the recipe for the next run. The simulation case studies, i.e., the fixed schedule process, the random schedule process, and the periodical schedule process, are conducted and the results show that CPTDE control scheme has best control performance when compared with three recently published control schemes. The method is also applied to the estimation of removal rate in mix-product CMP process. The results show that the proposed method has improvements over product-based EWMA control, CF-EWMA control and threaded PCC control by 9.52%, 2523.43% and 11.71% on average, respectively, for estimating removal rate of historical data in mix-product CMP process.

Original languageEnglish
Pages (from-to)471-481
Number of pages11
JournalInternational Journal of Precision Engineering and Manufacturing
Volume13
Issue number4
DOIs
StatePublished - 1 Apr 2012

Keywords

  • CMP process
  • Mix-product
  • Removal rate
  • Run-to-run control

Fingerprint Dive into the research topics of 'Combined product and tool disturbance estimator for the mix-product process and its application to the removal rate estimation in CMP process'. Together they form a unique fingerprint.

Cite this