Analysis of mrr and wiwnu in oxide cmp using sparse experimental data
Conference Paper
Overview
Additional Document Info
View All
Overview
abstract
Availability of limited experimental data impedes the ability of current CMP models to accurately capture and predict the underlying complex chemo-mechanical interactions. Modeling approaches that can effectively interpret the available sparse experimental data are therefore necessary. In this paper, characteristic relationships connecting the material removal rate (MRR) and within-wafer-nonuniformity (WIWNU) with salient machine and process parameters are investigated for different oxide CMP operations. Approaches based on combining statistical and nonlinear modeling principles are developed to model these characteristics using sparse experimental data together with known process relationship structures. The resulting models are validated using experimental data obtained from a laboratory and an industrial site. The results indicate the adequacy of simpler models to capture MRR relationships and motivate the need for nonlinear modeling approaches to capture the complex relationships that characterize WTWNU. Insights into some plausible physical mechanisms that underpin these relationships have also emerged based on these investigations.
