A new Virtual Metrology (VM) method for estimation of product quality characteristics will be presented using the recently introduced Growing Structure Multiple Model System (GSMMS) approach for modeling of non-linear dynamic systems. The underlying concept of local linear models enables representation of non-linear dependencies with non-Gaussian and non-stationary noise characteristics. In addition, localized analysis of VM inputs within the GSMMS framework enables detection of situations when the model is not adequate and needs to be improved. The newly proposed method was applied to an extensive dataset gathered from a plasma enhanced chemical vapor deposition tool operating in a major semiconductor manufacturing fab, with tool signatures being used to predict the mean film thicknesses on the wafers. The GSMMS based VM significantly decreased the number of measurements necessary for prediction, while improving VM accuracy, as compared to several linear and nonlinear benchmark VM methodologies. These beneficial results are credited to the GSMMS being able to store local models within its growing network of local VM models corresponding to various operating regimes of the underlying manufacturing machine, as well as to recognize situations when new physical measurements need to be taken and when new local VM models need to be added.Keywords: Virtual Metrology, nonlinear dynamic systems modeling, divide and conquer modeling of dynamic systems 0894-6507 (c)
In semiconductor fabrication processes, reliable feature extraction and condition monitoring is critical to understanding equipment degradation and implementing the proper maintenance decisions. This paper presents an integrated feature extraction and equipment monitoring approach based on standard built-in sensors from a modern 300mm-technology industrial Plasma Enhanced Chemical Vapor Deposition (PECVD) tool. Linear Discriminant Analysis was utilized to determine the set of dynamic features that are the most sensitive to different tool conditions brought about by chamber cleaning. Gaussian Mixture Models of the dynamic feature distributions were used to statistically quantify changes of these features as the condition of the tool changed. Data was collected in the facilities of a well-known microelectronics manufacturer from a PECVD tool used for depositing various thin films on silicon wafers, which is one of the key steps in semiconductor manufacturing. Dynamic features coming from the radio frequency (RF) plasma power generator, matching capacitors, pedestal temperature, and chamber temperature sensors were shown to consistently have significant statistical changes as a consequence of repeated cleaning cycles, indicating physical connections to the chamber condition.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.