K. HUI, J.C.S. LU & K.C. SUN, Taiwan Semiconductor Manufacturing Company Ltd., Hsin-Chu, Taiwan
ABSTRACT
This article presents a case study on the applications of modelling techniques to a diffusion furnace, and discusses some of the practical issues encountered during the identification process in semiconductor manufacturing. The emphasis is from the perspective of a wafer fabrication plant, with the objective of achieving tighter targets of statistical process control. The investigation uses actual process data from on-line production runs. Procedures are described on the identification process, starting from data collection, to preliminary analysis and selection of dynamic models for the diffusion furnace, to the revelation of underlying relationships between zone temperatures and inputs from heaters, pressure settings and mass flow controllers.