Yield Process Control based on the Production Data

Semiconductor device manufacturing process involves a large number of sophisticated production steps that have to be controlled precisely so to achieve and to maintain the required yield and product quality specifications. Therefore, it is critical to employ extensive data collection from numerous process-associated sensors. The data are then utilized by the process control system (PCS) to monitor and control the manufacturing process. This paper presents a real-world study performed at one of the major semiconductor device manufacturers where the massive data from the production line sensors were analyzed by PCS to (1) determine parameters that are significantly correlated with the product yield; and (2) perform a regression analysis to estimate the yield from PCS data. The study enabled to identify several key parameters, that can be optimized to improve the product yield.