Efficient Method for Testing the Batch-Processing Process Yield

Abstract

Batch manufacturing processes have been widely used in various manufacturing processes, such as wafer fabrication, IC fabrication, and gridline printing process in the solar battery fabrication. In a batch manufacturing process, products are produced batch-by-batch. Thus, total process variations are generally divided into batch-by-batch variation and within-batch variation. The main purpose of this study is to provide an efficient method for testing the batch-processing process yield. Base on the one-way random effect model, the generalized pivotal quantity is utilized to establish the generalized confidence interval for assessing the process yield index. By simulations, the proposed method shows that its empirical coverage probability is not affected by the batch effect, and is still close the nominal coverage probability as the batch size increases.