Abstract
The cumulative sum (CUSUM) control chart is much better than a C control chart when the process demonstrates small to moderate-sized shifts. This study provides the calculating formula for the CUSUM's reference values, calculates ARLs using a Markov chain approach for a semiconductor production. An actual data set from a wafer plant is used to demonstrate the operation of the proposed model. Simulation method is used to double-check the analytical ARL result. The results show that CUSUM control chart has significantly better performance. It provides a useful method to detect the process variations and improve process quality for wafer plants.