Application of GR&R for productivity improvement

Abstract

Testing is a check point in semiconductor manufacturing to ensure quality products are delivered to the customer. As demand and complexity of products increases, advanced equipments are used to perform the test. However, the outcome of the tests does not just depend on availability of advanced equipment, but also depends on the consistency and stability of the measuring system, or commonly called gage. Inconsistency and instability of the gage may result in good units being sorted as bad units and eventually being scrapped leading to production yield loss. As a result the gage performance has become an important issue in testing process. Gage Repeatability and Reproducibility (GR&R) is a statistical model to ensure consistency and stability of the measuring system. By performing GR&R study one could determine the uncertainty present in the gage and help to prevent production losses. This paper reports a study on GR&R on a 416 leads 16 bits micro-controller device to check the consistency and stability of the measuring system used. A total of 150 measurement data was collected and analyzed. Incidentally, the study was made at the time one of the test on the micro-controller suffered a comparatively high yield loss. This GR&R study came just at the right time and the results showed inconsistency in the measuring system. Further detailed analysis finally picked up the root cause of the poor GR&R performance in the measuring system. This indicated that GR&R could be used as a powerful tool to improve productivity even for complex semiconductor products.