Semiconductor manufacturing has always been a high-precision industry, where the precision levels have increased even further in recent years. For instance, most companies are now aiming for defects per billion instead of defects per million. In order to achieve these high levels of quality (defects per billion), semiconductor manufacturers are employing a number of statistical tools and techniques to catch defects.
One such tool that is gaining popularity is Gauge Repeatability and Reproducibility, abbreviated as Gauge R&R. Gauge R&R is now being widely used in the semiconductor industry to statistically measure the deviations and variability caused by the measuring system, the tool, or the operator, helping flag out inconsistent equipment and processes.
The repeatability and reproducibility of each measurement, taken by a measuring or testing device, should remain within a small limit of deviation from the specification tolerances of the process. This is to ensure that if the product deviates from its optimal performance because of a measurability error of the tool, the deviation still remains within the required performance bands. This is done primarily to guarantee that the required specifications of the devices, and optimal performance of the underlying applications performed by the devices, are met and there are no complaints or returns from the customers that may result in financial or reputation loss.
The Gauge R&R tool has become synonymous with semiconductor production and testing all around the world, and its popularity is increasing as more and more semiconductor devices are being connected to each other—IoT devices.
How does Gauge R&R work?
The underlying principles or statistical techniques used in Gauge R&R are fairly basic yet extremely powerful. The technique involves studying a dataset generated by a sample of control-group devices that meet the required specifications. These devices are tested persistently by the same machine to measure repeatability, and then tested by different machines to measure reproducibility of the testing devices.
The data is then collected and analyzed by a Semiconductor Data Analysis Software using various statistical techniques such as ANOVA, EMP (Evaluating the Measurement Process), and Average & Range Method, to calculate the variance from the control group and to determine a possible cause of the variance. Gauge R&R helps to quickly identify and mitigate issues relating to problem tests, sites, and equipment failures. Once the datasets are generated and statistically analyzed, the final measurements are added as guardbands to the test parameters to ensure that packaged devices are performing according to the required specifications, and faulty devices are identified before they are shipped to end customers.
Breaking down Gauge R&R
As mentioned earlier, Gauge R&R is the abbreviation of Gauge Repeatability and Reproducibility, where the variation in measurement of the testing device is repeatedly kept within the control range and reproduced by different tool sets.
- Repeatability is observed when the same tool or operator tests the same device several times using the same gauge and a similar test environment. If the variation is consistent between all the measurements, the test tool or operator is termed as repeatable and gauge repeatability is observed.
- Reproducibility is observed when different tools or operators test the same device several times using the same gauge and similar test environment. If the variation is consistent between all the measurements, the test tool or operator is termed as reproducible and gauge reproducibility is observed.
When combined, the two processes are collectively termed as Gauge R&R, which helps differentiate between changes in the old and new operators and provides guidelines to ensure the quality of measurement systems. Coupling Gauge R&R with other statistical tools, powered by an end-to-end yield management system, can ensure that the semiconductor manufacturing process is detecting defects per billion.
See related articles:
- Role of statistical process control in semiconductor yield management
- Part average testing’s popularity grows in the semiconductor industry
- Wafer-mapping software in modern-day high-volume manufacturing
- RMAs and yield management systems in semiconductor manufacturing
- Engineering data and the economics of semiconductor manufacturing
- Counterfeiting and semiconductor test-data management
