Determining equipment performance using analysis of variance

It is pointed out that one of the critical steps required to improve the overall quality of a wafer manufacturing operation is reducing the variability of the individual processing operations. The goal is to design a consistent method of monitoring equipment performance that samples the major sources of variation that can affect the process output. The authors discuss a statistical method that utilizes multiple types of variation (multi vari) analysis and the analysis of variance to achieve this goal. The data presented are taken from a commercially available silicon epitaxial reactor and FTIR (Fourier transform infrared spectrometry) film measurement tool, but the technique is generic and applicable to other wafer processing operations. The analysis method is presented along with some examples of the application of the calculated standard deviation and process variation. In addition to determining the performance of a piece of equipment over time, this method of analysis can be used to specify and qualify equipment. As an extension of this analysis, it is possible to determine the contribution of the measurement system to the observed variability in the process output.<<ETX>>