A comparison of Taguchi methods and response surface methodology for optimising a CMOS process

The influence of process variables (hereafter called control factors or factors) on critical product parameters (hereafter called responses) is necessary to understand and control the IC fabrication process. However investigating the effect of one control factor at a time from a source of n factors in a totally rigorous fashion requires n! experiments. For a large number of factors this is clearly impractical. Driven by the requirement to reduce the number of experiments, two techniques have become popular in recent years. These are commonly known as Taguchi and RSM analyses and this paper will use a current fabrication process to compare and contrast their performance