Yield variation is a source of constant investigation and improvement efforts in the semiconductor industry. Especially in the case of power MOSFET suppliers, margins dictate that even small yield improvement can be used as competitive advantages in the marketplace. Therefore, any useful technique to help yield managers to direct resources and analyze data should be considered. This paper evaluates the usefulness of multivariate statistical techniques aiming toward generation of insight that can be used to improve yields. Although many of the reported results are rather intuitive, some interesting findings are presented that may lead future work toward more significant improvement.
Lite, Kevin, "Multivariate Statistical Techniques: Application to Semiconductor Power Device Manufacturing Yield Enhancement" (2003). Engineering and Technology Management Student Projects. 1566.