Incorporating Core-to-Core Correlation to Improve Partially Good Yield Models

New partially good yield models have been developed by considering core-to-core correlation. The new models take into account the physical mechanisms which cause core-to-core correlation in advanced technologies. The improved models have clearly shown improved accuracy to actual data when compared to the existing models. First, existing partially good yield models are reviewed. Next, the physical mechanisms causing core-to-core correlation in advanced technology nodes are discussed. A simple correlated partially good yield model is given that is based on the physical mechanisms. A correlated partially good yield model based on Moody's correlated binomial distribution is proposed. Finally, actual partially good yield data of two 14 nm products are compared to both the existing models and the proposed new models. The new correlated partially good yield model better matches the actual yield data across a range of core counts compared to existing models.