Forecast of CMOS Imagers Yield Learning by the Gompertz Model

This paper presents a novel approach to modeling yield using the Gompertz function, which is widely used in biology to model the growth processes of plants, tumors, etc. We demonstrate that the yield-learning process in a semiconductor fab follows the same behavior of the growth of biological systems. We start with a simple time series model, which describes the learning process in terms of defect density reduction. Then we obtain the Gompertz growth model, which also fits the experimental data better than more traditional learning models such as the Gruber' s general yield learning model [38] .