Management of critical areas and defectivity data for yield trend modeling

This paper reports a yield model applied to a large set of defectivity data measured on a 0.35 micron process in ST Crolles plant and critical areas extracted by survey sampling. It takes into account all defect densities measured at the main process steps and determines their respective yield loss. The robustness of the model was tested week by week during three months of production for two high volume devices. The model was then applied wafer by wafer on a new process version.