Novel Technique to Identify Systematic and Random Defects during 65 nm and 45nm Process Development for Faster Yield Learning

Advanced bright field inspection tools available today applied on development wafer may often result in 100 k to 1 M defects per wafer. Such defect data consist of systematic and random defects that may be yield limiting or may be just cosmetic issue with low probability of yield impact. It is also difficult to identify systematic defects from random defects by using traditional defect classification method where 50 to 100 defects per wafer are sampled on the SEM review. Missing important systematic defect types can be very costly and cause delays in product introduction. In this paper, a new approach has been introduced to improve identification of systematic defect and to improve defect sampling for SEM review. By applying design data to defect inspection, many of the systematic defects have been identified and monitored for efficient management of systematic defects. Application of design data in defect inspection provides new capability in identifying systematic defects that a traditional random sampling or repeater analysis could not identify. Identification and characterization of an important process related defect type, STI cave defect, is described in this paper to illustrate the new approach. By using the design data to bin defect types, the STI cave defect was identified and quantified. The discovery was further confirmed using SEM review and FIB. An insufficient gap-fill during the deposition step was determined as culprit for this void defect type. The novel technique described here provided a way of detecting and identifying such systematic defect, enabling fab to quickly resolve the issue. Furthermore creating this capability embedded on inspection tool promises to provide a new paradigm in defect inspection technology.