Automatic in-line measurement for the identification of killer defects

This paper describes the use of Fast Static RAMs (FSRAM) to identify the locations of defects which are then used to wafer map defect sites. It describes a method to automate the correlation of in-line with end-of-line defect measurements to quickly identify the source of killer defects. In-line defects are identified using a Tencor Surfscan 7000 and these defect locations are overlayed onto failing bitmap data from functional testers. The analysis and overlay software has been written using a commonly available database, RS/1 to automatically generate histograms of defects-to-fail ratios for each monitoring stage. This rapidly identifies the the process steps that have a major impact upon yield and these can then be targeted as part of a defect reduction strategy. Using this approach large quantities of data can be quickly processed allowing realistic SPC defectivity limits to be set from large sample sizes