SEM Non-Visual Defect Reduction By Catherine Perry-Sullivan, Ph.D
-----------------------------------------------------------------------------------------
Cathy is a Product Marketing Manager at KLA-Tencor
Corporation, where she has specialized in applications
and marketing for wafer inspection products since 1996.
As design rules shrink to 45nm and below, defect and yield engineers are becoming increasingly concerned about the quality
of the defect Pareto coming from their review tools. They are finding that one of the largest categories on their defect Pareto is ‘SEM Non Visual (SNV),’ also called ‘Not Found.’ These SNV or NF categories are populated with defects that the inspection tools have found, but that the SEM review system has failed to re-detect and classify. The large number of SNVs interferes with the ability of defect or yield engineers to monitor the key defects of interest.  Figure 1. Percentage of SEM Non-Visual defects is growing as design rules advance. A new defect review and classification system has the unique ability to connect with compatible inspection systems, reducing SNVs counts by an order of magnitude or more. The result is a defect Pareto that more accurately represents the population of defects of interest on the wafer—generated in a shorter time. The new system has a five-fold approach to addressing SNV reduction: 1. Inspector Recipe Tuning on High Resolution SEM Review Systems.
Proprietary connectivity between the review system and compatible optical inspection systems enables 2X faster, more accurate inspector recipe development. The inspector’s recipe can be optimized on the SEM review tool, eliminating multiple transports of the wafer between the systems, and freeing the inspection tool for additional inspections. The efficiency of this method makes it practical to fully optimize the recipe—which often does not happen in today’s time-pressured fabs. The percentage of nuisance defects, such as color variation and other ‘real’ defects which do not affect yield, can be reduced by a factor of ten by using the new review system to help optimize the inspection system recipe. 2. High Positioning Accuracy. The new review and classification system is equipped with a high precision stage and advanced defect de-skewing algorithms. Together these can reduce SNV rates, by increasing the ability of the SEM review tool to capture <50nm defects—critical size at the 45nm node 3. Improved Automatic Defect Location. Low contrast or tiny defects can be completely overlooked, categorized as SNV by traditional defect re-detection approaches. The new tool introduces proven optical defect re-detection which contributes a considerable number of critical yield-limiting defects to the final Pareto.  Figure 2. KLA-Tencor’s new e-beam review and classification system, the eDR-5200, works with compatible inspection systems to reduce SNVs. 4. Very high resolution imaging. Defects smaller than 50nm require very high resolution in order to be imaged and classified correctly. The lack of resolution on some review tools can lead to SNV classification of subtle defects such has undersized contacts or very small pattern variations. High resolution imaging provides the user with enough defect resolution to classify the defect correctly. 5. Recognition of Previous-Layer Defects. Because an electron beam interacts only with the surface of a layer, previous-layer defects are traditionally classified as SNV. Anew approach accessing proprietary optical information from KLA-Tencor inspection tools allows characterization of previous-layer defects. Together these elements can reduce the percentage of SEM Non-Visuals by as much as an order of magnitude, or more. Given a defect Pareto that more accurately represents the population of defects of interest on the wafer, yield and defect engineers can more readily, and more quickly, take action to resolve the important defect issues—and protect their yield. Click here to learn more about the SEM non-visual defect reduction | 
