研究目的
To present a method for quantifying a risk for killer defects at layer level and estimating yield for substrate packages using information from design files.
研究成果
The paper presents a method for quantifying layer level risks using the risk distance metric, showing an inverse correlation between risk layer ranks and yield. The yield model matches with actual yield for different layers, and the estimated yield for a second design correlates with actual yield within less than 1% yield difference for given design and process conditions.
研究不足
The study focuses on only short killer defects, suggesting that the same concept should be able to apply for open killer defects as well. The method assumes that the short killer defect PDF remains the same between different designs when run through the factory line.
1：Experimental Design and Method Selection:
The study defines a risk distance as a key parameter extracted from designs using image processing techniques to calculate risk ranks and predicted yield.
2：Sample Selection and Data Sources:
Two different designs, each having multiple layers, are analyzed and compared with data from baseline lots.
3：List of Experimental Equipment and Materials:
MATLAB program is used for image processing and analysis.
4：Experimental Procedures and Operational Workflow:
The method involves image pre-processing, geometrical calculation for short and open risks, data post-processing, and layer level risk assessment and yield model calculation.
5：Data Analysis Methods:
The study compares the layer level risk ranks with actual yield values and validates the yield prediction model using killer defect PDF from baseline lots.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
