Systematic Defect Analysis for IC Chip Physical Sites
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systematic defect diagnosis methods in IC manufacturing are inefficient in analyzing trends and reliability of defects associated with physical features, often requiring costly and time-consuming physical failure analysis and failing to identify root causes of multiple defects effectively.
Innovation Solution
A systematic defect analysis method that partitions physical sites into groups based on various physical features, computes defect probabilities, and derives analysis results to identify culprit features by analyzing variance and trends in defect probabilities, using statistical techniques like simultaneous confidence intervals and variance calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical failure analysis is conducted to understand defect mechanisms, then reliability of defect analysis is improved, but time consumption and cost increase significantly
Solution Approach 1:
The patent segments the defect analysis process into two stages: (1) a screening stage using statistical analysis of defect data to identify potential systematic defects and candidate physical sites, and (2) a detailed physical failure analysis stage applied only to selected candidate sites. This segmentation reduces the number of samples requiring time-consuming PFA while maintaining analysis reliability.
Solution Approach 2:
The patent performs preliminary statistical analysis and defect pattern recognition before conducting physical failure analysis. By pre-identifying suspicious physical sites through defect data analysis and calculating defect probabilities, the method prepares a targeted list of candidates, thereby reducing the overall time required for complete defect characterization.
2Reliability
If physical failure analysis is conducted to understand defect mechanisms, then reliability of defect analysis is improved, but cost increases significantly
Solution Approach 1:
The patent segments the defect analysis population into two groups: a large screening group analyzed through statistical methods, and a small selected group undergoing expensive PFA. This segmentation ensures that costly PFA resources are allocated only to the most promising candidates, thereby maintaining reliability while reducing overall cost.
Solution Approach 2:
The patent uses statistical defect models and simulation results as proxies or copies of the actual physical failure mechanisms. By analyzing defect patterns, densities, and spatial distributions statistically, the method creates virtual representations of failure mechanisms that reduce dependence on expensive physical analysis for every case.
3Productivity
If statistical analysis techniques are used for systematic defect diagnosis, then productivity is improved, but measurement precision of defect trends deteriorates
Solution Approach 1:
The patent replaces traditional mechanical/statistical defect counting methods with a physics-based defect probability calculation model. By incorporating physical knowledge about defect generation mechanisms and using Bayesian inference, the method achieves more precise trend analysis while maintaining high diagnostic productivity.
Data Source
AI summary
A systematic defect analyzing method, includes: partitioning physical sites into groups to obtain a plurality of groups of physical sites according to a plurality of physical features of a chip corresponding to different potential systematic defects; utilizing a processor to compute at least one defect probability of each group of physical sites; and deriving an analysis result according to the plurality of defect probabilities corresponding to the plurality of groups of physical sites.