Wafer Defect Analysis Grid Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, defect classification in logic products with random patterns is challenging due to varying sensitivity regions, leading to inaccurate defect analysis and yield prediction, especially when defects are caused by multiple factors or have non-regional systematic causes.
Innovation Solution
A defect analyzer divides the wafer into grid squares, extracts pattern feature quantities, classifies regions, matches defect information, calculates defect size distribution, and compares it to a predetermined threshold to accurately identify random or systematic defects, enabling high-accuracy analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the wafer is tested at the same sensitivity level across all regions, then the testing process is simple and fast, but the accuracy of defect detection varies in different regions leading to poor defect analysis
Solution Approach 1:
The wafer surface is divided into multiple sensitivity regions based on pattern density and layout characteristics. Each region is assigned an appropriate sensitivity level for defect detection, allowing optimized testing parameters for each area rather than using a uniform sensitivity across the entire wafer. This segmentation enables both high productivity and accurate defect detection by matching test parameters to local pattern characteristics.
2Measurement precision
If manual sensitivity setting is used for different regions, then defect detection accuracy improves, but the process complexity and time consumption increase
Solution Approach 1:
The system automatically determines sensitivity settings for different regions by analyzing pattern data and layout characteristics. The automated sensitivity assignment eliminates the need for manual intervention while maintaining high defect detection accuracy. The system self-adjusts testing parameters based on region-specific pattern density and structural features, reducing both process complexity and time consumption compared to manual setting methods.
3Productivity
If defects from multiple causes are treated uniformly, then the analysis process is simple, but the accuracy of defect classification and yield prediction decreases
Solution Approach 1:
Defects are classified into different types (random defects and systematic defects) based on their distribution patterns and characteristics. The system segments defect data by analyzing spatial distribution, pattern matching, and statistical properties to distinguish between random and systematic causes. This classification enables targeted analysis and prediction strategies for each defect type, improving both accuracy and productivity by avoiding uniform treatment of all defects.
Data Source
AI summary
A defect analyzer system having a program storage device storing instructions and an arithmetic unit executing the instructions. The system divides a defect analyzing region of a wafer into a plurality of grid squares and classifies the grid squares into a plurality of groups. The system then matches the defect information with the defect analyzing region. The defect information includes defect positions and sizes detected in the defect analyzing region. The system calculates a defect size distribution for each defect size in each of the groups, compares the defect size distribution and a predetermined estimation distribution for each of the groups, and calculates a difference. The system then compares the difference of each of the groups and a predetermined threshold value, extracts the group having the difference equal to or smaller than the predetermined threshold value, and outputs the defect information corresponding to the extracted group.