Wafer Yield Trend Monitoring via Regional Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current yield management systems in semiconductor manufacturing struggle with timely calculations and storage of derived parameters and aggregate data due to limited computational bandwidth and storage capacity, leading to delays in identifying and rectifying defective trends in the manufacturing process, especially when dealing with vast amounts of data from thousands of integrated circuits on each wafer.
Innovation Solution
A method is introduced to calculate and store regional aggregated data, designating equal-area spatial regions on wafers to simplify statistical calculations, allowing quicker access to critical manufacturing statistics and trends, thereby reducing the time required for computations and providing timely feedback for process control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If calculations of derived parameters and aggregate data are performed on the complete set of raw test data, then the accuracy and completeness of manufacturing trend analysis is improved, but the computational time and resources required increase significantly
Solution Approach 1:
The patent divides the wafer into multiple spatial regions and performs aggregate data calculations separately for each region rather than processing the entire wafer as one unit. This segmentation reduces the computational complexity and time required while maintaining the ability to identify localized manufacturing trends and defects.
Solution Approach 2:
The patent calculates aggregate data for a selected number of spatial regions rather than all possible regions, and computes derived parameters for a subset of the complete test data set. This partial action approach provides sufficient information for manufacturing trend analysis without the excessive computational burden of processing every data point.
2Reliability
If regional aggregate data is calculated for every spatial region on each wafer, then the ability to identify localized manufacturing defects is improved, but the storage capacity and computational bandwidth required increase
Solution Approach 1:
The wafer is divided into a predetermined number of spatial regions, and aggregate data is calculated for each region. This segmentation enables localized defect identification while managing data volume by focusing on regional statistics rather than individual device data.
Solution Approach 2:
Different spatial regions on the wafer are analyzed with region-specific aggregate data, allowing the system to identify and characterize localized manufacturing defects in specific areas while using less storage capacity than analyzing every individual device across the entire wafer.
3Measurement precision
If a large number of spatial regions are designated on each wafer for detailed analysis, then the resolution of manufacturing trend detection is improved, but the arithmetic computation complexity increases
Solution Approach 1:
The wafer is divided into a predetermined number of spatial regions that balances detection resolution with computational feasibility. The patent selects a number of regions that provides sufficient detail for manufacturing trend detection without creating excessive arithmetic computation complexity.
Solution Approach 2:
The patent adjusts the number of spatial regions and the type of aggregate data calculated based on the specific manufacturing process and defect characteristics being monitored. This parameter optimization allows the system to achieve adequate detection resolution while controlling computational complexity.
Data Source
AI summary
A parametric parameter is selected, which has an upper specification limit and a lower specification limit. A storage percentile is determined. The storage percentile is equal to a product yield percentage if the number of the set of measurements greater than the upper specification limit exceeds the number of the set of measurements lower than the lower specification limit, and is equal to the product yield percentage subtracted from one hundred percent if the number of the set of measurements less than the lower specification limit exceeds the number of the set of measurements greater than the upper specification limit. A number of spatial regions on the wafer is designated. A first group of measurements from the set of measurements is obtained for a first spatial region of the spatial regions. A measurement closest to the storage percentile is stored.