Defect Point Clustering for Repeated Substrate Yield Alarms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems for manufacturing yield analysis in production lines face inefficiencies due to high costs, slow data processing, and difficulty in managing large volumes of data, leading to increased time and manpower consumption in identifying and addressing production defects.
Innovation Solution
A data processing system utilizing a distributed storage apparatus, data transformation apparatus, and data processing apparatus to analyze substrates for defect points, performing clustering calculations to identify repeated defects and generate alarm information for timely corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional data processing systems are used for manufacturing yield analysis, then comprehensive defect detection can be achieved, but data processing speed is slow and time consumption is high
Solution Approach 1:
The patent segments the data processing task by dividing substrates into analysis units based on production time and arrival rates. This segmentation allows parallel processing of different substrate batches, reducing overall processing time while maintaining comprehensive defect detection through systematic clustering analysis of defect points across multiple substrates.
2Measurement precision
If traditional data processing systems are used for manufacturing yield analysis, then defect analysis can be performed, but processing cost is high
Solution Approach 1:
The patent extracts only the essential defect point coordinates from comprehensive substrate inspection data, rather than processing all raw inspection data. This extraction approach maintains defect analysis capability by focusing on critical defect information while significantly reducing computational resources and processing costs.
3Measurement precision
If comprehensive inspection data of all substrates is processed, then all defects can be detected, but data management complexity increases
Solution Approach 1:
The patent merges defect point information from multiple substrates into unified analysis units, combining defect coordinates, substrate identifiers, and production metadata into structured datasets. This merging approach maintains comprehensive defect detection coverage while simplifying data management through organized clustering that groups related defect information together.
4Measurement precision
If traditional clustering methods are used on all defect points, then all defect patterns can be identified, but processing speed decreases
Solution Approach 1:
The patent performs preliminary actions by pre-segmenting substrates into analysis units based on production time sequences and arrival rates before clustering. This preliminary organization of data into manageable units with similar temporal characteristics enables faster clustering processing while maintaining accurate defect pattern recognition through systematic analysis of each unit's defect points.
Data Source
AI summary
A data processing method and system, and a computer-readable storage medium. The method includes: acquiring target data, the target data including data from a plurality of analysis units (21); acquiring, from each of the analysis units, inspection data of a plurality of substrates to be inspected that contain defect points (22); superimposing the defect points in the plurality of substrates to be inspected to obtain a data set of the defect points (23); performing a clustering calculation on the data set of the defect points to obtain a first cluster set of the defect points (24); and generating, based on the first cluster set of the defect points, alarm information for a device contact position corresponding to the first cluster set (25). With the method, the substrates to be inspected in each analysis unit are analyzed, and a time interval between the process of production and the process of analysis of repeated defects is shortened, which is conducive to improving the data processing efficiency. Moreover, with the method, it is possible to detect the defect points that are repeatedly defective at the same position, i.e., to obtain the first cluster set of the defect points, so as to achieve the effect of timely inspection of the repeated defects caused by the production device, which is conducive to improving a production yield of the substrates to be inspected.