Defect Classification Decision Algorithm for Semiconductor Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the semiconductor industry, the classification of defects detected during fabrication processes becomes increasingly complex and time-consuming due to shrinking design rules and new manufacturing processes, leading to a need for efficient methods to distinguish between defects of interest and nuisances, especially with high-sensitivity inspection tools reporting thousands of defects per wafer.
Innovation Solution
A computer-implemented method and system for classifying defects using a decision algorithm that processes defect and design data to sort defects into predefined bins, utilizing a sequence of classification operations, including design rule checks and design-based binning, to effectively categorize defects based on their attributes and proximity to patterned features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple classification operations are performed to accurately distinguish defects of interest from nuisances, then classification accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The classification process is divided into multiple sequential operations, each handling a specific aspect of defect classification. The decision algorithm segments the classification task into distinct operations that process defects in stages, allowing accurate distinction between defects of interest and nuisances while managing computational load through structured division of work.
Solution Approach 2:
The system performs preliminary classification operations to filter and prioritize defects before more complex analysis. By conducting initial sorting and categorization operations, the system pre-processes defect data to reduce the computational burden on subsequent classification operations, thereby maintaining accuracy while reducing overall processing time.
2Adaptability or versatility
If design-based classification operations are applied to handle new manufacturing processes, then classification capability is improved, but device complexity increases
Solution Approach 1:
The decision algorithm is designed as a universal framework that can handle multiple types of defects and classification operations through a single integrated system. The algorithm accommodates various manufacturing processes and defect types using common classification operations, reducing the need for separate specialized systems and thereby managing complexity while maintaining high adaptability.
Solution Approach 2:
The system introduces an intermediary decision algorithm layer between inspection data collection and final classification. This intermediary layer processes and mediates the complex interactions between new manufacturing processes and defect detection, translating diverse process-specific data into unified classification categories, thereby managing system complexity while enhancing classification capability.
3Productivity
If the number of defects processed by each classification operation is reduced, then processing efficiency is improved, but the total number of classification operations required increases
Solution Approach 1:
The classification process is segmented into operations that process smaller subsets of defects, improving processing efficiency for each operation. The decision algorithm divides the defect list into manageable groups based on characteristics and priorities, allowing each operation to process fewer, more targeted defects while maintaining overall throughput through parallel processing and efficient workflow management.
Data Source
AI summary
There is provided an inspection method capable of classifying defects detected on a production layer of a specimen. The method comprises: obtaining input data related to the detected defects; processing the input data using a decision algorithm associated with the production layer and specifying two or more classification operations and a sequence thereof; and sorting the processed defects in accordance with predefined bins, wherein each bin is associated with at least one classification operation, wherein at least one classification operation sorts at least part of the processed defects to one or more classification bins to yield finally classified defects, and wherein each classification operation, excluding the last one, sorts at least part of the processed defects to be processed by one or more of the following classification operations.