Defect Classification System for Semiconductor Wafer Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect classification methods in semiconductor manufacturing face challenges in accurately distinguishing between Defects of Interest (DOI) and nuisance defects, often resulting in false positives and reduced sensitivity, especially as semiconductor processes require high precision and uniformity with shrinking feature sizes.
Innovation Solution
A computerized system and method that utilize a Processing and Memory Circuitry (PMC) to classify defects by obtaining defect clusters with spatial attributes, applying a cluster classifier trained on pre-labelled data, and performing defect filtration using specified filtering parameters to identify DOI, thereby improving sensitivity and reducing false alarms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional defect classification methods are used, then the examination process can be performed, but false positives increase and sensitivity decreases
Solution Approach 1:
The defect classification process is segmented into multiple independent stages: defect detection, feature extraction, classification, and verification. Each stage processes specific attributes independently before combining results, allowing the system to maintain high sensitivity while reducing false positives through staged filtering.
Solution Approach 2:
The system dynamically adjusts classification parameters and thresholds based on defect characteristics, spatial location, and process conditions. By changing parameters adaptively rather than using fixed thresholds, the system maintains high detection sensitivity while minimizing false alarms through context-aware parameter optimization.
2Measurement precision
If detailed inspections are performed on all devices, then defect detection accuracy improves, but examination time and productivity decrease
Solution Approach 1:
The system performs detailed inspections selectively on only those defects that meet specific criteria (e.g., unusual characteristics, critical locations, high-risk patterns). Routine defects receive standard processing while suspicious cases trigger enhanced examination, achieving high accuracy for critical defects without sacrificing overall throughput.
Solution Approach 2:
Different examination intensities are applied to different spatial regions and defect types based on their risk profiles. Critical areas and unusual defects receive detailed inspection while normal areas use faster methods, optimizing the balance between detection accuracy and examination speed across the entire wafer surface.
3Reliability
If classification criteria are made more stringent to reduce false positives, then reliability improves, but sensitivity to detect actual defects decreases
Solution Approach 1:
Multiple classification rules and verification mechanisms act as intermediaries between initial defect detection and final classification. Defects pass through several filtering stages with progressively stricter criteria, allowing the system to maintain high sensitivity in early stages while ensuring reliability through subsequent verification layers.
Solution Approach 2:
The system performs preliminary classification and filtering based on easily measurable attributes before applying more stringent criteria. This preliminary action separates obvious false positives early in the process, allowing subsequent strict classification to focus only on ambiguous cases without losing sensitivity to actual defects.
Data Source
AI summary
There are provided system and method of classifying defects in a specimen. The method includes: obtaining one or more defect clusters detected on a defect map of the specimen, each cluster characterized by a set of cluster attributes comprising spatial attributes including spatial density indicative of density of defects in one or more regions accommodating the cluster, each given defect cluster being detected at least based on the spatial density thereof meeting a criterion. The defect map also comprises non-clustered defects. Defects of interest (DOI) are identified in each cluster by performing respective defect filtrations for each cluster and non-clustered defects.