Defect Detection Using Structural Segmentation for Semiconductor Wafers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in semiconductor manufacturing struggle to accurately differentiate between n-type metal-oxide-semiconductor (NMOS) and p-type MOS (PMOS) structures due to similar gray level values, leading to unclear indications about wafer structures and potential defects, especially as device dimensions decrease.
Innovation Solution
A system and method that utilize structural information to segment and detect defects on specimens by employing an inspection subsystem with an energy source and detector, and computer subsystems to separate output based on characteristics of structures, allowing for distinct defect detection in different segments, independent of gray level values, using SEM images or design files for structural identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gray level segmentation is used to divide images into different areas, then defect detection can be guided by structure information, but NMOS and PMOS structures with similar gray level values cannot be clearly differentiated
Solution Approach 1:
The patent changes the parameter basis for segmentation from gray level values to structural characteristics. Instead of relying on intensity values that may be similar across different structures, the system uses structural parameters such as pattern geometry, spatial arrangement, and morphological features to differentiate between NMOS and PMOS structures, thereby achieving clear structural identification without information loss
Solution Approach 2:
The patent introduces structural information as an intermediary between the image data and defect detection process. By using structural characteristics as a mediating layer, the system can accurately identify different device structures even when their gray level values are similar, enabling precise defect detection without losing structural detail information
2Productivity
If a single defect detection method is applied to the entire array area, then the inspection process is simplified, but defects in different structures cannot be detected with optimized sensitivity
Solution Approach 1:
The patent segments the array area into multiple regions based on structural characteristics, allowing different defect detection methods and sensitivity settings to be applied to each segment. This enables optimized defect detection for each structure type while maintaining overall inspection efficiency through automated segmentation and method assignment
Solution Approach 2:
The patent applies local quality by tailoring defect detection parameters and methods to specific structural regions. Each segmented area receives customized inspection settings appropriate to its structural characteristics, improving detection sensitivity for local defect types while maintaining overall productivity through efficient resource allocation
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances defect detection sensitivity and yield relevance by accurately identifying structures and suppressing nuisance events, improving the ability to detect defects in semiconductor wafers by using structural information for segmentation and defect analysis.
Implementation Method 1
The energy source is configured to generate energy that is directed to a specimen. The detector is configured to detect energy from the specimen and to generate output responsive to the detected energy.
Data Source
AI summary
Systems and methods for detecting defects on a specimen based on structural information are provided. One system includes one or more computer subsystems configured for separating the output generated by a detector of an inspection subsystem in an array area on a specimen into at least first and second segments of the output based on characteristic(s) of structure(s) in the array area such that the output in different segments has been generated in different locations in the array area in which the structure(s) having different values of the characteristic(s) are formed. The computer subsystem(s) are also configured for detecting defects on the specimen by applying one or more defect detection methods to the output based on whether the output is in the first segment or the second segment.