Charged Particle Image Filtering for Real-Time Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting defects in semiconductor IC chip manufacturing using charged particle beam systems face challenges in high-speed noise reduction and image processing, leading to inefficiencies in real-time defect detection due to high computational costs and unsuitable noise reduction techniques for scanning electron microscopes.
Innovation Solution
A computer-implementable method and device that applies a filter, specifically a convolution with a uniform kernel, to sample images from charged particle beam systems, comparing the filtered images to a reference image to detect defects, optimizing noise reduction and processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional noise reduction techniques are applied to charged particle beam images, then noise is reduced, but processing time and computational cost increase prohibitively for high-speed inspection
Solution Approach 1:
The image processing is segmented into distinct stages: acquisition of multiple frames, application of motion compensation using correlation techniques, differential processing to extract defect information, and noise reduction. Each stage processes specific aspects of the data independently, allowing optimization of processing speed at each step while maintaining overall noise reduction quality.
Solution Approach 2:
Motion compensation is performed preliminarily on the acquired image frames before defect detection and noise reduction steps. By pre-aligning the frames using correlation techniques, subsequent processing steps operate on already-stabilized data, reducing the computational burden and processing time required for later stages while preserving noise reduction effectiveness.
2Productivity
If high computational power is used for real-time image processing, then processing speed improves, but system complexity and cost increase
Solution Approach 1:
Complex mechanical or hardware-based processing systems are replaced with software-based correlation techniques and differential processing algorithms. The motion compensation uses mathematical correlation operations rather than complex hardware alignment mechanisms, and defect detection uses differential processing instead of complex real-time analysis hardware, reducing system complexity while maintaining processing speed.
Solution Approach 2:
The processing approach changes parameters such as working with difference images rather than original images, using correlation coefficients to quantify similarity, and applying motion vectors derived from parameter optimization. These parameter changes enable efficient software-based processing that achieves real-time performance without requiring complex hardware systems.
3Reliability
If standard image processing filters are applied to SEM images, then noise is reduced, but defect detection accuracy decreases due to unsuitable filtering for charged particle images
Solution Approach 1:
Differential processing serves as an intermediary step between image acquisition and noise reduction. By computing the difference between motion-compensated frames and reference images, the system creates intermediate difference images that contain defect information with reduced background complexity. Standard filtering then operates on these simplified difference images rather than complex original images, preserving defect detection accuracy while achieving noise reduction.
Solution Approach 2:
The processing applies different operations to different aspects of the images: motion compensation handles global frame alignment, differential processing extracts local defect features, and noise reduction filters are applied specifically to the difference images. This localized approach ensures that each processing step operates optimally on the data characteristics it encounters, maintaining defect detection accuracy while achieving effective noise reduction.
Data Source
AI summary
A data processing device for detecting defects in sample images generated by a charged particle assessment system, the device comprising: an input module, a filter module, a reference image module and a comparator. The input module is configured to receive a sample image from the charged particle assessment system. The filter module is configured to apply a filter to the sample image to generate a filtered sample image. The reference image module is configured to provide a reference image based on one or more source images. The comparator is configured to compare the filtered sample image to the reference image so as to detect defects in the sample image.