Semiconductor Defect Detection via Dual-Processor Bit Depth Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor defect detection methods face challenges in achieving high detection sensitivity while maintaining high throughput, particularly due to the high computing power required for processing images at high bit depths.
Innovation Solution
A computerized system with a hybrid architecture is proposed, utilizing two processing and memory circuits (PMCs) to perform distributed double-funnel defect detection. The first PMC converts high-bit-depth images to lower bit depth, enabling efficient processing by the second PMC, which performs the first funnel of defect detection. The first PMC then performs a second funnel of defect detection on image patches from the original high-bit-depth image, leveraging advanced processing power for improved sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-bit-depth images are processed for defect detection, then detection sensitivity is improved, but computing power requirements increase
Solution Approach 1:
The defect detection process is divided into two separate funnels: a first funnel that processes down-sampled low-bit-depth images for rapid defect candidate identification, and a second funnel that processes only extracted image patches from the original high-bit-depth image for precise defect characterization. This segmentation allows the system to maintain high detection sensitivity while significantly reducing overall computing power requirements.
Solution Approach 2:
The patent extracts only the necessary image patches containing defect candidates from the full high-bit-depth image for detailed analysis. By taking out and processing only these relevant regions in the second funnel, the system avoids the computational burden of processing the entire high-resolution image while maintaining detection sensitivity for the identified defect areas.
2Measurement precision
If high-resolution inspection is performed across the entire specimen, then defect detection accuracy is improved, but processing time increases
Solution Approach 1:
The inspection process is segmented into two phases: a first phase using low-bit-depth down-sampled images for rapid whole-specimen scanning to identify defect candidate locations, and a second phase using high-bit-depth image patches only at those specific locations. This segmentation dramatically reduces processing time while maintaining defect detection accuracy at the critical locations.
Solution Approach 2:
The system performs preliminary defect candidate identification using low-resolution images before conducting high-resolution analysis. This preliminary action filters out the majority of the specimen that does not contain defects, allowing the high-resolution inspection to be applied only where necessary, thus reducing overall processing time while maintaining accuracy.
3Measurement precision
If full-image processing is performed at high bit depth, then detection sensitivity is maintained, but throughput decreases
Solution Approach 1:
The processing workflow is segmented into two parallel streams: a fast low-bit-depth stream that processes the entire image at high throughput to identify defect candidates, and a selective high-bit-depth stream that processes only extracted patches. This segmentation enables the system to maintain high overall throughput while preserving detection sensitivity for the identified defect regions.
Solution Approach 2:
The system applies high-bit-depth processing only partially, specifically to extracted image patches containing defect candidates, rather than excessively processing the entire high-resolution image. This partial application of high-bit-depth processing maintains detection sensitivity where needed while significantly improving overall throughput by avoiding unnecessary high-computational processing of defect-free areas.
Data Source
AI summary
There is provided a system and method of defect detection of a semiconductor specimen. The method includes obtaining a first image of the specimen acquired at a first bit depth, converting by a first processor the first image to a second image with a second bit depth lower than the first bit depth, transmitting the second image to a second processor configured to perform first defect detection on the second image using a first defect detection algorithm to obtain a first set of defect candidates, and sending locations of the first set of defect candidates to the first processor, extracting, from the first image, a set of image patches corresponding to the first set of defect candidates based on the locations, and performing second defect detection on the set of image patches using a second defect detection algorithm to obtain a second set of defect candidates.