Semiconductor Image Segmentation Using Hybrid Probability Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image segmentation methods for semiconductor specimens lack repeatability, leading to inconsistent defect detection and classification during semiconductor fabrication, particularly due to variations in imaging conditions and tool processes.
Innovation Solution
A method using a processor and memory circuitry that generates a second label map by processing a first probability map and a simulated probability map, obtained through deep learning models and design data, to enhance the segmentation of semiconductor specimen images with improved repeatability, incorporating distance transforms and weighted probability combinations for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current image segmentation methods are used for semiconductor specimens, then the segmentation process can be performed, but the repeatability is poor leading to inconsistent defect detection and classification
Solution Approach 1:
The patent combines image-based probability maps from deep learning models with design-based probability maps from layout data to create a hybrid segmentation approach. This merging of two different data sources compensates for the weaknesses of each individual method, with image-based methods capturing actual variations and design-based methods providing theoretical consistency, thereby improving both repeatability and precision
Solution Approach 2:
The patent introduces probability maps as an intermediary representation that bridges the gap between raw images/design data and final segmentation labels. By operating in the probability space and combining multiple probability estimates before generating final labels, the system achieves more consistent and repeatable segmentation results
2Adaptability or versatility
If deep learning models are used to generate probability maps from images, then segmentation can be performed, but variations in imaging conditions and tool processes reduce repeatability
Solution Approach 1:
The patent applies different processing approaches to different data sources: image-based probability maps capture local variations and actual specimen characteristics, while design-based probability maps provide global theoretical expectations. By combining these complementary local and global information sources, the system maintains adaptability to imaging variations while improving repeatability
Solution Approach 2:
The patent pre-calculates design-based probability maps from layout data before image processing, creating a reference framework that cushions against variations in imaging conditions. This pre-established theoretical model serves as a stable baseline that compensates for uncertainties in the image-based segmentation process
Data Source
AI summary
There is provided a system and method of segmenting an image of a fabricated semiconductor specimen. The method includes: obtaining a first probability map corresponding to the image representative of at least a portion of the fabricated semiconductor specimen and indicative of predicted probabilities of pixels in the image to correspond to one or more first structural elements presented in the image, obtaining a first label map informative of one or more segments representative of second structural elements and labels associated with the segments, performing simulation on the first label map to obtain a second probability map indicative of simulated probabilities of pixels in the first label map to correspond to the one or more segments, and generating a second label map based on the first probability map and the second probability map, the second label map being usable for segmentation of the image with enhanced repeatability.