Specimen Defect Detection via Noise Map Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in semiconductor manufacturing lack sensitivity, particularly in distinguishing between real defects and noise, leading to false alarms and reduced efficiency in high-density, ultra-large scale integration processes.
Innovation Solution
A computerized system and method for defect detection that involves partitioning specimen regions based on image and design data, receiving noise maps, and performing segmentation to associate regions with noise levels, enabling improved defect detection by adjusting noise maps and setting detection thresholds based on calculated scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional defect detection methods are used, then the inspection process is simple, but the sensitivity to distinguish defects from noise is insufficient
Solution Approach 1:
The detection system divides the specimen into multiple regions of interest (ROIs) based on design data and image characteristics. Each ROI is further segmented into different noise level zones (high, medium, low noise) allowing differentiated detection strategies. This segmentation enables the system to focus computational resources on critical areas while maintaining overall detection sensitivity.
Solution Approach 2:
The patent implements local quality by applying different detection thresholds and noise tolerance levels to different regions of the specimen. High-noise regions receive higher thresholds while low-noise regions use lower thresholds, optimizing detection sensitivity locally rather than applying a uniform global threshold. This resolves the contradiction by making the detection system adaptive to local characteristics.
2Reliability
If uniform detection threshold is applied across all regions, then the detection process is simple, but false alarms increase due to varying noise levels
Solution Approach 1:
The detection threshold is made dynamic rather than static. The system calculates noise characteristics for each region and automatically adjusts the detection threshold based on local noise levels. This dynamic adaptation reduces false alarms in high-noise regions while maintaining sensitivity in low-noise regions, improving reliability without significantly impacting throughput.
Solution Approach 2:
The patent changes the detection parameter (threshold) based on the noise level parameter. By establishing a relationship between noise level and threshold value, the system adapts its detection sensitivity to match local conditions. This parameter change strategy resolves the contradiction by making the detection process responsive to varying noise characteristics across different specimen regions.
3Measurement precision
If high-resolution scanning is performed across the entire specimen, then defect detection sensitivity is improved, but inspection time increases
Solution Approach 1:
The specimen is segmented into regions of interest and non-critical regions. High-resolution scanning is applied only to ROIs where defects are most likely to occur or where critical features exist, while other areas receive lower-resolution scanning. This selective resolution strategy maintains detection sensitivity for critical defects while reducing overall inspection time.
Solution Approach 2:
The system applies high-resolution inspection partially rather than universally. By identifying and focusing computational resources on critical ROIs, the system achieves sufficient detection sensitivity for the most important areas without the time penalty of scanning the entire specimen at maximum resolution. This partial action resolves the time-sensitivity contradiction.
Data Source
AI summary
There are provided a system and method of defect detection on a specimen, the method comprising: performing partitioning for each of one or more portions of a first die; receiving one or more noise maps indicative of noise distribution on second images captured for one or more portions of a second die; performing segmentation for each noise map in runtime, the segmentation for a given noise map including: calculating a score for each region, the given noise map aligned with the regions and each region is associated with noise data aligned therein, the score for a given region calculated at least based on the noise data associated therewith; and associating each region with one segmentation label of a predefined set of segmentation labels indicative of noise levels based on the score, thereby obtaining a set of segments each corresponding to one or more regions associated with the same segmentation label.