Charged Particle Beam Imaging for Tilt Marker Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current charged particle beam apparatuses require significant time and effort for human-operated field-of-view search and imaging, especially when observing sample cross sections, which hampers the efficiency of semiconductor fabrication and other material analysis processes.
Innovation Solution
The development of a charged particle beam apparatus that automatically recognizes marker patterns in tilt images using machine learning models or template matching, allowing for automated field-of-view search and cross-section observation, thereby reducing the time and effort required for these processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human operators manually search for field of view and perform imaging operations, then observation accuracy can be maintained, but time consumption and operational effort increase significantly
Solution Approach 1:
The system enables automatic field-of-view search and imaging operations through machine learning models that autonomously identify marker patterns and determine observation positions, eliminating the need for manual operator intervention while maintaining observation accuracy
Solution Approach 2:
Manual visual search and operational control are replaced by automated image processing systems using machine learning algorithms, specifically employing template matching or neural network-based classifiers to detect marker patterns and calculate observation positions automatically
2Ease of operation
If automated field-of-view search is implemented using conventional pattern matching on top-view images, then operational effort is reduced, but detection reliability fails when marker patterns are not visible or positional relationships cannot be grasped
Solution Approach 1:
The system transitions from two-dimensional top-view image analysis to three-dimensional cross-sectional image analysis, enabling detection of marker patterns that are not visible or identifiable in planar views by utilizing depth information and structural characteristics in the cross-sectional dimension
3Productivity
If large amounts of observation data are acquired rapidly through automation, then productivity increases, but the complexity of automated recognition systems increases
Solution Approach 1:
The system uses template matching where a reference marker pattern template is created and compared against multiple images to identify marker positions, or employs pre-trained machine learning models that have learned marker characteristics during an offline training phase, enabling rapid automated detection without complex real-time processing
Solution Approach 2:
Marker pattern templates are prepared in advance during an offline training phase using sample images, and machine learning models are pre-trained with labeled data before deployment, allowing the automated system to rapidly process observation images without performing complex learning during actual data acquisition
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 solution significantly reduces the time and effort needed for field-of-view search and automated cross-section imaging, enhancing the speed and efficiency of semiconductor fabrication and other material analysis processes.
Implementation Method 1
an imaging device that acquires image data of a sample at a given magnification by radiating a charged particle beam to the sample
Data Source
AI summary
An imaging device is provided with a sample stage configured to be capable of transferring a sample by at least two drive shafts, and shifting an imaging field of view corresponding to positional information of the sample, the positional information being obtained by a computer system. The computer system includes a classifier which outputs, in response to input of image data of a tilt image where the sample is imaged in an inclined state with respect to a charged particle beam, positional information of one or more feature parts existing on the tilt image. The classifier is trained in advance by using training data in which input is image data of the tilt image, and output is positional information of the feature part. The computer system executes, with respect to new tilt image data inputted into the classifier, a processing to output positional information of the feature part.