3D Volume Image Reconstruction Accuracy via Feature-Based Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for 3D reconstruction of integrated semiconductor samples face challenges in achieving high precision due to limitations in fiducial-based alignment, such as coarse fiducial markers and measurement inaccuracies, which lead to artificially enhanced line edge roughness and positional errors, especially with feature sizes below 10 nm.
Innovation Solution
The method employs feature-based alignment using accurately known inner structures of the semiconductor sample, such as metal lines and HAR structures, for precise image registration and correction, combining statistical methods like centroid extraction and averaging to achieve subpixel accuracy and reduce errors from stage drift and imaging distortions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If fiducial-based alignment is used for 3D reconstruction, then the alignment process is simple and straightforward, but the measurement precision deteriorates due to coarse fiducial markers and measurement inaccuracies
Solution Approach 1:
The patent extracts and removes the fiducial markers from the alignment process, replacing them with direct feature-based registration using the semiconductor sample's own structural features (metal lines, HAR structures). This eliminates the intermediate fiducial layer that limited precision, allowing the system to achieve sub-10nm accuracy by working directly with the sample features themselves.
Solution Approach 2:
The patent introduces statistical methods (centroid extraction, averaging) as an intermediary processing layer between image acquisition and alignment. This statistical intermediary compensates for measurement noise and drift by aggregating information across multiple images and features, thereby improving overall measurement precision without complicating the fundamental alignment approach.
2Productivity
If feature sizes are reduced below 10 nm to increase circuit density, then the productivity and integration density improve, but the measurement precision deteriorates due to limitations in current imaging and alignment techniques
Solution Approach 1:
The patent replaces traditional mechanical/physical alignment methods (fiducial-based registration) with a computational/image-processing-based approach. By using statistical analysis of feature positions across multiple images, the system achieves precision beyond the physical limitations of fiducial markers, enabling accurate measurement of sub-10nm features that would otherwise be impossible to resolve.
Solution Approach 2:
The patent transitions from 2D fiducial marker positioning to 3D feature-based registration by utilizing the vertical stacking of metal layers and HAR structures. This dimensional approach allows the system to leverage the known geometric relationships between layers to improve measurement accuracy, using the third dimension (layer depth) as an additional reference framework.
3Speed
If conventional image registration methods are used, then the processing speed is fast and simple, but the manufacturing precision deteriorates due to stage drift and imaging distortions
Solution Approach 1:
The patent performs preliminary statistical analysis and feature extraction during the image acquisition phase, preparing centroid data and feature positions in advance. This preliminary processing allows the actual alignment computation to proceed quickly while already having pre-processed data that compensates for drift and distortion, achieving both speed and precision.
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 approach enables high-precision 3D reconstruction with improved line edge roughness and surface roughness measurements, reducing artificial misplacements and enhancing measurement accuracy by a factor of 1.5 to 3 compared to fiducial-based methods, suitable for feature sizes down to 7 nm or below.
Implementation Method 1
subsequently removing a cross section surface layer of the integrated semiconductor sample with a focused ion beam
Implementation Method 2
imaging the new cross section of the integrated semiconductor sample with an imaging device
Data Source
AI summary
A three-dimensional circuit pattern inspection technique includes cross sectioning integrated circuits for obtaining a 3D volume image of an integrated semiconductor sample. The method employs a feature based alignment of cross section images based on features of an integrated semiconductor sample. A computer program product and apparatus are provided.