Semiconductor Layer Thickness Detection via Tilt Correction and Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for measuring the thickness of semiconductor layers in fabrication processes are prone to high error rates and subjectivity due to down-sampling and manual data selection, leading to inconsistent results and low sample sizes.
Innovation Solution
A method and system that detect the tilt of an image of a stack of layers, perform rough boundary line detection, and fine thickness detection using a processor, allowing for precise thickness measurement without down-sampling and reducing user variability, thereby increasing data points and measurement accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If down-sampling is performed on the image, then the measurement process is simplified, but the measurement precision deteriorates due to loss of fine detail
Solution Approach 1:
The patent applies preliminary action by performing tilt detection and correction on the original high-resolution image before boundary line detection. This preliminary step prepares the image data in advance, allowing subsequent measurements to be performed on the full-resolution image without down-sampling, thereby maintaining measurement precision while simplifying the overall process through automated preprocessing.
2Adaptability or versatility
If manual data selection is performed, then the measurement process is flexible, but the reliability deteriorates due to subjectivity and inconsistency
Solution Approach 1:
The patent implements self-service through automated algorithms that perform tilt detection, boundary line detection, and thickness measurement without requiring manual data selection. The system automatically processes the image data through multiple detection stages, eliminating human subjectivity and inconsistency while maintaining flexibility through configurable measurement parameters and automated quality control.
3Measurement precision
If rough boundary line detection is performed first, then the fine thickness detection is facilitated, but the overall process time increases
Solution Approach 1:
The patent applies segmentation by dividing the measurement process into distinct stages: tilt detection, rough boundary line detection, and fine thickness detection. Each stage processes specific aspects of the image data independently, allowing for optimized computation at each step. The rough boundary detection provides initial guidance that accelerates the subsequent fine detection phase, reducing overall processing time while maintaining high precision.
4Measurement precision
If tilt detection is performed on the image, then the measurement accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent replaces mechanical alignment procedures with computational tilt detection and correction algorithms. Instead of physically adjusting the sample or imaging system to achieve proper orientation, the system uses image processing algorithms to detect and correct tilt automatically, maintaining measurement accuracy while reducing mechanical complexity.
Data Source
AI summary
Embodiments related to systems and methods for thickness measurement in semiconductor structures are disclosed. For example, a method for thickness detection in a semiconductor structure may include detecting, by at least one processor, a tilt of an image of a stack of layers in the semiconductor structure. The method may also include performing, by the at least one processor, rough boundary line detection on the layers of the stack in the image. The method may further include performing, by the at least one processor, fine thickness detection on the layers of the stack in the image. The rough boundary line detection may detect boundaries of the layers of the stack with a first precision and the fine thickness detection may detect thickness of the layers of the stack with a second precision greater than the first precision. The method may additionally include providing, by the at least one processor, output results of the fine thickness detection.