Inverse Linescan Model for Unbiased Roughness Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring roughness in pattern structures, particularly in noisy images from scanning electron microscopes, suffer from bias due to measurement noise, leading to inaccurate and unreliable results that affect the quality of manufacturing processes and device performance.
Innovation Solution
A system and method that utilize an inverse linescan model to detect edges in pattern structures without filtering, generating a biased power spectral density dataset, evaluating high-frequency noise, and subtracting predicted noise to obtain an unbiased roughness measure, thereby filtering out artifacts and revealing true feature roughness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If noise filtering is applied to images before roughness measurement, then measurement reliability is improved, but measurement precision deteriorates due to loss of high-frequency information
Solution Approach 1:
The PSD dataset is segmented into multiple frequency portions (low-frequency, mid-frequency, high-frequency ranges). This segmentation allows selective processing where high-frequency noise is identified and removed while preserving the low-frequency and mid-frequency components that contain the actual roughness information, thus resolving the contradiction between reliability and precision.
Solution Approach 2:
Different processing approaches are applied to different frequency portions of the PSD dataset. The high-frequency portion undergoes noise evaluation and removal, while the low-frequency and mid-frequency portions are preserved with minimal processing. This local quality approach ensures that noise is removed without compromising the precision of the actual roughness measurements.
2Reliability
If images are filtered to remove noise, then measurement reliability is improved, but manufacturing precision deteriorates due to loss of feature detail
Solution Approach 1:
The frequency spectrum of the measured linescan information is segmented into distinct portions. By identifying and removing only the high-frequency noise components while preserving the mid-frequency feature information, the method achieves reliable noise removal without compromising the precision of pattern structure measurements.
Solution Approach 2:
The method converts the harmful effect of noise in the high-frequency portion into a benefit by using it to identify and remove artifacts. The high-frequency noise, which initially degrades measurement quality, is utilized to detect and eliminate measurement artifacts, thereby improving both reliability and precision simultaneously.
3Measurement precision
If complex noise removal algorithms are applied, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The complex task of noise removal is simplified by segmenting the PSD dataset into frequency portions and applying targeted processing to each. This segmentation transforms a complex global processing problem into simpler, more manageable local processing tasks, reducing overall system complexity while maintaining measurement precision.
Solution Approach 2:
Instead of applying complex processing to the entire frequency spectrum, the method applies partial action by focusing noise removal efforts only on the high-frequency portion where noise dominates. This selective processing reduces computational complexity while achieving the desired measurement precision.
Data Source
AI summary
A method is disclosed. The method includes receiving measured linescan information describing a pattern structure of a feature, applying the received measured linescan information to an inverse linescan model that relates measured linescan information to feature geometry information, identifying, based at least in part on the applying the received measured linescan model to the inverse linescan model, feature geometry information that describes a feature that would produce a linescan corresponding to the received measured linescan information, determining, at least in part using the inverse linescan model, feature edge positions of the identified feature, and analyzing the feature edge positions to detect the presence or absence of defects in the pattern structure.