Line Edge Roughness Measurement Noise Separation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring line edge roughness (LER) using scanning electron microscopes face challenges in accurately separating true LER from measurement noise, as their spatial frequency characteristics are mixed, leading to deteriorated measurement accuracy.
Innovation Solution
A method involving multiple frame scans, integration of signals, regression analysis, and spectral component analysis to distinguish and remove noise, allowing for the derivation of a noise-free roughness evaluation value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple frame scans are performed to improve measurement accuracy, then measurement precision is improved, but measurement time increases
Solution Approach 1:
The patent performs preliminary actions by acquiring multiple frame signals before analysis, and uses regression analysis to extrapolate the result that would be obtained with infinite frames. This preliminary data collection combined with mathematical extrapolation reduces the need for excessive actual measurements, thereby improving precision without proportionally increasing measurement time.
Solution Approach 2:
The patent replaces the mechanical approach of continuously acquiring more and more frame data with a mathematical substitution approach. By using regression analysis and spectral decomposition, the system substitutes actual additional measurements with computational methods to achieve the same precision improvement, thereby reducing measurement time.
2Measurement precision
If spectral analysis is performed to separate LER from noise, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the spectral components into different frequency ranges - separating the LER signal components from the noise components based on their distinct spatial frequency characteristics. This segmentation allows precise identification and extraction of the true LER signal while filtering out measurement noise, improving measurement precision through systematic signal decomposition.
Solution Approach 2:
The patent introduces spectral analysis as an intermediary processing step between raw signal acquisition and final LER measurement. This intermediary spectral domain analysis acts as a mediator that transforms the mixed time-domain signal into separable frequency-domain components, enabling clean separation of LER from noise without requiring complex hardware modifications.
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 accurate separation of true LER from measurement noise, improving the measurement accuracy of line edge roughness and enhancing semiconductor device performance and production yield.
Implementation Method 1
a plurality of frame signals are acquired by scanning a beam a plurality of times on a pattern
Data Source
AI summary
The present invention is directed to a pattern measuring method and the like for deriving a roughness evaluation value from which measurement noise having a frequency component is removed. The pattern measuring method includes generating a plurality of integration signals having the different number of integration from the signals of the plurality of frames; acquiring edge position information of the pattern along an edge from the plurality of respective integration signals having the different number of integration; obtaining a difference between the edge position information and reference position information; obtaining a spectral component of a specific spatial frequency of the edge or a standard deviation based upon the difference; performing regression analysis in which a value relating to the number of integration is set as an explanatory variable, and a value relating to the spectral component or the standard deviation is set as a target variable; and outputting a value obtained by the regression analysis as a roughness index value of the pattern or a noise index value included in the signals.