Pattern Measurement Apparatus Noise Exclusion Method
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern measurement techniques fail to accurately measure semiconductor patterns when noises or dirt are present in the periphery, leading to false cursor location corrections and potential measurement failures, as they do not account for such interferences.
Innovation Solution
A pattern measuring apparatus and method that sets a measurement region and excludes overlapping areas with a predetermined region not to be measured, ensuring precise measurement by isolating noise and dirt from the measurement process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the measurement region is expanded to accommodate process fluctuations, then measurement robustness is improved, but noise and dirt from peripheral regions interfere with measurement accuracy
Solution Approach 1:
The image data is divided into multiple measurement regions, where each region is independently evaluated for noise and dirt content. This segmentation allows the system to selectively process only the clean portions of the pattern, maintaining measurement robustness while avoiding interference from contaminated areas.
Solution Approach 2:
The patent applies different processing qualities to different regions of the image data. Regions identified as containing noise or dirt are excluded or down-weighted in the measurement calculation, while clean regions are given full weight. This local quality differentiation ensures that measurement accuracy is maintained even when process fluctuations require expanded measurement regions.
2Measurement precision
If pattern matching is performed across the entire image including peripheral regions, then alignment accuracy is improved, but false corrections occur due to noise and dirt
Solution Approach 1:
The patent extracts and identifies noise and dirt components from the image data through image processing analysis. These unwanted elements are then separated from the valid pattern data, allowing pattern matching to be performed only on the clean, valid regions. This extraction process prevents false corrections while maintaining alignment accuracy.
Solution Approach 2:
Before performing pattern matching and measurement, the patent preliminarily analyzes the image data to identify and mark regions containing noise or dirt. This preliminary action allows the subsequent pattern matching process to automatically exclude these problematic regions, preventing false corrections before they occur.
3Adaptability or versatility
If measuring cursors are corrected based on overall pattern matching results, then measurement adaptability to process fluctuations is improved, but false cursor locations are generated due to peripheral noise
Solution Approach 1:
The patent extracts noise and dirt from the image data before performing cursor correction based on pattern matching. By removing these interfering elements in advance, the pattern matching results reflect only the true pattern variations due to process fluctuations, enabling accurate cursor correction without false location shifts.
Solution Approach 2:
The patent performs preliminary noise and dirt identification and exclusion before the cursor correction step. This preliminary action ensures that when cursors are adjusted to compensate for process fluctuations, the adjustments are based solely on valid pattern data, preventing false cursor locations while maintaining adaptability to real process variations.
Data Source
AI summary
In measuring pattern with large process fluctuation, correct measurement cannot be carried out if noises, such as pattern that is not the subject to be measured, and dirt, are present in periphery of pattern to be measured in previously registered measurement region. Among the image data of sample, predetermined region aligned by pattern matching is set as region not to be measured that is excluded from subjects of pattern measurement. For example, in measuring pattern with large process fluctuation, only region including pattern with small process fluctuation is used in pattern matching, while in measuring the pattern, predetermined region, which was used in pattern matching and aligned, is set as region not to be measured. Stable pattern measurement can be easily carried out with respect to pattern with large process fluctuation, without being affected by region where measurement region and region not to be measured overlap with each other.