Wavelet-Based Edge Detection for Semiconductor Pattern Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pattern inspection apparatuses face errors in detecting pattern edges due to profile changes caused by noise, charging, and focus deviations, especially when using fixed template and edge filters, which affects the accuracy of defect detection in ultrafine patterns on semiconductor wafers and exposure masks.
Innovation Solution
The apparatus performs a wavelet transform on dimensional profiles with varying scale variables and uses multiple mother wavelet functions to extract contour positions, which are then compared to reference contours to reduce edge position errors and improve detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If fixed template and edge filter are used to extract pattern edges, then the inspection process is simple, but edge position detection accuracy deteriorates due to profile changes from noise, charging, and focus deviation
Solution Approach 1:
The patent applies dynamics by making the template and edge filter adjustable rather than fixed. The template shape, size, and position can be dynamically modified based on actual pattern profiles, and the edge filter parameters can be adaptively changed to match varying pattern conditions. This allows the inspection system to adapt to profile changes caused by noise, charging, and focus deviation while maintaining operational simplicity through automated parameter adjustment.
Solution Approach 2:
The patent implements parameter changes by allowing modification of template parameters (shape, size, position) and edge filter parameters based on detected pattern characteristics. When profile changes are detected in the inspected patterns, the system automatically adjusts these parameters to optimize edge extraction accuracy, thereby resolving the contradiction between simple operation and precise measurement.
2Measurement precision
If layout data is required to extract pattern edges, then edge extraction can be performed, but the inspection system becomes more complex and less adaptable to profile changes
Solution Approach 1:
The patent applies self-service by enabling the inspection system to automatically extract pattern edges without requiring external layout data. The system uses its own captured images and applied templates to identify and extract edges through iterative optimization. This self-contained approach reduces system complexity while maintaining edge extraction capability by making the system independent of additional data sources.
Solution Approach 2:
The patent implements preliminary action by pre-loading adjustable template patterns that can be configured before inspection. These templates are prepared in advance with modifiable parameters that can be quickly adjusted during inspection based on detected profile changes, enabling edge extraction without requiring complex real-time processing or external layout information.
3Stability of the object's composition
If template and edge filter parameters are fixed, then the inspection process is stable, but accuracy deteriorates when image profiles change due to noise, charging, and focus deviation
Solution Approach 1:
The patent resolves this contradiction by implementing parameter changes that allow the template and edge filter to adapt to varying image conditions. The system monitors profile changes caused by noise, charging, and focus deviation, and automatically adjusts template parameters (shape, size, position) and filter characteristics accordingly. This maintains inspection process stability through automated control while improving measurement precision under varying conditions.
Solution Approach 2:
The patent applies feedback by continuously monitoring the inspection results and comparing extracted edges with expected pattern characteristics. When deviations are detected due to profile changes, the system feeds this information back to automatically adjust template and filter parameters, thereby maintaining both process stability and measurement accuracy across varying inspection conditions.
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 effectively reduces errors in edge position detection even with image changes, enhancing the accuracy of pattern inspection by adapting to variations in pattern profiles and noise, thereby improving the yield in LSI manufacturing.
Implementation Method 1
acquire an image of a substrate on which a figure pattern has been formed
Data Source
AI summary
According to one aspect of the present invention, a pattern inspection apparatus includes: a profile extraction circuit configured to extract each of a plurality of predetermined dimensional profiles for a plurality of pixels with a value of a differential intensity greater than or equal to a threshold value in the image; a wavelet transform circuit configured to perform, on each of the plurality of predetermined dimensional profiles, a wavelet transform while changing a scale variable of a mother wavelet function to a predetermined value; and a contour position extraction circuit configured to extract, for the each of the plurality of predetermined dimensional profiles, a maximum peak position as a contour position of the figure pattern from peak positions of a plurality of transformed profiles of after the wavelet transform in which the scale variable is set.