Pattern Inspection Using Local Filter Coefficients for Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pattern inspection methods for semiconductor manufacturing face challenges in accurately detecting defects on masks due to the use of unsuitable filter function coefficients, leading to false defect identification and reduced inspection accuracy as the dimensions of LSI patterns shrink to nanometers.
Innovation Solution
A pattern inspection method that calculates and uses specific filter function coefficients for each small region of a substrate, determining whether a difference in pixel values between optical and developed images exceeds a threshold to decide on the appropriate filter coefficient for generating a reference image, thereby improving inspection accuracy by avoiding unsuitable coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single filter function coefficient is used for the entire inspection region, then the device complexity is reduced, but the measurement precision deteriorates due to false defect identification in regions with different optical characteristics
Solution Approach 1:
The inspection region is divided into multiple small regions, and a separate filter function coefficient is calculated and applied to each small region. This segmentation allows the optical characteristics to be accurately adapted to each local area, preventing false defect identification while maintaining manageable device complexity through systematic regional processing.
Solution Approach 2:
Different filter function coefficients are assigned to different small regions based on their specific optical characteristics. This local quality approach ensures that each region is inspected with a coefficient optimized for its unique properties, thereby improving measurement precision without requiring a single overly complex global coefficient.
2Measurement precision
If the inspection region is divided into multiple small regions with separate filter coefficients, then the measurement precision is improved, but the device complexity and processing time increase
Solution Approach 1:
Filter function coefficients are pre-calculated for each small region before the actual defect inspection process. This preliminary action allows the complex coefficient calculation to be performed in advance, so that during the actual inspection, only the pre-computed coefficients need to be applied, reducing the processing time and operational complexity during the main inspection workflow.
3Ease of manufacture
If a filter coefficient calculated from a region with defects is used, then the calculation process is simplified, but the reliability deteriorates due to inclusion of error in the calculated coefficient
Solution Approach 1:
The system performs a preliminary inspection to identify regions without defects, and uses only those defect-free regions for calculating the filter function coefficient. This feedback mechanism ensures that the calculated coefficient is not contaminated by defect-related errors, thereby maintaining high reliability. The process automatically selects appropriate reference regions based on the inspection results.
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 enhances the precision of pattern defect inspection by accurately determining suitable filter coefficients, reducing false defect identification and improving the overall accuracy of the inspection process.
Implementation Method 1
a magnification optical system for imaging a pattern on a substrate at a predetermined magnification
Implementation Method 2
a pattern writing apparatus capable of writing or 'drawing' fine circuit patterns by using electron beams
Data Source
AI summary
A pattern inspection method includes calculating a first coefficient of a filter function by using data of optical images of plural small regions selected and data of developed images of the plural small regions based on design data; calculating a second coefficient of a filter function by using data of an optical image of a reference small region selected and data of a developed image of the reference small region selected; and determining, for each pixel, whether there exists a pixel for which a difference, between a first temporary reference image to be compared with the optical image of the reference small region generated using the filter function in which the first coefficient is defined and a second temporary reference image to be compared with the optical image of the reference small region generated using the filter function in which the second coefficient is defined, is larger than a threshold.