Dynamic Threshold Reticle Defect Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reticle defect inspection methods based on fixed particle size judgments are prone to errors, leading to inaccurate and untimely warnings, resulting in increased reprocessing rates and production costs due to measurement errors in semiconductor photolithography processes.
Innovation Solution
A reticle defect inspection system that continuously monitors reticles for defects, calculates a dynamic threshold based on historical defect information and measurement errors, and performs early-warning processing when the dynamic threshold exceeds unacceptable limits, thereby reducing reprocessing rates and production costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed particle size judgment is used for defect inspection, then inspection process is simple, but measurement precision deteriorates due to measurement errors
Solution Approach 1:
The patent transforms the static fixed threshold into a dynamic threshold that adapts based on historical defect data and measurement error characteristics. The system continuously updates the threshold value to reflect actual measurement conditions, making the inspection criterion flexible rather than rigid. This resolves the contradiction by maintaining measurement precision through adaptation while keeping the inspection process relatively simple through automated threshold adjustment.
Solution Approach 2:
The patent changes the inspection parameter from a fixed particle size threshold to a dynamic threshold that varies based on statistical analysis of historical data and measurement errors. By modifying the threshold parameter dynamically according to actual measurement conditions, the system maintains high measurement precision without significantly increasing process complexity, as the parameter adjustment is performed automatically through statistical calculations.
2Loss of time
If fixed threshold inspection is used, then early-warning timing is delayed, but system complexity is reduced
Solution Approach 1:
The patent performs preliminary statistical analysis of historical defect data and measurement errors to establish a dynamic threshold before actual inspection begins. This preliminary characterization of measurement conditions enables more accurate real-time warnings without requiring complex real-time adjustments during inspection. The system proactively prepares the adaptive threshold, improving warning timing while limiting complexity to the data collection and initial analysis phase.
Solution Approach 2:
The patent implements a feedback mechanism where historical defect inspection results and measurement error data are continuously fed back to adjust the threshold value. This feedback loop enables the system to learn from past measurements and improve warning timing accuracy over time. The automated feedback process maintains precision without proportionally increasing system complexity, as the feedback is processed through established statistical algorithms.
3Measurement precision
If dynamic threshold calculation is implemented, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent enables the inspection system to self-adjust its threshold through automated statistical analysis of its own historical data and measurement errors. The system serves itself by using its accumulated operational data to refine its own inspection criteria without requiring external intervention or complex manual calibration. This self-service approach improves measurement precision while limiting complexity growth, as the system autonomously manages its own parameter optimization.
4Productivity
If continuous defect monitoring is performed, then productivity is improved through timely warnings, but loss of time increases due to data processing
Solution Approach 1:
The patent performs preliminary statistical analysis of historical defect data to establish baseline characteristics and measurement error patterns before continuous monitoring begins. This pre-characterization enables the system to process real-time data more efficiently by comparing against pre-computed reference values rather than performing full statistical analysis on each new measurement. This approach improves productivity through timely warnings while minimizing additional data processing time.
Data Source
AI summary
Provided are a reticle defect inspection method and system. The reticle defect inspection method includes: a reticle is provided; a reticle defect inspection system is provided, and when the reticle is placed on a station or leaves the station, defect inspection is continuously performed on the reticle to obtain defect information of each defect; a dynamic threshold of each defect is obtained from the defect information of each defect; and whether the dynamic threshold of each defect belongs to a threshold unacceptable by the inspection system is judged, and if so, warning processing is performed.