Optical Metrology Library Generation Sampling Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional library generation methods in optical metrology waste significant computation power by simulating and training samples in regions outside the customer's Region of Interest (ROI), leading to inefficient sampling and library generation.
Innovation Solution
Automatically recommending a sampling shape or filter based on expected sample space usage and parameter correlation to reduce the sampling space, thereby improving efficiency and reducing unnecessary computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional library generation methods use a hypercube sampling range based on minimum and maximum of each parameter, then the sampling process covers all possible parameter regions, but a large amount of computation power is wasted in unneeded parameter regions outside the Region of Interest
Solution Approach 1:
The patent extracts and removes the unneeded parameter regions outside the Region of Interest from the sampling space. By identifying and eliminating hypercube regions that fall outside the customer's ROI, the method retains only the necessary sampling regions, thereby reducing computation power waste while maintaining coverage of all relevant parameter regions.
Solution Approach 2:
The patent applies different sampling strategies to different regions of the parameter space. Instead of uniform sampling across the entire hypercube, the method concentrates sampling efforts in the Region of Interest while reducing or eliminating sampling in regions outside ROI, thereby optimizing computation power allocation to where it is most needed.
2Manufacturing precision
If the sampling process includes all regions within the hypercube bound by minimum and maximum parameters, then no parameter region is missed, but the library generation time increases due to sampling in unneeded regions
Solution Approach 1:
The patent extracts and removes unneeded parameter regions from the sampling process. By identifying regions outside the customer's Region of Interest and excluding them from sampling, the method maintains accurate coverage of all relevant parameters while significantly reducing the time required for library generation.
Solution Approach 2:
The patent applies partial sampling action by focusing computational resources only on the necessary portion of the parameter space (the Region of Interest). Instead of performing excessive sampling across the entire hypercube, the method performs sampling only where needed, thereby reducing library generation time while maintaining sampling accuracy for relevant parameters.
3Productivity
If automatic recommendation of sampling shape and filter based on parameter correlation is implemented, then the sampling space is reduced and efficiency is improved, but the device complexity increases
Solution Approach 1:
The patent implements a self-service system where the sampling optimization process automatically analyzes parameter correlations and recommends optimal sampling shapes and filters without requiring manual intervention. The system uses the customer's own process control knowledge and parameter correlation data to autonomously determine the most efficient sampling strategy, thereby improving productivity while managing complexity through automation.
Solution Approach 2:
The patent changes the sampling parameters dynamically based on parameter correlations. By automatically adjusting sampling shapes and filters according to the correlations between parameters, the system optimizes the sampling space reduction while managing complexity through algorithmic parameter adaptation rather than fixed complex structures.
Data Source
AI summary
Embodiments include automatic selection of sample values for optical metrology. An embodiment of a method includes providing a library parameter space for modeling of a diffracting structure using an optical metrology system; automatically determining by a processing unit a reduced sampling set from the library parameter space, wherein the reduced space is based on one or both of the following recommending a sampling shape based on an expected sample space usage, or recommending a sampling filter based on correlation between two or more parameters of the library parameter space; and generating a library for the optical metrology system using the reduced sampling set.