Model-Based Critical Dimension Measurement for Reticle Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reticle inspection techniques are inadequate for accurately measuring critical dimensions due to sensitivity to errors in intensity scale and aberrations in the imaging system, and fail to accurately account for the interaction of illumination with the reticle, leading to unacceptable levels of CD measurement accuracy.
Innovation Solution
A method involving an imaging system that generates a calculated image based on a design database and a computational model of the imaging system, with iterative adjustment of critical dimension (CD) and uncertain parameters to minimize the difference between measured and calculated images, using multiple views and settings such as focus offset, pupil distribution, and polarization state, to achieve accurate CD measurement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional reticle inspection techniques are used, then inspection speed and simplicity are maintained, but measurement precision of critical dimensions deteriorates due to sensitivity to intensity scale errors and imaging system aberrations
Solution Approach 1:
The patent introduces a computational model as an intermediary between the imaging system and measurement results. This model accounts for illumination-reticle interactions and imaging system characteristics, serving as a mediator that corrects for intensity scale errors and aberrations without requiring physical modification of the imaging system. The computational model translates raw images into accurate CD measurements by compensating for system imperfections.
Solution Approach 2:
The patent changes the measurement approach from direct geometric measurement to model-based parameter estimation. By adjusting the computational model parameters (illumination conditions, reticle properties, imaging system characteristics) and comparing simulated images with actual images, the system achieves accurate CD measurements that are insensitive to intensity scale errors and imaging aberrations.
2Measurement precision
If conventional image-based measurement methods are used, then measurement process simplicity is maintained, but measurement precision deteriorates due to failure to account for illumination-reticle interactions
Solution Approach 1:
The patent performs preliminary computational modeling of illumination-reticle interactions before actual measurement. The computational model pre-calculates expected image characteristics based on known reticle patterns and illumination conditions, allowing rapid comparison with actual images during measurement. This preliminary action enables accurate CD extraction without time-consuming iterative adjustments during the measurement process itself.
Solution Approach 2:
The patent creates computational copies (simulated images) of the reticle patterns under various illumination conditions and compares these copies with actual measured images. By generating multiple simulated images with different CD values and selecting the best match, the system achieves accurate measurements while maintaining efficient processing through algorithmic comparison rather than physical measurement iteration.
3Measurement precision
If simple threshold-based CD measurement is used, then ease of operation is maintained, but measurement precision deteriorates due to sensitivity to intensity scale errors
Solution Approach 1:
The computational model serves as an intermediary that eliminates sensitivity to intensity scale errors. Instead of directly measuring feature dimensions from intensity images (which is sensitive to scaling), the model compares entire image patterns including intensity distributions. This intermediary approach automatically compensates for intensity scale variations without requiring manual calibration or complex user intervention.
Solution Approach 2:
The patent implements an iterative feedback process where the computational model generates simulated images, compares them with actual images, and adjusts CD parameters to minimize differences. This automatic feedback loop continuously refines the measurement until convergence, eliminating the need for manual threshold adjustment and providing robust measurements that are insensitive to intensity scale variations.
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 provides improved accuracy in CD measurement by iteratively fitting calculated images to measured images, reducing measurement uncertainty and enhancing the detection of defects, thereby improving reticle inspection and wafer yield.
Implementation Method 1
the measured image comprises a plurality of measured images of the structure obtained for a plurality of views for different operating parameters of the inspection tool... the views comprise reflected and transmitted light detection
Implementation Method 2
the views comprise reflected and transmitted light detection
Implementation Method 3
The model generates the calculated image based on optical properties of reticle materials of the structure, a computational model of the imaging system, and an adjustable CD
Implementation Method 4
the views comprise different settings for one or more of the following: focus offset, a pupil distribution of the illumination, a polarization state of the illumination
Data Source
AI summary
A reticle is inspected with an imaging system to obtain a measured image of a structure on the reticle, and the structure has an unknown critical dimension (CD). Using a model, a calculated image is generated using a design database that describes a pattern used to form the structure on the reticle. The model generates the calculated image based on: optical properties of reticle materials of the structure, a computational model of the imaging system, and an adjustable CD. A norm of a difference between the measured and calculated images is minimized by adjusting the adjustable CD and iteratively repeating the operation of generating a calculated image so as to obtain a final CD for the unknown CD of the structure. Minimizing the norm of the difference is performed simultaneously with respect to the adjustable CD and one or more uncertain parameters of the imaging system.