SEM Metrology Data Conversion for Cross-Tool CD Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Different metrology systems produce inconsistent measurement results for patterned substrates due to variations in mechanical, electrical, and algorithmic differences, leading to discrepancies in critical dimension (CD) measurements.
Innovation Solution
A machine learning model is trained to convert metrology data, such as SEM images, from one system to match the characteristics of another system, using a cost function that adjusts parameters to minimize differences in CD measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple metrology systems are used to improve measurement throughput and accuracy, then productivity and measurement precision are improved, but measurement consistency deteriorates due to tool variations in mechanical, electrical, and algorithmic components
Solution Approach 1:
The patent introduces a reference metrology system as an intermediary standard. All other metrology systems are calibrated and adjusted to match the reference system's measurements. This mediator (reference system) enables consistent measurements across multiple tools by providing a common baseline that all systems align to, thereby resolving the inconsistency caused by tool variations.
Solution Approach 2:
The patent adjusts measurement parameters and settings of non-reference metrology systems to match the reference system. By changing parameters such as measurement thresholds, signal processing settings, and calibration factors, the system aligns the measurement outputs of different tools, ensuring that the same pattern produces consistent measurements across all metrology systems.
2Measurement precision
If metrology data is converted using a trained machine learning model to match reference system characteristics, then measurement consistency is improved, but device complexity increases due to the ML model infrastructure
Solution Approach 1:
The patent replaces traditional mechanical and algorithmic adjustment methods with a machine learning model. Instead of manually calibrating each metrology system to match the reference, an ML model automatically learns the transformation needed to convert measurements from any system to match the reference system's characteristics. This substitution of ML for manual calibration reduces operational complexity while maintaining measurement consistency.
Solution Approach 2:
The patent creates a virtual copy of the reference metrology system's measurement characteristics through the ML model. The model learns to replicate the reference system's measurement behavior by training on paired data from the reference and target systems. This copying approach allows consistent measurements without physically modifying or recalibrating each tool to match the reference.
3Measurement precision
If a reference metrology system is established to ensure measurement consistency, then measurement precision is improved, but productivity may deteriorate due to additional calibration and conversion steps
Solution Approach 1:
The patent performs preliminary training of the machine learning model using measurement data from the reference and target metrology systems. This preliminary action (offline training) captures the transformation relationships between systems before actual measurements are taken. During production, the pre-trained model quickly converts measurements without requiring real-time calibration, thus maintaining productivity while ensuring consistency.
Solution Approach 2:
The patent replaces time-consuming manual calibration procedures with automated machine learning-based conversion. The ML model performs rapid mathematical transformations of measurement data, substituting for lengthy mechanical adjustment and recalibration processes. This automation maintains measurement consistency while significantly reducing the time required for calibration and data harmonization.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Described herein is a metrology system and a method for converting metrology data via a trained machine learning (ML) model. The method includes accessing a first (MD1) SEM data set (e.g., images, contours, etc.) acquired by a first scanning electron metrology (SEM) system (TS1) and a second (MD2) SEM data set acquired by a second SEM system (TS2), where the first SEM data set and the second SEM data set being associated with a patterned substrate. Using the first SEM data set and the second SEM data set as training data, a machine learning (ML) model is trained (P303) such that the trained ML model is configured to convert (P307) a metrology data set (310) acquired (P305) by the second SEM system to a converted data set (311) having characteristics comparable to metrology data being acquired by the first SEM system. Furthermore, measurements may be determined based on the converted SEM data.