SEM Metrology Data Conversion Using ML Tool Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, multiple metrology systems operating on similar principles often produce inconsistent measurement results due to differences in mechanical, electrical, and software components, leading to variations in measurements of patterned substrates across different tools.
Innovation Solution
A machine learning model is trained to convert metrology data from one scanning electron microscope (SEM) system to match the characteristics of another, using a cost function that adjusts parameters to align signal and critical dimension (CD) measurements, enabling consistent measurements across different metrology systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple metrology systems are used to improve measurement throughput and accuracy, then productivity and measurement quality improve, but measurement consistency deteriorates due to differences in mechanical, electrical, and software components across different tools
Solution Approach 1:
The patent transforms metrology data from different systems by changing its parameters (signal characteristics, measurement values) to match a reference system. The conversion model adjusts parameters such as signal intensity, critical dimension measurements, and other system-specific parameters to ensure consistency across different metrology tools while maintaining their individual operational characteristics
Solution Approach 2:
The patent introduces a conversion model as an intermediary between different metrology systems and the reference system. This intermediary component receives data from various metrology tools, processes it through learned transformation relationships, and outputs converted data that is consistent with the reference system, thereby mediating the inconsistency between multiple systems
2Measurement precision
If metrology data is converted using a machine learning model to ensure consistency across different systems, then measurement consistency improves, but device complexity increases due to the need for training and applying the conversion model
Solution Approach 1:
The patent performs the complex task of learning transformation relationships in advance during a training phase. The conversion model is trained offline using paired data from different metrology systems, capturing the systematic differences between systems before actual measurement operations. This preliminary action separates the complexity of model training from the operational phase, making the conversion process straightforward during production use
Solution Approach 2:
The patent creates a virtual copy or representation of the reference metrology system's data characteristics. The conversion model learns to reproduce the signal patterns, measurement values, and overall data structure of the reference system from input data of other systems. This copying approach allows consistent data format without requiring physical modification of the metrology systems
Data Source
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.