Metrology Tool Matching via Virtual Reference Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metrology systems face challenges in matching data measurements across different systems, leading to inconsistencies in overlay values and other physical characteristics during semiconductor manufacturing, particularly due to phase-related differences that are not captured by existing frameworks like the Jones Framework.
Innovation Solution
A method is developed to create a model that predicts physical characteristics of patterned substrates by using training data from one measurement tool and adjusting parameters based on data from another tool, employing techniques such as singular value decomposition and principal component analysis to generate mapping functions that align measurements across different metrology tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data is measured using different metrology systems, then measurement coverage is improved, but measurement consistency deteriorates
Solution Approach 1:
The patent introduces a reference system as an intermediary that mediates between different metrology systems. Measured data from various systems is transformed into the reference system's coordinate framework, enabling consistent comparisons. The reference system acts as a common language that reconciles differences between systems, allowing multiple systems to contribute data while maintaining measurement consistency through the unifying reference framework.
Solution Approach 2:
The patent transforms measurement parameters by applying coordinate transformations and calibration factors to convert data from different metrology systems into a common reference framework. By changing the parameter representation (coordinate systems, scaling factors, offset corrections), the system enables consistent comparison across different measurement tools while preserving the original measurement information.
2Device complexity
If existing frameworks like Jones Framework are used, then computational simplicity is maintained, but phase-related differences are not captured
Solution Approach 1:
The patent extends the measurement framework by adding a phase dimension to the traditional coordinate-based approach. While maintaining the computational structure of existing frameworks, the invention incorporates phase information as an additional measurement dimension, allowing phase-related differences to be captured and analyzed without completely redesigning the computational framework.
Data Source
AI summary
Methods and systems for determining a model configured to predict values of physical characteristics (e.g., overlay, CD) associated with a patterned substrate measured using different measurement tools. The method involves obtaining a first set of measured data for a first one or more patterned substrates using a first one or more measurement tools, and reference measurements of a physical characteristic. Also, a second set of measured data and virtual data for a second one or more patterned substrates measured using a second one or more measurement tools is obtained. A set of mapping functions between the second set and the virtual data are generated. The set of mapping functions is used to convert the first set. Then, a model is determined based on the reference measurements and the converted data such that the model predicts values of the physical characteristic that are within an acceptable threshold of the reference measurements.