Scatterometry Model Selection via Virtual Reference Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current model-based measurement techniques, such as scatterometry, are dependent on the selected model and require costly and time-consuming reference data for accurate results, leading to delays and increased complexity in process development, especially with shrinking dimensions and complex processes.
Innovation Solution
A method and system that predicts virtual reference data from measured optical data using a flexible and adaptive model, allowing for optimal model selection without the need for extensive reference metrology, by generating theoretical responses and creating a reference predictor to interpret measured data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If model-based scatterometry techniques are used for precise measurement, then measurement precision is improved, but device complexity increases due to model selection requirements
Solution Approach 1:
The patent creates virtual reference data by copying and simulating the measurement process through computational models. Instead of requiring physical reference samples and complex reference measurement systems, the invention generates synthetic reference data that mirrors real measurements, thereby reducing device complexity while maintaining measurement precision through accurate virtual replicas of the measurement process
Solution Approach 2:
The patent replaces the mechanical/physical reference measurement system with a computational model-based system. Instead of using physical reference standards and complex measurement apparatus, the invention substitutes these with algorithms that generate virtual reference data, eliminating the need for extensive physical referencing infrastructure while preserving measurement accuracy
2Reliability
If extensive reference data is collected for model qualification, then reliability is improved, but loss of time increases due to time-consuming qualification processes
Solution Approach 1:
The patent performs preliminary action by generating virtual reference data in advance through computational modeling. This pre-computed reference data can be stored and reused for multiple model qualification scenarios, eliminating the need to repeatedly collect and process extensive reference measurements for each new model or process condition, thereby reducing time loss while maintaining reliability through pre-validated reference standards
Solution Approach 2:
The patent creates virtual copies of reference measurement data through computational models. These synthetic reference datasets replicate the information content of extensive physical measurements without requiring the actual time-consuming measurement process to be repeated, enabling rapid model qualification while preserving the reliability that would result from extensive reference data collection
3Measurement precision
If multiple model configurations are evaluated, then measurement precision is improved, but productivity decreases due to increased complexity in model selection
Solution Approach 1:
The patent replaces the manual or semi-automated process of evaluating multiple model configurations with an automated computational system. The virtual reference data generation and comparison process is fully algorithmic, enabling rapid automated selection of optimal models without manual intervention, thereby maintaining measurement precision through thorough model evaluation while restoring productivity through automation
Solution Approach 2:
The patent systematically varies model parameters and configurations in the virtual reference data generation process. By automatically exploring parameter spaces and comparing different model configurations against the generated virtual references, the system efficiently identifies optimal models without manual effort, maintaining precision through comprehensive parameter exploration while improving productivity through automated parameter sweeping and comparison
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 enables accurate model selection and data interpretation similar to using real reference data, reducing the need for external referencing, thereby improving time-to-solution and cost efficiency in scatterometry applications.
Implementation Method 1
Scatterometry is known as the primary non-destructive method to address such detailed requirements for in-line profile metrology. Optical Critical Dimension (also termed 'optical CD' or 'OCD') measurement techniques (known also as Scatterometry) are known as efficient techniques for measuring parameters of patterned (periodic) structures.
Implementation Method 2
in model-based measurement techniques, such as scatterometry, measured data is analysed/interpreted by calculation of optical response from a geometrical/optical model (i.e. theoretical data), fitting a measured optical signal (measured data) to the theoretical data
Data Source
AI summary
A method and system are presented for use in model-based optical measurements in patterned structures. The method comprises: selecting an optimal optical model for interpretation of optical measured data indicative of optical response of the structure under measurements. The selection of the optimal optical model comprises: creating a complete optical model with floating parameters defining multiple configurations of said complete model including one or more model configurations describing an optical response of the structure under measurements, utilizing the complete model for predicting a reference optical response from the structure and generating corresponding virtual reference data, and using the virtual reference data for selecting the optimal optical model for interpretation of the optical measured data.