Stochastic Metric Inference via Optical Metrology and ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stochastic metrology in lithographic processes is slow, making it unsuitable for high-volume manufacturing due to reliance on scanning electron microscopes, which are time-consuming.
Innovation Solution
A method using a trained model to infer stochastic metrics from optical metrology data, correlating angularly resolved intensity parameter distributions of scattered radiation with stochastic metric values, allowing for faster and more efficient measurement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If scanning electron microscope (SEM) metrology is used to determine stochastic metrics, then measurement precision is improved, but productivity deteriorates due to slow measurement speed
Solution Approach 1:
The patent replaces the mechanical scanning electron microscope system with an optical metrology system that uses light-based diffraction measurements. This substitution enables faster measurements while maintaining the capability to determine stochastic metrics through computational analysis of diffraction patterns, thereby resolving the contradiction between measurement precision and productivity
Solution Approach 2:
The patent introduces a trained machine learning model as an intermediary between the optical metrology measurements and the stochastic metric determination. This intermediary processes the diffraction data to infer stochastic metrics, enabling the optical system to achieve precision previously only attainable through SEM while maintaining high throughput
2Productivity
If optical metrology is used for stochastic analysis, then productivity is improved through faster measurements, but measurement precision deteriorates compared to SEM
Solution Approach 1:
The patent changes the measurement parameters by using angularly resolved diffraction measurements across multiple diffraction orders instead of direct imaging. This parameter change, combined with machine learning analysis, enables the optical metrology system to extract stochastic metric information with precision comparable to SEM while maintaining the speed advantages of optical methods
Solution Approach 2:
The patent transitions from direct spatial imaging to measuring diffraction patterns in reciprocal space (angular domain). This dimensionality change allows the system to access structural information including stochastic variations through the diffraction pattern's angular distribution, achieving precision without the slow scanning process
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
Enables faster and more accurate determination of stochastic metrics compared to traditional SEM metrology, facilitating higher throughput in manufacturing by utilizing optical metrology data with machine learning models.
Implementation Method 1
measurement signals relating to a plurality of angularly resolved distributions of an intensity related parameter across a zero or higher order of diffraction comprised within radiation scattered from a plurality of training structures on a substrate
Implementation Method 2
angularly resolved distributions of an intensity related parameter across a zero or higher order of diffraction
Data Source
AI summary
A method of determining a stochastic metric, the method including: obtaining a trained model having been trained to correlate training optical metrology data to training stochastic metric data, wherein the training optical metrology data includes a plurality of measurement signals relating to distributions of an intensity related parameter across a zero or higher order of diffraction of radiation scattered from a plurality of training structures, and the training stochastic metric data includes stochastic metric values relating to the plurality of training structures, wherein the plurality of training structures have been formed with a variation in one or more dimensions on which the stochastic metric is dependent; obtaining optical metrology data including a distribution of the intensity related parameter across a zero or higher order of diffraction of radiation scattered from a structure; and using the trained model to infer a value of the stochastic metric from the optical metrology data.