Lithography Process Model Optimization via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lithographic processes face challenges in accurately reproducing patterns with dimensions smaller than the classical resolution limit, requiring sophisticated fine-tuning steps and resolution enhancement techniques to achieve desired electrical functionality and performance.
Innovation Solution
A method involving obtaining and processing input values for a lithographic process model, including characteristics of illumination, projection optics, and patterning devices, to determine a process model using training data and simulated values, allowing for optimization and improvement of pattern reproduction on substrates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional lithographic processes are used to reproduce patterns with dimensions smaller than the classical resolution limit, then manufacturing precision deteriorates, but using sophisticated fine-tuning steps and resolution enhancement techniques improves manufacturing precision while increasing device complexity
Solution Approach 1:
The patent applies parameter changes by systematically varying lithographic process parameters (illumination conditions, optical characteristics, patterning parameters) to optimize pattern reproduction. The computer-implemented method evaluates multiple parameter combinations to identify optimal settings that achieve sub-resolution pattern fidelity without requiring complex additional process steps.
Solution Approach 2:
The patent implements preliminary action through pre-computation and simulation of lithographic outcomes before actual manufacturing. By using training data and machine learning models to predict pattern results under various conditions, the system prepares optimal process parameters in advance, eliminating the need for sophisticated fine-tuning during production and reducing overall process complexity.
2Manufacturing precision
If more sophisticated fine-tuning steps and resolution enhancement techniques are applied, then manufacturing precision improves, but productivity decreases due to additional processing steps
Solution Approach 1:
The system performs preliminary computation and optimization using machine learning models trained on extensive simulation data. By determining optimal process parameters before production through automated evaluation of training data, the method eliminates multiple iterative fine-tuning steps during manufacturing, thereby maintaining high precision while improving productivity through reduced cycle time.
Solution Approach 2:
The patent implements self-service through automated parameter optimization using machine learning algorithms that independently evaluate training data and determine optimal settings without requiring manual intervention or iterative experimentation. This automation eliminates time-consuming fine-tuning steps while maintaining optimization quality, thus improving productivity without sacrificing precision.
3Manufacturing precision
If resolution enhancement techniques are used to achieve desired electrical functionality, then manufacturing precision improves, but device complexity increases
Solution Approach 1:
The patent resolves this contradiction by systematically exploring parameter space through automated evaluation of illumination, optical, and patterning parameters. The machine learning model identifies optimal parameter combinations that achieve target electrical performance characteristics without requiring additional process steps or complex resolution enhancement techniques, thereby maintaining simplicity while ensuring precision.
Solution Approach 2:
The patent extracts and isolates the critical parameters that most significantly influence pattern reproduction quality from the full set of lithographic parameters. By focusing optimization efforts on these key parameters identified through training data analysis, the system achieves high manufacturing precision without implementing complex multi-step resolution enhancement processes, thus reducing overall device complexity.
Data Source
AI summary
Methods of determining, and using, a process model that is a machine learning model. The process model is trained partially based on simulation or based on a non-machine learning model. The training data may include inputs obtained from a design layout, patterning process measurements, and image measurements.