Effective Medium Dispersion Models for Thin Film Stack Metrology
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for measuring thin film stacks, particularly in semiconductor manufacturing, face challenges such as insufficient precision, high parameter correlation, slow throughput, and inability to handle in-situ ALD processes, leading to inaccurate and unstable measurements that fail to meet semiconductor manufacturer specifications.
Innovation Solution
The development of a method using effective medium dispersion models to measure film stacks by substituting multiple layers with virtual targets, allowing for parallel fitting and feedback loops to optimize dispersion models, and utilizing data feedforward between models to break correlation and improve precision and throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional single angle of incidence all floating method is used to measure film stacks, then the measurement can be performed with simple equipment, but the measurement precision is insufficient and parameter correlation is high
Solution Approach 1:
The patent transitions from single angle of incidence measurement to multi-angle of incidence measurement, adding the dimension of angle variation. This enables the system to capture more information about the film stack by measuring at multiple angles, thereby improving measurement precision and reducing parameter correlation through the additional dimensional data.
Solution Approach 2:
The patent segments the film stack measurement into multiple independent layers by using effective medium dispersion models to substitute for all but one layer at a time. This segmentation approach allows each layer to be measured independently with high precision while reducing the correlation between parameters of different layers.
2Productivity
If multiple layers are measured sequentially using traditional methods, then each layer can be measured individually, but the throughput is too slow for commercial manufacturing
Solution Approach 1:
The patent implements continuous measurement of multiple layers simultaneously by using parallel fitting algorithms that process multi-angle data for all layers in one measurement cycle. This eliminates the need for sequential measurement steps while maintaining high precision through the continuous optimization of all layer parameters together.
Solution Approach 2:
The patent uses effective medium dispersion models as preliminary substitutes for multiple layers before actual measurement. This preliminary modeling approach pre-preps the measurement system with expected parameter ranges and correlations, enabling faster convergence during the actual measurement process and significantly improving throughput.
3Reliability
If fixed nominal values are assigned to pre-layer parameters to reduce correlation, then the measurement complexity is reduced, but the measurement accuracy deteriorates because pre-layer processes are not stable
Solution Approach 1:
The patent makes the measurement system dynamic by allowing pre-layer parameters to float and be optimized during the measurement process rather than fixing them at nominal values. The multi-angle measurement data provides sufficient constraints to enable this dynamic adjustment while maintaining measurement robustness and precision simultaneously.
Solution Approach 2:
The patent implements feedback loops where measurement results from multi-angle data are used to continuously refine and update the effective medium dispersion models. This feedback mechanism allows the system to adapt to process variations in real-time, maintaining both robustness and precision even when pre-layer processes are not stable.
4Measurement precision
If conventional measurement methods are used for in-situ ALD processes, then the process can be monitored, but the measurements are inaccurate and unstable due to process-specific challenges
Solution Approach 1:
The patent changes the measurement parameters by using multi-angle of incidence data instead of single-angle data. This parameter change provides additional independent measurements that are less sensitive to the specific challenges of in-situ ALD processes, thereby improving precision while maintaining adaptability to the process environment.
Solution Approach 2:
The patent introduces effective medium dispersion models as intermediaries between the raw optical measurement data and the final film thickness values. These models act as mediators that translate the complex multi-angle optical responses into accurate film thickness measurements, specifically addressing the challenges of in-situ ALD process monitoring.
Data Source
AI summary
Methods and systems disclosed herein can measure thin film stacks, such as film on grating and bandgap on grating in semiconductors. For example, the thin film stack may be a 1D film stack, a 2D film on grating, or a 3D film on grating. One or more effective medium dispersion models are created for the film stack. Each effective medium dispersion model can substitute for one or more layers. A thickness of one or more layers can be determined using the effective medium dispersion based scatterometry model. In an instance, three effective medium dispersion based scatterometry models are developed and used to determine thickness of three layers in a film stack.