Semiconductor Profile Measurement Using a Scanning Conditional Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing metrology techniques for semiconductor structures face challenges in accurately measuring parameters due to high computational cost, complexity, and multiple solution ambiguities, especially with small resolution requirements and complex geometric structures, leading to inaccuracies and inefficiencies in model training and parameter inference.
Innovation Solution
A scanning conditional measurement model is trained using Design Of Experiments (DOE) data with known and perturbed parameter values, minimizing error functions to estimate parameter values accurately, enabling efficient training with fewer samples and handling multiple solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physics-based measurement models are used to predict raw measurement signals, then measurement capability is provided, but computational cost and time consumption increase significantly
Solution Approach 1:
The patent pre-computes a library of measurement signals corresponding to various parameter values before actual measurement. This preliminary action creates a lookup table that enables rapid parameter estimation during actual measurement by simply matching the measured signal against the pre-computed library, avoiding real-time iterative calculations and significantly reducing measurement time while maintaining accuracy.
Solution Approach 2:
The patent creates copies of measurement signals for different parameter values and stores them in a library. Instead of performing complex calculations during actual measurement, the system copies the measured signal and compares it against the pre-computed signal copies in the library to identify the matching parameter value, thereby reducing computational burden and time consumption.
2Measurement precision
If physics-based measurement models with iterative regression are employed, then parameter values can be resolved, but computational complexity and cost increase
Solution Approach 1:
The patent creates copies of measurement signals for different parameter values and stores them in a library. Instead of performing complex calculations during actual measurement, the system copies the measured signal and compares it against the pre-computed signal copies in the library to identify the matching parameter value, thereby reducing computational burden and time consumption.
Solution Approach 2:
The patent pre-computes a library of measurement signals corresponding to various parameter values before actual measurement. This preliminary action creates a lookup table that enables rapid parameter estimation during actual measurement by simply matching the measured signal against the pre-computed library, avoiding real-time iterative calculations and significantly reducing measurement time while maintaining accuracy.
3Measurement precision
If traditional measurement models are used, then single solution is found, but measurement completeness is lost when multiple solutions exist
Solution Approach 1:
The patent dynamically adapts the measurement model by selecting from multiple candidate models based on the specific measurement scenario. The system evaluates multiple possible parameter values and their corresponding measurement signals, then dynamically determines which parameter value best matches the actual measurement. This dynamic approach allows the system to handle multiple solutions appropriately rather than being constrained to a single fixed model, thereby preserving measurement completeness.
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
The scanning conditional model provides accurate parameter estimation with reduced training data requirements, allowing for precise two- and three-dimensional imaging of semiconductor structures, enhancing measurement accuracy and efficiency.
Implementation Method 1
a beam of electrons is scanned over a surface of a semiconductor wafer and secondary electrons generated by interaction of the beam with the surface are detected
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Methods and systems for measuring semiconductor structures based on a trained scanning conditional measurement model are described herein. A scanning conditional model is trained based on Design Of Experiments (DOE) measurement data associated with known values of one or more parameters of interest and a set of perturbed values of the one or more parameters of interest. The trained conditional model minimizes the output of an error function characterizing the error between the known values of the perturbed values of the one or more parameters of interest for the given DOE measurement data. During inference, an error value associated with each candidate value of one or more parameters of interest is determined by the trained scanning conditional measurement model. The estimated value of the parameter of interest is the candidate value of the parameter of interest associated with the minimum error value.