Wafer Measurement Site Selection via Signal Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current metrology systems face challenges in effectively characterizing process variation across semiconductor wafers due to limited measurement sites and biased representation, especially with increasing complexity and small resolution requirements, leading to inaccuracies in identifying suitable measurement sites.
Innovation Solution
The method involves transforming high-throughput measurement signals into a new mathematical basis, selecting representative measurement sites using machine-learning techniques, and regularizing die locations for uniform spatial distribution, while eliminating outliers to minimize process anomalies and ensure robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a small number of measurement sites are selected to maintain cost and throughput goals, then productivity is improved, but measurement precision deteriorates due to insufficient representation of process variation
Solution Approach 1:
The system performs preliminary high-throughput measurements at a large number of measurement sites across the wafer before selecting representative sites for detailed characterization. This preliminary data collection enables subsequent identification of sites that best represent process variation, resolving the contradiction by preparing information in advance that guides efficient site selection.
Solution Approach 2:
The system transforms measurement signals into a new mathematical basis and uses machine learning techniques to identify representative measurement sites based on signal variation patterns. By changing the parameter selection criteria from uniform or heuristic-based selection to variation-driven selection, the system achieves both high throughput and accurate process variation characterization.
2Ease of operation
If measurement sites are selected uniformly across the wafer surface, then ease of operation is improved, but measurement precision deteriorates due to failure to effectively characterize process variation
Solution Approach 1:
The system changes the selection parameter from uniform spatial distribution to representation of measurement signal variation. By selecting sites based on their ability to capture process variation patterns in the measurement signals, the system achieves both operational simplicity and measurement precision.
Solution Approach 2:
The system introduces machine learning models and signal transformation as intermediaries between uniform site selection and process variation characterization. These intermediaries process the measurement signals to identify representative sites, bridging the gap between simple selection methods and accurate variation characterization.
3Ease of operation
If heuristic rules are used to select measurement sites based on empirical knowledge, then ease of operation is improved, but measurement precision deteriorates due to misrepresentation of process variation
Solution Approach 1:
The system replaces heuristic-based selection parameters with data-driven parameters derived from actual measurement signal variation. By using machine learning models trained on measurement signals to identify representative sites, the system maintains operational simplicity while achieving accurate process variation characterization.
Solution Approach 2:
The system substitutes empirical heuristic rules with machine learning-based algorithms that automatically identify representative measurement sites. This substitution replaces subjective or experience-based selection methods with objective, data-driven selection that accurately captures process variation.
4Measurement precision
If OCD measurement models are built and verified to select measurement sites, then measurement precision is improved, but productivity deteriorates due to significant time and effort required
Solution Approach 1:
The system performs preliminary high-throughput measurements across many sites to build a training dataset before final site selection. This preliminary action enables the machine learning model to learn from actual measurement variation patterns, achieving accurate site selection without requiring time-consuming model verification at each candidate site.
Solution Approach 2:
The system changes the approach from building and verifying complex OCD measurement models to using machine learning models that directly predict representative measurement sites from high-throughput signals. This parameter change reduces the time and effort required while maintaining or improving selection accuracy.
Data Source
AI summary
Methods and systems for selecting measurement locations on a wafer for subsequent detailed measurements employed to characterize the entire wafer are described herein. High throughput measurements are performed at a relatively large number of measurement sites on a wafer. The measurement signals are transformed to a new mathematical basis and reduced to a significantly smaller dimension in the new basis. A set of representative measurement sites is selected based on analyzing variation of the high throughput measurement signals. In some embodiments, the spectra are subdivided into a set of different groups. The spectra are grouped together to minimize variance within each group. Furthermore, a die location is selected that is representative of the variance exhibited by the die in each group. A spectrum of a measurement site and corresponding wafer location is selected to correspond most closely to the center point of each cluster.