Auto-Learning Tool Matching for Semiconductor Fabrication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tool matching processes in semiconductor fabrication are burdensome and infrequent due to the need for specialized wafers and intrusive procedures, leading to potential degradation in fabrication quality.
Innovation Solution
An auto-learning feedback loop method that updates a library of key parameters by monitoring deviations in tool conditions using a control wafer, allowing for more frequent tool matching without the need for frequent full runs, utilizing a measurement tool with a computing system to compare parameters against thresholds and add new parameters to the library for continuous monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If specialized monitor wafers are used for tool matching, then tool matching accuracy is improved, but process intrusiveness and operational burden increase
Solution Approach 1:
The patent uses control wafers as simplified copies or representatives of production wafers for tool matching purposes. These control wafers contain predefined control structures that enable parameter measurement without requiring specialized monitor wafers, thereby reducing operational burden while maintaining measurement accuracy.
Solution Approach 2:
The system automatically performs tool matching by measuring parameters on control wafers and comparing them against stored reference data. This self-service approach eliminates the need for manual intervention or specialized monitor wafer handling, reducing operational burden while maintaining matching accuracy.
2Measurement precision
If full parameter runs are performed frequently, then tool condition monitoring accuracy is improved, but productivity decreases
Solution Approach 1:
The patent extracts and monitors only the most critical parameters from the full set of measurable parameters. By identifying and focusing on key parameters that most strongly indicate tool condition changes, the system achieves accurate tool condition monitoring without requiring complete full parameter runs, thereby maintaining productivity.
Solution Approach 2:
Instead of performing complete full parameter runs frequently, the system performs partial measurements focusing only on critical parameters. This partial action approach provides sufficient tool condition information while minimizing the time and resource overhead, thus maintaining high productivity.
3Measurement precision
If comprehensive parameter monitoring is implemented, then tool condition detection accuracy is improved, but device complexity increases
Solution Approach 1:
The patent segments the comprehensive parameter monitoring into two levels: a library of all measurable parameters and a focused set of critical parameters for frequent monitoring. This segmentation allows the system to maintain comprehensive detection capability while simplifying the actual monitoring process to only essential parameters, reducing device complexity.
4Measurement precision
If control wafers with control structures are used, then parameter measurement accuracy is improved, but manufacturing complexity increases
Solution Approach 1:
The control wafers with control structures serve multiple functions: they enable parameter measurement, provide reference data for comparison, and can be used repeatedly across different tools and time periods. This multi-functionality justifies the additional manufacturing complexity by eliminating the need for separate specialized monitor wafers and reducing long-term operational costs.
Data Source
AI summary
The present disclosure is directed to a method of tool matching that employs an auto-learning feedback loop to update a library of key parameters. According to the method, measurements are performed on a control wafer to collect a set of parameters associated with the process/analysis tool that is being matched. When deviated parameters correlate to a correctable tool condition (i.e. a tool matching event), the parameters are added to the library of key parameters. These key or critical parameters may be monitored on a more frequent basis to identify deviations that have a strong likelihood of matching with a correctable tool condition. The tool matching methodology advantageously allows for monitoring of an automatically updated list of key parameters instead of needing to look at the full set of parameters collected from a control wafer each time. As such, tool matching can be performed on a more frequent basis.