Automated BEOL Yield Analysis via Machine Vision and Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor manufacturing processes for back end of the line (BEOL) analysis are slow, inefficient, and labor-intensive, requiring manual intervention and lacking rapid cost-effective methods for analyzing BEOL resistance and yield variations.
Innovation Solution
An automated method and system utilizing machine vision image processing and machine learning to measure BEOL structure dimensions and electrical performance, comparing them to a process assumption model, and providing scenarios for process modifications to maintain production within the model specifications, with a manufacturing prediction actuator implementing necessary changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual analysis methods are used for BEOL yield and performance analysis, then detailed examination can be performed, but the analysis process becomes slow and labor-intensive
Solution Approach 1:
The patent replaces manual mechanical analysis methods with automated machine vision image processing and machine learning systems. The machine vision system captures images of BEOL structures and automatically extracts dimensional measurements, while machine learning algorithms analyze the relationship between structural dimensions and electrical performance, eliminating the need for manual measurement and analysis while maintaining high precision.
Solution Approach 2:
The system enables self-service analysis by automatically performing measurements, comparisons, and root cause identification without human intervention. The machine learning models continuously learn from data and automatically generate insights about process deviations and their impact on yield and performance, allowing the system to serve itself in the analysis process.
2Reliability
If comprehensive BEOL analysis is performed to identify root causes of yield and performance variations, then manufacturing quality improves, but the complexity of the analysis system increases
Solution Approach 1:
The patent segments the complex analysis task into distinct functional modules: machine vision image processing for dimensional measurement, machine learning for correlation analysis between structure and performance, and root cause identification systems. Each module handles a specific aspect of the analysis, making the overall system more manageable and maintainable while comprehensively addressing yield and performance variations.
3Manufacturing precision
If real-time process adjustments are implemented to maintain BEOL production within specifications, then yield and performance consistency improves, but the extent of automation required increases system complexity
Solution Approach 1:
The patent implements feedback mechanisms where machine learning models continuously monitor the relationship between BEOL structural dimensions and electrical performance, compare actual measurements against specifications, and provide feedback for process adjustments. This automated feedback loop enables real-time process control to maintain consistency in yield and performance while managing automation complexity through systematic design.
Data Source
AI summary
A method of electrical device manufacturing that includes measuring a first plurality of dimensions and electrical performance from back end of the line (BEOL) structures; and comparing the first plurality of dimensions with a second plurality of dimensions from a process assumption model to determine dimension variations by machine vision image processing. The method further includes providing a plurality of scenarios for process modifications by applying machine image learning to the dimension variations and electrical variations in the in line electrical measurements from the process assumption model. The method further includes receiving production dimension measurements and electrical measurements at a manufacturing prediction actuator. The at least one of the dimensions or electrical measurements received match one of the plurality of scenarios the manufacturing prediction actuator using the plurality of scenarios for process modifications effectuates a process change.