Semiconductor Process Control Using Hierarchical Mathematical Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advanced semiconductor manufacturing processes face challenges in maintaining viable yields at nanometer scales due to increasing complexity, requiring integrated advanced equipment control and sensor-based process control beyond standalone tool-level control.
Innovation Solution
A method and system for monitoring manufacturing processes by acquiring metrology and process data to create mathematical models, comparing data against top-level models to detect faults and predict yields, and adjusting process variables to optimize performance, using multivariate analysis techniques like partial least squares and neural networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If stand-alone process tool control based on equipment state data is used, then device fabrication can be performed, but viable yields cannot be maintained at 65 and 45 nm due to increasing process complexity
Solution Approach 1:
The patent combines multiple data sources (equipment state data, sensor data, process data) and control methods (AEC, FDC, predictive models) into an integrated system-wide APC platform. This merging allows the system to handle increased process complexity while maintaining yield by collectively analyzing all available data rather than relying on stand-alone tool control.
Solution Approach 2:
The system implements comprehensive feedback mechanisms by continuously monitoring process data, comparing actual outcomes against predictive models, and adjusting process parameters in real-time. This feedback loop enables the system to respond to process variations and maintain yield despite increasing complexity at nanometer scales.
2Manufacturing precision
If new generation process equipment is purchased to meet advanced production specifications, then manufacturing capability is improved, but cost becomes staggering
Solution Approach 1:
The patent changes the parameters of existing legacy equipment by adding sensor-based monitoring and control systems. Instead of purchasing new equipment, the system modifies the operational parameters and control capabilities of existing tools through software and sensor integration, achieving advanced specification capability at lower cost.
Solution Approach 2:
The system introduces an intermediary layer (sensor-based AEC/APC platform) between the existing equipment and the manufacturing process. This intermediary enables legacy equipment to meet advanced specifications by mediating control and monitoring functions without requiring physical replacement of the equipment.
3Manufacturing precision
If sensor-based AEC/APC is implemented in legacy equipment, then tighter specifications are achieved, but system complexity increases
Solution Approach 1:
The patent creates a universal APC platform that can be applied across multiple legacy equipment types and process steps. By designing a multi-functional system that handles various data sources and equipment through a common architecture, the system achieves tight specifications without proportionally increasing complexity for each individual tool.
4Productivity
If comprehensive process monitoring and control systems are implemented, then yields and throughput are enhanced, but data processing complexity increases
Solution Approach 1:
The patent segments the comprehensive data processing task into hierarchical levels: tool-level AEC, system-level FDC, and plant-wide predictive models. This segmentation allows each layer to process data at an appropriate granularity, managing complexity while enabling enhanced throughput through coordinated control across all levels.
Data Source
AI summary
A method for monitoring a manufacturing process features acquiring metrology data for semiconductor wafers at the conclusion of a final process step for the manufacturing process (“Step a”). Data is acquired for a plurality of process variables for a first process step for manufacturing semiconductor wafers (“Step b”). A first mathematical model of the first process step is created based on the metrology data and the acquired data for the plurality of process variables for the first process step (“Step c”). Steps b and c are repeated for at least a second process step for manufacturing the semiconductor wafers (“Step d”). An nth mathematical model is created based on the metrology data and the data for the plurality of process variables for each of the n process steps (“Step e”). A top level mathematical model is created based on the metrology data and the models created by steps c, d and e (“Step f”). The top level mathematical model of Step f is based on those process variables that have a substantial effect on the metrology data.