Built-in Self Test Table for Thermal Processing Fault Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting fault conditions in semiconductor processing equipment, such as mass flow controllers and heater elements, are inadequate as they rely on scheduled preventative maintenance and do not effectively address emerging faults or provide system-level detection and prediction.
Innovation Solution
A method involving a built-in self-test (BIST) table is created to monitor thermal processing systems in real-time by changing process parameters, predicting responses, and comparing measured and predicted errors to operational thresholds, allowing for the detection, diagnosis, and prediction of fault conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If scheduled preventative maintenance is used, then equipment reliability is maintained based on average characteristics, but emerging fault conditions cannot be detected and system-level prediction is not provided
Solution Approach 1:
The patent implements a feedback mechanism where process data is continuously collected from multiple sources (MFC readings, chamber pressure, temperature profiles) and fed into a dynamic model that compares actual behavior against expected behavior. This feedback loop enables real-time detection of deviations indicating emerging faults, transforming static scheduled maintenance into dynamic condition-based monitoring that provides system-level prediction capabilities
Solution Approach 2:
The system performs self-diagnosis by automatically comparing measured process parameters against the dynamic model's predictions without requiring external intervention. The BIST software autonomously detects faults, identifies their locations, and predicts remaining useful life, enabling the equipment to monitor its own health status continuously rather than relying on periodic external maintenance checks
2Difficulty of detecting and measuring
If component vibration analysis is used, then mechanical equipment faults can be detected, but majority of critical components such as mass flow controllers and heaters cannot be monitored
Solution Approach 1:
The patent creates a universal monitoring system that uses process data from multiple components (MFCs, heaters, thermocouples, pressure sensors) to detect faults across diverse equipment types. Instead of component-specific methods like vibration analysis, the dynamic model approach universally applies to thermal processing systems by monitoring how changes in one component affect overall process behavior, enabling detection of faults in mass flow controllers, heaters, and other critical components through their process signatures
3Measurement precision
If statistical process control charts are used, then single component faults can be detected, but system-level detection and prediction are not provided
Solution Approach 1:
The patent merges data from multiple components and process parameters into a unified dynamic model that performs system-level analysis. Instead of separate SPC charts for each component, the system combines MFC readings, chamber pressure, temperature profiles, and power consumption into an integrated model that detects interactions between components and provides holistic system health assessment with prediction capabilities
4Reliability
If real-time dynamic modeling is implemented, then system-level fault detection and prediction are achieved, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent performs preliminary action by pre-developing the dynamic model using historical process data and system characteristics before real-time operation. The model structure, parameters, and expected behavior ranges are established in advance, allowing real-time monitoring to focus on comparing actual measurements against pre-computed predictions rather than performing complex calculations during each measurement cycle, thus reducing real-time computational burden while maintaining prediction capability
Data Source
AI summary
A method of creating and/or modifying a built-in self test (BIST) table for monitoring a thermal processing system in real-time that includes positioning a plurality of wafers in a processing chamber in the thermal processing system; executing a real-time dynamic model to generate a predicted dynamic process response; creating a measured dynamic process response; determining a dynamic estimation error; determining if the determined dynamic estimation error can be associated with a pre-existing BIST rule in the BIST table; creating a new BIST rule when the dynamic estimation error cannot be associated with any pre-existing BIST rule in the BIST table; and stopping the process when a new BIST rule cannot be created.