Plasma Processing Apparatus Dynamic Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing plasma processing technologies face challenges in maintaining high prediction accuracy for processing results due to changes in the state of the plasma processing apparatus, as they rely on single prediction models that fail to adapt to changing correlations between light emission data and processing results.
Innovation Solution
A plasma processing apparatus that includes a prediction model storage unit and a control device capable of selecting and switching prediction models based on light emission data and device data indicators of state change, using multiple explanatory variables and partial correlations to construct a multivariate prediction model that adapts to variations in processing results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single prediction model is used to predict plasma processing results, then the device complexity is reduced, but the prediction accuracy decreases when apparatus state changes
Solution Approach 1:
The patent implements a dynamic prediction model selection mechanism that automatically switches between multiple prediction models based on real-time apparatus state monitoring. The system monitors light emission data and device data to detect state changes, then selects the most appropriate prediction model from a plurality of stored models, enabling the system to adapt dynamically to changing conditions and maintain high prediction accuracy.
Solution Approach 2:
The patent changes the parameter of prediction model selection based on apparatus state parameters. By monitoring light emission data and device data as indicators of state change, the system selects different prediction models corresponding to different apparatus states, thereby maintaining prediction accuracy across varying operating conditions without requiring a single complex universal model.
2Measurement precision
If multiple prediction models are used to maintain accuracy under varying apparatus states, then the prediction accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent segments the prediction task by dividing it into multiple specialized prediction models, each optimized for specific apparatus states. Instead of using one complex model for all conditions, the system creates multiple simpler models that each handle particular state ranges effectively, reducing individual model complexity while maintaining overall system accuracy through selective deployment.
Solution Approach 2:
The patent introduces an intermediary mechanism (the selection unit) that manages the complexity of multiple prediction models. This intermediary monitors apparatus state and automatically selects the appropriate model, shielding the user from the complexity of having multiple models while enabling the system to leverage their combined predictive capabilities for high accuracy across varying conditions.
3Ease of operation
If light emission data is used as the only input for prediction, then the ease of operation is maintained, but the prediction accuracy decreases when correlation changes
Solution Approach 1:
The patent enhances the functionality of the prediction system by incorporating multiple types of input data (light emission data and device data) that serve different purposes. Light emission data provides real-time plasma state information, while device data provides contextual apparatus state information. This multi-functional input approach maintains ease of operation through automated data collection while improving prediction accuracy by capturing both plasma and apparatus state variations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables high-accuracy prediction and control of plasma processing results, even when the correlation between monitor data and processing results changes, by selecting suitable explanatory variables and switching prediction models according to apparatus state, thereby improving prediction accuracy and stability.
Implementation Method 1
a monitor that monitors light emission of plasma
Data Source
AI summary
Provided is a plasma processing apparatus including a processing unit in which a sample is plasma processed and which includes a monitor (optical emission spectroscopy) that monitors light emission of plasma, wherein the processing unit includes a prediction model storage unit that stores a prediction model predicting a plasma processing result, and a control device in which the plasma processing result is predicted by using a prediction model selected based on light emission data and device data as an indicator of state change of the processing unit.