Model Predictive Control Algorithm for Semiconductor Tool Drift
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current advanced process control (APC) strategies in semiconductor manufacturing, particularly in lithography processes, face challenges in efficiently compensating for tool drifts and disturbances, leading to reduced controller performance and increased production costs due to reprocessing of out-of-control substrates.
Innovation Solution
Incorporating machine constants into a model predictive control strategy to directly address tool drifts and disturbances, allowing for real-time adjustment of manipulated variables and maintaining process control efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine constants are not explicitly considered in the control algorithm, then the controller structure remains simple, but controller performance deteriorates after step disturbances due to tool drifts
Solution Approach 1:
The patent applies preliminary action by incorporating machine constant measurements into the control algorithm before they would cause performance deterioration. The measured machine constants are proactively integrated into the predictive control model to preemptively compensate for tool drifts and disturbances, maintaining controller performance without waiting for out-of-control situations to occur.
2Productivity
If conventional APC strategies are used without machine constant integration, then the control algorithm remains simple, but the number of pilot substrates and re-initialization events increases
Solution Approach 1:
The patent implements feedback by continuously measuring machine constants and feeding this information back into the predictive control algorithm. This closed-loop approach allows the controller to adapt to tool drifts in real-time, maintaining process control and reducing the need for pilot substrates and re-initialization events that would otherwise be required to compensate for unmonitored drifts.
3Manufacturing precision
If machine constants are explicitly considered in the control algorithm, then controller performance is maintained after step disturbances, but the control algorithm becomes more complex
Solution Approach 1:
The patent applies parameter changes by incorporating machine constant measurements as additional parameters into the predictive control algorithm. Rather than fundamentally redesigning the control structure, the solution integrates measured machine constants as input parameters that modify the control predictions, thereby improving manufacturing precision while maintaining algorithmic tractability through parameter augmentation.
Data Source
AI summary
A controller and a method of controlling a process tool is provided, in which machine constants used for calibrating manipulated variables of the control algorithm are explicitly introduced into the process model, thereby providing an enhanced controller behavior immediately after the introduction of new measurement values of the machine constants.