Advanced Process Control Using Measurement Uncertainty
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current advanced process control (APC) techniques in semiconductor manufacturing face challenges in accurately controlling process parameters due to measurement uncertainty and limited data availability, leading to inefficiencies and potential product loss from undetected parameter drifts and tool variations.
Innovation Solution
A control system that estimates the current process state using metrology data and a model, incorporating uncertainty into the control algorithm by weighting manipulated variables based on standard errors from measurement data, enhancing control accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If measurement data is used for process control, then control accuracy is improved, but measurement uncertainty reduces reliability
Solution Approach 1:
The patent implements feedback by using measurement data to update the process state estimate and adjust future control actions. The measurement data is fed back into the state estimator to correct deviations from the predicted process state, thereby improving control accuracy while accounting for measurement uncertainty through statistical methods.
Solution Approach 2:
The patent changes the parameter representation by using a state vector that includes both process parameters and their uncertainties. The measurement uncertainty is incorporated as a statistical parameter (covariance matrix) that modifies how measurement data is weighted in the state estimation, allowing the system to adapt to varying measurement quality.
2Manufacturing precision
If more measurement data is collected, then control accuracy is improved, but measurement time and equipment effort increase
Solution Approach 1:
The patent applies partial action by using a limited set of measurement data points rather than comprehensive measurements of all process parameters. The state estimator reconstructs the complete process state from these partial measurements, reducing measurement time and equipment effort while maintaining control accuracy through intelligent data utilization.
Solution Approach 2:
The patent makes the measurement system multi-functional by using the same measurement data for multiple purposes: updating the process state estimate, calculating control actions, and assessing measurement quality. This universal use of measurement data maximizes the value extracted from limited measurements, improving control accuracy without proportionally increasing measurement effort.
3Reliability
If measurement uncertainty is incorporated into control, then control reliability is improved, but control algorithm complexity increases
Solution Approach 1:
The patent implements self-service by having the control algorithm automatically assess and utilize measurement uncertainty without requiring external intervention. The state estimator autonomously weights measurement data based on their uncertainty levels and adjusts control actions accordingly, improving reliability while keeping the algorithm self-contained and manageable.
Solution Approach 2:
The patent introduces dynamics by making the control algorithm adaptive to varying measurement uncertainty levels. The weighting of measurement data in the state estimation is dynamically adjusted based on the current measurement uncertainty (covariance matrix), allowing the system to optimize control reliability in real-time without requiring a completely complex redesign for each scenario.
Data Source
AI summary
By taking into consideration the measurement uncertainties in the form of standard errors, the performance of APC controllers may be efficiently enhanced by using the standard errors as a control input. For example, the filter parameter of an EWMA filter may be efficiently scaled on the basis of a standard error of measurement data.