Laser Wavelength Control via Prism Position Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demands for precision and speed in semiconductor photolithography processes due to reduced semiconductor feature sizes and high laser firing rates pose challenges in accurately controlling the wavelength of laser light sources, particularly due to measurement delays and disturbances in the laser system.
Innovation Solution
An advanced system and method for controlling the wavelength of a laser that predicts the position of the prism at faster intervals than the laser firing rate, using a model of the prism's motion and disturbance behaviors, allowing for more frequent repositioning without waiting for output wavelength measurements, and updates the prediction using available measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system waits for output wavelength measurements before repositioning the prism, then measurement accuracy is improved, but control speed and response time deteriorate
Solution Approach 1:
The system performs preliminary actions by predicting the prism position and pre-calculating control signals before actual wavelength measurements are available. The control computer uses a prediction model to estimate where the prism will be positioned based on previous positions and control signals, allowing it to prepare compensation signals in advance rather than waiting for measurements.
Solution Approach 2:
The control process is segmented into prediction steps and measurement steps that can operate at different rates. The prediction model runs continuously at high speed to generate intermediate control signals, while wavelength measurements are taken at lower intervals to update and correct the prediction model, creating a multi-rate control architecture.
2Productivity
If the laser firing rate is increased to improve productivity, then output per unit time is improved, but the time available for measurement and control calculations decreases
Solution Approach 1:
The system performs preliminary control calculations by predicting prism position and pre-computing compensation signals before the laser fires. This allows the control computer to prepare correction signals in advance, reducing the critical path time during high-speed operation where every microsecond counts between laser pulses.
Solution Approach 2:
The system creates a virtual model (copy) of the prism's motion behavior through prediction algorithms that simulate how the piezoelectric transducer will move the prism. This digital twin allows the control system to forecast positions and calculate corrections without waiting for physical measurements, enabling faster control loops.
3Manufacturing precision
If prediction intervals are made faster than measurement intervals, then control precision is improved, but system complexity increases
Solution Approach 1:
The system implements feedback by using actual wavelength measurements to update and correct the prediction model. When measurements are available, they are compared with predicted values, and the differences (errors) are used to refine future predictions, creating a closed-loop system that improves accuracy without requiring equally fast measurements.
Solution Approach 2:
The system changes parameters by operating the prediction and measurement processes at different time scales. The prediction model runs at a higher frequency with smaller time steps for greater precision, while measurements occur at lower frequency but with sufficient accuracy to correct and validate the predictions, creating a multi-scale control approach.
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 more rapid and precise control of the laser wavelength, reducing performance degradation and allowing for faster laser firing rates by decoupling prediction intervals from measurement intervals, thereby improving the overall precision and speed of the photolithography process.
Implementation Method 1
The position of the prism in LNM 110 is controlled by a voltage applied to a piezoelectronic transducer (PZT) 140 connected to the prism in LNM 110
Implementation Method 2
the resulting light enters Line Narrowing Module (LNM) 110 where it shines through a prism (actually several prisms) and onto a grating within LNM 110. This acts as a light wavelength selector in that changing the position of the prism in LNM 110 changes the wavelength of the laser light
Implementation Method 3
shines through a prism (actually several prisms) and onto a grating within LNM 110
Data Source
AI summary
Laser light wavelength control is provided by periodically predicting a next position of a light controlling prism using a model of the prism's motion characteristics. The prediction is then updated if a measurement of laser output wavelength is obtained. However, because the predictions are made without waiting for a measurement, they can be made more frequently than the laser firing repetition rate and the prism can be repositioned at discrete points in time which can occur more frequently than the laser firing events. This also reduces performance degradation which may be caused by being one pulse behind a laser measurement and the resultant laser control signal being applied.