Excimer Laser Energy Model Identification Using Gated Recurrent Unit Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing excimer laser energy models struggle to accurately represent the complex nonlinear energy characteristics of excimer lasers, making it difficult to achieve precise control and simulation of laser energy in semiconductor photoetching processes.
Innovation Solution
A gated recurrent unit network is built and trained using energy harvesting conditions such as time intervals and discharge high voltage values to identify an excimer laser energy model, allowing for precise identification of pulse energy sequences and improving energy stability and dose precision control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If theoretical derivation methods are used to build excimer laser energy models, then the model structure can be established, but the model precision is insufficient due to the complicated nonlinear characteristics
Solution Approach 1:
The patent replaces traditional theoretical derivation methods with a data-driven neural network approach. Instead of relying on complex theoretical models that are difficult to derive accurately for nonlinear systems, the invention uses experimental data to train a neural network model that automatically learns the underlying patterns, achieving higher precision without requiring explicit theoretical formulations.
Solution Approach 2:
The patent transforms the modeling approach from fixed theoretical parameters to adaptive learned parameters. The neural network model learns optimal parameters from experimental data, allowing the model to adapt to the actual nonlinear characteristics of the excimer laser system, thereby improving precision while managing complexity through automated parameter optimization.
2Measurement precision
If complex nonlinear models are used to accurately represent excimer laser energy characteristics, then model precision improves, but model identification difficulty increases
Solution Approach 1:
The patent substitutes traditional model identification methods with neural network-based identification. Instead of attempting to analytically identify parameters in complex nonlinear models, the invention uses a neural network to automatically learn the mapping from input conditions to energy characteristics from experimental data, greatly simplifying the identification process while maintaining high precision.
Solution Approach 2:
The patent creates a virtual copy of the excimer laser system through the neural network model. This digital twin replicates the complex nonlinear energy characteristics without requiring direct analytical modeling, allowing for easy identification and simulation while accurately representing the physical system's behavior.
3Manufacturing precision
If traditional energy models are used, then model simplicity is maintained, but the ability to meet simulation requirements and control precision is insufficient
Solution Approach 1:
The patent replaces traditional simple energy models with a neural network-based model that can capture complex nonlinear relationships. This substitution enables the system to achieve the required energy control precision for semiconductor photoetching by learning from experimental data, overcoming the limitations of simplified traditional models.
Solution Approach 2:
The patent introduces a dynamic modeling approach using neural networks that can adapt to varying operating conditions. Unlike static traditional models, the neural network model dynamically adjusts its predictions based on learned patterns from experimental data, enabling precise energy control across different operating regimes while managing complexity through adaptive learning.
Data Source
AI summary
Disclosed in the present invention are an excimer laser energy model identification method and apparatus. The method comprises the following steps: establishing a gated recurrent network for excimer laser energy model identification; within a plurality of preset time periods, setting energy collection conditions in a single laser pulse mode, and collecting a training data set for excimer laser energy model identification; and using the training data set to train the established gated recurrent network, and when a training termination condition is satisfied, ending the training and obtaining an excimer laser energy model. By means of the method provided by the present invention, the maximum error between a pulse energy generated by the identified excimer laser energy model and an actual pulse energy is less than 1.5%, and thus, a simulation requirement of excimer laser energy characteristic control can be met.