Neural Network OPC Parameter Training for Faster Accurate Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing OPC technologies face challenges in efficiently selecting model parameters, particularly optical diameter, leading to either excessive computation time with large diameters or reduced precision with small diameters, affecting the accuracy and efficiency of optical proximity correction (OPC) modeling.
Innovation Solution
A method involving designing multiple OPC patterns to cover circuit layout requirements, collecting wafer pattern data, generating sample units, and using a neural network to train the data, thereby obtaining an optimal model parameter value for improved precision and speed in OPC modeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large optical diameter is used in OPC modeling, then the precision of the OPC model is improved, but the computation time increases exponentially
Solution Approach 1:
The patent applies preliminary action by pre-training a neural network model on comprehensive OPC modeling data before actual OPC operations. The neural network is trained in advance with various optical diameter parameters and their corresponding optimal values, enabling rapid parameter selection during runtime without performing time-consuming traditional OPC calculations. This pre-computation approach stores optimal solutions for quick retrieval, resolving the contradiction between precision and computation time.
2Productivity
If a small optical diameter is used in OPC modeling, then the computation time is reduced, but the precision of the OPC model deteriorates
Solution Approach 1:
The patent introduces a neural network as an intermediary between the optical diameter parameter selection and the OPC modeling process. Instead of directly using small optical diameters for fast computation at the cost of precision, the neural network intermediary predicts the optimal optical diameter based on input pattern characteristics. This intermediary enables the system to achieve both fast computation (by selecting appropriate parameters) and high precision (by using accurately predicted optimal parameters) simultaneously.
3Ease of operation
If traditional OPC modeling methods are used, then the modeling process is straightforward, but the selection of model parameters is difficult and time-consuming
Solution Approach 1:
The patent implements self-service by enabling the neural network model to automatically select optimal optical diameter parameters without requiring manual intervention or expert knowledge. The system autonomously performs parameter selection by querying the pre-trained neural network with pattern inputs and receiving optimal parameter recommendations. This self-service mechanism eliminates the complexity of manual parameter tuning while maintaining ease of operation for the overall modeling process.
Data Source
AI summary
The present disclosure provides a method for training an OPC modeling parameter, including: step 1: designing a plurality of OPC patterns according to a design requirement for a lithography process layer layout; selecting at least one model parameter as a selected model parameter requiring training, and setting a plurality of different values for the selected model parameter; and collecting wafer pattern data formed by performing a lithography process on each OPC pattern; step 2: generating a sample unit and a sample set based on the collected data; step 3: inputting data of the sample set into a neural network for training; and step 4: performing OPC modeling using the trained neural network, where during the OPC modeling, an optimal value of the selected model parameter is obtained using the trained neural network, and the optimal value of the selected model parameter is used to perform the OPC modeling.