Deep Learning Optical Proximity Correction for Corner Rounding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As semiconductor devices become increasingly complex and highly integrated, the reduced distances between layout patterns in photolithography processes lead to interference and diffraction issues, resulting in distorted patterns on the substrate, which existing optical proximity correction methods address inadequately, particularly in terms of accuracy and process margin.
Innovation Solution
A deep learning model is trained using sample input and reference images to perform corner rounding operations in optical proximity correction, leveraging data from real manufacturing processes to enhance the accuracy of photomask fabrication and improve process margins by learning from actual photomasks applied in semiconductor manufacturing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If existing optical proximity correction methods are used, then layout patterns can be corrected for interference and diffraction, but the accuracy and process margin are insufficient
Solution Approach 1:
The patent replaces traditional mechanical/optical correction methods with a deep learning-based computational approach. The neural network model learns from training data to predict and correct optical proximity effects, substituting physical correction mechanisms with intelligent algorithms that achieve higher accuracy and better process margins.
Solution Approach 2:
The patent changes the parameters of the correction process by using learnable weights and biases in the neural network. Instead of fixed correction rules, the system dynamically adjusts correction parameters based on input pattern characteristics, enabling adaptive optimization of correction accuracy and process margin.
2Manufacturing precision
If deep learning models are trained with comprehensive data, then corner rounding accuracy improves, but training time and computational resources increase
Solution Approach 1:
The patent performs preliminary action by collecting and preparing training data in advance from real photomasks and layout patterns. The training phase is conducted beforehand to establish the neural network model, which can then be deployed for rapid inference during actual optical proximity correction operations, separating the time-consuming training phase from the production phase.
Solution Approach 2:
The patent focuses training on specific critical features (corner portions of layout patterns) rather than attempting to perfect all aspects of the pattern. By concentrating computational resources on the most important correction tasks, the system achieves high corner rounding accuracy without requiring excessive training time for less critical features.
Data Source
AI summary
In a method of training a deep learning model for optical proximity correction, sample input images associated with sample layouts may be obtained, where the sample layouts are targets of the optical proximity correction. Sample reference images that correspond to the sample input images may be extracted from sample masks that are fabricated by performing the optical proximity correction on the sample layouts. A training operation may be performed on the deep learning model used in the optical proximity correction based on the sample input images and the sample reference images. The sample layouts may include sample layout patterns to form process patterns of a semiconductor device. The sample input images may include images of corner portions of the sample layout patterns. The deep learning model may be used to perform a corner rounding operation on the corner portions of the sample layout patterns.