GAN-Based Physical Design Layout Pattern Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating synthetic physical design layout patterns in semiconductor manufacturing are limited by their reliance on predefined building blocks and hardcoded rules, requiring significant manual effort and being prone to errors, especially when transitioning from unidirectional to bi-directional features, which are necessary for advanced patterning technology nodes.
Innovation Solution
The use of generative adversarial networks (GANs) to automatically generate synthetic physical design layout patterns by training on unidirectional data, allowing for the creation of bi-directional patterns that resemble real data, thus expanding pattern libraries and evaluating manufacturability without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predefined building blocks and hardcoded rules are used to generate synthetic physical design layout patterns, then the generation process is systematic and rule-based, but the method requires significant manual effort and is prone to errors
Solution Approach 1:
The system uses GANs to enable automated pattern generation where the neural networks self-learn from training data and automatically generate synthetic layout patterns without requiring manual rule definition or human intervention, eliminating errors associated with manual methods
Solution Approach 2:
The patent replaces the mechanical approach of manually defining building blocks and hardcoded rules with an intelligent system using GANs that learn patterns automatically from data, substituting manual mechanical processes with automated neural network-based generation
2Adaptability or versatility
If traditional methods are used to generate synthetic patterns, then the process is straightforward with predefined rules, but it cannot effectively generate bi-directional patterns needed for advanced patterning technology nodes
Solution Approach 1:
The GAN system changes the approach from fixed rule-based parameters to learned parameters from training data, enabling the generation of bi-directional patterns by learning from unidirectional training data and automatically adapting to create patterns suitable for advanced patterning technology nodes
Solution Approach 2:
The GAN-based system provides universal pattern generation capability that can handle both unidirectional and bi-directional patterns, as well as various feature types (lines, spaces, fills), replacing multiple specialized tools with a single versatile system
3Productivity
If manual methods are used to expand pattern libraries, then the process allows for careful evaluation of each pattern, but it is time-consuming and limits the quantity of patterns that can be generated
Solution Approach 1:
The GAN system enables continuous automated generation of synthetic layout patterns without interruption or manual intervention, producing large quantities of patterns rapidly by continuously sampling from the learned distribution, eliminating the time losses associated with manual pattern creation
Data Source
AI summary
A method for generating physical design layout patterns includes selecting as training data a set of physical design layout patterns of features in a given layer of a given patterned structure and converting the physical design layout patterns into two-dimensional (2D) arrays comprising entries for different locations in the given layer of the given patterned structure with values representing presence of the features at the different locations. The method also includes training, utilizing the 2D arrays, a generative adversarial network (GAN) comprising a discriminator neural network and a generator neural network. The method further includes generating one or more synthetic 2D arrays utilizing the trained generator neural network of the GAN, a given synthetic 2D array comprising entries for different locations in the given layer of a new physical design layout pattern with values representing presence of the features at the different locations of the new physical design layout pattern.