Hotspot detection focused on lithography induced defects becomes crucial at advanced node due to the increasing complexity of the design and manufacture process. Compared with traditional lithography simulation techniques for hotspot detection, machine-learning-based methods have shown significant advantages attributing to the efficiency and generality of their model. However, most convolutional neural network-based hotspot detector can only inference a layout pattern at once. Therefore, sampling clip patterns from the detected layout is the bottleneck of the whole process and determines the performance of hotspot detection. We designed a flow to generate filter rules by clustering analysis of known hotspots, which can efficiently extract layout clips as detected samples to hotspot classifier. We further propose a feature parametric optimization method to extract valuable graphic features for classifiers and reduce redundancy from context patterns. Experimental results demonstrate that these techniques improve the accuracy of hotspots detection.
With the VLSI technology shrinking to 7nm and beyond, the Redundant Local Loop (RLL), also known as via pillar, becomes a promising candidate of redundant via insertion due to its compatibility with the unidirectional layout style. Existing RLL insertion approaches only leverage rule-based heuristics for manufacturing constraints, which can no longer obtain a large enough Process Window (PW) in advanced technology nodes. It is imperative to develop new techniques to optimize lithography process window while inserting RLL to achieve a good yield. In this paper, we propose a machine learning-based litho-aware RLL insertion framework. Conventional lithography simulation requires tremendous computational resources to evaluate the lithography quality accurately, which is not feasible for process window exploration. We formulate the lithography simulation as a regression task and develop a customized Conventional Neural Network (CNN) architecture to predict the Depth of Focus (DOF), a standard metric for evaluating process window. We propose a complete ow for litho-aware RLL insertion based on the CNN model for process window evaluation. The commercial lithography simulator evaluates the effectiveness of the proposed framework. Experimental results demonstrate that our lithography model can predict the DOF with high accuracy and generalize well on unseen patterns while achieving orders of magnitude speedup compared to conventional lithography simulation. Our litho-aware RLL insertion framework can effectively improve the lithography process window with comparable runtime and insertion rate compared to the state-of-the-art method.
Background: In datasets for hotspot detection in physical verification, data are predominantly composed of non-hotspot samples with only a small percentage of hotspot ones; this leads to the class imbalance problem, which usually hinders the performance of classifiers.
Aim: We aim to enrich datasets by applying a data augmentation technique.
Approach: We propose a data augmentation flow-based generative adversarial network (GAN) to generate high-resolution hotspot samples.
Results: We evaluated our flow with the current state-of-the-art convolutional neural network hotspot classifier by comparison with conventional data augmentation techniques. Experimental results demonstrate that the accuracy improvement of our work can reach 3% at the same false alarm rate and the false alarm rate reduction can reach 5% at the same accuracy.
Conclusions: Our study demonstrates that rational hotspot classification can improve the efficiency of data. It also highlights the potential of GAN to generate complicated layout patterns.
With the development of process technology nodes, hotspot detection has become a critical process in integrated circuit physical design flow. The machine learning-based method has become a competitive candidate for layout hotspot detector with easy training and high speed. Classic methods usually define hotspot detection as a binary classification problem. However, the designer hopes to further divide the hotspot patterns into a series of levels according to their severity to identify and fix these hotspots. In this paper, we designed a multi-classifier based on the convolutional neural network to realize the detection of various levels of hotspot patterns. Unlike classic cross-entropy loss, we proposed a custom loss function to reduce the difference between false predicted levels and corresponding true levels, reducing the adverse effects caused by misclassified samples. Experimental verification results show that our hotspot detector can correctly classify various hotspots levels and has potential advantages for physical designers to fix hotspots.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.