Device Modeling Based on Cost-Sensitive Densely Connected Deep Neural Networks

Device Modeling Based on Cost-Sensitive Densely Connected Deep Neural Networks

Engineers used TCAD tools for semiconductor devices modeling. However, it is computationally expensive and time-consuming for advanced devices with smaller dimensions. Therefore, this work proposes a machine learning-based device modeling algorithm to capture the complex nonlinear relationship betwe...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiaoying Tang, Zhiqiang Li, Lang Zeng, Hongwei Zhou, Xiaoxu Cheng, Zhenjie Yao
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Journal of the Electron Devices Society
Subjects:
Machine learning
device modeling
technology computer-aided design (TCAD) simulation
cost-sensitive densely connected DNN
Online Access:https://ieeexplore.ieee.org/document/10643157/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Engineers used TCAD tools for semiconductor devices modeling. However, it is computationally expensive and time-consuming for advanced devices with smaller dimensions. Therefore, this work proposes a machine learning-based device modeling algorithm to capture the complex nonlinear relationship between parameters and electrical characteristics of gate-all-around (GAA) nanowire field-effect transistors (NWFETs) from technology computer-aided design (TCAD) simulation results. This method utilizes a densely connected deep neural networks (DenseDNN), which establishes direct connections between layers in the neural networks, provides stronger feature extraction and information transmission capabilities. By incorporating cost-sensitive learning methods, the proposed model focus more on the critical data that determines device characteristics, leading to accurate prediction of key device characteristics under various parameters. Experimental results on a test dataset of 116 NWFETs demonstrate the effectiveness of this method. The DenseDNN model with cost-sensitive learning exhibits better performance than traditional deep neural networks (DNN) with various widths and depths, with a prediction error below 1.62%. Moreover, compared to TCAD simulation results, the model can speedup <inline-formula> <tex-math notation="LaTeX">$10^{6}\times$ </tex-math></inline-formula>.
ISSN:2168-6734

Similar Items