A Neural Network Approach for Parameterizations of Hot Carrier Degradation Models

This study develops a parameter extraction flow for Hot Carrier Degradation (HCD) model in advanced technology based on the neural network (NN). Four types of parameters of the BSIM-CMG model are proposed to comprehensively capture the aged device characteristics. As verified by 16/14 nm FinFET data...

Full description

Saved in:
Bibliographic Details
Main Authors: Cong Shen, Yu Li, Zirui Wang, Lining Zhang, Runsheng Wang, Ru Huang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of the Electron Devices Society
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10749970/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This study develops a parameter extraction flow for Hot Carrier Degradation (HCD) model in advanced technology based on the neural network (NN). Four types of parameters of the BSIM-CMG model are proposed to comprehensively capture the aged device characteristics. As verified by 16/14 nm FinFET data, HCD-induced degraded characteristics can be well extracted with the NN approach. Furthermore, the dataset size, sampling schemes of dataset, activation functions and network structure are evaluated and optimized. Taking the PSO algorithm as the baseline, the accuracy and efficiency of the NN-based approach to parameter extraction are comprehensively studied and compared. The developed NN-based method is shown to be more effective for applications in the test data-intensive scenarios of reliability modeling.
ISSN:2168-6734