Paper Info
Title
Identification of characteristics after soft breakdown with GA-Based neural networks
Abstract
In this research, we analyze the low-frequency noise power spectrum of drain current (Sid) in electrically stressed SiO2 film, and then propose the evolutionary neural networks-based model named ENN-SBD to identify the highly nonlinear degraded characteristics of low frequency noise around the soft breakdown (SBD). The Sid data follow the 1/fγ relationship with different value of power exponent γ. The spatial oxide traps distribution is proposed to account for the different γ value. It is found that the Sid correlates closely with the gate fluctuations via the trapping and detrapping processes and hence it is feasible to build the model represents the behavior of soft breakdown. The results also indicate that ENN-SBD has more precisely identification capability than typical Lorentzian spectrum method. Besides, it is superior to the backpropagation neural networks-based model (BNN-SBD) while the system identification is proceeding. This paper is helpful for breakdown detection and saving the cost of testing from quality assurance in the process of advanced CMOS technology.
Year
DOI
Venue
2006
10.1007/11779568_61
IEA/AIE
Keywords
Field
DocType
sid data,power exponent,breakdown detection,low frequency noise,identification capability,ga-based neural network,backpropagation neural networks-based model,soft breakdown,evolutionary neural networks-based model,different value,low-frequency noise power spectrum,neural network,system identification,power spectrum,spectrum,quality assurance
Data mining,Nonlinear system,Evolutionary algorithm,Computer science,CMOS,Electronic engineering,Spectral density,Artificial intelligence,Backpropagation,Artificial neural network,System identification,Genetic algorithm
Conference
Volume
ISSN
ISBN
4031
0302-9743
3-540-35453-0
Citations 
PageRank 
References 
0
0.34
2
Authors
1
Name
Order
Citations
PageRank
Hsing-Wen Wang123.96