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 Wang | 1 | 2 | 3.96 |