| Presentation | 2022-11-11 Modeling of Super Steep Subthreshold Slope Device by using Neural Network Nakata Kengo, Mori Takayuki, Ida Jiro, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this study, we examined how PN-Body Tied (PNBT) Silicon On Insulator (SOI)-FETs, a novel device structure with steep subthreshold slope (SS) characteristics, should be modeled in a neural network. As a result, it was found that four or more hidden layers are required to represent super steep SS. The smoothness of the current derivative (gm), which is important for analog applications, was also examined. The results show that gm can be smoothed by using a Sigmoid function for the activation function. However, the Sigmoid function is known to have a gradient vanishing problem and is difficult to use in PNBT SOI-FETs because it is difficult to make it multi-layered. Therefore, we tested whether the model's gm could be smoothed by using a Rectified Linear Unit (ReLU), which is computationally less expensive than the Sigmoid function. We found that the combination of both functions may result in a smoother gm. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | neural network / SPICE / steep subthreshold slope |
| Paper # | SDM2022-73 |
| Date of Issue | 2022-11-03 (SDM) |
| Conference Information | |
| Committee | SDM |
|---|---|
| Conference Date | 2022/11/10(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Process, Device, Circuit simulation, etc. |
| Chair | Shunichiro Ohmi(Tokyo Inst. of Tech.) |
| Vice Chair | Tatsuya Usami(ASM Japan) |
| Secretary | Tatsuya Usami(Tohoku Univ.) |
| Assistant | Takuji Hosoi(Kwansei Gakuin Univ.) / Takuya Futase(SanDisk) |
| Paper Information | |
| Registration To | Technical Committee on Silicon Device and Materials |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Modeling of Super Steep Subthreshold Slope Device by using Neural Network |
| Sub Title (in English) | |
| Keyword(1) | neural network |
| Keyword(2) | SPICE |
| Keyword(3) | steep subthreshold slope |
| 1st Author's Name | Nakata Kengo |
| 1st Author's Affiliation | Kanazawa Institute of Technology(Kanazawa Inst. of Tech.) |
| 2nd Author's Name | Mori Takayuki |
| 2nd Author's Affiliation | Kanazawa Institute of Technology(Kanazawa Inst. of Tech.) |
| 3rd Author's Name | Ida Jiro |
| 3rd Author's Affiliation | Kanazawa Institute of Technology(Kanazawa Inst. of Tech.) |
| Date | 2022-11-11 |
| Paper # | SDM2022-73 |
| Volume (vol) | vol.122 |
| Number (no) | SDM-247 |
| Page | pp.pp.44-48(SDM), |
| #Pages | 5 |
| Date of Issue | 2022-11-03 (SDM) |