Abstract
An I-V global parameter extraction technique for the industry standard FinFET compact model BSIM-CMG using deep learning (DL) is presented in this paper. The training data of 750k is generated by Monte Carlo simulation of key BSIM-CMG Parameters and gate length (LG) for multiple devices. The created deep learning parameter extractor is trained to use I-V and LG data to predict the BSIM-CMG parameters. The DL parameter extractor is verified using measured device data, with LG ranging from 50n to 970nm. The created global model was able to create an accurate fitting for the input characteristics of multiple devices while capturing the trends in key electrical parameters. The results show the tremendous potential of using DL to create accurate global models instantly where measurement and manufacturing errors are present.
| Original language | English |
|---|---|
| Title of host publication | 2023 IEEE International Symposium on Radio-Frequency Integration Technology (RFIT) |
| Place of Publication | Piscataway, NJ |
| Publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| Pages | 20-22 |
| Number of pages | 3 |
| ISBN (Electronic) | 9798350324402 |
| ISBN (Print) | 9798350324419 |
| DOIs | |
| Publication status | Published - 2023 |
| Event | 2023 IEEE International Symposium on Radio-Frequency Integration Technology, RFIT 2023 - Cairns, Australia Duration: 14 Aug 2023 → 16 Aug 2023 |
Conference
| Conference | 2023 IEEE International Symposium on Radio-Frequency Integration Technology, RFIT 2023 |
|---|---|
| Country/Territory | Australia |
| City | Cairns |
| Period | 14/08/23 → 16/08/23 |
Keywords
- Parameter Extraction
- Berkeley Short-channel IGFET Model – Common Multi-Gate (BSIM-CMG)
- fin fieldeffect transistor (FinFET)
- deep learning
- compact model