14 June 2022 in IEEE Transactions on Electron Devices
A new deep learning (DL)-based parameter extraction method is presented in this brief; 50k training cases are generated by Monte Carlo simulations of these preselected parameters in Berkeley short-channel IGFET model (BSIM)-common multigate (CMG). DL models are trained using backward propagation with ${C} _{\text {gg}} - {V} _{g}$ and ${I} _{d} - {V} _{g}$ as the input and selected BSIM-CMG parameters as the output. A TCAD simulated FinFET device, calibrated to Intel 10-nm node, is used to test the DL models. The DL-based parameters extraction results show an excellent fit to capacitance and drain current data, with 0.16% rms error in ${C} _{\text {gg}} - {V} _{g}$ and 6.1% rms error in ${I} _{d} - {V} _{g}$ (0.69% rms error in above-threshold-voltage ${I} _{d} - {V} _{g}$ ), respectively. In addition, devices with a 10% variation in gate length and oxide thickness are successfully modeled with the trained DL model. The results show tremendous promise in using the DL-based models for parameter extraction.
10 August 2022 in IEEE Electron Device Letters
A fast and accurate deep learning (DL) based ASM-HEMT I-V model parameter extraction is presented for the first time. DL-based extraction starts with 120k training data-sets comprising of 374 million I-V data points. Training data-sets are generated through Monte Carlo simulations. The trained DL-model is demonstrated to successfully model 114 GaN HEMTs from a typical GaN fabrication process. The predicted parameters show an excellent fit for the I-V data. In addition, the root-mean-square(RMS) error incurred for key electrical parameters such as pinch-off voltage, linear condition current and the maximum current is 2.2%, 17.6%, and 2.4% respectively. The proposed approach is verified for multiple GaN HEMTs of different sizes. The developed technique can provide a very fast means for parameter extraction with a reasonable accuracy.
7 September 2023 in Solid-State Electronics
4 January 2024 in IEEE Microwave and Wireless Technology Letters
A new machine learning (ML)-based large-signal parameter extraction for ASM-HEMT model has been presented for the first time. The proposed technique uses a 20k training sample generated by Monte Carlo simulations. The training samples of simulated output power $P_{\text{out}}$ and power-added efficiency (PAE) are used to train an ML extractor to extract the ASM-HEMT model parameters. The trained ML extractor has been evaluated on measurements performed on a commercial GaN device which was previously modeled using ASM-HEMT using manual extraction. The results show that the ML extractor could extract ASM-HEMT large-signal parameters to model $P_{\text{out}}$ , gain, and PAE, producing a level of accuracy comparable to the conventional manual parameter extraction. The proposed parameter extraction technique takes less than a second while removing the complexity and the need for expertise for extraction. This shows the promise of ML toward parameter extraction for large-signal models.
14 June 2025 in Solid-State Electronics
19 February 2026 in Microelectronics Reliability