A new technical paper, “A Device-Physics-Informed Artificial Neural Network Approach for Thermal-Aware I-V and C-V Modeling of GAA FETs,” was published by researchers at National Yang Ming Chiao Tung University.
Abstract
“This work introduces a device-physics-informed neural network framework for simultaneous modeling of thermal-aware I-V and C-V characteristics of gate-all-around (GAA) field-effect transistors (FETs). The approach integrates the Grove–Frohman I-V and Meyer C-V analytical expressions within artificial neural networks (ANNs), enforcing a shared temperature-dependent threshold voltage and embedding temperature effects in both ANN parameters and analytical expressions. A Gummel symmetry test is applied across temperatures to ensure current symmetry. Experimental validation demonstrates that the proposed method maintains high predictive accuracy even with limited training data, and generalizes well beyond the training conditions, while achieving significantly faster circuit simulation speed and reduced modeling effort compared to the well-known BSIM-CMG model. These results highlight the framework’s potential as a scalable, physically interpretable, and manufacturing-aware solution for advanced transistor modeling and circuit-level simulations.”
Find the technical paper here. May 2026.
Tai, Yun, Yiming Li, and Min-Hui Chuang. 2026. “A Device-Physics-Informed Artificial Neural Network Approach for Thermal-Aware I-V and C-V Modeling of GAA FETs.” IEEE Access, 1–1. https://doi.org/10.1109/access.2026.3689756.
Leave a Reply