Transformer Neural Network-Based Design and Analysis of Nano-Scale Ferroelectric NCFETs With Low Subthreshold Swing

In this work, we employ Transformer Neural Network (TNN) to predict optimal materials and design parameters of nano-scale Ferroelectric Negative Capacitance Field-Effect Transistors (NCFETs) targeting low subthreshold swing (SS). The TNN model is trained on a dataset, encompassing various ferroelectric materials, gate oxide thicknesses, channel lengths, and other critical parameters. The predicted parameters are subsequently validated and refined using Finite Element Method (FEM) tools, which provide detailed insights into the electrostatic potential distribution, polarization effects, and device scaling challenges. This integrated TNN-FEM approach not only accelerates the design process but also enhances the accuracy of material and design selection, ensuring that the proposed NCFET structures meet the stringent requirements of next-generation low-power, high-speed electronics. The results indicate significant potential for NCFETs in overcoming fundamental limitations of conventional FETs.