Analysis and Prediction of Nanowire TFET’s Work Function Variation

Abstract

The research investigates the electrical effect of Work Function Variation (WFV) in Tunnel Field-Effect Transistors (TFETs), with Titanium Nitride (TiN) gate as a common Metal Gate material. Employing advanced Machine Learning (ML) techniques, this study seeks to establish causal relationships among various parameters, optimize ML models, and predict exceptional scenarios. Through an in-depth analysis of diverse data, the study uncovers insights into TFET’s performance variations. The ML model was optimized using the elimination method, checking each R2 value. After discovering the relevant output parameters (e.g., turn-on voltage (Von), threshold voltage (Vth)), it was observed that WFV at particular gate regions heavily affects current variation. Furthermore, ML demonstrated the ability to predict output parameters for exceptional cases, not present in the training data, such as gates composed of the 4.4-eV grain, which exhibited a high R2 value (0.9927).