Analysis of Negative-Bias Temperature Instability Utilizing Machine Learning Support Vector Regression for Robust Nanometer Design

Negative bias temperature instability, abbreviated as NBTI, has emerged as the primary reliability challenge for nanoscale PMOS transistors. In this paper, a complete and comprehensive analysis of 22nm PMOS transistor has been carried out for the purpose of determining the degradation of NBTI across both static and dynamic operations. The Support Vector Regression Machine Learning Model has been trained and tested across a set of a thousand data points, with eighty percent of the data serving as training and twenty percent serving as testing. A dataset has been gathered over the course of ten years, and the best-fit machine learning model earned the highest score of 0.99 by utilising Radial Basis (RBF) Function. In comparison, linear and polynomial-based modelling approaches achieved scores of 0.87 and 0.78, respectively.