An Artificial Neural Network Model for Electro-Thermal Effect Affected Hot Carrier Injection Reliability in 14-nm FinFETs

Abstract

FinFETs have a bright future for advanced analog/RF and digital circuit design in 5G/6G technology. As the integration density increases, the self-heating effect (SHE) becomes more severe, and the hot carrier injection (HCI) reliability of FinFETs further deteriorates since it is highly dependent on temperature. In this article, an artificial neural network (ANN) model for HCI prediction is proposed for 14-nm FinFET with consideration of SHE under various voltage stresses and environment temperature. The HCI numerical simulation can be mainly divided into two parts: 1) the temperature responses for the device under different stresses are obtained by numerically solving time-dependent thermal conduction equation with heat generation in the nanoscale FinFET and 2) based on the transient temperature response to different stress voltages, HCI-induced threshold voltage shift (TVS) as a function of time is captured numerically. Our HCI-induced TVS simulation results agree well with others’ experiment results, which verifies the simulation methods. An ANN model is then developed to predict the TVS induced by HCI with the training data obtained from numerical simulations. The ANN model has good accuracy with an average relative error of 0.36% in predicting TVS and greatly reduces simulation cost once the ANN model is built for the HCI prediction of FinFETs, which has great potential for reliability design in circuits and systems.