Prediction of Single Event Effects in FinFET Devices Based on Deep Learning

Abstract

The Single Event Effect (SEE) of FinFET devices has become one of the challenging issues affecting the reliability of modern electronic systems in space and terrestrial applications. However, the conventional FinFET device simulation steps are tedious and take a long time. This paper proposes a method based on a convolutional neural network (CNN) to predict the single event effect (SEE) of FinFET devices. By entering different particle incidence conditions, the SEE profiles, as well as the characteristic parameters, can be obtained quickly and accurately. The neural network model used in the experiments has a high prediction accuracy. The error of our trained network model in predicting the drain transient current pulse profile is only 0.012, the Mean Square Error (MSE) for predicting the peak drain transient current and total collected charge are only 0.00207 and 0.00084. The total time for training, validation and prediction of the neural network model in this study is 352 seconds, and the prediction time is much shorter, which is much lower than the simulation time of TCAD software. The minimum simulation time of the TCAD simulation software is 1901 seconds, and the simulation requires further modification of the resultant plots of the single event effect transient current curves.

Keywords

• Deep learning
• neural networks
• FinFET
• single event effect (SEE)
• drain transient current pulse