[22p-P05-1~4] 13.1 Fundamental properties, surface and interface, and simulations of Si related materials

[22p-P05-1] Improving the Accuracy of Modeling Electron Dynamics in Semiconductor Devices Using 3D Convolutional Neural Networks

〇Satoru Sugiyama¹, Yoshitaka Itoh¹, Tota Suko², Masakazu Muraguchi¹ (1.Hokkaido Univ. of Sci., 2.Waseda Univ.)

Keywords:

convolutional neural network,electron dynamics,semiconductor device design

We have been working on developing a model using machine learning to predict the distribution of impurity potentials from the electron conduction process in a two-dimensional nanowire system. In this study, our objective was to improve the accuracy of the model by incorporating the time evolution data of electron density through convolution with a 3D filter. Evaluation results of predictions made using models based on 2-D and 3-D convolutional neural networks will be presented.