Neural network model for implementation of electron–phonon scattering in nanoscale device simulations based on NEGF method

We propose a neural network (NN) model to implement the electron–phonon scattering in quantum mechanical transport simulations based on the nonequilibrium Green’s function (NEGF) method. As a representative example of nanoscale device, we consider the nanowire field effect transistors (NWFETs), although the proposed scheme can be applied any device structures. In particular, given the spectrum of lesser and greater Green’s functions at the source and the drain edges of the channel in the absence of the electron-phonon scattering, we constructed the NN to predict the lesser and greater Green’s functions in the presence of scattering, which are used to calculate the terminal current of the device. The proposed scheme has been successfully implemented to predict the trend of current against the scattering strength with meaningful accuracy.