Machine-Learning-Enhanced NEGF Solver of Interband Cascade Laser

NEGF simulation of the interband cascade laser is performed taking into account various scattering proceses. The Hartree potential, important for device operation, is calculated by solving the Poisson equation with an excess charge approach. Support vector machine technique is used to effectively improve the charge classification. Results are compared with other methods of classification, showing a significant improvement.

I. Introduction

An interband cascade laser (ICL), which utilizes a type-II quantum well active region, has gained significant and continuously growing interest as a source of coherent light in the mid-infrared (MIR) wavelength region [1]. It can be used in various applications such as: trace gas detection, industrial process control, medical diagnostics, and free-space communication. Additionally, an area where the tunable properties of ICLs can be utilized is tunable diode laser absorption spectroscopy [2]. High quantum efficiency of ICLs is achieved thanks to the broken-gap alignment and cascading scheme similar to that of intersubband quantum cascade lasers (QCLs). Furthermore, ICL design can reduce the Auger recombination through the band-structure engineering.