Machine-learning-accelerated quantum transport study on the effects of superlattice disorder and strain in a midwave-infrared curved sensor
Phys. Rev. Applied 25, 054076 – Published 29 May, 2026
DOI: https://doi.org/10.1103/bz8d-9f4q
Abstract
An emerging device architecture for infrared imaging is the curved focal-plane array, which benefits from several optical advantages over the traditional flat design. However, the curving process introduces additional strain in the active region, which must be taken into account. Type-II superlattices, a promising alternative to traditional bulk materials for use in infrared photodetectors, are candidate materials for use in these devices, but the transport properties of these highly heterogeneous materials are not straightforward and can be affected by different material conditions, such as superlattice (SL) disorder and external strain. We present a comprehensive study of the internal quantum efficiency (QE) calculated for a curved device that incorporates finite-element-analysis (FEA) modeling, nonequilibirium Green’s function (NEGF) calculations, and Gaussian process (GP) regression. FEA is used for predicting the strain configuration throughout the active region induced by the curving procedure of the device. NEGF is used to calculate the vertical hole mobility for a select set of strain configurations, from which the internal quantum efficiency of the device is approximated to predict performance under strained conditions. This data set is then used to train a GP model that maps the QE predictions onto the spatial coordinates of the curved device, based on the strain configuration predicted using FEA. This analysis is performed for ideal and disordered SLs to understand both the fundamental and practical limitations of the performance of these materials in curved devices.
