Modeling Nanowire Single-Photon Avalanche Detectors via Deep Neural Networks

Recently, single-photon avalanche detectors based on III-V compound semiconductor nanowires (NWs) have shown promising performance in emerging photonic quantum information technologies. However, their design and fabrication are ex-pensive and time-consuming. Predicting key device performance metrics for different nanowire structures before experimental prototyping is critical. We introduce a deep learning-based modelling framework named NW-SPAD-Net that predicts SPAD performance for given nanowire configurations several orders of magnitude faster than the conventional drift-diffusion modelling approach.