Parametrization of emitter photoluminescence towards AI-based color center quantification

Quantum technology devices based on single photon emitters close to the surface or within photonic structures are affected by parasitic interface photoluminescence (PL) and interaction of the interfaces with the functional single photon emitters. However, a systematic surface emitter characterization for quantitative statistics is still missing. A recent approach to analyze color centers via deep learning techniques was limited to emitters generated deep in the bulk, since training data including a sophisticated model of the surface emitter statistics is missing. Our novel approach provides advanced statistical analysis of emitters in PL maps and a software framework with modules for such analysis and for the generation of model data on a large scale. We investigate the emitter statistics in pristine as well as He ion-irradiated 4H-SiC epitaxial layers and we statistically resolve distinct classes of surface-related and irradiation-related emitters. We show how to extract model parameters for training data generation and demonstrate the generation of accurate model data fitting the statistical characteristics of the experimental data.