Correlative Fluorescence and Transmission Electron Microscopy Assisted by 3D Machine Learning Reveals Thin Nanodiamonds Fluoresce Brighter

Abstract

Nitrogen vacancy (NV) centers in fluorescent nanodiamonds (FNDs) draw widespread attention as quantum sensors due to their room-temperature luminescence, exceptional photo- and chemical stability, and biocompatibility. For bioscience applications, NV centers in FNDs offer high-spatial-resolution capabilities that are unparalleled by other solid-state nanoparticle emitters. On the other hand, pursuits to further improve the optical properties of FNDs have reached a bottleneck, with intense debate in the literature over which of the many factors are most pertinent. Here, we describe how substantial progress can be achieved using a correlative transmission electron microscopy and photoluminescence (TEMPL) method that we have developed. TEMPL enables a precise correlative analysis of the fluorescence brightness, size, and shape of individual FND particles. Augmented with machine learning, TEMPL can be used to analyze a large, statistically meaningful number of particles. Our results reveal that FND fluorescence is strongly dependent on particle shape, specifically, that thin, flake-shaped particles are up to several times brighter and that fluorescence increases with decreasing particle sphericity. Our theoretical analysis shows that these observations are attributable to the constructive interference of light waves within the FNDs. Our findings have significant implications for state-of-the-art sensing applications, and they offer potential avenues for improving the sensitivity and resolution of quantum sensing devices.