Measuring with a high accuracy the size distribution of small metallic nanoparticles loaded in a mesoporous metal oxide matrix is of particular interest for many studies related to new generations of interesting metamaterials. Transmission electron microscopy (TEM) is a powerful tool to determine the nature and morphology of very small particles, but their reliable and automatic identification in an inhomogeneous environment where the nanoparticle/background contrast locally varies is not straightforward. Here, we present how a quantitative analysis of high-angle annular dark field scanning TEM (HAADF STEM) images, accounting for the chemical sensitivity of the technique, can improve the accuracy of semiautomatic segmentation methods based on morphological processing to calculate size histograms. The paper also provides an estimate of the reliability of this method through the analysis of numerically synthesized images. The latter are based on the simulation of HAADF STEM projections of a volume filled with titania, pores and silver particles, whose morphological features, such as dimensions, shapes and densities are evaluated from experimental measurements of real samples. The results obtained with synthesized images prove the performances of the quantitative analysis to suppress nonsilver nanoparticles from the statistics and allow to infer empirical rules to determine imaging parameters that ensure a good reliability of histograms.
Keywords: Ag:TiO2 composite films; HAADF-STEM; image segmentation; image synthesis; nanoparticle; size histogram; statistical analysis.
© 2017 The Authors Journal of Microscopy © 2017 Royal Microscopical Society.