Résumé
Nanoparticles are defined as elementary particles with a size between 1 and 100 nm for at least 50% (in number). They can be made from natural materials, or manufactured. Due to their small sizes, novel toxicological issues are raised and thus determining the accurate size of these nanoparticles is a major challenge. In this study, we performed an intercomparison experiment with the goal to measure sizes of several nanoparticles, in a first step, calibrated beads and monodispersed SiO$_2$ Ludox®, and, in a second step, nanoparticles (NPs) of toxicological interest, such as Silver NM-300 K and PVP-coated Ag NPs, Titanium dioxide A12, P25(Degussa), and E171(A), using commonly available laboratory techniques such as transmission electron microscopy, scanning electron microscopy, small-angle X-ray scattering, dynamic light scattering, wet scanning transmission electron microscopy (and its dry state, STEM) and atomic force microscopy. With monomodal distributed NPs (polystyrene beads and SiO$_2$ Ludox®), all tested techniques provide a global size value amplitude within 25% from each other, whereas on multimodal distributed NPs (Ag and TiO$_2$) the inter-technique variation in size values reaches 300%. Our results highlight several pitfalls of NP size measurements such as operational aspects, which are unexpected consequences in the choice of experimental protocols. It reinforces the idea that averaging the NP size from different biophysical techniques (and experimental protocols) is more robust than focusing on repetitions of a single technique. Besides, when characterizing a heterogeneous NP in size, a size distribution is more informative than a simple average value. This work emphasizes the need for nanotoxicologists (and regulatory agencies) to test a large panel of different techniques before making a choice for the most appropriate technique(s)/protocol(s) to characterize a peculiar NP.