acquire many insights in interesting topics such as interparticle interactions and colloidal self-assembly. [1][2][3] Liquid-cell (scanning) transmission electron microscopy [4,5] (LC(S)TEM) has recently emerged as a powerful tool to observe dynamic processes of nanoparticles (NPs) in liquid with nanometer spatial resolution. However, the electron beam significantly influenced the observed phenomena in many cases. So far, strongly slowed down diffusion of NPs was observed in LC(S)TEM studies. Possible explanations for this phenomenon, apart from trivial difficulties such as the imaging system not being fast enough to image free Brownian motion, include hydrodynamic slowing down near the window's surface, [10,16] a highly viscous ordered liquid layer near the windows, [10,15] and strong (sometimes beam-induced) interactions with the liquid-cell windows. [6,10,15,16,18] Observing 3D Brownian motion in the electron microscope that is not significantly altered by the electron beam and/or the presence of the windows would open the way for many experiments, including studies on colloidal self-assembly of NP dispersions. [30] The objective of this work is to find conditions and identify key experimental parameters for which 3D Brownian motion is observable in LC(S)TEM.In this study, we combine a low dose scanning transmission electron microscopy (STEM) technique with viscous liquid media having a high dielectric constant to observe bulk diffusion of gold NPs and titania particles in LC(S)TEM. The significantly faster diffusion of particles in comparison to many previous liquid-cell electron microscopy studies that we report on in this work underlines the importance of choosing a suitable electron microscopy imaging technique, electron dose rate and solvent in order to study dynamic processes in LC(S)TEM without artefacts.
Results and DiscussionFor this work, we studied two different systems. One with bigger particles in a less viscous solvent and one with smaller particles in a more viscous solvent. The bigger particles serve as a first check whether free diffusion is at all possible within the
In theory, liquid-cell (scanning) transmission electron microscopy (LC(S)TEM) is the ideal method to measure 3D diffusion of nanoparticles (NPs) on a single particle level, beyond the capabilities of optical methods. However, particle diffusion experiments have been especially hard to explain in LC(S) TEM as the observed motion thus far has been slower than theoretical predictions by 3-8 orders of magnitude due to electron beam effects. Here, direct experimental evidence of undamped diffusion for two systems is shown; charge-neutral 77 nm gold nanoparticles in glycerol and negatively charged 350 nm titania particles in glycerol carbonate. The high viscosities of the used media and a low electron dose rate allow observation of Brownian motion that is not significantly altered by the electron beam. The resulting diffusion coefficient agrees excellently with a theoretical value assuming free diffusion. It is confirmed that the par...