Strategy for optimizing experimental settings for studying low atomic number colloidal assemblies using liquid phase scanning transmission electron microscopy

The effect of nanochannel length on in situ loading times of diffusion-propelled nanoparticles in liquid cell electron microscopy

Liquid cell transmission electron microscopy is a powerful tool for visualizing nanoparticle (NP) assemblies in liquid environments with nanometer resolution. However, it remains a challenge to control the NP concentration in the high aspect ratio liquid enclosure where the diffusion of dispersed NPs is affected by the exposed surface of the liquid cell walls. Here, we introduce a semi-empirical model based on the 1D diffusion equation, to predict the NP loading time as they pass through the nanochannel into the imaging volume of the liquid cell. We show that loading of NPs into the imaging volume of the liquid cell may take several days if NPs are prone to attach to the surface of the mm-long nanochannel when using an industry-standard flat microchip. As a means to facilitate mass transport via diffusion, we tested a liquid cell incorporating a microchannel geometry resulting in a NP loading time in the order minutes that allowed us to observe the formation of a randomly oriented self-assembled monolayer in situ using scanning transmission electron microscopy.

Automated calculations for computing the sample-limited spatial resolution in (scanning) transmission electron microscopy

MATLAB scripts were designed to compute the sample-limited spatial resolution in transmission electron microscopy (TEM) and scanning TEM (STEM) as a function of different microscopy parameters including the electron dose eD, sample geometry, and materials parameters. The scripts can be used to select the optimum microscopy modality and optimize the experimental conditions to achieve the best possible resolution considering the limitations set by both the electron optics and the examined sample. The resolution can be computed as function of the objective opening semi-angle α for TEM and detector opening semi-angle β for STEM. Optional code for computing a range over the sample thickness t or eD are provided as well, whereby the opening angle is optimized for each data point. The spatial resolution depends on the type of material of the nanoscale object (for example, gold or carbon nanoparticles), the type of matrix holding the objects (for example, water or ice), the depth of the nanoscale object inside the matrix, and eD. The optimization is consistent with the typical situation that carbon nanoparticles are best examined with TEM embedded in a thin matrix (t = 0.1 µm), while STEM is better suited for high atomic number objects such as gold nanoparticles in water, irrespective of t. The script also calculates the reduction of beam broadening in thick samples (t > 1 µm) using bright field STEM.