Depth Dependence of the Spatial Resolution in Scanning Transmission Electron Microscopy Experiments

Abstract

Annular dark field scanning transmission electron microscopy (STEM) is capable of imaging thick specimens. The capability to image thick specimens is relevant, for example, for studying cells embedded in plastic section, polymeric materials in which nanoparticles are embedded, metallic samples containing several phases, and for research on liquid specimens. However, at sample thicknesses larger than the mean free path length for elastic scattering in the materials under investigation, significant scattering of the beam occurs that leads to beam broadening. This beam broadening results in a reduction of the spatial resolution that becomes more pronounced the deeper the focus of the electron probe is within the sample. In addition, the spatial resolution even of objects focused at the top of the specimen (with respect to a downward traveling electron beam) is reduced by thick materials underneath as the contrast and hence the signal‐to‐noise ratio decreases in thicker specimen at a given electron dose in STEM experiments. The dependency of the spatial resolution on the specimen thickness was already determined experimentally, via simulation, and in various analytical models [1] for objects at the top and at the bottom of thick specimen. An initial theoretical model was developed about the effective resolution obtained for an object at a certain depth within a scattering matrix [2] but experimental verification is lacking. In this work, we examine the effect of beam broadening on nanoparticles at specific vertical positions within thick samples. For our experiment, we chose to study gold nanoparticles embedded in a solid aluminum film as experimental model system. We deposited multiple layers of aluminum by physical vapor deposition on silicon chips featuring thin (50 nm), electron beam transparent silicon nitride windows in the center, through which the imaging was done. Gold nanospheres of 5–10 nm in diameter were placed between individual layers. By using solid aluminum as support material we benefit from immobilized gold nanoparticles at specific vertical positions in an electrically conducting and stable matrix. Gold nanorods were deposited on top and at the bottom of the aluminum film enabling us to determine the thickness of the aluminum film by tilting the specimen holder. The setup of our experiments is illustrated in Figure 1. The experiments were conducted using a C S ‐corrected STEM/TEM (ARM200f, JEOL, Japan) at 200 kV acceleration voltage. Our experiments confirmed that the vertical position of the gold nanoparticles within the aluminum matrix determines the spatial resolution. Particles positioned deeper within the Al matrix were imaged with a lower spatial resolution than those closer to the top surface, where the electron beam entered the specimen. This observation is illustrated in the electron micrographs in Figure 2. The same particles are shown from the different sides of the silicon chip with respect to the direction of the electron beam. In a) the Au nanoparticles are in a depth of 0.48 µm, while in b) they are below only 0.18 µm of Al in. The total thickness of the Al matrix is 0.62 µm. The spatial resolution was determined by analyzing intensity profiles over the particles. Interestingly, the change in the spatial resolution was only reflected by the distance from 25% to 75% of the maximum intensity (d 25‐75 ), but not in the full widths at half maximum of the intensity profiles (see Figure 2c). Thus the scattering mainly led to increased beam tails.