Advances in electron microscopy have enabled visualizations of the three-dimensional (3D) atom arrangements in nano-scale objects. The observations are, however, prone to electron-beam-induced object alterations, so tracking of single atoms in space and time becomes key to unravel inherent structures and properties. Here, we introduce an analytical approach to quantitatively account for atom dynamics in 3D atomic-resolution imaging. The approach is showcased for a Co-Mo-S nanocrystal by analysis of time-resolved in-line holograms achieving ~1.5 Å resolution in 3D. The analysis reveals a decay of phase image contrast towards the nanocrystal edges and meta-stable edge motifs with crystallographic dependence. These findings are explained by beam-stimulated vibrations that exceed Debye-Waller factors and cause chemical transformations at catalytically relevant edges. This ability to simultaneously probe atom vibrations and displacements enables a recovery of the pristine Co-Mo-S structure and establishes, in turn, a foundation to understand heterogeneous chemical functionality of nanostructures, surfaces and molecules.

The relation between the energy-dependent particle and wave descriptions of electron-matter interactions on the nanoscale was analyzed by measuring the delocalization of an evanescent field from energy-filtered amplitude images of sample/vacuum interfaces with a special aberration-corrected electron microscope. The spatial field extension coincided with the energy-dependent self-coherence length of propagating wave packets that obeyed the time-dependent Schrödinger equation, and underwent a Goos-Hänchen shift. The findings support the view that wave packets are created by self-interferences during coherent-inelastic Coulomb interactions with a decoherence phase close to Δφ = 0.5 rad. Due to a strictly reciprocal dependence on energy, the wave packets shrink below atomic dimensions for electron energy losses beyond 1000 eV, and thus appear particle-like. Consequently, our observations inevitably include pulse-like wave propagations that stimulate structural dynamics in nanomaterials at any electron energy loss, which can be exploited to unravel time-dependent structure-function relationships on the nanoscale.

Carbon deposition on MgO-supported Pt nanoparticles from ethylene decomposition was studied by in situ transmission electron microscopy (TEM) at the atomic level. An imaging strategy was established for controlling beam-gas-sample interactions that minimizes beam-induced changes of the reaction. Using this strategy, it was possible to observe how carbon encapsulation occurs on Pt nanoparticles and the role of the Pt surface morphology. The evidence suggests that multiple partial layers grew simultaneously prior to full encapsulation of the nanoparticle. The growth of carbon on Pt nanoparticles was found to induce significant changes in the nanoparticle shape, resulting in particles becoming rounder as coking progressed. Closer examination of the surface structure revealed that in some cases carbon growth induced step formation.

The location of Co and Ni promoter atoms in industrial-style hydrotreating catalysts is examined by combining aberration-corrected scanning transmission electron microscopy and electron energy loss spectrum imaging. The observations unambiguously demonstrate that both Co and Ni promoter atoms occupy sites at all low-indexed edge terminations of hexagonally shaped multi-layer MoS2 nanocrystals. In contrast, similar observations for single-layer MoS2 nanocrystals show that Co-promoter atoms preferentially attach at the (-1 0 0) S-edge termination and are absent at the (1 0 0) Mo-edge termination. The apparent discrepancy between single- and multi-layer MoS2 nanocrystals can be explained by the 2H-MoS2 crystal structure, for which successive MoS2 layers alternatingly expose Mo- and S-edge terminations in any of the low-indexed directions. Thus, the multi-layer Co-Mo-S and Ni-Mo-S nanocrystals, formed in the present type of industrial-style hydrotreating catalyst, are consistently described as a superposition of single-layer Co-Mo-S and Ni-Mo-S structures, and in turn, provide promoted edge sites with different steric accessibility for the organic compounds in mineral oil distillates.

Reconstruction of the exit wave function is an important route to interpreting high-resolution transmission electron microscopy (HRTEM) images. Here we demonstrate that convolutional neural networks can be used to reconstruct the exit wave from a short focal series of HRTEM images, with a fidelity comparable to conventional exit wave reconstruction. We use a fully convolutional neural network based on the U-Net architecture, and demonstrate that we can train it on simulated exit waves and simulated HRTEM images of graphene-supported molybdenum disulphide (an industrial desulfurization catalyst). We then apply the trained network to analyse experimentally obtained images from similar samples, and obtain exit waves that clearly show the atomically resolved structure of both the MoS₂ nanoparticles and the graphene support. We also show that it is possible to successfully train the neural networks to reconstruct exit waves for 3400 different two-dimensional materials taken from the Computational 2D Materials Database of known and proposed two-dimensional materials.