Show Abstract ▼

Abstract: Electron ptychography has recently shown great promise as an atomic-resolution imaging technique for all elemental species in both two and three dimensions. However, each ptychographic step forward is typically a result of months (or years) of sample preparation, data acquisition and reconstruction refinement. In this presentation, we will summarize three of the main experimental difficulties associated with performing electron ptychography, and provide several examples of unsuccessful experiments which emphasize these challenges. Detection: The correct use of fast pixelated detectors at appropriate electron dose levels and experimental settings is crucial to obtaining data that yields high-fidelity reconstructions. For example, by lowering the dynamic range of certain detectors and collecting sparse diffraction patterns, acquisition speeds can be increased tenfold [1]. Initial ptychography experiments using sparse data acquisition were performed on an MoS2 monolayer (80 kV, 22.5 mrad) at an electron dose of 10,000 e- Å-2, resulting in low signal-to-noise ratios and poor fidelity phase reconstructions (Figure 1A). However, upon choosing a thicker sample with larger periodic features (e.g. ZSM-5) and using a smaller convergence angle (7.5 mrad at 300 kV), the contrast improved significantly, even at electron doses of 1,000 e- Å-2 (inset of Figure 1A) [2]. For 2D materials such as graphene and MoS2, multiple frame acquisition and alignment can provide sufficient signal to image single atoms [3]. With larger defocus values of the probe, sparse data acquisition becomes insufficient for sampling the fine features in the diffraction patterns. Dimensions: The extension of ptychography to three dimensions has provided the promise of imaging single light atoms inside materials via inverse multislice methods, tomographic routines, or a combination of both [4–6]. Focused-probe ptychography experiments of monolayer graphene yielded an undiscovered ‘defect’ revealed by electron ptychography and optical sectioning, as shown in Figure 1B. However, the depth resolution was insufficient to determine accurate atomic z-coordinates, and data acquisition speeds may have been slower than the movement of atoms within the defect system. Damage: Obtaining sufficient signal while avoiding beam-induced sample changes is one of the greatest challenges to the ptychographic researcher. Recent experiments on gold nanoparticle superlattice assemblies were performed in an attempt to reveal the inter-particle ligand distribution (Figure 1C). Although some regions suggest ordering of ligands close to the nanoparticles (inset of Figure 1C), aggregation and ligand motion was evident during data acquisition. Future work will involve performing experiments under cryogenic conditions to mitigate ligand movement and prevent damage. To conclude the presentation, recent progress on deep learning diffractive imaging [7] and ptychographic atomic electron tomography will be discussed [8].

Show Abstract ▼

Abstract: Annular-dark field imaging is one of the most readily interpretable techniques in scanning transmission electron microscopy (STEM). The ADF detector integrates the intensity of incoherent, high-angle scattered electrons providing images with strong atomic number sensitivity that can be directly interpreted in terms of the sample’s atomic structure. The ADF contrast increases monotonically with the number of atoms in projection. This allows direct atomic counting to solve three-dimensional (3D) structures or determine composition at a given thickness. However, ADF STEM is not efficient for light elements detection because they scatter less strongly at high scattering angles. More suitable methods to image both light and heavy atoms are based on bright-field or phase-contrast techniques that collect low-angle scattered electrons. However, because these electrons scatter coherently, image quantification, in this case, is non-trivial. This talk will discuss quantification methods for two main (quasi-)coherent STEM imaging modes: annular- bright field (ABF) imaging and 4D STEM electron ptychography. In the case of ABF, image quantification is applied to the characterisation of the oxygen framework in lithium- rich Li1.2Ni0.2Mn0.6O2 (LNMO) layered cathodes. In this material, short-range oxygen sublattice distortions are evidence for oxygen participation in charge compensation mechanisms. These movements are of the order of picometres and, to be measured, require well-designed quantification methods. However, achieving picometre accuracy and precision is challenging in these materials because of the limited signal-to-noise ratio of the images imposed by the electron dose. The first part of this talk discusses how to achieve high-quality ABF quantification using a combination of experimental design and computational data processing including simultaneous ADF atom counting and multi-slice image simulations. The second part of the talk shows how these methods can be extended to the quantification of 4D STEM electron ptychography data using fast electron detectors. A four-dimensional dataset consist of a two- dimensional (2D) convergent beam electron diffraction pattern (CBED) recorded at every pixel of a 2D scan array. The exit wave resulting from the electron-specimen interaction can be restored from this dataset using electron ptychography. In analogy with the contrast in ADF, the exit wave phase is fully quantitative and is highly sensitive to the number of atoms in the sample and their position along the column. Here, phase and ADF quantification are used to count the number of light and heavy elements simultaneously. The technique is applied to determine the composition in a reduced CeO2-x nanoparticle by simultaneous counting the number of Ce and O atoms. Experimental design is fundamental for any quantification technique. The last part of this talk discusses design strategies for comparing ptychographic techniques at low dose and imaging of light biological materials. Also, it examines how to efficiently compare restored data based on the effect of the contrast transfer function for the recovered spatial frequencies.

Show Abstract ▼

Abstract: Hybrid organic-inorganic perovskites have high potential as materials for solar energy applications, but their microscopic properties are still not well understood. Atomic-resolution scanning transmission electron microscopy has provided invaluable insights for many crystalline solar cell materials, and we used this method to successfully image formamidinium lead triiodide [CH(NH2)2PbI3] thin films with a low dose of electron irradiation. Such images reveal a highly ordered atomic arrangement of sharp grain boundaries and coherent perovskite/PbI2 interfaces, with a striking absence of long-range disorder in the crystal. We found that beam-induced degradation of the perovskite leads to an initial loss of formamidinium [CH(NH2)2+] ions, leaving behind a partially unoccupied perovskite lattice, which explains the unusual regenerative properties of these materials. We further observed aligned point defects and climb-dissociated dislocations. Our findings thus provide an atomic-level understanding of technologically important lead halide perovskites.

Show Abstract ▼

Abstract: We report the application of focused probe ptychography using binary 4D datasets obtained using scanning transmission electron microscopy (STEM). Modern fast pixelated detectors have enabled imaging of individual convergent beam electron diffraction patterns in a STEM raster scan at frame rates in the range of 1000–8000 Hz using conventional counting modes. Changing the bit depth of a counting detector, such that only values of 0 or 1 can be recorded at each pixel, allows one to decrease the dwell time and increase the frame rate to 12.5 kHz, reducing the electron exposure of the sample for a given beam current. Atomically resolved phase contrast of an aluminosilicate zeolite (ZSM-5) is observed from sparse diffraction patterns with isolated individual electrons, demonstrating the potential of binary ptychography as a low-dose 4D STEM technique.