Electron Microscopy (EM) has become a tool of choice to quantify materials structure at high resolution. Four-Dimensional Scanning Transmission Electron Microscopy (4D-STEM) is a versatile diffraction EM technique that can provide a high spatial resolution map of material structures and properties. However, 4D-STEM datasets can contain hundreds of thousands of patterns and can thus be challenging to manually interpret. Robust computational tools are required to gain insights from 4D-STEM datasets.

In this work, we present a set of computational methods for interpreting 4D-STEM datasets. There are two broad categories these methods fall into. The first category, unsupervised learning, allows for similarities across a dataset to be mapped with no prior knowledge regarding the structures in the dataset. To perform unsupervised learning on 4D-STEM datasets, we implement feature engineering protocols that distill the information in each individual diffraction pattern down. We then analyze the performance of both Non-negative Matrix Factorization (NNMF) and consensus clustering on a set of simulated datasets to highlight the versatility of different feature extraction techniques.

The second category of computational methods presented in this work is based on template-matching procedures. If the crystal structure(s) in a material are known, templates can be used to quantify the structures present in each diffraction pattern. Here, we present preliminary work towards mapping the spatial distribution of phases in multi-phase materials using Non-negative Least Squares (NNLS). Together, unsupervised and template-matching approaches allow researchers using 4D-STEM to extract information from their datasets more readily.

Precisely controlled synthesis of nanostructures is heavily emphasized in the field of nanoscience, in large part due to the desire to control the size, shape, and terminating facets of nanoparticles for applications in catalysis, optics, and medicine. Direct control of the size and shape of solution-grown nanoparticles relies on an understanding of how synthetic parameters alter nanoparticle structures during synthesis. Synthesis conditions rarely yield uniform particles, but the heterogeneity in these populations is hard to quantify especially in respect to atomic structure. Defects are a particularly challenging element to quantify the influence of synthesis on due to the limited number of methods which can give insight into these features. In this thesis we analyze the development of defect structures in multiply twinned metal nanoparticles. We find that non-classical growth mechanisms occur during standard colloidal growth. In order to further study the conditions which cause these non-classical growth modes we analyze methods to automate the analysis of defects in nanoparticles. To this end we propose a machine learning based pipeline for the segmentation and stacking fault classification in nanoparticles. We demonstrate a flexible two step pipeline for analysis of high resolution transmission electron microscopy data, which uses a U-Net for segmentation followed by a random forest for detection of stacking faults. Our trained U-Net is able to segment nanoparticle regions from amorphous background with a Dice coefficient of 0.8 and significantly outperforms traditional image segmentation methods. Using these segmented regions, we are then able to classify whether nanoparticles contain a visible stacking fault with 86% accuracy. We further explore the role of architecture in achieving better segmentation of HRTEM micrographs. These studies in turn illuminate the challenges in generating a large enough dataset for neural network training in electron microscopy. To solve this we evaluate methods to minimize labelling requirements by using automated labeling methods. This strategy recasts the classification problem as a label noise problem, with the fraction of label noise problem. We analyze the impact of label noise rate and training method on the classification error of neural networks for SEM data. We compare the noise resistance of three training methods: standard network training, pretraining with a tiny, label noise-free dataset, and co-teaching. We find that the pretraining approach yields the most accurate results across label error rates. These developments will help in enabling automated analysis of nanomaterials.

Transmission electron microscopy (TEM) is an indispensable and versatile characterizationtool, offering high resolution and a wide range of techniques for probing the intricate details of materials at the atomic scale. It is the Swiss Army knife of characterization tools. TEM utilizes a beam of high-energy electrons transmitted through a thin specimen to produce highly resolved images, diffraction patterns, and energy spectra. This enables researchers the exploration of morphology, crystal structure, defects, composition, and electronic structure of materials with high spatial and energy resolution.

In traditional TEM imaging, bright-field and dark-field techniques provide valuable insightsinto the crystallographic orientations, defects, and strain within a material. Bright-field imaging generates a high-contrast image of the specimen’s features, while dark-field imaging selectively scatters electrons from specific crystal planes, making defects and dislocations more visible. These methods allow for the characterization of defects, grain boundaries, and phase boundaries, aiding in the understanding of material properties and behavior. Modern methods in TEM have expanded the capabilities even further and increasingly rely on custom data processing and “big data” analysis. Scanning nanodiffraction, performed in the scanning TEM (STEM) configuration, enables the local mapping of crystallographic information with nanoscale resolution, making it possible to analyze the strain distribution and lattice distortions in materials. Additionally, electron energy-loss spectroscopy enables the investigation of the energy losses experienced by the electron beam as it interacts with the sample, offering valuable information about elemental composition, chemical bonding, and electronic structure.

TEM’s ability to reveal atomic-scale details and its versatility in utilizing a variety of tech-niques make it an invaluable tool for scientists and researchers across numerous disciplines. The work presented in the following chapters relies on a selection of TEM methods, including scanning nanodiffraction, high-resolution TEM, Lorentz TEM, and monochromated electron energy-loss spectroscopy (EELS) and takes advantage of access to advanced microscopes at the National Center for Electron Microscopy. Elegant and accurate analysis of the data produced with these techniques required the development of custom codes or the utilization of emergent data processing pipelines.

In the subsequent chapters of this work, amorphous Ta (tantalum) and Tb-Co (terbium-cobalt) are analyzed using fluctuation electron microscopy (FEM), a technique that exhibits unique sensitivity to medium-range ordering in disordered materials. Simulated and de- posited Ta films are used to demonstrate the applicability of FEM data for measuring ori- ented strain. Through various analysis pathways, FEM allows us to determine the degree of medium-range ordering (MRO) and strain in amorphous thin films. FEM is applied to the study of magnetron sputtered amorphous Ta thin films at different tilt angles, reveal- ing isotropic strain and validating the approach through electron diffraction simulations. A custom workflow and analysis code enable the acquisition of tilted scanning nanodiffraction data, facilitating the determination of strain and ordering in different directions while ad- dressing potential strain-induced artifacts. Detection limitations for low-strain films are also determined through this method.

Additionally, changes in the perpendicular magnetic anisotropy of amorphous thin films ofTb-Co grown through magnetron co-sputtering are observed and tuned by varying growth and annealing temperatures. The magnetic anisotropy constant increases with higher growth temperatures but decreases when annealing temperatures exceed the growth temperature. Once again, FEM is employed to reveal that MRO increases with higher growth temperatures and decreases with higher annealing temperatures, indicating a correlation between magnetic anisotropy and local atomic ordering. These findings suggest that temperature-mediated adatom configurations during deposition result in preferential ordering along the growth direction, with oriented MRO aligning with larger anisotropy constants. Shifting the focus to obsidian glass in Chapter 4, we move away from previous discussions on lab-made amorphous materials and delve into the study of obsidian, a volcanic glass formed through the rapid cooling of silica-rich melt. Within the amorphous matrix of obsidian, nano-scale crystallites offer valuable insights into the flow kinetics and composition of the melt. To analyze these crystallites, we prepare samples with nanometer-thin edges using the conchoidal fracture of obsidian for TEM analysis. Employing techniques such as energy- dispersive spectroscopy, electron diffraction, and RDF analysis, we examine the nanolites within the amorphous matrix. Our analysis reveals patterns of specific cation depletion near Fe-oxide nanolites, highlighting the influence of nanolites on the nearby short-range ordering and atomic characteristics of the matrix. These effects result in measurable localized changes in the amorphous structure, evident in decreased mean nearest neighbor distances compared to the bulk matrix.

The final chapters of this thesis concentrate on deepening our understanding of the elec-tronic structure of lanthanide intermetallics and its connection to their bulk properties. We present a survey of lanthanide M4,5 edge data obtained using EELS from various members of the Ln2Co3Ge5 series, with a particular emphasis on Ce, Sm, Pr, Gd, and Yb. The EELS measurements are conducted on crystalline samples using an aberration-corrected TEAM I transmission electron microscope (TEM), and we analyze the collected data to establish re- lationships between the electronic structures of lanthanides and the behavior of Co in these compounds. Understanding the electronic and atomic structures of lanthanide systems is crucial for tailoring their properties to suit different scientific applications. Recent stud- ies have demonstrated that the conventional tight-binding model for f -electron materials is not universally accurate. Therefore, investigating the electronic and atomic structures of lanthanide systems provides valuable insights for designing and manipulating quantum materials with desired properties, considering the influence of itinerancy and crystal field interactions on magnetic properties and other characteristics. Furthermore, a specific case study focusing on monoclinic Sm2Co3Ge5, a member of the Ln2Co3Ge5 series, is presented. This study investigates the impact of twin grain boundaries on local atomic distortion and electronic structure. The measured magnetic moment exceeds the expected moment of trivalent Sm, implying a magnetic contribution from Co. Twinning, observed along (100) planes, is examined using monochromated EELS and high-resolution TEM. The analysis uncovers shear strain at the twin boundaries and distinct fine structure in the Co L2,3 edges, indicating 3d -electron hybridization and the occupation of states that are absent in bulk grain spectra. These findings offer valuable insights into manipulating the bulk properties of lanthanide-transition metal systems with complex magnetism. The observed change in electronic structure also provides an explanation for the elevated magnetic moment measured in the system.

A multitude of TEM methods are applied to the study of amorphous and magnetic materials,showcasing the versatility of TEM in diverse material systems. Amorphous materials, with their lack of long-range atomic ordering, present challenges in structural interpretation. To overcome this, statistical methods are employed to define average values that are correlated to bulk properties, providing valuable insights into the behavior of these materials. On the other hand, magnetic materials provide an exciting avenue for TEM characterization, allowing for investigations not only into crystal structure, but also electronic and magnetic domain structures. This multifaceted approach offers a clearer picture of the factors influencing magnetism across multiple length scales, facilitating a deeper understanding of magnetic materials and their properties.

Diamine-appended Mg2(dobpdc) (dobpdc4− = 4,4′-dioxidobiphenyl-3,3ʹ-dicarboxylate) and diamine-appended Mg2(olz) (olz4– = (E)-5,5′-(diazene-1,2-diyl)bis(2-oxidobenzoate)) metal– organic frameworks have emerged as promising candidates for carbon capture owing to their exceptional CO2 selectivities, high separation capacities, and step-shaped adsorption profiles, which arise from a unique cooperative adsorption mechanism resulting in the formation of ammonium carbamate chains. The cooperative CO2 adsorption behavior enables the application of CO2 separation using relatively small temperature swings with large CO2 working capacities. More importantly, this cooperativity can be tuned via the critical design of the frameworks and diamine structure. This dissertation discusses the development of new diamine functionalized variants of Mg2(dobpdc) and Mg2(olz) for point source CO2 capture, such as from coal flue gas and biogas, as well as direct air capture. Design and discovery of the novel materials are explored, as well as the construction of the new characterization system.Chapter 1 explores the new frameworks of the type diamine–Mg2(olz) that feature diverse diamines with bulky substituents and display desirable single-stepped CO2 adsorption across a wide range of pressures and temperatures. Previous materials appended with primary,secondary- diamines featuring bulky substituents in Mg2(dobpdc) exhibit excellent stabilities and CO2 adsorption properties. However, these frameworks display double-step adsorption behavior arising from steric repulsion between ammonium carbamate chains, which ultimately results in increased regeneration energies. Chapter 1 also discusses how the basicity of the pore-dwelling amine—in addition to its steric bulk—is an important factor influencing adsorption step pressure; furthermore, the amine steric bulk is found to be inversely correlated with the degree of cooperativity in CO2 uptake. One material, ee-2–Mg2(olz) (ee-2 = N,N-diethylethylenediamine), adsorbs >90% of the CO2 from a simulated coal flue stream and exhibits exceptional thermal and oxidative stability over the course of extensive adsorption/desorption cycling, placing it among the top-performing adsorbents to date for CO2 capture from a coal flue gas. Spectroscopic characterization and van der Waals-corrected density functional theory calculations support that diamine–Mg2(olz) materials capture CO2 via the formation of ammonium carbamate chains. These results point more broadly to the opportunity for fundamentally advancing materials in this class through judicious design. Chapter 2 discusses a new framework, pip2–Mg2(dobpdc) (pip2 = 1-(2- aminoethyl)piperidine), that exhibits two-stepped CO2 uptake and achieves an unusually high CO2 capacity approaching 1.5 CO2 per diamine at saturation. Previous CO2 adsorption mechanisms of diamine-appended Mg2(dobpdc) metal–organic frameworks have been shown to occur predominantly via a chemisorption mechanism involving CO2 insertion at the amine-appended metal sites, a mechanism that limits the capacity of the material to ~1 equivalent of CO2 per diamine. Analysis of variable-pressure CO2 uptake in pip2–Mg2(dobpdc) using solid-state nuclear magnetic resonance (NMR) spectroscopy and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) reveals that it captures CO2 via an unprecedented mechanism involving initial insertion of CO2 to form ammonium carbamate chains at half of the sites in the material, followed by tandem cooperative chemisorption and physisorption. Powder x-ray diffraction analysis, supported by van der Waals-corrected density functional theory, reveals that the physisorbed CO2 occupies a pocket formed by adjacent ammonium carbamate chains and the linker. Based on breakthrough and extended cycling experiments, pip2–Mg2(dobpdc) exhibits exceptional performance for CO2 capture under conditions relevant to the separation of CO2 from landfill gas. More broadly, these results highlight new opportunities for the fundamental design of diamine–Mg2(dobpdc) materials with even higher capacities than predicted based on CO2 chemisorption alone. Finally, Chapter 3 explore the application of diamine-functionalized Mg2(olz) for direct air capture, which involves capturing CO2 from dilute humid 400 ppm CO2 conditions. We designed a new material, 2-ampd–Mg2(olz) (2-ampd = 2-(aminomethyl)piperidine), that exhibits a cooperative CO2 adsorption profile with full capacity under direct air capture conditions. Analysis of the 1 bar 13CO2-dosed solid-state NMR spectroscopy and in situ DRIFTS data reveal that the material captures CO2 via a cooperative mechanism involving the formation of ammonium carbamate chains in the framework under the direct air capture conditions. In addition, breakthrough and long-term cycling experiments preformed with 2-ampd–Mg2(olz) reveal an exceptionally high capacity and operational stability. The oxidative experiments further confirm the resistance of the material toward degradation in operational environments. These results indicate that 2-ampd–Mg2(olz) is a promising candidate for achieving durable and high CO2 adsorption capacity in direct air capture applications.