Localized plasmon resonances in 2D transition metal dichalcogenides (TMDs) offer a powerful means to enhance light–matter interactions at the nanoscale, making them ideal candidates for advanced optoelectronic applications. However, disentangling the complex plasmonic interactions in these materials, especially in the low-energy regime, presents significant challenges. Herein, localized plasmon resonances in chemical vapor deposition-grown tungsten disulfide (WS2) nanotriangles, using a combination of advanced spectral analysis and simulation techniques, is investigated. By combining non-negative matrix factorization with electron energy loss spectroscopy, distinct plasmonic modes to provide a comprehensive analysis of the plasmonic landscape of individual and stacked WS2 nanotriangles are identified and characterized. Furthermore, the dispersion relation of these localized plasmon resonances is quantified and their evolution across different WS2 triangular geometries is evaluated. Experimental characterization of plasmonic resonances in WS2 through dedicated numerical simulations based on the PyGDM package is validated. The findings highlight the critical role of localized plasmon resonances in modulating the electronic and optical properties of WS2, offering new insights into the design and optimization of TMD-based devices for optoelectronic and nanophotonic applications.@en

Two-dimensional (2D) layered materials are integral to modern condensed matter research due to their remarkable electronic and optical properties. A key feature of these materials is that their properties can be adjusted bymaking small changes to their structure at the nano- and atomic scale. Understanding and linking these electronic and optical properties to structural features at the nanoscale is crucial for unlocking the full potential of 2D layered materials and maximizing their use in advanced devices. This thesis uses electron-based microscopy and spectroscopy to achieve the high spatial and energy resolution required for this goal. These techniques address the limitations of optical and X-ray spectroscopy, which, while offering excellent spectral resolution, lack the spatial precision needed to resolve nanoscale morphologies and atomic structures critical for understanding 2Dmaterials. To achieve this,we employ two advanced electron microscopy methodologies: probe corrected Scanning Transmission Electron Microscopy (STEM) and monochromated Electron Energy-Loss Spectroscopy (EELS). Together, these techniques enable the acquisition of high-quality Spectral Images (SIs) with both exceptional spatial and spectral resolution, providing a powerful platform for the detailed characterization of 2D layered materials. To further enhance the potential of STEM-EELS, we integrate Machine Learning (ML)-based approaches. These approaches introduce innovative solutions such as the removal of the dominant Zero Loss Peak (ZLP) background in the low-loss energy region, peak identification and multivariate techniques to separate overlapping signals and so fully leverage the rich information contained in STEM-EELS SIs Chapter 2 establishes the data processing methodology used in this work. It provides an overview of STEM-EELS SIs, detailing how they are acquired, interpreted, and the challenges involved in processing these high-dimensional datasets. A key focus is on ourML-based approach for image-wide subtraction of the ZLP in SIs. This step is crucial for isolating spatially localized information in the low-loss energy region, which would otherwise be obscured by the ZLP tail. The methodology incorporates ML techniques originally developed in high-energy physics for probing the interior structure of protons, demonstrating the adaptability of these methods to electron microscopy. This analysis framework, named EELSFITTER, serves as the foundation for the remainder of the thesis, where it is applied to the characterization of 2D layered materials. Developed in Python, the framework is open-source and freely available for use by the research community. In Chapter 3, the framework EELSFITTER is applied to investigate Indium Selenide (InSe) nanosheets and Tungsten Disulfide (WS2) flakes with mixed polytipism (2H/3R). The thickness and stacking order of layers are critical structural features that influence the optoelectronic properties of 2D materials, including their band gap. For InSe, the stacking order or crystalline phase determines whether the band gap is direct or indirect and affects its value. In the case of WS2, a member of the Transition Metal Dichalcogenides (TMDs) family, thickness plays a direct role in tuning the band gap, making it an ideal benchmark for validating the ML-based approach. Using robust ZLP subtraction in the SIs of these materials, we achieve nanoscale precision in spatially resolving their band gap and dielectric function. Additionally, we correlate the electronic properties to structural features, with a particular focus on local specimen thickness, demonstrating the effectiveness of this methodology. We extend the data processing techniques and analytical methods to tackle automated feature identification within the energy-loss and energy-gain region of EELS in Chapter 4. The first part of this chapter focuses on one-dimensional (1D) Molybdenum Disulfide (MoS2) nanostructures. As a TMD material similar to WS2, MoS2 in a 1D morphology allows us to study the effects of curvature-induced strain on its optoelectronic properties. We characterise excitonic and plasmonic resonances, revealing how these features are influenced by the 1D geometry. Additionally, we investigate excitonic behaviour and the band gap value in relation to localized curvature-induced strain, comparing the properties at the tips of the 1D structures with those at the body. The second part of the chapter examines the layered topological insulator Bismuth Telluride Bi2Te3. Here, we focus on the energy-gain region, applying ML-based techniques originally developed formodelling the loss region of the ZLP.Using this approach, we extract a well-defined collective excitation at -0.8 eV on the energy-loss axis. By relying on the energy-gain region, we avoid complications from multiple scattering, enabling the characterization of this excitation with enhanced spectral precision. This chapter highlights the versatility of our methods for analyzing diverse materials and morphologies. In Chapter 5, we focus on WS2 nanotriangles, examining localised plasmonic resonances that form along their edges. By employing non-negative matrix factorization (NMF), we identify the spatial distribution of these resonances and successfully separate them from signals originating from overlapping WS2 nanotriangles. The results of the NMF analysis are compared with electrodynamical simulations, which reveal strong agreement with the observed localized plasmonic resonances. Further,we quantify these resonances by analysing their dispersion relation through a 1D Fabry-Perot model. This analysis demonstrates a quadratic dispersion characteristic of surface plasmonic phenomena, offering deeper insights into the optical behaviour of WS2 nanotriangles. This thesis presents the development of novel strategies for processing and interpreting STEM-EELS SIs in both the low energy-loss and energy-gain regions. Through these advancements, we provide valuable insights into the relationship between structural and physical properties across various morphologies and material types of layered materials. Importantly, all computational frameworks developed during this work are open-source and freely available, ensuring that the methodologies and approaches can be easily adopted by other researchers...@en

Among the many potential applications of topological insulator materials, their broad potential for the development of novel tunable plasmonics at THz and mid-infrared frequencies for quantum computing, terahertz detectors, and spintronic devices is particularly attractive. The required understanding of the intricate relationship between nanoscale crystal structure and the properties of the resulting plasmonic resonances remains, however, elusive for these materials. Specifically, edge- and surface-induced plasmonic resonances, and other collective excitations, are often buried beneath the continuum of electronic transitions, making it difficult to isolate and interpret these signals using techniques such as electron energy-loss spectroscopy (EELS). Here we focus on the experimentally clean energy-gain EELS region to characterise collective excitations in the topologically insulating material Bi₂Te₃ and correlate them with the underlying crystalline structure with nanoscale resolution. We identify with high significance the presence of a distinct energy-gain peak around −0.8eV, with spatially-resolved maps revealing that its intensity is markedly enhanced at the edge regions of the specimen. Our findings illustrate the reach of energy-gain EELS analyses to accurately map collective excitations in quantum materials, a key asset in the quest towards new tunable plasmonic devices.

@en

Twisted 2D materials present an enticing platform for exploring diverse electronic properties owning to the tunability of their bandgap energy. However, the intricate relationship between local heterostrain fields, thickness, and bandgap energy remains insufficiently understood, particularly at the nanoscale. Here, it presents a comprehensive nanoscale study elucidating the remarkable sensitivity of the bandgap energy to both thickness and heterostrain fields within twisted WS₂ nanostructures. This approach integrates electron energy-loss spectroscopy (EELS) enhanced by machine learning with 4D scanning transmission electron microscopy (STEM). Through this synergistic methodology, enhancements up to 20% in the bandgap energy is unveiled depending on the specimen thickness. This phenomenon is traced back to sizable deformation angles present within individual layers, which can be directly linked to distinct variations in local heterostrain fields. The findings represent a significant advancement in comprehending the electronic behavior of twisted 2D materials and introduce a novel methodological framework with far-reaching implications for twistronics and the investigation of other materials within the nanoscience domain.

@en

This study presents an in-depth investigation of the electronic properties and bandgap energy distribution in 1D molybdenum disulfide (1D-MoS₂) nanostructures. Through a combination of high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and electron energy-loss spectroscopy (EELS), it reveals significant differences between 1D-MoS₂ nanostructures and their 2D counterparts, shedding light on their localized exciton behavior and their bandgap energy modulation within the nanostructures. Excitonic peaks at around 2 and 3 eV appear localized at the ends or along the sides of the 1D-MoS₂ nanostructures, while the plasmonic resonance at 8.3 eV retains its inner-region localization. It demonstrates the spatial dependence of the bandgap energy, with the central region exhibiting a bandgap of approximately 1.2 eV, consistent with bulk MoS₂, while regions characterized by curvature-induced local strain fields exhibit instead a noticeable reduction. The findings provide valuable insights into the intricate relationship between excitonic behavior and bandgap sensitivity in 1D-MoS₂ nanostructures, streamlining the design and optimization of nanophotonic and optoelectronic devices.

@en

The electronic properties of two-dimensional (2D) materials depend sensitively on the underlying atomic arrangement down to the monolayer level. Here we present a novel strategy for the determination of the band gap and complex dielectric function in 2D materials achieving a spatial resolution down to a few nanometers. This approach is based on machine learning techniques developed in particle physics and makes possible the automated processing and interpretation of spectral images from electron energy loss spectroscopy (EELS). Individual spectra are classified as a function of the thickness with K-means clustering, and then used to train a deep-learning model of the zero-loss peak background. As a proof of concept we assess the band gap and dielectric function of InSe flakes and polytypic WS2 nanoflowers and correlate these electrical properties with the local thickness. Our flexible approach is generalizable to other nanostructured materials and to higher-dimensional spectroscopies and is made available as a new release of the open-source EELSfitter framework.

@en