Organic two-dimensional (2D) crystals, including 2D polymers and layer-stacked structures of 2D covalent organic frameworks (COFs), show promise in various applications, from organic field-effect transistors (OFETs) and organic solar cells to gas filtration and catalysis. Understanding structure-function relationships requires atomic-scale structural elucidation. A key characteristic of 2D polymers and COFs is the designability of pore interfaces, where functional side groups can be incorporated to enable functionalities such as catalysis, ion transport, energy storage, and gas adsorption. Porosity can be tuned by incorporating side groups of different sizes without altering the skeleton structure. However, precise characterization of pore interfaces remains challenging. Furthermore, characterizing amorphous organic 2D materials is difficult due to the absence of long-range ordering and the lack of suitable characterization techniques. Aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) offers sub-Ångström resolution, but electron irradiation damage often limits its application to organic materials. The achievable resolution is restricted by specimen stability, preventing visualization of local structures like defects, grain boundaries, and delicate features such as side groups. In amorphous materials, the lack of long-range order reduces Bragg scattering, decreasing the signal-to-noise ratio and image visibility. While high electron fluence can improve contrast in amorphous inorganic materials, this approach is impractical for organic materials due to severe radiation damage. Various techniques, such as the low-dose approach combined with direct electron detectors, and methods to restrain damage (vitrification and encapsulation), have been developed for high-resolution imaging of 3D organic crystalline materials. However, most TEM studies on organic 2D crystals are conducted at 300 keV despite the benefits of lower voltages for inorganic 2D materials. The use of high acceleration voltages stems from the predominance of radiolysis (inelastic damage) in organic materials, where the inelastic scattering cross-section is inversely proportional to the square of the electron velocity. High voltage reduces radiolysis, but also reduces the ratio of elastic to inelastic scattering. Recent studies suggest that lower voltages (around 100 kV) can improve this ratio, potentially outweighing the detrimental effects of radiolysis in thin organic 2D crystals. This study investigates the optimal acceleration voltage for imaging 2D polymer thin films.

The literature review section extensively cites previous research on high-resolution imaging of organic and inorganic materials using TEM. It highlights the challenges posed by electron beam radiation damage to organic materials and the various strategies employed to mitigate this issue, such as low-dose approaches, sample vitrification, and encapsulation techniques. The review also discusses the existing understanding of the energy dependence of electron radiation damage in organic materials and the trade-off between reducing radiolysis (inelastic damage) and maintaining a sufficient proportion of elastically scattered electrons (carrying structural information). It emphasizes the limited use of lower acceleration voltages in imaging organic 2D crystals compared to inorganic counterparts and points out the recent findings suggesting potential advantages of lower voltages in improving the efficiency of electron usage in thin organic samples.

This study systematically investigated the optimal electron acceleration voltage for high-resolution AC-HRTEM imaging of two highly crystalline imine-based 2D polymer thin films (2D-PI-BPDA and 2D-PI-DhTPA), with thicknesses up to 60 nm. The optimization considered the critical fluence (the maximum electron dose before significant sample damage) and the proportion of elastically scattered electrons at different acceleration voltages (300 kV, 200 kV, 120 kV, and 80 kV). The critical fluence was determined by monitoring the fading of reflection intensities in selected area electron diffraction (SAED) patterns as the electron fluence increased. Machine learning techniques (a U-Net type neural network) were employed to automate the analysis of numerous SAED patterns, identifying Bragg reflections and generating integrated intensity profiles to determine critical fluence for various resolution ranges. The efficiency of electron usage was quantified as the ratio of the intensity of Bragg reflections to the total integrated intensity (including the central beam). An 'information coefficient' was defined to balance the critical fluence and electron usage efficiency, identifying the optimal voltage for obtaining the highest resolution. AC-HRTEM imaging was performed at the optimal acceleration voltage to achieve improved image resolution and contrast. Images were acquired with a fluence close to the predetermined critical fluence for the desired resolution range. Density functional tight-binding (DFTB) calculations were used to generate atomic models for image simulation, allowing comparison with experimental results. To investigate the nature of observed structural discrepancies, quantum mechanical calculations, based on SCC-DFTB theory, were performed by introducing additional TAPP molecules (molecular interstitial defects) into the 2D polymer framework to simulate various stacking modes. The results were then used to generate simulated images for comparison with experimental images. A similar methodology, including the use of a U-Net based neural network for automated node position identification, was used to analyze the short-range order in an amorphous polyimine thin film (a-PI).

The key findings of this study are: 1. **Optimal Acceleration Voltage:** The study identified 120 kV as the optimal acceleration voltage for achieving near-atomic resolution (1.9 Å) in AC-HRTEM imaging of the 2D polymer thin films. This voltage maximized the 'information coefficient', balancing the critical fluence and the proportion of elastically scattered electrons. 2. **Enhanced Image Contrast and Resolution:** Imaging at 120 kV resulted in significantly enhanced image contrast and resolution compared to higher voltages (300 kV, 200 kV). The improved contrast allowed image acquisition with lower defocus values, minimizing contrast delocalization and enabling sharper localization of the image signal on the molecular framework. 3. **Discovery of Molecular Interstitial Defects:** The enhanced resolution revealed unexpected molecular interstitial defects in both 2D polymer materials. These defects, identified as intercalated TAPP molecules in specific orientations (R-type in 2D-PI-BPDA and A-type in 2D-PI-DhTPA), were confirmed by DFTB calculations and image simulations. This finding represents a new type of defect not previously reported in 2D polymers. 4. **Distinguishing Functional Groups:** The high resolution at 120 kV enabled the distinction of linker molecules with and without hydroxyl groups (DhTPA vs. BPDA) based on differences in their full width at half maximum (FWHM) values, highlighting the potential for visualizing functional groups at the pore interface. 5. **Structural Analysis of Amorphous Material:** The optimized conditions were successfully applied to an amorphous polyimine thin film (a-PI). A U-net based neural network allowed quantitative analysis of short-range order, providing data on node positions, bond lengths, angles, and polygon distributions, demonstrating the potential of this approach for characterizing amorphous organic 2D materials.

The findings address the research question by demonstrating that a lower acceleration voltage (120 kV) is superior to conventionally used higher voltages (300 kV) for high-resolution imaging of 2D polymer thin films. This optimization improves both resolution and contrast, enabling the visualization of subtle structural features not previously observable. The discovery of molecular interstitial defects highlights the importance of using optimal imaging parameters to reveal unexpected structural features that could influence material properties. The ability to distinguish functional groups based on subtle differences in FWHM opens up new possibilities for directly correlating structure and function in these materials. The successful application of the optimized parameters to an amorphous material expands the applicability of this method to a broader range of 2D organic materials. This work contributes significantly to the understanding of 2D polymer structure and defect chemistry, providing valuable insights for the design and synthesis of new materials with tailored properties.

This study demonstrates the crucial role of acceleration voltage optimization in achieving high-resolution AC-HRTEM imaging of organic 2D materials. The optimal voltage of 120 kV yielded a 1.9 Å resolution, unveiling unexpected molecular interstitial defects and allowing differentiation of functional groups. The findings highlight the importance of revisiting established imaging parameters and utilizing advanced image analysis techniques, such as machine learning, to obtain a more comprehensive understanding of complex 2D material structures. Future research could focus on exploring the influence of these defects on material properties and expanding the method to a wider range of organic 2D materials with different thicknesses and functionalities.

While the study successfully identified an optimal acceleration voltage for high-resolution imaging, the results are specific to the studied imine-based 2D polymers and the amorphous polyimine. The optimal voltage may vary for other organic 2D materials with different chemical compositions, thicknesses, and degrees of crystallinity. The machine learning methods used for automated analysis rely on trained neural networks and the quality of the training data influences the accuracy of the results. The study focuses primarily on structural characterization and further investigations are needed to fully understand the impact of the identified defects on the physical and chemical properties of the materials.