Visibility of Nanoparticles in Liquid-Phase SEM via Monte Carlo Simulations

Liquid-phase scanning electron microscopy (LP-SEM) is a novel imaging technique that enables the in situ observation of specimens in their native liquid environment [1]. This technique utilizes specialized liquid cells enclosed with ultrathin electron-transparent membranes to introduce liquid-phase samples into the high vacuum specimen chamber of the microscope [2]. Compared to its counterpart in transmission electron microscopy (TEM) [3], LP-SEM benefits from having fewer sample size restrictions, which allows for the use of bulk specimens and larger liquid volumes that are more representative of the specimen’s natural environment. In addition, LP-SEM is advantageous for its lower electron beam energies, which reduce radiation damage via the direct displacement of atoms [4]. One of the main challenges in LP-SEM is ensuring that sufficient electron transmission is achieved through the membranes of the liquid cell apparatus. Electron transmission is influenced by many factors, including the atomic number of the specimen due to the Z-filtering effect [2]. Previous studies have compared the relative contrast of Au nanoparticles and bentonite clay through 20 nm SiN membranes [5]. However, there is a lack of systematic and quantitative study of the visibility of different materials in liquid that is directly linked to the physical properties, such as atomic number, density, band structure, and size. Such information can be critical for the feasibility of the experimental design, where an inappropriate combination of membrane composition and thickness may either impede the visualization of features of interest or lead to a catastrophic spill of the liquid into the specimen chamber. We investigated the secondary electron (SE) and backscattered electron (BSE) image contrast of nanoparticles in front of and behind a Si3N4 membrane using Nebula, a Monte Carlo simulation Package [6]. A point electron probe at a dose of 10 e-/Å2 was used to generate SE and BSE images at acceleration voltages of 3, 7, 10, 20, and 30 kV, but a higher dose of 4000 e-/Å2 was used to generate the line profiles through the centers of the particles to determine the contrast and spatial resolution. Electrons were filtered by energy and emission semi-angle (from 0 to 60°) to generate images of 181 by 121 nm with a pixel size of 1 nm. The signal-to-noise ratio (SNR) of the images were estimated using the SMART plugin adapted to ImageJ [7]. The spatial resolution was determined using the width at the 35th and 65th percentiles of the line profile. If resolution higher than 50 nm cannot be achieved, liquid SEM may not be viable or competitive as an option for the visualization of nanoscale features. We define contrast as the difference of the detected electrons per pixel divided by the maximum. Figure 1 (a) shows the geometry of the specimen for the simulations. The diameter of the particle and thickness of the membrane were based on a previous study by San Gabriel et al [5]. All interfaces were separated by a small gap of 0.1 nm to prevent multiple material specifications at the interface due to meshing. The materials investigated include C, Al, Ti, Fe, Cu, Ag, and Au to cover a broad range of atomic numbers. Figure 1 (b) and (c) present an example of the simulated BSE and SE micrographs of gold particles of 45 nm diameter at 5 kV, a default setting for routine SEM imaging. The images show that the clarity of the edges and overall contrast can be used to readily distinguish particles on top of the membrane in vacuum from the ones below in water, especially in SE (Figure 1 (c)). Figure 2 shows the simulated line profiles along the 45 nm particles of different materials in water at the same 5 kV. The FWHM values were between 37.5 and 39.5 nm, which are less than the actual size of the particle of 45 nm by over 10 %. The difference may be attributed to the combined effects of additional scattering in water and Si3N4 of the incident electrons and the reduced thickness along the incident electron paths near the edge of the particle. Contrast below 0.1 was found to be easily overwhelmed by noise and prominent features nearby such that it becomes imperceivable. In such cases, the full width half maximum (FWHM) and spatial resolution (Table 1) would not be available (NA). The spatial resolution limit for all materials with perceptible contrast was above 10 nm, but optical aberrations and detector efficiency will lower it in practice. The optimal acceleration voltage that balances resolution, contrast, and SNR for most materials is found to be at 7 kV as highlighted in Table 1. The simulation also suggests that carbon materials would have low contrast, which is consistent with the literature [2]. Figure 3 shows the trends of contrast and SNR of the particles of different materials in water. The contrast of BSE images is maximized for all materials at 5 or 7 kV and significantly reduced at 20 kV and above. This observation may be attributed to the suitable penetration depth of the electrons such that most scattering events contributing to BSE generation occur in the particles. The contrast and SNR increase with atomic number for both BSE and SE, suggesting that the observed SE contrast may be mostly generated by the BSEs. We also performed quantitative analyses on the experimentally captured images of Au nanoparticles at 5 kV (Figures 4 (a) and (b)) using the line profile method (Figures 4 (c) and (d)). The water and 20 nm Si3N4 membrane would reduce the spatial resolution and contrast (Table 2), and the resolution of the particles in water agree with the simulation results (Table 1). However, the SNR of the BSE micrograph is lower than that of the SE micrograph, which contradicts with the simulations. The discrepancy between the theoretical and experimental results can be attributed to the differences in the shape and size of the Au nanoparticles, the carbon contamination on the particle above the membrane, the optical aberrations and the electrostatic fields for SE signal enhancement of the instrument, and the contrast and gain settings used in the experiment. This work provides insights into the visibility of nanoparticles in an SEM liquid cell using BSE and SE signals, particularly in the control of contrast and SNR through accelerating voltage and electron dose. Despite the limitations of software, which takes no consideration of radiolysis, charging, contamination, optical aberrations, and external field effects, the simulated images still show qualitative similarities when compared to the images from real experiments. Future work may explore the effects of membrane design, liquid composition, particle size, and distance of the particle in liquid from the membrane. A detailed study on energy and angular distribution along with the contributions of different scattering mechanisms may facilitate the optimization of liquid cell designs and the standardization of imaging protocols [8]. (a) Schematic of the specimen geometry for simulations. Examples of simulated BSE and SE images of Au particles at 5 kV are (b) and (c), respectively. Scale bars: 50 nm. (a) BSE and (b) SE line profiles through the center of 45 nm particles in water made of different materials, obtained at an acceleration voltage of 5 kV. Simulated Spatial Resolution of the Nanoparticles of Different Elements below the Membrane Simulated Spatial Resolution of the Nanoparticles of Different Elements below the Membrane Simulated variations of (a) BSE contrast, (b) SE contrast, (c) BSE SNR, and (d) SE SNR with accelerating voltage for the nanoparticles of different elements below the membrane in water. BSE and SE micrograph of Au particles (top-left vacuum side, and bottom-right water side) imaged at an acceleration voltage of 5 kV using a Hitachi SU7000 SEM are (a) and (b), respectively. The images were generated with their corresponding in-lens detectors at a pixel size of 0.66 nm. Scale bars: 50 nm. BSE and SE line profile for the Au particles are (c) and (d), respectively. Experimental Spatial Resolution, Contrast, and SNR of the Au Nanoparticles at 5 kV Experimental Spatial Resolution, Contrast, and SNR of the Au Nanoparticles at 5 kV

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.