Quantitative Analyses of Mesopores in Carbon Electrode Materials via Low-Voltage SEM

Porous carbons are promising electrode materials for energy storage devices due to their high surface areas, tailored pore structures and good physicochemical stability. In many cases, changes in the pore structure can greatly impact the energy and power density of the overall device [1,2]. Traditional gas physisorption methods for pore size and specific surface area can obtain an average over the entire sample, and local variations and spatial arrangements of the pores are not available. While preparing electron-transparent thin sections of porous carbons for transmission electron microscopy (TEM) is feasible, the process can be time-consuming and expensive, especially when the objective is to analyze a variety of samples. The transmitted electrons (TE) contain sub-surface information, but the phase contrast can complicate the interpretation of data, especially for materials that are both porous and amorphous. Moreover, the high incident electron energies in conventional TEM often exceed the knock-on damage threshold of carbon [3]. There exists a need for high-throughput and spatially specific porosity analysis to gain a holistic understanding of the processing-structure-property relationship. Scanning electron microscopy (SEM) has the unique advantage of intuitively bridging macro and nanoscale information with simple preparation. SEM reveals the surface topography because secondary electrons (SE) cannot escape from the bulk of the material [4]. Simplifying and automating the SEM analysis protocol can enhance its repeatability and reliability to accelerate the discovery and optimization of new materials. We investigated the feasibility of using low-voltage SEM for quantitative mesopore analysis by imaging commercial mesoporous carbon CMK-8 (ACS Material) synthesized from a three-dimensionally ordered silica template [5]. The sample was sprinkled on a 1-inch aluminum stub and fixed with carbon paint. Figure 1 shows an example of the mesopore analysis workflow. SE images were captured in line integration mode using a Hitachi SU7000 Schottky field emission SEM in beam deceleration mode at a landing voltage of 200 V, with an image size of 508 by 356 nm, a probe current of 7 pA, and a total scan time of 34 s (Figure 1 (a)). The spatial resolution and signal-to-noise ratio (SNR) of the original images were estimated using the methods by D. C. Joy implemented in FIJI (Figure 1 (b)) [6, 7]. All images need to have a spatial resolution that exceeds the expected pore size before subsequent analyses. Image denoising was achieved via a median filter at a radius of 2 pixels, and Contrast Limited Adaptive Histogram Equalization (CLAHE) [8] was applied with a block size of 65 and a maximum slope of 1.5 (Figure 1 (c)) to enhance the accuracy of both manual and automatic identification of the pores. The pore contours were then extracted using the Weka segmentation [9] (Figure 1 (d)) model trained on manually labelled micrographs. The extracted contours can then be used to calculate the pore size (major axis of an elliptical fit of the pore shape), aspect ratio (major axis divided by minor axis), and the nearest neighbor distance [10]. The entire process can be performed using a suite of Fiji macro functions for rapid automated analyses. We obtained the pore size distribution of CMK-8 from 4 different regions (Figure 2(a)). The mean (μ) and standard deviation (σ) of them have differences of less than 10%, as expected for a templated material. In addition to the high consistency, the peaks of the distributions from SEM have the same overall shape when compared to the one from N2 adsorption-desorption isotherms using quenched solid density functional theory (QSDFT) pore approximations (Figure 3). The differences between the positions of the peaks are within 1 nm, which further supports the reliability of the SEM method. The possible sources of the remaining difference include the pore identification model, which can underestimate small pores or create false positives (Figure 1 (d)), and the uneven fracture surface, which can change the apparent shape of the pores. Another advantage of pore analysis via SEM over gas physisorption is the access to more detailed numerical descriptors of the mesopores, such as aspect ratio (Figure 2(b)) and nearest neighbor distance (Figure 2(c)). These descriptors also have high consistency for the CMK-8 tested and may further contribute to the understanding of kinetics and durability of the material when accompanied by electrochemical characterizations such as cyclic voltammetry and galvanostatic charge-discharge. The automated quantitative workflow in this study enhances the utility and throughput of SEM imaging for the rapid screening of electrode materials. Despite the known limitations of the spatial resolution of SEM, the accuracy of contour extraction, and lack of sub-surface information, the preliminary results of pore size distribution still agree well with the data from gas physisorption. More accurate contour extraction and robust validation methods may be developed and tested on different reference pore morphologies from other materials [11]. Schematic of the quantitative analysis workflow: (a) crop of an original SE image of CMK-8 mesoporous carbon, with an SNR of 30 determined from the SMART plugin, (b) FFT of the original image for determining the spatial resolution, which was 2.1 nm, (c) denoised image with local contrast enhancement, and (d) mesopore contours extracted using trainable Weka segmentation. Distribution of (a) mesopore size, (b) aspect ratio, and (c) nearest neighbor distance of CMK-8. Pore size distribution of CMK-8 using N2 adsorption isotherms.

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.