Abstract
Scanning transmission electron microscopy (STEM) is a powerful technique for analysing materials at atomic scale, enabling the relationship between local atomic structures and materials properties to be clarified. One of the most challenging topics in STEM analysis is the observation of beam-sensitive materials (e.g., porous materials, energy materials, and organic materials, etc.). Under the electron irradiation, these materials are easily damaged, requiring the use of low-dose condition for observation. While low-dose conditions are effective in reducing irradiation damage, this condition also leads to a poor signal-to-noise ratio (SNR) due to substantial shot noise, which hinders detailed structure analysis. Here, we developed an optimum bright-field (OBF) STEM using a segmented-type detector as a low-dose imaging technique [ 1]. Conventionally, STEM images are formed by single annular detectors such as annular dark-field (ADF) or annular bright-field (ABF) detectors. Although these imaging methods offer robust and straightforward interpretations of observed atomic structures, their imaging efficiency is relatively low because they use only a limited fraction of transmitted or scattered electrons on the diffraction plane. To obtain richer information about the diffraction pattern, segmented and pixelated type detectors have recently been used extensively [ 2]. In particular, segmented detector allow read-out speeds as fast as conventional detectors, making them especially useful for actual experimental data acquisition. Figure 1 shows the workflow of OBF STEM reconstruction using the segmented detector. Spatial frequency filters, designed to maximise the SNR of the final reconstructed image, are applied to each channel image in Fourier space. The filtered channel images are summed and then inverse Fourier transformed to obtain the OBF image in real space. Figure 1 Schematic illustration of the OBF STEM reconstruction workflow using a segmented detector [ 3]. For simplicity, the case of four-segmented detector is shown. After acquisition of channel images from the segmented detector, each image is Fourier transformed and multiplied by frequency filters, which is designed to maximise the SNR of OBF image. These images are summed and subsequently inverse Fourier transformed to produce the OBF image.