November_EDFA_Digital

edfas.org 25 ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 23 NO . 4 advantageous (Fig. 3d). Amore advanced implementation enablingmotorized samplemanipulation (i.e., tilt, rotate, CL) could be realized by combining this detector with a six degree-of-freedom apparatus described recently. [13] 4D STEM-IN-SEM IMAGING AND DIFFRACTION As mentioned, numerous transmission imaging tech- niques can be implemented in an SEM. Seeing as many readers are familiar with common solid-state detectors and the concomitant imaging modes, those applications will not be covered. An emerging method that has gained popularity is 4D STEM (or 4D STEM-in-SEM as it pertains here), a recently coined name for a technique encompass- ing several applications. [21] Here, a DP is recorded at each raster position as the electron beam is scanned across a sample, and those patterns are stored in a data cube for offline analyses (Fig. 5a). As an introduction to 4DSTEM-in-SEM, oneway of visu- alizing the data in real and reciprocal space (i.e., k-space) is shown in Fig. 5. Here, a MATLAB script was used to determine anddisplay the averageDPof the entire dataset from a multilayer two-dimensional MoS 2 sample (Fig. 5b) and an annular image was compiled from the dataset (Fig. 5c). The average DP generally provides a broad overview of the sample structure, but it may also be overwhelming, so the script allows the user to select an ROI from the real space image to limit the dataset for subsequent analyses. For example, a less complex DP from the region specified in Fig. 5c is shown in Fig. 5d. A virtual aperture can thenbedrawnon the reduceddataset, and a real-space image showing regions of the sample that scatter electrons into the user-selected region can be extracted from the reduced dataset (Fig. 5e). Like the p-STEMdetector, the 4DSTEM-in-SEMvirtual aperture can be any shape. Perhapsmore importantly though, the data can be analyzed in any number of ways without concern for incurring further damage to beam sensitive samples. Although difficult to show in print, a convenient way to survey the data/sample is to create animated image sequences that the user can specify. For example, the script used here enables the user to draw a freehand line on any DP and to specify a virtual aperture. The script scans the virtual aperture along each beam raster point in the freehand line, generates a real-space image at each point, and writes the image to an image stack that can be replayed in an animated format. Alternatively, a freehand line can be drawn on a real-space image and theDPs along that line can be visualized in the same manner. 4D STEM-in-SEM is not without challenges, however, two of which are stage drift and data storage. Because dif- fraction cameras available for SEMs are currently limited to approximately 1000 frames per second, scan times can last from a few minutes to more than an hour depending on the camera binning and the number of points in the raster array. Almost any SEMsample stage candrift several tens of nanometers in an hour. Data files can also be large, making storage and data analysis potentially cumber- some. For example, a single dataset file size can easily comprise several tens of gigabytes depending on the scan resolution, camera binning, camera ROI, etc. Processing times can be improved by setting aside unneeded data, by using every other DP, or by binningDPs together. However, the vast amount of diffraction data is a major benefit of the technique, and numerous applications have been devised to take full advantage of the information including monolayer graphene grain orientationmapping, [22] strain mapping, [23] and non-contact nanoscale temperature mapping, [24] for example. Furthermore, numerous python scripts are available online to analyze 4D datasets (i.e., LiberTEM, pyxem, py4Dstem, hyperspy, pycroscopy, etc.). SUMMARY STEM-in-SEM imaging and diffraction techniques are accessible today, and their utility continues to expand. Although there will always be the need for high-energy STEM/TEM, the SEMcan fulfill many transmission electron imaging and diffraction needs provided that the sample is sufficiently thin. In addition to the solid-state detectors for imaging, newdiffractiondetectors continue to appear that will inevitably be standard options on all SEMs. Regarding 4DSTEM-in-SEM, the vast amount of data collected in each dataset offers many opportunities for further exploration beyond the existing applications, and it is only amatter of time before faster diffraction cameras are implemented in an SEM. ACKNOWLEDGMENT Identification of commercial products does not imply recommendationor endorsement by theNational Institute of Standards and Technology, nor does it imply the prod- ucts are the best suited for the application. REFERENCES 1. E. Buhr, et al.: “Characterization of Nanoparticles by Scanning Electron Microscopy in Transmission Mode,” Meas. Sci. Technol., 2009, 20, p. 084025. 2. M. Kuwajima, et al.: “Automated Transmission-Mode Scanning Electron Microscopy (tSEM) for Large Volume Analysis at Nanoscale Resolution,” PLOS ONE, 2013, 8, p. e5957.

RkJQdWJsaXNoZXIy MTMyMzg5NA==