Answers are grounded in this paper and cite their sources.
Due to the recent advances in pixelated detector technology, the field of 4D STEM has grown into one of the major areas of electron microscopy. In the early days of this development, I worked on the design and development of a segmented detector, in which the detector plane is divided into many segments capable of atomic-resolution imaging [1]. I have mainly focused on applying them to differential phase contrast (DPC) and center of mass (CoM) imaging techniques for visualizing electromagnetic fields in materials and devices at high spatial resolution [2]. We have shown that these techniques are powerful for visualizing electromagnetic structures inside materials: electric field imaging of p-n junction [3] and single atoms [4], magnetic field imaging of magnetic domains [5], magnetic skyrmions [6] and antiferromagnet [7]. However, these observations were limited to electromagnetic fields inside individual crystalline grains. When trying to extend these imaging techniques to crystal interfaces and lattice defects, which are very important local structures for actual material and device properties, we have encountered a serious obstacle, namely the superposition of diffraction contrast. This problem has long hindered the application of the electromagnetic field imaging to interfaces and defects in real materials and devices. In recent years, it has been shown that the diffraction-contrast problem in DPC imaging can be effectively reduced [8-10]. This is achieved by averaging a large number of DPC images acquired with slightly tilting the specimens. This effect arises from the fact that, whereas diffraction contrast changes sensitively with even slight specimen tilts, the true electromagnetic-field contrast is robust against such small tilts, so that averaging effectively suppresses only the diffraction-contrast contribution. Since this approach requires the acquisition and averaging of a large number of DPC images, it was not well suited to practical applications. Recently, however, the tilt-scan averaging DPC method (tDPC) has been developed to make this type of averaging much more efficient [11], and it has been demonstrated that it can substantially reduce diffraction contrast in practice [12]. This tDPC technique has been successfully applied to the observations of two-dimensional electron gases accumulated at heterointerfaces [13], space-charge layers formed at ceramic grain boundaries [14], charge distributions at ferroelectric domain walls [15], spintronic devices [16] and the precise measurement of domain-wall widths in hard magnets [17,18]. However, if one were to point out a drawback of this method, the tilt-scan averaged DPC method requires a long pixel dwell time and generates an enormous data volume, which makes it technically very challenging to implement in practical materials and device analysis using pixelated detectors. In our case, we employ a 40-segment detector that enables both high-speed data processing and quantitative CoM measurements. In our normal observations, tDPC imaging is performed under 61-tilt conditions and a pixel dwell time of up to approximately 25 μS / pixel can be achieved, making live tDPC imaging possible. Such high-speed observations are still considered very difficult with current pixelated detectors due to the enormous data-processing load. The use of a very fast segmented detector with a sufficiently large number of segments is therefore regarded as an effective approach to the practical implementation of tDPC. In this talk, I will first briefly review the development history of our segmented detectors, and then introduce specific examples showing that segmented detectors are still extremely powerful for practical electromagnetic field imaging using tDPC. I will also introduce our latest segmented-detector developments, which follow a bottom-up approach to make the segmentation design increasingly similar to that of a pixelated detector [19].
No takes yet. Share an insight, caveat, or question.
Naoya Shibata (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: