A Reconfigurable Detector for Spectroscopy and Diffraction
Microscopy and Microanalysis · 27.07.2026 · 32 (Supplement_1)
DOI: 10.1093/mam/ozag053.012 ↗Kurz gefasst
Fachartikel in der Zeitschrift Microscopy and Microanalysis, online erschienen am 27.07.2026. Verfasst von Peter Denes. Die Arbeit nennt 5 Literaturangaben.
Abstract (Originalsprache)
The ability to capture pixilated diffraction data in Scanning Transmission Electron Microscopy (STEM) provides additional information beyond just the scattered intensity. During the past 1.5 decades the use of “direct” electron detectors based on CMOS Image Sensors (CIS) [1] or hybrid pixels [2] has become common. Initially, such detectors operated at frame rates of less than a few kHz, which is considerably slower than the typical 1-10 μs STEM scan dwell times. The 4D Camera [3], initially deployed in 2019 at Lawrence Berkeley National Laboratory is a CIS sensor with 576x576 pixels reading out at 2.9x1010 pixels/s ≈ 87,000 frames/s. In addition to the very high pixel rate, the detector was designed to stream data directly to the NERSC [4] supercomputer. The high acquisition rate has proven to be useful for 4D STEM methods [5] and the attached computational power has greatly shortened the time from experiment to result. A new CIS-based sensor, called DoEELS, has recently been fabricated with the goal of building on the 4D Camera and expanding capabilities in spectroscopy and diffraction. To accomplish this, the sensor incorporates a novel reconfigurable architecture: As shown in Fig. 1, the sensor always reads 3.6x1010 pixels/s. In spectroscopy mode (a) most of the sensor reads ∼10,000 frames/s. The sensor can be electronically reconfigured to diffraction mode (b) where only the central 960x960 pixels are read out at ∼40,000 frames/s (or higher rates for smaller regions of interest). The sensor has improved signal/noise relative to the 4D Camera due to a thicker sensitive volume, but still maintains high electron sensitivity to energies >5 kV. The data acquisition for DoEELS, while inspired by the 4D Camera, adds significant local processing through the use of the Nvidia IGX GPU platform. Data are assembled by an FPGA-based aggregator which is connected to a GPU array capable of up to 800 Gb/s data bandwidth to GPU memory. GPU processing will be used not only for low-level detector signal processing, but also for initial analysis. DoEELS is planned to be deployed on a new cryo-STEM this year at NCEM. The development status and initial results will be presented. In the same way that the 4D Camera enabled virtual experiments by collecting a broad range of scattering angles, from which certain ones could be selected afterwards, reconfigurable detectors could open new experiment designs, tailoring the detector geometry experiment-by-experiment. Reconfigurable DoEELS detector: (a) 3264 x 960 pixel EELS sensor at 3.6x1010 pixels/s ≈ 10,000 frames/s (b) Sensor internally reconfigured as 960 x 960 pixel diffraction sensor also at 3.6x1010 pixels/s ≈ 40,000 frames/s (c) An optical mask illustrating the two operating modes: spectroscopy (top) and diffraction (bottom). The lower image is darker for the same illumination as the readout rate is higher.
Zitieren
APA
Denes, P. (2026). A Reconfigurable Detector for Spectroscopy and Diffraction. Microscopy and Microanalysis, 32(Supplement_1). https://doi.org/10.1093/mam/ozag053.012
Vancouver
Denes P. A Reconfigurable Detector for Spectroscopy and Diffraction. Microscopy and Microanalysis. 2026;32(Supplement_1). doi:10.1093/mam/ozag053.012
BibTeX
@article{Denes2026,
author = {Denes, Peter},
title = {A Reconfigurable Detector for Spectroscopy and Diffraction},
journal = {Microscopy and Microanalysis},
year = {2026},
volume = {32},
number = {Supplement_1},
doi = {10.1093/mam/ozag053.012}
}
EUVIMED