Reducing Electron Beam Damage with Multipass Transmission Electron Microscopy

Abstract

With the introduction of hardware aberration correction, direct electron detectors, ultra-bright electron sources and highly precise spectrometers, it seems like we are approaching the pinnacle of transmission electron microscopy (TEM) imaging and spectroscopy. However, the field of electron microscopy is still far from the ultimate signal-to-noise and efficiency limits imposed by electron scattering physics. Current TEM experiments where images are measured with high quantum efficiency detectors can approach the shot noise (or Poisson noise) limits, but the true upper bound for signal-to-noise for any electron imaging measurement is the Heisenburg limit [1]. Some initially proposed methods for approaching this measurement limit were quantum techniques such as highly-entangled measurement particles, which are impractical to use in a large-scale imaging experiments using photons. Entangling large numbers of electrons is even more difficult and seems unlikely to be achieved with existing technology. However, there is an alternative method to approach the Heisenburg limit: allowing the electron probe to interact with the sample many times before the image intensity measurement [2], as in the recently proposed design for a multipass transmission electron microscope [3,4].In this talk, we will describe a multipass TEM instrument designed using the principles of quantum metrology. A simplified schematic of the experimental geometry is shown in Figure 1. A multipass TEM is very similar to a conventional TEM, except for the addition of two electron mirrors and the use of a pulsed electron source. The biasing of these electron mirrors can be reduced to allow the pulsed electron beam to enter the center portion of the instrument. The central portion is composed of two mirror-image configurations of a TEM instrument with mirrors on each end. The electron beam can therefore be passed through the sample multiple times. After a set number of passes, the electron beam can be passed out through the second mirror shown in the Figure 1 schematic. Finally, the electron pulse can be imaged using a standard electron detector, either after using a phase plate or applying a defocus to generate a phase contrast image. We assume that the mirrors and lenses can compensate for each other's aberrations in this configuration.The advantage of using a multipass instrument is demonstrated in Figure 2. In this example, we have used multislice simulations to show how a multipass instrument could transform a low-dose imaging experiment of an isolated, small protein sample that is representative of cryo-EM experiments, the hexameric unit of the immature HIV-1 Gag CTD-SP1 lattice [5]. The structure and projected potential of this protein sample is shown in two different orientations in Figures 2a and b. Simulated TEM images at different electron doses for different numbers of passes are plotted in Figure 2c. The reported dose is the total dose seen by the specimen, which is approximately equal to one divided by the number of passes. These simulation...

Imaging Beam‐Sensitive Materials by Electron Microscopy

“…d) Model and the simulated TEM images of HIV‐1 Gag protein at different electron doses for different numbers of passes. Reproduced with permission 177. Copyright 2017, Cambridge University Press.…”

“…Compared with the typical TEM, multipass TEM add two more electron mirrors and a pulsed electron source (Figure 6c). The two mirrors allow the pulsed electron beam to pass through the sample multiple times,177 the total dose can thus be reduced to obtain an image with the same SNR. After a specific number of passes, the electron could be collected by the detectors.…”

“…After a specific number of passes, the electron could be collected by the detectors. Figure 6d shows the simulated TEM images of HIV‐1 Gag protein at different electron doses for different numbers of passes 177. It is observed that the SNR of the image reaches maximum at neither too many passes nor too few passes.…”

“…It is observed that the SNR of the image reaches maximum at neither too many passes nor too few passes. It is straightforward that few passes lead to low SNR, while the low SNR at a high number of passes is possibly ascribed to the phase wrapping or inelastic losses 177…”

USID and Pycroscopy – Open Source Frameworks for Storing and Analyzing Imaging and Spectroscopy Data

“…We evaluated existing scientific data exchange models within and beyond the STEM community such as NeXus 48 , PaNData 49 , HMSA 36 , SPMML 35 , MatML 50 , Data Exchange 51 , Open Microscopy Environment's (OME) version of the Tagged Image File Format (TIFF) -OME-TIFF 52 , Electron Microscopy Data (EMD) 53 , multidimensional eXtensible archive (MXA) 52,54 , and the model that used by DREAM.3D 55 , etc. Most of these were rigidly designed around specific kinds of instruments, modalities, coordinate systems (Cartesian only with no option for polar coordinates etc.…”