Developments in High Speed Structural Imaging of Low Dimensional Materials
Prof. Angus Kirkland
University of Oxford
16:15 - 17:15 Tuesday 18 May 2021 Graz University of Technology - remote online I will describe recent work using high speed direct electron detectors and artificial intelligence / machine learning to automatically map defect transitions in graphene. I will also discuss the use of similar detectors in electron ptychography, in particular under extremely low dose conditions using binary counting.
The development of high speed direct electron detectors suitable for use at intermediate electron energies has led to significant progress in imaging, diffraction and spectroscopy [1].
To usefully deploy Graphene and related materials in electronic applications [2-4] it is essential to understand the behavior of defects, which have been the subject of extensive research in silicon devices for decades. Detector advances make it possible to image these defects at primary energies below those that cause significant specimen damage whilst retaining sufficient spatial resolution to resolve local atomic configurations around the defect site [5].
However, the extremely large datasets (typically 106 images or greater) that can be acquired makes conventional manual image processing intractable. I will describe how this can be overcome using a deep learning neural network [6] for atomic model abstraction from low dose high framerate graphene images (Fig. 1). Using this approach, it is possible to identify many instances of defect transitions and to map the lifetimes of defect states. In turn these can be used as input to density functional theory to model the potential energy landscape for the transitions.
Finally, I will highlight the use of fast detectors for electron Ptychography at low dose. At low dose the sampling of the diffraction pattern in the far field is sparse and a counting direct electron detector can be operated in a binary mode to provide an effective speed increase.
Figure 1. (a) Low-dose image of a graphene sheet recorded with a 1ms exposure at 80KV containing defects; (b) Automatically generated annotation with 5-membered rings (pink), 6-membered rings (green), and 7-membered rings (yellow) overlaid on top of the experimental image; (c) enlarged defect area with carbon atom positions marked by blue circles.
References
1. Mir, J.A. et al. (2017) Ultramicroscopy, 182, 44.
2. Novoselov, K. S. et al. (2012) Nature 490, 192.
3. Schwierz, F. (2010) Nature Nanotech. 5, 487.
4. Britnell, L. et al. (2012) Science 335, 947.
5. Robertson, A.W. et al. (2012) Nature Comm. 3, 1144.
6. LeCun, Y. et. Al. (2015) Nature 521, 436
Gastgeber: Gerald Kothleitner, FELMI TU Graz
WEBEX-Link: Click here
|
