Probing electron beam induced effects at the single atom and defect level
ORAL
Abstract
Control of matter at the atomic scale is the key to building structures atom-by-atom. Currently only a handful of tools exist which have the precision to affect structures at the single atom level – the scanning transmission electron microscope (STEM) is one of these. The STEM provides the capability to routinely detect single atoms, due to an atom-sized electron beam. Typically any beam-induced effects in matter is regarded by microscopists to be “beam damage” however if proper control is attained then this may indeed be an excellent route toward atomic fabrication.
In order to realize this, fast and accurate extraction of atomic coordinates from image data is required, followed by control of the electron beam in relation to those coordinates. The extraction of coordinates has not been reliable or fast enough for live operation using standard blob detection or even neural networks, however here we utilize deep ensembles to robustly predict atom coordinates on the order of milliseconds. Given this, we then position the electron beam to selectively remove single atoms or create single defects from the system with extreme precision. This is demonstrated in both graphene and MoS2.
In order to realize this, fast and accurate extraction of atomic coordinates from image data is required, followed by control of the electron beam in relation to those coordinates. The extraction of coordinates has not been reliable or fast enough for live operation using standard blob detection or even neural networks, however here we utilize deep ensembles to robustly predict atom coordinates on the order of milliseconds. Given this, we then position the electron beam to selectively remove single atoms or create single defects from the system with extreme precision. This is demonstrated in both graphene and MoS2.
–
Publication: 1. Roccapriore K.M.., Boebinger M.G., Dyck O., Ghosh A., Unocic R.R., Kalinin, S.V., Ziatdinov M."Probing Electron Beam Induced Transformations on a Single Defect Level via Automated Scanning Transmission Electron Microscopy." ACS Nano 2022 10.1021/acsnano.2c07451
Presenters
-
Kevin M Roccapriore
Oak Ridge National Lab, Oak Ridge National Laboratory
Authors
-
Kevin M Roccapriore
Oak Ridge National Lab, Oak Ridge National Laboratory
-
Maxim Ziatdinov
Oak Ridge National Lab
-
Matthew Boebinger
Oak Ridge National Laboratory
-
Ondrej Dyck
Oak Ridge National Laboratory
-
Ayana Ghosh
Oak Ridge National Lab
-
Raymond Unocic
Oak Ridge National Laboratory
-
Sergei V Kalinin
University of Tennessee, University of Tennessee, Knoxville