Charting the low-loss region in electron energy loss spectroscopy with machine learning

Journal article (2021)

Authors

Laurien I. Roest Science Park 105, Kavli institute of nanoscience Delft
S.E. van Heijst QN/Conesa-Boj Lab - AS , Kavli institute of nanoscience Delft
L.A. Maduro Kavli institute of nanoscience Delft, QN/Conesa-Boj Lab - AS
Juan Rojo Science Park 105, Vrije Universiteit Amsterdam
S. Conesa Boj QN/Conesa-Boj Lab - AS , Kavli institute of nanoscience Delft
Research Group
QN/Conesa-Boj Lab (AS) (TU Delft)
Copyright: © 2021 Laurien I. Roest, S.E. van Heijst, L.A. Maduro, Juan Rojo, S. Conesa Boj
DOI
help_outline
https://doi.org/10.1016/j.ultramic.2021.113202
Machine learning Neural networks Bandgap Transmission electron microscopy Transition metal dichalcogenides Electron energy loss spectroscopy
More Info
expand_more

Published Date

2021

Language

English

Faculty

Applied Sciences

Department

QN/Quantum Nanoscience

Research Group

QN/Conesa-Boj Lab

Abstract

Exploiting the information provided by electron energy-loss spectroscopy (EELS) requires reliable access to the low-loss region where the zero-loss peak (ZLP) often overwhelms the contributions associated to inelastic scatterings off the specimen. Here we deploy machine learning techniques developed in particle physics to realise a model-independent, multidimensional determination of the ZLP with a faithful uncertainty estimate. This novel method is then applied to subtract the ZLP for EEL spectra acquired in flower-like WS2 nanostructures characterised by a 2H/3R mixed polytypism. From the resulting subtracted spectra we determine the nature and value of the bandgap of polytypic WS2, finding EBG=1.6−0.2+0.3eV with a clear preference for an indirect bandgap. Further, we demonstrate how this method enables us to robustly identify excitonic transitions down to very small energy losses. Our approach has been implemented and made available in an open source PYTHON package dubbed EELSfitter.