Charting the low-loss region in Electron Energy Loss Spectroscopy with machine learning

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 modelindependent, 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 WS 2 nanostructures characterised by a 2H/3R mixed polytypism. From the resulting subtracted spectra we determine the nature and value of the bandgap of polytypic WS 2 , finding E BG = 1.6 +0.3 −0.2 eV 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.

Illuminating the Electronic Properties of WS2 Polytypism with Electron Microscopy

“…Exploiting this information requires the subtraction of the zero-loss peak (ZLP) contribution. Here the band gap of polytypic 2H/3R WS2 nanoflowers is investigated by means of a recently developed model-independent method based on machine learning 27 . This approach, developed in the context of high energy physics 28 , combines artificial neural networks for an unbiased parametrization of the zero-loss peak profile with Monte Carlo sampling for a faithful uncertainty estimate.…”

“…Further discussions on the technical procedure adoped for the band gap determination can be found in ref. 27…”