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

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