Low‐energy electron microscopy intensity–voltage data – Factorization, sparse sampling and classification.

Low‐energy electron microscopy (LEEM) taken as intensity–voltage (I–V) curves provides hyperspectral images of surfaces, which can be used to identify the surface type, but are difficult to analyse. Here, we demonstrate the use of an algorithm for factorizing the data into spectra and concentrations of characteristic components (FSC3) for identifying distinct physical surface phases. Importantly, FSC3 is an unsupervised and fast algorithm. As example data we use experiments on the growth of praseodymium oxide or ruthenium oxide on ruthenium single crystal substrates, both featuring a complex distribution of coexisting surface components, varying in both chemical composition and crystallographic structure. With the factorization result a sparse sampling method is demonstrated, reducing the measurement time by 1–2 orders of magnitude, relevant for dynamic surface studies. The FSC3 concentrations are providing the features for a support vector machine‐based supervised classification of the surface types. Here, specific surface regions which have been identified structurally, via their diffraction pattern, as well as chemically by complementary spectro‐microscopic techniques, are used as training sets. A reliable classification is demonstrated on both example LEEM I–V data sets. Lay description: Low‐energy electron microscopy (LEEM) is a powerful experimental method to image surfaces, thin films and nanoparticles. An incident beam of low energy electrons (<50eV) is reflected from the surface and used to create an image of the investigated sample. The structure of the first few atomic layers of the investigated sample is encoded in the energy dependent electron reflectivity of the surface, so called intensity versus electron energy, or in short LEEM I‐V, spectra, which however are difficult and time‐consuming to interpret. In this paper we present a factorization method to describe the LEEM I‐V hyperspectral data as a combination of characteristic components which are defined by their concentrations and spectra. Using the concentration maps, we demonstrate a supervised classification method which provides a fast and reliable classification of surface reconstructions, as shown on two examples, ruthenium oxide (RuO2), and praseodymium oxide (PrOx). For PrOx, the factorization and classification reveals that the surface consists of a flat substrate with bands of coalesced oxide islands which nucleated at the atomic step edges of the Ru(0001) substrate. The PrOx regions comprise a complex substructure of five distinguishable phases. For RuO2, the method reveals the different types of islands that exist in the rich RuO2/Ru system, where different RuO2 orientations characteristic of the Ru oxidation can be separated by their I‐V spectra. Furthermore, using the extracted component spectra and the classification of the concentrations, demonstrate a sparse sampling method to reduce the number of acquired spectral points required for classification. A reduction of the acquisition time by a factor of 30 per classification is achieved for the example data. [ABSTRACT FROM AUTHOR]