Quasi Instantaneous ELNES Mapping of Multi Element Compounds

Electron energy loss spectroscopy (EELS) measures the energy distribution of the inelastic scattered electrons. In combination with scanning transmission electron microscopy (STEM) it is possible to obtain high spatial resolution spectroscopic data. This method is widely used to map the spatial elemental distribution from the core-loss region (> 50 eV). Moreover, especially with the use of a monochromators, the energy resolution gives access to the energy loss near structure (ELNES) of the core-loss edges which is connected to the electronic structure of the material [1]. Here the energy resolution of the spectrum is only limited by the dispersion of the optics and sampling of the detector. The under sampling poses a problem for simultaneous ELNES mapping of multiple elements in heterogenic compounds with multiple edges distributed over a wide energy range. Even though the optical setup enables mapping the ELNES of every element (< 10 keV) it is challenging to acquire multiple ELNES edges and the low loss simultaneously. For example, 12,050 channels/pixels would be needed to measure an energy range of 2000 eV with a dispersion of 0.17 eV/pixel. A method to overcome this issue is by taking a snippet of the interesting energy loss region by applying an offset voltage to the drift tube. This procedure can be repeated by changing the offset voltage and scanning again the same region of interest. When the offset voltage is changed, the spectrum needs to be refocused which adds overhead to the acquisition procedure. Due to spatial drift or beam damage, it will become very challenging to reliably correlate the spatial information of the multiple maps. For this reason, DualEELS has been developed where two energy regions are acquired at each probe position by varying the drift tube voltage [2]. However only two snippets are available per probe position which is still not sufficient in a lot of cases to map the ELNES of multiple edges. We developed a method to overcome these challenges by applying any number of offsets (2,3,4,…) without adding overhead on the dead time of the detector. Additionally, the defocus in the different offset windows is minimized to maintain a good energy resolution. This enables easy correlation of the different ELNES and low loss structure(s) since the different spectra are acquired quasi-instantaneously. The acquisition time for each offset can be chosen independently to optimize the signal-to-noise (SNR) per energy region. A detailed description of the proposed methodology is given in this presentation. In this work, the ELNES of K-L, Ti-L, O-K, Eu-L and Ta-M are mapped together with the low loss. These different edges could be measured by using 4 different offset energies at a dispersion of 0.17 eV/pixel. In Fig. 1, a schematic view of the 4 snippet regions is shown on a simulated ETO/KTO spectrum. For each offset, the acquisition time is indicated making the total dwell time 0.2 s. A 128×40 map was acquired with these settings and averaged perpendicular from the interface to increase the SNR per spectrum which is necessary to investigate the fine structure. The line plots across the interface of the different elements are shown in Fig. 2. At the top of ETO (purple box), there is presence seen of a different oxidation state in the fine structure of Ti, O and Eu simultaneously when comparing it to the fine structure of the ETO layer (red box). With this example we show the power of the new approach for quality, speed and ease of use in ELNES applications [3]. A simulated ETO/KTO EEL spectrum (blue) to indicate the regions of interest. The four snippets, which are used during the experiment, are indicated in red. The 2d map converted into a line map where the low loss and different fine structures are shown across the interface. The information from the different ELNES can be correlated since they are acquired quasi-instantaneous.