Yi Wang1 Michael R. S. Huang1 Ute Salzberger1 Kersten Hahn1 Wilfried Sigle1 Peter van Aken1

1, Max Planck Institute for Solid State Research, Stuttgart, , Germany

Electron energy-loss spectroscopy (EELS) and energy-dispersive X-ray spectroscopy are two of the most common means for chemical analysis in scanning transmission electron microscopy (STEM). The marked progress of the instrumentation hardware has made chemical analysis at atomic resolution readily possible nowadays. However, the acquisition and interpretation of atomically resolved spectra can still be problematic due to image distortions and poor signal-to-noise ratio (SNR) of the spectra, especially for investigation of energy-loss near-edge fine structures.

By combining multi-frame spectrum imaging and automatic energy-offset correction, we developed a spectrum imaging (SI) technique [1] implemented into customized DigitalMicrograph scripts for suppressing the image distortion and improving the signal-to-noise ratio of the EELS spectra. As widely used in STEM imaging, the former technique can efficiently remove scan distortions in the ADF and ABF images. The latter technique helps significantly reducing correlated noise [2], as for successive spectra different camera pixels are exposed which precludes amplification of small gain normalization errors. We implemented these techniques into STEM spectrum imaging for atomically resolved EELS elemental and fine structure mapping. Available methods for realizing the energy-offset on modern Gatan GIF quantum energy filters, and their reliability and influences on the final EELS energy resolution are tested and discussed. Using practical examples, we show that multi-frame SI and post-alignment can efficiently suppress image distortions and improve the final elemental map quality. We demonstrate that the energy-offset correction method reduces the correlated noise and helps resolving weak features of the near-edge fine structures. The final SI with improved SNR enables extracting individual component maps of the Ti-L2,3 near-edge fine structure and of a Ti-O-Ti bonding direction map at atomic resolution, which has been theoretically predicted but extremely difficult to detect experimentally due to the poor SNR of the spectrum [3]. Combining the multi-frame SI and auto energy-offset-correction we demonstrate that these techniques will open new opportunities for atomically-resolved EELS fine structure mapping.

[1] Y. Wang, M. R. S. Huang, U. Salzberger, K. Hahn, W. Sigle, P. A. van Aken, Ultramicroscopy, DOI: 10.1016/j.ultramic.2017.10.014.
[2] M. Bosman, V.J. Keast, Ultramicroscopy, 108 (2008) 837–846.
[3] M.J. Neish, N.R. Lugg, S.D. Findlay, M. Haruta, K. Kimoto, L.J. Allen, Phys. Rev. B, 88 (2013) 115120.