Abstract text (incl. figure legends and references)

Multimetallic nanocrystals are of great scientific and technological interest because of their unique electronic, optical, or catalytic properties. These properties are largely determined by the atomic structure and composition of the nanocrystal. Therefore, quantitative structure determination is essential for the development of new nanocrystals. For homogeneous nanoparticles, the number of atoms can be counted from high resolution annular dark field scanning transmission electron microscopy (HAADF STEM) images [1] where intensities scale with the number of atoms and the atomic number Z. For heterogeneous structures, however, the presence of this so-called Z-contrast complicates this atom-counting procedure since the different elements and their exact 3D arrangement in each atomic column will contribute differently to the image intensity [2]. Therefore, we developed a new methodology, which combines HAADF STEM imaging and elemental mapping by energy dispersive X-ray (EDX) spectroscopy.

The so-called scattering cross-section (SCS), corresponding to the total intensity of electrons scattered by a single atomic column, has been shown to be a successful performance measure for atom-counting and composition determination in HAADF STEM [1,4-6]. Similarly, EDX STEM SCSs can be defined from elemental maps. Since both HAADF STEM and EDX imaging are incoherent techniques, a linear relationship between the EDX and HAADF STEM SCSs exists. By exploiting this linear relationship, the experimental SCSs are matched to the simulated SCSs by estimating normalization constants for the EDX SCSs, using an iterative weighted least squares minimization.

As a proof of concept, the combination of EDX and HAADF STEM images is used to count the number of atoms for a Au@Ag nanorod. Figure 1(a) shows a times series for both the STEM images and the Ag and Au EDX maps. The resulting number of Ag and Au atoms for each atomic column are shown in Figure 1(b). With this methodology we also explored the possibility of the characterization of a simulated Au@Pt nanorod, with adjacent atomic numbers (Figure 2). Those analyses quantitatively demonstrate the new opportunities to count the number of atoms corresponding to each specific element, even when the difference in atomic number is only one [7].

References

[1] S. Van Aert et al., Physical Review B 87 (2013), p. 064107.

[2] K.H.W. van den Bos et al., Physical Review Letters 116 (2016), p. 246101.

[3] A.J. D"Alfonso et al., Physical Review B 81 (2010), p. 100101.

[4] S. Van Aert et al., Ultramicroscopy 109 (2009), p. 1236.

[5] H. E et al., Ultramicroscopy 133 (2013), p. 109.

[6] G.T. Martinez et al., Ultramicroscopy 187 (2018), p. 84.

[7] This work was supported by the European Research Council (Grant 770887 PICOMETRICS to SVA and Grant 815128 REALNANO to SB, Grant 823717 ESTEEM3). The authors acknowledge financial support from the Research Foundation Flanders (FWO, Belgium) through project fundings and postdoctoral grants to ADB and EB.

Figure 1 (a) Experimental HAADF STEM images and EDX elemental maps for Ag and Au. (b) The total number of atoms and the number of Au and Ag atoms.

Figure 2 (a) Example of a simulated time-series of HAADF STEM images and Pt and Au elemental maps with 5% collection efficiency and an incident dose of 5×10⁴e^-/Ų per frame for the HAADF STEM images and 5×10⁶e^-/Ų per frame for the EDX elemental maps. (b) Root mean squared error per atomic column for Pt, Au, and the total number of atoms.