2D Atomic Mapping of Oxidation States in Transition Metal Oxides by Scanning Transmission Electron Microscopy and Electron Energy-Loss Spectroscopy

Using a combination of high-angle annular dark-field scanning transmission electron microscopy and atomically resolved electron energy-loss spectroscopy in an aberration-corrected transmission electron microscope we demonstrate the possibility of 2D atom by atom valence mapping in the mixed valence compound ${\mathrm{Mn}}_3{\mathrm{O}}_4$. The Mn $L_{2,3}$ energy-loss near-edge structures from ${\mathrm{Mn}}^{2+}$ and ${\mathrm{Mn}}^{3+}$ cation sites are similar to those of MnO and ${\mathrm{Mn}}_2{\mathrm{O}}_3$ references. Comparison with simulations shows that even though a local interpretation is valid here, intermixing of the inelastic signal plays a significant role. This type of experiment should be applicable to challenging topics in materials science, such as the investigation of charge ordering or single atom column oxidation states in, e.g., dislocations.

Figure 1

High resolution HAADF-STEM image of ${\mathrm{Mn}}_3{\mathrm{O}}_4$ along the [100] direction. The cation columns are indicated by red (${\mathrm{Mn}}^{3+}$) and green (${\mathrm{Mn}}^{2+}$) circles. $A$ simulated STEM image at 16 nm sample thickness is inserted into the experimental image.

Figure 2

Average Mn $L_{2,3}$ edge spectrum from atomic column types $A$ (red, averaged from 24 spectra) and $B$ (green, averaged from 54 spectra) together with references from ${\mathrm{Mn}}_2{\mathrm{O}}_3$ and MnO (black lines). Total average Mn $L_{2,3}$ edge spectrum (blue) together with a single pixel spectrum (gray) from an $A$-type site to demonstrate the noise level. The inset shows the HAADF signal acquired simultaneously during spectrum acquisition.

Figure 3

Map of the Mn oxidation states in ${\mathrm{Mn}}_3{\mathrm{O}}_4$ along the [100] zone axis orientation. (a) Shows the spectral weight of component $A$ (${\mathrm{Mn}}^{3+}$) and component $B$ (${\mathrm{Mn}}^{2+}$) obtained from spectrum fitting. $A$ color map is displayed at the bottom (${\mathrm{\text{Red}}=\mathrm{Mn}}^{3+}$, ${\mathrm{\text{Green}}=\mathrm{Mn}}^{2+}$). (b) Maps after low-pass filtering. (c) Simulated maps of the ${\mathrm{Mn}}^{3+}$ and ${\mathrm{Mn}}^{2+}$ signals.

Figure 4

(a),(b) Linear profiles of the spectral weight of component $A/B$ (${\mathrm{Mn}}^{3+}/{\mathrm{Mn}}^{2+}$) taken from the single pixel line indicated by an arrow in Fig. 3. The solid lines are profiles through the simulations for a sample thickness of 16 nm scaled either as fraction of signal $A/B$ (left axis) or fraction of ${\mathrm{Mn}}^{2+}/{\mathrm{Mn}}^{3+}$ (right axis). The ${\mathrm{Mn}}^{2+}$ dumbbells (2.36 Å) are resolved in both experiment and simulation. (c) Derived Mn valence (blue circles) obtained through allocation of spectrum $A$ to ${\mathrm{Mn}}^{3+}$ and spectrum $B$ to ${\mathrm{Mn}}^{2+}$. The solid line represents the simulated result. The effect of intermixing becomes clear through the discrepancy between the experimental and the simulated curve as explained below.