Local coordination of Mn atoms at the Mn:Ge(111) interface from photoelectron diffraction experiments

X-ray photoelectron diffraction (PED) experiment on the metallic Mn:Ge(111) surface alloy has been carried out in the initial stage of mixed phase growth. Our findings show that the ${\mathrm{Mn}}_5{\mathrm{Ge}}_3$ phase is not yet formed for a 1.3 ML (monolayer) coverage, as well as for a 2 ML coverage, followed by an annealing at temperature $(\sim 300°\mathrm{C})$. Rather, we observed the formation of an ordered surface alloy by Mn occupation of the hollow ${\mathrm{H}}_3$ sites in the topmost layer, while about 10% of Mn atoms are found in subsurface layers, partially confirming the theoretical expectations by Zhu et al. [Phys. Rev. Lett. 93, 126102 (2004)]. However, the contribution to PED patterns from subsurface Mn atoms is found compatible with the occupation of interstitial sites only for the 1.3 ML coverage and with the occupation of both interstitial and unexpected substitutional sites for the 2 ML coverage. These findings thus open questions about the determination of the kinetic path to Mn subsurface migration in diluted alloys.

Figure 1

(Color online) Plot of the measured anisotropy $\chi$ of the $\mathrm{Mn}2p_{3∕2}$ emission (kinetic energy of $305\mathrm{eV}$) from the as-deposited thin film (top). The experimental polar range extends from 0° to 62° with respect to the surface normal. The simulated anisotropy for the best-fit model (${\mathrm{H}}_3+\mathrm{I}_2{}^{+}$) is also shown for comparison (bottom).

Figure 2

(Color online) Plot of the measured anisotropy $\chi$ of the $\mathrm{Mn}2p_{3∕2}$ emission (kinetic energy of $305\mathrm{eV}$) from the $(\sqrt{3}\times \sqrt{3})R30°$ surface of the Mn:Ge(111) system (top). The experimental polar range extends from 0° to 62° with respect to the surface normal. The simulated anisotropies for the three best-fit models, (${\mathrm{H}}_3+\mathrm{I}_2{}^{+}$, ${\mathrm{H}}_3+{\mathrm{S}}_8$, and ${\mathrm{H}}_3+{\mathrm{S}}_9$) are also shown for comparison.

Figure 3

(Color online) Top and side view of the interface structures around the emitting Mn atom for each model considered in the simulated PED anisotropies.

Figure 4

(Color online) Simulated anisotropies $\chi$ for Mn atoms in different substitutional and interstitial lattice sites: the ${\mathrm{H}}_3$ site, the ${\mathrm{T}}_4$ site, and the ${\mathrm{Mn}}_5{\mathrm{Ge}}_3$ bulk model. For the ${\mathrm{S}}_8$, ${\mathrm{I}}_1$, and ${\mathrm{I}}_2$, additional simulations (labeled $\mathrm{S}_8{}^{+}$, $\mathrm{I}_1{}^{+}$, and $\mathrm{I}_2{}^{+}$) have been carried out, considering also a not-emitting Mn atom in ${\mathrm{H}}_3$ site. The ${\mathrm{S}}_8$ and ${\mathrm{I}}_2$ simulations have been omitted, since the differences to the $\mathrm{S}_8{}^{+}$ and $\mathrm{I}_2{}^{+}$ are negligible. Please note the much different amplitude of the anisotropy from one site to the others. In particular, the ${\mathrm{H}}_3$ site displays an anisotropy at least ten times lower than the substitutional sites.