Magnetic and chemical nonuniformity in ${\mathrm{Ga}}_{1-x}{\mathrm{Mn}}_x\mathrm{As}$ films as probed by polarized neutron and x-ray reflectometry

We have used complementary neutron and x-ray reflectivity techniques to examine the depth profiles of a series of as-grown and annealed ${\mathrm{Ga}}_{1-x}{\mathrm{Mn}}_x\mathrm{As}$ thin films. A magnetization gradient is observed for two as-grown films and originates from a nonuniformity of Mn at interstitial sites, and not from local variations in Mn at Ga sites. Furthermore, we see that the depth-dependent magnetization can vary drastically among as-grown ${\mathrm{Ga}}_{1-x}{\mathrm{Mn}}_x\mathrm{As}$ films despite being deposited under seemingly similar conditions. These results imply that the depth profile of interstitial Mn is dependent not only on annealing, but is also extremely sensitive to initial growth conditions. We observe that annealing improves the magnetization by producing a surface layer that is rich in Mn and O, indicating that the interstitial Mn migrates to the surface. Finally, we expand upon our previous neutron reflectivity study of ${\mathrm{Ga}}_{1-x}{\mathrm{Mn}}_x\mathrm{As}$, by showing how the depth profile of the chemical composition at the surface and through the film thickness is directly responsible for the complex magnetization profiles observed in both as-grown and annealed films.

Figure 1

(Color online) XRR data, and fits for the as-grown and annealed films. The scattering length density depth profiles $\rho \left(z\right)$ used to fit the data are shown in the insets.

Figure 2

(Color online) Resonant XRR data for the as-grown and annealed set A films. The annealed film features pronounced O and Mn peaks while the as grown does not.

Figure 3

(Color online) Spin-up and spin-down neutron reflectivities and fits for the as-grown and annealed set A and set C samples. The spin-up data and fits are shown offset by $6\times {10}^{-10}{\mathrm{\AA }}^{-4}$. The data and fits recast as spin asymmetry is shown in the inset.

Figure 4

(Color online) Scattering length density depth profiles $\rho \left(z\right)$ used to fit the set A and set C data in Fig. 3. For each panel, the data above the break in the vertical axis are the chemical scattering length densities ${\rho }_{\mathit{Chem}}$, and the data below the break are the magnetic scattering length densities ${\rho }_{\mathit{Mag}}$.

Figure 5

(Color online) Change in best-fit ${\chi }^2$ as a function of $M$ gradient falloff $R_M$ (left), and spatial extent $R_T$ (right), for the set A as-grown and annealed samples. The inset $M$ models correspond to fits that reproduce the data with one standard deviation uncertainty (circled data points). $M$ is normalized by the maximum value of $M$ for each sample to allow for direct comparison. The inset models still show a clear smoothing of $M$ after annealing.