O $K$-edge x-ray absorption study of ultrathin $\mathrm{NiO}$ epilayers deposited in situ on $\mathrm{Ag}\left(001\right)$

In a recent contribution [Phys. Rev. Lett. 91, 046101 (2003)] we used polarization-dependent, $\mathrm{Ni}$ $K$ edge, x-ray absorption spectroscopy (XAS) to probe the structure of ultrathin $\mathrm{NiO}$ epilayers deposited on $\mathrm{Ag}\left(001\right)$. In that experiment samples were measured ex-situ and a $5\mathrm{ML}$-thick $\mathrm{MgO}$ cap was used to avoid the hydroxylation of the $\mathrm{NiO}$ film. In the present paper we report complementary $\mathrm{O}$ $K$ edge XAS data on the same system; $\mathrm{NiO}$ epilayers, in the $3–50\mathrm{ML}$ thickness range, were grown in situ in the end station of the ALOISA beamline of the ELETTRA facility. A quantitative analysis of the data in the extended energy range has been performed using multiple scattering simulations. We found that, even in the ultrathin limit, the local structure of the film is rock-salt and we obtained a quantitative evaluation of the average in-plane and out-of-plane film strain as a function of the film thickness $T$. An in-plane compressive strain, due to lattice mismatch with the $\mathrm{Ag}$ substrate, is clearly present for the $3\mathrm{ML}$ film, being the in- and out-of-plane nearest neighbor distances equal to $r_{\Vert }=2.048\pm 0.016\mathrm{\AA }$ and $r_{\perp }=2.116\pm 0.018\mathrm{\AA }$. These values are in agreement with the expected behavior for a tetragonal distortion of the unit cell. The growth-induced strain is gradually released with increasing $T$: it is still appreciable for $10\mathrm{ML}$ but is completely relaxed at $50\mathrm{ML}$. Any significant intermixing with the $\mathrm{Ag}$ substrate has been ruled out. Combining $\mathrm{O}$ and $\mathrm{Ni}$ $K$ edge results we can conclude that $\mathrm{NiO}$ films grow on $\mathrm{Ag}\left(001\right)$ in the O-on-$\mathrm{Ag}$ configuration, with an interface distance $d=2.28\pm 0.08\mathrm{\AA }$. This expansion of the interplanar distance is in agreement with recent ab initio simulations. A comparison with the similar $\mathrm{MgO}∕\mathrm{Ag}\left(001\right)$ system is also performed.

Figure 1

(a) Raw $\mathrm{O}$ $K$ edge $\chi \left(\mathrm{k}\right)$ data of $\mathrm{NiO}∕\mathrm{Ag}\left(001\right)$ films as a function of both film thickness and beam polarization $s(\theta =90°)$ and $p(\theta =90°)$, where $\theta$ is the angle between the electric field vector and the vector normal to the (001) surface. (b) A comparison between experimental (points) and best fit (solid line) for the modulus of the $k^2$-weighted, phase uncorrected FTs of the $\chi \left(k\right)$ data reported in part (a). (c) XPS spectra collected in situ (from top to bottom: $T=50$, 10 and $3\mathrm{ML}$).

Figure 2

Out-of-plane $\left(r_{\perp }\right)$ and in-plane $\left(r_{\Vert }\right)$ nearest neighbor distances obtained by the analysis of $\mathrm{Ni}$ (full markers) and $\mathrm{O}$ $K$ edge (open markers) XAS data for the 10 (triangles) and 3 (squares) ML films and for the $\mathrm{NiO}$ reference samples, with the relative uncertainties. Full and open stars represent the $r_{\perp }$ and $r_{\Vert }$ values for unstrained $\mathrm{NiO}$ bulk (XRD) and for $\mathrm{NiO}$ perfectly epitaxial with $\mathrm{Ag}\left(001\right)$, respectively. The open diamond represents the value obtained by ab initio calculations performed on a $2\mathrm{ML}$ $\mathrm{NiO}∕\mathrm{Ag}\left(001\right)$ system by a CRYSTAL code. Note that the simulated $\mathrm{NiO}∕\mathrm{Ag}\left(001\right)$ results less strained than the experimental one owing to a theoretical half lattice parameter of $\mathrm{Ag}$ bulk of $2.07\mathrm{\AA }$. The solid line represents the theoretical $r_{\perp }$ vs $r_{\Vert }$ relationship predicted by the elastic theory, adopting the $\gamma$ constant of $\mathrm{NiO}$ bulk.[34] The top right corner hosts a scheme of the $\mathrm{Ag}∕\mathrm{NiO}$ interface determined by combined $\mathrm{O}$ and $\mathrm{Ni}$ $K$ edge XAS data. Full and dotted lines represent the $r{}_{\mathrm{O}\text{−}\mathrm{Ag}}=d$ ($\mathrm{O}$ $K$ edge) and the $r_{\mathrm{Ni}\text{−}\mathrm{Ag}}$ ($\mathrm{Ni}$ $K$ edge) distances, respectively.

Figure 3

A comparison between experimental (points) and best fit (solid line) for the modulus of the $k^2$-weighted, phase uncorrected FTs of the experimental $\chi \left(k\right)$ of $3\mathrm{ML}(\theta =0°)$. The upper curves refer to a fit including the signals from the tetragonally distorted $\mathrm{NiO}$ film only (i.e., with the same approach used in the fits of all other signals). Bottom curves refer to the fit performed by including the $\mathrm{O}\text{−}\mathrm{Ag}$ signal from the $\mathrm{Ag}$ substrate (dashed dotted line). Also reported are the most relevant contribution to the simulated curve.