Total-Reflection Inelastic X-Ray Scattering from a 10-nm Thick ${\mathrm{La}}_{0.6}{\mathrm{Sr}}_{0.4}{\mathrm{CoO}}_3$ Thin Film

To study equilibrium changes in composition, valence, and electronic structure near the surface and into the bulk, we demonstrate the use of a new approach, total-reflection inelastic x-ray scattering, as a sub-keV spectroscopy capable of depth profiling chemical changes in thin films with nanometer resolution. By comparing data acquired under total x-ray reflection and penetrating conditions, we are able to separate the O $K$-edge spectra from a 10 nm ${\mathrm{La}}_{0.6}{\mathrm{Sr}}_{0.4}{\mathrm{CoO}}_3$ thin film from that of the underlying ${\mathrm{SrTiO}}_3$ substrate. With a smaller wavelength probe than comparable soft x-ray absorption measurements, we also describe the ability to easily access dipole-forbidden final states, using the dramatic evolution of the La $N_{4,5}$ edge with momentum transfer as an example.

Figure 1

The geometry of the incident and inelastically scattered x rays is shown. The momentum transfer $q$ is defined primarily by the angle between $k_1$ and $k_2$ while the energy loss is given by the monochromator energy $E_1$ and the analyzer energy $E_2$. The incident angle $\alpha$ can be used to tune the penetration depth. The inset depicts the Bragg scattering geometry of the sample analyzer and detector that defines $E_2$.

Figure 2

(a) The IXS spectrum for $q$ averaged over $8.3–9.5{\AA }^{-1}$ at $\alpha >{\alpha }_c$ and $\alpha <{\alpha }_c$, is shown. As demonstrated by the diagram in the upper left corner, the $\alpha =5°$ is bulk sensitive since incident beam refracts $1\mu \mathrm{m}$ into the sample whereas the grazing $\alpha =0.2°$ spectrum only penetrates 5 nm. In each case, the broad, dispersive Compton profile (valence IXS), peaked at 300 eV, dominates the spectrum, while the fixed-energy core-shell features indicate the sensitivity to the substrate and film at angles above and below ${\alpha }_c$. (b) To highlight the individual core-shell excitations, the same spectrum is replotted with the Compton scattering removed by a polynomial fit. The individual peaks are labeled and error bars based on Poisson statistics from the total IXS signal are given in the residual intensity. In both parts, the STO spectrum has been normalized and offset from the LSCO TRIXS data.

Figure 3

(a) Normalized oxygen $K$ edge for LSCO and STO measured at $\alpha =0.2$ and 5°, respectively. Data have been averaged over all $q$. Splines through the data points are provided as a guide. (b) FEFF8.4 calculations for LSCO and STO spectra.

Figure 4

(a) La $N_{4,5}$-edge contribution to measured IXS from LSCO ($\alpha =0.25°$) as a function of $q$. Splines through the data points are provided as a guide. (b) Multiplet calculation for $4d^{10}{f}^0$ atomic states as a function of $q$ (top) and the angular momentum components (bottom). The calculation, originally compared to Ce $4f^0$ data [15], has been shifted by $-9\mathrm{eV}$ to match the lower binding energy of the lanthanum edge. Data shown in (a) and (b) have been offset and scaled for comparison. Intensity from the calculated dipole resonance has been truncated to highlight higher order multiplets at intermediate $q$.