Final-state effect on x-ray photoelectron spectrum of nominallyd1andn-dopedd0transition-metal oxides

We investigate the x-ray photoelectron spectroscopy (XPS) of nominally ${d}^{1}$ and $n$-doped ${d}^{0}$ transition-metal oxides including ${\mathrm{NbO}}_{2},\phantom{\rule{0.16em}{0ex}}{\mathrm{SrVO}}_{3}$, and ${\mathrm{LaTiO}}_{3}$ (nominally ${d}^{1}$), as well as $n$-doped ${\mathrm{SrTiO}}_{3}$ (nominally ${d}^{0}$). In the case of single phase ${d}^{1}$ oxides, we find that the XPS spectra (specifically photoelectrons from Nb $3d$, V $2p$, Ti $2p$ core levels) all display at least two, and sometimes three distinct components, which can be consistently identified as ${d}^{0},\phantom{\rule{0.16em}{0ex}}{d}^{1}$, and ${d}^{2}$ oxidation states (with decreasing order in binding energy). Electron doping increases the ${d}^{2}$ component but decreases the ${d}^{0}$ component, whereas hole doping reverses this trend; a single ${d}^{1}$ peak is never observed, and the ${d}^{0}$ peak is always present even in phase-pure samples. In the case of $n$-doped ${\mathrm{SrTiO}}_{3}$, the ${d}^{1}$ component appears as a weak shoulder with respect to the main ${d}^{0}$ peak. We argue that these multiple peaks should be understood as being due to the final-state effect and are intrinsic to the materials. Their presence does not necessarily imply the existence of spatially localized ions of different oxidation states nor of separate phases. A simple model is provided to illustrate this interpretation, and several experiments are discussed accordingly. The key parameter to determine the relative importance between the initial-state and final-state effects is also pointed out.