Layer-resolved band bending at the $n-\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)/p-\mathrm{Ge}\left(001\right)$ interface

Journal Article · Sat Sep 01 00:00:00 EDT 2018 · Physical Review Materials

DOI:https://doi.org/10.1103/PhysRevMaterials.2.094602· OSTI ID:1485314

Du, Y.; Sushko, P. V.; Spurgeon, S. R.; Bowden, M. E.; Ablett, J. M.; Lee, T. -L.; Quackenbush, N. F.; Woicik, J. C.; Chambers, S. A.

The epitaxial n-STO/p-Ge(001) interface shows considerable promise for harvesting visible sunlight to drive the hydrogen evolution reaction (HER) associated with water splitting [1]. The Ge band gap is small (0.66 eV) and band alignment with STO is favorable for electron-hole pair creation and separation. Moreover, the conduction band minimum of STO is above the HER half-cell potential, allowing for energetically favorable electron transfer across the heterojunction/liquid interface and coupling to the HER. In determining the band alignment for n-STO/p-Ge by XPS, we have previously noted a curious broadening in the Ge 3d core-level line shape relative to that for pure, clean p-Ge(001)-(2x1) [2]. In principle, such broadening could come from a chemically shifted feature resulting from an interface formation product, or from the presence of a built-in potential within the Ge. Data taken on heterostructures with thinner (< ~10 u.c.) STO films using AlK x-rays (1.5 keV) are inconclusive. However, spectra measured on thicker STO/Ge heterojunctions with 6 keV x-rays, along with scanning transmission electron microscopy (STEM) imaging, allow the interface chemical shift explanation to be ruled out. Rather, these data are best explained by band bending in the Ge that spans the entire band gap. We have developed an algorithm that fits the Ge 3d spectrum to a linear combination based on a model function describing the spectrum for flat-band p-Ge. The spectrum for each layer is attenuated in intensity according to its depth. The fitting algorithm starts by assigning randomly generated energy shifts to the spectra for all layers, subject to the boundary condition that the potential profile be a monotonic function of depth. The model spectra are then summed over all layers to generate a simulated heterojunction spectrum and a cost function (i.e. standard deviation) is calculated. Incremental changes are made in the energy shifts and the process is repeated until the standard deviation is minimized. This approach allows binding energies to be assigned to each Ge layer to a depth prescribed by the probe depth (~30 nm). The resulting analysis yields a potential energy profile for the entire heterojunction. The Ge 3d binding energy for the interfacial layer is markedly higher than that for layers deeper into the substrate, consistent with the interface structure determined by STEM imaging and first-principles modeling. The conduction band minimum at the interface is degenerate with the Fermi level. The depletion width is ~3 nm wide, and the valence band maximum at the bottom is degenerate with the Fermi level. This method allows characteristics of built-in potentials at buried interfaces to be probed in a unique and powerful way, and should be useful for other complex oxide heterostructures.

• Research Organization: Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1485314
• Report Number(s): PNNL-SA-134900; PRMHAR
• Journal Information: Physical Review Materials, Vol. 2, Issue 9; ISSN 2475-9953
• Publisher: American Physical Society (APS)
• Country of Publication: United States
• Language: English