BTU overview
Faculty 1
Chair of Applied physics / sensor technology Prof. Dr. rer. nat. habil. Dieter Schmeißer

Polarons in oxidic semiconductors- the SrTiO3 case

Thomas Heller and Dieter Schmeißer
Angewandte Physik - Sensorik, Brandenburgische Technische Universität Cottbus

Oxidic semiconductors with their fascinating physical phenomena are an important field with powerful technological applications such as smart windows and superconductivity, to name a few. In this contribution we focus on SrTiO3(100) and stress the nature of a defect state within the electronic gap which appears upon doping. Doping is induced by either heating in vacuum or by Na exposure and causes the intensity of the defect state D to increase (Fig.1). The nature of this defect state is topic of many studies, it is commonly described to be indicative of metal (Ti 3d) orbitals as claimed from resonant photoemission results [1-3].

Fig.1: The valence band spectra of SrTiO3(100) (h*ν = 45eV) taken from the bare surface (solid line), after heating (dash dotted), and after Na-exposure (dashed).

We vary the surface preparation and obtain the valence band spectrum after several sputter annealing cycles which shows almost no emission in the gap (solid line). After heating (dash-dotted curve) we see an emission 0.7eV below EFermi in the gap and a band bending of about 0.3eV. For the Na exposed surface (dotted curve) the structures A and B are not definitely separated and the band bending is about 0.4eV.
We measure the photo-ionization cross-section (PIX) of that defect state D and compare it to the characteristic valence band features A an B (see Fig.1). We record the apparent intensity of these defined initial states (ADI) in a CIS-like experiment upon varying the incident photon energy. The ADI curves monitor the final state contributions that occur for the defined initial states A, B, and D, respectively. These curves represent the PIX of the corresponding atomic parent wave function as long as there is no additional final state contribution.
Consequently, such curves should be different if the valence states A and B are from O2p atomic levels and the defect state D has a main contribution from the metal (Ti3d) states. However, our data (Fig.2) give evidence that there are almost no differences between these states. The spectra of the as obtained surface shows smoothly decaying curves as expected for p- or d- like atomic wave functions. There are distinct changes upon doping which are very similar for the states labeled A and D as the curves show a significant increase at elevated photon energies. There appear new features with an onset at around 40eV and a broad maximum next to 60eV and 70eV. There is a weak modulation right at the Ti3p threshold (37eV). It might be assigned to a resonant photoemission process. However, it occurs for the valence state A as well as for the defect state. In both cases it is not the dominating contribution at all.

Fig.2: Comparison of the ADI-A of SrTiO3 (100) for the bare oxide (lower curve) and after Na exposure (upper curve).

The absence of sharp resonances (the Na2p resonance at 31eV is assigned to occur from an excitonic excitation and will be discussed elsewhere) indicates that there are no dominating contributions due to resonant photoemission involving the Ti3d/Ti3p threshold. In particular, the increase in the ADI curves around 40eV has been observed earlier repeatedly and assigned to that process. According to our data that assignment is questionable as that onset is part of a very broad resonance and as the curves of the valence states and that of the defect states are very similar. In particular, there is no difference at the Ti3p threshold (37eV). Hence, the defect state D has a significant contribution from the O2p levels which dominate in the valence band and determine the PIX of the state A. Such defect states may be formed by a displacement of the oxygen atoms in a local environment. These states are called polarons and are stabilized by electron-phonon interactions. Their existence is established for oxidic surfaces as derived from optical data. Similar data are found for doped and undoped TiO2(110) surfaces as well as for WO3 films.


Consequently, we postulate that polarons may appear upon doping and are a major defect type in these oxidic semiconductors. [1] PACox, V.E.Henrich, "The Surface Science of Metal Oxides", Cambridge Univ. Press (1994).[2] S.Hüfner, "Photoelectron Spectroscopy", Springer Series in Solid-State Sciences (1982).[3] R.Courths, S.Cord, H.Saalfeld, Solid State Communications 70, 1047(1989).[4] E.K.H.Salje, A.S.Alexandrov, W.Y.Liang, "Polarons and Bipolarons in High-Tc Superconductors and Related Materials", Cambridge University Press (1995). We study the core level and the valence band emission by photoelectron spectroscopy (TGM7, ADES) and find characteristic differences between the undoped and the doped oxidic semiconductor.After Na exposure a broad feature appears between 18.5eV and 25eV superimposed on a clearly increased intensity with several weak but distinct features at 37eV and between 45eV and 64eV. There is a heavy structure at 31eV near at the binding energy of the Na 2p. The ADI-A show, Fig.2) even for the fresh surface, a strong increase for photon energies above 18eV which results in pronounced structures in the ranges around 33eV and 50eV. In the first range there are maxima at 27eV, 33eV and in the second by 47eV. Na exposure causes significant changes as There is a shoulder in that the emission from the second range is also increased. Similarly things we observe for the heated surface.We observe for the derived measurements similar results for ADI-D and ADI-A has the D-state in the ADI-D at least significant contributions from the valence band (A-state). Therefore has the defect state not only d-band character and these behaviour is expected for Polarons [4].Similar data we obtained for WO3-films and ZnO (000-1) crystals.