Switching Ti Valence in SrTiO3by a dc Electric Field
T. Leisegang,1,*H. Sto ¨cker,1A.A. Levin,1T. Weißbach,1M. Zschornak,1E. Gutmann,1
K. Rickers,2S. Gemming,3and D.C. Meyer1
1Institut fu ¨r Strukturphysik, Technische Universita ¨t Dresden, 01062 Dresden, Germany
2Hamburger Synchrotronstrahlungslabor HASYLAB am Deutschen Elektronen-Synchrotron DESY,
Notkestrasse 85, 22603 Hamburg, Germany
3Institut fu ¨r Ionenstrahlphysik und Materialforschung, Forschungszentrum Dresden-Rossendorf, 01314 Dresden, Germany
(Received 8 May 2008; revised manuscript received 19 December 2008; published 23 February 2009)
A (001) SrTiO3wafer has been investigated in situ at room temperature under application of a static
electric field of varying polarity by fluorescence x-ray absorption near edge structure (XANES) analysis at
the Sr-K and Ti-K absorption edges. The XANES spectra show a clear shift of the Ti-K absorption edge
energy. The shift is attributed to a change of the Ti valence state in a volume invoked by diffusion of the
oxygen ions and vacancies. No shift was observed for the Sr-K absorption edge energy. Theoretical
calculations support these findings.
DOI: 10.1103/PhysRevLett.102.087601 PACS numbers: 77.84.Bw, 61.05.cj, 66.30.Qa, 71.55.?i
During the last decade, extensive investigations have
focused on ferroelectric, ferromagnetic, and ferroelastic
materials . Their ferroic properties arise from a sponta-
neous long-range ordering of, e.g., electric dipoles or
magnetic moments or strained orientation domains.
Important characteristics of these materials are, e.g., high
dielectric constants, piezo- or pyroelectricity or giant mag-
netoresistance. A coupling of at least two ferroic proper-
ties, named multiferroicity, allows for a large diversity of
upcoming applications, e.g., microelectromechanical sys-
tems, magnetoelectric transducers, microwave electronics,
ferroelectric field effect transistors, as well as data storage
and random access memory devices [1–4].
Among ferroic or multiferroic materials, oxides with
perovskite-type structure have gained great importance.
To maintain structural integrity, they are mostly grown
epitaxially. SrTiO3(STO) is widely used as a substrate
material due to its compatible lattice parameters, structure,
and comparatively low chemical reactivity . Pristine
STO is characterized by a perovskite-type cubic structure
[lattice parameter a0¼ 3:905?A at room temperature
(RT)]  and exhibits no crystallographic phase transitions
down to a temperature of ?105 K and no ferro- or piezo-
electric behavior down to helium temperatures .
The growth as well as processing of oxide single crys-
tals, such as STO, introduces a specific density of Schottky
point defects. Although several defect reducing approaches
are utilized, in particular, near-surface regions still exhibit
a distorted structure [3,7]. Essentially oxygen vacancies,
introducing lattice distortions and charge carriers, show a
strong impact on structural and electronic properties, e.g.,
phase transitions,ionicconductivity,and resistanceswitch-
In comparison with the ideal crystal structure, a statis-
tical distribution of oxygen vacancies in STO results in an
expansion of the average cubic unit cell . However,
they can also arrange periodically forming linear or planar
defects causing the formation of intergrowth structures
. Hence, specific well-defined oxygen-deficient te-
tragonal phases, i.e., SrTiO3-?and SrTi1-xO3-y[x¼1=ðnþ
1Þ, y ¼ 2=ðn þ 1Þ, n ¼ 1;2;...], can be formed [12,13].
Application of a static (dc) electric field to a STO crystal
results in a redistribution of oxygen vacancies and subse-
quently in structural changes. Early experiments revealed
ionic conductivity in STO and the influence of a dc electric
field thereon . It has been suggested that either oxygen
vacancies or SrO complexes gain an enhanced diffusion
coefficient when transported along dislocation lines or
planar defects [9,15]. Recently, for (001) STO wafers we
have found a reversible change of the x-ray reflection
profiles using wide-angle x-ray diffraction under applica-
tion of a dc electric field at RT . This phenomenon has
been attributed to structural variations observed beneath
the anode (usually at the as-cut unpolished side) and prom-
ises application in the field of adaptive x-ray optics .
In the present report we focus on the characterization of
the near-surface structure of STO under the influence of an
electric field. We performed dedicated fluorescence x-ray
absorption near edge structure (XANES) investigations at
the Sr-K and Ti-K absorption edge of a (001) STO single-
crystal plate (supplied by Crystec GmbH, Germany, etched
a dc electric field. The measurements were carried out at
beam line C (CEMO)  of the Hamburger Synchrotron-
strahlungslabor (HASYLAB) at Deutsches Elektronen-
Synchrotron (DESY) using grazing incidence of the excit-
ing x rays. This experiment allows investigating changes of
the valence states of the Sr and Ti atoms based on the
binding energy of the resonantly excited 1s K-shell elec-
trons. The STO crystal plate (size 10 ? 10 ? 0:5 mm3)
used for this experiment was coated with W electrodes
(thickness t ? 70 nm) on top of B4C diffusion barriers
PRL 102, 087601 (2009)
27 FEBRUARY 2009
? 2009 The American Physical Society
(t ? 30 nm) on both sides of the plate, so that a voltage of
U ¼ ?500 V,correspondingtoan electricfieldstrengthof
Ez¼ ?106V=m, could be applied (see Fig. 1). The elec-
tric field was applied for 115 min (U ¼ þ500 V) and
19 min (U ¼ ?500 V) in case of the Ti and at least 3 times
longer in case of the Sr-K XANES measurements. The
voltage source used (Physik Instrumente E-507) ensured a
constant voltage and thus a constant electric field over
time. The electric current through the sample, monitored
throughout the whole experiment, increased generally but
remained below 10?7A. A detailed description of the
behavior of the electric current will be given elsewhere.
The XANES spectra were collected repeatedly using a
fixed-exit Si (111) double crystal monochromator 
for an energy range of the exciting photons of E ¼
16000...16150 eV and E ¼ 4960...5000 eV (Sr-K
and Ti-K absorption edge) with an energy step width of
?E ¼ 0:5 eV and ?E ¼ 1 eV, respectively. An angle of
incidence ! ¼ 0:15?was chosen to probe the distorted
near-surface region (attenuation depth of about 200 and
20 nm for Sr-K and Ti-K edge energies of x-ray radiation,
respectively) with most significance. An energy resolving
solid state detector (Kevex PSI) was arranged perpendicu-
lar to the plane of scattering so that the fluorescence yield
IFcould be collected in the direction of the polarization
vector of the synchrotron radiation enhancing the signal-
to-noise ratio. XANES spectra were obtained by integra-
tion of the fluorescence spectra over energy ranges of
interest (ROI) 13.3...15.3 keV (Sr-K?) and 4.3...5.0 keV
(Ti-K?), respectively, and normalization relative to the
primary beam intensity. Calibration of the energy scale
was done by testing absorption edge energies of several
metal foils. The exciting photons were guided through an
evacuated (p < 10?6mbar) beam guide. Three different
XANES spectra were recorded, one in zero-electric field
and two during application of the electric field with differ-
ent polarity. After every change of the electric field
strength, a realignment procedurewas performed to correct
for height and tilt changes of the irradiated sample surface,
both caused by the electric field .
For the investigated side, the in situ XANES spectra
exhibited no differences within the estimated error in the
case of the Sr-K absorption edge [Fig. 2(a)] indicating an
unchanged or at most marginally affected valence state of
the Sr atoms under the influence of the electric field
applied. In case of the Ti atoms a significant shift of the
Ti-K absorption edge energy (determined by the maximum
of first derivative of IF;normðEÞ which was fitted with a
Gaussian profile) of ð1:29 ? 0:06Þ eV due to the electric
field was observed [Fig. 2(b)]. In particular, the voltage
applied to the sample surface increased (U ¼ þ500 V) or
decreased (U ¼ ?500 V) the Ti-K absorption edge energy
within the crystal volume probed in the experiment. In case
of a positive voltage the Ti-K absorption edge energy
shifted to higher values over time with a rate of ð0:005 ?
0:001Þ eV=min. Within the time period investigated
(115 min) no plateau was observed. The short duration of
the electric field application with negativevoltage (19 min)
did not allow elucidating the time dependence.
A rather qualitative interpretation of the results can be
drawn in terms of ionic conductivity. The near-surface
FIG. 1 (color online).
mental setup with incident angle ! of the exciting x rays.
Schematic representation of the experi-
FIG. 2 (color online).
(a) the Sr-K and (b) the Ti-K (ROI: 4.3...5.0 keV) absorption
edges under application of a dc electric field. The inset in (a)
gives an enlarged view of the Sr-K absorption edge region. An
exemplary XANES spectrum for the completely integrated en-
ergy range (ROI: 3.1...8.5 keV) of the fluorescence yield show-
ing the well-known preedge features of Ti is depicted in the inset
XANES spectra for photon energies at
PRL 102, 087601 (2009)
27 FEBRUARY 2009
regionof the unpolished STO plate is considered to contain
an increased amount of oxygen vacancies. By applying a
is forced to migrate from the bulk to the anode so that it
compensates the intrinsic oxygen vacancies at the irradi-
ated near-surface region. Thereby, the coordination of Ti
atoms changes from square-planar or pyramidal (in case of
oxygen vacancy) to octahedral symmetry, the formal Ti
valence state shifts from Tið4??Þþnext to a vacancy back to
Ti4þand the ideal STO structure is recovered in the near-
surface volume probed. Thus, the shift of the Ti-K absorp-
tion edge energy to a higher value in Fig. 2(b) can be
attributed to a filling of oxygen vacancies by oxygen
diffusion mediated by the electric field.
Inversion of the polarity (U ¼ ?500 V), i.e., the near-
surface region acts as cathode, reverses the oxygen anion
transport and the Ti-K absorption edge energy decreases,
implying a change of the formal oxidation state from Ti4þ
to Tið4-?Þþ[Fig. 2(b)]. The final state at the cathode in-
dicates that the Ti-K absorption edge energy is shifted to
energies lower than that observed for the state without an
electric field. Consequently, more oxygen atoms diffuse
into the bulk material thereby increasing the density of
near-surface vacancies. Referring to previous investiga-
tions planar and line defects are assumed to be the most
significant diffusion paths [9,15,17,21].
Comparing the shift of the Ti-K absorption edge energy
with literature data on various Ti containing compounds
with different Ti oxidation states (shift of 3.83 eV between
formal Ti3þand Ti4þ) , we can note that our data are
consistent with a change of the Ti valence state. In addi-
tion, the form of the XANES spectra at the Ti-K absorption
edge also depends on structural disorder. Comparing the
white lineregionofthegraphsinFig. 2(b)withtherecently
reported spectra in Refs. [23–25] we can state that our
model is consistent with a highly crystalline surface in case
of positive voltage and with a comparatively more per-
turbed surface for negative voltage.
The well-known Ti preedge features are visualized by
using the completely integrated fluorescence spectra [ROI:
3.1...8.5 keV, inset of Fig. 2(b)]. However, because of a
low signal-to-noise ratio, these data do not allow for inter-
pretation of changes of XANES spectra induced by appli-
cation of the electric field. Therefore, the limited data
(ROI: 4.3...5.0 keV for Ti-K?) were used for analysis of
the XANES spectra described above. By using a limited
integrated spectral range, preedge features can disappear
 as was observed here [Fig. 2(b)].
In the following we substantiate the qualitative sugges-
tions with a more quantitative approach. Low-lying core
states such as the Ti 1s state can be employed as local
probe for the chemical potential in the vicinity of a given
site. Shifts of such levels are commonly related to changes
of the local crystal potential due to changes of the coordi-
nation geometry or the formal oxidation state . Core-
level shifts are also accessible by all-electron density-
functional calculations and changes of ionization energies
can be obtained and interpreted with this methodology
Here, we model the influence of neutral oxygen vacan-
cies VOon the Ti 1s core level by scalar-relativistic all-
electron local-spin-density-functional (LSDA þ U) calcu-
lations with the FPLO-5 code . A supercell with four
SrTiO3formula units and cell dimensions of a ¼ b ¼ 2a0
and c ¼ a0was employed and repeated by periodic bound-
ary conditions. An ordered array of oxygen vacancies was
simulated by removing one oxygen atom from this super-
cell (see Fig. 3) resulting in a composition of SrTiO2:75;
i.e., about 8% of oxygen sites are vacant.
This supercell reproduces the favored chainlike accu-
mulation of oxygen vacancies along a h001i direction
obtained in a previous valence-only density-functional
study . For the present calculation we adopted all
numericalsettings accordingto thevalues established there
, especially the on-site Coulomb and exchange terms
with an effective parameter Ueff ¼ 4:46 eV. In the super-
cell each vacancy chain and its periodic replica are at least
7.802 A˚apart and separated by a volume with ideal bulk-
type structure. Three Ti sites can be distinguished in this
model: Ti1, the most strongly perturbed, square-planar
coordinated neighbor site within the vacancy chain at a
distance dðTi1-VOÞ ¼ 1:95 ?A and the more distant six-
fold coordinated sites Ti2 at dðTi2-VOÞ ¼ 4:36?A and Ti3
at dðTi3-VOÞ ¼ 5:85?A. As the calculation does not yield
absolute XANES edge onsets, the 1s orbital of the most
bulklike Ti3 atom was employed as an internal standard for
the Ti 1s core level. We obtained a Ti 1s binding energy
reduction of 1.1 eV for the Ti1 atom in the vacancy chain
oxygen vacancy VO
Layer ALayer B
FIG. 3 (color online).
LSDA þ U calculation. The different Ti environments and the
oxygen vacancy are indicated.
Sketch of the supercell used for the
PRL 102, 087601 (2009)
27 FEBRUARY 2009
and of 0.3 eV for Ti2. These values reflect very well the Download full-text
experimentally observed shift of the Ti-K absorption edge
energy and, in particular, the formation of the shoulder-
type XANES in the case of U ¼ ?500 V compared to
ideal STO at U ¼ þ500 V. For the Sr 1s binding energy
no prediction can be made because only a unique Sr site
was incorporated in our model. Thus, we ascribe the ob-
served Ti-K absorption edge energy shift to the influence
of O vacancies on the local electronic structure at the
adjacent Ti sites.
In conclusion, XANES measurements proved a shift of
the Ti-K absorption edge energy of STO under applied dc
electric fields, which has not been reported so far to the
best of our knowledge. The results of theoretical calcula-
tions show that such shifts of the Ti-K absorption edge
energy can be caused by oxygen vacancy redistribution.
Taking into account the observed energy shift of the Ti-K
absorption edge, a change of vacancy concentration in the
order of 8% in the near-surface region was estimated. The
Sr-K absorption edge energy, in contrast, is not affected.
These investigations provide new insight into the origin
of the remarkable changes of the properties occurring
in STO in an applied electric field at the atomic scale.
Thus, a controlled switching of the Ti valance gives rise to
a large variety of interesting applications and physical
phenomena, e.g., dedicated valence states for controlled
catalytic behavior on STO surfaces or tuning of super-
conductivity  or of insulator-to-metal transition .
The authors are indebted to D. Novikov and E. Welter
(HASYLAB) for supporting the experiments, DESY for
granting beam time and to DFG (research unit FOR 520
and Projects No. ME 1433/4-4, No. GE 1037/9-1, and
No. GE 1202/5-1) for financial support.
 N. Setter et al., J. Appl. Phys. 100, 051606 (2006).
 K. Do ¨rr and C. Thiele, Phys. Status Solidi B 243, 21
 Physics of Ferroelectrics: A Modern Perspective (Topics
in Applied Physics), edited by K. Rabe, Ch.H. Ahn, and
J.-M. Triscone (Springer, Berlin, 2007), Vol. 105.
 S. Gemming et al., J. Comput.-Aided Mater. Des. 14, 211
 H.E. Swanson and R.K. Fuyat, Natl. Bur. Stand. Circ.
(U.S.) 539, 44 (1954); ICDD-PDF2, No. 035-0734.
 H. Unoki and T. Sakudo, J. Phys. Soc. Jpn. 23, 546 (1967).
 U. Ru ¨tt, A. Diedrichs, J.R. Schneider, and G. Shirane,
Europhys. Lett. 39, 395 (1997).
 J.P. Buban, H. Iddir, and S. Ogu ¨t, Phys. Rev. B 69, 180102
 K. Szot, W. Speier, G. Bihlmayer, and R. Waser, Nature
Mater. 5, 312 (2006).
 R. Perez-Casero et al., Phys. Rev. B 75, 165317 (2007).
 L.E. Smart and E.A. Moore, Solid State Chemistry
(Taylor & Francis, Boca Raton, 2005).
 W. Gong et al., J. Solid State Chem. 90, 320 (1991).
 S.N. Ruddlesden and P. Popper, Acta Crystallogr. 11, 54
 E. Hegenbarth, Phys. Status Solidi 6, 333 (1964).
 M. Bobeth et al., J. Ceram. Soc. Jpn. 114, 1029 (2006).
 D.C. Meyer et al., Appl. Phys. A 80, 515 (2005).
 D.C. Meyer et al., Appl. Phys. A 84, 31 (2006).
 K. Rickers et al., in X-Ray Absorption Fine Structure–
Absorption Fine Structure, Stanford, 2006, edited by
Britt Hedman and Piero Pianetta, AIP Conf. Proc.
No. 882 (AIP, New York, 2007), p. 905.
 K. Rickers et al., in Proceedings of the 9th International
Conference on Synchrotron Radiation Instrumentation,
Daegu SRI 2006, edited by Jae-Young Choi and
Seungyu Rah, AIP Conf. Proc. No. 879 (AIP, New York,
2007), p. 907.
 T. Leisegang et al., HASYLAB Ann. Rep. 1, 1109 (2007).
 W.T. Lee, E.K.H. Salje, L. Goncalves-Ferreira, M.
Daraktchiev, and U. Bismayer, Phys. Rev. B 73, 214110
 J. Graetz, J.J. Reilly, J. Johnson, A. Yu. Ignatov, and T.A.
Tyson, Appl. Phys. Lett. 85, 500 (2004).
 A.I. Frenkel et al., Phys. Rev. Lett. 99, 215502 (2007).
 C. Sabathier, J. Chaumont, S. Rouziere, and A. Traverse,
Nucl. Instrum. Methods Phys. Res., Sect. B 234, 509
 F. Farges, G.E. Brown, Jr., and J.J. Rehr, Phys. Rev. B 56,
 F.M.F. de Groot, in X-Ray Absorption Fine Structure–
XAFS13 (Ref. ), p. 37.
 A. Baraldi, J. Phys. Condens. Matter 20, 093001 (2008).
 A.T. Paxton, A.J. Craven, J.M. Gregg, and D.W.
McComb, J. Microsc. 210, 35 (2003).
 S. Nufer, T. Gemming, C. Elsa ¨sser, S. Ko ¨stlmeier, and M.
Ru ¨hle, Ultramicroscopy 86, 339 (2001).
 K. Koepernik and H. Eschrig, Phys. Rev. B 59, 1743
 Do Duc Cuong et al., Phys. Rev. Lett. 98, 115503 (2007).
 J.F. Schooley, W.R. Hosler, and E. Ambler, Phys. Rev.
Lett. 12, 474 (1964).
 P. Calvani, M. Capizzi, and F. Donato, Phys. Rev. B 47,
PRL 102, 087601 (2009)
27 FEBRUARY 2009