LEED-IV study of the rutile TiO
2
„110…-1 Ã 2 surface
with a Ti-interstitial added-row reconstruction
M. Blanco-Rey,
1,
* J. Abad,
2
C. Rogero,
3
J. Méndez,
1
M. F. López,
1
E. Román,
1
J. A. Martín-Gago,
1
and P. L. de Andrés
1
1
Instituto de Ciencia de Materiales (CSIC), Cantoblanco, 28049 Madrid, Spain
2
Centro de Investigación en Optica y Nanofísica, Universidad de Murcia, Campus Espinardo, 30100 Murcia, Spain
3
Centro de Astrobiología (CSIC-INTA), Carretera de Ajalvir km. 4, 28850 Torrejón de Ardoz, Madrid, Spain
Received 22 November 2006; revised manuscript received 22 December 2006; published 9 February 2007
Upon sputtering and annealing in UHV at 1000 K, the rutile TiO
2
110 surface undergoes a 1 1 →1
2 phase transition. The resulting 1 2 surface is Ti rich, formed by strands of double Ti rows as seen on
scanning tunneling microscopic images, but its detailed structure and composition have been subject to debate
in the literature for years. Recently, Park et al. Phys. Rev. Lett. 96, 226105 2006 have proposed a model
where Ti atoms are located on interstitial sites with Ti
2
O stoichiometry. This model, when it is analyzed using
LEED-IV data Phys. Rev. Lett. 96, 0055502 2006, does not yield an agreement between theory and
experiment as good as the previous best fit for Onishi and Iwasawa’s model for the long-range 1 2 recon-
struction. Therefore, the Ti
2
O
3
added row is the preferred one from the point of view low-energy electron
diffraction.
DOI: 10.1103/PhysRevB.75.081402 PACS numbers: 61.14.Hg, 68.35.Bs, 68.47.Gh
Metal oxide surfaces are a subject of strategic interest due
to their wide range of technological applications. They are
popular as a support for heterogeneous catalysis, but are also
highly appreciated for their use in photocatalysis, anticorro-
sion coatings, pigments, gas sensing, biocompatibility, envi-
ronmental applications, etc.
1
Among the different rutile sur-
faces, TiO
2
110 is the most popular due to its stability and
favorable morphology, and it is considered in the literature as
a benchmark for modeling the structure and electronic prop-
erties of metal oxides. The bulk rutile is characterized by
rows of alternatively oriented oxygen octahedra centered at
titanium atoms. The 110 cut produces a relatively flat sur-
face; however, only recently has a structural determination
by low-energy electron diffraction LEED been considered
accurate enough to bring wide consensus to the quantitative
values for the different atomic positions.
2
Most interesting,
moreover, is the 1 2 reconstruction that appears on the
clean surface upon annealing up to 1000 K in ultrahigh
vacuum UHV. From scanning tunneling microscopy
STM it has been well established that this phase consists of
long stripes running along the 001 crystallographic direc-
tion. By careful formation of this new stable phase, these
chains can be formed nearly without interruption on the typi-
cal distances of terraces—i.e., 100– 1000 Å long. These
chains are known to depart from the surface stoichiometry
and are usually characterized as a reduced Ti rich 1 2
reconstruction.
3
From a combined approach using STM,
LEED and density functional theory DFT, recently we pre-
dicted that these stripes may possess interesting quasi-one-
dimensional properties.
4
The complexity of the reconstruc-
tion, however, is enough to allow for many different
geometrical configurations; even a possible reversible trans-
formation between the 1 1 and the 1 2 phases has been
reported,
5
not to mention the existence of cross-links be-
tween chains with an entirely different stoichiometry or other
defects on the surface.
6
Some of these phenomena might
well depend on the careful conditions to recreate the 1 2
reconstruction, and the fact is that this system is a challenge
for both theory and experiment. Therefore, it is important to
realize that the formation of a 1 2 phase may certainly be
influenced by external parameters e.g., UHV or not, the par-
ticular ambience if any, different temperatures and times for
annealing, etc. The geometrical structure of the 1 2 phase
has been much studied in the literature by STM, but these
experiments cannot give a clear description of the atomic
positions or even the surface stoichiometry. Hence, different
proposals can be found in the literature: i a missing-row
model
7
where bridging oxygen atoms are desorbed into
vacuum in competition with diffusion of bulk lattice oxygen
atoms moving to the surface, ii an added-row model with a
Ti
2
O
3
stoichiometry
3
where the Ti cations move to octahe-
dral sites and oxygen atoms stay near their bulklike posi-
tions, iii an added-row model with a Ti
3
O
5
stoichiometry
8
where all the atoms remain near their bulk positions, and iv
a stoichiometric model
6
where the added rows proposed by
Pang et al. have been modified by adding oxygen to create
an added row of Ti
3
O
6
.
LEED is known to be very sensitive to the final disposi-
tion of the atoms on the surface. However, by analyzing only
the Bragg spots, we obtain information about the ordered
long-range structures. To study the geometry of disordered
short-range structures, defects, etc. one should focus on the
diffuse LEED intensities measured at low-temperatures,
which is beyond our experimental setup. We have used
LEED-IV for the 1 2 reconstruction to test all these differ-
ent models against full multiple-scattering calculations.
4
An
objective quantitative comparison between experimental and
theoretical LEED-IV curves can be obtained by computing a
suitable R factor that can be used to identify relevant best-fit
models. We have used Pendry’s R factor,
9
a widely accepted
relevant and accurate number to characterize real
structures.
10
Pendry’s R factor R
P
takes values between 0 and
2; values below 0.3 yield a reasonable good correlation be-
tween theory and experiment, while values over 0.5 mark the
onset of spurious chaotic correlations.
11
Recently, a new structural candidate for the 1 2 recon-
PHYSICAL REVIEW B 75, 081402R2007
RAPID COMMUNICATIONS
1098-0121/2007/758/0814024 ©2007 The American Physical Society 081402-1