Formation of aluminum oxide thin films on FeAl(1 1 0) studied by STM

Article abstract of DOI:10.1016/S0039-6028(03)00271-1

The surface morphology and atomic structure of clean and oxidized FeAl(1 1 0) surfaces have been investigated with scanning tunneling microscopy (STM). An incommensurate reconstructed structure, having FeAl₂ stoichiometry confined to the outmost layer, is observed on the clean surface due to preferential Al segregation upon annealing to 1125 K. When the reconstructed clean surface is exposed to oxygen at elevated temperatures, an ordered ultra-thin aluminum oxide film is formed. Based on STM data, a structural model of the oxide film is proposed, which exhibits a quasi-hexagonal oxygen layer and accommodates an even mix of octahedral and tetrahedral occupancy of Al ions arranged in an alternating zigzag-stripe structure. STM imaging with tunnel voltages in the range of the bulk band gap implies that the thin film oxide electronic structure differs substantially from the bulk oxide, and indicates a local density of states around the oxide constituents within the bulk band gap.

Full text of DOI:10.1016/S0039-6028(03)00271-1

Surface Science 529 (2003) 223–230
www.elsevier.com/locate/susc
Formation of aluminum oxide thin films on FeAl(110)
studied by STM
a
b
a,
*
O. Kizilkaya , D.A. Hite a,1, D.M. Zehner , P.T. Sprunger
a
Department of Physics and Astronomy, Center for Advanced Microstructures and Devices, Louisiana State University,
6980 Jefferson Highway, Baton Rouge, LA 70803, USA
b
Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
Received 14 November 2002; accepted for publication 6 February 2003
Abstract
The surface morphology and atomic structure of clean and oxidized FeAl(1 1 0) surfaces have been investigated with
scanning tunneling microscopy (STM). An incommensurate reconstructed structure, having FeAl2 stoichiometry
confined to the outmost layer, is observed on the clean surface due to preferential Al segregation upon annealing to
1125 K. When the reconstructed clean surface is exposed to oxygen at elevated temperatures, an ordered ultra-thin
aluminum oxide film is formed. Based on STM data, a structural model of the oxide film is proposed, which exhibits a
quasi-hexagonal oxygen layer and accommodates an even mix of octahedral and tetrahedral occupancy of Al ions
arranged in an alternating zigzag–stripe structure. STM imaging with tunnel voltages in the range of the bulkband gap
implies that the thin film oxide electronic structure differs substantially from the bulkoxide, and indicates a local density
of states around the oxide constituents within the bulkband gap.
Ó 2003 Elsevier Science B.V. All rights reserved.
Keywords: Aluminum; Iron; Metal–insulator interfaces; Scanning tunneling microscopy; Oxidation; Surface relaxation and
reconstruction
1. Introduction
oxides, alumina has been the focus of numerous
studies due to its desirable physical and chemi-
cal properties. Bulkalumina gives rise to sev-
eral phases, all of which are comprised of a close
packed oxygen sublattice. The differences in oc-
cupancy of tetrahedral and/or octahedral coordi-
nation determine the crystal phase of alumina, and
consequently, the overall properties. Despite years
of investigation, developing a fundamental un-
derstanding of their complex surface/thin-film
electronic and atomic properties has been hindered
by the lackof high quality oxide surfaces that can
be probed with standard surface/thin-film tech-
niques.
Due to their importance in a broad range of
technological applications such as catalytic sup-
ports, microelectronics devices, and metal ceramic-
based sensors, studies of oxide surfaces continue to
draw considerable attention [1–3]. Among the
* Corresponding author. Tel.: +1-225-5784621; fax: +1-225-
5786954.
E-mail address: phils@lsu.edu (P.T. Sprunger).
1
Present address: National Institute of Standards and
Technology, Magnetic Technology Division, Boulder, CO
80305, USA.
0039-6028/03/$ - see front matter Ó 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0039-6028(03)00271-1

224
O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
These experimental obstacles, including sur-
observed for FeAl(1 1 0), and XPS indicates that
Al is the only cation in the oxide film, indicating
formation of Al2O3, rather than any iron-oxide
complex. Still, the detailed atomic and electronic
structure of these oxide films has not been ex-
plored.
In the present paper, we report on an investi-
gation of the nucleation, growth, and atomic
structure of this ultra-thin oxide film grown on
FeAl(1 1 0). In order to interpret the resulting ox-
ide structure, STM results of the clean, high-tem-
perature reconstructed phase of the FeAl(1 1 0)
surface will also be presented. Because of strong
similarities, the electronic and atomic structure of
the resulting alumina film on FeAl(1 1 0) will be
discussed in light of both experimental and theo-
retical results for aluminum oxide growth on
NiAl(1 1 0) and other substrates.
face charging, can be minimized by the growth of
ultra-thin oxide films on well-characterized metal
substrates. Due to lattice mismatch effects, the
production of alumina films from direct oxidation
of aluminum results in morphologically rough
surfaces and short-range crystallographic growth.
However, previous studies have shown that ultra-
thin alumina films, possessing long-range crystal-
lographic structure have been successfully grown
on other metal and alloy substrates.
Most notably, alumina growth on NiAl surfaces
has been the subject of many studies, which have
characterized the resulting dynamic, geometric and
electronic structure of oxide surfaces, thin-films,
and metal–oxide interfaces. Specifically, upon ox-
idation of clean alloy surfaces at elevated temper-
atures, different phases of well-ordered thin Al2O3
films have been identified on Ni3Al(1 1 1) [4,5],
NiAl(1 1 1) [6,7] and NiAl(1 1 0) [8,9]. X-ray pho-
toelectron spectroscopy (XPS), low energy elec-
tron diffraction (LEED) and scanning tunneling
microscopy (STM) studies have demonstrated that
2. Experimental details
The preliminary STM experiments were con-
ducted at OakRidge National Laboratory using
an Omicronâ room temperature STM, followed
by workcompleted at the Center for Advanced
Microstructure and Devices at Louisiana State
University, using a variable temperature STM,
described in detail elsewhere [11]. Both STM
measurements were performed in ultra-high vac-
uum chambers, equipped with LEED optics, cy-
lindrical mirror analyzers for AES, as well as other
surface cleaning/characterization instruments. The
tunneling voltages refer to the bias on the sample
with the tip at virtual ground. The lattice constant
of the FeAl single crystal, grown at the Max Plank
Institut fur Eisenforschung, was determined with
ꢀ
a well-ordered 5 A thick c-Al2O3 film has been
produced on NiAl(1 1 0) after dosing 1200 L of O2
at elevated temperatures and annealing to 1150–
1250 K for a short time. Two layers of oxygen and
aluminum ions were proposed for these epitaxially
grown oxides. High-resolution electron energy-loss
spectroscopy experiments have confirmed that this
oxide has c-Al2O3 like, rather than a-Al2O3-like,
structure, and is terminated with a hexagonal ar-
rangement of oxygen ions [9]. Due to the symme-
try of the substrate, two domains of oxide have
been reported in accordance with STM images and
LEED patterns [8,9].
Like NiAl, FeAl is also an ordered intermetallic
alloy that crystallizes in the same CsCl-type crystal
structure. In contrast to the well investigated oxi-
dation behavior of NiAl(1 1 0), there are only
preliminary Auger electron spectroscopy (AES),
LEED and XPS studies of oxide films grown on
FeAl(1 1 0) [10]. Graupner et al. reported that oxy-
gen dosing at room temperature leads to the for-
mation of amorphous Al2O3 with a thickness of
ꢀ
X-ray diffraction to be 2.903 A, corresponding to a
constituency of Fe0:53Al0:47 [12]. This composition
of the constituents prevents the phase transition,
seen at 50–50% of the bulkalloy, between FeAl
and FeAl2 structures at low temperatures [13]. The
(1 1 0) sample was aligned to ꢁ0.3° using Laue
diffraction and mechanically polished [12]. The 5
mm diameter sample was mounted on a Ta sample
platen with a K-type thermocouple attached to
determine the sample temperature. The substrate
was cleaned by repeated cycles of sputtering at
ꢀ
ꢀ5 A. However, elevated temperature oxidation
results in a well-ordered oxide film. LEED pat-
terns for similarly oxidized NiAl(1 1 0) are also

O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
225
room temperature (Neþ ions, 1 keV, 20 lA), fol-
lowed by e-beam annealing to 1125 K. The oxide
thin film was prepared by dosing, at 1 Â 10À6
Torr, the sample with oxygen, typically at 1125 K
and consequently annealed for 5 min at this tem-
perature.
3. Results and discussion
3.1. Atomic structure of the clean, reconstructed
FeAl(110) surface
In starkcontrast to NiAl surfaces, previous
LEED and AES studies indicate that clean sur-
faces of low-index and vicinal surfaces of FeAl
undergo surface reconstructions [14,15]. In the
case of the (1 1 0) surface, it has been shown that
preferential sputtering creates an attenuated Al
concentration profile in the selvage region; how-
ever sequential annealing promotes aluminum
segregation to the surface. Annealing to the tem-
perature range of 1075–1175 K results in a surface
reconstructed phase on the clean surface, which is
the lowest energy configuration. Previous LEED
studies by Graupner et al. [14] characterized the
diffraction spots of the reconstructed surface as an
incommensurate superstructure, which is com-
ꢁ
mensurate in the [1 1 0] direction and incommen-
surate along the [0 0 1] direction of the FeAl (1 1 0)
surface. In addition, their AES data indicate an
enhanced concentration of aluminum at the sur-
face, compared to the bulkcomposition. The sur-
face concentration of Al was calculated to be
0.67 ꢁ 0.06, corresponding to FeAl2 surface stoi-
chiometry assuming the segregated Al is confined
to the topmost layer. They adopted a structural
model of an ordered quasi-hexagonal overlayer
which has a unit mesh containing two Al and one
Fe atoms. This proposed surface structural model
from LEED data has been corroborated by a more
recent surface X-ray diffraction/reflectivity (XRD)
study [16].
Fig. 1. (a) STM image (15 Â 15 nm2, Vs ¼ 30:2 mV, It ¼ 5:4
nA), acquired at room temperature, of clean FeAl(1 1 0) after
annealing to 1125 K. The atomically resolved image shows Fe
atoms surrounded by six Al atoms forming a FeAl2 overlayer,
as depicted in the hard ball model. (b) Darkballs represent Fe
atoms with the unit cell indicated by a dashed line.
tween ꢁ30 mV and ꢁ1 V. Variation of tunneling
parameters (e.g. bias) did not appreciably change
the observed structure. The STM results for this
reconstructed FeAl(1 1 0) phase are consistent with
the structural model determined from previous
LEED and XRD studies [14,16], shown in Fig. 1b
(for comparison, a bulkterminated surface ball
Correspondingly, the STM image, shown in
Fig. 1a, reveals the atomic structure of this clean,
incommensurate phase. STM images of the clean
surface were acquired with tunneling currents be-
tween ꢁ0.5 and ꢁ5.5 nA, and bias voltages be-

226
O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
model is shown in the bottom of the model). As
seen in the model, a quasi-hexagonal arrangement
of atoms is atomically resolved in the STM image.
The hexagon mesh drawn with a solid line in the
ball model implies that the imaged atoms are Fe,
surrounded by six Al atoms. The unit mesh, shown
with a dashed line adopting the 1:2 ratio of Fe:Al,
confirms the FeAl2 stoichiometry. The measured
STM image that reveals the overall structure
and morphology of this high-temperature phase of
ultra-thin alumina formed on reconstructed FeAl-
(1 1 0), upon exposure of 1000 L of O2 (i.e. satu-
ration coverage) at 1125 K. The large-area STM
image, acquired after the exposed sample was re-
cooled to room temperature, shows a periodic and
nearly hexagonal superstructure on the ordered
oxide. The spot-to-spot distance of the super-
ꢀ
distance of 8.22 A between two successive Fe
ꢁ
atoms in the [1 1 0] direction of substrate confirms
ꢀ
structure is approximately 19 A. STM images re-
the commensurability; however, the periodicity
ꢀ
veal that with saturation coverages (as in Fig. 2a),
the thin-film alumina grows homogeneously across
the surface in a relatively flat morphology with a
lateral disorder primarily perpendicular to the
[0 0 1] direction. LEED confirms this disorder, with
along the [0 0 1] direction (4.2 ꢁ 0.05 A) lacks
commensurability with the underlying substrate
ꢀ
(2.9 A). Upon oxygen exposure at high tempera-
tures, this reconstructed surface becomes the
template for nucleation and growth of well-
ordered ultra-thin alumina films.
ꢁ
streaking along the [1 1 0] direction. The unit cell
2
ꢀ
of the oxide film (18.6 Â 19.4 A ), shown by a solid
line, is much larger than that of the substrate
2
ꢀ
3.2. Atomic and electronic properties of alumina
thin films on FeAl(110)
(2.9 Â 4.1 A ), and is rotated by 30° with respect to
ꢁ
the [1 1 0] direction of substrate. Due to the two-
fold symmetry of the substrate, two rotational
domains are expected and are indeed observed for
the aluminum oxide on NiAl(1 1 0) [9,18]. How-
ever, as in the case of STM studies of Al2O3/
Ni3Al(1 1 1) [5], our STM data reveals only one
domain in the present case. The measured corru-
Ultimately, the composition and structure of
the oxide that forms on FeAl is dictated by the
overall energetics. Since the enthalpy of formation
of Al2O3 (DHf ¼ À1676 kJ/mol) is lower than that
for Fe based oxides (Fe2O3 ¼ )826 kJ/mol,
FeO ¼ )278 kJ/mol, Fe3O4 ¼ )1118.4 kJ/mol)
[17], it is thermodynamically more favorable that
aluminum be preferentially oxidized on the FeAl-
(1 1 0) surface forming an ultra-thin Al2O3 film.
From previous experimental studies [10], it is
known that the oxidation of FeAl(1 1 0) can result
in three different structural phases, as determined
by LEED, of alumina films on the surface, de-
pending on the substrate temperature during oxy-
gen exposure. Although an amorphous Al2O3
structure is formed at temperatures up to 775 K
primarily due to kinetic limitations, a ‘‘streaked
oxide’’ phase with short-range order is observed
with a saturation exposure of oxygen within the
775–975 K temperature range. Tentatively, the
structure of this latter oxide phase is related to
the missing-row reconstructed FeAl(1 1 0) surface
that occurs within this temperature range.
ꢀ
gation height is 0.55 ꢁ 0.05 A; this large value is
not solely due to variations in the structural
morphology, but also includes strong lateral vari-
ations in the local electronic structure.
The STM image in Fig. 2b reveals large ho-
mogeneous films separated by a step edge. In the
ꢀ
present case, the measured step height is ꢀ8 A (see
linescan at the bottom of figure). It is believed that
this height corresponds to a large step in the un-
derlying metal substrate; STM images of the clean
ꢀ
FeAl(1 1 0) surface reveals step heights of 2–8 A.
Only a few defects in the oxide film are observed.
Typically they are imaged as ‘‘holes’’ as identified
by the arrow in Fig. 2b. If it is assumed that this
hole indicates a lackof oxide growth, then the
measured depth should be indicative of the thin-
film oxide thickness. As seen in the linescan, the
depth of the hole, employing the given tunneling
ꢀ
As opposed to these lower temperature phases,
a well-ordered aluminum oxide phase can be pro-
duced at or above 1125 K and is the primary focus
of this STM study. Fig. 2a shows a large-scale
parameters, is ꢀ4 A; previous LEED studies of
ꢀ
this system determine the thickness to be ꢀ6 A
[10]. Although quite similar, it is acknowledged
that the measured depth of the hole with STM

O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
227
Fig. 2. (a) STM image (70 Â 70 nm2, Vs ¼ À1:25 V, It ¼ À0:7 nA) of oxidized FeAl(1 1 0) after exposure of 1000 L of O2 (saturation). A
2
ꢀ
unit mesh (18.6 Â 19.6 A ) is displayed with a solid line. Propensity of ordering along the close packed direction of the substrate is
observed. (b) STM image (69 Â 100 nm2, Vs ¼ 1:25 V, It ¼ 0:17 nA) shows a step edge and void hole formation, marked by an arrow.
ꢀ
ꢀ
ꢀ
Line scan, below the image, indicates a step height (ꢀ8 A) and nanoscale hole formation (4 A depth width and 80 A across). (c) Low
coverage of O2 exposure (125 L) reveals local order with a larger unit cell than the saturated surface (72 Â 72 nm2, Vs ¼ À0:4 V,
It ¼ À3:69 nA). (d) STM image (50 Â 50 nm2, Vs ¼ 3:02 V, It ¼ 1:04 nA), total exposure of 1000 L, reveals an electronic structure effect,
showing a line pattern with positive bias voltage.
may not accurately represent the true oxide
thickness due to electronic effects. For example,
from previous studies on the NiAl(1 1 0) substrate,
it is known that apparent height of oxide step
edges and hole defects depends critically on tun-
neling parameters [18,19]. However, in these
studies the apparent height attenuated when tun-
neling bias values within the ‘‘gap’’ were employed.
In contrast, we do not observe this effect.
range order. Fig. 2c shows an image of the surface
produced with an exposure of only 125 L. The
spot-to-spot distance is, on average, larger than on
the saturated surface. Because of the somewhat
disordered surface, STM does not yield direct in-
formation about the oxide structure of the sub-
saturated alumina film. However, LEED reveals
that, in addition to the oxide superstructure, in-
commensurate diffraction spots corresponding to
the clean reconstructed FeAl(1 1 0) surface are
concurrently observed. This implies that while
Subsaturation coverages of oxygen at 1125 K
results in an oxide superstructure that lacks long-

228
O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
short-range order of the oxide film is maintained,
either patches of clean FeAl(1 1 0) or a non-equi-
librium structure of oxide exists. Higher exposure
of oxygen (corresponding to Fig. 2a) results in the
disappearance of the incommensurate hexagonal
diffraction spots, indicating the interfacial layer
between the oxide thin film and substrate does not
maintain its reconstructed structure. XPS data [10]
imply that the presence of a smaller Al 2p peak, in
addition to the major one, is indicative of Al atoms
between the oxide and the substrate. As corrobo-
rated by STM data, the disappearance of intense
superstructure diffraction spots in the LEED pat-
tern demonstrates that the ultra-thin film of alu-
minum oxide extends homogeneously across the
surface at saturation coverage.
In order to investigate the influence of the oxide
electronic structure on STM images, data were
acquired under reverse bias conditions. Fig. 2d
shows an image of the saturated oxide film ac-
quired with +3 V bias voltage (opposite tunneling
polarity compared to the one used in Fig. 2a and
c). As seen, tunneling with positive sample bias
voltage reveals a different electronic configuration
showing a line pattern structure of the oxide. The
atomic structure of the oxide surface is only re-
vealed upon tunneling into (out of) conductance
(valence) states. However, this assumes that the
electronic properties of the thin film have direct
analogs to the bulkoxide. A recent theoretical
study by Jennison et al. [21] of supported alumina
thin films determined a non-zero local density of
states (LDOS) of aluminum and oxygen within the
expected bulkband gap. This result is a conse-
quence of the film thickness and the expected
quantum mechanical mixing of the non-bulk-like
chemical environment within the oxide layers.
Similarly, it is believed that due to the unique
atomic and chemical structure of ultra-thin Al2O3
films grown on FeAl(1 1 0), STM can indeed be
used to elucidate the surface atomic structure of
the Al–O thin film. However, the results are best
interpreted with information about the bulkstruc-
ture and theoretical predictions focused on the
ultra-thin films.
The structure of bulk c-Al2O3 has been a matter
of discussion for nearly 40 years. The debate
mainly stems from questions regarding the posi-
tion of vacancies required for charge balance in the
defect structure. Most experimental studies [22,23]
suggest that a small number of cations occupy
tetrahedral sites, whereas the normal preference
for Al ions is considered to be in octahedral sites.
However, a recent transmission electron diffrac-
tion study [24] has revealed a reverse assessment,
indicating all tetrahedral sites are occupied by ca-
tions and vacancies are found at the octahedral
sites. Again, these controversial findings corre-
spond to the bulk structure.
ꢀ
lines, with a periodicity of nearly 30 A, are rotated
approximately 40° with respect to the [0 0 1] di-
rection. The apparent structural changes seen with
opposite tunneling polarity are produced by elec-
tronic band structure effects of the oxide film.
Since the unoccupied bands may have a different
density of states and symmetry, changing the bias
voltage may result in tunneling into different
states, whose symmetries differ from those of the
occupied bands. We assert that the difference in
the local electronic structure, i.e. density of states
and symmetry of the un/occupied bands, results in
a substantial difference in the apparent morpho-
logy with different polarity. The structure of
the oxide, shown in Fig. 2d with positive polar-
ity, is very similar to the STM/AFM results of
NiAl(1 1 0) [20].
Regarding the ultra-thin alumina films, al-
though experimental studies have not adequately
addressed the question of the coordination of Al
ions, ab initio calculations have revealed much as
to the atomic structure of the thin film Al2O3 and
corresponding electronic structure [21]. Specifi-
ꢀ
cally, calculations performed on simulated 5 A
Al2O3 films, supported on Al(1 1 1), Mo(1 1 0) and
Ru(0 0 0 1) substrates, have demonstrated that the
tetrahedrally coordinated Al3þ ions are located
nearly in the same plane as the oxygen ions for
electroneutrality. These calculations concluded
that an overall structure composed of two oxide
layers, wherein the Al ions are tetrahedrally co-
In earlier studies on Al2O3/NiAl(1 1 0) [9,19], it
was argued that due to the wide band gap of the
bulkoxide (8 eV), STM images obtained by tun-
neling in this gap correspond to the contributions
from both the oxide and, more importantly, from
the interfacial electronic structure. That is, the

O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
229
ordinated to O ions, describe the lowest energy
configuration. Even though the occupied density
of states of the bulkoxide begins at about 4 eV
below the Fermi energy, they calculated that a
non-zero LDOS from aluminum and oxygen for
the two oxide layers lie within the bulkband gap
of the oxide. Due to this finite LDOS near EF, at
small tunnel voltages, we tentatively propose that
STM directly reveals the atomic structure of the
thin film.
Considering a larger unit cell [25], Jennison
et al. recently showed a propensity of an even mix
of tetrahedral and octahedral site Al cations, con-
stituting the lowest energy stable film. The result-
ing structure is composed of an alternating zigzag
structure of equal number of tetrahedral and oc-
tahedral Al ions with an increase in the interlayer
Al–Al spacing. They also described a structure
consisting of an alternating stripe and zigzag
structure, but due to a decrease in the Al–Al sep-
aration resulting in higher energy, this structure is
not favored. In Fig. 3a, a well-ordered structure
consisting of a composite zigzag and stripe struc-
ture, alternating every two atoms along the [0 0 1]
direction of the substrate, is clearly seen. Based on
the similarities with the theoretical model [25], we
propose that the atomic topography observed here
with STM is due to Al ions that are assumed to be
coplanar with oxygen atoms. The ball model at the
top of Fig. 3b depicts this proposed structure. For
comparison, the bottom part of Fig. 3b shows the
pure octahedral arrangement of Al ions, consistent
with a bulkAl O3 arrangement (see unit cell at
2
Fig. 3. (a) STM constant current image (20 Â 20 nm2,
Vs ¼ À1:9 V, It ¼ À2:54 nA) depicts Al ions in a zigzag–stripe
structural arrangement along the [0 0 1] direction of FeAl(1 1 0).
(b) Zigzag–stripe structure is shown on the upper part of the
ball model and a pure octahedral structure is included (lower)
to compare the unit cells drawn with solid line.
bottom). Relative to the unit cell of bulkAl O3,
2
the resulting surface structure adopted from these
STM measurements is consistent with a larger unit
cell. Although not definitive, the unit cell shown at
the top of Fig. 3b reveals a (4 Â 2) structure with
respect to the bulk-oxide unit cell and is consistent,
both rotationally and spatially, with the quasi-
hexagonal superstructure observed in the STM
image of Fig. 2a. Assuming this model is valid, the
occupancy of cations in tetrahedral sites is equal to
that of octahedral sites, and questions the as-
sumption of a surface Al2O3 stoichiometry. How-
ever, some questions still remain concerning this
proposed structure. For example, atomically re-
solved STM images of the superstructure are
needed to confirm the model. Indeed, assuming an
attenuated Al occupancy in the surface layer, in
order to maintain proper stoichiometry, periodic
oxygen vacancies may be present, perhaps oriented
along the [0 0 1] direction of substrate. Because we
were unable to obtain atomically resolved images
to determine quantitative vacancy ratios of the
outmost layer, this model is only tentative and
calls upon additional theoretical input.

230
O. Kizilkaya et al. / Surface Science 529 (2003) 223–230
4. Conclusion
References
[1] V.E. Henrich, P.A. Cox, The Surface Science of
Metal Oxides, Cambridge University Press, Cambridge,
1994.
The atomic and electronic structure of ultra-thin
aluminum oxide films formed on the FeAl(1 1 0)
surface has been studied with STM. Upon anneal-
ing above 1125 K, the clean surface reconstructs
into an incommensurate structure consistent with a
FeAl2 stoichiometry confined to the outermost
layer. Upon saturating this surface with oxygen, an
ordered ultra-thin aluminum oxide film is formed,
in direct equivalence with previous studies on
low-index surfaces of NiAl. At this coverage, the
incommensurate diffraction spots of the clean re-
constructed surface have disappeared, implying
that the substrate surface does not retain its re-
constructed phase. Furthermore, the Al ions im-
aged by STM were assumed to be nearly in the
same plane as the oxygen anions, necessary for
electroneutrality, which is in agreement with theo-
retical calculations performed by Jennison et al. on
Al2O3(0 0 1)/Al(1 1 1). Confirming the previous
calculated electronic structure of Al2O3, we as-
serted that due to limited thickness of the thin oxide
film and rearrangement of Al ions, a LDOS of
aluminum and oxygen resided in the region of the
bulkband gap of the oxide. A combination zigzag
and stripe structure of equal octahedrally and tet-
rahedrally coordinated Al ions was observed.
[2] V.E. Henrich, Rep. Prog. Phys. 48 (1985) 1481.
[3] R. Franchy, Surf. Sci. Rep. 38 (2000) 195.
[4] U. Bardi, A. Atrei, G. Rovida, Surf. Sci. 268 (1998)
87.
[5] S.G. Addepalli, B. Ekstrom, N.P. Magtoto, J.-S. Lin, J.A.
Kelber, Surf. Sci. 442 (1999) 385.
[6] R. Franchy, M. Wutting, H. Ibach, Surf. Sci. 189/190
(1987) 438.
[7] R. Franchy, J. Masuch, P. Gassmann, Appl. Surf. Sci. 319
(1994) 95.
[8] R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wutting,
W. Hoffmann, R. Franchy, H. Ibach, Surf. Sci. 259 (1991)
235.
€
[9] J. Libuda, F. Winkelmann, M. Baumer, H.-J. Freund, Th.
Bertrams, H. Neddermeyer, K. Muller, Surf. Sci. 318
€
(1994) 61.
[10] H. Graupner, L. Hammer, K. Heinz, D.M. Zehner, Surf.
Sci. 380 (1997) 335.
[11] F. Besenbacher, Rep. Prog. Phys. 59 (1996) 1737.
€
[12] Landolt-Bornstein, Crystallograhic and Thermodynamic
Data of Binary Alloys, New Series, iV/5a, Phase Equilibria,
Springer, Berlin, 1991.
€
[13] W. Koster, T. Godecke, Z. Metallkd. 71 (1980) 765.
[14] H. Graupner, L. Hammer, K. Muller, D.M. Zehner, Surf.
€
€
Sci. 322 (1995) 103.
[15] L. Hammer, H. Graupner, V. Blum, K. Heinz, G.W.
Ownby, D.M. Zehner, Surf. Sci. 412 (1998) 69.
[16] A.P. Baddorf, S.S. Chandavakar, Phys. B. Cond. Mat. 221
(1996) 141.
[17] D.R. Lide, Handbookof Chemistry and Physics, CRC
Press, Boca Raton, 1994.
Acknowledgements
[18] K.H. Hansen, T. Worren, E. Lægsgaard, F. Besenbacher,
I. Stensgaard, Surf. Sci. 475 (2001) 96.
[19] Th. Bertrams, A. Brodde, H. Neddermeyer, J. Vac. Sci.
Technol. B 12 (1994) 2122.
We would like to acknowledge the staff of
CAMD for their support in the photoemission
studies and G.W. Ownby and J.F. Wendelken of
the Solid State Division of ORNL for the support
in the crystal preparation and STM studies, re-
spectively. Research at ORNL was sponsored by
the Department of Energy managed by UT-Bat-
telle, LLC under contract DE-AC05 00OR22725.
The research was also supported by U.S. DOE
contract no. DE-FG02-98ER45712.
[20] C.L. Pang, H. Raza, S.A. Haycock, G. Thorton, Phys.
Rev. B 65 (2002) 201401.
[21] D.R. Jennison, C. Verdozzi, P.A. Schultz, M.P. Sears,
Phys. Rev. B 59 (1999) 15605.
[22] R. Dupree, M.H. Levis, M.E. Smith, Philos. Mag. A 53
(1986) L17.
[23] V. Jayaram, C.G. Levi, Acta Metall. 37 (1989) 569.
[24] B. Ealet, M.H. Elyakhloufi, E. Gillet, M. Ricci, Thin Solid
Films 250 (1994) 92.
[25] D.R. Jennison, A. Bogicevic, Surf. Sci. 464 (2000) 108.