The growth of protective ultra-thin alumina layers on γ-TiAl(1 1 1) intermetallic single-crystal surfaces

Article abstract of DOI:10.1016/j.susc.2005.09.011

An XPS and AES study of the early stages of oxidation of γ-TiAl(1 1 1) surfaces at 650 °C under 1.0 × 10^-7-1.0 × 10 ^-6 mbar O₂ is reported. The data evidence a first regime of oxidation characterized by the growth of a pure alumina layer followed by a second regime of simultaneous oxidation of both alloying elements. In the first regime, continuous alumina layers from ～0.4 to ～1.5 nm thick have been observed by angle-resolved XPS. The composition of the metallic phase underneath the growing oxide is modified by a depletion of Al and the injection of Al vacancies in the metal during the growth of the transient alumina formed at 650 °C. The onset of Ti oxidation was repeatedly observed for a critical concentration in the modified region of the alloy underneath the alumina layer: Ti_75±2Al_25±2 (Ti₅₀Al _17±2V(Al)_33±2), showing that decreasing the number of Ti-Al bonds in the modified intermetallic region increases the activity of Ti up to a critical point where its oxidation at the oxide/metal interface becomes competitive with that of Al. The growth of Ti³⁺ and Ti⁴⁺ oxide particles observed above the alumina layer by angle-resolved XPS indicates the transport of titanium cations trough the alumina layer and their subsequent reaction with oxygen at the outer gas/oxide interface. Improving structural ordering in the intermetallic phase slows down the growth kinetics of the alumina layer and the related Al-depletion of the substrate, and increases the resistance of the alloy to the subsequent oxidation of Ti. This is assigned to two combined effects: a slower diffusion of Al in the better ordered metallic phase and the growth of less defective alumina layers allowing to slow down the ionic transport through the oxide. Highly stable and corrosion resistant alloy surfaces covered by a 0.4 nm thick alumina layer have been obtained by slowly oxidizing the alloy at lower partial pressure (<5.0 × 10^-10 mbar O₂).

Full text of DOI:10.1016/j.susc.2005.09.011

Surface Science 596 (2005) 61–73
www.elsevier.com/locate/susc
The growth of protective ultra-thin alumina layers
on c-TiAl(111) intermetallic single-crystal surfaces
V. Maurice a,*, G. Despert a,b, S. Zanna , P. Josso , M.-P. Bacos , P. Marcus
a
b
b
a
a
Laboratoire de Physico-Chimie des Surfaces, CNRS (UMR 7045), Universit e´ Pierre et Marie Curie, Ecole Nationale
Sup e´ rieure de Chimie de Paris, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France
ONERA, D e´ partement Mat e´ riaux M e´ talliques et Proc e´ d e´ s, BP 72, 29 avenue de la Division Leclerc 92322 Ch aˆ tillonCedex, Fra nc e
b
Received 6 April 2005; accepted for publication 2 September 2005
Available online 29 September 2005
Abstract
ꢀ
7
An XPS and AES study of the early stages of oxidation of c-TiAl(111) surfaces at 650 ꢀC under 1.0 · 10
–
ꢀ
6
1
.0 · 10 mbar O is reported. The data evidence a first regime of oxidation characterized by the growth of a pure
2
alumina layer followed by a second regime of simultaneous oxidation of both alloying elements. In the first regime,
continuous alumina layers from ꢁ0.4 to ꢁ1.5 nm thick have been observed by angle-resolved XPS. The composition
of the metallic phase underneath the growing oxide is modified by a depletion of Al and the injection of Al vacancies
in the metal during the growth of the transient alumina formed at 650 ꢀC. The onset of Ti oxidation was repeatedly
observed for a critical concentration in the modified region of the alloy underneath the alumina layer: Ti75± 2Al25± 2
(
Ti Al17± 2V(Al)33± 2), showing that decreasing the number of Ti–Al bonds in the modified intermetallic region
50
increases the activity of Ti up to a critical point where its oxidation at the oxide/metal interface becomes competitive
3
+
4+
with that of Al. The growth of Ti and Ti oxide particles observed above the alumina layer by angle-resolved XPS
indicates the transport of titanium cations trough the alumina layer and their subsequent reaction with oxygen at the
outer gas/oxide interface. Improving structural ordering in the intermetallic phase slows down the growth kinetics of
the alumina layer and the related Al-depletion of the substrate, and increases the resistance of the alloy to the subse-
quent oxidation of Ti. This is assigned to two combined effects: a slower diffusion of Al in the better ordered metallic
phase and the growth of less defective alumina layers allowing to slow down the ionic transport through the oxide.
Highly stable and corrosion resistant alloy surfaces covered by a 0.4 nm thick alumina layer have been obtained by
ꢀ
10
slowly oxidizing the alloy at lower partial pressure (<5.0 · 10
2005 Elsevier B.V. All rights reserved.
2
mbar O ).
ꢁ
Keywords: X-ray photoelectron spectroscopy; Intermetallics; Aluminium oxide; Titanium oxide; Corrosion; Protection; Single crystal
surfaces; Surface defects
*
Corresponding author. Tel.: +33 1 44276736; fax: +33 1 46340753.
E-mail addresses: vincent-maurice@enscp.fr (V. Maurice), philippe-marcus@enscp.fr (P. Marcus).
0
039-6028/$ - see front matter ꢁ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.09.011

62
V. Maurice et al. / Surface Science 596 (2005) 61–73
1
. Introduction
The use of lightweight TiAl intermetallic alloys
2. Experimental
2.1. Materials
in aeronautic applications is partly limited by their
low resistance to high temperature corrosion (their
mechanical properties is also a key issue). This is
due to the growth of mixed oxide layers forming
by competitive oxidation of the Ti and Al alloying
elements [1–5], which prevents the formation of a
continuous and dense a-alumina that would
provide a more effective oxidation barrier in high
temperature applications. In order to better under-
stand the oxidation properties of these alloys, sev-
eral studies of the initial stages of oxidation of the
c-TiAl intermetallic phase have been carried out
on polycrystalline surfaces [6–12]. At room tem-
perature, oxidation was reported to be preceded
by the dissociative adsorption of oxygen, which
was proposed to occur preferentially on the Ti
atoms [7]. Titanium oxidation was observed after-
wards and preceded aluminium oxidation [7,12].
The oxide was reported to be Ti(II) species [10].
For 100 ꢀC < T < 600 ꢀC, the simultaneous growth
of titanium and aluminium oxides was reported
The c-TiAl single-crystal used in this study was
supplied by Pr. Masaharu Yamaguchi (Depart-
ment of Materials Science and Engineering, Kyoto
University, Japan). Its bulk composition was
Ti Al (at%) corresponding to the c phase. The
4
9
51
chemically ordered structure of the c-TiAl inter-
metallic is based on a FCC lattice of parameter
a = 0.402 nm. The stacking sequence along the
[001] axis alternates pure Al or Ti atomic planes,
thus creating a very small asymmetry along the c
axis (c = 0.406 nm instead of 0.402 nm). The crys-
tal was oriented along the [111] direction within
± 1ꢀ as controlled by X-ray La u¨ e back diffraction.
Each (111)-oriented plane is stoichiometric (Ti -
5
0
Al ) and alternates close-packed rows of Ti and
5
0
Al. The surface was prepared by mechanical pol-
ishing with SiC paper down to a final grade of
4000 (4 lm) for the XPS study and further with
diamond paste down to 0.25 lm for the AES
study. Afterwards, the sample was cleaned ultra-
sonically with acetone and ethanol and then
thoroughly rinsed and dried before mounting on
the sample holders for in situ treatments and
analysis.
[
8,10,11] with the Ti(II), Ti(III), Ti(IV) and Al(III)
oxidation states [10,11]. The amounts of Ti(II) and
Ti(IV) decreased and increased with increasing
temperature, respectively [10]. At 850 ꢀC, the oxi-
dation of aluminium was reported to precede the
oxidation of titanium. Only Ti(IV) and Al(III)
were observed in the following stage of simulta-
neous oxidation of the two alloying elements
2.2. Insitu surface preparation
Most oxidation experiments and analyses were
performed in a VG ESCALAB Mark II spectrom-
eter (VG Scientific). After introduction in the XPS
preparation chamber, the surface was further
[
9].
In this paper, we report a study of the initial
stages of growth of the oxide layers at 650 ꢀC
on (111)-oriented c-TiAl single-crystal surfaces.
X-ray photoelectron spectroscopy (XPS) and
Auger electron spectroscopy (AES) were used.
The sample was exposed to controlled doses of gas-
+
treated by five cycles of Ar ion bombardment
of 15 min and subsequent annealing (20 min)
under the residual pressure of the UHV chamber
ꢀ
10
+
(ꢁ5 · 10
mbar). Bombardment with Ar ions
ꢀ
7
ꢀ6
eous oxygen at low pressure (P(O ) = 1.0 · 10
–
(P(Ar) = 1 · 10 mbar) accelerated at 3 keV and
giving a sample current of 100 lA was used to elim-
inate the presence of light contaminants (O, C and
N). Annealing at 800 ꢀC was used to eliminate the
traces of Ar and C detected after ion bombard-
ment and to re-crystallize the surface. During the
last cycle, a milder ion bombardment treatment
2
ꢀ
6
1
.0 · 10 mbar). The composition and thickness
of the oxide layers as well as the modifications of
the alloy substrate surface underneath the growing
oxide scale were analysed by angle-resolved XPS.
The results emphasize the effect of surface ordering
on the mechanism of selective oxidation of the
alloy to form a protective ultra-thin layer of
alumina.
ꢀ
6
(P(Ar) = 1 · 10 mbar, 3 keV, 3 lA, t = 15 min)
was performed and the surface was flash-annealed

V. Maurice et al. / Surface Science 596 (2005) 61–73
63
between 900 and 1000 ꢀC. This last treatment at
3. Results
higher temperature was necessary in order to
eliminate the residual traces of carbon (assigned
to titanium carbide [13]) always observed after
sputtering and/or annealing at 800 ꢀC. An uptake
of oxygen (mostly due to surface segregation from
the bulk) was always observed after annealing in
UHV. The analysis of the trace amounts of oxygen
observed after preparation revealed the presence
of two surface oxygen species adsorbed on the sur-
face and of oxygen dissolved in the sub-surface
3.1. Selective oxidationof alumi ni um
Fig. 1 shows a selection of spectra characteristic
of the evolution of the Al 2p, O 1s and Ti 2p core
level regions during the first 71 min of oxidation of
the c-TiAl(111) sample (pre-annealed for a total
period of 110 min). The Al 2p spectrum consists
of two peaks as revealed by the curve fitting shown
in Fig. 2. The first peak, at a binding energy (BE)
of 72.2 eV, is also observed before oxidation and
[
13,14].
The oxidation experiments were performed by
background filling the UHV chamber to P(O ) =
2
ꢀ
7
1
· 10 mbar after setting the sample temperature
at 650 ꢀC. At the end of the exposure, oxygen inlet
and sample heating were simultaneously stopped.
The surface was cooled down to room temperature
for analysis. The oxidation kinetics was measured
by XPS by performing cumulative oxidation steps
from 20 s up to a total time of 210 min. Their
recording was ended when the metallic compo-
nents corresponding to the substrate alloy could
no longer be observed in the Ti 2p and Al 2p spec-
tra due to the attenuation of the photoelectrons by
the oxide scale. Different series of measurements
were performed by varying the total pre-annealing
time of the substrate before oxidation (110, 200,
3
50 and 550 min).
2
.3. XPS analysis
The XPS analyses were performed using the
Al Ka X-ray source (1486.6 eV). Survey spectra
were recorded with a pass energy of 100 eV and
high resolution spectra of the Al 2p, Ti 2p and O
1
s core level regions were recorded with a pass en-
ergy of 20 eV. Angle-resolved measurements were
performed by varying the take-off angle from 90ꢀ
to 40ꢀ (angle of the analysed photoelectrons with
respect to the surface plane). Peak fitting allowing
to decompose the XPS spectra in different compo-
nents assigned to different surface species was per-
formed using the ECLIPSE software and a Shirley
background. The procedure was established on
reference spectra obtained on well-defined samples
Fig. 1. XPS spectra of the Al 2p, O 1s and Ti 2p core levels for
ꢀ7
2
the c-TiAl(111) sample exposed to P(O ) = 1.0 · 10 mbar at
650 ꢀC for 0, 2, 16 and 71 min.
[
13].

64
V. Maurice et al. / Surface Science 596 (2005) 61–73
7
1
1 min of exposure (following pre-annealing for
10 min), the two peaks at 454.6 and 460.6 eV,
spin-orbit
doublet for the metallic state Ti , are strongly
corresponding to the 2p3/2–2p
1/2
0
2
p
attenuated by the oxide layer. Instead, two sets
of 2 peaks at 457.5 eV/463.2 eV and 459.6 eV/
465.2 eV are observed, that are characteristic of
the 2p3/2–2p1/2 spin-orbit doublet for the Ti(III)
and Ti(IV) oxidation states, respectively [8] (the
curve fitting is described further on). At 71 min,
the O 1s peak is centred at a BE of 531.4 eV char-
2
ꢀ
acteristic of O anions in titanium oxides [8], indi-
cating a predominant growth of titanium oxides at
this point. These spectra evidence a transient stage
of selective oxidation of the alloy to form a pure
alumina layer prior to the competitive oxidation
of titanium, in agreement with the results obtained
for polycrystalline samples [9,13,14].
The thickness, d, of the alumina layer can be
3
þ
calculated from the intensity ratio of the Al
2p
0
and Al signals with:
2p
Al2O3
d ¼ kAl2p  sin h
Fig. 2. Curve fitting of the XPS Al 2p and Ti 2p spectra for the
"
#
ꢀ7
3þ
TiAl
Al2p
c-TiAl(111) sample exposed to P(O
50 ꢀC for 2 min.
2
) = 1.0 · 10 mbar at
TiAl
Al
Al2O3
k
IðAl Þ
D
2
p
6
 ln 1 þ
Â
Â
;
0
Al2O3
k
Al2p
IðAl Þ DAl
2p
Al2O3
TiAl
where kAl2p (k ) is the attenuation length of the
Al2p
corresponds to the metallic signal of the substrate
Al2p photoelectrons in the oxide layer (substrate),
0
(
Al ). The second peak located at 75.5 eV is grow-
h the take-off angle of the photoelectrons with re-
2p
Al2O3
ing with increasing exposure to oxygen and corre-
spect to the surface plane, and DAl
the atomic
3
þ
sponds to oxidised aluminium (Al ), in good
concentration of aluminium in the oxide layer
2p
TiAl
agreement with previous observations on similar
substrates [10,11,15–17]. The O 1s spectrum shows
the growth of one peak centered at a BE of
[13]. DAl is the concentration of aluminium in
the modified alloy underneath the oxide (see be-
low). In this expression, it is assumed that the
oxide layer is continuous and has an homogeneous
thickness. Fig. 3 shows the equivalent thickness of
the alumina layer obtained after 2 min of exposure
to oxygen (for the sample pre-annealed for
110 min) calculated from the XPS spectra recorded
at various take-off angles h. The measured value is
constant, which confirms the model of a continu-
ous oxide layer.
Fig. 4 shows the evolution of the equivalent
thickness of the alumina layer with increasing
exposure to oxygen after pre-annealing time peri-
ods of 110, 200, 350 and 550 min. The plot is lim-
ited to the stage of selective oxidation during
5
32.1 eV assigned to the anions of the growing
2ꢀ
oxide layer (O ). At 2 min of oxidation, the Ti
1s
2
p spectrum remains unchanged, showing no
higher BE component that could correspond to
Ti(II), Ti(III) or Ti(IV) oxidation states. It consists
of the two peaks at 454.4 and 460.6 eV (see Fig. 2
for curve fitting), corresponding to the 2p3/2–2p1/2
spin-orbit doublet (intensity ratio of 2:1) for
0
2
the metallic signal of the substrate (Ti
and
p3=2
0
Ti p1=2, respectively). After longer exposures
2
(
>3 min, see below), the Ti 2p spectra show the
growth of the additional higher BE components
characteristic of the oxidation of titanium. At

V. Maurice et al. / Surface Science 596 (2005) 61–73
65
can be calculated from the intensity ratio of the
0
0
Ti to Al signals (assuming an homogeneous
2p
2p
depth distribution in the metallic phase) with:
0
TiAl
Ti2p
TiAl
Al
TiAl
IðTi Þ T Ti2p rTi2p
k
k
1 ꢀ D
2p
¼ T
 r
Â
Â
0
TiAl
Al2p
IðAl Þ
Al2p
Al2p
D
Al
2
p
!
2
3
ꢀ
d
exp
6
6
6
Al
2
O
3
7
7
kTi
Âsin h
2
p
7
Â
! ;
6
4
7
5
ꢀ
d
exp
Al
O
2
2
3
Fig. 3. Equivalent thickness of the alumina layer and modified
concentration of the underlying metallic phase as a function of
k
Âsin h
Al
p
2
the take-off angle h for the c-TiAl(111) exposed to P(O ) =
ꢀ7
where TTi2p (TAl2p ) and rTi2p (rAl2p ) are the transmis-
1.0 · 10 mbar at 650 ꢀC for 2 min.
sion factor and photoionisation cross section of
TiAl
Al
the Ti2p (Al ) electrons, respectively [13]. D
is
2p
the concentration of aluminium in the modified
alloy region underneath the oxide. Fig. 3 shows
the Al concentration calculated from measure-
ments at various take-off angles of the photoelec-
trons, after 2 min of exposure to oxygen (i.e.,
below the 0.8 nm alumina film). The measured val-
ues are lower than 50 at%, showing that the metal-
lic phase underneath the oxide layer is depleted in
Al, which is consistent with the selective oxidation
of the alloy. Since the escape depth of the photo-
electrons (ꢂ3 · k · sinh) decreases with decreasing
take-off angle, the decrease of the average value of
the Al concentration with decreasing take-off angle
indicates that the Al depletion is limited to the
nearest substrate layers below the metal/oxide
interface and that there is a concentration gradient.
Fig. 5 shows the evolution of the composition
of the modified alloy region underneath the grow-
ing oxide layer. The relative concentration in Al
Fig. 4. Kinetics of growth of the alumina layer on c-TiAl(111)
ꢀ7
exposed to P(O
measurements performed after pre-annealing for (a) 110 min,
b) 200 min, (c) 350 min and (d) 550 min are shown. The inset
2
) = 1.0 · 10 mbar at 650 ꢀC. Four series of
(
shows the first 9 min of oxidation.
TiAl
TiAl
Ti
which no titanium oxide particles are detected in
the oxide layer. In all series, the thickness of the
alumina layer increases rapidly after the first expo-
sure to oxygen to reach 0.7–0.8 nm after less than
(DAl þ D
¼ 1) is plotted. A decrease of the
Al concentration is observed, from 50 at%, mea-
sured prior to oxidation, to 25 ± 2 at% measured
at the end of the stage of selective oxidation (i.e.,
before the oxidation of titanium). The fact that
this decrease is observed shows that the growth
of the oxide is too fast to allow the metallic phase
to re-homogenize, Al being consumed too rapidly
by the growth of the oxide. This indicates that
the growth of the alumina layer is limited by the
diffusion of Al to the metal/oxide interface, and
not by its diffusion in the oxide layer (the transient
alumina formed in this temperature regime is
3
min of exposure, and levels off afterwards to
reach 1.3–1.5 nm at the most. This range of thick-
ness is typical for the alumina layers formed under
UHV conditions on other aluminide alloys (NiAl)
[
16–22]. It amounts to oxide films consisting of 3–7
monolayers of oxide. The effect of the pre-anneal-
ing time is discussed further on.
The composition of the modified alloy region
TiAl
TiAl
TiAl
underneath the oxide, DAl , and DTi ¼ 1 ꢀ DAl
,

66
V. Maurice et al. / Surface Science 596 (2005) 61–73
Fig. 5. Changes of the composition of the modified alloy region
underneath the alumina layer for c-TiAl(111) exposed to
ꢀ7
P(O
ments performed after pre-annealing for (a) 110 min, (b)
00 min, (c) 350 min and (d) 550 min are shown.
2
) = 1.0 · 10 mbar at 650 ꢀC. Four series of measure-
2
growing predominantly by outward cation diffu-
sion [23,24]).
3
.2. Onset of Ti oxidation
Fig. 6. Ti 2p XPS spectra recorded at 90ꢀ and 45ꢀ take-off angle
ꢀ7
for the c-TiAl(111) sample exposed to P(O ) = 1.0 · 10 mbar
2
The detection of the onset of oxidation of tita-
at 650 ꢀC for 3 min (a) and 4 min (b). The spectra obtained after
nium was based on angle-resolved XPS measure-
ments. Fig. 6a and b show the superposition of
Ti 2p spectra measured at 90ꢀ and 45ꢀ take-off an-
gles after 3 and 4 min of exposure to oxygen,
respectively. For titanium oxides growing above
the metallic substrate, the corresponding higher
BE components of the Ti 2p signal must increase
at lower take-off angles, for which the signal is
more surface sensitive. This is not the case after
background subtraction are shown.
3
min of oxidation (the two spectra nearly per-
fectly superimpose (Fig. 6a)), whereas after 4 min
of oxidation (Fig. 6b), the spectrum obtained at
4
5ꢀ take-off angle reveals a significant increase on
the high BE side of the peaks, suggesting the
growth of titanium oxide particles at the surface
of the alloy and therefore the onset of oxidation
of titanium.
In Fig. 7 we have plotted the full width at half
maximum (FWHM) of the experimental Ti 2p3/2
peak as a function of d · sinh, i.e., the distance
in the aluminium oxide that the collected photo-
electrons must go through. The two series of mea-
surements performed at 90ꢀ and 45ꢀ during the
Fig. 7. Full width at half maximum of the Ti 2p3/2 peak
measured at 90ꢀ and 45ꢀ take-off angles during the initial stages
of oxidation of the c-TiAl(111) sample exposed to
ꢀ7
P(O ) = 1.0 · 10 mbar at 650 ꢀC.
2
growth of the oxide layer are shown. The thick-
ness, d, of the alumina layer was obtained as de-

V. Maurice et al. / Surface Science 596 (2005) 61–73
67
scribed above. It can be seen that the two series of
measurements superimpose up to a value of
d · sinh = 0.9. The increase of the FWHM (i.e.,
broadening of the Ti 2p signal) observed up to this
point is assigned to the inelastic scattering of the
photoelectrons through the alumina layer, increas-
ing with increasing thickness of the oxide. Above
d · sinh = 0.9, the two series of measurements in
Fig. 7 diverge, indicating that the broadening of
the Ti 2p3/2 cannot then be explained by the in-
crease of the inelastic scattering only, but also cor-
responds to the growth of titanium oxides at the
surface This confirms the onset of oxidation of
titanium since the thickness of 0.9 nm was attained
after 3 min of oxidation (for the sample pre-an-
nealed for 110 min). Note that above the onset of
Ti oxidation the thickness of the alumina layer still
increases indicating the simultaneous oxidation of
the two alloying elements.
Angle-resolved measurements were also used to
elucidate the location of the nucleation and growth
of the titanium oxides: below the alumina layer
(oxide/metal interface) or at the alumina surface
(gas/oxide interface). Assuming the growth of
two distinct continuous layers of aluminium and
TiOx
AlOx
titanium oxides, the density ratio DTi =DAl of
the two layers can be extracted from the intensity
Ox
Ox
ratio IðTi Þ=IðAl Þ. If the titanium oxide layer
2p
2p
grows below the alumina layer, the expression is
Ox
TiOx
TiOx
Ti
AlOx
Al
IðTi Þ
D
rTi2p
k
Ti2p
TTi2p
2p
¼
Â
Â
Â
AlOx
T
Al2p
Ox
IðAl Þ
D
r
Al2p
k
Al2p
2p
ꢀ
ꢀ
ꢁꢁ
ꢀ
ꢁ
ꢀ
dTiOx
ꢀdAlOx
Ti
1
ꢀ exp
 exp
TiOx
Ti
AlOx
k
Âsinh
k
Âsinh
2p
2
p
Â
ꢀ
ꢀ
ꢁꢁ
;
ꢀ
dAlOx
1 ꢀ exp
AlOx
Al
k
Âsinh
p
2
where dAlOx (dTiOx) is the thickness of the alumin-
ium (titanium) oxide layer calculated as described
above. If the titanium oxide layer grows above
the alumina layer, the expression is
Fig. 8 shows the curve fitting of the Ti 2p XPS
spectrum obtained after 16 min of exposure of the
c-TiAl sample to oxygen. It can be seen that the
experimental curves can be fitted by 3 sets of
Ox
TiOx
Ti2p
TiOx
Ti
AlOx
Al
2
p3/2–2p1/2 peaks (2:1 intensity ratio) at 454.6
and 460.6 eV, 457.5 and 463.2 eV, and 459.6 and
65.2 eV. These 3 sets correspond to the Ti(0),
Ti(III) and Ti(IV) oxidation states, respectively
8]. Their observation evidences the transient
IðTi Þ
rTi2p
k
TTi2p
2p
D
¼
Â
Â
Â
ꢀ
Ox
AlOx
Al2p
IðAl Þ
D
r
Al2p
k
T
Al2p
2p
4
ꢀ
ꢁꢁ
ꢀ
ꢀdTiOx
1
ꢀ exp
TiOx
k
Âsinh
Ti
[
2p
 ꢀ
ꢀ
ꢁꢁ
ꢁ:
3
+
growth of Ti oxides after the onset of oxidation
of titanium. After prolonged exposure to oxygen,
the formation of Ti
dominate.
ꢀd
ꢀd
AlOx
TiOx
1
ꢀ exp
 exp
AlOx
TiOx
k
Âsinh
k
Âsinh
Al
Al
2
p
2p
4
+
oxides was found to
In both cases, the values of the density ratio
obtained by varying the take-off angle h must
remain constant if the assumption made is valid.
TiOx
AlOx
The values of DTi =DAl obtained after 8 min of
exposure to oxygen are 0.73 and 1.07 at 90ꢀ and
45ꢀ, respectively, if one assumes that titanium
oxide grows below the alumina layer, and 0.36
and 0.37 at 90ꢀ and 45ꢀ, respectively, if one as-
sumes that titanium oxide grows above the alu-
mina layer. The significant difference obtained in
the first case allows us to exclude this possibility.
In the second case, the values are nearly equal
which validates the assumption that titanium
oxide is formed above the alumina layer. This re-
sult is in agreement with other results obtained
on polycrystalline samples [13,14], where the
growth of titanium oxide particles above the
Fig. 8. Curve fitting of the XPS Ti 2p spectrum for the c-
ꢀ7
TiAl(111) sample exposed to P(O
50 ꢀC for 16 min.
2
) = 1.0 · 10 mbar at
6

68
V. Maurice et al. / Surface Science 596 (2005) 61–73
alumina layer was determined from ion-sputtering
experiments. The equivalent thickness (0.17 nm) of
the titanium oxide layer obtained after 8 min of
exposure to oxygen is extremely low, smaller that
the equivalent thickness of ꢁ0.2 nm estimated for
one monolayer of oxide. This indicates that islands
partially covering the substrate are formed in this
initial stage of growth following the nucleation
of the titanium oxide.
The composition of Ti75± 2Al25± 2 obtained in the
present study right before the oxidation of titanium
is in excellent agreement with the value of Ti Al
25
obtained on polycrystalline samples [13,14], show-
ing that this is really a critical concentration for
the onset of oxidation of titanium on c-TiAl.
7
5
3.4. Corrosionprotectionu nd er low oxygen
pressure
3
.3. Effect of pre-annealing
Longer exposures of the alloy at 650 ꢀC and
2
ꢀ
7
ꢀ6
P(O ) = 1 · 10 mbar and 1 · 10 mbar were
In the experiment described above the alloy was
performed in a separate UHV system. In these
experiments, the sample was pre-annealed between
650 and 750 ꢀC for 20 h under a residual pressure
annealed for 110 min prior to oxidation. The effect
of the pre-annealing time has been investigated.
Figs. 4 and 5 show respectively the evolution of
the equivalent thickness of the alumina layer and
of the composition of the modified alloy region
with increasing exposure to oxygen after pre-
annealing time periods of 110, 200, 350 and
ꢀ
10
of ꢁ3 · 10
mbar in order to optimise the sur-
face structural ordering. UHV-STM revealed the
terrace and step topography of the surface after
this pre-treatment, allowing atomic scale measure-
ments on the terraces [13,25]. Fig. 9 shows the AES
spectrum obtained after this pre-treatment. The
peak at 506 eV (OKLL transition) shows the surface
5
50 min. It is observed in Fig. 4 that the longer
the pre-annealing time of the sample, the slower
the growth kinetics and the longer the duration
of the stage of selective oxidation of the alloy
uptake of oxygen. The peaks at 43 and 56 eV
Ox
LMM
(Al
) correspond to oxidized Al detected in
(
i.e., growth of Al O only). For the sample pre-
addition to the metallic Al signal at 68 eV
2
3
Met
annealed for 550 min, a slightly smaller thickness
of 1.3 nm is obtained after 210 min of exposure
at which point no oxidation of titanium was ob-
served. In this case, the slow growing alumina
layer provided a better protection of the alloy
against the oxidation of titanium, under the low
oxygen pressure used here.
(Al
). Some residual carbon contamination is
LMM
detected at 272 eV (CKLL). The small carbon
contamination is due to the fact that, in this
UHV chamber, the sample could not be flash-
annealed between 900 and 1000 ꢀC like in the
XPS chamber. The peaks at 30 eV (TiKLL),
386 eV (TiLMM) and 416 eV (TiLMM) are assigned
Fig. 5 shows that a striking effect of the pre-
annealing time is to decrease the kinetics of Al
depletion in the modified metallic region under
the oxide. This is consistent with the decrease of
the growth kinetics of the oxide layer. Note also
that for the sample pre-annealed for 550 min, for
which no titanium oxidation was observed even
after 210 min of exposure to oxygen, the Al con-
centration in the modified metallic phase is higher
(
(
Ti Al ) than that measured in the other series
70 30
Ti Al ). The alumina layer formed in this case
7
5
25
being slightly thinner than the less protective lay-
ers formed after shorter pre-annealing periods, it
can be concluded that the onset of oxidation of
titanium is related to a critical concentration of
the modified metallic phase underneath the oxide.
Fig. 9. AES spectrum of the c-TiAl(111) sample after pre-
annealing at 650 ꢀC under UHV for 20 h.

V. Maurice et al. / Surface Science 596 (2005) 61–73
69
Ox
LMM
Met
LMM
to Ti. The detection of the Al
signals, com-
tions of the TiLMM=Al
ratio. They are attrib-
bined with the selective oxidation of the alloy evi-
denced by XPS, indicates the presence of an ultra-
uted to some dispersion resulting from the mea-
Met
surement of the low intensity Al
peak at
LMM
thin layer of alumina covering the alloy surface
68 eV rather than to actual fluctuations of the Al
concentration in the modified metallic phase
underneath the alumina layer. This assignment is
supported by the structural analysis of this surface
[25] that showed an unchanged structure of the
pre-annealed surface after subsequent exposure
to oxygen and also after further annealing under
UHV, thus confirming the high stability of this
surface in these low pressure oxidizing conditions.
The equivalent thickness of the alumina layer, cal-
Met
(
the low-energy Al
signal at 68 eV is still ob-
LMM
served). Its formation results from the interaction
of the surface with residual oxygen traces out-gas-
sing from the sample holder during the prolonged
annealing treatment.
Fig. 10 shows the peak-to-peak height ratio
Ox
LMM
Met
(
i.e., intensity ratio) for the Al
and Al
,
LMM
Met
TiLMM and OKLL, and TiLMM and Al
signals
LMM
obtained after pre-treatment, and their evolution
after subsequent exposures to oxygen at P(O ) =
Ox
Met
culated from the Al
=Al
intensity ratio as
LMM
2
LMM
ꢀ
7
ꢀ6
1
· 10
and 1 · 10 mbar at 650 ꢀC (up to a
described above and assuming an unmodified con-
centration in the metallic phase, is estimated to
ꢁ0.4 nm. This suggests the formation of an extre-
mely thin alumina layer (ꢁ2 oxide monolayers)
protecting the alloy against the oxidation of tita-
nium.
cumulated exposure time of 20 h). The result ob-
tained after annealing the oxidized surface at
7
50 ꢀC for 15 h under UHV is also included in this
figure. A slow increase from ꢁ1.5 to ꢁ2.0 is ob-
Ox
Met
served for the Al
=Al
ratio with some fluc-
LMM
LMM
tuation in the data. This suggests a slight
thickening of the alumina layer covering the alloy.
However, this trend is not confirmed by the
TiLMM/OKLL ratio which remains constant, indi-
cating that there is no further uptake of oxygen
by the alumina covered surface during these oxida-
tion treatments and that the Ti concentration of
the metallic phase underneath the alumina layer
does not vary. The fluctuations of the data for
4. Discussion
The results obtained in the present study on
c-TiAl(111) single-crystal surfaces evidence that
a stage of selective oxidation of the alloy to form
a pure alumina layer precedes the oxidation of tita-
nium. Other data obtained in the same oxidizing
conditions on polycrystalline surfaces are consis-
tent [13,14]. In the previous studies of the initial
stages of oxidation of c-TiAl alloys at low oxygen
pressure, this stage of selective oxidation of alu-
minium has only been observed for experiments
performed at 850 ꢀC [9]. We note that, for the oxi-
dation experiments performed at 600 ꢀC [8,10,11],
the first analyses were performed after several min-
utes of oxidation, very likely not allowing the
observation of the transient stage of selective oxi-
dation which can be short (ꢁ3 min) depending
on the pre-annealing treatment of the sample as
shown in the present study.
Ox
Met
the Al
=Al
ratio are paralleled by fluctua-
LMM
LMM
Our results also show that the alumina layers
formed in this transient stage of selective oxidation
are extremely thin (61.5 nm). Thinner (ꢁ0.4 nm)
and protective layers can be prepared by pro-
longed pre-annealing of the substrate. Angle-
resolved XPS measurements performed on a
Fig. 10. AES intensity ratios for the c-TiAl(111) sample pre-
annealed at 650 ꢀC under UHV for 20 h, and subsequently
ꢀ7
ꢀ6
exposed to (a) P(O
2
) = 1.0 · 10 mbar, (b) P(O ) = 1.0 · 10
2
mbar at 650 ꢀC for a cumulated time of 20 h, and (c) finally
annealed at 750 ꢀC under UHV for 15 h.

70
V. Maurice et al. / Surface Science 596 (2005) 61–73
0
.8 nm thick film give evidence of the formation of
In the (111) orientation, each atomic plane of
a continuous layer of alumina. The formation of
islands covering partially the metallic substrate
was not observed, although it is expected after
nucleation of the oxide particles. This range of
thickness is consistent with that observed on poly-
crystalline surfaces [13,14] and typical for the alu-
mina layers formed under UHV conditions on
other aluminide alloys (NiAl) [16–22]. It corre-
sponds to oxide films consisting of 2–7 monolayers
of oxide.
the c-TiAl structure has a Ti Al concentration
5
0
50
ꢀ
and alternates close-packed h110i rows of Al
and Ti atoms. Assuming, in first approximation,
that the Ti concentration is not modified in an
Al-depleted metallic phase, the concentration of
injected Al vacancies (V(Al)) can be deduced from
the measured relative concentration of the modi-
fied phase and compared with the number of in-
jected vacancies required to form the alumina
layer. For the 0.8 nm thick alumina layer formed
after 2 min of oxidation (Fig. 3), the modified rel-
ative concentration measured at 45ꢀ take-off angle
is Ti Al which amounts to an average value of
The data obtained after different pre-annealing
treatments show that the thickness of the alumina
layer does not seem to be critical to optimise the
protection of the alloy by a pure alumina layer. In-
7
2
28
the concentration of Al vacancies of 31%
(Ti Al V(Al) ) for each atomic plane of the
ꢀ
7
deed, no oxidation of titanium at P(O ) = 1 · 10
2
50 19
31
mbar and 650 ꢀC was observed by XPS for a 1.3 nm
thick film whereas it was observed for 1.5 nm films.
The protective film analysed by AES was only
alloy below the 0.8 nm oxide film. The probed
depth in the XPS measurement can be estimated
to 2 · k · sinh (corresponding to 89% of the
signal). It amounts to 2.8 nm in this experiment
0
.4 nm thick. This is consistent with the fact that
Al2O3
TiAl
Al2p
the more protective the oxide layer the slower its
growth for equal conditions of oxidation.
(kAl2p ¼ k
¼ 2 nm [13,14]; h = 45ꢀ). The oxide
layer being 0.8 nm thick, a depth of 2.0 nm is
therefore measured in the alloy, which amounts
to ꢁ9 atomic planes given the reticular distance
of 0.232 nm of the (111)-oriented c-TiAl structure.
The total amount of Al vacancies is therefore
The onset of oxidation of titanium was repeat-
edly observed for a critical value of the concentra-
tion of the modified metallic phase underneath the
oxide film. The critical value of Ti75± 2Al25± 2 for
the relative concentration measured in the present
study is also found for polycrystalline alloy sur-
faces [13,14].
1
5
ꢀ2
estimated to 3.75 · 10 cm (2.7 equivalent mon-
olayers (ML)). If one takes the relative concentra-
tion of Ti Al (Ti Al V(Al) ) measured at
6
3
37
50 29
21
A depletion of Al in a modified metallic phase
underneath the oxide film has previously been ob-
served in oxidation studies performed at atmo-
spheric pressure [1–3,5]. In the present work we
have shown that the Al depletion under the oxide
film requires not only the selective oxidation of Al
but also a growth kinetics of the alumina layer lim-
ited by the diffusion of Al in the substrate (other-
wise, a near equilibrium Al concentration would
be measured in a re-homogenized metallic phase).
At 650 ꢀC transient alumina layers are formed and
have been actually observed to grow on TiAl sub-
strates [26]. They grow by a cationic outward dif-
fusion mechanism requiring the injection of Al
vacancies at the metal/oxide interface [23,24].
Therefore, the Al depletion of the metallic phase
measured in our study can be related to the injec-
tion of Al vacancies by the growth mechanism of
the oxide.
90ꢀ take-off angle, one obtains a value of
15
ꢀ2
4.06 · 10 cm
of vacancies (2.9 ML). These
values are in good agreement with the total
number of injected vacancies estimated from the
thickness of the alumina layer. Indeed the value
of 0.8 nm corresponds to ꢁ4 oxide monolayers,
which, given the Al O stoichiometry, amounts
2
3
1
5
ꢀ2
to 3.73 · 10 cm of Al vacancies (2.66 ML) in-
jected to grow the oxide layer.
These calculations confirm that the modifica-
tion of the metallic phase underneath the oxide
layer results from the presence of Al vacancies in-
jected in the metal by the growth mechanism of the
oxide. The observed accumulation of the vacancies
at the metal/oxide interface shows that the growth
kinetics is limited by the diffusion of Al to the
metal/oxide interface, in a process, which is too
slow to allow the re-homogenisation of the metal-
lic phase during the oxidation. Recent STM results

V. Maurice et al. / Surface Science 596 (2005) 61–73
71
obtained on this system have confirmed the pres-
ence of 2Dand 3Dinterfacial defects assigned to
the presence of metal vacancies at the metal/oxide
interface [25]. These measurements provided direct
observations of the interfacial vacancies produced
by injection at the metal/oxide interface, in agree-
ment with the conclusions of the present study.
The concentration of the interfacial vacancies
was significantly lower than the total number of in-
jected vacancies required to grow the ultra-thin
alumina layer because, in the reported case, the
metallic phase was re-homogenized by prolonged
annealing treatments.
as observed on Ni Al(111) [28–31] and NiAl(111)
3
[21] surfaces. Optimising the epitaxial growth of
the oxide layer decreases the density of defects that
form preferential diffusion pathways for the ionic
transport through the film.
The results reported in this paper show that the
oxidation of titanium is characterized by the for-
3
+
4+
mation of Ti and Ti oxide species, indicating
the transition from a selective oxidation regime
to a simultaneous oxidation regime of both ele-
ments. This simultaneous oxidation regime was
also observed in previous studies performed at
600 ꢀC under low oxygen pressure [8,9,11]. No
2
+
A marked effect of pre-annealing the substrate
before oxidation is to slow down the growth rate
of the alumina layer. Pre-annealing was performed
to improve structural ordering by re-crystallization
of the highly defective surfaces obtained after the
mechanical polishing and Ar ion bombardments.
The LEEDmeasurements performed in the sepa-
rate structural study confirmed that at least 10 h
of annealing at 750 ꢀC were necessary to produce
relatively sharp diffraction spots [13,25]. A higher
structural ordering of the substrate is consistent
with the decrease of the growth kinetics of the
oxide measured in the present study. Diffusion of
the alloying elements is slow in intermetallics due
to their strong interaction that provides structural
and chemical ordering in these phases (the diffu-
Ti species are observed in our study in contrast
with other observations [8,10]. This may be an ef-
2
+
fect of temperature on the stability of the Ti
oxide species at 650 ꢀC. Indeed, a previous study
2
+
[10] has shown that whereas Ti
species are
3
+
4+
formed at 30 ꢀC, Ti and Ti species are formed
2
+
above 600 ꢀC, indicating the instability of the Ti
oxide species with increasing temperature.
The angle-resolved XPS measurements per-
formed in this study allowed us to precisely deter-
mine the onset of oxidation of titanium, and to
characterize the oxide islands partially covering
the surface of the alumina layer. The formation
of the titanium oxide particles at the outer oxide/
gas interface suggests that titanium atoms, oxi-
3
+
dized at the internal interface, diffuse as Ti cat-
ions through the alumina lattice to react with
oxygen at the external (oxide/gas) interface. Thus,
when slower growing and better ordered alumina
scales are formed, the lower density of preferential
diffusion pathways in the oxide film decreases the
rate of transport of cations to the outer interface,
thus decreasing the growth rate of the oxide and
the Al depletion in the underlying metallic phase.
If the critical concentration of Ti Al (Ti Al -
ꢀ
22
2
sion coefficient of Al in c-TiAl is 5 · 10
m /s
at 650 ꢀC [27]). Diffusion is expected to signifi-
cantly increase in the highly defective non-ordered
surface layer produced by mechanical polishing
and ion bombardments. Therefore, increasing
structural order will decrease the rate of diffusion
of Al from the bulk to the alloy/oxide interface
where Al is consumed by oxidation.
Another effect of the increase in structural
ordering of the alloy surface achieved by annealing
is to optimise the epitaxial growth of ordered tran-
sient alumina layers. The exact structure of the
alumina layer is not determined in this system.
Based on the identical symmetry and small lattice
mismatch (1.4%) of c-TiAl(111) planes and
7
5
25
50 17
V(Al) ) is reached in the metallic phase, the re-
3
3
lated increase of the activity of titanium leads to
a competitive oxidation of the Ti and Al atoms
with transfer of both cations to the outer interface
to form Al(III) and Ti(IV) oxides. The comparison
with similar measurements performed on polycrys-
talline a2-Ti Al samples, for which the onset of
(
111) planes of the FCC oxygen sub-lattices of
7
5
25
the transient alumina, the parallel (or anti-parallel)
orientation of the FCC oxygen sub-lattice of a
c-like Al O layer can be expected on c-TiAl(111),
Ti oxidation is observed for a modified concentra-
tion of Ti Al [13,14], shows that the decrease
of the number of Al–Ti bonds in the Al-depleted
8
2
18
2
3

72
V. Maurice et al. / Surface Science 596 (2005) 61–73
c-phase, rather than just the nominal Ti concentra-
tion of 75 at%, determines the Ti activity in the
modified intermetallic.
concentration of Ti Al (Ti Al V(Al) ), corre-
72 28 50 19 31
1
5
sponding to the injection of 3.75 · 10 vacancies/
ꢀ
2
cm , has been measured, in excellent agreement
with the total number of Al vacancies required to
In contrast, if the growth kinetics is slow en-
ough to keep the Ti content below the critical con-
centration of Ti Al for the modified c-phase,
1
5
grow 4 monolayers of oxide (3.73 · 10 vacan-
ꢀ
2
cies/cm ). Improving the structural ordering of
the substrate by prolonged annealing pre-treat-
ments slows down the growth kinetics of the oxide
and the related Al-depletion of the substrate, and
increases the resistance of the alloy to the subse-
quent oxidation of titanium. This is assigned to a
combined slower diffusion of Al in the less defec-
tive ordered metallic phase and to the growth of
better ordered alumina layers allowing to slow
down the ionic transport through the oxide. Good
protection of the alloy against the oxidation of
titanium has been observed at 650 ꢀC under
7
5
25
only Al atoms are oxidized and their transport to
the outer interface is hindered by the structural
ordering of the oxide layer, eventually nearly stop-
ping the growth of the oxide. Ordering of the sur-
face of the oxide layer may also decrease the
adsorption of the oxygen species at the surface,
thus also contributing to hinder the growth of
the oxide. Full protection of the alloy against the
oxidation of titanium under low oxygen pressure
conditions was observed in our study for ultra-thin
layers of pure alumina (1.3 and 0.4 nm), covering
completely the substrate, and annealed to optimise
their structural order.
ꢀ
7
1.0 · 10 mbar of gaseous oxygen for a 1.3 nm
thick alumina layer, despite an average modified
concentration of Ti Al (Ti Al V(Al) ) in the
7
0
30
50 21
29
metallic phase underneath the oxide. Highly stable
and corrosion resistant alloy surfaces covered by a
5
. Conclusion
0.4 nm thick alumina layer have been obtained by
The early stages of oxidation of c-TiAl(111)
slowly oxidizing the alloy at lower partial pressure
ꢀ
10
surfaces have been studied at 650 ꢀC under low
(<5.0 · 10
mbar).
ꢀ
7
ꢀ6
oxygen pressure (1.0 · 10 –1.0 · 10 mbar).
The thickness of the oxide layers, the modifica-
tions of the alloy substrate underneath the growing
oxide scale, and the corrosion protection were
investigated by XPS and AES. The data evidence
two distinct oxidation regimes.
The second oxidation regime corresponds to the
simultaneous oxidation of aluminium and tita-
nium. The determination of the onset of oxidation
of titanium was based on angle-resolved XPS data
allowing to discriminate the influence of inelastic
scattering through the growing alumina layer from
the formation of titanium oxide particles, both
contributing to increase the high BE signal in the
Ti2p3/2–Ti2p1/2 spectra. Angle-resolved XPS data
also provide evidence that the Ti oxide particles
grow above the alumina layer at the oxide/gas inter-
face. The onset of titanium oxidation was repeat-
edly observed for a critical concentration of Ti in
the modified region of the alloy underneath the alu-
The first regime is characterized by the growth
of a pure alumina layer resulting from the selective
oxidation of aluminium. Continuous alumina lay-
ers completely covering the substrate have been
obtained, as evidenced by angle-resolved XPS
data. Their thickness ranges from ꢁ0.4 to ꢁ1.5
nm, which amounts to 2–7 monolayers of oxide.
The composition of the metallic phase underneath
the growing oxide is modified by a depletion of Al,
forming a concentration gradient below the oxide/
metal interface. The observation of vacancies
underneath the growing oxide indicates a growth
kinetics limited by the diffusion of Al to the
oxide/metal interface, and is consistent with the
injection of Al vacancies in the metal during
the growth of the transient alumina at 650 ꢀC.
Underneath a 0.8 nm thick oxide layer, a modified
mina layer: Ti75± 2Al25± 2 (Ti Al17± 2V(Al)33± 2),
50
showing that the depletion in Al decreases the num-
ber of Ti–Al bonds and increases the activity of the
Ti atoms up to a critical point where the oxidation
of Ti at the oxide/metal interface becomes compet-
itive with that of Al. The growth of titanium oxide
particles above the alumina layer indicates the
3
+
transport of Ti cations through the alumina layer
and subsequent reaction with oxygen at the outer

V. Maurice et al. / Surface Science 596 (2005) 61–73
73
interface. Slower growing and better ordered alu-
mina layers decrease the density of preferential
pathways and the ionic transport through the film
thus allowing some re-homogenisation of the
underlying metallic phase limiting the Al depletion
and maintaining the Ti concentration below a crit-
ical value. This is a way to obtain improved resis-
tance of the alloy to high temperature oxidation.
[12] P.K. Datta, Z. Klusek, S. Pierzgalski, H.L. Du, in:
Proceedings of CORROSION 2003, San Diego, CA, 16–
2
0 March 2003, NACE International, p. 9.
[
13] G. Despert, Study of the initial stages of oxidation of
titanium aluminides: gas turbine applications, PhDThesis,
University Paris VI, Paris, 2003.
[14] V. Maurice, G. Despert, S. Zanna, P. Josso, M.-P. Bacos,
P. Marcus, to be submitted.
[15] I. Shimizu, Y. Setsuhara, S. Miyake, M. Kumagai, K.
Ogata, M. Kohata, K. Yamaguchi, Surf. Coat. Technol.
1
31 (2000) 187.
16] N. Fr e´ my, V. Maurice, P. Marcus, Surf. Interf. Anal. 34
2002) 519.
[
Acknowledgements
(
[
[
17] V. Maurice, N. Fr e´ my, P. Marcus, Surf. Sci. 581 (2005) 88.
18] C. Becker, J. Kandler, H. Raaf, R. Linker, T. Pelster, M.
Draeger, M. Tanemura, K. Wandelt, J. Vac. Sci. Technol.
A 16 (1998) 1000.
19] M. Baumer, H.-J. Freund, Prog. Surf. Sci. 61 (1999) 127.
20] S.G. Addepalli, B. Ekstrom, N.P. Magtoto, J.-S. Lin, J.A.
Kelber, Surf. Sci. 442 (1999) 385.
The partial financial support of this work by the
‘D e´ l e´ gation G e´ n e´ rale pour lÕArmement (DGA)’’
is acknowledged.
‘
[
[
References
[
[
21] R. Franchy, Surf. Sci. Rep. 38 (2000) 195.
22] A. Stierle, V. Formoso, F. Comin, R. Franchy, Surf. Sci.
467 (2000) 85.
[
1] A. Rahmel, W.J. Quadakkers, M. Sch u¨ tze, Mat. Corr. 46
1995) 271.
2] S. Becker, A. Rahmel, M. Schorr, M. Sch u¨ tze, Oxid. Met.
8 (1992) 425.
(
[23] R. Presscott, D.F. Mitchell, M.J. Graham, J. Doychak,
Corros. Sci. 37 (1995) 1341.
[24] B.A. Pint, J.R. Martin, L.W. Hobbs, Solid State Ionics 78
(1995) 99.
[25] V. Maurice, G. Despert, S. Zanna, M.-P. Bacos, P.
Marcus, Nature Mater. 3 (2004) 687.
[
3
[3] C. Lang, M. Sch u¨ tze, Oxid. Met. 46 (1996) 255.
[4] C. Lang, M. Sch u¨ tze, Mat. Corr. 48 (1997) 13.
[5] M. Schmitz-Niederau, M. Sch u¨ tze, Oxid. Met. 52 (1999)
2
25.
[26] F. Dettenwanger, E. Schumann, M. R u¨ hle, J. Rabowski,
G.H. Meier, Mater. Corros. 48 (1996) 23.
[
[
[
[
6] M.R. Shanabarger, Mat. Sci. Eng. A 153 (1992) 608.
7] M.R. Shanabarger, Appl. Surf. Sci. 134 (1998) 176.
8] T.N. Taylor, M.T. Paffett, Mat. Sci. Eng. A 153 (1992) 584.
9] J. Geng, G. Gantner, P. Oelhafen, P.K. Datta, Appl. Surf.
Sci. 158 (2000) 64.
[27] Y. Mishin, C. Herzig, Acta Mater. 48 (2000) 589.
[28] U. Bardi, A. Atrei, G. Rovida, Surf. Sci. 239 (1990) L511.
[29] U. Bardi, A. Atrei, G. Rovida, Surf. Sci. 286 (1992) 87.
[30] C. Becker, J. Kandler, H. Raaf, R. Linke, T. Pelster, M.
Dr a¨ ger, M. Tanemura, K. Wandelt, J. Vac. Sci. Technol. A
16 (1998) 1000.
[
[
10] M. Schmiedgen, P.C.J. Graat, B. Baretsky, E.J. Mittemei-
jer, Thin Sol. Films 415 (2002) 114.
11] K. Kovacs, I.V. Perczel, V.K. Josepovits, G. Kiss, F. Reti,
P. Deak, Appl. Surf. Sci. 200 (2002) 185.
[31] S.G. Addepalli, N.P. Magtoto, J.A. Kelber, Surf. Sci. 458
(2000) 123.