Growth, electronic properties and thermal stability of the Fe/Al2O3 interface

Article abstract of DOI:10.1016/S0039-6028(02)02306-3

Soft X-ray photoelectron spectroscopy and resonant photoemission have been used to study the growth and electronic properties of Fe ultrathin films deposited on Al₂O₃ substrates. A simultaneous multilayer growth mode has been found for Fe growth at room temperature. For iron coverages below 1 ML, Fe²⁺ species are formed at the Fe/Al₂O₃ interface, followed by the formation of a metallic iron overlayer. The bonding of Fe at very low coverages occurs by charge transfer from Fe to surface oxygen atoms, and neither hybridisation of Fe and Al states nor reduction of the Al₂O₃ substrate are observed. The thermal stability of the interface has been also studied in the range 673-873 K. Annealing produces Fe agglomeration in such a way that some areas of the Al₂O₃ substrate become fully Fe-depleted. In these Fe-depleted areas, Fe²⁺ completely disappears and Al⁰ reduced species are formed. This behaviour would explain the decrease in the magnetoresistance performance of magnetic tunnel junctions after annealing above 573 K.

Full text of DOI:10.1016/S0039-6028(02)02306-3

Surface Science 521 (2002) 77–83
www.elsevier.com/locate/susc
Growth, electronic properties and thermal stability of the
Fe/Al2O3 interface
a,b
A. Arranz , V. Perez-Dieste , C. Palacio
b
a,
*
ꢀ
a
Departamento de Fꢀısica Aplicada, Facultad de Ciencias, C-XII, Universidad Autoꢀnoma de Madrid, Cantoblanco, 28049Madrid, Spain
b
LURE, Centre Universitaire Paris-Sud, Ba^t. 209 D, B.P. 34, 91898 Orsay Cedex, France
Received 9 July 2002; accepted for publication 15 September 2002
Abstract
Soft X-ray photoelectron spectroscopy and resonant photoemission have been used to study the growth and elec-
tronic properties of Fe ultrathin films deposited on Al2O3 substrates. A simultaneous multilayer growth mode has been
found for Fe growth at room temperature. For iron coverages below 1 ML, Fe2þ species are formed at the Fe/Al2O3
interface, followed by the formation of a metallic iron overlayer. The bonding of Fe at very low coverages occurs by
charge transfer from Fe to surface oxygen atoms, and neither hybridisation of Fe and Al states nor reduction of the
Al2O3 substrate are observed. The thermal stability of the interface has been also studied in the range 673–873 K.
Annealing produces Fe agglomeration in such a way that some areas of the Al2O3 substrate become fully Fe-depleted.
In these Fe-depleted areas, Fe2þ completely disappears and Al0 reduced species are formed. This behaviour would
explain the decrease in the magnetoresistance performance of magnetic tunnel junctions after annealing above 573 K.
Ó 2002 Elsevier Science B.V. All rights reserved.
Keywords: Growth; X-ray photoelectron spectroscopy; Photoemission (total yield); Metallic films; Magnetic films; Iron; Aluminum
oxide
1. Introduction
interfacial microstructure and chemical composi-
tion at the interface.
During the last years there has been an in-
creasing interest in the study of metal–ceramic
systems as a consequence of their use in a variety
of technologies requiring specific electronic, optical,
magnetic, mechanical or catalytic properties [1–13].
The above-mentioned properties of metal–ceramic
components are in general strongly affected by the
In order to understand bonding and adhesion
mechanisms and other phenomena that are specific
of the metal–ceramic interfaces, several funda-
mental studies have been carried out recently on
interfaces formed by growing ultrathin metallic
layers of different metals on oxide substrates [1]. In
particular, ultrathin iron films grown on Al2O3
surfaces have a great potential in the field of
nanoscale magnetism, where the Fe/Al2O3/Fe sys-
tem is used as a magnetic tunnel junction (MTJ).
The magnetoresistance of such MTJÕs depends
critically on the spin polarisation at the interfaces
between the magnetic film and the insulator layer,
* Corresponding author. Tel.: +34-91-3975266; fax: +34-91-
3974949.
E-mail address: carlos.palacio@uam.es (C. Palacio).
0039-6028/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved.
PII: S0039-6028(02)02306-3

78
A. Arranz et al. / Surface Science 521 (2002) 77–83
and therefore on the composition and sharpness of
the interface [14–16]. In spite of its promising in-
terest not only in magnetic but also in catalytic
applications, only few studies on the formation of
the Fe/Al2O3 interface are available [17,18].
In this work we present a study of the growth
and electronic properties of Fe ultrathin films de-
posited on Al2O3 substrates at room temperature
(RT), by means of soft X-ray photoelectron spec-
troscopy (SXPS) and resonant photoemission
(RPE). We also analyse the thermal stability of the
interface up to 873 K.
2. Experimental
Fig. 1. Photoemission spectra measured with hm ¼ 150 eV of (a)
Al 2p band, (b) Fe 3p band, and (c) VB, respectively, for dif-
ferent iron deposition times. The deconvolution of Fe 3p spec-
tra carried out as explained in the text is also shown.
The experiments were performed at LURE
(Orsay, France) using the PES2 experimental sta-
tion connected to the SU8 ondulator beamline of
the Super-Aco storage ring. The measurements
were carried out in a purpose-built ultrahigh vac-
uum system, with a base pressure better than 1 Â
10À10 Torr, equipped with an angle resolving 50
mm hemispherical VSW analyser coupled on a gon-
iometer inside the chamber. For a photon energy
(hm) of 150 eV, the overall energy resolution, in-
cluding the analyser, was estimated to be better
than 100 meV. Photoemission spectra have been
normalised to the incident current measured in a
gold grid located at the entrance of the chamber.
The Al2O3 substrate was made by annealing
(ꢀ723 K) an aluminium-foil substrate of 99.994%
purity in 5N oxygen at a partial pressure of 1Â
10À5 Torr for about 30 min. Complete oxida-
tionwas assured by the absence of a metallic com-
ponent in the Al 2p band. Iron was evaporated by
electron bombardment of a high purity Fe-wire tar-
get at constant heating power, in such a way that
the thickness of the Fe layer is controlled by the
deposition time. During Fe evaporation the resid-
ual pressure was always kept below 1 Â 10À9 Torr.
150 eV for different iron deposition times, after
background subtraction based on a modified
Shirley method [19]. The spectra labelled t ¼ 0 min
are representative of the Al2O3 substrate. The
Al 2p is characterised by a single symmetric band,
and no reduced Al species are observed in the low
binding energy side upon iron deposition. The
Al 2p spectra have been aligned at 75.6 eV, as
observed in Fig. 1(a), using the applied energy shift
to correct surface charging effects in the Fe 3p and
VB spectra. The full width at half-maximum of the
Al 2p band at 0 min is 2 eV, decreasing to 1.6 eV
at 45 min. This decrease is a consequence of
the formation of an iron-conducting layer on the
surface which leads to a more homogeneous sur-
face potential [20]. With increasing iron deposition
time, the Al 2p strongly attenuates and practically
disappears for higher deposition times.
For small deposition times, the Fe 3p band
shows a complex shape. In particular, for t ¼ 15
min, the main peak of the Fe 3p band is observed
at ꢀ55.2 eV, being attributed to the formation of
Fe2þ species during the very first stages of the Fe/
Al2O3 interface formation [13]. With increasing Fe
deposition time, the Fe2þ peak strongly attenuates
and an asymmetric peak at 53 eV, that can be at-
tributed to the formation of metallic iron, Fe0,
emerges. In addition to that, another peak, Fes,
3. Results and discussion
3.1. Deposition at room temperature
Fig. 1 shows (a) the Al 2p, (b) Fe 3p and
(c) valence band (VB) spectra measured with hm ¼

A. Arranz et al. / Surface Science 521 (2002) 77–83
79
can be observed at 50.7 eV for all deposition times.
The separation between the Fes and Fe0 peaks
is independent of the photon energy, therefore
indicating that the Fes peak should be associated
to a photoemission peak rather than to an Auger
transition. The intensity of Fes peak remains nearly
constant for t P 45 min. This behaviour suggests
that Fes species are located at the surface, and
probably correspond to Fe atoms with a lower
atomic coordination, that is to surface atoms in
the outer atomic layer. The deconvolution of the
Fe 3p spectra in terms of Fe0, Fe2þ and Fes com-
ponents is illustrated in Fig. 1(b).
Fe thickness (ML)
0
2
4
6
8
10
1.0
0.8
0.6
0.4
0.2
0.0
Al 2p
Fe 3p
Fe0
Fe2+
Fes
0
50
100
150
200
250
The VB spectra of the Al2O3 substrate is char-
acterised by two broad peaks at ꢀ6.2 and 10 eV,
that are attributed to p and r molecular orbitals,
respectively, formed by hybridisation of Al 3sp
and O 2p atomic orbitals [21]. It should be pointed
out that the Al2O3 VB of the thermal film presents
a bandgap greater than 3 eV with no electronic
states. Since oxygen vacancies are characterised by
the formation of occupied states between 2–3 eV
[9,22], this should indicate the formation of a
highly stoichiometric oxide film. During iron
deposition, a strong attenuation of the p and r
peaks as well as the growth of the metallic Fe 3d
band at ꢀ0.8 eV below the Fermi level, EF, are
observed.
Fig. 2 shows the normalised Al 2p (j) and total
Fe 3p (ꢁ) signal intensities as a function of iron
deposition time. In addition, the Fe0, Fe2þ and Fes
signal intensities obtained from the deconvo-
lution of spectra in Fig. 1(b) are also shown.
With increasing iron deposition time, the Al 2p
band strongly attenuates and nearly disappears
for higher deposition times, therefore suggesting
that the Al2O3 surface is completely covered with
iron. This behaviour rules out either Volmer We-
ber (VW) or Stranski-Krastanov (SK) growth
modes [11,23]. On the other hand, the continuous
variation of the Al 2p and Fe 3p signals to reach
saturation suggests a simultaneous multilayer (SM)
growth mode, rather than pure layer-by-layer. In
this model iron condensates in layers in such a way
that at a layer starts to grow before the preceding
layer is completed. Therefore, the surface consists
on a series of terraces at different atomic levels
[24]. According to the SM growth mode, the Al
t (min)
Fig. 2. Normalised Al 2p (j) and total Fe 3p ( ) signal in-
ꢁ
tensities as a function of iron deposition time. The evolution of
the Fe0, Fe2þ and Fes signal intensities obtained from the de-
convolution of spectra in Fig. 1(b) are also shown.
and total Fe normalised intensities can be de-
scribed by Eqs. (1) and (2), respectively [23]
!
ÀaFeRt
IAl ¼ exp
ð1Þ
ð2Þ
kFAel cos u
!
ÀaFeRt
kFFee cos u
Fe
IFe ¼ 1 À exp
Fe
Fe
where k and k are the attenuation lengths of
Fe 3p and Al 2p photoelectrons in Fe, respectively,
R is the evaporation rate in ML/min, u is the take-
Al
ꢁ
off angle, and aFe ꢂ 2:3 A is the iron monolayer
thickness. Eqs. (1) and (2) have been used to fit
the data of Fig. 2, using R as the only adjustable
Fe
Fe
ꢁ
parameter and assuming kFe ¼ kAl ¼ 4:4 A [25].
Experimental measurements were carried out at
normal emission, that is u ¼ 0°. The best fit pa-
rameter is R ¼ 0:039 ML/min for the Al 2p data
and R ¼ 0:041 ML/min for the Fe 3p data. There-
fore, the evaporation rate, R ¼ 0:040 Æ 0:001 ML/
min, was used to calibrate the top axis in Fig. 2.
VW-like growth mode has been reported for the
growth of Ag, Pt, Pd, Rh, Co and Al on Al2O3
substrates [10], and its explanation was formulated
in terms of surface free energy [1]. According to
this criterion, a VW growth mode should be ex-
pected for the Fe/Al2O3 system [1]. However, as

80
A. Arranz et al. / Surface Science 521 (2002) 77–83
pointed out before, the Al and Fe signal evolutions
of Fig. 2 allow to rule out such a growth mode.
The SM growth mode has been also observed
during the growth of vanadium oxide on Al2O3
[20]. Such a type of growth mode is indicative of
very low or negligible lateral mobility of the atoms
impinging on the surface. To our knowledge, this
is the first time that the growth mode of Fe de-
posited on Al2O3 substrates at RT has been de-
termined.
orbitals of the transition metals (Ag, Cu, Ni and
Fe) may establish interactions with the p-orbitals
of the O2À anions of the Al2O3 surface, whereas
interactions between the metal ions of the support
and the deposited metal are not predicted. There-
fore, as pointed out by Diebold et al. [13] for the
Fe/TiO2 system, bonding of Fe at very low cov-
erages should take place by charge transfer from
Fe to surface oxygen atoms, whereas hybridisation
of Fe and Al states should not take place. This is in
agreement with results of Fig. 1, where Fe2þ spe-
cies are formed for Fe coverages below 1 ML, but
no reduction or formation of metallic Fe–Al spe-
cies at lower binding energies can be observed.
It is well established that the affinity of the de-
posited metal to oxygen is a crucial factor deter-
mining the interaction strength at the metal/oxide
interface, the wetting characteristics of the depos-
ited metal, and the mechanical properties of the
interface [3,10]. A strong interaction between
Fe, at low fractional monolayer coverages, and
TiO2 and MnO substrates has been observed in
the literature [12,13]. In particular, Diebold et al.
[13] have observed the formation of an oxidized-
Fe-reduced-TiO2 interface followed by the forma-
tion of a metallic Fe overlayer during the growth
of Fe on TiO2(1 1 0) substrates at RT. The oxida-
tion of several metals (Ti, Nb, V, Cr, Ni, Cu)
supported on alumina substrates has been also
observed [5,6,8–10]. Ealet and Gillet [9] have
found that metals with low electronegativity like
Cr or Ni promote the formation of an oxide phase
at the interface, whereas atoms with a higher
electronegativity (close to oxygen) induce an elec-
tronic transfer from the oxide substrate to the
metal deposit. Varma et al. [6] have found that Cu
donates charge to surface oxygen with the forma-
tion of Cu–O bonds at the Cu/Al2O3 interface,
being the reaction layer limited to one atomic
layer. In contrast with TiO2, Al2O3 cannot be
easily reduced. However, Al reduced species have
been observed during Ti and Nb deposition on
Al2O3 substrates at RT [5]. For iron (see Fig. 1(b)),
the formation of Fe2þ species takes place during
the first deposited Fe ML, as expected from the
high chemical reactivity of iron to oxygen, but no
Al reduction can be observed in Fig. 1(a), in good
agreement with the Auger electron spectroscopy
(AES) results of Colaianni et al. [17]. A theoretical
study of Johnson and Pepper [2] on the interfaces
between alumina and transition metals, using the
molecular orbital approach, predicts that the d-
3.2. Resonant photoemission
Further complementary information on the
electronic structure of the deposited films can be
obtained through RPE experiments. To carry out
this task, VB spectra have been measured near the
3ML Fe/Al2O3
h
(eV)
61
60
59
58
57
56
55
54
53
52
51
50
49
48
47
46
45
44
43
42
41
40
O 2p
6
Fe 3d
2
12 10
8
4
0
Binding energy (eV)
Fig. 3. VB spectra of a 3 ML Fe film deposited on stoichio-
metric Al2O3 as a function of the photon energy in the range
hm ¼ 40–61 eV.

A. Arranz et al. / Surface Science 521 (2002) 77–83
81
Fe 3p ! 3d excitation threshold. Fig. 3 shows the
VB spectra of an iron film 3 ML thick deposited
on stoichiometric Al2O3 as a function of the
binding energy, for different photon energies in the
range hm ¼ 40–61 eV. Spectra have been norma-
lised to the incident photon flux. Similar series
have been also measured for iron films of thickness
1.8 and 5.4 ML, respectively. RPE is usually in-
terpreted as due to the interference between the
direct photoemission process from the Fe 3d levels,
3p63dn þ hm ! 3dnÀ1 þ eÀ, and the autoionization
emission of a highly localised excited state created
by photoabsorption, 3p63dn þ hm ! ½3p53dnþ1ꢃꢄ !
3p63dnÀ1 þ eÀ [26]. In iron oxides, RPE is ex-
plained as due to the hybridisation between the
O 2p and Fe 3d orbitals and therefore RPE can be
used to isolate the Fe 3d-derived states contribu-
tion to the VB [27].
Constant-initial-state (CIS) curves of (a) Fe 3d
states at ꢀ0.8 eV, and (b) O 2p derived states at
ꢀ6.2 eV below EF for the 1.8, 3 and 5.4 ML Fe
films deposited on Al2O3 at RT are shown in Fig.
4. The shape of the CIS curves for the Fe 3d states
is similar to that observed by Kaurila et al. [28]
on bulk iron and iron thin films deposited on Cu
substrates, in good agreement with the mainly
metallic behaviour observed in the Fe 3p spectra of
the Fe films thicker than 1 ML. A careful analysis
of the position of the destructive minimum before
the enhancing resonance carried out fitting the
data of Fig. 4(a), in the range hm ¼ 51–55 eV, to a
parabola shows that the minimum is at 53.5 eV for
the film 1.8 ML thick and at 52.7 eV for the films 3
and 4.5 ML thick. Since FeOx single crystals have
Fe 3d-derived states at ꢀ1.2 eV below EF that
show resonance profiles characterised by a mini-
mum at hm ¼ 54 eV [27], the above-mentioned
difference should be attributed to the initial for-
mation of Fe2þ species for low iron coverages.
With increasing the film thickness, the influence of
Fe2þ species in the resonance profile decreases
in such a way that the minimum of the CIS curve
shifts to lower energies to reach the expected value
for metallic iron [27,28].
As shown in Fig. 4(b), a resonance is observed
at hm ¼ 55 eV for the O 2p VB derived states for
the different Fe films. According to previous pub-
lished results [29], the metallic Fe MVV Auger
transition overlaps with the O 2p derived states at
hm ꢀ 55–56 eV, and therefore the observed reso-
nance in Fig. 4(b) should be attributed to the Fe
MVV Auger transition, especially for higher Fe
thicknesses when the contribution of metallic iron
becomes dominant. However, the intensity of the
resonance is practically independent on the Fe
thickness. This fact suggests that for low Fe cov-
erages (1.8 ML), the resonance should be also re-
lated to the Fe2þ species, since the FeOx VB has
Fe 3d-derived states at ꢀ6.9 eV that present a
maximum in their resonance profile at hm ¼ 57 eV
[27].
(a)
(b)
1.8ML
3ML
5.4ML
3.3. Thermal annealing
It has been reported that annealing MTJÕs up
to ꢀ503 K strongly increases the magnetoresis-
tance performance of the device, being the Al2O3
barrier height and effective thickness also modified
[30,31]. The origin of the annealing enhancing
effect is related to the modification of the interface
between the ferromagnetic electrode and the in-
sulating barrier layer [31]. However, for annealing
temperatures above 573 K, the magnetoresistance
performance is degraded. According to Keavney
et al. [31], this behaviour suggests that some dis-
order created during deposition is annealed out,
up to a certain optimal temperature, at which point
40
45
50
55
60
40
45
50
55
60
h (eV)
Fig. 4. CIS curves of (a) Fe 3d states at ꢀ0.8 eV, and (b) O 2p
derived states at ꢀ6.2 eV below EF for the Fe films 1.8, 3 and
5.4 ML thick deposited on Al2O3 at RT.

82
A. Arranz et al. / Surface Science 521 (2002) 77–83
Colaianni et al. [17] have studied, using AES
and high resolution electron energy loss spectro-
ꢁ
scopy, the thermal modification of the Fe(3.2 A)/
10ML Fe/Al2O3
(c)
h =150eV
Fe0
(a)
(b)
Al2O3
T (K)
RT
ꢁ
Al2O3(18.5 A) interface deposited on a Mo(1 1 0)
substrate in the temperature range 90–1200 K.
According to these authors, annealing up to 900 K
causes agglomeration of Fe leading to large clus-
ters, in such a way that the Al2O3 substrate is
reexposed in some areas after annealing. Such
process (Ostwald ripening) has been also observed
during annealing of Cu films deposited on TiO2
[32]. Since the diffusion coefficient of Fe on Al2O3,
D ꢀ 10À20 cm2 sÀ1 at ꢀ873 K [33], is too small to
explain the decrease of iron species and the in-
crease of the aluminium oxide species, respectively,
the above-mentioned agglomeration should be
responsible for the evolution of the Al 2p, Fe 3p
and VB spectra in Fig. 5. At ꢀ673 K, Fe begins to
agglomerate into larger clusters, being the Al2O3
substrate and the Fe2þ species formed during
the first stages of Fe deposition reexposed in some
areas. Above this temperature, annealing causes
further agglomeration of iron in such a way that
the Fe2þ species are destroyed and Al0 reduced
species are formed in Fe-depleted areas. It should
be pointed out that the AES results of Colaianni
et al. [17] do not show the reduction of the Al2O3
oxide in their Fe/Al2O3 interface. However, in our
case it exists a straightforward connection between
the elimination of Fe2þ species and the reduction
of the Al2O3 substrate in the Fe-depleted areas. A
rough estimation of the Al2O3 substrate reexposed
area calculated from the Al 2p signal intensity
gives ꢀ50% at 873 K. The above-mentioned
modification of the Fe/Al2O3 interface could ex-
plain the reduction in the magnetoresistance per-
formance of MTJ devices with annealing above the
optimal temperature.
RT
673
773
RT
Fes
Fe2+
673
673
773
873
773
873
Al0
873
78 76 74 72
58 56 54 52 50
Binding energy (eV)
12
8
4
0
Fig. 5. Photoemission spectra measured with hm ¼ 150 eV of (a)
Al 2p band, (b) Fe 3p band, and (c) VB, respectively, of a 10
ML Fe film on Al2O3 measured after annealing during 15 min
at 673, 773 and 873 K. Spectra measured after deposition at RT
are also given for comparison.
other processes begin to reduce the spin-polarised
current.
In order to get more insight into the processes
taking place at the interface during annealing
above the optimal temperature (ꢀ503 K), we have
studied the thermal stability of the Fe/Al2O3 in-
terface in the temperature range of 673–873 K.
Fig. 5 shows (a) the Al 2p, (b) the Fe 3p and (c) the
VB spectra measured with hm ¼ 150 eV of an iron
film 10 ML thick deposited on Al2O3. Measure-
ments were carried out after annealing during 15
min at 673, 773 and 873 K. Spectra measured at
RT are also given for comparison. The deconvo-
lution of the Fe 3p spectra is illustrated in Fig.
5(b). With increasing temperature, the Al 2p and
O 2p VB states related to the Al2O3 substrate in-
crease considerably. Also a decrease of the Fe 3p
and Fe 3d VB intensities is observed. It is impor-
tant to note that Fe2þ species appears with an-
nealing at 673 K. Above this temperature, Fe2þ
species disappear, along with the formation of Al0
reduced species in the low binding energy side of
the Al 2p band. The annealing also produces a
shift of ꢀ )1 eV of the Fes peak, therefore sug-
gesting an increase of the atomic coordination of
the outer Fe surface atoms probably due to the
agglomeration of Fe.
4. Conclusions
SXPS and RPE have been used to study the
growth and electronic properties of Fe ultrathin
films deposited on Al2O3 substrates at RT. A SM
growth mode has been found for Fe growth. For
iron coverages below 1 ML, Fe2þ species are
formed at the Fe/Al2O3 interface, followed by the

A. Arranz et al. / Surface Science 521 (2002) 77–83
83
[8] B. Ealet, E. Gillet, V. Nehasil, P.J. Møller, Surf. Sci. 318
(1994) 151.
[9] B. Ealet, E. Gillet, Surf. Sci. 367 (1996) 221.
formation of a metallic iron overlayer. Bonding of
Fe at very low coverages occurs by charge transfer
from Fe to surface oxygen atoms, whereas hy-
bridisation of Fe and Al states does not take place.
No reduction of the Al2O3 substrate upon iron
deposition at RT is observed.
€
[10] M. Baumer, J. Biener, R.J. Madix, Surf. Sci. 432 (1999)
189, and references therein.
€
[11] S. Andersson, P.A. Bruhwiler, A. Sandell, M. Frank, J.
Libuda, A. Giertz, B. Brena, A.J. Maxwell, M. Baumer,
€
ꢁ
H.J. Freund, N. Martensson, Surf. Sci. 442 (1999) L964.
The thermal stability of the interface has been
examined above the optimal annealing temperature
reported in the literature for MTJ devices. Results
are consistent with the Fe agglomeration to form
large clusters. As a consequence of Fe agglomera-
tion, some areas become fully Fe-depleted, in such
a way that Fe2þ species are eliminated and Al0 re-
duced species are formed in these areas. This an-
nealing behaviour of the Fe/Al2O3 interface could
explain the decrease in the magnetoresistance per-
formance of MTJ devices when annealed above
their optimal temperature.
[12] V. Di Castro, G. Polzonetti, S. Ciampi, G. Contini, O.
Sahko, Surf. Sci. 293 (1993) 41.
[13] U. Diebold, H. Tao, N.D. Shinn, T.E. Madey, Phys. Rev.
B 50 (1994) 14474.
[14] H. Chen, Q.Y. Xu, G. Ni, J. Lu, H. Sang, S.Y. Zhang,
Y.W. Du, J. Appl. Phys. 85 (1999) 5798.
[15] R. Schad, K. Mayen, J. McCord, D. Allen, D. Yang, M.
Tondra, D. Wang, J. Appl. Phys. 89 (2001) 6659.
[16] S.R. Teixeira, M.A.S. Boff, W.H. Flores, J.E. Schmidt,
M.C.M. Alves, H.C.N. Tolentino, J. Magn. Magn. Mater.
233 (2001) 96.
[17] M.L. Colaianni, P.J. Chen, J.T. Yates, Surf. Sci. 238 (1990)
13.
[18] H. Poppa, C.A. Papageorgopoulos, F. Marks, E. Bauer, Z.
Phys. D3 (1986) 279.
[19] A. Proctor, M.P.A. Sherwood, Anal. Chem. 54 (1982) 13.
Acknowledgements
€
[20] J. Biener, M. Baumer, R.J. Madix, P. Liu, E.J. Nelson, T.
Kendelewicz, G.E. Brown Jr., Surf. Sci. 441 (1999) 1.
[21] S. Thomas, P.M.A. Sherwood, Anal. Chem. 64 (1992)
2488.
A part of this work has been financially sup-
ported by the Large Scale Facilities of the Euro-
pean Community to LURE, project no. ES814-01.
One of us (A. Arranz) acknowledges financial
ꢀ
support from the Spanish Ministerio de Educacion
y Cultura.
ꢀ
[22] A. Arranz, V. Perez-Dieste, C. Palacio, Phys. Rev. B 66
(2002) 075420.
[23] C. Argile, G.E. Rhead, Surf. Sci. Rep. 10 (1989) 277.
ꢂ
[24] M.G. Barthes, A. Rolland, Thin Solid Films 76 (1981)
45.
[25] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interf. Anal. 17
(1991) 911.
References
[26] L.C. Davis, J. Appl. Phys. 59 (1986) R25.
[27] R.J. Lad, V.E. Henrich, Phys. Rev. B 39 (1989) 13478.
€ € €
[28] T. Kaurila, L. Saisa, J. Vayrynen, J. Phys. Condens.
Matter 6 (1994) 5053.
[1] C.T. Campbell, Surf. Sci. Rep. 27 (1997) 1, and references
therein.
[2] K.H. Johnson, S.V. Pepper, J. Appl. Phys. 53 (1982) 634.
[3] M.W. Finnis, J. Phys. Condens. Matter 8 (1996) 5811.
[4] C. Verdozzi, D.R. Jennison, P.A. Schultz, M.P. Sears,
Phys. Rev. Lett. 82 (1999) 799.
[29] H. Kato, T. Ishii, S. Masuda, Y. Harada, T. Miyano, T.
Komeda, M. Onchi, Y. Sakisaka, Phys. Rev. B 32 (1985)
1992.
[30] M. Sato, H. Kihuchi, K. Kobayashi, J. Appl. Phys. 83
(1998) 6691.
[5] F.S. Ohuchi, M. Kohyama, J. Am. Ceram. Soc. 74 (1991)
1163.
[31] D.J. Keavney, S. Park, C.M. Falco, J.M. Slaughter, Appl.
Phys. Lett. 78 (2001) 234, and references therein.
[32] A.K. See, R.A. Bartynski, Phys. Rev. B 50 (1994) 12064.
[33] O. Lenoble, J.F. Bobo, L. Hennet, H. Fischer, Ph. Bauer,
M. Piecuch, Thin Solid Films 275 (1996) 64.
[6] S. Varma, G.S. Chottiner, M. Arbab, J. Vac. Sci. Technol.
A 10 (1992) 2857.
[7] S.M. Mukhopadhyay, C.S. Chen, J. Vac. Sci. Technol. A
10 (1992) 3545.