Interfacial effects in ultra-thin nickel deposits on yttria-stabilized zirconia

Article abstract of DOI:10.1016/j.susc.2004.06.002

Ultra-thin films with 0.2 ML of Ni vapor-deposited in UHV near RT on yttria-stabilized-zirconia YSZ(100), were studied by X-ray photoelectron spectroscopy (XPS). Upon extensive annealing of the deposits in UHV between 480 and 850 K, it was found that there exists a critical temperature, below which the deposits tend to be fully oxidized via the substrate and above which they are stable in the metallic state. Nickel oxide formed on YSZ by either substrate or RT gas-phase oxidation is stable in UHV at least up to the critical temperature. This critical temperature appears to be sensitive to the history of the clean substrate treatments, as expressed by the O1s/Zr3d XPS intensity ratio, being higher than 850 K for the fully oxidized state of the clean YSZ surface and falling even below 635 K after repeated Ar+ sputtering cycles and extensive heating of the clean substrate in UHV at 870 K leading to an oxygen-deficient surface.

Full text of DOI:10.1016/j.susc.2004.06.002

Surface Science 566–568 (2004) 698–702
www.elsevier.com/locate/susc
Interfacial effects in ultra-thin nickel deposits on
yttria-stabilized zirconia
L. Sygellou, S. Ladas *
Department of Chemical Engineering, University of Patras and FORTH/ICE-HT, POB 1414, Gr-26504 Rion, Achaia, Greece
Available online 15 June 2004
Abstract
Ultra-thin films with 0.2 ML of Ni vapor-deposited in UHV near RT on yttria-stabilized-zirconia YSZ(1 0 0), were
studied by X-ray photoelectron spectroscopy (XPS). Upon extensive annealing of the deposits in UHV between 480 and
8
50 K, it was found that there exists a critical temperature, below which the deposits tend to be fully oxidized via the
substrate and above which they are stable in the metallic state. Nickel oxide formed on YSZ by either substrate or RT
gas-phase oxidation is stable in UHV at least up to the critical temperature. This critical temperature appears to be
sensitive to the history of the clean substrate treatments, as expressed by the O1s/Zr3d XPS intensity ratio, being higher
than 850 K for the fully oxidized state of the clean YSZ surface and falling even below 635 K after repeated Ar+
sputtering cycles and extensive heating of the clean substrate in UHV at 870 K leading to an oxygen-deficient surface.
ꢀ
2004 Elsevier B.V. All rights reserved.
Keywords: Yttrium; Zirconium; X-ray photoelectron spectroscopy; Nickel; Nickel oxides
1
. Introduction
heterogeneous catalysts and electrodes in electro-
chemical cells. Therefore, it is important to have a
fundamental understanding of the interaction of
metal ultra-thin films with the substrate, the ther-
mal stability of the interface and the effect of the
support treatment.
The Ni/YSZ system is interesting both for cat-
alytic and electrochemical applications and model
studies of ultra-thin Ni films on poly- and single-
crystalline YSZ have recently appeared [4–7].
Sub-monolayer amounts of Ni on single crystal
Zirconia attracts technological and scientific
attention for ceramic, optical, and thin film
applications. Undoped ZrO is not stable up to its
melting point and can crystallize into monoclinic,
tetragonal and cubic phases. The cubic is stabilized
at all temperatures by the addition of solutes, such
2
2 3
as CaO, MgO and Y O . Zirconia fully stabilized
with yttria (YSZ) has a good ionic conductivity
and is the most promising material as an electro-
lyte for oxygen sensors and solid oxide fuel cells
2
yttria-stabilized ZrO (YSZ) were found to be
(
SOFCs) [1–3]. Metal/YSZ systems, are used as
oxidized via the substrate by extensive annealing in
UHV at 580 K [6], whereas ultra-thin NiO films
reverted to metallic Ni upon annealing at tem-
peratures below the expected thermodynamic
dissociation limit for NiO [5]. The proposed
interpretation was that YSZ, via its built-in ion
*
Corresponding author. Tel.: +30-2610997631; fax: +30-
610993255.
E-mail address: ladas@iceht.forth.gr (S. Ladas).
2
0
039-6028/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.06.002

L. Sygellou, S. Ladas / Surface Science 566–568 (2004) 698–702
699
vacancies, can act both as a source and a sink of
reactive oxygen and this was supported experi-
source at an XPS-estimated deposition rate of
about 0.03 ML/min, with 1 monolayer (ML) cor-
responding to 2 · 10 atoms/cm .
15
2
mentally by comparison with the Ni/a-Al
tem, where no such effects were observed [7].
2
O
3
sys-
This work presents a systematic and more
extensive XPS study of the Ni/YSZ and NiO/YSZ
interfaces for sub-monolayer Ni amounts and their
behavior upon heat treatment in UHV, as well as a
correlation of the observed interactions with
published oxygen transport properties for YSZ [8].
3. Results and discussion
Fig. 1 shows a series of Ni2p XP spectra for 0.2
ML Ni deposited on clean YSZ (O1s/Zr3d ratio
0.78) at 340 K following: heating in UHV for
increasing periods of time at 635 K and then 30
min annealing intervals at 745, 780 and 870 K.
Even after 8 min of heating in UHV Ni has begun
to oxidize and the process is complete after 70 min
(spectrum d). The oxidation is evidenced by the
gradual replacement of the Ni2p3=2 peak at ꢁ853
eV, corresponding to dispersed metallic Ni, by a
new peak at 856.1 eV with a pronounced satellite
at ꢁ862 eV, corresponding to a thin NiO layer.
This substrate-induced oxidation is similar,
2
. Experimental
The experiments were performed in an ultra-
ꢀ
10
high vacuum chamber (base pressure 5 · 10
mbar), equipped with a Leybold LHS-12 hemi-
spherical electron energy analyzer and a dual
anode X-ray source for XPS. The non-mono-
chromatized AlKa radiation (1486.6 eV) was used
and the analyzer was operated at the constant DE
mode with 100 eV pass energy resulting in a ana-
2
lyzed area of 3 · 5 mm . Binding energy values
were referenced to the respective value of 182.6 eV
Ni2p
XPS
856.1eV
oxide
for Zr3d5=2 [5]. The YSZ(1 0 0) (9% Y
3
2
O
3
),
8
53.0eV
1
2 · 12 · 1 mm , substrate crystals (Mateck, Ger-
dispersed
Ni
many) were mounted on a Mo plate, which could
be heated up to 1170 K. The maximum tempera-
ture at the surface of the YSZ crystal was found in
a calibration experiment to be 870 K and the
temperatures reported here are all estimated sur-
face temperatures. The specimen was cleaned in
g
f
e
þ
situ by a few cycles of RT Ar sputtering (2 keV, 1
d
c
lA) and annealing at 635 K for at least 1 h in
order to remove carbon contamination. The O1s/
Zr3d XPS intensity ratio on the clean surface was
0
.77 ± 0.01, which is taken to approach the fully
b
a
oxidized state of the surface. Repeated sputtering
cycles and extensive heating of the clean YSZ in
UHV at the maximum temperature of 870 K re-
sulted in a gradual decrease of the O1s/Zr3d ratio
and a light gray coloration of the specimen bulk
indicating increased number of oxygen vacancies
9
00
890
880
870
860
850
840
BINDING ENERGY E B / eV
[
be restored to its initial O1s/Zr3d ratio by expo-
9]. The substrate surface, but not the bulk, could
Fig. 1. Ni2p XP spectra for 0.2 ML Ni/YSZ(1 0 0), following
heating at 635 K for: (a) 8 min, (b) 23 min, (c) 48 min, (d) 68
min and subsequent annealing for 30 min at: (e) 745 K, (f) 780
K, (g) 870 K. (Clean YSZ intensity ratio O1s/Zr3d ¼ 0.78.)
ꢀ
6
sure to 5 · 10 mbar gas phase oxygen (7200 L) at
35 K. Nickel was deposited on the clean surface
at ꢁ340 K from an electron beam evaporation
6

700
L. Sygellou, S. Ladas / Surface Science 566–568 (2004) 698–702
although much faster, to that previously observed
for sub-monolayer amounts of Ni/YSZ(1 0 0) after
then increases less steeply up to 850 K. Remark-
ably, when two of these sequences (at 635 and 685
K) were repeated with the specimen far from the
X-ray gun during heating, the time for complete
oxidation increased by about a factor of 15. Upon
subsequent 30 min annealings of the oxidized Ni
deposits up to 870 K in UHV far from the X-ray
gun, the oxide was in all cases stable, with no
traces of reduction.
8
h annealing at 580 K in UHV [6,7]. The oxide is
stable up to the maximum achievable temperature
of 870 K (spectrum g in Fig. 1), contrary to pre-
vious work, where NiO produced either via the
substrate or by gas-phase oxidation decomposed
at 700 K [5,7].
The experiment of Fig. 1 was repeated for 0.2
ML Ni on clean YSZ with O1s/Zr3d ratios
In order to understand the results shown in Fig.
2, it is necessary to refer to the well-known ion-
conducting properties of the substrate. Yttria-sta-
bilized-zirconia is an oxygen ion conductor,
0
.77 ± 0.01, by heating between 480 and 850 K in
UHV for increasing periods of time. During
heating the specimen was kept in front of the X-
ray gun and XPS data were taken periodically to
follow oxidation. Following complete oxidation,
the specimen was subjected to 30 min intervals of
annealing up to 870 K. Fig. 2 shows the time re-
quired for complete oxidation of 0.2 ML of Ni
versus temperature. Oxidation time decreases
sharply, as temperature increases from 480 to 560
K, goes through a shallow minimum at 635 K and
2ꢀ
whereby the movement of mobile O through the
crystal lattice occurs by thermally activated hop-
ping through a network of built-in oxygen ion
vacancies [8,10]. Upon electrically biasing the
2ꢀ
crystal, the movement of O ions produces an ion
current, provided the temperature is high enough
(practically above several hundred Kelvin) to ob-
tain significant ionic conductivity. The latter in-
creases exponentially with temperature. In the
2ꢀ
absence of electric bias, the mobile O ions can
self-diffuse within the lattice, thus transporting
oxygen where it can be consumed [8]. The obser-
vation that the oxidation time is much shorter
when the specimen is in front of the X-rays (Fig.
2
2
1
1
1
1
1
200
000
800
600
400
200
000
annealing infront of X-Ray gun
annealing far from X-Ray gun
2), that is the oxygen transport to the surface of
YSZ is enhanced has, to our knowledge, not been
reported before. We attribute this enhancement to
the positive charging of the surface, as a result of
photoemission and to the neutralization of this
charge by an ion current flowing between the side
of the YSZ crystal in contact with the grounded
heating plate and that of the Ni-deposited surface.
When the ionic conductivity is small (e.g. at 480 K)
it is difficult to supply the necessary ion current,
the specimen surface is electrostatically charged (as
observed in XPS) and part of the neutralization is
achieved by collection of stray electrons, thus the
8
00
00
00
00
0
6
4
2
2ꢀ
transported O ions are few. As the temperature
increases towards 635 K, the ionic conductivity
and the ion current sharply increase, thus
explaining the sharp decrease of the oxidation time
in front of the X-rays (Fig. 2, r). The ion current
saturates to some value, which is expected to be of
4
00
500
600
700
800
900
Temperature (K)
Fig. 2. Time for complete oxidation of Ni deposits (ꢁ0.2 ML)
versus temperature of annealing in UHV in front (r) and far
2
the order of a few tens of nA/cm for typical
conditions of an XPS experiment. A simple cal-
(
ꢂ
) from the X-ray gun.

L. Sygellou, S. Ladas / Surface Science 566–568 (2004) 698–702
701
2
1
ꢀ3
culation for the oxidation time (70 min) at 635 K,
demonstrates that in order to fully oxidize 0.2 ML
the order of 10 cm and that the effect of va-
cancy association [8] can severely limit the number
of free vacancies contributing to diffusion, we have
a reasonable agreement with the experiment. The
mild increase of oxidation time from 635 to 850 K,
although both the ionic conductivity and the dif-
fusion coefficient increase further, is not straight-
forward to explain. Most likely, it is the result of
clean-off reactions with the UHV background
14
2
or ꢁ4 · 10 Ni atoms/cm the necessary current of
ions is about 30 nA/cm . On the other hand,
far from the X-rays, the oxygen must be supplied
2ꢀ
2
O
2
ꢀ
by thermal diffusion of O from the bulk towards
the surface. If a simple model is adopted, whereby
2
ꢀ
O
are supplied by diffusion from a bulk source of
ꢀ
3
constant concentration C
face barrier [11], the time t to supply N (cm ) O
0
(cm ) across an inter-
2ꢀ
ꢀ
2
2
(mainly H ), which increasingly consume oxygen
2
ions is given by t ¼ ðp=DÞðN=C
0
Þ , where D is the
transported to the surface.
2
ꢀ
O
YSZ (9% Y
self-diffusion coefficient, which at 635 K in
ꢀ
In order to explain the observed discrepancy
with previous results [5,7], concerning the thermal
stability of the oxide upon annealing up to 870 K,
we investigated the influence of possible YSZ
surface changes. It was found that, upon heating
at 870 K in UHV for 7 h, following sputtering, the
O1s/Zr3d ratio of the clean surface had dropped to
0.67. However, the shape of both the Zr3d, Y3d
11
2
ꢀ1
2
O
3
) is about 1 · 10
cm s [8]. A
simple calculation for the oxidation time (1100
min) at 635 K (Fig. 2, ) demonstrates that in
2
ꢂ
1
4
order to supply ꢁ4 · 10 O atoms/cm , the nec-
1
8
ꢀ3
essary value of C
0
is ꢁ10 cm . This value must
represent the concentration of free oxygen vacan-
cies in YSZ. Taking into account that the maxi-
2 3
mum number of vacancies in YSZ (9% Y O ) is of
Fig. 4. Ni2p XP spectra for ꢁ0.2 ML Ni/YSZ(1 0 0), following
heating at 635 K for: (a) 8 min, (b) 23 min, (c) 48 min, (d) 70
min and subsequent annealing for 30 min at: (e) 800 K, (f) 850
K, (g) 870 K. Prior to Ni deposition the substrate was exposed
Fig. 3. Ni2p XP spectra for ꢁ0.2 ML Ni/YSZ(1 0 0), following
heating at 635 K for: (a) 8 min, (b) 223 min, then: (c) exposure
to 3600 L O
d) 745 K, (e) 780 K, (f) 850 K, (g) 870 K. (Clean YSZ intensity
ratio O1s/Zr3d ¼ 0.67.)
2
at 340 K and subsequent annealing for 30 min at:
(
2
to 7200 L O at 635 K. (Clean YSZ intensity ratio O1s/
Zr3d ¼ 0.76.)

702
L. Sygellou, S. Ladas / Surface Science 566–568 (2004) 698–702
and O1s peaks did not exhibit detectable changes
by XPS, indicating loss of oxygen without exten-
sive substrate reduction. Repeating the experiment
of Fig. 1 on this altered surface, produced the re-
sults shown in Fig. 3. After heating 0.2 ML Ni for
4. Conclusions
There exists an upper critical temperature for
oxidizing 0.2 ML of Ni on YSZ by heating in
UHV, which is very sensitive to the O1s/Zr3d XPS
intensity ratio of the clean surface. With a ratio of
0.77, the critical temperature is above 850 K,
whereas on an oxygen-deficient surface with a
ratio of 0.67, resulting from extensive heating of
YSZ in UHV, the critical temperature drops be-
tween 580 and 635 K. Experiments still in progress
indicate that the critical temperature is around 820
K for a ratio of 0.74.
The O1s/Zr3d ratio also affects the thermal
stability of NiO on YSZ, as with a ratio of 0.77 a
sub-monolayer oxide film remains stable at least
up to 850 K, whereas a ratio of 0.67 destabilizes
the oxide film around 750 K. Thus, previous re-
sults [5], whereby the oxide started to decompose
at 700 K, must be accounted for by an even more
oxygen-deficient YSZ surface.
2
(
23 min at 635 K in front of the X-ray gun
spectrum b) the deposit remained metallic, in
contrast to Fig. 1, with only a small intensity de-
crease of the Ni2p peaks due to coalescence within
the film. At 580 K, however, the deposit could be
oxidized in agreement with previous work [6]. In
order to investigate the Ni oxide stability on the
altered surface, the deposit was oxidized by expo-
ꢀ
6
sure to 5 · 10 mbar gas phase oxygen (3600 L at
340 K) and then subjected to 30 min annealing at
ꢁ
increasing temperatures up to 870 K. After 30 min
at 745 K, the oxide starts to decompose to metallic
Ni and at 850 K it has been completely reduced. It
is clear that this behavior is very close to that
previously observed [5,7].
The original state of the clean YSZ surface
ꢀ
6
could be regenerated by exposure to 5 · 10 mbar
gas phase oxygen (7200 L) at 635 K, whereby the
O1s/Zr3d ratio increased again to 0.76. The same
oxygen treatment at ꢁ340 K did not cause signif-
icant increase of the O1s/Zr3d ratio. Fig. 4 shows
the results of repeating the experiment of Fig. 1 on
the regenerated surface. Upon heating 0.2 ML of
Ni in front of the X-ray gun in UHV at 635 K, Ni
oxidizes completely after 70 min, as in Fig. 1. The
oxide formed is stable up to 800 K and it starts to
decompose to metallic Ni only after heating at 870
K (spectrum g, Fig. 4). This is probably due to
somewhat incomplete regeneration of the inner
layers of the substrate surface under the conditions
used.
References
[1] T.H. Etsell, S.N. Flengas, Chem. Rev. 70 (1970) 339.
[
[
[
[
[
2] S.P.S. Badwal, Solid State Ionics 52 (1992) 23.
3] N.Q. Minh, J. Am. Ceram. Soc. 76 (3) (1993) 563.
4] D. Sotiropoulou, S. Ladas, Surf. Sci. 408 (1998) 182.
5] S. Zafeiratos, S. Kennou, Surf. Sci. 482–485 (2001) 266.
6] S. Zafeiratos, S. Kennou, Surf. Sci. 532–535 (2003) 402.
[7] L. Sygellou, S. Zafeiratos, N. Tsud, V. Matolin, S.
Kennou, S. Ladas, Surf. Interf. Anal. 34 (2002) 545.
[8] P.S. Manning, J.D. Sirman, R.A. De Souza, J.A. Kilner,
Solid State Ionics 100 (1997) 1.
[9] Q. Ge, P.J. Moller, Thin Solid Films 254 (1995) 10.
10] F. Hund, Z. Electrochem. 55 (5) (1951) 363.
[
[11] J.F. Mojica, L.L. Levenson, Surf. Sci. 59 (1976) 447.