Model reaction studies on vanadium oxide nanostructures on Pd(111)

Article abstract of DOI:10.1063/1.2336770

Deuterium desorption and reaction between deuterium and oxygen to water has been studied on ultrathin vanadium oxide structures prepared on Pd(111). The palladium sample was part of a permeation source, thus enabling the supply of atomic deuterium to the sample surface via the bulk. Different vanadium oxide films have been prepared by e-beam evaporation in UHV under oxygen atmosphere. The structure of these films was determined using low energy electron diffraction and scanning tunneling microscopy. The mean translational energy of the desorption and reaction products has been measured with a time-of-flight spectrometer. The most stable phases for monolayer and submonolayer VO _x are particular surface-V₂O₃ and VO phases at 523 and 700 K, respectively. Thicker films grow in the form of bulk V ₂O₃. The mean translational energy of the desorbing deuterium species corresponds in all cases to the thermalized value. Apparent deviations from this energy distribution could be attributed to different adsorption/desorption and/or accommodation behaviors of molecular deuterium from the gas phase on the individual vanadium oxide films. The water reaction product shows a slightly hyperthermal mean translational energy, suggesting that higher energetic permeating deuterium contributes with higher probability to the water formation.

Full text of DOI:10.1063/1.2336770

THE JOURNAL OF CHEMICAL PHYSICS 125, 074703 ͑2006͒
Model reaction studies on vanadium oxide nanostructures on Pd„111…
M. Kratzer
Institute of Solid State Physics, Graz University of Technology, Petersgasse 16, A-8010 Graz, Austria
S. Surnev and F. P. Netzer
Institute of Physics, Surface and Interface Physics, Karl-Franzens University Graz, A-8010 Graz, Austria
a͒
A. Winkler
Institute of Solid State Physics, Graz University of Technology, Petersgasse 16, A-8010 Graz, Austria
͑
Received 17 March 2006; accepted 20 July 2006; published online 18 August 2006͒
Deuterium desorption and reaction between deuterium and oxygen to water has been studied on
ultrathin vanadium oxide structures prepared on Pd͑111͒. The palladium sample was part of a
permeation source, thus enabling the supply of atomic deuterium to the sample surface via the bulk.
Different vanadium oxide films have been prepared by e-beam evaporation in UHV under oxygen
atmosphere. The structure of these films was determined using low energy electron diffraction and
scanning tunneling microscopy. The mean translational energy of the desorption and reaction
products has been measured with a time-of-flight spectrometer. The most stable phases for
monolayer and submonolayer VO are particular surface-V O and VO phases at 523 and 700 K,
x
2
3
respectively. Thicker films grow in the form of bulk V O . The mean translational energy of the
2
3
desorbing deuterium species corresponds in all cases to the thermalized value. Apparent deviations
from this energy distribution could be attributed to different adsorption/desorption and/or
accommodation behaviors of molecular deuterium from the gas phase on the individual vanadium
oxide films. The water reaction product shows a slightly hyperthermal mean translational energy,
suggesting that higher energetic permeating deuterium contributes with higher probability to the
water formation. © 2006 American Institute of Physics. ͓DOI: 10.1063/1.2336770͔
I. INTRODUCTION
different from that of bulk vanadium oxides. The most fre-
quently observed surface oxide shows a 2ϫ2 honeycomb
structure. This structure has been identified by DFT calcula-
tions as a surface vanadium oxide ͑s-V O ͒ phase and con-
Many modern technologies take advantage of the spe-
cific physical and chemical properties of nanostructured sur-
faces. In particular, the catalytic activity of surfaces depends
strongly on the chemical composition and the atomic scale
structure of the surface. In this context ultrathin oxide films
on metal supports have attracted considerable interest, be-
cause they represent a model system of an “inverse catalyst.”
Vanadium oxide nanostructures in the monolayer regime
have recently been studied extensively because they form a
variety of interesting nanostructures with varying oxidation
2
3
sists of a layer of two V atoms per unit cell located in three-
fold palladium hollow sites and three oxygen atoms above
5
V–V bridge sites. Exposing this surface to further oxygen
changes the LEED pattern to a 4ϫ4 structure. The structure
of this phase has also been clarified by STM and DFT cal-
culations and can be described by a surface oxide with V O
5
14
stoichiometry. In this case the oxide layer is oxygen termi-
nated at both the metal interface and the oxide surface and
contains V atoms in tetrahedral O coordination. This layer is
quite unstable and can easily be reduced to the so-called
zigzag structure, as observed with STM. From DFT calcula-
tions a structure model for the zigzag phase is derived which
1
–7
states.
For the model systems VO /Pd͑111͒ and
x
VO /Rh͑111͒ a number of experimental studies using low
x
energy electron diffraction ͑LEED͒, scanning tunneling mi-
croscopy ͑STM͒, x-ray photoelectron spectroscopy ͑XPS͒,
near-edge x-ray-absorption fine structure, and high-
resolution electron-energy-loss spectroscopy ͑HREELS͒
have been performed to reveal the electronic and geometric
structures of the individual oxides. Density functional theory
4
contains V O units. Further reduction by hydrogen yields
6
14
again the stable 2ϫ2 s-V O phase. In the case of higher
2
3
vanadium coverage ͑1 ML͒ other vanadium oxide surfaces
can be produced. The observed “flower LEED pattern” can
be ascribed to a structure of rotational domains of rectangu-
͑DFT͒ calculations have been carried out in this context and
a quite comprehensive understanding of the oxide structures
lar unit cells and the formal stoichiometry of this phase is
5
,6
3
is available.
Evaporation of 0.5 ML vanadium on a Pd͑111͒ surface at
VO . In addition, also a hexagonal oxide phase with formal
2
VO stoichiometry has been observed in the 1 ML range. For
2
−
7
5
23 K under oxygen pressure, typically 2ϫ10 mbar, leads
thicker oxide films ͑5 ML vanadium͒ a ͑ͱ3ϫ ͱ3͒R30° LEED
to a number of surface oxides with structures that are quite
pattern is found which can be described as being due to the
3
͑
0001͒ face of the corundum V O bulk phase. The reason
2
3
a͒Author to whom correspondence should be addressed. Fax: 43-316-873-
for this LEED pattern is that the lattice mismatch between
8463. Electronic mail: a.winkler@tugraz.at
the V O bulk lattice and the ͱ3 direction of Pd͑111͒ is only
2
3
0021-9606/2006/125͑7͒/074703/9/$23.00
125, 074703-1
© 2006 American Institute of Physics

074703-2
Kratzer et al.
J. Chem. Phys. 125, 074703 ͑2006͒
3
.7%. The ͑ͱ3ϫ ͱ3͒R30° LEED pattern of the vanadium
vacuum soldered with gold onto a high purity nickel cylin-
der. This cylinder was soldered onto a stainless steel disk,
which in turn was welded to a stainless steel tube for gas
inlet. The whole assembly was tightly covered by two con-
centric ceramic tubes containing a coil of molybdenum wire
for resistive heating up to 1000 K. Several layers of tantalum
foil were wrapped around the outer ceramics for radiation
shielding. A NiCr–Ni thermocouple for temperature mea-
surements was spot welded to the rim of the sample. The
oxide is therefore a pseudosuperstructure of the Pd͑111͒
surface.
In this work we investigate the reactions of V oxide/Pd
inverse catalyst surfaces by focusing on specific model reac-
tions taking place on various surface vanadium oxides on
Pd͑111͒ and we explore the stability of these ultrathin oxide
films under reaction conditions. In particular, we present ex-
perimental data on the desorption of deuterium and the reac-
tion of deuterium with oxygen to form water. For these stud-
ies we use a Pd͑111͒ sample which is part of a permeation
source, that allows the supply of deuterium ͑or hydrogen͒
atoms by permeation from the bulk phase to the surface. This
possibility opens new reaction channels to take place and
both the reaction kinetics ͑reaction rates͒ as well as the re-
action dynamics ͑energy distribution of the reaction prod-
ucts͒ may differ from those when deuterium ͑hydrogen͒ is
supplied in the conventional way from the gas phase. Stud-
ies of this reaction type are relevant for heterogeneous ca-
talysis, hydrogen storage, and solid fuel cell technology. The
main emphasis of this work is to study the translational en-
ergy distribution of desorbing deuterium and of the reaction
product water as a function of different vanadium oxide
structures on Pd͑111͒, and to analyze the stability of these
oxides under reaction conditions. The energy distribution has
been investigated with a time-of-flight spectrometer and the
permeation source was attached to a LN cooled sample
2
holder which allowed the positioning of the sample in
8
front of all necessary analytical devices. A continuous
permeation/desorption flux of deuterium could be achieved
with this device by applying a proper backpressure ͑typically
500–1500 mbars͒ and a convenient choice of the palladium
sample temperature ͑typically 523–700 K͒. For pressure
measurement in the gas inlet system of the permeation
source a piezoelectric membrane gauge was applied. A
Knudsen cell, used for calibration of the TOF spectrometer,
was also mounted on the manipulator.
8
,9
The Pd͑111͒ surface was cleaned prior to the deposition
of vanadium oxide by 600 eV argon ion sputtering at 300 K,
followed by annealing at 900 K. The cleanliness was
checked by AES and LEED. Vanadium oxide was prepared
by electron beam evaporation of vanadium onto the Pd
sample at 523 K in a background oxygen atmosphere of
2ϫ10− mbar. The vanadium flux was measured by a quartz
microbalance. The vanadium oxide layer is characterized by
the amount of vanadium on the surface, which is given in
monolayer equivalents ͑MLE͒. 1 MLE of vanadium oxide
on the Pd͑111͒ surface therefore is equivalent to 1.52
7
VO structures under reaction conditions have been verified
x
by LEED. In addition, STM and LEED investigations on
similarly prepared vanadium oxide films on a separate
Pd͑111͒ crystal have been performed to get a more complete
picture of the real surface structures being involved.
1
5
2
ϫ10 V atoms/cm .
The TOF spectra have been obtained by using a chopper
frequency of 200 Hz ͑yielding a pulse frequency of 400 Hz
due to the double slit chopper͒ for water and a chopper fre-
quency of 400 Hz ͑800 Hz pulse frequency͒ for deuterium.
The time delay and the slit opening gate function ͑which are
needed to obtain the fit temperature of the Maxwellians͒
were obtained from calibration with a Knudsen beam. For
data evaluation it has been taken into account that the mass
spectrometer is a density detector, which means that the mea-
sured signal is indirectly proportional to the velocity of the
molecules. More details on the data acquisition and the fit
II. EXPERIMENT
The reaction experiments were carried out in an ultra-
high vacuum ͑UHV͒ chamber with a base pressure of
−
10
10
mbar, equipped with a LEED optics, an Auger electron
spectrometer ͑AES͒, a multiplexed quadrupole mass spec-
trometer ͑QMS͒, and a time-of-flight ͑TOF͒ spectrometer. An
extractor ion gauge and the QMS, calibrated with the help of
a spinning rotor gauge, allowed the quantitative partial pres-
sure determination in the vacuum system. A calibrated stan-
dard sample ͑tungsten filament͒ for thermal desorption al-
lowed the quantitative measurement of deuterium desorption
rates, as well as the determination of the effective pumping
11
procedures can be found elsewhere.
1
0
speed in the vacuum chamber. Thus, quantitative desorp-
tion and reaction rate measurements were possible. The time-
of-flight spectrometer was used to determine the translational
energy distribution of desorbing products, such as deuterium
and water. The TOF spectrometer comprises two differen-
tially pumped UHV chambers, housing the chopper motor
and the QMS detector, respectively. A detailed description of
III. RESULTS
A. Stability of vanadium oxide on Pd„111…
The crucial question in the context of reaction studies on
modified or nanostructured surfaces concerns the real struc-
ture of the surface layer under reaction conditions. In this
work we have used LEED to follow the structure during
reaction and we have compared the corresponding LEED
patterns with those obtained at room temperature under
vacuum conditions. We have studied the stability of
vanadium oxides for 0.3, 1, and 5 MLE, respectively,
for different temperatures and different reaction conditions:
11
the TOF spectrometer can be found elsewhere. The STM
measurements have been carried out in a second UHV sys-
2
tem, at the University of Graz, also described elsewhere.
The Pd͑111͒ sample was part of a permeation source,
which allowed to study the deuterium or hydrogen desorp-
tion and water reaction in a continuous way. The high purity
palladium disk ͑10 mm in diameter, 1 mm thickness͒ was

074703-3
VO nanostructures on Pd͑111͒
J. Chem. Phys. 125, 074703 ͑2006͒
FIG. 1. LEED patterns for different 0.3 MLE VO structures on Pd͑111͒: ͑a͒
x
after preparation at 523 K, 4ϫ4 structure, E=67 eV; ͑b͒ during reaction of
deuterium with oxygen at 523 K, 2ϫ2 structure, E=60 eV; ͑c͒ during deu-
terium desorption at 700 K, VO structure, E=56 eV; ͑d͒ during reaction of
deuterium with oxygen at 700 K, 2ϫ2 structure, E=64 eV.
͑
a͒ permeation/desorption of pure deuterium yielding a typi-
−
7
cal D pressure of 8ϫ10 mbar and ͑b͒ permeation of deu-
2
terium and reaction with impinging oxygen ͑typically
1
−
6
ϫ10 mbar͒ to form water.
FIG. 2. STM images of various VO structures on Pd͑111͒: ͑a͒ 4ϫ4 V O
x
5
14
2
1
. 0.3 MLE vanadium oxide
2
͑
160ϫ160 Å , U=2.0 V, I=0.1 nA͒; ͑b͒ 2ϫ2 s-V O3 ͑78ϫ78 Å ,
2
2
U=0.04 V, I=1.0 nA͒; ͑c͒ “wagon-wheel” VO ͑70ϫ70 Å , U=0.1 V,
After the preparation of 0.3 MLE vanadium oxide at
2
I=1.0 nA͒; ͑d͒ rectangular VO ͑200ϫ200 Å , U=0.25 V, I=0.5 nA͒; ͑e͒
2
2
−
7
T=523 K in an oxygen atmosphere of 2ϫ10 mbar we ob-
served a 4ϫ4 LEED pattern ͓Fig. 1͑a͔͒. This pattern is very
unstable and changes within a few minutes to a 2ϫ2 struc-
ture. This is due to the fact that in our chamber the hydrogen
and/or deuterium residual partial pressure is rather high. The
bulk V O ͑0001͒ ͑200ϫ200 Å , U=−0.3 V, I=0.1 nA͒.
2
3
to a deuterium desorption flux of about 6ϫ1015 D /cm s
2
2
͑
at a pumping speed of 140 l/s, effective surface area of
2
0
.5 cm ͒. After the oxygen partial pressure in the chamber is
4
ϫ4 structure has been recently described as a rather open
−
6
increased to 1ϫ10 mbar ͑corresponding to an impinge-
ment rate of 2.7ϫ10 O /cm s͒, water is formed on the
network structure where the surface unit cell consists of 5 V
atoms and 14 O atoms, forming a honeycomb-like arrange-
ment. A high-resolution STM image of the 4ϫ4 structure is
shown in Fig. 2͑a͒. The 4ϫ4 phase reduces easily by
hydrogen. The stable structure which is quite resistant to
hydrogen at 523 K is the 2ϫ2 structure ͓Fig. 1͑b͔͒, which
has also been described in detail in a previous paper; see
Fig. 2͑b͒ for a STM image. In this case a V O surface oxide
is formed that also exhibits a honeycomb-like structure.
Since at a vanadium coverage of 0.3 MLE the s-V O layer
covers only about 60% of the surface one is confronted with
1
4
2
2
vanadium oxide covered Pd͑111͒ surface which desorbs im-
mediately at this temperature. As the amount of permeating/
desorbing deuterium is much larger than the amount of im-
pinging oxygen, still a considerable amount of permeating
deuterium desorbs associatively in the form of D2, in addi-
tion to the formation of water with the impinging and ad-
sorbed oxygen. It is generally accepted that water formation
on Pd͑111͒ proceeds via the formation of hydroxyl
3
,5
2
2
3
2
3
1
2
intermediates. In this reducing atmosphere the 2ϫ2 struc-
ture is retained ͓Fig. 1͑b͔͒. Heating the 0.3 MLE vanadium
oxide surface to 700 K in vacuum or under reducing condi-
tions leads to a change of the LEED pattern ͓Fig. 1͑c͔͒. This
structure can be assigned to a surface oxide of VO stoichi-
ometry. Similar LEED patterns for vanadium oxide on
Pd͑111͒ and Rh͑111͒ and corresponding STM investigations
the coexistence of well ordered 2ϫ2 phases and bare
3
,4
Pd͑111͒ areas, as verified by STM measurements. Such a
surface layer is particularly interesting for reaction investiga-
tions because one could expect a special influence of the
island borders on the reaction. The 2ϫ2 structure remains
stable for deuterium permeation/desorption at 523 K for
extended time ͑many hours͒ at a D partial pressure of
8
and DFT calculations have described this structure as the
2
−
7
1,2,6,7
ϫ10 mbar during permeation. This pressure corresponds
“wagon-wheel” structure.
Figure 2͑c͒ displays a high-

074703-4
Kratzer et al.
J. Chem. Phys. 125, 074703 ͑2006͒
FIG. 4. LEED patterns of 5 MLE VOx structures on Pd͑111͒: ͑a͒ after
preparation at 523 K, bulk V O structure ͑ͱ3ϫ ͱ3͒R30°, E=57 eV;
2
3
͑
b͒ during reaction of deuterium with oxygen at 523 K, ͑ͱ3ϫ ͱ3͒R30°,
E=56 eV; ͑c͒ during deuterium desorption at 700 K, ͑ͱ3ϫ ͱ3͒R30°,
E=63 eV; ͑d͒ during reaction of deuterium with oxygen at 700 K,
͑
ͱ3ϫ ͱ3͒R30°, E=42 eV.
oxide surface to 700 K in vacuum leads to a LEED pattern
similar to that in Fig. 1͑c͒, which is again indicative of the
wagon-wheel VO structure. Pure permeation of deuterium
does not change this pattern. Exposing this surface to oxygen
without permeation of deuterium leads to a LEED pattern
which is a superposition of the VO structure and the 2ϫ2
structure ͓Fig. 3͑d͔͒. However, exposing oxygen to the sur-
face when deuterium is permeated concomitantly, a change
in the LEED pattern appears which has not been observed so
far ͓Fig. 3͑e͔͒. In addition to the 2ϫ2 structure a new super-
structure appears which can be explained by the superposi-
tion of rotational and mirror domains of a rectangular struc-
ture. This structure is similar to the flowerlike pattern but
with a different size of the unit cell. A more detailed descrip-
tion of this phase will be given in the discussion. After the
oxygen exposure has been terminated and only deuterium
permeation still exists, the 2ϫ2 structure and the rectangular
structure disappear and the remaining pattern is again indica-
tive of the VO structure ͓Fig. 3͑f͔͒.
FIG. 3. LEED patterns for 1 MLE VOx structures on Pd͑111͒: ͑a͒ after
preparation at 523 K, “flower pattern,” E=39 eV; ͑b͒ during deuterium de-
sorption at 523 K, 2ϫ2 structure, E=40 eV; ͑c͒ during reaction of deute-
rium with oxygen at 523 K, 2ϫ2 structure, E=60 eV; ͑d͒ during oxygen
exposure at 700 K, 2ϫ2 structure+VO structure, E=58 eV; ͑e͒ during re-
action of deuterium with oxygen at 700 K, 2ϫ2 structure+rectangular
structure, E=57 eV; ͑f͒ after reaction at 700 K, VO structure, E=62 eV.
resolution STM image of the wagon-wheel phase. Exposing
this surface again to oxygen changes the LEED pattern back
to the 2ϫ2 structure ͓Fig. 1͑d͔͒. The background signal in
the LEED pattern is increased compared with Fig. 1͑b͒ due
to the Debye-Waller factor. Apparently, under these reaction
conditions the surface vanadium oxide layer s-V O is con-
2
3
served at 700 K.
2
. 1 MLE vanadium oxide
Evaporation of 1 MLE vanadium under oxygen atmo-
sphere at 523 K leads to a LEED pattern as shown in Fig.
3
. 5 MLE vanadium oxide
3
͑a͒. This “flowerlike” pattern has recently been described as
Evaporation of 5 MLE vanadium under oxidative condi-
being the superposition of diffraction from domains with
rectangular unit cells of VO stoichiometry and the 2ϫ2
V O structure. This is clearly apparent from the STM im-
tions at 523 K leads to a ͑ͱ3ϫ ͱ3͒R30° LEED pattern ͓Fig.
4͑a͔͒. This pattern has been identified as being due to diffrac-
2
3
1
tion from the ͑0001͒ plane of V O bulk phases. No palla-
2
3
2
3
age in Fig. 2͑d͒. After initiation of the deuterium permeation
the LEED pattern changes to the pure 2ϫ2 structure ͓Fig.
dium signal is seen in the AES, indicating a continuous oxide
film on the substrate. However, from the fact that the LEED
spots are rather broad in this case we conclude that the film
consists of quite small V O crystallites. The STM measure-
3
͑b͔͒. The VO structure is reduced to the more stable V O
2 2 3
13
surface oxide. Adding oxygen at 523 K retains the 2ϫ2
structure, but additionally a faint ringlike structure appears in
the LEED pattern ͓Fig. 3͑c͔͒. Heating the 1 MLE vanadium
2
3
ments on such surfaces produced under similar conditions
1
indicate a mean crystallite diameter of about 50–100 Å.

074703-5
VO nanostructures on Pd͑111͒
J. Chem. Phys. 125, 074703 ͑2006͒
FIG. 5. Time-of-flight spectra for deuterium desorbing from clean Pd͑111͒
at 523 and 700 K, respectively. The best fit temperatures for a Maxwellian
distribution are also given, indicating a thermalized desorption.
The STM image presented in Fig. 2͑e͒ shows the hexagonal
pattern of the corundum V O ͑0001͒ top facets of these
2
3
crystallites. The LEED structure remains stable at 523 K,
both under reducing conditions as well as under reaction
conditions ͓Fig. 4͑b͔͒. This is also true for a sample tempera-
ture of 700 K, without ͓Fig. 4͑c͔͒ and with ͓Fig. 4͑d͔͒ oxy-
gen exposure. It should be noted that after the reaction ex-
periment AES revealed a significant palladium signal, but the
LEED pattern still shows the ͑ͱ3ϫ ͱ3͒R30° structure, albeit
with increased spot size. Apparently, the crystallites have
changed their shape and size during the reaction.
FIG. 6. ͑a͒ Time-of-flight spectra for pure deuterium desorption ͑curve 1͒
and deuterium desorption during concomitant oxygen exposure ͑curve 2͒
from 0.3 MLE VOx on Pd͑111͒ at 523 K. ͑b͒ Time-of-flight spectrum of
D O resulting from the reaction of permeating D and impinging O on
B. Permeation of deuterium and reaction with O2
on the surface VO on Pd„111…
x
2
2
0
.3 MLE VO on Pd͑111͒ at 523 K.
Hydrogen or deuterium permeates easily through palla-
x
1
4
dium samples, even at relatively low temperature. It there-
fore offers a convenient way to expose a surface to atomic H
or D, which originates from the interior of the sample. One
could imagine that this permeating deuterium or hydrogen
reacts quite differently with other adsorbed species than mo-
lecular H or D impinging from the gas phase. This is also
fits to the data points is approximately 5% throughout the
experiment and will not be cited explicitly for all the indi-
vidual fit temperatures in the following.
2
2
1
. 0.3 MLE vanadium oxide
expected to hold for modified palladium surfaces, e.g., modi-
fied by chemical additives or by a nanostructuring of the
The TOF spectrum for deuterium from a Pd͑111͒ sample
9
surface. Surface vanadium oxides, which exhibit a large va-
covered with 0.3 MLE vanadium oxide at 523 K is shown in
Fig. 6͑a͒ ͑curve 1͒. The surface vanadium oxide film in this
particular case is characterized by the s-V O 2ϫ2 structure.
riety of different stoichiometries and form different nano-
structures on Pd͑111͒, are excellent model systems which
can be taken to study the permeation and desorption of deu-
terium and its reaction with impinging oxygen to water. We
have focused particularly on the translational energy distri-
bution of the desorption flux and the reaction products,
2
3
A least squares fit of a Maxwellian TOF distribution to the
experimental data points leads to T =481 K. Although this
fit
value is not far off the surface temperature of 523 K, this
decrease in the mean translational energy was reproducibly
measured within the margin of error. The TOF measurement
11
which have been measured by TOF spectroscopy. In Fig. 5
TOF spectra of permeating/desorbing deuterium from a clean
Pd͑111͒ surface at two different surface temperatures are de-
picted for comparison. Least squares fits to the data points
using Maxwellian distributions yield mean translational en-
−
6
of D during molecular oxygen exposure ͑1ϫ10 mbar͒
2
leads to a TOF spectrum as depicted in Fig. 6͑a͒ ͑curve 2͒.
As already outlined above not all deuterium atoms react with
oxygen to water. In addition, due to the much higher deute-
rium permeation rate compared to the oxygen impingement
rate, associative desorption of deuterium molecules still
ergies of the desorption flux, ͗E͘=2kT , where the fit tem-
fit
perature Tfit is close to the sample temperatures. This shows
that associative desorption of deuterium from Pd͑111͒ is
largely unactivated, as already described for hydrogen de-
takes place. The best fit temperature is now T =476 K. Ap-
fit
parently the TOF spectrum is not changed significantly. In
the latter case we have also measured the TOF distribution of
9
,15
sorption in previous work.
The error of the Maxwellian

074703-6
Kratzer et al.
J. Chem. Phys. 125, 074703 ͑2006͒
FIG. 8. Maxwellian fit temperatures for desorbing deuterium as function of
VO coverage on Pd͑111͒ at 523 K: ͑ᮀ͒ pure deuterium desorption and ͑᭡͒
x
deuterium desorption during oxygen exposure.
2
. 1 and 5 MLE vanadium oxides
The 1 MLE surface vanadium oxide at 523 K is charac-
terized by a 2ϫ2 structure, with additional spots in the
LEED pattern, resulting in the flowerlike pattern. From STM
measurements it is known that this surface consists of a full
layer of surface V O with additional islands of VO2
2
3
3
stoichiometry. This surface should exhibit only few border
lines between oxide islands and bare Pd͑111͒. We have car-
ried out pure deuterium desorption as well as deuterium
+
oxygen reaction TOF studies on this surface, both at 523
and 700 K. Since the shape of the spectra and the signal to
noise behavior are similar as in the above described circum-
stances we will not present all the individual TOF spectra but
quote only the determined fit temperatures. The data are
compiled, together with the data for the 0.3 and 5 MLE
surface vanadium oxides in Figs. 8–10. Pure deuterium de-
sorption from the 1 MLE surface at 523 K ͑LEED shows the
FIG. 7. ͑a͒ Time-of-flight spectra for pure deuterium desorption ͑curve 1͒
and deuterium desorption during concomitant oxygen exposure ͑curve 2͒
from 0.3 MLE VOx on Pd͑111͒ at 700 K. ͑b͒ Time-of-flight spectrum of
D O resulting from the reaction of permeating D and impinging O on
2
2
0
.3 MLE VO on Pd͑111͒ at 700 K.
x
the water reaction product ͓Fig. 6͑b͔͒. A best fit temperature
Tfit=544 K is obtained, indicating that water desorption is
close to thermal, but slightly hyperthermal.
The same experiments were performed at a sample tem-
perature of 700 K. The backpressure of deuterium in the
permeation source was changed in such a way that the per-
meation flux was approximately the same as for the
2
ϫ2 structure͒ yields a mean translational energy corre-
sponding to a temperature of T =461 K, which is again sig-
fit
nificantly smaller than the surface temperature ͑Fig. 8͒. In
the case of concomitant oxygen exposure the D TOF yields
2
523 K situation. The TOF spectrum for pure deuterium de-
sorption is depicted in Fig. 7͑a͒ ͑curve 1͒. A best fit yields
Tfit=735 K. From the LEED investigations we know that in
this case the surface consists of a surface vanadium oxide
with VO stoichiometry. After adding the appropriate oxygen
pressure the surface is changed to the 2ϫ2 vanadium oxide
structure with s-V O stoichiometry. Deuterium desorption
2
3
during oxygen impingement yields a TOF spectrum
which now indicates a significantly smaller value for the
mean translational energy of Tfit=629 K ͓Fig. 7͑a͒,
curve 2͔. Apparently, the specific oxide layers which are
stable at the individual reaction conditions influence the TOF
spectra. The corresponding water-TOF spectrum ͓Fig. 7͑b͔͒
FIG. 9. Maxwellian fit temperatures for desorbing deuterium as function of
yields a best fit temperature of T =754 K, again slightly
fit
VO coverage on Pd͑111͒ at 700 K: ͑ᮀ͒ pure deuterium desorption and ͑᭡͒
x
hyperthermal.
deuterium desorption during oxygen exposure.

074703-7
VO nanostructures on Pd͑111͒
J. Chem. Phys. 125, 074703 ͑2006͒
at 523 K can be approximated by a Maxwellian of Tfit
=
481 K. It would be tempting to ascribe this feature of a
“
cooled” deuterium flux to particular changes of the potential
energy surface ͑PES͒ for the desorption process. However,
careful examination of the experiment leads to a different
and quite simple explanation of this result. As outlined above
the deuterium permeation/desorption flux of about 6
1
5
2
ϫ10 D /cm s leads to a pressure increase in the chamber
2
−
7
of 8ϫ10 mbar ͑at a pumping speed of 140 l/s as deter-
1
0
mined with the desorption standard ͒. This pressure equals
an impingement rate of D2 on the surface of
6
1
4
2
ϫ10 D /cm s. This means that about 10% of the deute-
2
rium flux which leaves the surface originates from reflected
or adsorbed/desorbed deuterium. On the bare Pd͑111͒ surface
the sticking coefficient for hydrogen ͑deuterium͒ is quite
high ͓S =0.45 ͑Ref. 16͔͒. In this case the permeated/
0
recombined deuterium molecules and the adsorbed/desorbed
deuterium molecules are close to thermal, as observed ex-
perimentally ͑Fig. 5͒. In the case of the 0.3 MLE vanadium
oxide about 60% of the surface is covered by the oxide. We
assume that deuterium molecules impinging on the oxide
areas do not adsorb but are reflected without losing much of
their kinetic energy, corresponding to 300 K. This contribu-
tion to the total desorption flux yields on the average a mean
translational energy which is equivalent to a smaller tem-
perature than the corresponding sample temperature ͑Fig. 8͒.
For the 1 MLE oxide surface at 523 K, which is com-
FIG. 10. Maxwellian fit temperatures for desorbing D O as function of VOx
coverage on Pd͑111͒ at ͑᭡͒ 523 K and ͑᭺͒ 700 K.
2
a mean desorption temperature of T =455 K. The TOF
fit
spectrum for D O can be approximated by a mean desorption
2
temperature of 567 K ͑Fig. 10͒. The TOF measurements at
700 K yield the following best fits to the spectra: pure deu-
terium desorption ͑LEED shows VO structure͒, T =714 K;
fit
deuterium during oxygen exposure ͓LEED shows ͑2ϫ2͒
+
rectangular VO structure͔, T =649 K ͑Fig. 9͒; water de-
2 fit
sorption, T =727 K. Again, for 1 MLE vanadium oxide dif-
fit
posed of a full layer of 2ϫ2 s-V O and islands with the
ferent structures are stable under different reaction condi-
tions which have an influence on the TOF spectra.
2
3
rectangular oxide structure, no bare palladium is exposed and
the contribution of reflected molecules with 300 K Maxwell
temperature is even larger, as seen in Fig. 8. The result of
these considerations is that the mean translational energy dis-
tribution of permeating/desorbing deuterium, which is ther-
malized on the Pd͑111͒ surface, is also largely thermalized
for the vanadium oxide layer. In the case of concomitant
oxygen exposure the kinetic energy distribution of deuterium
is not changed significantly. The reaction product water is
also close to thermal, but slightly hyperthermal ͑by about
10%͒.
In the case of the 5 MLE vanadium oxide the structure is
characterized by the LEED pattern indicating bulk V O .
2
3
The desorption and reaction studies yield the following
results: At 523 K pure deuterium desorption gives Tfit
=
531 K, deuterium concomitant with oxygen yields a best fit
temperature T =523 K ͑Fig. 8͒, and the water TOF can be
fit
described by T =548 K ͑Fig. 10͒. At 700 K pure deuterium
fit
desorption gives T =779 K, deuterium concomitant with
fit
oxygen yields T =720 K ͑Fig. 9͒, and the water TOF can be
fit
described by T =744 K ͑Fig. 10͒.
fit
On the 5 MLE vanadium oxide at 523 K, which consists
of small islands of bulk vanadium V O , pure deuterium
2
3
IV. DISCUSSION
desorption and D desorption during water reaction are again
2
The 0.3 MLE vanadium oxide prepared under oxidizing
thermalized ͑Fig. 8͒. This is somewhat puzzling because this
would mean that impinging deuterium is fully thermalized
either by dissociative adsorption/recombinative desorption or
by inelastic scattering with a high accommodation coeffi-
cient. Apparently, the deuterium adsorption behavior on the
surface vanadium oxide and the bulk vanadium oxide is quite
different. Here the question of the influence of the surface
morphology at the accommodation process has to be consid-
ered. The morphology of the 5 MLE bulk-type V O ͓Fig.
conditions at 523 K forms a layer of 4ϫ4 V O which fully
5
14
covers the Pd͑111͒ surface ͓Figs. 1͑a͒ and 2͑a͔͒. However,
under our experimental conditions, where a quite high deu-
terium ͑hydrogen͒ partial pressure exists, the 4ϫ4 phase
quickly reduces to the 2ϫ2 s-V O phase. This phase covers
2
3
4
about 60% of the Pd͑111͒ surface and is stable at 523 K,
both under reducing conditions as well as under reaction
conditions ͑deuterium+oxygen͒. Note, however, that the re-
2
3
action conditions are also largely reducing conditions ͑p
2͑e͔͒ is quite rough at the atomic scale, compared to the
monolayer surface vanadium oxide ͓Fig. 2͑d͔͒. Multiple col-
lisions in the interaction region on the bulk V O surface
O
2
−
6
14
2
=
1ϫ10 mbar equals 2.7ϫ10 O /cm s impinging on
2
the surface, deuterium pressure increase during permeation
2
3
−
7
16
2
of 8ϫ10 mbar equals about 1.2ϫ10 D/cm s emanating
from the bulk͒. It is known from STM that at this tempera-
ture under reducing conditions the s-V O phase forms com-
may give rise to the more pronounced accommodation. We
will investigate this particular feature with more appropriate
adsorption techniques in the future.
The permeation experiments at 700 K yield some differ-
ences compared to the 523 K case. Under reducing condi-
2
3
4
pact two-dimensional islands on the bare Pd͑111͒ surface.
The TOF spectrum for deuterium desorbing from this surface

074703-8
Kratzer et al.
J. Chem. Phys. 125, 074703 ͑2006͒
ergy in the energy distribution react with higher probability
with the intermediate hydroxide species on the surface, to
form water.
Finally, we would like to address a special feature of the
1
MLE structure at 700 K under reaction conditions. In ad-
dition to the 2ϫ2 LEED pattern extra diffraction spots ap-
pear which have not been observed so far ͓Fig. 3͑e͔͒. These
additional spots can be fully explained by rotational and mir-
ror domains of a rectangular structure with similar surface
unit cell as that proposed for the flowerlike pattern. The
schematic LEED pattern together with a schematic geometric
model is depicted in Fig. 11. The direction of the short axis
of the superstructure is rotated with respect to the ͑01͒ direc-
tion of the substrate unit cell by ±6° and the length ratio of
the rectangular unit cell is 1:1.2. This is close to the ratio of
the unit cell axes of the flowerlike pattern. The only differ-
ence is a contraction of the reciprocal unit cell by about 10%,
equivalent to an expansion of the geometric unit cell by 10%.
This leads to a better matching of the atoms of the rectangu-
lar structure along the diagonal with respect to the ͗110͘
rows of the palladium substrate. On the other hand the direc-
tion of the diagonal is 3.5° off the ͗110͘ direction ͓Fig.
1
1͑b͔͒. Apparently, at 700 K the VO rectangular structure
2
relaxes to an energetically more favorable expanded struc-
ture. In this case a larger number of oxygen atoms can oc-
cupy on-top positions which have been shown to be the en-
3
ergetically most favorable ones.
V. SUMMARY AND CONCLUSION
Various vanadium oxides in different oxidation states
have been prepared on a Pd͑111͒ surface and their stabilities
under specific reaction conditions have been analyzed. The
palladium sample was part of a permeation source, thus
enabling the permeation of deuterium through the sample
onto the surface. The permeation/desorption of deuterium
from the surface oxide layer and the reaction of permeating
deuterium with impinging oxygen on the oxide layer has
been investigated, using LEED and TOF spectroscopy.
The main results of our study can be summarized as
follows.
FIG. 11. ͑a͒ LEED pattern of Fig. 3͑e͒ with calculated LEED spots and
rectangular reciprocal unit cells and ͑b͒ corresponding geometrical model.
tions, i.e., for pure deuterium desorption, the LEED patterns
indicate the existence of the VO structure, both for 0.3 and 1
MLE. In this case the mean translational energy distribution
of deuterium is close to thermal ͑Fig. 9͒. Following the con-
siderations made above leads to the conclusion that deute-
rium molecules impinging from the gas phase on the VO
surface leave the surface fully accommodated, either due to
adsorption/desorption or due to a large accommodation coef-
ficient for the inelastically reflected molecules. Under reac-
tion conditions, i.e., permeation of hydrogen during exposure
to oxygen, LEED reveals the existence of the 2ϫ2 s-V O
͑
1͒ The stable phase of a 0.3 MLE vanadium oxide at
5
23 K, during deuterium desorption as well as during
reaction between deuterium and oxygen, is the 2ϫ2
s-V O surface oxide. The experimentally obtained
2
3
2
3
phase, both for 0.3 and 1 MLE. On this vanadium oxide
layer with higher oxidation state again a cooled kinetic en-
ergy distribution is observed, due to the negligible sticking
coefficient for deuterium on this oxide phase. For the 5 MLE
case at 700 K the desorption features are similar to those for
the 523 K, case showing that the bulk vanadium oxide is less
mean translational energy for desorbing deuterium is
smaller than the thermal value. However, this is just
due to a considerable contribution of reflected ͑nonac-
commodated͒ deuterium molecules that are at room
temperature. The translational energy distribution of as-
sociatively desorbing deuterium itself is thermal.
influenced by temperature and oxygen pressure. The D O
2
desorption flux at 700 K is for all VO phases slightly hy-
perthermal, as for the 523 K case ͑Fig. 10͒. This behavior
suggests that those deuterium atoms with higher kinetic en-
͑2͒ The stable 0.3 MLE vanadium oxide at 700 K is the
“wagon-wheel” VO phase. TOF shows a thermal distri-
bution of deuterium from this surface, indicating
x

074703-9
VO nanostructures on Pd͑111͒
J. Chem. Phys. 125, 074703 ͑2006͒
that the reflected contribution is also thermal. The ad-
sorption and accommodation behaviors of impinging
deuterium are apparently quite different on the
s-V O surface and on the VO surface. Concomitant
ACKNOWLEDGMENT
This work has been supported by the Austrian Science
Fund in the framework of the Joint Research Project “Nano-
science on Surfaces.”
2
3
oxygen adsorption changes the vanadium oxide to
s-V O . The TOF spectrum is again “cooled” due to the
changed adsorption behavior of deuterium.
2
3
1
F. P. Leisenberger, S. Surnev, L. Vitali, M. G. Ramsey, and F. P. Netzer,
J. Vac. Sci. Technol. A 17, 1743 ͑1999͒.
S. Surnev, L. Vitali, M. G. Ramsey, F. P. Netzer, G. Kresse, and J.
͑
͑
͑
3͒ For the 1 MLE VO scenario more different stable ox-
x
2
ide phases exist as a function of temperature and deu-
terium and/or oxygen pressures. However, the main
feature is again that the different oxides existing at 523
and 700 K exhibit different adsorption behaviors for
deuterium, resulting in different apparent TOF distribu-
tions.
Hafner, Phys. Rev. B 61, 13945 ͑2000͒.
S. Surnev, G. Kresse, M. Sock, M. G. Ramsey, and F. P. Netzer, Surf. Sci.
495, 91 ͑2001͒.
S. Surnev, M. Sock, G. Kresse, J. N. Andersen, M. G. Ramsey, and F. P.
Netzer, J. Phys. Chem. B 107, 4778 ͑2003͒.
C. Klein, G. Kresse, S. Surnev, F. P. Netzer, M. Schmid, and P. Varga,
Phys. Rev. B 68, 235416 ͑2003͒.
J. Schoiswohl, M. Sock, S. Eck, S. Surnev, M. G. Ramsey, and F. P.
Netzer, Phys. Rev. B 69, 155403 ͑2004͒.
J. Schoiswohl, S. Surnev, and F. P. Netzer, Top. Catal. 36, 91 ͑2005͒.
G. Pauer, M. Kratzer, and A. Winkler, Vacuum 80, 81 ͑2005͒.
G. Pauer, M. Kratzer, and A. Winkler, J. Chem. Phys. 123, 204702
͑2005͒.
A. Winkler, J. Vac. Sci. Technol. A 5, 2430 ͑1987͒.
C. Eibl and A. Winkler, J. Chem. Phys. 117, 834 ͑2002͒.
T. Mitsui, M. K. Rose, E. Fomin, D. F. Ogletree, and M. Salmeron, Surf.
Sci. 511, 259 ͑2002͒.
S. Surnev, G. Kresse, M. G. Ramsey, and F. P. Netzer, Phys. Rev. Lett.
87, 086102 ͑2001͒.
J. Völkl and G. Alefeld, in Hydrogen in Metals I, edited by J. Völkl and
G. Alefeld ͑Springer, Berlin, 1978͒, p. 321.
G. Comsa and R. David, Surf. Sci. Rep. 5, 145 ͑1985͒.
M. Beutl, M. Riedler, and K. D. Rendulic, Chem. Phys. Lett. 247, 249
͑1995͒.
3
4
5
6
4͒ The 5 MLE vanadium oxide forms a bulklike V O .
2
3
Deuterium desorption from this surface yields thermal-
ized TOF spectra, both for 523 and 700 K. This shows
that the adsorption behavior for deuterium is different
on the surface vanadium oxide and the bulk vanadium
oxide. The influence of the different surface morpholo-
gies may be a possible reason.
7
8
9
1
0
11
1
1
1
2
3
4
5͒ The mean translation energy distribution of the water
reaction product is on the average slightly hyperther-
mal. A possible explanation for this behavior is that
those deuterium atoms which possess higher kinetic en-
ergy after permeation react more easily with the hy-
droxyl intermediate and channel their energy into the
reaction product.
15
1
6