Vanadium oxide overlayers on rhodium: Influence of the reduction temperature on the composition and on the promoting effect in CO hydrogenation

Article abstract of DOI:10.1006/jcat.2002.3594

Vanadia adlayers on a Rh surface promote the hydrogenation of CO. This promotion was investigated with particular emphasis on the effect of the oxidation state of vanadium. Inverse model catalysts were prepared by UHV deposition of submonolayer amounts of metallic vanadium on a Rh foil in an oxygen atmosphere. After oxidation at 673 K the catalyst was reduced in hydrogen under specified conditions and the surface composition was determined by Auger electron spectroscopy. The kinetics of the reaction were then measured at 573 K and atmospheric pressure. The reaction rate is always increased with respect to the bare Rh surface but depends strongly on the reduction temperature. After reduction up to 673 K the observed rate increase can be correlated to the existence of two VO_x adlayer phases of different oxygen content. Enhanced CO dissociation at the perimeter sites of the VO_x islands leads to enhanced initial hydrogenation rates but also to enhanced deactivation of the catalyst. Raising the reduction temperature above 773 K results in the formation of metallic V, which is partially dissolved in the bulk, resulting in a V/Rh subsurface alloy with a Rh-terminated surface. This surface is particularly active for CO hydrogenation and less susceptible to deactivation by carbon deposition.

Full text of DOI:10.1006/jcat.2002.3594

Journal of Catalysis 208, 422–434 (2002)
doi:10.1006/jcat.2002.3594
Vanadium Oxide Overlayers on Rhodium: Influence of the Reduction
Temperature on the Composition and on the Promoting
Effect in CO Hydrogenation
Wolfgang Reichl and Konrad Hayek1
Institut fu¨r Physikalische Chemie, Leopold-Franzens-Universita¨t Innsbruck, Innrain 52a, A-6020 Innsbruck, Austria
Received December 10, 2001; revised February 26, 2002; accepted February 26, 2002
in model studies on noble metal–titania systems (2–6).
Blocking of metal sites by migrating overlayers must be
taken into account whenever mobile suboxides are formed
upon reduction. No evidence of decorating layers has been
obtained after reduction of ceria-supported noble metals
under comparable conditions, which has led to the conclu-
sion that the inhibiting effects observed on ceria and other
refractory oxides are merely “electronic” (7).
On the other hand, certain reactions will benefit from the
interaction between metal particles and supporting oxide
due to an increased number of noble metal–oxide interface
sites with high reaction probability. For instance, the CO hy-
drogenation on noble metal particles is generally promoted
by transition metal oxides (8), and likewise by deposition
of oxide submonolayers on a noble metal surface (9). In a
number of studies Somorjai and coworkers (10–14) inves-
tigated this reaction on polycrystalline Rh surfaces partly
covered with different reducible oxides. They concluded
that mainly sites at the perimeter of the two-dimensional
oxide islands are involved in the promotion. Although the
importance of these sites is generally recognised, the ori-
gin of the promotion is not completely clear. Boffa et al.
(14) observed a connection between the promotional ef-
fect and the Lewis acidity of the oxide in its highest valence
state.
Vanadia adlayers on a Rh surface promote the hydrogenation of
CO. This promotion was investigated with particular emphasis on
the effect of the oxidation state of vanadium. Inverse model cata-
lysts were prepared by UHV deposition of submonolayer amounts
of metallic vanadium on a Rh foil in an oxygen atmosphere. After
oxidation at 673 K the catalyst was reduced in hydrogen under
specified conditions and the surface composition was determined
by Auger electron spectroscopy. The kinetics of the reaction were
then measured at 573 K and atmospheric pressure. The reaction
rate is always increased with respect to the bare Rh surface but
depends strongly on the reduction temperature. After reduction
up to 673 K the observed rate increase can be correlated to the
existence of two VOx adlayer phases of different oxygen content.
Enhanced CO dissociation at the perimeter sites of the VOx islands
leads to enhanced initial hydrogenation rates but also to enhanced
deactivation of the catalyst. Raising the reduction temperature
above 773 K results in the formation of metallic V, which is par-
tially dissolved in the bulk, resulting in a V/Rh subsurface alloy
with a Rh-terminated surface. This surface is particularly active
for CO hydrogenation and less susceptible to deactivation by carbon
c
deposition.
ꢀ 2002 Elsevier Science (USA)
Key Words: CO hydrogenation; vanadium oxide; V/Rh subsur-
face alloy; model catalyst; metal-support interaction.
1. INTRODUCTION
However, one must anticipate that the interaction is de-
termined by the actual valence state of the transition metal
during the reaction. The morphology and composition of
reducible oxides are very sensitive to the experimental
conditions during deposition (e.g., oxygen partial pressure
and temperature). Furthermore, restructuring usually oc-
curs during annealing and oxidation–reduction procedures
applied for catalyst activation and regeneration. Vanadium
oxides tend to change structure and composition very eas-
ily. This has also been verified in a number of studies of
VOx deposits on metal surfaces (10, 13, 15). In recent STM
and LEED studies Surnev and coworkers (16–19) observed
that on Pd(111) a variety of surface structures of ultra-
thin vanadium oxide films are formed at varying coverage
and temperature. The stoichiometry and morphology of the
Reducible transition metal oxides like titania, ceria, and
vanadia have a strong effect on the catalytic properties of
group VIII metals which can be promoting or inhibiting,
depending on the reaction type and on the valence state of
the transition metal. High-temperature reduction (>773 K)
usually transforms the system into the “SMSI” state, char-
acterised by a strong decrease in the CO and hydrogen
chemisorption capacity and a decreased activity for skeletal
hydrocarbon reactions (1). Some of the observed phenom-
ena can be explained by a partial decoration of the noble
metal surface by substoichiometric oxide layers, as revealed
1 To whom correspondence should be addressed. Fax: +43 512 507 2925.
E-mail: konrad.hayek@uibk.ac.at.
422
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
423
low-coverage 2D phases depends on the preparation con- the chamber via a gas dosing manifold. The sample transfer
ditions. Over a wide range of temperature and composi- is achieved by means of a magnetically coupled transfer rod
tion a surface V2O3 layer is the energetically most stable with a Pyrex glass sample car acting also as sample holder
configuration (18, 19). Upon high-temperature annealing in the reaction cell. With this all-glass setup no wires or ther-
these vanadium oxide layers decompose and are converted mocouples are connected to the sample and background
to island structures of a V/Pd subsurface alloy with a activity is negligible. The main chamber is pumped by a
palladium-terminated surface. There are many indications turbomolecular pump, an ion getter pump, and a titanium
that layers of vanadium oxide resp. vanadium metal will sublimation pump to a base pressure in the low 10−10 mbar
behave similarly on palladium and on rhodium surfaces range.
(20–23).
A polycrystalline rhodium foil (0.125-mm thickness; pu-
The following questions now arise: Is it possible to quan- rity, 99.9%) and vanadium wire (1-mm ꢀ; purity, 99.996%)
tify an effect of the reduction temperature on the catalytic were obtained from Goodfellow. The rhodium foil was
activity of a vanadium oxide-covered rhodium surface, and cleaned by successive cycles of Ar ion bombardment
if so, is it related to chemical or structural properties of (500 eV, 0.8-µA sample current), oxidation (2.0×10−7 mbar
the overlayer? Furthermore, how does the transition from O2, T up to 1000 K), and annealing in hydrogen (2.0 ×
the oxide surface to the reduced (probably Rh terminated) 10−7 mbar H2, T up to 700 K) and in vacuum (at 1000 K)
surface affect the catalytic activity?
until no impurities were detected by AES. Vanadium ox-
Previously we could show that the promotion of the reac- ide overlayers were prepared by reactive evaporation of
tion on a vanadia-modified rhodium surface depends on the vanadium from a well-degassed electron beam evaporator
temperature of prereduction with hydrogen (20). In order in a background pressure of 2.0 × 10−7 mbar oxygen. After
to provide a more detailed answer to the above questions deposition the samples were oxidised in 2.0 × 10−7 mbar
we performed a quantitative study on the influence of hy- O2 at 673 K for 10 min. The deposition rate was monitored
drogen reduction on the composition of vanadium oxide by a quartz microbalance. Auger signal vs deposition time
deposits on rhodium and on its effect on the CO hydro- plots clearly indicated a layer-by-layer growth of the ox-
genation. Following the classical approach by the Somorjai ide. The vanadium oxide coverage is always given in mono-
group, “inverse” model catalysts have been prepared by layer equivalents (MLE) and normalised to a stoichiome-
UHV deposition of vanadium on a polycrystalline sub- try of V2O3. One MLE V2O3 corresponds to 9.5 × 1014
V
strate in an oxygen atmosphere. The submonolayer cov- atoms/cm2 (16).
erage range, with bare Rh and vanadium oxide patches co-
Both sides of the rhodium foil were exposed to the vana-
existing on the surface, is best suited to study metal–oxide diumfluxsothatthecatalystsurfaceareawasrelativelyhigh
interactions. Oxidative and reductive treatments (temper- (7.8 cm2). After preparation the samples were reduced at a
ature range between 300 and 1200 K and pressure range given temperature and hydrogen pressure (for 10 min un-
from 10−7 to 200 mbar) were applied to the samples prior less stated otherwise). Every preparation step was followed
to reaction. Changes in the surface composition were de- by Auger electron spectroscopy. The Auger spectra were
termined by AES and the catalytic activity was measured acquired at a 3-keV primary energy in the pulse-counting
in situ under exceptionally “clean” conditions (24). As in- mode, normalised with respect to the background level at
dicated, the properties of VOx overlayers on rhodium and the high-energy side of the respective peak, and differenti-
palladium surfaces will be quite similar. The results were ated electronically.
therefore related to those for the Pd(111) surface obtained
by Surnev and coworkers (16–19).
After characterisation the samples were transferred to
the reactor and exposed to the reactant mixture. The reac-
tants were circulated by means of a metal bellows pump.
Samples were taken periodically and analysed by gas chro-
matography with mass selective detection. Initial reaction
rates at 573 K were determined at low conversion from con-
version versus time plots. The turnover frequency (TOF)
was calculated with respect to the geometric area of the
rhodium substrate (7.8 cm2) assuming a Rh atom density
of 1.58×1015 atoms/cm2 (average of close-packed Rh(111)
and Rh(100) facets).
After every reaction the sample was oxidised in 2.0 ×
10−7 mbar O2 at 673 K for 10 min in order to remove
carbonaceous deposits. Vanadium oxide deposits were re-
moved by cycles of oxidation, Ar ion bombardment, and
annealing. Before every step the cleanliness of the sample
2. EXPERIMENTAL
The experiments were performed in an ultrahigh vac-
uum system combined with a high-pressure reaction cell
(HPC) described in detail in (24). This experimental setup
consists of a conventional UHV chamber with a long travel
z-manipulator and a small-volume Pyrex glass reactor
(60 ml) attached to the UHV chamber via a sample transfer
port. The UHV chamber is equipped with two electron
beam evaporators, a quartz microbalance, an electron
beam heater, an ion sputter gun, a mass spectrometer, and
an Auger spectrometer (DESA 100 Staib Instruments) ar-
ranged in three different levels. Gases are introduced into

424
REICHL AND HAYEK
was checked by AES and by running a CO hydrogenation exposure to hydrogen at 373 K, while the intensity and peak
reaction under “standard” conditions.
shapes of the vanadium signals at 468 and 435 eV (V468 and
435) are not effected by this reduction. In Fig. 3 the oxy-
V
gen loss upon reduction at 373 K (determined from the
decrease of the two Auger signals, O508 and O490) is shown
as a function of coverage. In the submonolayer range it de-
creases with increasing coverage of vanadium oxide, but
it is small (∼5%) and constant for coverages in excess of
1 MLE. This decrease is therefore due to the removal of
adsorbed O from the bare Rh patches, and the remaining
oxygen is associated with the vanadium oxide overlayer,
which in turn can be characterised by the ratio of vanadium
and oxygen signals (O508/V468). After reduction at 373 K
O508/V468 is 2.9 0.15 in the range of submonolayer cov-
erage, but it is notably smaller for coverages above 1 MLE
(e.g., it is 2.5 for Rh + 2.2 MLE V2O3). In Fig. 4 the depen-
dence of the Auger signals of Rh, V, and O on the reduction
temperature is shown. Raising the reduction temperature
in steps from 373 to 573 K does not lead to further changes.
However, increasing the temperature of hydrogen re-
duction (in 2.0 × 10−7 mbar H2) above 573 K results in a
markedly decreased oxygen signal and concomitantly a de-
crease of O508/V468 (Fig. 4; see also Figs. 1 and 2). In Fig. 3 it
is shown that in the submonolayer range O508 and O490 de-
crease by about 1/3 between reduction at 373 (or 573) and
673 K. Below 1 MLE coverage the reduction of the signal is
constant and must therefore be connected with a change in
theVOx stoichiometry. Afterreductionat673KO508/V468 is
1.9 0.1 compared to 2.9 after reduction at or below 573 K.
3. RESULTS
3.1. Composition (AES Data)
In a two-dimensional structure the Auger intensities are
directly proportional to the surface concentration, so that
their changes will reflect changes in the stoichiometry of
the overlayer. The surface composition of the sample was
monitored by AES after preparation, after subsequent re-
duction and after every reaction, and the changes of the
relevant oxygen and vanadia signals were recorded.
In Figs. 1 and 2 Auger spectra are shown for (Rh + 0.55
MLE V2O3) and (Rh + 0.89 MLE V2O3), respectively,
recorded before and after reduction in 2.0 × 10−7 mbar H2
for 10 min at different temperatures. (All reduction steps
were separated by oxidation in 2.0×10−7 mbar O2 at 673 K
for 10 min!)
After vanadium deposition and annealing in 2.0 ×
10−7 mbar O2 at 673 K oxygen remains adsorbed on the
bare rhodium patches, as confirmed by a study of the bare
rhodium surface. Therefore, the O508/V468 Auger signal ra-
tio measured after oxidation, which varies with VOx cover-
age, does not provide information about the vanadium ox-
ide stoichiometry. However, oxygen adsorbed on rhodium
is easily removed by exposure to hydrogen (2.0×10−7 mbar
at 373 K). Consistently, the oxygen signal decreases upon
FIG. 1. Auger spectra of the bare Rh surface (a), and of (Rh + 0.55 MLE V2O3) immediately after preparation (b) and after reduction at 373 (c),
673 (d), 753 (e), and 823 K (f). Reduction conditions: 2.0 × 10−7 mbar H2, 10 min; the reduction steps were separated by oxidation in 2.0 × 10−7 mbar
O2 at 673 K for 10 min.

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
425
FIG. 2. Auger spectra of (Rh+0.89 MLE V2O3): after preparation (a) and after reduction at 373 (b), 673 (c), and 1200 K (d). Reduction conditions:
2.0 × 10−7 mbar H2, 10 min (except 1 min in (d)); reduction steps separated by oxidation in 2.0 × 10−7 mbar O2 at 673 K for 10 min.
At the same time, the ratio of V430 to V468 has changed from (at or below 673 K) result in an identical Auger fingerprint
0.78 0.1 to 1 0.15. In excess of 1 MLE coverage only a and in the same O508 and V468 intensities. Furthermore, it is
slight decrease of the oxygen signals is observed upon re- noted that the hydrogen pressure during reduction, either
duction at 673 K, and O508/V468 is not significantly affected. 2.0 × 10−7 or 200 mbar, does not influence the composition
Thechangesdescribedsofararefullyreversiblebyoxida- of the resulting overlayer.
tion in 2.0×10−7 mbar O2 at 673 K. Two reductions at 373 K
Thus, it is concluded that under the given conditions sub-
separated by one or more oxidation and reduction cycles monolayers of vanadium oxide on polycrystalline Rh exist
FIG. 3. Oxygen loss of Rh+VOx between preparation and reduction
at 373 K (ꢀ) and between reduction at 373 and 673 K (ꢁ) as a function of
FIG. 4. (Rh + 0.55 MLE V2O3): Rh, V, and O Auger signals as a
initial vanadium oxide coverage. Reduction conditions: 2.0 × 10−7 mbar
function of the reduction temperature. Intensities are normalised to the
main Rh peaks of the bare Rh surface (Rh0). Reduction conditions: 2.0 ×
H2, 10 min, at indicated temperatures. The oxygen loss is determined from
10−7 mbar H2, 10 min; reductions separated by oxidation in 2.0×10−7 mbar
the intensity decrease of the O490 and O508 oxygen Auger signals upon
raising the reduction temperature.
O2 at 673 K for 10 min.

426
REICHL AND HAYEK
in two different phases, characterised by O508/V468 = 2.9
and 1.9, respectively. The transition between these two
phases occurs upon reduction in hydrogen between 623 and
673K. Itmustbenotedthatthisholdsonlyforthesubmono-
layer regime of vanadium oxide, while for higher coverage
the change in the O signals is small, indicating that the VOx
stoichiometry is much less affected.
Raising the reduction temperature above 723 K leads
to further significant changes in the Auger spectrum. In
this case not only the oxygen signals but also the vanadium
signals are altered (Fig. 1). Furthermore, the modifications
are no longer completely reversible by cycles of oxidation
and reduction.
After reduction above 723 K the oxygen signals O508 and
O490 diminish. On (Rh + 0.55 MLE V2O3) oxygen is re-
moved after reduction at 823 K, and V430 and V468 decrease
to about 1/3, as shown, for example, for the sample in Fig. 1f.
For a higher coverage (Rh+0.92 MLE) the changes of V468
and O508 after reduction at 823 K are shown in Fig. 5. In this
case the vanadium and oxygen signals are decreased, but
oxygen is not completely removed. A reduction at 873 K is
now necessary to remove oxygen entirely from the surface.
Concomitantly, V468 and V430 decrease again to about 1/3
of the initial value. In the intermediate temperature range,
from 753 to 773 K, O508 (and O490) as well as V468 and V435
are decreased, but oxygen is still present on the surface and
FIG. 6. Surface composition after HTR (2.0 × 10−7 mbar H2, 823–
873 K, 10 min): V468/Rh300 Auger signal ratio as a function of initial vana-
dium oxide coverage.
reduction to metallic vanadium must have occurred. (The
apparent O508/V468 < 0.2 is due to a vanadium peak at
510 eV.) At the same time the intensities of V468 and V430
are markedly decreased, which can be explained by diffu-
sion of vanadium into near-surface layers.
The ratio V468/Rh300 can be used to quantify the vana-
dium concentration within the information depth of AES
(approximately three to five layers). As shown in Fig. 6,
O
508/V468 is about 1 (Fig. 5).
Since after reduction above 823 K (“high-temperature
reduction” (HTR)) no oxygen is detected on the surface,
V468/Rh300 increases with the amount of vanadium oxide
initially present on the surface.
After reduction above 1100 K no V468 or V430 peaks
are detected, as shown in Fig. 2, after 1 min reduction
of (Rh + 0.89 MLE V2O3) at 1200 K. The same is ob-
served upon heating samples with lower coverage above
1100 K. Apparently, the vanadium concentration in the
near-surface region has vanished and vanadium is now com-
pletely dissolved in the rhodium bulk.
In Fig. 7 it is documented that vanadium segregates
from the subsurface region to the surface upon reoxida-
tion (2.0 × 10−7 mbar O2, 673–873 K, 10–30 min), forming
again an oxide overlayer. Auger spectra of (Rh + 0.55 MLE
V2O3) are shown after two reductions at 373 K (Figs. 7a and
7c) separated by a high-temperature reduction (Fig. 7b) (re-
ductionalwaysafteroxidationat673K). OxidisingtheHTR
sample does not reestablish the original vanadium cover-
age. V468 decreases by 10–15% and O508/V468 is now 2.7
compared to 2.9 on the original deposit. If we assume that
reduction at 373 K will always result in the same vanadium-
to-oxygen stoichiometry, part of the vanadium must have
remained dissolved in the rhodium bulk after oxidation.
However, the resegregation of most of the vanadium upon
oxidation, together with the fact that the Auger V468 signal
is still observed after HTR, supports the conclusion that af-
ter reduction, between 723 and 1000 K vanadium remains
FIG. 5. Auger spectra of (Rh + 0.92 MLE V2O3): after reduction at
373 (a), 773 (b), 823 (c), and 873 K (d). Reduction conditions: 2.0 ×
10−7 mbar H2, 10 min; reductions separated by oxidation in 2.0×10−7 mbar
O2 at 673 K for 10 min.

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
427
rate vs coverage, we obtain a volcano curve as described by
Boffa et al. (13). A further rise in the reduction tempera-
ture in steps to 573 K does not significantly affect the rates.
However, raising it from 573 to 673 K leads to a modest
but significant and reproducible rate increase. The relative
rate of methane formation for any reduction temperature
passes a maximum at a coverage of 0.5 0.05 MLE vanadia,
corresponding to a 2.7-fold activity enhancement after re-
duction at 673 K. Above 0.85 MLE coverage the rate drops
below that measured on the bare rhodium surface and the
difference between the two prereduction temperatures di-
minishes. Samples completely vanadium oxide covered are
less active than the bare rhodium surface by more than two
orders of magnitude.
The selectivity on clean and vanadium oxide-promoted
rhodium is summarised in Table 1. With increasing vana-
dium oxide coverage, the selectivity towards methane de-
creases from 97 to 55% (on Rh + 0.89 MLE V2O3), and
morehigheralkanesareformed. Atthesametimetheselec-
tivity towards unsaturated products (ethene and propene)
increases significantly. At constant coverage, a reduction at
673 K decreases the methane selectivity compared to that
after LTR, and increases the ratio of unsaturated to satu-
rated hydrocarbons. The formation of methanol and higher
alcohols is well-known under different conditions of tem-
perature and pressure (25, 26) but was not observed under
the present experimental conditions.
A further increase in the reduction temperature (HTR,
823–873 K) under 2.0 × 10−7 mbar H2 leads to another,
rather unexpected rate increase. In Fig. 9 the initial TOF
after HTR is shown as a function of the initial V2O3 cover-
age. A strong increase is observed for initial coverages up
to 0.5 MLE, at which point the TOF levels off, while the
rate enhancement is about sixfold compared to the bare
FIG. 7. Auger spectra of (Rh + 0.55 MLE V2O3) after sequential
reduction at 373 (a), 823 (b), and 373 K (c). Reduction conditions:
2.0 × 10−7 mbar H2, 10 min; reduction steps separated by oxidation in
2.0 × 10−7 mbar O2 at 673 K for 10 min.
mainly in the near-surface rhodium layers. After reduction
above 1100 K, no vanadium signals are observed, and vana-
diumdoesnotsegregatetothesurfaceevenuponprolonged
oxidation, which is consistent with complete dissolution in
the bulk.
3.2. CO Hydrogenation
The kinetic experiments were performed immediately af-
ter reduction in hydrogen at a given temperature and pres-
sure (“prereduction”) and after subsequent characterisa-
tion by AES. The rate of methane formation from a mixture
of 40 mbar CO and 120 mbar H2 (with He added to 1 bar)
on the unpromoted rhodium foil was 0.017 molecules/s·site
at 573 K, and no deactivation was observed over long re-
action times. Methane is the main product, with a selectiv-
ity of more than 96%. The reproducibility of the turnover
frequencies (TOFs) is 4%. The bare Rh surface was ex-
posed to the same gas exposures and treatments as the VOx -
covered samples, but no influence of pretreatment on the
activity was observed in this case.
The initial TOF for methane formation on vanadium
oxide-promoted rhodium is presented in Fig. 8 as a function
of vanadium oxide coverage for prereduction at and below
673 K. Again, the results are shown for the reaction of a
stoichiometric CO/H2 mixture (40 mbar CO and 120 mbar
H2, He added to 1 bar) at 573 K. Each reduction was per-
formed after oxidation in 2.0 × 10−7 mbar O2 at 673 K for
10 min and immediately prior to the reaction. After pre-
reduction at 373 K the initial rate of methane formation
FIG. 8. CO hydrogenation activity after LTR as a function of vana-
dium oxide coverage. Reaction conditions: 40 mbar CO, 120 mbar H2, He
added to 1000 mbar, T = 573 K. Prereduction: 2.0 × 10−7 mbar H2 at the
increases with respect to that on the bare surface. Plotting stated temperature (373 or 673 K), 10 min.

428
REICHL AND HAYEK
TABLE 1
Selectivity of (Rh + VOx) after LTR (Measured at Conversions below 2%)a
Selectivity (%)
TOF
Sample
(1/s · site)
Methane
Ethane
Ethene
Propane
Propene
Bare Rh
0.017
0.035
0.039
0.042
0.047
0.013
0.0006
97.5
79
68
72
64
2.5
9
0.24 MLE V2O3 (Tred = 373 K)
0.24 MLE V2O3 (Tred = 673 K)
0.55 MLE V2O3 (Tred = 373 K)
0.55 MLE V2O3 (Tred = 673 K)
0.89 MLE V2O3 (Tred = 373 K)
2.2 MLE V2O3 (Tred = 373 K)
2
8
3
7
15
28
5
4
4
6
2
5
7
7
11
15
18
13
14
12
13
14
55
40
a Reaction conditions: 40 mbar CO, 120 mbar H2, He added to 1 bar, T = 573 K. Reduction in 2.0 × 10−7 mbar H2 for
10 min.
Rh surface. The TOFs and selectivities for HTR samples reversible if the reduction temperature is 673 K or below:
are summarised in Table 2. After high-temperature reduc- After two low-temperature reductions (≤573 K) which are
tion the selectivity towards methane is between 96 and 93% separated by one or more cycles of oxidation and reduction
and does not depend significantly on the initial coverage.
at 673 K, identical conversion vs time plots are obtained.
Finally, raising the reduction temperature further, up to This is also demonstrated in Fig. 11, where the initial TOFs
≥1100 K, leads to a strong activity decline (Fig. 9). The se- of three samples with different V2O3 coverage are shown
lectivitytowardsmethaneis>96%, andTOFandselectivity after a series of reduction steps (separated by oxidation in
are almost identical to those of the bare rhodium surface.
In Fig. 10 conversion versus time plots are shown for
2.0 × 10−7 mbar O2 at 673 K).
Incontrast, ifalow-temperaturereductionfollowsahigh-
(Rh + 0.55 MLE V2O3). The reduction steps (again sepa- temperature reduction (673–873 K, for 10 min) (with oxi-
rated by oxidation at 673 K) are consecutively numbered. dation in 2.0 × 10−7 mbar O2 in between) a higher activity
In Fig. 10, the hydrogen pressure during reduction was is observed than after low-temperature reduction of an as-
2.0 × 10−7 mbar in steps 2 and 3, and 200 mbar in steps prepared VOx /Rh sample at the same temperature. Thus,
4 and 5. For equal reduction temperature the conversion vs after HTR the activity is not reversible by oxidation. In
time plots are identical, i.e., the hydrogen pressure during Fig. 10 it is also seen that on the bare rhodium surface the
prereduction does not influence the activity or the compo-
sition of the overlayer, as mentioned before. On the other
hand, the activity changes induced by reduction are fully
FIG. 10. Conversion vs time plots for (Rh + 0.55 MLE V2O3). Re-
action conditions: 40 mbar CO, 120 mbar H2, He added to 1000 mbar,
FIG. 9. CO hydrogenation activity after HTR as a function of vana-
T
= 573 K. Prereduction: (ꢂ, step 2) 373 and (ꢃ, step 3) 673 K at
2.0 × 10−7 mbar H2 for 10 min; (ꢁ, step 4) 373 and (ꢀ, step 5) 673 K
at 200 mbar H2 for 10 min; and (ꢄ, step 6) 823 K at 2.0 × 10−7 mbar H2
for 10 min. For comparison: (ꢅ, step 1) bare Rh.
dium oxide coverage. Reaction conditions: 40 mbar CO, 120 mbar H2, He
added to 1000 mbar, T = 573 K. Prereduction: 2.0 × 10−7 mbar H2 at
823/873 K for 10 min and at 1100 K for 1 min.

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
429
TABLE 2
Selectivity of (Rh + VOx) after HTR (Measured below 3% Conversion)a
Selectivity (%)
TOF
Sample
(1/s · site)
Methane
Ethane
Ethene
1
Propane
Propene
0.5
Bare Rh
0.016
0.059
0.089
0.10
98.5
96
94
93
95.5
1.5
3
3
6
4.5
0.24 ML V2O3 (Tred = 823 K)
0.55 ML V2O3 (Tred = 823 K)
0.92 ML V2O3 (Tred = 873 K)
0.89 ML V2O3 (Tred = 1200 K)
0.5
2
1
0.022
a Reaction conditions: 40 mbar CO, 120 mbar H2, He added to 1 bar, T = 573 K. Reduction at indicated T in 2.0 ×
10−7 mbar H2 for 10 min.
methane formation increases linearly with time, while on
the oxide-promoted sample a noticeable deactivation is ob- V2O3) are shown. On the HTR samples carbon formation
served with increasing reaction time. On the HTR sample is generally much less than on oxide-promoted samples.
In Fig. 12 postreaction Auger spectra for (Rh+0.55 MLE
deactivation is much less effective.
An increased oxygen signal is observed after reaction, but
Postreaction surface analysis by AES indicates the pres- flashing the samples briefly to 623 K removes adsorbed CO
ence of carbon and small quantities of oxygen-containing and/or other adsorbed oxygen-containing species and de-
species. On the vanadium oxide-promoted surface the creases the oxygen signal to the level measured on bare
“steady state” carbon content is much higher than on the rhodium under identical conditions. Thus there is no indi-
bare surface and it increases with increasing oxide cover- cation for the oxidation of vanadium to vanadium oxide
age. The reactivity of the deposited carbon was tested by or the formation of vanadium carbide during reaction. As
subjecting the sample to 1000 mbar H2 at 573 K. Only a mentioned before, the above results were all obtained from
negligible amount of methane formation was observed and a CO : H2 mixture of stoichiometry of 1 : 3 at 573 K (40 mbar
the intensity of the carbon peak decreased by only about CO and 120 mbar H2 and with He added to 1 bar total
10–20% even after several hours of reaction. Hence, the pressure). Raising the hydrogen partial pressure to 400 and
carbon formed during reaction is not reactive towards hy- 1000 mbar at constant CO pressure and coverage leads to
drogen but is responsible for the observed deactivation by an increase in the promotional effect. In agreement with
site blocking.
previous work (13, 15) a reaction order of about 1 with
FIG. 11. Initial turnover frequency during a sequence of reduction steps. Reaction conditions: 40 mbar CO, 120 mbar H2, He added to 1000 mbar,
T = 573 K. Prereduction: 2.0 × 10−7 mbar H2 at the stated temperature, 10 min; reductions separated by oxidation at 673 K, 2.0 × 10−7 mbar O2 for
10 min.

430
REICHL AND HAYEK
cally most stable configuration. Upon heating in vacuum
above 673 K the oxide decomposes and is converted in is-
land structures of a V/Pd subsurface alloy with a palladium-
terminated surface, without ever adopting the valence state
V2+ (18, 19).
In our experiments the plots of the respective Auger in-
tensities vs the amount of deposited vanadium (determined
with the quartz microbalance) exhibit linear segments with
slope changes agreeing with two-dimensional island growth
in the submonolayer range. Upon deposition of one nomi-
nal monolayer V2O3 Rh300 decreases linearly to 0.65 of its
initial value. We can assume that on the polycrystalline Rh
surface preferential growth will occur along step edges, but
also along the grain boundaries.
It is convenient to separate two different situations—
vanadium oxide overlayers after low-temperature reduc-
tion (LTR, ≤673 K) and vanadium metal in the surface
region after HTR at 823 K and above. (The status of the
surface after reduction between 673 and 823 K is not re-
producible and mainly determined by the duration of the
reduction.)
After reduction at or below 673 K vanadium oxide over-
layers are stable and were identified in two different com-
positions:
FIG. 12. Postreaction spectra of (Rh + 0.55 MLE V2O3): reaction
after LTR (673 K) (a), reaction after HTR (823 K) (b), and reaction after
HTR followed by flashing to 623 K (c). For comparison: bare Rh surface
after reaction and flashing to 623 K (d).
• Reduction between 373 and 573 K leads to a composi-
tion characterised by O508/V468 = 2.9 0.15 (cf., phase I).
• Reduction between 623 and 723 K leads to an oxide
overlayer with O508/V468 = 1.9 0.1 (phase II).
respect to hydrogen and a slightly negative order with re-
spect to CO were determined.
In principle this is consistent with the findings of Boffa
et al. (13) and Hartmann and Kno¨zinger (15), who mea-
sured a changing proportion of V3+ to V2+ ions on the
surface depending on the reduction conditions. However,
4. DISCUSSION
4.1. Structure and Composition
The preparation procedure applied in this study is sim- the fact that the calculations by Kresse and coworkers
ilar to that used by Boffa et al. (13) for the preparation (19) exclude the existence of a (thermodynamically) sta-
VOx on polycrystalline rhodium and to that by Hartmann ble VO phase on Pd(111), and possibly also on a Rh(111)
et al. (15) for the preparation of VOx on Rh(111). They both surface, has led us to consider other possible stoichiome-
observed two-dimensional island growth and assigned a va- tries for phases I and II. In fact, a recent STM investiga-
lence state 3+ to vanadium in the as-grown oxide overlayer tion by Netzer and coworkers (23) revealed that vanadia
on the basis of core level shifts in XPS, corresponding to adlayers on Rh(111) may adopt a variety of as yet un-
a “V2O3”-type stoichiometry. Depending on the reduction determined surface structures under different preparation
temperature they found a changing proportion of V3+ to conditions.
V2+ ions in the oxide overlayer, indicating a reduction of
In an earlier study Szalkowski and Somorjai (27, 28) in-
the initial V2O3 oxide. On Pd(111) Surnev and coworkers vestigated oxide powder samples of different stoichiome-
(16–19) observed various surface structures with increasing try and related the Auger intensities to the oxidation state
coverage, and with three-dimensional growth starting only of vanadium. They measured the ratio O508/V468, but also
just below monolayer oxide coverage. In the submonolayer the ratio of the vanadium valence band transition (VLMV
range VOx grows in two dimensions either along steps of
=
V
468) to the vanadium inner shell transition (VLMM = V430).
the Pd surface or as circular islands on the terraces. Upon V468/V430 increases with decreasing oxidation state due to
annealing the layers become very mobile and may change the different occupation of the vanadium valence band. Us-
from more open to more closed surface structures. The sto- ing Szalkowski’s findings and the relative sensitivity fac-
ichiometry of the low-coverage 2D phases is usually V3+
,
tors tabulated in the Handbook of Auger Electron Spec-
but V4+ may also be present depending on the preparation troscopy (29) we obtain a formal stoichiometry of V3O7 for
conditions. Ab initio density functional theory calculations phase I and of V2O3 for phase II. However, for many rea-
(18) confirmed that a surface V2O3 layer is the energeti- sons, such as the different electron escape depths, different

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
431
FIG. 13. Scheme of the change of the surface composition of Rh/VOx with increasing reduction temperature.
backscattering factors, and adsorbate–substrate interac- A schematic presentation of the surface composition as a
tions, Auger data measured on bulk samples and on thin function of reduction temperature is given in Fig. 13.
films are not readily comparable even if the same experi-
mental setup is used. On the other hand, structures which
are thermodynamically unfavourable on a single crystal sur-
4.2. Catalytic Activity
face may be kinetically stable on the polycrystalline Rh
This section relates the measured catalytic activity to the
surface. Hence, at present we must conclude that the initial composition and structure of the sample before and after
stoichiometry is between V3O7 and V2O3 (phase I) and that reaction.
reduction at 673 K leads to a kinetically stable phase II of
From Fig. 8, where the activity is shown as a function of
a formal composition between V2O3 and VO. Upon tran- oxide coverage, we see that the maximum rate enhance-
sition between the two phases no change of the vanadium ment is reached near 0.5-ML coverage, as in the earlier ob-
concentration on the surface is observed, and the position servations (13). In contrast, thick layers of vanadium oxide
of the activity maximum with respect to the nominal (V2O3) are about two orders of magnitude less active and therefore
coverage is not altered.
vanadium oxide alone cannot be responsible for the activ-
A high-temperature hydrogen treatment (823–873 K) ity enhancement. Hence, upon reduction between 373 and
leads to a complete reduction of the vanadium oxide to 573 K the reaction will occur at the perimeter of vanadium
metallic vanadium. The V468 signal is markedly attenu- oxide islands. These islands (phase I) have a defined struc-
ated and hence the vanadium concentration on the sur- ture, probably similar to that observed on Pd(111). With in-
face is decreased (see, e.g., Fig. 4). Metallic vanadium and creasing reduction temperature they loose oxygen (Fig. 5)
rhodiumarewellmiscibleandseveralalloysofVandRhare and are transformed to phase II. The kinetic experiments
known. In separate experiments (30) it could be shown that reveal a small but reproducible enhancement of the initial
vanadium metal overlayers on rhodium exchange with the activity upon raising the reduction temperature from 573
rhodium substrate upon annealing in UHV above 700 K. to 673 K (Fig. 8), which is clearly related to the changing
Thus, after reduction of the oxide at elevated tempera- vanadium oxide stoichiometry.
ture, the resulting metallic vanadium diffuses into the bulk.
As mentioned before, promotion of the CO hydrogena-
The fact that V468 and V430 are still detected by AES indi- tion by reducible oxides has been observed long ago on
cates that it remains in the near-surface region. The con- different noble metals (e.g., 8, 31, 32). Burch and Flambard
centration of vanadium in the near-surface region can be (33, 34) were among the first to propose that the interface
quantified by the V468/Rh300 signal ratio which increases between metal and oxide provides highly active sites and
almost linearly with increasing initial vanadium oxide cov- that the activation of adsorbed CO occurs via an interaction
erage upon reduction between 823 and 873 K (Fig. 6). On of the oxygen end with cationic centres or oxygen vacancies
the Rh(111) surface a rhodium–vanadium subsurface alloy present at the interface. On regular catalysts the study of
of defined stoichiometry has been recently characterised the effect of metal–oxide interactions is impeded by the dif-
by STM (22). Its rhodium-terminated surface contains a ficulties in controlling and defining the structure and mor-
high density of atomic steps. If the reduction temperature phology. On metal surfaces decorated by oxide overlayers
is increased above 1000 K, no vanadium is present in the Boffa et al. (14) demonstrated that the extent of promotion
near-surface region, but it is dissolved in the rhodium bulk. by reducible oxides is related to the reducibility and the

432
REICHL AND HAYEK
Lewis acidity of the oxide. The idea of describing the inter- higher alkanes. The TOF increases initially with coverage,
action between CO and reduced oxides as Lewis acid–base as after LTR, but it levels off above about 0.5 MLE (Fig. 9).
complexes was first suggested by Sachtler et al. (35, 36).
On HTR samples methane is the main product (93–96%),
The postreaction analysis by AES reveals that the occurring in amounts similar to that measured on the bare
amount of carbon is significantly greater on vanadium rhodium surface (Table 2). Since V468/Rh300 also increases
oxide-covered rhodium than on the bare rhodium surface. with initial coverage it is concluded that the extent of pro-
The amount of carbon depends on the oxide coverage and motion is related to the concentration of vanadium in the
on the hydrogen excess in the reactant gas. On the bare subsurface region. Postreaction Auger spectra show that
rhodium surface the unreactive carbon species formed after segregation of vanadium to the surface and oxidation of
long reaction times is clearly responsible for catalyst de- vanadium to VOx do not occur upon reaction, at least not
activation. On the vanadium oxide-promoted catalyst the inmeasurableamounts. Therefore, thepromoting effectob-
rate of carbon formation is higher and deactivation there- served after HTR cannot be attributed to the presence of
fore much faster and more pronounced. Therefore, if the vanadium oxide. However, a synergistic effect of subsur-
CO/hydrogen ratio is relatively high (e.g., 1 : 3, correspond- face vanadium and small patches of vanadium oxide, which
ing to the stoichiometry of the reaction) the promotion is may be produced during reaction, cannot be completely
mainly an initial effect. Vanadium oxide at the interface excluded.
promotes the dissociation of the C–O bond; thus the rates
Upon oxidation of HTR samples at 673 K vanadium
of both carbon formation and carbon hydrogenation are segregates back to the surface and forms vanadium ox-
increased. In the steady state the vanadium oxide-covered ide (Fig. 7). V468 is reduced by 10–15% with respect to
samples are generally less active because the carbon de- the original LTR sample. Therefore, the segregation is not
posits block the surface.
quantitative and the original vanadium oxide coverage is
The postreaction spectra also indicate that the oxygen- not regained. Nevertheless, although the oxide coverage
to-vanadium stoichiometry has decreased during the reac- is reduced, the catalytic activity is still higher than on the
tion, almost independently of the reaction conditions. A original VOx -covered surface independent of the initial
similar decrease has been observed upon treatment of the coverage (Fig. 11). Furthermore, these catalysts are less
sample in 40 mbar CO at reaction temperature. The reduc- susceptible to deactivation and exhibit a higher selectivity
ing potential of CO is high enough to partially reduce the towards methane compared to the original LTR catalyst.
vanadium oxide. After a reaction, or after a reduction in The O508/V468 ratio after HTR–LTR is also smaller than
CO, the ratio of O508/V468 decreases to 1.65 0.15, fairly on the original LTR sample, e.g., 2.7 vs 2.9 for (Rh + 0.55
independent of its initial value after prereduction. (A more MLE V2O3). If we assume that the vanadium oxide over-
detailed analysis of this surface is impeded because the car- layer has the same stoichiometry in both cases (LTR and
bon deposition during reduction affects the overall Auger HTR–LTR), the incomplete surface segregation indicates
intensities and does not allow comparison with the clean that some vanadium remains in near-surface layers under
sample). This means that the initial state (phase I or II) in- the stated conditions. The most likely structure of the re-
fluences mainly the initial activity but not the steady state. sulting surface is thus a vanadium oxide overlayer on top
At a lower CO/hydrogen ratio the original reduction state of a V/Rh subsurface alloy (Fig. 13). Such a mixed surface
can be maintained over longer reaction times and deacti- composition may well explain why the catalytic properties
vation by carbon is considerably reduced.
(activity, selectivity, deactivation) of these surfaces are be-
As is also well-known from bulk catalysts (25, 26) the tween those of the HTR and the LTR samples. It seems
promotion by vanadium oxide results in an increased selec- that in this case the promotion is a synergistic or at least
tivity of rhodium towards longer-chain hydrocarbons. This a combined effect of vanadium oxide and of subsurface
is also reflected in the present results. Moreover, the se- vanadium.
lectivity is determined by the choice of the prereduction
Postreaction Auger spectra (Fig. 12) show that the carbon
temperature. Reduction at 673 K decreases the selectivity content of the HTR samples after reaction is very low and
to methane and increases the ratio of unsaturated to satu- comparable to that of the bare rhodium surface, i.e., much
rated hydrocarbons. Higher pressures, lower temperature, smaller than on vanadium oxide-covered surfaces. In con-
and/or a lower H2/CO ratio would have to be applied in or- nection with the low carbon level the deactivation is almost
der to produce appreciable amounts of oxygenates (25, 26). negligible.
A completely different catalytic behaviour is obser-
For a comparison, rhodium surfaces with top layers of
ved when after high-temperature reduction a rhodium- metallic vanadium were also prepared by deposition of sub-
terminated surface is formed with vanadium occupying sub- monolayers of vanadium at room temperature, and their
surface layers. If two samples of equal initial coverage are catalytic activity in the CO hydrogenation was tested. At
compared after LTR and HTR, the HTR sample shows the start of the reaction a slight promotion was observed
higheractivity, lessdeactivation, andlessselectivitytowards but deactivation occurred very rapidly. Strong carbon and

PROMOTION OF CO HYDROGENATION ON Rh BY VANADIUM OXIDE
433
oxygen Auger signals indicate that the vanadium overlayer above 723 K, VOx is reduced to metallic V and is partially
was converted to vanadium oxide and/or carbide during the dissolved in the bulk, resulting in a V/Rh subsurface alloy
reaction. This means that on-top vanadium has completely with a Rh-terminated surface. This surface is particularly
different catalytic properties and cannot be responsible for active and less susceptible to deactivation by carbon depo-
the observed promotion.
sition.
The origin of the promotion is still a matter of debate. In
the case of VOx overlayers it is very likely that the dissocia-
tion of CO remains the rate-limiting step, and that this step
is favoured by the presence of low-valent vanadium species
and/or oxygen vacancies at the metal–oxide boundary. If
the dissociation probability is much enhanced it is conceiv-
able that the rate-limiting step shifts to CHx hydrogenation
(11), but under the present experimental conditions no in-
dication of such a shift (e.g., a change in the reaction order,
etc.) was observed. On the other hand, other mechanisms,
such as electrostatic activation of the molecule centre (37),
cannot be excluded as long as an experimental proof of CO
predissociation is lacking.
ACKNOWLEDGMENTS
This work was supported by the Austrian Science Fund within the Joint
Research Project “Gas Surface Interactions” (S 8105). We also thank the
members of the research project for their continuous cooperation.
REFERENCES
1. Tauster, S. J., Fung, S. C., and Garten, R. L., J. Am. Chem. Soc. 100,
170 (1978).
2. Singh, A. K., Pande, N. K., and Bell, A. T., J. Catal. 94, 422 (1985).
3. Sadeghi, H. R., and Henrich, V. E., J. Catal. 87, 279 (1984).
4. Braunschweig, E. J., Logan, A. D., Datye, A. K., and Smith, D. J.,
J. Catal. 118, 227 (1989).
On the subsurface alloy CO is more weakly bound than
on clean Rh (20). Hence CO dissociation will not be en-
hanced on a flat surface which excludes this as a reason
for the observed promotion of CO hydrogenation by sub-
surface vanadium. Nevertheless, the drastic increase in hy-
drogenation rates may be explained by a combination of
CO dissociation on step edges and increased hydrogen
coverage on the terraces: As observed by STM on the
V subsurface alloys of Pd (21) as well as of Rh (22) the
alloy surfaces are never flat on an atomic scale but con-
tain a high density of irregular steps and other defects
which arise during their formation. It is very likely that
CO dissociation, and therefore also CO hydrogenation,
are enhanced along these step edges. On the other hand,
the lower binding energy leads to a lower surface concen-
tration of CO on the terraces. Reduced site blocking by
CO, in turn, will increase the relative surface concentra-
tion of hydrogen and hence result in a higher reaction rate
in the frame of a Langmuir–Hinshelwood mechanism. To
prove this hypothesis a more detailed investigation of the
alloy surface structure and morphology will of course be
necessary.
5. Bernal, S., Botana, F. J., Calvino, J. J., Lopez, C., Perez-Omil, J. A.,
and Rodr´ıguez-Izquierdo, J. M., J. Chem. Soc. Faraday Trans. 92, 2799
(1996).
6. Pesty, F., Steinru¨ck, H. P., and Madey, T. E., Surf. Sci. 339, 83
(1995).
7. Bernal, S., Calvino, J. J., Cauqui, M. A., Cifredo, G. A., Jobacho, A.,
and Rodr´ıguez-Izquierdo, J. M., Appl. Catal. A 99, 1 (1993).
8. Vannice, M. A., J. Catal. 37, 449 (1975).
9. Demmin, R. A., and Gorte, R. J., J. Catal. 105, 373 (1987).
10. Levin, M. E., Salmeron, M., Bell, A. T., and Somorjai, G. A., J. Catal.
106, 401 (1987).
11. Williams, K. J., Boffa, A. B., Lahtinen, J., Salmeron, M., Bell, A. T.,
and Somorjai, G. A., Catal. Lett. 5, 385 (1990).
12. Williams, K. J., Boffa, A. B., Salmeron, M., Bell, A. T., and Somorjai,
G. A., Catal. Lett. 9, 41 (1991).
13. Boffa, A. B., Bell, A. T., and Somorjai, G. A., J. Catal. 139, 602
(1993).
14. Boffa, A. B., Lin, C., Bell, A. T., and Somorjai, G. A., Catal. Lett. 27,
243 (1994).
15. Hartmann, Th., and Kno¨zinger, H., Z. Phys. Chem. 197, 113
(1996).
16. Leisenberger, F. P., Surnev, S., Vitali, L., Ramsey, M. G., and Netzer,
F. P., J. Vac. Sci. Technol. A 17, 1743 (1999).
17. Leisenberger, F. P., Surnev, S., Koller, G., Ramsey, M. G., and Netzer,
F. P., Surf. Sci. 444, 211 (1999).
18. Surnev, S., Vitali, L., Ramsey, M. G., Netzer, F. P., Kresse, G., and
Hafner, J., Phys. Rev. B 61, 13495 (2000).
19. Surnev, S., Kresse, G., Ramsey, M. G., and Netzer, F. P., Phys. Rev. Lett.
87, 086102 (2001) .
5. CONCLUSIONS
20. Hayek, K., Jenewein, B., Klo¨tzer, B., and Reichl, W., Top. Catal. 14, 25
(2001).
Vanadia adlayers on a Rh surface promote the hydro-
genation of CO. In the present study this promotion was
investigated, with emphasis on the effect of the oxidation
state of vanadium on the catalytic activity. The initial reac-
tion rates depend strongly on the temperature of hydrogen
21. Konvicka, C., Jeanvoine, Y., Lundgren, E., Kresse, G., Schmid, M.,
Hafner, J., and Varga, P., Surf. Sci. 463, 199 (2000).
22. Konvicka, C., and Varga, P., et al., manuscript in preparation.
23. Schoiswohl, H., Surnev, S., and Netzer, F. P., et al., manuscript in prepa-
ration.
reduction. The observed increase of the reaction rate by 24. Reichl, W., Rosina, G., Rupprechter, G., Zimmermann, C., and Hayek,
K., Rev. Sci. Instrum. 71, 1495 (2000).
25. Van der Lee, G., Bastein, A. G. T. M., van den Bogert, J., Schuller,
B., Luo, H. Y., and Ponec, V., J. Chem. Soc. Faraday Trans. 1 83, 2103
(1987).
26. Beutel, T., Alekseev, O. S., Ryndin, Yu. A., Likholobov, V. A., and
reduction up to 673 K can be correlated to the existence of
two adlayer phases of different composition. CO dissocia-
tion is enhanced at the perimeter sites of the VOx islands,
leading to enhanced hydrogenation rates but also to en-
hanced deactivation. If the reduction temperature is raised
Kno¨zinger, H., J. Catal. 169, 132 (1997).

434
REICHL AND HAYEK
27. Szalkowski, F. J., and Somorjai, G. A., J. Chem. Phys. 61, 2064
(1974).
28. Szalkowski, F. J., and Somorjai, G. A., J. Chem. Phys. 56, 6097 (1972).
29. Davis, L. E., MacDonald, N. C., Palmberg, P. W., Riach, G. E., and
32. Haller, G. L., and Resasco, D. E., Adv. Catal. 36, 173 (1989).
33. Burch, R., and Flambard, A. R., J. Catal. 78, 389 (1982).
34. Burch, R., and Flambard, A. R., J. Catal. 86, 384 (1982).
35. Sachtler, W. M. H., and Ichikawa, M., J. Phys. Chem. 90, 4752 (1986).
Weber, R. E., Eds., “Handbook of Auger Electron Spectroscopy.” 36. Sachtler, W. M. H., Shriver, D. F., Hollenberg, W. B., and Lang, A. F.,
Physical Electronic Ind., Eden Prairie, MN, 1978.
30. Reichl, W., and Hayek, K., manuscript in preparation.
31. Pande, N. K., and Bell, A. T., J. Catal. 98, 7 (1986).
J. Catal. 92, 429 (1985).
37. Bonac˘ic´-Koutecky´, V., Koutecky´, J., Fantucci, P., and Ponec, V., J.
Catal. 111, 409 (1988).