MultiTRACK and operando Raman-GC study of oxidative dehydrogenation of propane over alumina-supported vanadium oxide catalysts

Article abstract of DOI:10.1039/b305813c

Combined operando Raman-GC and MultiTRACK studies provide new insights into the interaction of propane with a V/alumina catalyst. The Raman-GC analysis showed that the catalyst is essentially in the oxidized state during oxidative dehydrogenation reaction conditions, while stable intermediates and/ or carbonaceous deposits are not observed on the catalyst surface. In the absence of oxygen, the catalyst is reduced by propane, and two types of carbonaceous deposits can be observed: one with a more aliphatic character at low temperatures, and one with a graphitic character at higher temperatures, of which the particle size increases as a function of increasing temperature. In agreement with the Raman studies, evidence for carbonaceous deposits was also provided by the MultiTRACK experiments. From CO₂ response profiles of the oxidation of these deposits, it was concluded that increasing temperature of operation and increasing propane/oxygen ratio enhance the amount and stability of the surface carbonaceous species formed. Based on the MultiTRACK studies also the participation of two types of oxygen species in the reaction of propane was evident: a highly reactive super-surface oxygen mainly yielding CO₂, and a catalytic oxygen, associated with the vanadia phase. In MultiTRACK conditions, the extent of participation of vanadia-associated oxygen increases with reaction temperature and/or the propane/oxygen ratio, enhancing the selectivity of the reaction to propene.

Full text of DOI:10.1039/b305813c

View Article Online / Journal Homepage / Table of Contents for this issue
MultiTRACK and operando Raman-GC study of oxidative
dehydrogenation of propane over alumina-supported vanadium
oxide catalystsy
a
b
b
a
b
G. Mul, M. A. Ba n˜ ares,* G. Garcia Cort e´ z,z B. van der Linden, S. J. Khatib and
a
J. A. Moulijn
a
Delft University of Technology, Reactor and Catalysis Engineering, Delft, The Netherlands
Instituto de Cat a´ lisis y Petroleoqu ı´ mica, CSIC, Campus Cantoblanco, E-28049, Madrid,
Spain. E-mail: mbanares@icp.csic.es; Fax: +34-91-5854760
b
Received 22nd May 2003, Accepted 25th July 2003
First published as an Advance Article on the web 21st August 2003
Combined operando Raman-GC and MultiTRACK studies provide new insights into the interaction of propane
with a V/alumina catalyst. The Raman-GC analysis showed that the catalyst is essentially in the oxidized state
during oxidative dehydrogenation reaction conditions, while stable intermediates and/or carbonaceous deposits
are not observed on the catalyst surface. In the absence of oxygen, the catalyst is reduced by propane, and two
types of carbonaceous deposits can be observed: one with a more aliphatic character at low temperatures, and
one with a graphitic character at higher temperatures, of which the particle size increases as a function of
increasing temperature. In agreement with the Raman studies, evidence for carbonaceous deposits was also
2
provided by the MultiTRACK experiments. From CO response profiles of the oxidation of these deposits, it
was concluded that increasing temperature of operation and increasing propane/oxygen ratio enhance the
amount and stability of the surface carbonaceous species formed. Based on the MultiTRACK studies also
the participation of two types of oxygen species in the reaction of propane was evident: a highly reactive
super-surface oxygen mainly yielding CO , and a catalytic oxygen, associated with the vanadia phase. In
2
MultiTRACK conditions, the extent of participation of vanadia-associated oxygen increases with reaction
temperature and/or the propane/oxygen ratio, enhancing the selectivity of the reaction to propene.
1
3
for such reaction. It does appear that the V–O–S (S ¼ cation
of the support) bond is directly involved in the rate-determin-
ing step for the ODH of propane. Operando studies are cur-
rently used to further study the effect of the interaction of
propane and propene on the structure of alumina-supported
vanadia catalysts.
Introduction
The fundamental understanding of a structure–activity rela-
tionship requires the simultaneous determination of both,
structure and activity on the same sample while it is actually
working. Many authors have presented such need with differ-
1
–23
ent emphasis.
has been proposed to express a methodology that combines
The term ‘‘operando’’ (working in Latin)
In view of the above-described summary, significant knowl-
edge has been gained about the structure of the active vanadia
sites during propane ODH. However, the kinetics of the reac-
tion have been less extensively studied. Only recently a study
by Khodakov et al. describes the kinetics and mechanism of
the reaction. The studies were conducted under propane-rich
conditions. These authors suggest a sequential (via propene) as
well as a parallel (from propane) formation of CO . Further-
2
reaction in situ spectroscopic characterization and simulta-
neous activity measurement in a single experiment,
the first conference on operando spectroscopy has been orga-
nized, and in less than two years a paper has been published
providing a perspective about the operando approach. Selec-
tive oxidation catalysis is a field where operando methodology
is developing rapidly. These reactions have attracted many
9
,10,13–15
2
9
2
2
more, the dissociative O adsorption and the C–H bond acti-
2
vation steps are proposed to be irreversible. The possible
2
4–28
people active in the field of heterogeneous catalysis.
Many
studies evaluate the structural requirements of vanadia-
containing catalysts to be active for oxidative dehydrogenation
formation (and oxidation) of carbonaceous deposit was not
included in the mechanistic scheme, probably because signifi-
cant formation of those deposits during steady-state oxidative
dehydrogenation reaction conditions does not appear to be
important. However, it is well known that coke can be easily
(
ODH) reaction. By variation of the catalyst loading, and
determination of the corresponding activity, conclusions have
been drawn on the role in the ODH reaction of the various
vanadium oxide entities in the catalyst. The use of in situ
Raman and operando Raman-GC rules out the participation
of the terminal V ¼ O bond as the critical site for propane
3
0
formed over bifunctional (acidic and redox) materials, such
as alumina-supported vanadia.
This work further evaluates the mechanism of the propane
ODH reaction over a vanadia/alumina catalyst, as determined
by operando Raman-GC and an advanced TAP (Transient
Analysis Product) reactor setup, the MultiTRACK (Multiple
Transient Analysis of Catalytic Kinetics). The formation of
carbonaceous deposits on this catalyst will be demonstrated
and evaluated. Specifically the role of temperature and the
applied propane/oxygen ratio on the amount and nature of
1
3,15
oxidative dehydrogenation.
Also, operando Raman-GC
studies suggest that bridging V–O–V bonds may not be critical
y Presented at the International Congress on Operando Spectroscopy,
Lunteren, The Netherlands, March 2–6, 2003.
z On leave from Departmento de Engenharia Quı
´
mica, FAENQUIL;
Rod. Itajub a´ -Lorena, Km 74,5; 12600-000, Lorean, SP, Brazil.
4
378
Phys. Chem. Chem. Phys., 2003, 5, 4378–4383
This journal is # The Owner Societies 2003
DOI: 10.1039/b305813c

View Article Online
the carbonaceous deposit will be discussed. Based on the Multi-
TRACK studies also the participation of two types of oxygen
in the reaction of propane will be demonstrated.
Experimental
Catalyst synthesis
The catalysts were prepared by impregnation of a commercial
2
ꢀ1
g-Al O (Girdler S u¨ dchemie 160 m g ) support with an aqu-
2 3
eous solution of ammonium metavanadate. Due to modest
solubility of the vanadium salt, the solution was heated to
3
33 K. The content of vanadium in the catalysts corresponds
to half monolayer coverage of VO , i.e., near 4 V atoms
nm (16 wt.% V O ); this catalyst is labelled 16VAl. After
x
ꢀ
2
2
5
Fig. 1 Schematic overview of the MultiTRACK set-up.
drying at 393 K for 2 h, the samples were calcined at 273 K
for 4 h. The chemical analyses of the catalysts were carried
out by inductively coupled plasma analysis (ICP-AES) using
a Perkin-Elmer Optima 3300 DV spectrometer.
components in the exit gas stream with a maximum sample fre-
quency of 1 MHz. In the experiments described in this paper,
1
8
MS 1 was used to detect
detect propane (m/z 29) and MS 3 to simultaneously detect
products (CO (m/z 44, m/z 46, or m/z 48), CO (m/z 28), or
2
O (m/z 36), MS 2 to simultaneously
Operando Raman experiments
2
The Raman spectra during propane ODH (oxidative dehydro-
genation) and DH (non-oxidative dehydrogenation) were run
in an operando Raman cell with a Renishaw Micro-Raman
System 1000 equipped with a thermoelectrically cooled CCD
detector (200 K) and a holographic super-Notch filter that
removes the elastic scattering. The samples were excited with
the 514 nm Ar line. The spectra acquisition consisted of 1000 s
propene (m/z 41). MS 4 was not used.
As the signal-to-noise ratio in this system is excellent, no
averaging of pulses is necessary to obtain good peak signals.
This is an important aspect, as transient phenomena may
remain unobserved when the peaks have to be averaged.
Before the measurements, the catalysts were evacuated in
the MultiTRACK reactor at 723 K for 3 h. The mechanistic
studies of the propane ODH were performed using so-called
pump–probe experiments. An example of the experimental
(
10 accumulations of 100 seconds). The spectral resolution
ꢀ1
is 3 cm . To prevent local heating of the sample during the
acquisition due to the laser beam, the laser power was low
2
procedure is illustrated in Fig. 2. The CO response is shown
(
ca. 8 mW), which lowers the signal-to-noise ratio. Thus, the
at the time of the oxygen pulse, as well as at the time of the
propane pulse. A time interval of 1 s between the O and pro-
spectra are taken at representative spots, with no appreciable
heating with respect to the reaction temperature of the catalyst
bed. The operando Raman-GC spectra were run with a home-
made cell-reactor that consists of a fixed-bed quartz-micro-
reactor contained by quartz wool plugs on both ends. The
operando cell-reactor walls have optical quality. The operando
cell-reactor is heated to reaction temperature and the gas feed
and outlet is heated so that there are no cold points in the way
from the operando reactor-cell to the on-line gas chromato-
graph. More details on the operando Raman-GC setup can
2
pane pulse was applied, to obtain a good resolution between
the (sometimes tailing) responses of the system to the oxygen
and propane pulses. At the time of the oxygen pulse, the shape
and intensity of the CO (and CO) response provide informa-
2
2
tion on the amount (signal intensity), type (CO /CO ratio),
and stability (broadness response) of the carbonaceous deposit
formed at the time of the propane pulse. At the time of the pro-
pane pulse the propane conversion can be determined, as well
as the selectivity of the reaction (propene formed) and kinetic
pathway (by the order of appearance of the products). The
broadness of the pulse response indicates the extent of inter-
action (number of adsorption/desorption cycles) of reactant
9
,10,13–15
be found elsewhere.
3 8 2
Propane oxidation (C H /O /
3
He ¼ 1/6/4) was performed with a total flow rate of 67 cm
ꢀ
1
min and 150 mg of catalyst. Activity data from a fixed bed
conventional reactor and the operando Raman reaction cell
shows no significant differences. During the operando
Raman-GC study, the catalyst structure and its performance
were analysed (Raman and on-line GC) every 10 degrees.
The GC analyses were run in a Varian 3800 series gas chro-
matograph with a TCD detector and automatic on-line sam-
pling through a 6-way valve. The analytic system comprises
two columns, Porapak-Q and Molecular Sieve 4A, in a
column-isolation configuration.
(
propane) and products.
With successive alternations of propane and oxygen pulses,
the catalysts were ‘screened’ for activity as a function of tem-
perature, using a propane to O ratio of 1 : 1 and pulse size
2
MultiTRACK setup and description of experiments
An overview of the MultiTRACK (Multiple Transient Analy-
sis of Catalytic Kinetics) system is presented in Fig. 1. A cata-
lytic reactor is located in an ultra-high vacuum system
operating at a pressure of 3 Pa. The catalyst bed (100 mg,
pellet size 125–200 mm) packed between two layers of SiC-
particles (each containing 200 mg particles of 230 mm in size),
is exposed to a gas pulse with adjustable size of the order of
1
0
5
17
1
to 10 molecules. In the reactor, the shape and composi-
tion of the gas pulse changes due to processes such as diffusion,
adsorption, and reaction. At the reactor exit, the reaction
products are analysed by four quadrupole mass spectrometers.
All four mass spectrometers are able to analyse one of the
2
Fig. 2 Example of the CO response to the experimental procedure.
At t ¼ 0.1 s the oxygen pulse is dosed, followed by the propane pulse
at t ¼ 1.1 s. This sequence could be repeated with reproducible CO
2
responses at least 150 times.
Phys. Chem. Chem. Phys., 2003, 5, 4378–4383
4379

View Article Online
1
6
ꢀ1
2
assigned to surface layers of sp carbon species.
33–37
of 1.5 ꢁ 10 molecules pulse . At 673 K the effect of the
Two
ꢀ
1
propane/oxygen ratio was studied, by variation of the propane
new Raman bands near 1589 and 1327 cm become evident
as the reaction temperature increases. These bands are very
close to those indicative of sp carbon in graphitic materi-
1
6
pulse size, with a fixed oxygen pulse size of 1.5 ꢁ 10 molecules
ꢀ
1
2
pulse . Because of the contribution of propane and CO
2
to
was used, to facilitate product identification
m/z 46 was used to identify CO production). The only pro-
and propene.
1
8
38–40
ꢀ1
The Raman band at 1580 cm corresponds to the
m/z 44,
O
2
als.
E2g C–C stretching mode, and is the only possible band in sin-
gle crystal graphite. The Raman band at 1355 cm has been
(
2
3
8
ꢀ1
ducts observed were CO, CO
2
3
8
assigned to the A1g mode. The position of this band may
change depending on the nature of the carbon material.
3
7
Results and discussion
Furthermore, Tuinstra and Koenig showed a linear correlation
˚
between 1000/La (La is crystal size in A) and the Raman
intensity ratio of the bands at 1327 cm over 1580 cm .
It is therefore possible to estimate that the graphitic aggregates
Operando Raman-GC study
ꢀ1
ꢀ1 38
Fig. 3 and 4 illustrate the operando Raman-GC study of the
6V/Al catalyst during propane oxidative dehydrogenation
ODH) and non-oxidative dehydrogenation (DH). During
propane oxidative dehydrogenation (C + O + He) the cat-
˚
formed on 16VAl at 823 K and above are ca. 79 A. The
1
(
ꢀ
1
increase of the Raman band near 1355 cm indicates the
4
1
structural disorder of the graphitic deposit.
3
H
8
2
Summarizing, the operando Raman-GC studies show the
formation of two types of carbonaceous deposits under pro-
pane DH conditions: one with a more aliphatic character at
lower temperatures, and one with a graphitic type character
at higher temperatures. The particle size of graphitic-like
deposits increases with temperature. It should be noted that
the propane DH conversion becomes measurable concomi-
tantly with the formation of graphite-like deposits (Fig. 4).
During the operando propane DH reaction, propylene is the
most important reaction product recorded, with selectivity
values very close to 100% (Fig. 4). It appears likely that the
formation of the graphite-like deposits is a consequence of
the propene formed during propane DH reaction.
alyst is active, as the simultaneous activity data illustrate in
Fig. 3. The conversion increases with reaction temperature,
while selectivity appears to decrease. The activity data are con-
sistent with those observed for this catalyst in a conventional
fixed-bed microreactor. During propane ODH, the catalyst
ꢀ
1
exhibits a Raman band at 1012 cm and a broad Raman band
ꢀ
1
near 890 cm , which correspond to the terminal V=O and
V–O–V bonds vibration modes of surface dispersed mono-
oxo vanadium oxide species, respectively. These Raman bands
of surface vanadium oxide species do not appear to be signifi-
cantly affected by the propane ODH reaction in the 573 K to
6
73 K temperature range (Fig. 3). Therefore, it can be con-
cluded that the catalyst remains essentially oxidized during
the ODH reaction, which is consistent with previous reaction
3
1
in situ UV-Vis studies during hydrocarbon ODH, where only
a very small fraction of the V species are reduced. Thus, the
reoxidation rate of surface vanadium oxide species appears
to be faster than its reduction, which is in line with TPR/
TPO studies by L o´ pez Nieto et al.. The operando Raman-
GC study does not show the formation of any carbon deposit
on the surface of the catalyst during propane ODH reaction.
Fig. 4 shows the Raman spectra of the operando Raman-GC
study of the same catalyst during propane DH reaction
MultiTRACK reactor studies
Effect of temperature on catalyst performance. The effect of
temperature on the formation of products during the oxygen
and propane pulses, respectively, is shown in Figs. 5a and 5b.
3
2
1
8
As indicated in the experimental procedures,
determine the peak shape of CO (m/z 46), to prevent interfer-
ence of propane at m/z 44. At 673 K, the product signals are
small. At oxygen pulse, CO is the predominant product, while
2
O was used to
2
2
(
C H + He feed) and the simultaneous activity data. No activ-
3
the CO profile reaches its maximum faster than CO . At the
2
8
ity is recorded at low temperature (below 773 K). However, the
catalyst exhibits a moderate decrease of the Raman bands of
propane pulse, very low quantities of propene leave the reac-
tor, followed by CO (main product) and CO. Propene was
only formed at the time of propane pulse, and never at the
oxygen pulse.
At 773 K the product profiles become more intense. At the
time of the oxygen pulse, the CO profile passes through a max-
2
ꢀ
1
surface vanadium oxide species (1012 and 900 cm ) as the
temperature increases from 473 K to 673 K. This trend indi-
cates a reduction of surface vanadium oxide species by pro-
pane. In addition, Raman spectra also exhibit new features
ꢀ
1
in the 1100–1700 cm region. A new broad Raman band near
1
2 2
imum significantly earlier than CO . The CO profile is very
broad, not returning to its base line within 1 s after the oxygen
ꢀ
1
600 cm becomes evident at 673 K. This band has been
Fig. 3 Operando Raman-GC study of 16V/Al during propane ODH reaction at different reactions temperatures (left). Simultaneous GC activity
data are presented (right). Catalyst weight: 150 mg, total flow 67 ml min , reaction feed propane/oxygen/helium: 1/6/4.
ꢀ1
4
380 Phys. Chem. Chem. Phys., 2003, 5, 4378–4383

View Article Online
Fig. 4 Operando Raman-GC study of 16V/Al during propane DH reaction vs. reaction temperature. In situ reoxidized at 773 K after propane
DH and operando Raman-GC during propane ODH at 673 K. Catalyst weight: 150 mg, total flow 67 ml min , reaction feed propane/oxygen/
helium: 1/6/4 for ODH reaction and 1/0/6 for DH reaction.
ꢀ1
pulse. At the propane pulse, significantly higher amounts of
propene and CO are formed. The CO profile appears, more
CO is the primary product of the oxidation, suggesting that
at least part of the CO is formed as a consecutive product
2
2
evidently than at 673 K, to be composed of two overlapping
signals, one with a relatively fast response time, maximizing
before propene and CO products, the other showing a broad
and tailing response. Opposite to what was observed for steady
state experiments, a higher temperature induces a relatively
high selectivity towards propene. The various observations
can be explained as follows.
The formation of carbon oxides at oxygen pulses (Fig. 5) is
indicative of the deposition of carbonaceous species on the
catalyst surface during the propane pulse. The carbonaceous
material might consist of simple molecular species, such as
propoxides, or polymerised graphite-like deposits. From the
of CO oxidation. Carbonaceous deposits were not observed
by the Mirodatos’ group in pump–probe TAP experiments
of propane ODH over V/MgO catalysts. This is probably
related to the basic character of the MgO support vs. the
4
2
acidic character of alumina support. Acidity promotes the
formation of graphitic deposits.
3
0
In experiments similar to those shown in Fig. 5, the Miro-
datos’ group observed a relatively long retention time of CO2
(compared to the other products – propene and CO) at the
time of the propane pulse on V/MgO catalysts. This was
argued to be related to: (a), a slow process of oxidation of
propane to CO
propoxy-carboxylate formation and decomposition; or (b), a
chromatographic effect of CO by adsorption and desorption
2
, involving various elementary steps such as
2
significant tailing of the CO response at 773 K, which is
absent at 673 K, it can be concluded that the nature of the
deposited material (at the propane pulse) is changing as a func-
tion of temperature from more aliphatic (possibly containing
some oxygen) to more graphitic, which is in agreement with
the operando Raman-GC studies. Moreover, it appears that
2
on specific sites of the catalyst. Rather than a longer ‘retention
time’ in our experiments, especially at 773 K, an initial fast
2
response of CO is observed. The absence of the initial quick
CO response at Mirodatos’ V/MgO system, might again be
2
a result of the basic character of the MgO support vs. the acidic
nature of the alumina support used in the present study. The
tail in the CO profile in Fig. 5B is most likely related to the
2
consecutive reaction of propene. This is in agreement with
the relatively high concentration of CO at 0.3 s after the
propane pulse, which is consistent with a sequential reaction
of propene to CO.
The apparently contradictory results of steady state experi-
ments (where selectivity is decreasing) and MultiTRACK
experiments (where selectivity is increasing with temperature)
are most likely related to the effective partial pressure of pro-
pane. We have demonstrated in a recent MultiTRACK study
of the reduction of an oxidized Pt that sites of low reducibility
are not reactive towards H
whereas under pretreatment conditions usually applied for
CO chemisorption experiments (20% H in Ar at atmospheric
pressures) these sites are effectively reduced. Such a phenom-
2
in MultiTRACK conditions,
2
2
1
enon is often referred to as the ‘‘pressure gap’’ effect. Extra-
polating these results to the current study, and assuming that
extraction of oxygen from the catalyst is necessary for propene
formation, the relatively small amount of propene formed at
6
73 K is explained by the inability of propane molecules to
extract oxygen species from the catalyst structure under Multi-
TRACK conditions, while this is occurring at steady-state con-
ditions. It is proposed that propane is only reacting with highly
reactive super-surface oxygen species during MultiTRACK
Fig. 5 Effect of temperature on the product response profiles of
1
6V/Alat the time of the oxygen pulse (at t ¼ 0.1 s) and at the time
of the propane pulse (at t ¼ 1.1 s). Top: 673 K, Bottom: 773 K. Pro-
ducts as indicated in the figure. Conditions: propane:O
¼ 1 : 1, 25 nmol
2
ꢀ1
pulse
.
conditions at 673 K, yielding mainly CO and CO
2
along with
Phys. Chem. Chem. Phys., 2003, 5, 4378–4383
4381

View Article Online
labile carbonaceous deposits. The differences generated by the
‘pressure gap’’ can be interpreted as an indirect indication of
the reactivity of the sites.
‘
Effect of propane pulse size on catalyst performance
The effect of propane pulse size on the shape and intensity of
the CO
lyst. Please note that the oxygen pulse size amounts to 25 nmol
2
response signal is shown in Fig. 6 for the V/Al cata-
ꢀ
1
.
while that of propane ranges from 3.8 to 28.4 nmol pulse
Obviously, at a small propane pulse, no CO was detected at
2
1
8
the time of the
naceous deposit was formed. For small propane pulses, a
rather broad CO response is obtained at the propane pulse.
The CO profile at the propane pulse sharpens with the pro-
2
O pulse, suggesting that hardly any carbo-
2
2
Fig. 7 Product distribution as a function of amount of propane
pulsed at 673 K over 16V/Al. Curves as indicated in the figure. Vari-
able propane:O ratio, O pulse size: 25 nmol pulse .
pane pulse size. The yield to CO
propane pulse size, with a relatively sharp profile. A further
increase of the propane pulse induces a sharper CO profile
profile at the oxygen
2 2
at O pulse increases with
ꢀ1
2 2
2
at the propane pulse but a broader CO
pulse.
The broadening of the CO profile at the oxygen pulse with
2
2
species vs. the vanadia associated oxygen to the formation of
products (O* vs. O , respectively, in Fig. 7). As expected, this
2ꢀ
increasing propane pulse size can be explained by a change in
the nature of the carbonaceous deposit. At large propane
pulses, the carbon deposits on the surface of the catalyst
must become more organized for they take more time to be
oxidized.
leads to a change in selectivity (increasing propene amounts,
decreasing CO amounts). In agreement with the experiment
2
shown in Fig. 5, a relatively high selectivity is obtained if cat-
alytic lattice oxygen participates, which is induced either by a
higher temperature or relatively high propane/oxygen ratios in
MultiTRACK conditions. Such a conclusion is also consistent
with the activity data afforded by 16V/Al during propane
ODH and DH in the operando Raman-GC study. In the
presence of oxygen (ODH) total conversion is higher but the
system presents a rather low selectivity to propene (Fig. 4).
Conversely, in the absence of molecular oxygen feed, the sys-
tem is significantly less reactive and requires higher reaction
temperatures to achieve propane conversion. However, such
propane DH reaction affords selectivity values to propene
close to 100% and the yield to propene approaches 11% (Fig.
4), which is significantly higher than during propane ODH
reaction. Such activity and selectivity differences may be due
to the absence of highly reactive super-surface oxygen species,
which must account for the non-selective oxidation. The phe-
nomena observed in the present paper related to the nature
of the oxygen species involved in the reaction, are in agreement
with observations in the oxidation of CO with for V-bronzes.
Here a transition in the activation energy from low to high
values was observed as a function of temperature. Van den
Berg et al. suggested that such change in activation energy
was related to a change in mechanism from associative (invol-
ving super-surface oxygen species) to redox, involving lattice
The relatively broad CO
pulses, is probably the result of a chromatographic effect of
CO by adsorption and desorption on specific sites of the cat-
2
response at small propane
2
4
2
alyst, as previously discussed. The sharpening of the CO2
response at larger propane pulses is most likely related to a
higher coverage of the CO2 adsorption sites with propane
and/or carbonaceous deposit. Therefore, the lower inter-
action of CO2 with the catalytic material surface sharpens
2
the CO profile.
The effect of the propane pulse size on the amount of CO2
m/z 46) and propene formed is further illustrated in Fig. 7.
(
CO formation passes through a maximum with the size of
2
2
the propane pulse, whereas the formation of CO at the time
of the oxygen pulse continuously increases. Furthermore, pro-
pene was only observed at propane pulse larger than 10 nmol.
The continuously increasing amount of CO at the time of
2
the oxygen pulse suggests that the amount of carbonaceous
deposit increases with propane pulse size. Since the oxygen
pulse amount is constant, more oxygen is consumed to remove
these deposits, leaving less ‘active’ oxygen on the catalyst sur-
face at the time of the next propane pulse. This leads in turn
to a smaller relative contribution of highly reactive oxygen
4
3
oxygen.
An intuitive summary of the phenomena occurring is pre-
sented in Fig. 8. The left side illustrates the state of surface
species during large propane pulses and high temperatures, and
the right side, during small propane pulses and low tempera-
tures. The higher part illustrates the scenario at the time of
the propane pulse and the lower part, at the time of the oxygen
pulse. Thus, for large propane pulses (and high temperatures),
propylene forms along with carbon deposits at the time of the
propane pulse, inducing the reduction of the catalyst. At the
time of the oxygen pulse, these deposits are burnt, but it
requires a longer time due to their relative high stability. More-
over, oxygen is also consumed to reoxidise the catalyst. How-
ever, reactive surface oxygen remains present at small propane
pulses (and low temperatures); thus, only small aggregates
form on the surface at the propane pulse, since most propane
is oxidized to CO and CO
by carbonaceous deposit, adsorption/desorption cycles of
CO are possible (explaining the broad response). At the time
of the oxygen pulse, these small carbon deposits are burnt
2
. Due to the low surface coverage
Fig. 6 Effect of propane pulse size at 673 K on the product response
profiles of 16V/Al at the time of the oxygen pulse (at t ¼ 0.1 s) and
at the time of the propane pulse (at t ¼ 1.1 s). Conditions: Variable
2
ꢀ1
propane:O
2 2
ratio, O pulse size: 25 nmol pulse .
4
382
Phys. Chem. Chem. Phys., 2003, 5, 4378–4383

View Article Online
References
1
J. M. Stencel, in Raman Spectroscopy for Catalysis, Van
Nostrand, New York, 1990.
2
3
J. M. Thomas, Angew. Chem. Int. Ed., 1999, 38, 3588.
I. E. Wachs, F. D. Hardcastle, in Catalysis–A Specialist Periodical
Report, Royal Society of Chemistry, Cambridge, UK, 1993,
vol. 10, p. 102–152.
4
5
6
I. E. Wachs, Catal. Today, 1996, 27, 437.
I. E. Wachs, Topics Catal., 1999, 8, 57.
J. M. Thomas, in Characterization of Catalysts, ed. J. M. Thomas
and R. M. Lambert, John Wiley and Sons, 1980.
I. E. Wachs, in Raman Scattering in Materials Science, ed
W. H. Weber and R. Merlin, Springer, Berlin, 2000, p. 271.
I. E. Wachs and K. Segawa, in Characterization of Catalytic
Materials, ed. I. E. Wachs, Butterworths, Stoneham, MA,
p. 69–88, 1992.
7
8
Fig. 8 Overview of the processes occurring over the V/Al catalyst by
the interaction of propane with the catalyst surface. Left: large pulses,
high temperature. Right: small pulses, low temperature. Top: at the
time of the propane pulse, Bottom: at the time of the oxygen pulse.
9
M. O. Guerrero-P e´ rez and M. A. Ba n˜ ares, Chem. Commun., 2002,
1292.
10 M. A. Ba n˜ ares and I. E. Wachs, J. Raman Spectrosc., 2002,
33, 359.
1
1
1
2
B. M. Weckhuysen, Chem. Commun., 2002, 97.
S. Kuba and H. Kn o¨ zinger, J. Raman Spectrosc., 2002, 33, 325.
away rapidly, while oxygen regenerates the highly reactive
surface oxygen.
13 G. Garc ´ı a Cort e´ z and M. A. Ba n˜ ares, J. Catal., 2002, 209, 197.
1
4
5
M. A. Ba n˜ ares, M. O. Guerrero-P e´ rez, G. Garc ´ı a-Cort e´ z and
J. L. G. Fierro, J. Mater. Chem., 2002, 12, 3337.
M. A. Ba n˜ ares, in In situ Characterization of Catalysts, ed.
B. M Weckhuysen, American Scientific Publishers, 2003 in press.
M. J. Weaver, J. Raman Spectrosc., 2002, 33, 309.
G. Mestl, J. Raman Spectrosc., 2002, 33, 333.
W. H. Weber, Springer Ser. Mater. Sci., 2000, 42, 233.
H. Kn o¨ zinger, Catal. Today, 1996, 32, 71.
P. Rybaczyk, H. Berndt, J. Radnik, M. M. Pohl, O. Buyevskaya,
M. Baerns and A. Br u¨ ckner, J. Catal., 2001, 202, 45.
1
Conclusions
16
17
18
19
20
The combined operando Raman-GC and MultiTRACK ana-
lyses lead to the following conclusions with respect to propane
oxidation over alumina-supported vanadia catalysts:
At steady-state conditions, the vanadia/alumina catalyst is
essentially in its oxidized state, and no evidence for surface
intermediates/carbonaceous deposits is obtained. This is likely
due to the presence of highly reactive surface oxide species.
In the absence of oxygen, two types of carbonaceous depos-
its form by the interaction of propane with the catalyst: (a), a
labile form, which is favoured at lower temperatures and at
low propane/oxygen ratios; (b), a graphite-like phase that
forms at relatively high temperatures, which is favoured by
high propane/oxygen ratios.
21 H. Topsøe H, Studs. Surf. Sci. Catal., 2000, 130, 1.
22 H. Topsøe, J. Catal., 2003, in press.
2
3
G. J Hutchings, A Desmartin-Chomel, R. Olier and JC Volta,
Nature, 2000, 368, 41.
24
25
26
M. A. Ba n˜ ares, Catal. Today, 1999, 51, 319.
T. Blasco and J. M. L o´ pez Nieto, Appl. Catal., 1997, 157, 117.
E. A. Mamedov and V. Cort e´ s Corber a´ n, Appl. Catal., 1995,
127, 1.
27 M. O. Guerrero P e´ rez, J. L. G. Fierro and M. A. Ba n˜ ares,
J. Catal., 2002, 206, 339.
2
8
9
I. E. Wachs, J.-M. Jehng, G. Deo, B. M. Weckhuysen, V.
Guliants, J. B. Benziger and S. J. Sundaresan, J. Catal., 1997,
Temperature affords different effects on the reaction selectiv-
ity for the MultiTRACK and for the steady-state propane
ODH operando experiments. In the 673–773 K range, the selec-
tivity to propene increases with temperature in the Multi-
TRACK experiment; while an opposite effect is observed
during steady-state propane ODH. This is most likely related
1
70, 75.
2
A. Khodakov, B. Olthof, A. T. Bell and E. Iglesia, J. Catal., 1999,
181, 205.
30 M. Guisnet and P. Magnoux, Appl. Catal. A, 2001, 212, 83.
31
32
33
X. Gao, M. A. Ba n˜ ares and I. E. Wachs, J. Catal., 1999,
188(2), 325.
J. M. L o´ pez Nieto, J. Soler, P. Concepci o´ n, J. H. Herguido,
M. Men e´ ndez and J. Santamar ´ı a, J. Catal., 1999, 185, 324.
F. Kong, R. Kostecki, G. Nadeau, X. Song, K. Zaghib, K.
Kinoshita and F. McLarnon, J. Powder Sources, 2001, 97–98, 58.
2
1
to the so-called experimental ‘‘pressure gap’’. This ‘‘pressure
gap’’ effect suggests the participation of two types of oxygen
species: super-surface oxygen species and vanadia associated
oxygen. In MultiTRACK conditions, the oxygen associated
with the vanadia monolayer may only participate in the
reaction at high temperatures.
34 N. Vidano and D. B. Fischbach, J. Am. Ceram. Soc., 1978,
1, 13.
6
35
36
37
G. Katagiri, H. Ishida and A. Ishitani, Carbon, 1988, 2, 565–571.
R. J. Nemanich and S. A. Solin, Phys. Rev. B, 1979, 20, 392.
A. Cuesta, P. Dhamelincourt, J. Laureyns, A. Mart ´ı nez-Alonso
and J. M. D. Tasc o´ n, Carbon, 1994, 32, 1523.
Acknowledgements
3
39 J. L. Lauer, in Handbook of Raman Spectroscopy, ed. I. R. Lewis
8
F. Tuinstra and J. L. Koenig, J. Chem. Phys., 1970, 33, 1126.
The research is part of a COST action (D15/0021/01) spon-
sored by the European Union and by project MAT-2002-
and H. G. M. Edwards, Marcel Dekker Inc., 2001.
4
4
4
0
1
2
Y. T. Chua and P. C. Stair, J. Catal., 2003, 213, 39.
M. A. Montes-Mor a´ n and R. J. Young, Carbon, 2002, 40, 845.
A. Pantizidis, S. A. Bucholz, H. W. Zanthoff, Y. Schuurman and
C. Mirodatos, Catal. Today, 1998, 40, 207.
J. van den Berg, J. H. L. M. Brans-Brabant, A. J. van Dillen,
J. C. Flach and J. W. Geus, Ber. Bunsenges. Phys. Chem., 1982,
86, 43.
0
400-C02-01 by MCyT, Spain. GM gratefully acknowledges
a fellowship granted by the Royal Netherlands Academy of
Arts and Sciences. SJK gratefully acknowledges an I3P grant
by CSIC-EU for doctorate research. Prof. A. O. I. Krause,
Prof. B. M. Weckhuysen and Dr. E. Harlin are gratefully
acknowledged for fruitful discussions.
4
3
Phys. Chem. Chem. Phys., 2003, 5, 4378–4383
4383