Monitoring transition metal ions (TMI) in oxide catalysts during (re)action: The power of operando EPR

Article abstract of DOI:10.1039/b305884k

The benefits that can be drawn from operando and in situ ESR studies for elucidating structure-reactivity relationships in heterogeneous catalytic gas-phase reactions performed over transition metal oxide catalysts were presented. A short section presenting suitable experimental setup was followed by selected examples in which the information potential and the limits of application of operando ESR are illustrated. These comprised the action of different transition metal ions (TMI): structure and function of V ions in vanadia-based catalysts used for the oxidative dehydrogenation of propane to propylene and the selective oxidation of toluene to benzaldehyde; activity and stability of Cr sites in supported chromium oxides during dehydrogenation of propane; and structure and function of Mn ions in MnO_x storage catalysts for SCR of NO_x. The novel combination provided an excellent opportunity to monitor the reaction-dependent interconversion of paramagnetic and diamagnetic TMI, and thus, widened the variety of spectroscopically accessible species. The ESR signal intensity of paramagnetic species was inversely proportional to the temperature.

Full text of DOI:10.1039/b305884k

View Online / Journal Homepage / Table of Contents for this issue
Monitoring transition metal ions (TMI) in oxide catalysts
during (re)action: the power of operando EPRy
A. Br u¨ ckner*
Institut f u¨ r Angewandte Chemie Berlin-Adlershof e.V., R.-Willst a¨ tter-Str. 12,
D-12489, Berlin, Germany. E-mail: brueckner@aca-berlin.de
Received 23rd May 2003, Accepted 20th August 2003
First published as an Advance Article on the web 5th September 2003
Tailoring the properties of transition metal oxide catalysts so as to provide optimum performance for the
desired reaction is a major goal in catalysis research that can only be reached by deep understanding of
structure–reactivity relationships of active sites. Operando EPR is a versatile tool for that purpose since it
provides simultaneous information on valence state, coordination geometry and electronic interactions of
paramagnetic TMI in relation to activity and selectivity of the catalysts. This is demonstrated by illustrative
examples comprising the action of different TMI: (i) structure and function of V ions in vanadia-based catalysts
used for the oxidative dehydrogenation of propane to propene (ODP) and the selective oxidation of toluene to
benzaldehyde; (ii) activity and stability of Cr sites in supported chromium oxides during dehydrogenation of
propane; (iii) structure and function of Mn ions in MnO storage catalysts for selective catalytic reduction of
x
x
NO . Benefits arising from combined operando EPR/UV-vis/on-line GC studies are also shown. This
combination provides an excellent opportunity to monitor the reaction-dependent interconversion of
diamagnetic and paramagnetic TMI.
interactions and details of the local geometry are not accessi-
ble. However, it will be shown below that in those cases infor-
mation about the interaction of TMI with each other and with
reactant molecules can be obtained by evaluating the line
shape and intensity of the EPR signals.
1
. Introduction
Transition metal oxides are of outstanding importance as con-
stituents of catalysts used in heterogeneous catalytic redox
processes such as partial oxidation and hydrogenation/dehy-
drogenation of hydrocarbons. They are applied in high diver-
sity, e.g., as unsupported crystalline pure and mixed oxide
phases or in amorphous form supported on different carriers.
Activity, selectivity and stability of such catalysts depend sen-
sitively on the properties of the catalytically active transition
metal ions (TMI), e.g., on their valence state, coordination
geometry and dispersion. Tailoring those properties so as to
provide optimum performance for the desired reaction is a
major goal in catalysis research that can only be reached by
a deep understanding of structure–reactivity relationships of
active sites.
Fortunately, EPR can be applied under true operating con-
ditions, i.e., at elevated temperatures, under reactant gas flow
and by using flow cells the geometry of which does not differ
much from conventional catalytic fixed-bed microreactors
9
,13
used in lab scale.
The outlet of such EPR reactors can be
connected to a product analysis device (gas chromatograph
or mass spectrometer). Thus, it is possible to follow the beha-
viour of active sites in catalysts simultaneously together with the
catalytic performance in the same experiment. For those types
of investigations the term ‘‘operando spectroscopy’’ has been
1
4–16
recently introduced,
to distinguish from ‘‘in situ spectro-
Electron paramagnetic resonance (EPR) is a unique tool for
analyzing structural and electronic peculiarities of paramag-
scopy’’ which is commonly used in a broader sense not neces-
sarily implying product analysis and sometimes performed
under conditions quite far away from those of the catalytic
reaction.
Despite the above described benefits, in situ and operando
EPR investigations of TMI are still being performed by a
few research groups only. Most important examples comprise
the behaviour of
1
–9
netic TMI
since it provides simultaneous information on
valence state, coordination geometry and electronic inter-
actions of TMI between each other and with reactant mole-
cules and with the support. While the valence state of TMI
determines the total spin which must be different from zero
for EPR detection, differences in the coordination geometry
are sensitively reflected by intrinsic interactions of the electron
spin with the orbital momentum (g-tensor), as well as with the
spin of the nucleus (A-tensor) and, if present, with other elec-
trons in the same atom (D-tensor). Thus, it is possible to dis-
1
7,18
ꢀ unsupported VPO catalysts during selective oxidation
1
8–20
and ammoxidation
of n-butane and toluene,
2
2
1,22
ꢀ supported vanadia catalysts during oxidation of SO
2
3,24
and oxidative dehydrogenation of propane,
ꢀ supported molybdena-based catalysts during propylene
4
+
tinguish between V species supported on different positions
1
0
25
oxidation and oxidative dehydrogenation of methanol,
26
of diamagnetic oxides such as Al
TiO
2
O
3
,
or zeolites, since anchoring of VO
WO
3
/TiO
2
and
species on support
1
2
1
12
x
ꢀ supported chromium oxides during aromatization of
2
octane and dehydrogenation of propane,
7
28
surfaces leads to distortions of their local geometry. The above
described splittings are not resolved in EPR spectra of closely
neighbouring TMI due to spin–spin dipolar and/or exchange
2
9
30
31–33
ꢀ TMI such as Fe, Mn, Cu
in molecular sieve
matrices during interaction with gas phase molecules such as
O , H , H O, NH , NO and hydrocarbons.
2
2
2
3
Besides, there are numerous examples in which TMI-
containing catalysts were pretreated with certain gaseous
y Presented at the International Congress on Operando Spectroscopy,
Lunteren, The Netherlands, March 2–6, 2003.
DOI: 10.1039/b305884k
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
This journal is # The Owner Societies 2003
4461

View Online
control with these devices is done by automatic tuning of the
heater current and the gas flow velocity implying that a certain
time is required to adjust the desired temperature value. This
should be kept in mind when measuring catalyst samples that
undergo fast temperature-dependent changes.
The experiments described in this paper were performed by
an ELEXSYS 500-10/12 c.w. spectrometer (Bruker) in X-band
using a home-made fixed-bed quartz reactor heated by a bifilar
winding of 0.2 mm Pt wire (Fig. 1). This principle was first
reactants at elevated temperatures but the EPR spectra were
recorded after cooling to room or lower temperatures (e.g. refs.
6
and 34). Although those experiments provide valuable
information about persisting changes of the respective TMI,
they do not fulfill the condition of in situ EPR and, less than
ever, of operando EPR. Therefore, they are not subject of this
paper.
Another group of important in situ EPR studies comprises
the detection of hydrocarbon and oxygen radical intermediates
formed upon contact of hydrocarbon reactants with oxide sur-
described in the late 1950s for the investigation of carbon radi-
3
5–38
42,43
faces.
However, these systems do not contain TMI and
are, therefore, not considered in this paper.
cal species during pyrolysis of carbonaceous materials
and
has later been adapted also for the study of heterogeneous cat-
3
8,44
The reason for the limited activities in operando EPR may be
due to two aspects: On the one hand, the opportunities of this
technique seem to be still not adequately recognized. Thus, it is
frequently stated that useful information can only be obtained
from non-interacting isolated paramagnetic species (a situation
which is hardly realized in real transition metal oxide catalysts)
while paramagnetic bulk phases frequently present in those
materials are poorly suitable for EPR detection due to strong
magnetic interactions between paramagnetic centres giving rise
alytic reactions.
The heatable reaction tube (i.d. 3 mm) was
placed by ground joints into an evacuated quartz Dewar for
protecting the EPR cavity from high temperatures. The tem-
perature was controlled using an Eurotherm 902-904 tempera-
ture programmer and a Pt/Rh thermocouple. The reactant gas
flow was mixed by a gas/liquid dosing system containing mass
flow controllers (Bronckhorst) and thermostated saturators.
For on-line product analysis in the case of propane dehydro-
genation, the reactor outlet was connected to a GC 17AAF
capillary gas chromatograph (Shimadzu) equipped with a
FID and a 30 m ꢁ 0.32 mm Silicaplot column (Chrompack).
In the case of toluene oxidation, product analysis was per-
formed off line after collecting the products in a cold trap filled
with ethanol using a 25 m ꢁ 0.25 mm SE-54-CB column.
For simultaneous recording of UV-vis reflectance spectra
during oxidative and non-oxidative dehydrogenation of pro-
pane, a fibre optics AVS-PC-2000 plug-in spectrometer
(Avantes) equipped with a CCD array detector responsive
from 200 to 1100 nm was used. The two channels of the spec-
trometer (master: 200–500 nm; slave: 500–1100 nm) as well as
a DH-2000 deuterium–halogen light source were connected to
a cylindrical quartz sensor (Optran WF, length 200 mm, dia-
meter 1.5 mm) by fibre optic cables (length 2 m) consisting
of a core of pure silica (diameter 0.4 mm) coated with poly-
imide. The sensor fits into the reactor through a Teflon sealing
disk which is fixed by a screwing at the top end of the reaction
tube. The tip of the sensor is a plane polished surface. It is
placed within the catalyst bed. The feed-through of the ther-
mocouple at the bottom end of the reaction tube is realized
in the same way (Fig. 1).
3
9
to line broadening. It will be demonstrated below that it is
just this property which can be exploited to learn more about
catalyst–reactant interactions in paramagnetic bulk phases
and clusters of TMI. On the other hand it is true, that operando
EPR at elevated temperatures is restricted to the detection of
paramagnetic TMI with sufficiently long relaxation times
and does not see diamagnetic TMI in high oxidation states
5
+
6+
6+
(
e.g. V , Cr , Mo ) which, however, may play an impor-
tant role in the catalytic cycle. It will be shown below that this
drawback can be beneficially circumvented by simultaneously
coupling EPR and UV-vis diffuse reflectance spectroscopy
2
8
(
DRS).
It is the aim of this paper to demonstrate the benefits that
can be drawn from operando and in situ EPR studies for eluci-
dating structure–reactivity relationships in heterogeneous cat-
alytic gas-phase reactions performed over transition metal
oxide catalysts. A short section presenting suitable experimen-
tal setup is followed by selected examples in which the infor-
mation potential as well as limits of the application of
operando EPR are illustrated. These comprise the action of dif-
ferent TMI: (i) structure and function of V ions in vanadia-
based catalysts used for the oxidative dehydrogenation of
propane to propene (ODP) and the selective oxidation of
toluene to benzaldehyde; (ii) activity and stability of Cr sites
in supported chromium oxides during dehydrogenation of pro-
x
pane; (iii) structure and function of Mn ions in MnO storage
catalysts for selective catalytic reduction of NO . Benefits aris-
x
ing from coupled operando EPR/UV-vis/on-line GC studies
are also shown. This novel combination provides an excellent
opportunity to monitor the reaction-dependent interconver-
sion of paramagnetic and diamagnetic TMI and, thus, widens
the variety of spectroscopically accessible species.
2
. Experimental
The vast majority of heterogeneous catalytic reactions that use
transition metal oxide catalysts, proceed at elevated tempera-
tures. To follow these processes by in situ EPR spectroscopy,
suitable reaction cells have to be implemented into the EPR
cavity that can be heated and connected to some equipment
for dosing reactant mixtures and analyzing product composi-
tions. Different experimental solutions being restricted to X-
9
,40,41
band continuous wave EPR have been recently reviewed.
In some cases, commercially available variable-temperature
control units (e.g. Wilmad Glass Co.) or high-temperature cav-
ities (e.g., Bruker) equipped with some specially designed
home-made flow tubes have been used to which heat is trans-
Fig. 1 Experimental set-up for simultaneous operando EPR/UV-vis/
on-line GC measurements consisting of a fixed-bed flow reactor heated
with a bifilar winding of Pt wire and an implemented UV-vis fibre optic
sensor.
9
ferred by a pre-heated stream of inert gas. Temperature
4
462
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472

View Online
1
9
3
. Application examples
during the ammoxidation of toluene. It was attributed to a
perturbation of the spin–spin exchange between neighbouring
2
+
3.1. Structure and function of V ions in vanadia-based
selective oxidation catalysts
2 2 7
VO ions in the vanadyl double chains of the (VO) P O
structure. This perturbation is caused by transfer of electron
density from the adsorbed aromatic ring to surface vanadyl
sites and further delocalization over a certain range of the
vanadyl chains. From Fig. 2(b) and (c) it is seen that the effect
is even more pronounced when the basicity of the aromatic
ring is enhanced by additional electron donating substituents
such as a 4-methoxy group.
While the spectral changes occur immediately upon repla-
cing the nitrogen flow by the reactant mixture, the shape of
the EPR signal is only slowly reversible when changing from
the reactant mixture (toluene/air) back to nitrogen flow (Fig.
Selective oxidation of toluene to benzaldehyde. From an
industrial point of view, benzaldehyde is a very important
aromatic aldehyde. A possible way of production is selective
gas-phase oxidation of toluene using catalysts based on vana-
dia. However, low conversion rates have to be maintained to
avoid deeper oxidation and even in this case the benzaldehyde
selectivities are low. Thus, improvement of the catalytic per-
formance by rational catalyst design is of considerable indus-
trial interest. Comprehensive studies of the ammoxidation of
toluene to benzonitrile proceeding with high yields over unsup-
ported vanadium phosphorus oxide catalysts, e.g. over (VO)2-
4
5
2
(c)). This indicates that the aromatic reactant and/or product
is very strongly adsorbed and difficult to remove from the
catalyst surface. The temporal decrease of the EPR signal
intensity is the more pronounced, the higher the toluene con-
version and, as a consequence, the lower the benzaldehyde
selectivity (Fig. 3) indicating that more and more surface
2 7
P O , revealed that benzaldehyde is an intermediate in this
4
6,47
reaction.
absence of NH
Thus, it was expected that (VO)
2 2 7
P O in the
3
should be a suitable material, too, to catalyze
the selective formation of benzaldehyde. However, in con-
trast to the ammoxidation of toluene, in which selectivities
as high as 90% were obtained at conversions of up to 70%,
maximum benzaldehyde selectivities in the selective oxidation
of toluene did not exceed 15–47%, and these values were only
2
+
VO sites are involved in the reaction cycle and, thus, contri-
bute to the spin-spin exchange perturbation. FTIR studies per-
formed on the same system revealed that, besides adsorbed
benzaldehyde, cyclic anhydrides are formed by deeper oxida-
tion of the aromatic ring which finally might be converted to
4
8
achieved at toluene conversions as low as 5%. By operando
EPR studies, supplemented by FTIR investigations, it was
possible to elucidate reasons for this disappointing result and
to suggest opportunities for improving benzaldehyde yields.
Vanadyl pyrophosphate is a paramagnetic bulk phase in
4
8
CO
x
.
As a reason for this undesired strong adsorption it
was found that P–O–P and/or P–O–V bonds of the catalyst
surface are hydrolyzed by water which is a product of the oxi-
dation reaction. This hydrolysis creates Brønsted acidic sites
that form hydrogen bonds with the carbonyl group of benzal-
dehyde and, thus, hinder its desorption and facilitate total
4
+
which the EPR-active V
6
O octahedra form ladder-like dou-
ble chains along which they are coupled by magnetic spin–spin
exchange interactions. Due to these interactions, splittings
arising from the spin–orbit and electron-nuclear spin inter-
action (reflected by the principle components of g and A ten-
sors, respectively) are not resolved and a rather narrow
isotropic singlet is observed in the operando EPR spectrum
at 658 K (Fig. 2). As soon as the inert nitrogen atmosphere
is replaced by the feed mixture (1 mol% toluene or 4-methoxy-
toluene, respectively, in air) the apparent height of the EPR
signal decreases (Fig. 2(a) and (b)). This is also true for the
relative double integrals (Fig. 2(c)) since marked line broaden-
ing leads to the disappearance of intensity in the outer wings of
the spectrum. Thus, only a part of the EPR intensity is com-
prised by double integration. This effect has also been observed
in previous operando EPR studies of vanadyl pyrophosphate
4
8
oxidation.
For 4-methoxytoluene and its corresponding aldehyde the
interaction with the Brønsted acidic sites is even stronger as
for pure toluene due to the inductive effect of the substituent.
2
+
The VO operando EPR signal is superimposed by a very nar-
row line which arises from carbon radical species generated by
oxidative degradation of the aromatic ring system (Fig. 2(b)).
It is interesting to note that, on the catalysts discharged from
the reactor after reaction, no coke deposits are detected. Thus,
the carbon radical species have to be regarded as intermediates
in the total oxidation of the aromatic hydrocarbons for which
operando EPR is a unique detection method. They are only
hardly visible in toluene/air flow (Fig. 2(a)), for which total
oxidation is less pronounced. Analysis of the product mixture
leaving the EPR flow reactor revealed that benzaldehyde was
formed with 20% selectivity at a toluene conversion of 26%
(
8
Fig. 3) while 4-methoxybenzaldehyde was obtained with only
% selectivity at the same degree of conversion. In contrast to
ꢂ
Fig. 2 EPR spectra of (VO)
2
P
2
O
7
(70 mg) measured at 385 C in N
flow and reactant feed: (a) 1 mol% toluene/air, (b) 1 mol% 4-methoxy-
2
Fig. 3 Relative decrease of the EPR signal intensity (double integral)
of (VO) (70 mg) in a flow of 1 mol% toluene/air (total flow: 34
ml min ) (white bars, Irel(N
hyde selectivity (black bars) versus toluene conversion obtained after
1 h time on stream (values determined by GC analysis of the product
mixture).
2 2 7
P O
ꢃ1
ꢃ1
toluene/air, total flow: 34 ml min ) and relative intensity of the EPR
2
) ¼ 100%) and corresponding benzalde-
signals (double integrals) as a function of time on stream and feed
composition (open symbols: 1 mol% toluene/air, filled symbols: 1
mol% 4-methoxytoluene/air).
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
4463

View Online
dehydrogenation of propane to propene. This could be an
2
+
toluene, the VO EPR signal is not reversible when switching
the gas flow from 4-methoxytoluene/air back to nitrogen con-
firming much stronger product adsorption in the latter case
interesting alternative to the currently applied cracking pro-
cesses since it uses cheap and environmentally friendly starting
materials. However, there is a major drawback that keeps the
ODP reaction still far from being attractive for industrial
application: Maximum propene yields obtained so far hardly
exceed 20% since total combustion of both propane and pro-
pene leads to low selectivities, in particular at higher degrees
of conversion. Based on a comprehensive study of structure–
reactivity relationships in bulk MeVO phases (Me ¼ Mg,
(
Fig. 2(c)). Flowing air is needed to burn off the adsorbates.
It is very probable that the higher benzonitrile selectivities
observed in the ammoxidation of toluene over (VO) P O
2 2 7
are due to facilitated product desorption which might arise
from blocking of the generated Brønsted sites by ammonia.
Following this idea, pyridine was added to the feed in the selec-
tive oxidation of toluene to act as a competitive adsorbate
which displaces benzaldehyde from the acidic surface sites
and, thus, prevents its total oxidation. This approach led to
an increase of the benzaldehyde selectivity by a factor of 2.2
at the same degree of toluene conversion. From Fig. 4 it is seen
that the operando EPR signal intensity decreases as expected
x
Zn, Pb) and first promising results with a supported VO /
2
3
MCM catalyst it was assumed that highly dispersed, prefer-
ably tetrahedrally coordinated V sites supported on high sur-
face area carriers of low surface acidity could improve the
propene selectivity. In this section, simultaneous operando
EPR/UV-vis/on-line GC measurements performed with the
novel setup shown in Fig. 1 are presented, by which key proper-
ties of good ODP catalysts could be identified.
when switching from N
2
to toluene/air flow. No further
changes are observed when water vapour is added to the feed
by passing a part of the air flow through a water-filled satura-
tor. However, when a 4 wt% pyridine/water solution is used
instead of pure water, the EPR signal intensity starts returning
to its initial value. This indicates that benzaldehyde may be
Catalysts containing VOx species on mesoporous SiO2
(SBA-15 and MCM-48) and mesoporous Al
prepared by wet impregnation with NH VO as described else-
2 3
O supports were
4
3
2
+
24
displaced from the VO surface adsorption sites removing
the temporal perturbation of the spin–spin exchange is reversi-
ble. This conclusion was also confirmed by in situ FTIR mea-
surements, in which adsorbed benzaldehyde could be no longer
where. In the case of SBA-15, support materials with two dif-
ferent pore diamters (190.5 and 52.6 nm) were used to check
the influence of the latter on the catalytic performance. In
the following, the SBA-15-supported catalysts are denoted as
4
8
detected in the presence of pyridine.
x x
VO /SBA-15/200 and VO /SBA-15/50. Structural and cata-
In summary, it has been shown that operando EPR investi-
gations supplemented by FTIR measurements were able to
identify strong product adsorption as the major selectivity lim-
iting factor in the selective oxidation of toluene to benzalde-
hyde. Blocking of Brønsted surface sites formed under
reaction conditions by competitive adsorbates such as pyri-
dine, which is not oxidized under the applied reaction condi-
tions, has been shown to be a possible way of improving the
catalytic performance in this reaction. Alternatively, other
catalysts which do not tend to the easy formation of Brønsted
acidic sites have been suggested following the conclusions of
the above described studies. First promising results were
lytic properties of these materials are listed in Table 1. In the
fresh catalysts, the vanadium species are mainly pentavalent
giving rise to intense charge-transfer (CT) transitions in the
UV-vis diffuse reflectance spectra (Fig. 5). From the position
of the low-energy CT band conclusions on the coordination
number and the degree of V site agglomeration can be derived.
In the spectra of fresh VO /MCM-48 and VO /SBA-15/200
x
x
samples which are very similar these bands occur above 450
nm suggesting octahedral coordination (Fig. 5(A)). After
4
9
ꢂ
heating in air to 500 C these signals disappear since the octa-
hedral V sites lose coordinated water ligands and become
tetrahedral. Accordingly, the most intense low-energy CT
band occurs at 320 nm being characteristic of mainly isolated
1
9
2 5
obtained with V O modified by alkali ions. Moreover, it
has been clearly shown that operando EPR is a valuable tool
not only for monitoring more or less isolated vanadyl ions,
as will be demonstrated in the following section, but also for
following reaction-dependent changes of pure paramagnetic
bulk phases by analyzing their intrinsic magnetic interactions.
4
tetrahedral VO sites. This process is reversible upon rehydra-
tion at room temperature. A similar reversible change between
tetrahedral and octahedral symmetry of the V centres due
to desorption/readsorption of water molecules was also
5
0,51
observed by other authors using in situ-UV-vis-DRS.
VO /SBA-15/200, the small band around 380 nm indicates
the presence of some VO sites connected via V–O–V bridges,
too (Fig. 2(a) ). The latter band is not visible in VO /MCM-
8, probably due to its very high surface area (Table 1) which
In
x
Oxidative dehydrogenation of propane to propene (ODP).
Vanadia-based catalysts are used, too, for the oxidative
4
5
2
x
4
facilitates high V dispersion. These spectral changes are com-
pletely reversible by rehydrating the samples in ambient atmo-
sphere suggesting that all V sites are exposed on the surface
and accessible to water ligands and, thus, also to potential
x
reactant molecules. Thus, it is justified to calculate VO surface
densities and turnover frequencies (TOF) assuming that all V
sites are exposed (Table 1).
Virtually no spectral changes are observed when heating
VO
are tetrahedrally coordinated as evidenced by CT bands
x 2 3
/Al O in air (Fig. 5). In this catalyst, most of the V sites
4
9,52
located below 400 nm.
However, a weak band at 455 nm
points to the presence of a small amount of octahedrally coor-
dinated, rather oligomeric vanadium sites which do not
become tetrahedral upon heating (Fig. 5(B)).
From the UV-vis-DRS measurements it can be concluded
that the degree of V sites connected via V–O–V bridges
Fig. 4 Relative intensity (double integral) of the EPR signal of 70 mg
ꢂ
(VO)
experiments (1: white symbols, 2: black symbols) upon switching the
2
P
2
O
7
at 385 C as a function of reaction time in two different
increases in the order VO
/MCM-48 ꢄ 0 < VO
x
/SBA-15/
x
2
the supports which decreases in the same order (Table 1).
00 < VO /Al O . This might be due to the surface area of
ꢃ1
x
2
3
feed composition (total flow: 34 ml min ) subsequently from N
2
to
air/toluene ¼ 100/1 to air/toluene/water vapour ¼ 100/1/1 (experi-
ment 1, white symbols) or air/toluene/water vapour (containing 4
wt% pyridine) ¼ 100/1/1 (experiment 2, black symbols) and again
Operando EPR/on-line-GC/UV-vis results under ODP con-
. UV-vis spectra of
ditions are plotted in Fig. 6 for VO
ꢂ
x
/Al
2
O
3
5+
to N
2
.
the fresh catalyst at 20 C are dominated by CT bands of V
.
4
464
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472

View Online
ꢂ
Table 1 Structural properties of catalysts with 2.8 wt% of V and catalytic results obtained at 500 C: Turn-over frequencies (TOF) and propene
selectivities (S) measured at propane conversions of 2.5–3.9%, maximum propene yields (Y) and space-time yields (STY) measured at propane
conversions between 20 and 30%
a
Surface density /
Pore
volume/cm g
Mean pore
˚
diameter/A
Y
max
(%)
STYmax/kgC H6
3
ꢃ2
2
ꢃ1
3
ꢃ1
a
TOF /s
ꢃ1
ꢃ1
Sample
V nm
S
BET/m g
S (%)
(kgcat h)
VO
VO
VO
VO
x
x
x
x
/Al
2
O
3
1.0
273
645
421
889
0.38
0.70
1.74
0.39
48.2
52.6
0.44
0.18
0.26
0.21
73.3
83.3
82.3
80.1
12.3
14.5
12.4
18.0
5.8
6.8
/SBA-15/50
/SBA-15/200
/MCM48
0.43
0.7
190.5
26.2
5.8
12.6
0.37
a
Apparent values, calculated assuming exposure of all V sites
ꢂ
By raising the temperature stepwise to 400 C, light absorption
increases gradually above 500 nm due to partial reduction of
conditions. From Fig. 7 it can be seen, that carbon deposits
can be at least partly removed by hydrogenation.
5
+
4+
to V , the d–d transitions of which fall in the higher
V
Catalytic tests revealed that the intrinsic activity of the VO
sites reflected by TOF values as well as the propene selectivities
do not differ much for the three silica-supported VO catalysts
x
5
3
wavelength range of the spectrum (Fig. 6, left). In the corre-
sponding EPR spectra, this reduction which starts already well
below the onset of propene formation, is reflected by a growing
EPR signal of octahedrally coordinated interacting and
x
(Table 1). This agrees well with the fact that their local struc-
ture and valence state under reaction conditions is also very
similar. Due to the much higher surface area of sample VOx/
MCM-48, the maximum propene yield achieved with this cat-
2
+
isolated VO species (Fig. 6, middle).
ꢂ
Further heating above 400 C gives rise to a strong increase
of absorbance in the whole visible range of the UV-vis spec-
trum (Fig. 6, left). This is caused by the formation of carbon
deposits which have been detected, also, by FTIR spectro-
scopy. Following the product composition by on-line-GC
shows that propene selectivity is strongly increasing in the
initial period of the reaction. Moreover, the catalyst is not
deactivated by the carbon deposits (Fig. 6, right). Obviously,
those carbonaceous residues are mainly deposited on the sup-
alyst is higher in comparison to the two VO /SBA-15 samples.
x
The intrinsic activity of VO
higher in comparison to that of VO /SBA-15/50, VO /SBA-
x 2 3
/Al O (TOF values, Table 1) is
x
x
15/200 and VO
described above, a certain amount of octahedral vanadium
sites is present under reaction conditions in VO /Al . These
species and the higher number of V–O–V bonds might be the
reason for the higher intrinsic activity of the VO sites in com-
parison to VO /SBA-15 and VO /MCM-48. However, pro-
pene selectivities over VO /Al are lower than over the
x
/MCM-48. As shown by the operando studies
x
2 3
O
x
port material while the active VO
x
species remain free. This
x
x
might reduce the surface acidity of the support under reaction
conditions and, thus, favour rapid desorption of the basic
propene molecules which reflects itself in increasing selectivity.
A rather similar behaviour in the operando-EPR/on-line-
GC/UV-vis experiments was also observed for the silica-
supported samples with the exception that for these catalysts
x
2 3
O
silica-supported samples. This might be due to the higher con-
centration and strength of Lewis acidic sites that have been
detected on VO
adsorption.
x 2 3
/Al O by FTIR spectroscopy of pyridine
From the above described operando experiments in connec-
tion with results of catalytic tests, the following conclusions
can be derived: under reaction conditions, the V sites being
mainly pentavalent in the fresh catalysts are considerably
reduced which might improve the propene selectivity due to
the lower redox potential. Another beneficial effect on selectiv-
ity is obviously caused by carbon deposits formed under pro-
pane-rich reaction conditions which do not deactivate the
catalysts since they might rather cover acidic sites of the
support materials while active V sites remain free. These
are unique observation that are only possible by operando
only a very small (VO /SBA-15/200, Fig. 7) or even no
x
x
EPR signal (VO /MCM-48) could be observed upon reduc-
5
+
tion of V (evidenced in the corresponding UV-vis spectra).
This is due to the fact that in these samples the active V sites
are essentially tetrahedrally coordinated at reaction tempera-
4
+
tures (Fig. 5) and V
in tetrahedral symmetry has short
relaxation times and is only visible at low temperatures. To
illustrate the strong influence of carbon deposits on the UV-
vis spectra, the gas flow over sample VO /SBA-15/200 was
x
2
switched to pure H after 1 h time on stream under ODP
Fig. 5 Room-temperature UV/VIS-DRS spectra of samples 2.8% V/Al
ꢂ
O
2 3
(left), 2.8% V/SBA-15/200 (middle) and 2.8% V/MCM-48 (right)
before (A) and after heating in air at 500 C (B).
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
4465

View Online
Fig. 6 Operando EPR/UV-vis/on-line GC measurement of sample VO
, W/F ¼ 1.25 g h mol ).
x
/Al
2
O
3
(5.8 wt% V) during ODH of propane (feed: 60% C
3
H
8
2
/30% O /
ꢃ1
N
2
EPR/UV-vis/on-line-GC experiments. Furthermore it is
obvious that V sites with more than four oxygen ligands are
more active but less selective than tetrahedral VO species. Iso-
well-defined and contains foam-like large-sized cavities which
are three-dimensionally interconnected. Moreover, the mean
pore diameter in the two SBA-15 catalysts is markedly higher
than in MCM-48 which leads to higher pore volumes (Table 1).
It is possible that, at the same flow rate, more propane passes
the larger pores of the SBA-15-supported catalysts without
conversion in comparison to VO /MCM-48. This is illustrated
x
by the values in Table 2. Thus, lower yields and STY values
result for SBA-15 samples.
4
lated VO
nected via V–O–V bonds. The beneficial effect of supported
VO catalysts with high surface area is first of all reflected
by maximum achievable space–time yields. In particular, for
VO /MCM-48 the STY value (Table 1) is much higher than
for bulk VMgO catalysts which, so far, belong to the best cat-
x
sites are selective but less active than V sites con-
x
x
2
3
alytic systems for the ODP. However, it is also roughly twice
as high compared to the two SBA-15-supported catalysts
although the structure of the V sites (Fig. 5) as well as the
selectivities at low conversion (Table 1) are rather similar.
The main difference between MCM-48 and SBA-15 is related
to the pore structure. While MCM-48 contains a three-dimen-
sional cubic system of unimodal mesopores, SBA-15/50 is
characterized by well-ordered hexagonal arrays of cylindrical
parallel-oriented pores which are interconnected by smaller
sized windows. The expanded SBA-15/200 structure is less
4
In summary, catalysts containing highly dispersed VO sites
with a mean valence state close to +4 on non-acidic supports
of very high surface area seem to be promising materials on
which future catalyst development should be focused.
3.2. Activity and stability of Cr sites in supported chromium
oxides during dehydrogenation of propane
Supported chromium oxides are industrially important cata-
lysts being used, e.g., in several industrial processes for the
Fig. 7 Operando EPR/UV-vis/on-line GC measurement of sample VO
O
x
/SBA-15/200 (2.8 wt% V) during ODH of propane (feed: 60% C
3
H
8
/30%
ꢃ1
2
/N
2
, W/F ¼ 1.25 g h mol ).
4
466
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472

View Online
2
of sample Cr/La,ZrO is due to the support.) This reduction is
Table 2 Propane conversion and propene selectivity for the three
ꢂ
mesoporous silica supported catalysts at 500 C under identical reac-
ꢃ1
well reflected in the related EPR spectra in which two different
0
tion conditions (W/F ¼ 1.4
helium ¼ 2/1/2)
g
h
(molC3H8
)
,
propane/oxygen/
signals can be distinguished at effective g-values of g ꢄ 4.3 and
0
g ꢄ 2. They are assigned to coordinatively unsaturated, iso-
3
lated Cr ions in distorted geometry on the catalyst surface
+
x x x
VO /MCM-48 VO /SBA-15/50 VO /SBA-15/200
3+
and to magnetically interacting Cr ions in Cr
2
O
3
clusters,
The former are most pronounced in Cr/
and almost not visible in Cr/Al . This points to
the stabilizing effect of La dopants being able to retain Cr spe-
5
9,60
respectively.
La,ZrO
X
C3H8 (%) 32.8
29.2
48.9
25.9
47.8
2 3
O
SC3H6 (%) 53.0
2
2
7,57
cies in high dispersion on the surface.
reduced catalysts in oxygen flow at 500 C shows that Cr spe-
Treatment of the
5
4
ꢂ
dehydrogenation of propane. Moreover, they revealed to be
promising materials for the aromatization of alkanes leading
to considerably higher selectivities than obtained with com-
cies in Cr/La,ZrO are almost completely reoxidizable while
2
this ability is partially lost for Cr/La,Al O and even more
2
3
5
5,56
mon reforming catalysts based on noble metals.
nately, they deactivate rather quickly by coke formation.
Previous operando EPR investigations of CrO /ZrO catalysts
in the aromatization of octane revealed that isolated CrO sites
are less active but more stable against coke deposition than
Unfortu-
2 3
for Cr/Al O . This is most probably due to the fact that
3
+
Cr ions formed upon reduction tend to be incorporated in
Al lattice positions of the catalyst volume and, thus, are no
longer accessible by reactants. It seems that La-doping can
suppress this effect only partly.
x
2
x
2
7
Cr
alysts with lanthanum stabilizes the CrO
2
O
3
clusters. Moreover, it was found that doping the cat-
sites on the sur-
When the catalysts are heated in a mixture of 23% propane/
N2 (W/F ¼ 16.2 g h mol ), the CT bands of chromate
ꢃ1
x
5
face. The operando EPR/UV-vis/on-line-GC technique was
7
3+
around 370 nm vanish and d–d bands of Cr appear in the
used to study the influence of the support material on activity
UV-vis spectra (Figs. 9 and 10, left) as observed, too, upon
treatment in H (Fig. 8). Simultaneously, the narrow EPR
and stability of the CrO
Three different catalysts have been studied. 1 wt% Cr/Al
x
sites.
2
5
+
0
2
O
3
2
singlets of Cr disappear and a broad singlet at g ꢄ 2 arises
2
ꢃ1
3+
from weakly interacting Cr species (Figs. 9 and 10, middle).
(
g
7
S
¼ 200 m g ) and 0.5 wt% Cr/La,ZrO (S
¼ 104 m
BET
ꢃ1
2
BET
) were prepared by impregnating g-Al
2
O
3
and a commercial
wt% La O /Zr(OH) support (MEL Chemicals), respec-
By comparing these results with the catalytic data (Figs. 9 and
10, right) it is clearly evident that the reduction of Cr and
6
+
2
3
4
5
+
3+
tively, with an aqueous (NH
ꢂ
4
)
cination in air at 600 C. Cr/La,Al O (S
2
CrO
4
solution followed by cal-
Cr to Cr occurs already at temperatures well below the
onset of the dehydrogenation reaction. This information can-
not be derived post mortem from measurements of quenched
catalysts and shows that Cr sites in oxidation states higher
than +3 do not exist under reaction conditions. With increas-
ing time on stream at 810 K the propane conversion over the
2
ꢃ1
¼ 345 m g ) was
2
3
BET
obtained by thermal decomposition of ammonium dawsonite,
NH Al(OH) CO , doped with 10 wt% Cr and 3.9 wt% La.
4
2
3
Prior to the operando EPR/UV-vis/on-line-GC experi-
ments, the redox behaviour of the catalysts was studied sepa-
rately by EPR and UV-vis spectroscopy (Fig. 8). UV-vis
spectra of as-synthesized catalysts are characterized by intense
Cr/La,ZrO catalysts drops markedly (Fig. 9). This is due to
2
3
the partial coverage of active Cr sites by coke deposits.
+
27
5
8
3+
CT bands of hexavalent chromate species at 370 nm. In the
corresponding EPR spectra a narrow line is detected arising
As a consequence, the EPR singlet of Cr loses intensity since
magnetic interaction with paramagnetic coke species may
cause line broadening. In the UV-vis spectra, deactivation
leads to a gradual increase of absorbance in the visible range
of the spectrum with a maximum around 470 nm arising from
higher condensed carbon species such as polyaromatics and
5
from traces of isolated Cr ions. After heating in hydrogen
+
ꢂ
flow to 500 C, the chromate CT bands in the UV-vis spectra
3
disappear and very broad bands of d–d transitions of Cr
+
5
8
6+
appear around 630 nm indicating that Cr is completely
reduced. (The remaining band around 290 nm in the spectrum
6
1
condensed rings.
Fig. 8 Room-temperature UV-vis (left) and EPR spectra (right, 3 mT < B
ꢂ
0
< 690 mT) of Cr/La,ZrO
2
(A), Cr/Al
2
O
3
(B) and Cr/La,Al
ꢂ
2
O
3
(C) in
flow (dashed lines) and after 1 h reoxidation at 500 C and cooling
as-synthesized form (solid lines), after 1 h reduction at 500 C and cooling in H
2
2
in O flow (dotted lines).
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
4467

View Online
Fig. 9 Operando EPR/UV-vis/on-line GC measurement of sample Cr/La,ZrO
, W/F ¼ 16.2 g h mol ).
2
during dehydrogenation of propane (feed composition: 23%
ꢃ1
C
3
H
8
/N
2
3
The recoverage of Cr in oxidative atmosphere is surpris-
ing. Under these conditions, Cr ions can only be formed
by oxidation of deeper reduced Cr species which are EPR-
silent. This means that during propane dehydrogenation Cr
species on Al O supports may be reduced to lower oxidation
2 3
states than +3. This is supported by the results of quasi-in situ-
XPS measurements, in which the catalyst has been treated in a
+
In comparison to sample Cr/La,ZrO
2
, the Cr/La,Al
2
O
3
cat-
3
+
alyst is, despite higher BET surface area and Cr content, less
active but more stable against deactivation (Fig. 10). Here,
the maximum of absorbance arising from carbon deposits is
shifted to lower wavelength around 360 nm being characteristic
of less condensed rather linear polyenylic species which might
be less deactivating. Besides the different nature of the carbon
species, the lower activity of the Cr/La,Al O catalyst is most
6
1
flow of 23% propane/N in a reaction cell installed in the lock
2
2
3
probably due to the fact that a considerable part of the Cr ions
migrates into the bulk of the support and, thus, is not accessi-
ble by reactants. As suggested by the non-reversibel redox
behaviour, this effect should be most pronounced for lantha-
to the analysis chamber. After subsequent cooling to room
temperature the sample was transferred to the analysis cham-
ber without contact to ambient atmosphere. In the fresh cata-
lyst, the Cr 2p3/2 peak was found to be a superposition of
6
+
3+
num-free Cr/Al
tion/regeneration cycles have been followed by operando
2
O
3
(Fig. 8). For this sample, repeated reac-
contributions from Cr
and Cr
at binding energies of
579.4 and 576.8 eV (corrected with respect to the C 1s signal
at 284.5 eV). After treatment in the reactant gas mixture, the
peak is shifted to a binding energy of 574.5 eV which is mark-
ꢂ
EPR at 550 C (Fig. 11). It can be seen that during regeneration
3
+
5+
in air, only the signals of isolated Cr and Cr species reap-
pear the concentration of which is small. In contrast, the signal
at g ꢄ 2 due to interacting Cr is not influenced. This suggests
that only a minor amount of rather isolated Cr species remains
accessible on the surface accounting for the catalytic activity.
3
+
edly lower than expected for pure Cr suggesting that deeper
reduced Cr species must have been formed. However, the latter
are not stable in air at ambient atmosphere. After storing for
20 h in air at room temperature the Cr 2p3/2 peak is again
0
3+
Fig. 10 Operando EPR/UV-vis/on-line GC measurement of sample Cr/La,Al
, W/F ¼ 16.2 g h mol ).
2
O
3
during dehydrogenation of propane (feed composition: 23%
ꢃ1
C
3
H
8
/N
2
4
468
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472

View Online
ꢂ
Fig. 11 Left: Operando EPR spectra of sample 1 wt% Cr/g-Al
O
2 3
at 500 C after 30 min under reactant feed (thick lines, feed composition: 23%
ꢃ1
C , W/F ¼ 16.2 g h mol ) and subsequently after 30 min in air flow (thin-lines). Cycles were repeated three times. Right: propene yields
determined after 2, 15 and 30 min within each period under reactant feed.
3
H
8
/N
2
3
+
shifted back to 576.8 eV, indicating that Cr species of lower
valence states have been reoxidized to Cr
temperature. Upon heating in a flow of 0.5% NO/N
2
(100
0
3
+
ꢃ1
.
mg catalyst, total flow 7.5 ml min ) an EPR signal at g ꢄ 2
6
5
Based on the above described results, it can be concluded
that the lower activity of Cr/Al O and Cr/La,Al O in com-
appears (Fig. 12(a)). By spectra simulation it was found, that
this signal is a superposition of two sub-lines with hyperfine
structure characteristic of isolated octahedrally coordinated
2
3
2
3
parison to Cr/La,ZrO
face sites which is very likely caused by migration of Cr into
the volume of the Al support. This, in turn, leads to the
2
is due to the dilution of active Cr sur-
3
+
2+
Mn ions and a broad singlet arising from magnetically inter-
acting Mn ions. The starting temperature of Mn forma-
2
+
66
2+
2
O
3
ꢂ
formation of weaker condensed, less deactivating carbon
deposits. In addition, reduction of the Cr sites to valence states
lower than +3 may also contribute to the activity decrease.
Since isolated Cr surface sites revealed to be rather stable
against deactivation, a possible way of catalyst improvement
could be their stable dispersion on supports of high surface
area. These should expose a higher amount of such isolated
sites and guarantee sufficient overall activity.
tion (ꢄ225 C) agrees very well with the onset of NO
x
6
4
conversion observed during catalytic tests of these materials.
By related in situ FTIR measurements it has been shown that
ꢃ
intermediate NO
to NO3 ions which are stored at the Na ions of the zeolite
2
species are formed and further converted
ꢃ
6
support (eqn. (1)). From EPR results it is clear that this
process goes along with a reduction of Mn to Mn . This
means that lattice oxygen participates in this redox process.
5
4
+
2+
Osurface2
ꢃ
2ꢃ
Osurface
ꢃ
ꢃ
NO ƒƒƒƒƒƒƒ! NO2 ƒƒƒƒƒƒƒ! NO
ð1Þ
3
3
catalysts for selective catalytic reduction of NO
.3. Structure and function of Mn ions in MnO storage
x
ꢂ
x
After passing a maximum at ꢄ355 C the intensity of the EPR
2
+
signal decreases again suggesting that Mn is re-oxidized.
This is accompanied by thermal decomposition of the
adsorbed nitrate species.
The removal of NO from exhaust gases by selective catalytic
x
reduction (SCR) is one of the most important issues in envir-
onmental catalysis. In the case of diesel and lean burn gasoline
engines, special problems arise from the oxygen excess in the
exhaust gas. For effective NO abatement, periodic operation
x
under oxidative and reductive reaction conditions is applied
requiring catalysts which are able to store NO at lower tem-
x
peratures in oxygen excess and desorb it at higher tempera-
tures under the conditions of selective catalytic reduction.
6
2–64
Supported MnO
2
revealed to be a suitable catalyst.
ever, the nature of adsorbed NO species, the mechanism lead-
How-
x
ing to their formation and, in particular, the role of the
catalysts components in this process are not fully understood
so far. A further important issue is the stability of the storage
catalysts against deactivation by impurities present in the
exhaust gases such as sulfur-containing compounds or water
vapour. In situ EPR studies in connection with in situ FTIR
measurements have been performed to investigate the inter-
6
5
action of NO, O , H O and SO (used as reducing agent).
2
2
2
2
The 5 wt% MnO /NaY catalyst used in this study was
prepared by precipitating MnO on commercial zeolite NaY
2
6
4
2
or SiO (Aerosil).
Fig. 12 EPR spectra of MnO /NaY (100 mg) recorded during heat-
ing in a flow (7.5 ml min ) of (a) 0.5% NO/N
2
ꢃ1
The as-synthesized catalyst contains tetravalent manganese
ions which do not give rise to an EPR signal at ambient
2
, (b) 0.5% NO, 50%
.
O /N
2 2
and (c) 0.5% NO, 50% O
2
, 2.3% H
2
O/N
2
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
4469

View Online
When oxygen is added to the feed in a molar ratio of NO/
2
+
O ¼ 1/100, Mn ions start to form at lower temperature and
2
2
in a higher amount than in the absence of O (Fig. 12(b)). This
was quite surprising since the opposite was expected in oxidiz-
ing atmosphere. Corresponding FTIR experiments have
shown that nitrate formation is also strongly enhanced in the
6
5
presence of O
2
. The results of both methods support the con-
4
+
clusion that NO oxidation and, as a consequence, Mn
2
+
reduction to Mn proceeds via lattice oxygen of the MnO
component not only in NO/N flow but also in an excess of
gas-phase oxygen. In the latter case, Mn ions formed upon
NO oxidation can be readily re-oxidized by O and, thus, par-
ticipate in the redox cycle again. In the absence of O , gradual
reduction of MnO obviously leads to a saturation of the
surface with Mn being the reason why nitrate formation
does not exceed a certain limit.
When 2.3% water vapour is additionally added to the feed,
2
2
2
+
2
2
2
+
2
2
+
Mn formation is markedly suppressed (Fig. 12(c)). This sug-
gests that water is a competitive adsorbate for NO and blocks
a certain part of the active sites. A similar conclusion has also
been drawn from the FTIR results which indicated that less
nitrate is formed when pre-adsorbed water is present on the
surface. However, the inhibiting effect of water was found to
Fig. 13 EPR spectra of MnO
2
/NaY (100 mg) recorded during heat-
ing in a flow (7.5 ml min ) of 0.3% NO, 30% O /N (a) without SO ,
ꢃ1
2
2
2
(b) with 40 ppm SO
in a flow of 600 ppm SO
2
, (c) with 550 ppm SO
, 33% O
2
, and (d) after pretreatment
ꢂ
2
2
/He at 250 C.
nitrate. Since this conversion proceeds via lattice oxygen spe-
cies, it must involve reduction of Mn to Mn . These results
reflect one of the main problems of these NO storage cata-
be less pronounced in the presence of O
formation of HNO and finally HNO as sources for nitrite
2
which favours the
4+
2+
2
3
6
5
x
and nitrate (eqns. (2)–(4)).
lysts, namely, their sensitivity against sulfur-containing impu-
rities in fuels which poison the active Mn sites by sulfate
deposition.
By critically evaluating the results described above it is
evident that the major benefit does not arise from in situ
EPR alone but from the combination with in situ FTIR. Using
this approach, it was possible to follow both the behaviour of
the active Mn sites by EPR and the formation of adsorbed
intermediates by FTIR spectroscopy. Based on this complex
information, mechanistic insights could be obtained.
NO þ 0:5 O2 ! NO2
NO2 þ H2O ! HNO2 þ HNO3
NO2 þ H2O ! 2 HNO3 þ NO
2
ð2Þ
ð3Þ
ð4Þ
2
3
+
The spin-Hamiltonian parameters of the Mn centres have
6
5
been derived by spectra simulation. To achieve a satisfactory
fit of experimental and calculated spectrum, two slightly differ-
2
+
ent hfs signals for octahedrally coordinated isolated Mn
6
ions and one isotropic singlet for interacting Mn
7
2+
sites
had to be superimposed. The spin-Hamiltonian parameters
as well as the relative intensities of the three sub-signals are
very similar during treatment with the various feed composi-
tions indicating that the local structure of the respective
4. Summary: Opportunities and limitations of
operando EPR
Operando EPR, although still rather seldom used in catalysis
research so far, is a unique tool to follow the action of para-
magnetic transition metal ions (TMI) in oxide catalysts during
the catalytic process. This has been demonstrated by the appli-
cation examples presented in this paper. Experimental condi-
tions can be chosen almost identical to those in fixed-bed
microreactors, thus, making sure that the obtained results are
properly reflecting the true reaction behaviour. The method
provides a great deal of information on the role of paramag-
netic TMI in oxide catalysts which cannot be adequately
obtained by other methods. Thus, reaction-dependent changes
of coordination and valence state of TMI as well as their elec-
tronic interactions between each other and with electrons of
reactant molecules can be elucidated simultaneously. Using
vanadia-based selective oxidation catalysts as an example, it
has been shown that such investigations are not only restricted
to well isolated TMI, as frequently supposed. Rather it has
been demonstrated that closely neighbouring TMI which are
coupled by effective spin-spin exchange interactions in oxidic
clusters or even in bulk phases can be studied, too, by analyz-
ing temporal changes of the EPR line shape and intensity. In
fact, it is just these interactions which have been exploited to
derive conclusions on catalyst-reactant interactions in the
selective oxidation of toluene. The application examples pre-
sented above comprise only a limited selection out of the wide
variety of catalytically relevant TMI accessible by operando
EPR. A more comprehensive treatise of this issue is given in
ref. 40
2
+
Mn centres remains almost unchanged. The only difference
comes from the total EPR signal intensity (Fig. 12) which
4
+
reflects the extent of Mn reduction.
In the presence of SO , the extent of Mn formation by
reduction of Mn decreases markedly with increasing amount
of SO present in the feed (Fig. 13(a)–(c)). This is surprising
since by FTIR measurements it was found that, upon adsorp-
2
+
2
4
+
2
2
ꢃ
tion, SO
2
is converted to SO
4
(stored on both the Mn and
Na ions of the catalyst) the formation of which must involve
6
a redox step. A possible explanation for the partially sup-
6
2
+
pressed Mn formation can be given by eqns. (5)–(9). From
these equations it is seen that, in the presence of NO, O and
to sulfate ions needs not
2
traces of water, oxidation of SO
necessarily to involve Mn ions.
2
SO2 þ 2 NO3ꢃ þ 2 Hþ ! SO3 þ 2 NO2 þ H2O
ð5Þ
ð6Þ
ð7Þ
ð8Þ
ð9Þ
SO3 þ H2O ! H2SO4
SO2 þ H2O ! H2SO3
NO þ 0:5 O2 ! NO2
H2SO3 þ NO2 ! H2SO4 þ NO
From Fig. 13(b) and (c) it is also seen that, the more SO is
2
4
in the feed, the lower is the degree of Mn reduction. Further-
+
2
more, Mn formation in NO/O flow starts at higher tem-
+
2
peratures when the catalyst had been pretreated in a flow of
6
00 ppm SO /33% O /He (Fig. 13(d)). This suggests that the
2 2
sulfate species formed during this pre-treatment block active
Mn sites which are required for the conversion of NO to
Despite the undisputable benefits of operando EPR, the
major drawbacks of this technique have to be mentioned,
4
470
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472

View Online
too. The EPR signal intensity of paramagnetic species is inver-
sely proportional to the temperature. Although EPR is a
highly sensitive method, this may lead to poor signal-to-noise
ratios when catalytic reactions have to be studied at high tem-
peratures and the concentration of paramagnetic species is
low, i.e., in the case of isolated TMI. Moreover, certain TMI
can have very short relaxation times depending on their parti-
cular coordination symmetry and electronic configuration so
that at ambient or elevated temperatures no EPR signal can
References
1
2
3
Z. Sojka, Catal. Rev.-Sci. Eng., 1995, 37, 462.
K. Dyrek and M. Che, Chem. Rev., 1997, 97, 305.
K. Dyrek and M. Che, in Spectroscopy of Transition Metal Ions
on Surfaces, ed. B. M. Weckhuysen, P. Van Der Voort and
G. Catana, Leuven University Press, Leuven, 2000, p. 45.
M. Che and E. Giamello, Stud. Surf. Sci. Catal. B, 1989, 57, 265.
J. C. Vedrine, in Characterization of Heterogeneous Catalysts,
ed. F. Delannay, Marcel Dekker, Inc., New York, 1984, p. 161.
M. Labanowska, Chem. Phys. Chem., 2001, 2, 712.
4
5
4
+
6
7
be observed. This holds, e.g., for V in weakly distorted octa-
Z. Sojka and M. Che, C. R. Acad. Sci. Paris, Ser. IIc, Chim., 2000,
3, 163.
3
+
3+
hedral or tetrahedral symmetry, for V , Mn and for various
other TMI with an even number of unpaired electrons as well
as for most of the rare earth ions for which strong spin–orbit
coupling shortens the relaxation times. Weak dipolar or weak
spin–spin exchange interactions between TMI of medium
mutual distance which do not give rise to exchange-narrowing
can broaden the EPR signal beyond the limit of detection com-
pletely excluding the respective catalysts from in-situ EPR
measurements.
8
9
D. M. Murphy and C. C. Rowlands, Curr. Opin. Solid State
Mater. Sci., 2001, 5, 97.
A. Br u¨ ckner, in Spectroscopy of Transition Metal Ions on Surfaces,
ed. B. M. Weckhuysen, P. Van Der Voort and G. Catana, Leuven
University Press, Leuven, 2000, p. 69.
1
1
0
1
K. Nowinska and A. B. Wieckowski, Z. Phys. Chem. NF, 1989,
162, 231.
M. C. Paganini, L. Dall’Acqua, E. Giamello, L. Lietti, P. Forzatti
and G. Busca, J. Catal., 1997, 166, 195.
12 A. M. Prakash and L. Kevan, J. Phys. Chem., 1999, 103, 2214.
Besides certain paramagnetic TMI that escape EPR detec-
tion for reasons discussed above, all TMI in highest oxidation
states which do not possess unpaired electrons are not EPR-
1
1
1
3
4
5
A. Br u¨ ckner, Catal. Rev.-Sci. Eng., 2003, 45, 97.
B. M. Weckhuysen, Chem. Commun., 2002, 97.
M. A. Banares, M. O. Guerriero-P e´ rez, J. L. G. Fierro and
G. Garcia Cortez, J. Mater. Chem., 2002, 12, 3337.
H. Topsøe, J. Catal., 2003, 216, 155.
5
+
6+
active. However, those species (e.g. V , Cr ) may well be
involved in catalytic reaction steps. Visualizing their changes
requires the combination with other suitable spectroscopic
techniques. It has been demonstrated above that the coupling
of operando EPR/UV-vis-DRS/on-line GC is very beneficial
in this respect since it extends the variety of simultaneously
detectable valence states of TMI and their interconversion dur-
ing reaction. Other information that cannot be gained from
operando EPR concerns, e.g., the particular structure of dia-
magnetic intermediates and products adsorbed on the catalyst
surface which, however, is provided by operando FTIR spec-
troscopy. For elucidating comprehensive structure–reactivity-
relationships it is, thus, very helpful to study the respective
catalytic system by different in-situ methods. The advantages
1
6
17 A. Br u¨ ckner, B. Kubias, B. L u¨ cke and R. St o¨ ßer, Colloids Surf. A:
Physikochem. Eng. Aspects, 1996, 115, 179.
1
1
8
9
A. Br u¨ ckner, Appl. Catal. A: General, 2000, 200, 287.
A. Br u¨ ckner, A. Martin, N. Steinfeldt, G.-U. Wolf and B. L u¨ cke,
J. Chem. Soc., Faraday Trans., 1996, 92, 4257.
2
0
A. Br u¨ ckner, A. Martin, B. L u¨ cke and F. K. Hannour, Stud. Surf.
Sci. Catal., 1997, 110, 919.
21 K. M. Eriksen, R. Fehrmann and N. J. Bjerrum, J. Catal., 1991,
132, 263.
2
2
C. Oehlers, R. Fehrmann, S. G. Masters, K. M. Eriksen, D. E.
Sheinin, B. S. Bal’zhinimaev and V. I. Elokhin, Appl. Catal. A:
General, 1996, 147, 127.
2
3
P. Rybarczyk, H. Berndt, J. Radnik, M.-M. Pohl, O. Buyevskaya,
M. Baerns and A. Br u¨ ckner, J. Catal., 2001, 202, 45.
1
3
of this approach have been reviewed very recently.
24 A. Br u¨ ckner, P. Rybarczyk, H. Kosslick, G.-U. Wolf and M.
Baerns, Stud. Surf. Sci. Catal. B, 2002, 142, 1141.
All operando EPR investigations published so far have been
performed in X-band (microwave frequency: ꢄ9.5 GHz) using
the continuous wave (cw) technique. Although modern pulse-
EPR methods have rapidly developed in recent years, their
application with respect to TMI in solid catalysts is limited
since both, low concentration of paramagnetic species and
low recording temperatures being intrinsic requirements of
pulse-EPR echo techniques are far beyond the conditions
which have to be met for investigations of real catalysts during
reaction. Operando EPR measurements at higher microwave
frequencies (e.g. at ꢄ35 GHz in Q-band), which would be help-
ful in certain cases for assigning complex spectra, have not
been performed so far since the dimensions of the EPR cavity
set an upper limit for the diameter of a catalytic flow reactor.
With a similar equipment as shown in Fig. 1, operando mea-
surements in Q-band would require reaction tube diameters
of less than 1 mm which are not suitable for catalytic flow
reactors.
2
5
K. M. Sancier, T. Dozono and H. Wise, J. Catal., 1971,
3, 270.
Z. Sojka and M. Che, J. Phys. Chem., 1995, 99, 5418.
2
2
2
6
7
A. Br u¨ ckner, J. Radnik, D.-L. Hoang and H. Lieske, Catal. Lett.,
1
999, 60, 183.
28 A. Br u¨ ckner, Chem. Commun., 2001, 2122.
29 A. Br u¨ ckner, U. Lohse and H. Mehner, Microporous Mesoporous
Mater., 1998, 20, 207.
3
0
U. Lohse, A. Br u¨ ckner, E. Schreier, R. Bertram, J. J a¨ hnchen and
R. Fricke, Microporous Mater., 1996, 7, 139.
A. V. Kucherov, J. L. Gerlock, H.-W. Jen and M. Shelef, Zeolites,
3
1
1
995, 15, 9.
32 A. V. Kucherov, J. L. Gerlock, H.-W. Jen and M. Shelef, Zeolites,
1995, 15, 15.
3
3
A. V. Kucherov, J. L. Gerlock, H.-W. Jen and M. Shelef,
J. Catal., 1995, 152, 63.
K. Dyrek, D. Kiessling, M. Labanowska, G. Wendt and J.
Widziszewska, Colloids Surf. A: Physikochem. Eng. Aspects,
3
4
1
993, 72, 183.
35 M. Hunger and J. Weitkamp, Angew. Chem., Int. Ed., 2001,
40, 2954.
Despite the above mentioned limitations, operando EPR has
been shown to be a very useful tool for investigating, in parti-
cular, heterogeneous catalytic gas-phase reactions over transi-
tion metal oxide catalysts. Therefore, it is desirable that this
technique will receive more future attention.
3
6
W. Matir and J. H. Lunsford, J. Am. Chem. Soc., 1981, 103,
728.
J. H. Lunsford, Catal. Today, 1990, 6, 235.
3
3
3
7
8
H. G. Karge, J.-P. Lange, A. Gutsze and M. Laniecki, J. Catal.,
1
988, 114, 144.
3
4
9
J. C. Vedrine, in Characterization of Heterogeneous Catalysts,
ed. F. Delannay, Marcel Dekker, Inc., New York, 1984.
A. Br u¨ ckner, in In situ Spectroscopy of Catalytic Solids, ed.
B. M. Weckhuysen, Academic Science Publishers, in press.
Z. Sojka and M. Che, Appl. Magn. Reson., 2001, 20, 433.
D. J. E. Ingram, in Proceedings of the 3rd Conference on Carbon,
Pergamon Press, New York, 1959, p. 94.
0
Acknowledgements
4
4
1
2
The author thanks U. Bentrup, A. Martin, J. Radnik, D. L.
Hoang, I. Pitsch, R. Fricke and M. Baerns for fruitful coop-
eration. Financial support by the German Federal Ministry
of Education and Research (project no. 03C0280) is gratefully
acknowledged.
4
3
L. S. Singer, W. H. Smith and G. Wagoner, Rev. Sci. Instrum.,
1961, 32, 213.
44 A. Bruckner, B. Kubias and B. Lucke, Catal. Today, 1996,
¨
32, 215.
¨
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472
4471

View Online
57 A. Trunschke, D. L. Hoang, J. Radnik and H. Lieske, J. Catal.,
2000, 191, 456.
58 B. M. Weckhuysen, L. M. De Ridder and R. A. Schoonheydt,
J. Phys. Chem., 1993, 97, 4756.
59 S. De Rossi, G. Ferraris, S. Fremiotti, V. Indovina and A.
Cimino, Appl. Catal. A: General, 1993, 106, 125.
60 B. M. Weckhuysen, R. A. Schoonheydt, F. E. Mabbs and D.
Collison, J. Chem. Soc., Faraday Trans., 1996, 92, 2431.
61 H. G. Karge, M. Laniecki, M. Ziolek, G. Onyestyak, A. Kiss,
P. Kleinschmit and M. Siray, in Zeolites: Facts, Figures, Future,
ed. P. A. Jacobs and R. A. van Santen, Elsevier, Amsterdam,
1989, p. 1327.
62 F. Kapteijn, L. Singoredjo, M. van Driel, A. Andreini, J. A.
Moulijn, G. Ramis and G. Busca, J. Catal., 1994, 150, 105.
63 K. Eguchi, M. Watabe, S. Ogata and H. Arai, Bull. Chem. Soc.
Jpn., 1995, 68, 1739.
4
4
4
5
6
7
F. Br u¨ hne and E. Wright, in Ullmann, 6th edn., 1998, Electronic
Release (benzaldehyde entry).
Y. Zhang, A. Martin, H. Berndt, B. L u¨ cke and M. Meisel, J. Mol.
Catal.: A, 1997, 118, 205.
A. Martin, A. Br u¨ ckner, Y. Zhang and B. L u¨ cke, in Heteroge-
neous Catalysis and Fine Chemicals IV, ed. H. U. Blaser, A. Baiker
and R. Prins, Elsevier Science B. V., Amsterdam, 1997, p. 377.
A. Martin, U. Bentrup, A. Br u¨ ckner and B. L u¨ cke, Catal. Lett.,
48
49
50
1999, 59, 61.
G. Centi, S. Perathoner, F. Trifir o´ , A. Aboukais, C. F. Aissi and
M. Guelton, J. Phys. Chem., 1992, 96, 2617.
M. Mathieu, P. Van Der Voort, B. M. Weckhuysen, R. R. Rao,
G. Catana, R. A. Schoonheydt and E. F. Vansant, J. Phys. Chem.
B, 2001, 105, 3393.
51
M. Morey, A. Davidson, H. Eckert and G. Stucky, Chem. Mater.,
1996, 8, 486.
5
5
5
2
3
4
X. Gao and I. E. Wachs, J. Phys. Chem., 2000, 104, 1261.
C. J. Ballhausen and H. B. Gray, Inorg. Chem., 1962, 1, 111.
F. Buonomo, D. Sanfilippo and F. Trifir o´ , in Handbook of Hetero-
geneous Catalysis, ed. G. Ertl, H. Kn o¨ zinger and J.Weitkamp,
Wiley, 1996, p. 2140.
64 M. Richter, H. Berndt, R. Eckelt, M. Schneider and R. Fricke,
Catal. Today, 1999, 54, 531.
65 U. Bentrup, A. Br u¨ ckner, M. Richter and R. Fricke, Appl. Catal.
B: Environmental, 2001, 32, 229.
66 A. Br u¨ ckner, U. Bentrup, J. Radnik, M. Richter and R. Fricke,
4th Tokyo Conference on Advanced Catalytic Science and
Technology (TOCAT), July 14th–19th, 2002.
67 Z. Levi, M. Raitsimring and D. Goldfarb, J. Phys. Chem., 1991,
95, 7830.
5
5
6
K. Arata, M. Hino and M. Matsuhashi, Appl. Catal. A: General,
1
993, 100, 19.
H. Lieske and D. L. Hoang, Ger. Pat., DE 196 12 000 A1
18 March 1996).
5
(
4
472
Phys. Chem. Chem. Phys., 2003, 5, 4461–4472