The structure of active sites in Me-V-O catalysts (Me = Mg, Zn, Pb) and its influence on the catalytic performance in the oxidative dehydrogenation (ODH) of propane

Article abstract of DOI:10.1006/jcat.2001.3251

Me-V-O catalysts (Me = Zn, Mg, Pb) of different Me:V ratios were characterized with respect to phase composition, structure, and valence state of vanadium sites, as well as composition and acid-base characteristics of the surface. V⁵⁺ and V

Full text of DOI:10.1006/jcat.2001.3251

Journal of Catalysis 202, 45–58 (2001)
doi:10.1006/jcat.2001.3251, available online at http://www.idealibrary.com on
The Structure of Active Sites in Me –V–O Catalysts (Me = Mg, Zn, Pb)
and Its Influence on the Catalytic Performance in the Oxidative
Dehydrogenation (ODH) of Propane
P. Rybarczyk, H. Berndt, J. Radnik, M.-M. Pohl, O. Buyevskaya, M. Baerns, and A. Bru¨ckner1
Institut fu¨r Angewandte Chemie Berlin-Adlershof e. V., Richard-Willsta¨tter-Straße 12, D-12489 Berlin, Germany
Received November 10, 2000; revised April 21, 2001; accepted April 21, 2001
current knowledge on these subjects is still incomplete and
frequently contradictory.
Me–V–O catalysts (Me = Mg, Zn, Pb) of different Me : V ratios
were characterized with respect to phase composition, structure and
valence state of vanadium sites as well as composition and acid–
base properties of the surface using XRD, TEM, TPR, TPD/TPRS,
FTIR, potentiometric titration, XPS, EPR, and UV/VIS-DRS. The
latter two techniques were also applied in situ under catalytic re-
action conditions. The following structure–reactivity relationships
were derived by comparing the results of catalyst characterization
Mixed oxides based on vanadia and, in particular, Mg–
V–O compounds belong to the most effective catalysts for
the ODH of propane and have been frequently studied in
this reaction (1–11). Generally, vanadium ions are assumed
as active sites. However, their structure, in a particular co-
ordination and valence state, are still controversially dis-
cussed. Depending on the V/Mg ratio magnesium ortho-,
Mg3(VO4)2, pyro-, Mg2V2O7, and metavanadate, MgV2O6,
have been observed as crystalline phases besides MgO and
V2O5 in Mg–V–O materials. Kung and co-workers iden-
tified isolated tetrahedral VO34 units in Mg orthovana-
date to be the active and selective species in the ODH
of propane. They explained the higher selectivity of these
sites in comparison to dimeric tetrahedral V5+ in Mg py-
rovanadate by the presence of Mg–O–V units in the former
in which the vanadium ions are assumed to be less eas-
ily reducible than in V–O–V units in pyrovanadate (4, 5).
In contrast, Siew Hew Sam et al. regarded -Mg2V2O7 as
an active and selective phase and Mg3(VO4)2 as respon-
sible for total oxidation (6). High catalytic performance
of -Mg2V2O7 has been observed, too, by Guerrero-Ruiz
et al. who stated that the easier a catalyst is reduced, the
higher the propene selectivity (7). However, the opposite
relationship has been established by Kung and Kung (8)
who correlated dropping ODH selectivities to increasing
oxidation potentials of the metal cations in a number of
metal orthovanadates. They concluded that the easier it is
to reduce the metal vanadate, i.e., to remove lattice ox-
ide ions, the more likely the vanadate will react with the
hydrocarbon to result in carbon oxides. Pantazidis et al.
attributed high catalytic performance to the presence of
mixed and inhomogeneous MgO/Mg–V–O phases contain-
ing anion vacancies rather than to pure Mg vanadates
(9). A synergistic effect on the catalytic performance has
also been ascribed to the co-existence of two phases in
MgO/Mg3(VO4)2 and Mg2P2O7/Mg3(VO4)2 biphasic sys-
tems, although this phenomenon is not understood up
to now (10). In a recent in situ study using combined
5+
4+
with those of catalytic tests: Both V and V catalyze the ODH
4+
of propane; however, V seems to be more selective though less
active than V5+. V sites in octahedral or square pyramidal coor-
dination are more active but less selective than VO tetrahedra.
4
Isolated VO and V O units as present in crystalline metal ortho-
4
2 7
and pyrovanadates, respectively, are more selective but less active
than VOx species in amorphous clusters or even in crystalline chain-
or layer-like structures. The differences in the catalytic properties
arising from different metal cations are governed not only by the
oxidation potential of the latter but also to a major degree by the
acid–base properties, the crystal size and the extent of structural
disorder. Catalytic activity and selectivity of the Me–V–O catalysts
decrease in the order Mg > Pb > Zn. The lower activity of Zn–V–O
and Pb–V–O catalysts in comparison with Mg–V–O samples of sim-
ilar composition is assumed to be due to large crystallites with low
c
2001 Academic Press
surface defect concentration.
Key Words: metal vanadate catalysts; oxidative dehydrogenation
of propane; structure–reactivity relationships; in situ EPR; in situ
UV/VIS; quasi in situ XPS.
INTRODUCTION
The oxidative dehydrogenation (ODH) of lower alkanes
to the respective olefins being of high industrial interest has
been extensively studied since the 1970s and comprehen-
sively discussed in several review papers (1–3). Neverthe-
less, in propane ODH, maximum propene yields obtained
so far do not markedly exceed 20% . For rational catalyst
design, detailed information on the nature of the active sites
and their role in the catalytic cycle is needed. However, the
1 To whom correspondence should be addressed. Fax: +49 (0) 30 63 92
43 50. E-mail: ab@aca-berlin.de.
45
0021-9517/01 $35.00
c
Copyright
2001 by Academic Press
All rights of reproduction in any form reserved.

46
RYBARCZYK ET AL.
EXAFS/XRD measurements, the formation of a spinel- aqueous solution of NH4VO3. After evaporation in a rotary
type phase, MgV2O4, with trivalent vanadium has been evaporator, the resulting powder was dried at 383 K for 18
identified under reaction conditions (11).
h and calcined in air at 873 K for 3 h.
The discrepancies in the results obtained by different au-
Pb vanadate catalysts of Pb/V atomic ratios 3/2 (Pb3V2),
thors from investigations of similar catalysts might arise 1/1 (Pb1V1), and 5/4 (Pb5V4) were prepared by adding ap-
from different preparation procedures and conditions of propriate amounts of solid Pb acetate in portions to a hot
catalytic tests, in particular, different oxygen partial pres- aqueous solution of NH4VO3. After evaporation in a rotary
sures of the feed. Against this background, it is the aim evaporator, the resulting powder was dried at 383 K for 12
of this work to derive structure–reactivity relationships for h and calcined in air at 873 K for 3 h.
metal vanadate catalysts used in the ODH of propane. In
The BET surface area of the catalysts was determined by
particular, it shall be elucidated how coordination, valence N2 adsorption at 77 K using a Gemini III 2375 surface area
state, and redox behavior of the vanadium sites present in analyzer (Micromeritics). The vanadium valence state was
different phases influence activity and propene selectivity determined by potentiometric titration using a variant of
ofthe variouscatalysts. Thisisdone, on the one hand, usinga the method developed by Niwa and Murakami (12).
variety of ex situ characterization methods and, on the other
hand, by monitoring the behavior of the catalysts under re-
action conditions using in situ EPR, in situ UV/VIS, and
2. Catalyst Characterization
quasi in situ XPS. In comparison to Mg–V–O catalysts used
EPR spectra were recorded by the c.w. spectrometer
as a model system, the properties of Zn and Pb vanadates ELEXSYS 500-10/12 (Bruker) in X-band. In situ investiga-
have been studied. The structure of the anions and, thus, the tions under catalytic reaction conditions were performed
vanadium coordination in corresponding vanadate phases using a homemade flow reactor equipped with a temper-
of the three different metals is rather similar. However, the ature programmer and connected to a gas dosing appara-
redox behavior and the acid–base properties are expected tus (13, 14). In each run 100 mg of catalyst particles (0.31–
1
to be different. This provides a good possibility in study- 0.63 mm) were treated with a total gas flow of 3.6 L h
ing the particular influence of the various metal cations on consisting of 13.5 mol% C3H8/6.8 mol% O2/N2 at 773 K
these properties and, in turn, on the catalytic behavior in (Mg–V–O) or 823 K (Zn–V–O, Pb–V–O). For on-line prod-
the ODH of propane.
uct analysis the reactor outlet was connected to a GC
17AAF capillary gas chromatograph (Shimadzu) equipped
with a 30 m 0.32 mm Silicaplot column (Chrompack) and
a FID.
METHODS
1. Catalysts
The spin concentration of EPR-active V4+ was deter-
Mg–V–O catalysts were prepared by two different pro- mined by comparing the intensities of the EPR signals
cedures. Samples of Mg/V atomic ratios of 9/1, 4/1, 1/1, and (double integrals) of the catalysts measured at room tem-
1/4 were obtained by a precipitation method using NH4VO3 perature with that of a VOSO4 standard prepared by
and Mg(OH)2 precipitated from a 0.25 M MgCl2 6H2O by diluting VOSO4 3 H2O with K2SO4 (15). As indicated
adding 0.5 M KOH (6). The Mg(OH)2 powder was added by potentiometric titration (12), the standard contained
in portions to an aqueous solution of NH4VO3 and 1 wt% 3.97 1020 spins g 1. EPR measurements of sample and
of NH4OH. After evaporation in a rotary evaporator, the standard were performed using the Bruker X-band double-
resulting powder was dried at 383 K for 18 h and calcined cavity ER 4105 DR.
in air at 873 K for 3 h. Samples obtained by this proce-
dure are denoted as Mg9V1(p), Mg4V1(p), Mg1V1(p), and VG ESCALAB 220 iXL spectrometer (VG Instruments)
Mg1V4(p). at room temperature with a monochromated AlK source.
The X-ray photoelectron spectra were recorded using a
Samples of Mg/V atomic ratios of 3/2, 1/1, and 5/4 Charging effects at insulators were reduced by a flood gun.
were prepared by the citrate method, details of which are The spectra were corrected with respect to the C 1s signal at
described elsewhere (10). Briefly, citric acid was added 284.5 eV. Signal intensities were normalized using the sensi-
to a transparent aqueous solution containing appropriate tivity factors of Scofield (16) and the transmission function
amounts of Mg(NO3)2 and NH4VO3. After evaporation, of the spectrometer. In certain cases, spectra were recorded
the resulting precursor was decomposed in air at 653 K for after treating the catalysts in a flow of 13.5 mol% C3H8/
18 h and finally calcined in air at 873 K for 3 h. Catalysts pre- 6.8 mol% O2/N2 at 773 K using a reaction cell installed
pared by this method are denoted as Mg3V2(c), Mg1V1(c), in the lock to the analysis chamber. After subsequent cool-
and Mg5V4(c).
ing to room temperature, the samples were transferred into
Zn vanadate catalysts of Zn/V atomic ratios of 3/2 the analysis chamber of the spectrometer without contact
(Zn3V2), 1/1 (Zn1V1), and 5/4 (Zn5V4) were prepared by to ambient atmosphere. In due course XPS measurements
adding an appropriate amount of ZnCO3 powder to a hot performed in this way are labeled “quasi in situ.”

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
47
UV/VIS-DRS measurements were performed using a (Tonset) was used. This parameter was defined as the tem-
Cary 400 UV/VIS spectrometer (Varian) equipped with perature of the 5% increase of the TCD signal compared
a diffuse reflectance accessory (praying mantis, Harrick). to the maximum of the TPR peak.
To reduce light absorption, the catalysts were diluted with
3. Catalytic Tests
BaSO4 (Merck, white standard). For in situ measurements
under reaction conditions, the diffuse reflectance accessory
was equipped with a heatable reaction chamber (Harrick).
In these cases -Al2O3 (calcined at 1473 K for 4 h) was used
as dilution material. A flow of 13.5 mol% C3H8/6.8 mol%
O2/N2 was passed through the catalyst bed.
Catalytic tests were performed at atmospheric pressure
in a fixed-bed U-type quartz reactor (internal diameter
4 mm) at 773 K (Mg–V–O) or 823 K (Zn–V–O, Pb–V–O)
using an educt mixture of 40 vol% C3H8/20 vol% O2/N2.
The total flow rate was varied between 10 and 100 ml min
1
.
SEM micrographs were obtained by means of a Amray
1000B microscope (KYKY). Samples were deposited on a
metal disc and made conductive with gold using a MED
020 Coating System (BAL-TEC). TEM micrographs were
recorded at an acceleration voltage of 200 kV using a CM20
microscope (Philips). Samples were deposited on Lacey
carbon film on copper grids. EDX analysis was performed
using a PV 9900 spectrometer (EDAX). XRD powder pat-
terns were recorded using a STADI P transmission diffrac-
tometer (STOE) using CuK radiation at room tempera-
ture.
The sample particles(250–380 m) were diluted with quartz
particles in a ratio of 1 : 5. The catalyst mass was 0.05 g (Mg–
V–O) and 0.15–0.3 g (Zn–V–O and Pb–V–O). The reaction
products were analyzed by on-line gas chromatography us-
ing a Satochrom gas chromatograph. A molecular sieve 5A
column and a thermal conductivity detector were used to
determine the content of O2, N2, and carbon oxides while
the hydrocarbons were analyzed on a Poraplot-Q capillary
column using a flame ionisation detector.
RESULTS AND DISCUSSION
Acidic surface sites were determined by pyridine adsorp-
tion using a FTIR spectrometer (Bruker IFS 66) equipped
with a heatable adsorption cell. Self-supporting wafers were
pretreated in vacuum at 673 K. FTIR spectra were recorded
after pyridine adsorption at room temperature and subse-
quent evacuation. Additionally, surface acidity was evalu-
ated by temperature-programmed desorption of ammonia
(TPDA) since, in the case of Zn vanadates, it was not pos-
sible to obtain self-supporting wafers from these materials.
Samples of 0.8–1.0 g saturated with CO2 during their stor-
age in air were loaded with ammonia at 373 K for 0.5 h using
a 3 vol% NH3/He flow (90 ml/min). After being flushed with
He (50 ml/min) at 373 K for 1 h, the samples were heated
at 5 K/min to 873 K in He flow. Desorption/reaction of am-
monia as well as the desorption of carbon dioxide were
followed using the multi-ion-detection mode of the mass
spectrometer QMG 420C (Balzers) by monitoring the ion
current profiles at m/z = 2 (H+2 ), m/z = 14 (N+ from NH3
and N2), m/z = 15 (NH+ from NH3), m/z = 28 (N2+, CO+
from N2 and CO2, respectively), m/z = 44 (CO+2 from CO2,
N2O+ from N2O), m/z = 30(NO+ from NO), and m/z = 46
(NO+2 from NO2). N2 pulses (500 l) were used to calibrate
the peak areas considering the different ionization proba-
bilities and relative ion currents of the fragments.
1. Phase Composition and Surface Properties
Phase composition and particle size. Depending on the
Mg/V atomic ratio in the synthesis mixture, different crys-
talline phases are observed by XRD in Mg–V–O cata-
lysts (Table 1). MgO is formed as a major component in
Mg9V1(p) or as a minor component besides magnesium or-
thovanadate in Mg4V1(p) when magnesium isin excess. Sto-
ichiometric Mg/V atomic ratios lead to the crystallization
of magnesium ortho- and/or pyrovanadate while an excess
of vanadium in the synthesis mixture gives rise to V2O5 as
a major phase in addition to a small amount of magnesium
metavanadate. TEM micrographs indicate that crystallites
of Mg–V–O samples prepared by precipitation (15–50 nm)
are smaller than those of catalysts obtained by the citrate
method (30–200 nm) (Figs. 1a and 1b), which might give rise
to the stronger XRD reflections of the latter. EDX analysis
of the Mg–V–O crystallites revealed that their composition
is rather homogeneous corresponding virtually to the Mg/V
atomic ratios in the synthesis mixtures.
Zn–V–O catalysts with the same metal/V atomic ratio
contain crystalline phases similar to those of the respective
Mg–V–O samples (Table 1). However, in contrast to the lat-
ter, XRD reflectionsofZn–V–O samplesare about sixtimes
more intense, probably due to the much larger crystallites
( 1 m) as confirmed by TEM micrographs (Fig. 1c).
In Pb–V–O catalysts, crystalline ortho- and pyrovana-
dates of lead have been detected, too, by XRD (Table 1).
However, the preparation method does obviously not pro-
vide as good control of phase composition as was observed
with Mg–V–O and Zn–V–O catalysts. While lead ortho-
vanadate is the only detectable crystalline phase for atomic
TPR measurements were performed by heating samples
of about 20 mg at 10 K/min up to 1173 K in a 5 vol% H2/Ar
flow (40 ml/min) and holding 0.5 h. The formed water was
removed from the H2/Ar flow by trapping with molecular
sieve 4A. Changes in the H2 concentration were measured
by a thermal conductivity detector (TCD). The areas of the
TPR peaks were calibrated by Ar pulses (500 l) into the
H2/Ar flow. To compare the reducibility of the catalysts,
the temperature of the onset of the hydrogen consumption

48
RYBARCZYK ET AL.
TABLE 1
Crystalline Phases, Content of Total and Reduced Vanadium, Mean Vanadium Valence State, and BET
Surface Area of the Catalysts
1
Sample
Crystalline phases (XRD)
Total V/at% a
SBET/m2
g
Mean V valencea
V4+/at% a
V4+/at% b
Mg9V1(p)
Mg4V1(p)
Mg3V2(c)
Mg5V4(c)
Mg1V1(c)
Mg1V1(p)
M1V4(p)
Zn3V2
Zn5V4
Zn1V1
Pb3V2
Pb5V4
MgO, Mg3V2O8
Mg3V2O8, MgO
Mg3V2O8
Mg3V2O8, Mg2V2O7
Mg2V2O7
Mg2V2O7
V2O5, MgV2O6
Zn3V2O8
Zn2V2O7, Zn3V2O8
Zn2V2O7
Pb3V2O8
Pb3V2O8
Pb2V2O7, Pb3V2O8
4.7
8.75
15.5
16.8
18.4
18.3
25.3
15.5
16.8
18.3
15.9
17.1
18.55
87.4
46.9
21.6
17.7
9.5
7.9
1.9
1.6
0.7
1.0
0.3
0.32
1.2
4.68
4.86
4.90
4.92
4.87
4.91
4.92
4.89
4.91
4.92
4.57
4.71
4.78
1.5
1.2
1.6
1.3
2.4
1.7
2.1
1.6
1.5
1.5
6.8
4.9
4.0
0.08
0.06
—
—
—
—
0.09
—
—
—
—
—
—
Pb1V1
a
Determined by potentiometric titration.
Derived from determination of spin concentration using EPR.
b
ratios of Pb/V = 3/2 and 5/4, a mixture of ortho- and py- indicates that in addition to the reduction of V5+ and V4+
rovanadate is obtained with a ratio of Pb/V = 1/1. The to V3+, Pb2+ is also reduced to Pb0. In the case of the zinc
XRD reflections of Pb vanadate phases are less intense than vanadates, Zn2+ is at least partially reduced because a gray
those of Mg vanadates. Unfortunately, no information on deposition of metallic zinc was observed at the cold reactor
the structure of the crystallites is available since the TEM outlet. However, this effect seems to play only a minor role,
microcraphs are only poorly resolved due to the strong ab- as can be seen from the H2 consumption values in Table 2.
sorption of lead.
Surface composition. XPS data of as-synthesized cata-
lysts are summarized in Table 3. The metal/vanadium
atomic ratios on the surface are similar to those derived
from the bulk composition, pointing to a rather homoge-
neous elemental distribution in the whole samples. This is
in agreement with the EDX results. In magnesium and zinc
vanadate samples with metal/vanadium molecular ratios of
Reducibility. In all as-synthesized catalysts a certain
percentage of tetravalent vanadium has been detected by
potentiometric titration (Table 1). Within each group of
metal/vanadium oxides the V4+ content increases with ris-
ing metal/V atomic ratio. In as-synthesized Mg–V–O cat-
alysts the mean V valence is lowest for sample Mg9V1(p).
Pb–V–O catalysts are markedly more reduced than the cor-
responding Mg–V–O and Zn–V–O samples. It seems rather
unlikely that crystalline Pb2V2O7 and Pb3V2O8 can accom-
modate such large amounts of tetravalent vanadium with-
out degradation of the crystal structure. This suggests the
presence of additional amorphous components containing
mainly tetravalent vanadium, which could be the reason for
the lower intensity of the XRD reflections in comparison
to those of Zn–V–O and Mg–V–O samples.
In the TPR profiles, Tonset is shifted to higher tempera-
tures with decreasing metal/vanadium ratio (Figs. 2 and 3).
The stepwise reduction of V2O5 via V6O13, V2O4, to V2O3
(17) dominates the TPR profile of the Mg1V4(p) sample,
which agrees well with XRD and TEM results showing the
presence of an excess of V2O5 in this sample. The profiles of
Pb–V–O catalysts show shoulders on the high-temperature
flank of the peaks of the respective Pb vanadates. The rather
high V4+ content (Table 1) suggests the presence of addi-
tional amorphous phases, which could be one reason for the
asymmetric TPR profile. Moreover, the difference between
the experimental and calculated H2 consumptions (Table 2)
TABLE 2
TPR Results
1
H2 consumption/mmol g
Sample
Calculateda Experimental
Tonset/K
Mg9V1(p)
Mg4V1(p)
Mg3V2(c)
Mg5V4(c)
Mg1V1(c)
Mg1V1(p)
Mg1V4(p)
Zn3V2
Zn5V4
Zn1V1
Pb3V2
Pb5V4
1.90
3.65
6.10
6.74
6.99
7.33
9.46
4.39
5.09
5.38
1.79
1.52
2.59
1.67
3.49
5.93
6.46
7.15
7.09
9.04
4.01
5.10
5.36
5.53
5.13
5.90
750
813
829
821
823
832
879
781
785
813
827
831
845
Pb1V1
a Calculated assuming that all V5+ and V4+ determined
by potentiometric titration (Table 1) is reduced to V3+
.

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
49
FIG. 2. TPR profiles of Mg–V–O samples.
FIG. 1. TEM micrographs of (a) sample Mg1V1(p), (b) sample
Mg1V1(c), and (c) sample Zn1V1.
1/1, 3/2, and 5/4 only pentavalent vanadium represented by
V 2p3/2 binding energies between 518.6 and 519.9 eV has
been detected on the surface, although the mean vanadium
valence state is slightly less than +5 (Table 1). A reason
may be that V4+ is readily oxidized on the surface of
these samples upon contact with air. The V 2p3/2 binding
FIG. 3. TPR profiles of Zn–V–O and Pb–V–O samples.

50
RYBARCZYK ET AL.
TABLE 3
Binding Energies, EB, and Surface Content for Catalyst Constituents Derived by XPS
Sample
Signal
EB/eV
Content/at% a
Sample
Signal
EB/eV
Content/at% a
Mg9V1(p)
O 1s
V 2p3/2
529.6
517.0
514.9
46.7
60.3
1.9
1.3
Mg1V4(p)
O 1s
V 2p3/2
Mg 2p
530.2
517.4
50.1
61.7
19.6
3.6
Mg 2p
29.7
Mg4V1(p)
O 1s
V 2p3/2
529.6
517.1
515.4
46.8
58.1
3.1
2.5
Mg1V4(p)b
O 1s
V 2p3/2
529.9
516.9
515.4
49.5
78.7
7.5
9.2
4.6
Mg 2p
25.8
Mg 2p
Mg3V2(c)
Mg5V4(c)
Mg1V1(c)
Mg1V1(c)b
O 1s
V 2p3/2
Mg 2p
532.6
519.8
51.9
50.7
10.2
19.4
Zn3V2
Zn5V4
Zn1V1
Pb3V2
O 1s
V 2p3/2
Zn 2p3/2
532.7
519.6
1024.2
48.3
6.0
10.8
O 1s
V 2p3/2
Mg 2p
532.6
519.9
52.0
49.7
10.5
16.9
O 1s
V 2p3/2
Zn 2p3/2
532.6
519.7
1024.2
52.9
7.9
8.7
O 1s
V 2p3/2
Mg 2p
532.4
519.8
52.0
51.3
8.4
10.3
O 1s
V 2p3/2
Zn 2p3/2
532.5
519.8
1024.3
39.7
3.6
5.0
O 1s
V 2p3/2
532.5
519.8
517.9
52.9
57.2
11.4
8.6
O 1s
529.0
530.1
515.7
516.6
137.6
43.3
12.3
11.0
2.8
V 2p3/2
Pb 4 f7/2
Mg 2p
18.2
18.0
Mg1V1(p)
O 1s
V 2p3/2
Mg 2p
531.6
518.6
51.0
42.5
10.0
12.2
Pb5V4
Pb1V1
O 1s
V 2p3/2
528.0
513.4
514.9
138.4
53.3
4.0
8.6
Pb 4 f7/2
O1s
17.2
528.3
529.4
515.6
137.8
19.6
36.7
13.3
14.0
V 2p3/2
Pb 4 f7/2
a
Difference to 100% due to carbon impurities.
Quasi in situ XPS data after treatment under propane ODH conditions.
b
energies in these vanadates are higher than the value ex- air. Alternatively, the shift to lower binding energies could
pected for pure V2O5 (517.4 eV (18)). This trend was ob- be caused by the presence of the heavy lead ion. A similar
served by other authors, too, and has been explained in shift of the V 2p3/2 peak is observed, too, within the group
terms of increasing ionic character of the V–O bond in the of alkali vanadates from Li3VO4 (517.5 eV) to Cs3VO4
vanadates (6). In sample M1V4(p), which contains V2O5 as (516.9 eV) (18). With respect to this consideration we at-
a separate phase, the V 2p3/2 peak occurs at 517.4 eV, which tribute the V 2p3/2 peaks at 515.6 and 515.7 eV to V4+ and
is very close to the value determined for pure V2O5 (18). the one at 516.6 eV to V5+. In sample Pb5V4 partial charg-
When the mean vanadium valence decreases as in the case ing effects cannot be excluded, which could be the reason
of samples M9V1(p) and M4V1(p), a certain percentage of for the unusually low V 2p3/2 binding energies (Table 3).
reduced vanadium (EB = 514.9 and 515.3 eV, Table 3) is Such effects might also be responsible for the deviations of
also detected on the surface. The same holds for Pb–V– the binding energy observed for Mg in samples Mg9V1(p)
O samples, showing a rather low mean vanadium valence and Mg4V1(p) and for V and Zn in Zn–V–O catalysts. They
(Table 1). In general, the V 2p3/2 peak in these samples ap- can be compensated for by depositing conducting gold par-
pears at rather low binding energies. Based on a comparison ticles of 20 nm in size as a reference on the samples (19).
with pure vanadium oxides, values of 515.6 and 515.7 eV This has been done for samples Mg4V1 and Zn5V4. Un-
should be assigned to V3+ and 516.6 eV to V4+ (18). How- der these conditions the peaks in Mg4V1 occur at 531.7 eV
ever, this would mean that the surface of lead vanadates (O 1s), 517.3 eV (V 2p3/2), and 49.7 eV (Mg 2p) and in
should be markedly more reduced than the bulk, which Zn5V4 at 530.2 eV (O 1s), 517.1 eV (V 2p3/2), and 1021.6 eV
seems to be improbable since the samples were stored in (Zn 2p3/2).

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
51
Acid–base properties. Acid–base properties of the cata- viously, the catalysts contain both weak and strong acidic
lyst surface have been studied by FTIR spectroscopy us- sites, the latter fixating ammonia molecules on the surface
ing pyridine adsorption and by temperature-programmed until their oxidation at higher temperatures.
desorption of ammonia (TPDA). In all Mg–V–O catalysts
Apart from the desorption of ammonia and nitrogen,
Lewis surface sites evidenced by a FTIR band around the desorption of carbon dioxide can be observed. The
1440 cm 1 were detected. The band area normalized on the m/z = 44 profiles of both samples show a peak with a shoul-
BET surface area of the catalysts rises with increasing vana- der on its low-temperature flank, pointing to basic sites of
dium content (arbitrary units): Mg9V1(p): 4.6 < Mg3V2(c): different strength. The profile m/z = 44 can be clearly at-
7.4 < Mg5V4(c): 11.9 Mg1V1(c): 10.5 < Mg1V4(p): 26.3. tributed to CO2 because N2O can be excluded due to the
This indicates that Lewis acidity is related to the vana- noncorresponding shapes and intensities of the m/z = 30
dium surface sites. The band position is slightly shifted to profile (NO+ from N2O). Furthermore, shoulders observed
higher wavelengths in the order Mg3V2(c) < Mg9V1(p) < on both flanks of the m/z = 28 peaks cannot be related
Mg5V4(c) < Mg1V1(c) < Mg1V4(p), pointing to increasing to CO2 desorption (CO+ fragments) because of the same
strength of the Lewis sites (20). Despite sample Mg1V4(p) shape of the m/z = 14 and m/z = 28 profile in this range.
in which a weak band at 1536 cm 1 arises from pyridine ad-
The total amount of adsorbed ammonia, i.e., the total
sorbed at Brønsted sites, no such sites have been detected amount of acidic sites, was calculated from the ammonia
in all other Mg–V–O catalysts. On the surface of Pb–V–O and nitrogen desorption (Table 4). The amount of the basic
catalysts no pyridine adsorption could be detected by FTIR, sites, i.e., the amount of desorbed CO2, is rather low with
indicating the absence of both Lewis and Brønsted sites. respect to that of acidic sites. However, it has to be consid-
Since Zn–V–O catalysts could not be pressed into self- ered that CO2 was adsorbed during storage in air, i.e., in
supporting discs, the surface acidity of Zn1V1 has been stud- competition with water vapor. The basicity of these sites is
ied by TPDA in comparison to that of Mg1V1(c).
rather high since the major part of CO2 desorbs only be-
TPD/TPRS profiles of the samples Mg1V1(c) and Zn1V1 tween 573 and 873 K. Upon comparison of the surface area
are shown in Fig. 4. The ion current profiles m/z = 15 and of specific amounts of desorbed NH3 and CO2, it is readily
14 reveal the desorption of weakly bound ammonia be- seen that the concentration of both acidic and basic sites
tween 400 and 700 K for both samples. However, a con- is markedly higher for sample Zn1V1 than for Mg1V1(c).
siderable part of NH3 is so strongly bound that it does not This holds in particular for strong acidic sites comprised by
reversibly desorb but is oxidized to nitrogen as indicated the amount of formed N2. In contrast, the strength of acidic
by comparable shapes and peak maxima of the m/z = 28 and basic sites is virtually similar for both samples. From the
and m/z = 14 profiles at ca. 700 K (Mg1V1(c)) and 670 described investigations it turns out that the concentration
K (Zn1V1). NO or NO2 were observed in traces only. Ob- of acidic sites on the surface of metal vanadates increases
in the order 0 Pb–V–O
Mg–V–O < Zn–V–O.
2. Structure of Vanadium Sites
The catalysts have been studied by EPR and UV/VIS-
DRS spectroscopy being complementary methods to ob-
tain information on the local coordination geometry of
VOx species. By EPR, only V4+ in distorted octahedral
or square pyramidal symmetry can be observed at tem-
peratures higher than 77 K while V3+ and V4+ in tetra-
hedral symmetry are not detected due to very short relax-
ation times [20]. In contrast, UV/VIS spectra show intense
charge-transfer (CT) bands for V5+ while V4+ and V3+ are
hardly detectable especially in the presence of V5+ as major
component.
UV/VIS–DRS measurements. The UV/VIS spectrum of
sample Mg3V2(c) shows three bands at 218, 280, and 341
nm (Fig. 5), which can be assigned to CT transitions of iso-
lated tetrahedrally coordinated V5+ (21). This is in agree-
ment with the crystal structure of Mg3(VO4)2 (6) being
the main phase of this catalyst. In sample Mg1V1(c), CT
bands are red-shifted to 225, 300, and 370 nm being charac-
teristic of V5+O4 tetrahedra connected via V–O–V bonds
(22). This agrees well with the fact that the main phase
FIG. 4. Ion current profiles of temperature-programmed desorp-
tion/reaction of carbon dioxide and ammonia from catalyst Mg1V1(c) (1)
and Zn1V1 (2), m/z = 15 (A), m/z = 28 (B), m/z = 14 (C), and m/z =
44 (D).

52
RYBARCZYK ET AL.
TABLE 4
Evaluation of the Peak Areas of the TPD/TPRS Measurements
NH3
adsorbed/ mol m
NH3
desorbed/ mol m
N2
CO2
2
2
2
2
Sample
formed/ mol m
desorbed/ mol m
Mg1V1(c)
Zn1V1
27.6
85.0
20.6
51.0
3.5
17.0
2.9
9.0
of Mg1V1(c) is Mg2V2O7, the crystal structure of which ponent, the CT band at lowest energy is blue shifted and
contains dimers of VO4 tetrahedra (2). The spectrum of less intense, suggesting only a minor however not negligible
sample Mg1V4(p) contains bands at 290, 311, 394, and 480 influence of connected VO4 tetrahedra.
nm, which can be assigned to CT transitions of crystalline
The UV/VIS spectra of the Pb–V–O catalyst are rather
V2O5, being the main component of this catalyst (22). Ad- similar to those of the respective Zn–V–O samples (Fig. 6)
ditionally, the catalyst contains a small amount of MgV2O6, since the local structure of the vanadium ions (VO34 and
which forms chains of VO6 octahedra (4). Their CT bands V2O47 ) in the crystalline Pb vanadates is the same as that
have positions similar to those of V2O5 except the band in the corresponding Zn vanadates. However, the Pb–V–O
at 480 nm, which is characteristic only for crystalline V2O5 samples show a slightly higher absorption in the wavelength
(22). In samples Mg9V1(p) and Mg4V1(p) weak bands at range below 300 nm in which the CT bands of V4+ are ex-
380 and 386 nm are observed in addition to those arising pected (21, 24). This could be due to the rather high V4+
from Mg3(VO4)2. They indicate the presence of oligomeric content (Table 1). Moreover, the lowest energy CT band
tetrahedrally and probably some octahedrally coordinated in samples Pb3V2 and Pb5V4 containing Pb orthovanadate
V5+
.
with isolated VO4 units as the only crystalline phase falls at
The UV/VIS bands for Zn1V1 and Zn5V4 catalysts 368 nm. This indicates the presence of some connected VO4
(Fig. 6) fall at similar wavelengths as in Mg1V1(c) due to species probably within an additional amorphous compo-
the fact that Zn2V2O7 is the main component of these sam- nent discussed above.
ples containing dimers of VO4 tetrahedra like Mg2V2O7.
In the spectrum of sample Zn3V2 containing crystalline
EPR measurements. In the EPR spectra of Mg–V–O
Zn3(VO4)2 with isolated VO4 tetrahedra as the main com- catalysts, a signal with hyperfine structure (hfs) being
FIG. 5. UV/VIS-DRS spectra of as-synthesized Mg–V–O catalysts at room temperature; thick line, experimental, thin lines, deconvoluted bands.

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
53
FIG. 6. UV/VIS-DRS spectra of as-synthesized Zn–V–O and Pb–V–O catalysts at room temperature; thick line, experimental; thin lines, decon-
voluted bands.
characteristic of isolated, axially distorted VO2+ ions in lowest energy CT band in the UV/VIS spectra (Fig. 5). This
square pyramidal or octahedral coordination is superim- seems not to be the case for samples Mg1V1(p), Mg1V1(c),
posed on a broad more or less isotropic singlet arising from Mg5V4(c), and Mg3V2(c) consisting of vanadate phases
weakly interacting VO2+ species (Fig. 7). An exception is with tetrahedral V5+ species (Fig. 5). Accordingly, the EPR
sample Mg1V4(p), which shows a narrow singlet being due signal of VO2+ species supposed to be formed by partial re-
to closely neighboring VO2+ species coupled by effective duction of VO3+ is markedly smaller than that in the spec-
spin–spin exchange interactions. Since sample Mg1V4(p) tra of Mg9V1(p), Mg4V1(p), and Mg1V4(p) (Fig. 7). Par-
contains crystalline V2O5, it is very probable that the VO2+ tial reduction, which is evident also in samples Mg1V1(p),
clusters are formed by partial reduction of V2O5, being a Mg1V1(c), Mg5V4(c), and Mg3V2(c) (Table 1), leads prob-
frequently observed phenomenon in this material (24). The ably mainly to tetrahedral V4+, which is not EPR active
values of g = 1.943, g = 1.955, A = 17.1 mT, and A
=
due to short relaxation times. This is most obvious for the
k
⊥
k
⊥
5.0 mT derived by spectra simulation from the hfs spectra Pb–V–O catalysts and for sample Zn3V2 (pure Zn ortho-
of samples Mg9V1(p) and Mg4V1(p) containing MgO as a vanadate) for which the EPR signal is too weak to be prop-
separate phase are similar to those observed for VO2+ in erly detected. The other two Zn vanadates show only a
V2O5/MgO catalysts (g = 1.949, g = 1.959, A = 17.7 mT, very small signal resulting mainly from interacting VO2+
k
⊥
k
and A = 4.8 mT (25). Thus, it seems likely that in samples species.
⊥
Mg9V1(p) and Mg4V1(p) some VO2+ ions are located on
Upon comparison of the V4+ content derived by poten-
octahedral Mg lattice positions in the MgO phase or in- tiometric titration with that resulting from the determina-
corporated in an amorphous vanadia overlayer. Scattering tion of the spin concentration using EPR (Table 1), it is
of such an overlayer over the surface of MgO has been de- readily evident, even for samples Mg1V4(p), Mg4V1(p), and
tected, too, by Pantazidis et al. in a Mg7V1 catalyst prepared Mg9V1(p), which show rather strong EPR signals (Fig. 7),
in a way similar to sample Mg9V1(p) (26). The formation that the former value is much higher, indicating that a con-
of such an amorphous oligomeric VOx component is also siderable part of the tetravalent vanadium is EPR silent.
suggested in the UV/VIS results described above.
In all other samples the EPR signal is too weak to ob-
In samples Mg9V1(p) and Mg4V1(p), a certain percent- tain proper double integrals. Besides the tetrahedral co-
age of the V5+ species is also in square pyramidal or oc- ordination discussed above, strong line broadening due to
tahedral coordination, as indicated by the position of the weak magnetic dipolar interactions between less closely

54
RYBARCZYK ET AL.
ODH, samples Mg1V4(p) and Mg1V1(c) have been studied
by in situ EPR, in situ UV/VIS-DRS, and quasi in situ XPS.
Upon heating of sample Mg1V4(p) under feed at 773 K, the
EPR singlet of magnetically interacting VO2+ species be-
ing present already in the fresh catalyst increases strongly
(Fig. 8). This indicates that the V2O5 phase contained in
this sample is markedly reduced under ODH conditions.
This is also confirmed by a decrease of the mean vana-
dium valence state determined by potentiometric titration
from 4.918 in the fresh catalyst to 4.604 after the cata-
lytic test. Indeed, crystalline VO2 has been detected in the
XRD powder pattern of the used catalyst while the reflec-
tions of V2O5 roughly disappeared. Interestingly, this re-
duction goes along with an improvement of the catalytic
performance (Fig. 8). Quasi in situ XPS measurements af-
ter pretreatment of sample Mg1V4(p) under ODH condi-
tions indicate that 55% of the surface V5+ sites are reduced
to V4+ (Table 3). In situ UV/VIS measurements of sample
Mg1V4(p) during heating in nitrogen flow show a decrease
of CT bands around 400 and 480 nm, being characteristic
of crystalline V2O5 (23) (Fig. 9a). Simultaneously, absorp-
tion increases below 250 nm in the range of CT bands of
tetravalent vanadium and above 600 nm where d–d bands
of V4+ have to be expected (21). This indicates that sam-
ple Mg1V4(p) is already thermally reduced by heating in
an inert atmosphere. Upon switching to reactant-gas mix-
ture at 773 K, the band intensity increases in the region
of 300–350 nm, which is probably due to the deposition
of weakly condensed polyenyl species being coke precur-
sors (27) (Fig. 9b). Such carbonaceous residues have been
detected, too, on the used catalysts by FTIR microscopy.
From the different in situ studies it can be concluded that
the VO3+ sites in catalyst Mg1V4(p) are markedly reduced
to VO2+ species, which obviously improve the catalytic per-
formance.
FIG. 7. EPR spectra of as-synthesized catalysts at room temperature
in the range 150 < B0/mT < 550.
neighboring VO2+ species could be another reason for this
result.
3. Structural Changes under Reaction Conditions
The situation is different for catalyst Mg1V1(c). By
in situ EPR only a small broad singlet of weakly interact-
ing VO2+ species is detected at the beginning of the ODH
To elucidate whether the structure and valence state
of the active vanadium sites are modified during propane
FIG. 8. In situ EPR spectra of sample Mg1V4(p) during propane ODH at 773 K and corresponding catalytic data obtained by on-line GC.

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
55
surface V5+ sites is reduced to V4+ (Table 3). Since these
species do obviously not contribute to the EPR spectrum, it
is supposed that their coordination is probably tetrahedral.
With respect to the overall vanadium content of sample
Mg1V1(c), the percentage of V4+ formed upon reaction is
much smaller than that in sample Mg1V4(p) since almost
no changes are observed in the respective in situ UV/VIS
spectra. Moreover, the mean vanadium valence in sample
Mg1V1(p) showing a behavior very similar to that of sam-
ple Mg1V1(c) changes only from 4.91 before to 4.87 after
the catalytic test. All these results suggest that tetrahedrally
coordinated V5+ species in Mg pyrovanadate are less eas-
ily reduced than octahedral VO3+ species, being present in
V2O5 and MgV2O6.
4. Relations between Catalytic Behavior
and Structural Properties
Surface area specific rates of propane conversion and
propene formation as well as propene selectivities are plot-
ted for all catalysts in Figs. 10 and 11. Since catalytic activ-
ity is a function of the surface vanadium concentration, the
values have been normalized with respect to the total vana-
dium content for deriving explicit relations between the
nature of the individual vanadium sites and the catalytic
properties by eliminating the influence of different surface
vanadium concentrations. This could be done since XPS
measurements indicated that the metal/vanadium ratio on
the surface is roughly the same as that in the bulk, pointing
to a homogeneous distribution throughout the whole sam-
ple. Thus, the values on the left ordinate axis in Figs. 10 and
FIG. 9. In situ UV/VIS-DRS spectra of Mg1V4(p) at room tempera-
ture (A) before (1) and after heating in nitrogen flow to 773 K (2) and
under propane ODH conditions (B) at 773 K after 5 min (1), 60 min (2),
120 min (3), and 240 min (4) time on stream.
experiment, the intensity of which remains almost constant 11 are proportional to turnover numbers, which could not
upon time on stream, indicating that no additional VO2+ be determined directly. Furthermore, it has to be noted that
species are formed. In contrast, quasi in situ XPS measure- the catalytic tests for Zn–V–O and Pb–V–O samples were
ments revealed that after the ODH experiment 43% of the performed at 823 K since these solids were found to be not
FIG. 10. Area specific rates of propane conversion and propene formation normalized on the total vanadium content and propene selectivities at
773 K for Mg–V–O catalysts (degree of conversion listed in the bottom line).

56
RYBARCZYK ET AL.
FIG. 11. Area specific rates of propane conversion and propene formation normalized on the total vanadium content and propene selectivities at
823 K for Zn–V–O and Pb–V–O catalysts (degree of conversion listed in the bottom line).
yet active at a temperature of 773 K used to examine the nor trace component besides Mg orthovanadate in sample
catalytic behavior of Mg–V–O catalysts.
Mg4V1(p).
Within the group of Mg–V–O catalysts, sample
Pure Mg ortho- and pyrovanadate, which both contain,
Mg1V4(p), which consists of V2O5 and MgV2O6, both con- besides traces of octahedral VO2+ (detected by EPR), only
taining octahedrally coordinated polymerized V5+ sites, is V5+O4 tetrahedra either as isolated units (Mg3V2(c)) or
most active but poorly selective (Fig. 10). The next active connected in pairs (Mg1V1(c, p)), are markedly less ac-
catalyst is Mg9V1(p) for which EPR and UV/VIS measure- tive but show higher selectivities than the two catalysts
ments indicated, besides tetrahedral V5+, the presence of discussed above (Fig. 10). Mg1V1(c) and Mg1V1(p) are
certain amounts of octahedrally coordinated V5+ and V4+
,
somewhat more selective than Mg3V2(c). Mg5V4(c), which
which are probably located in an VOx overlayer on the contains a mixture of Mg ortho- and pyrovanadate, takes an
MgO phase. However, in contrast to sample Mg1V4(p) in intermediate position. The slightly higher activity of sam-
which only 8.3% of all vanadium sites are tetravalent, the ple Mg1V1(p) in comparison to that of Mg1V1(c) may be
relative percentage of V4+ is markedly higher in sample due to the smaller crystallite size of the former. However,
Mg9V1(p). Moreover, the VOx species in the latter cat- the different preparation procedures are obviously only of
alyst are less polymerized. Both properties, less effective minor influence on the catalytic properties.
clustering and high relative V4+ content, which lowers the
Upon comparison of the catalytic results of catalysts
oxidation potential, might be regarded as a reason for the Me1V1, Me3V2, and Me5V4 (Me = Mg, Zn, Pb) contain-
higher propene selectivity though lower activity in compar- ing vanadate ions of similar structure, it is evident that
ison to sample Mg1V4(p). In addition, pyridine adsorption the different metal cations have a strong influence. Both
experimentsbyFTIR have shown that surface acidityislow- activity and selectivity decrease in the order Mg–V–O >
est for sample Mg9V1(p) and highest for Mg1V4(p), which Pb–V–O
Zn–V–O. This is in contrast to the relationship
might hinder the propene desorption and lower the selec- between dropping ODH selectivities and increasing oxida-
tivity of this catalyst. This is in line with results of Blasco tion potentials of the metal cations established by Kung
and Lo´pez-Nieto (1) who observed decreasing olefin selec- and Kung (8) for a number of metal orthovanadates in the
tivities with increasing surface acidity on a variety of sup- ODH of butane. With respect to this relationship, a selec-
ported VOx catalysts in the ODH of propane and n-butane tivity order of Mg–V–O > Zn–V–O > Pb–V–O should be
while this effect is not as pronounced for ethane ODH expected since the oxidation potentials of the cations in-
since desorption of the less basic ethylene is also favored crease in the order Mg/Mg2+ < Zn/Zn2+ < Pb/Pb2+ (28).
from weakly acidic sites. The lower catalytic performance This is in contrast to the experimental results. It is also not
of sample Mg4V1(p) in comparison to that of Mg9V1(p) consistent with the reducibility of the catalysts reflected by
could be due to the fact that crystalline MgO, assumed to the temperature of the TPR onset (Table 2). If this property
be associated with an active VOx overlayer, is only a mi- was of major influence on the propene selectivity, Zn–V–O

OXIDATIVE PROPANE DEHYDROGENATION OVER ME VANADATES
57
samples should be the most selective catalysts. However, it
Vanadium catalyzes the ODH of propane in both valence
has to be taken into account that in most of the catalysts states, V5+ and V4+; however, V4+ seems to be more selec-
studied in this work vanadium does form not only Me–O–V tive though less active than V5+ due to the lower oxidation
bonds like in pure orthovanadates but also V–O–V bonds, potential. This can be concluded by comparing the two most
which could be reduced easier. Thus, the redox properties active catalysts Mg1V4(p) and Mg9V1(p) with the lowest
cannot only be discussed in terms of the oxidation poten- and highest fraction of V4+ in the total vanadium content.
tial of the metal cations. However, the reducibility of the Moreover, in situ reduction of V5+ to V4+ might optimize
catalysts reflected by the temperature of the TPR onset the surface redox potential under ODH conditions. This
(Table 2) is also not sufficient to explain the differences in issue is frequently not considered in the literature since in
the catalytic performance. Obviously, there are more strik- most papers only V5+ sites are discussed as being active and
ing factors than the redox properties that govern the selec- selective sites (1).
tivity of metal vanadate catalysts, e.g., the surface acidity.
V sites in octahedral or square pyramidal coordination
As revealed by TPDA investigations (Table 4), the con- are more active but less selective than VO4 tetrahedra,
centration of acidic sites, in particular of strong ones, is probably due to the higher number of coordinating oxygen
markedly higher on the surface of Zn1V1 in comparison atoms available for the reaction with the hydrocarbon
to that on Mg1V1(c). This could be the reason for the lower molecule. This can be concluded by comparing the activity
selectivity. In contrast, no acidic sites have been detected of samples Mg1V4(p) and Mg9V1(p) with a certain amount
on Pb1V1 by pyridine adsorption. Moreover, the concen- of higher coordinated vanadium sites in comparison to
tration of basic sites is by about a factor of 3 higher on that of samples Mg1V1(c, p), Mg3V2(c), and Mg5V4(c) in
Zn1V1 than on Mg1V1(c). Such sites are assumed to pro- which almost only tetrahedrally coordinated vanadium has
mote H-abstraction from the propane molecule in the rate- been detected. It is in line with observations of Blasco and
determining step and, thus, improve catalytic activity (9). Lo´pez Nieto for a variety of supported VOx catalysts (1).
However, despite more basic surface sites and higher ox-
Isolated VO4 and V2O7 units as present in crystalline
idation potential, Zn–V–O catalysts are poorly active in metal ortho- and pyrovanadates, respectively, are more se-
comparison to Mg–V–O catalysts. A reason may be that lective but less active than VOx species connected in clus-
Zn–V–O samples form much larger crystallites and, thus, ters (Mg9V1(p)) or even in crystalline chain- or layer-like
show lower defect concentration and structural disorder as structures (Mg1V4(p)).
detected by XRD and TEM measurements.
The differences in the catalytic properties brought about
Pb–V–O catalysts show comparable selectivities as the by the different metal ions (Mg2+, Zn2+, Pb2+) are not only
corresponding Mg–V–O samples; however, they are less ac- governed by the oxidation potential of the latter, as con-
tive (Fig. 11). This could be due to the rather high amount cluded by Kung and Kung (8) for a number of metal or-
of V4+ in these samples, which is most probably tetrahe- thovanadates in the ODH of n-butane, but to a major de-
drally coordinated. Thus, it is likely that the oxidation po- gree also by the acid–base properities and the crystal size
tentials of Pb–V–O catalysts are lower than those of the and the extent of structural disorder. Thus, it seems likely
respective Mg–V–O compounds, which in turn, should give that the low activity of Zn–V–O catalysts is mainly a conse-
rise to lower activities. In general, in situ reduction of V5+ quence of the large crystal size and low defect concentra-
to V4+ under ODH conditions as detected experimentally, tion, notwithstanding the fact that the oxidation potential of
e.g., for samples Mg1V4(p) and Mg1V1(c), is presumably of Zn/Zn2+ is higher than that of Mg/Mg2+ and the concentra-
important influence on the catalytic performance insofar as tion of basic surface sites promoting the rate-determining
it tailors the redox potential of the active surface sites.
H-abstraction is higher in Zn–V–O, too. The selectivities of
Pb–V–O catalysts are only slightly lower compared to those
of Mg–V–O compounds, although the oxidation potential
CONCLUSIONS
of Pb/Pb2+ is about 20 times larger than that of Mg/Mg2+
.
By comparison of the results of catalyst characteriza- A possible reason therefore could be the rather high per-
tion with those of the catalytic tests, it is evident that centage of tetravalent vanadium in the Pb–V–O samples,
structure–reactivity relationships in metal vanadate cata- which retains high selectivities, however, at the expense of
lysts are rather complex and require consideration of the activity.
combined action of different properties such as concentra-
As a main conclusion of the investigations in this work, it
tion of acidic and basic surface sites, crystal size, disorder, appears promising to use supported vanadia catalysts with
coordination, and valence state of the vanadium species, highly dispersed, preferably tetrahedrally coordinated, V5+
which in turn depend on the particular metal ion. Although sites to achieve high selectivities. The inherently lower ac-
there is no simple relation between these properties and tivities should be compensated for by using high surface
the catalytic performance, some key features can be de- area supports of low surface acidity. First promising results
rived from the results presented in this work.
were obtained with a supported VOx /MCM catalyst (29),

58
RYBARCZYK ET AL.
which is known to possess the properties mentioned above 14. Karge, H. G., Lange, J. P., Gutze A., and Laniecki, M., J. Catal. 114,
144 (1988).
(30).
15. Dyrek, K., Madej, A., Mazur, E., and Rokosz, A., Colloids Surf. 45,
135 (1990).
ACKNOWLEDGEMENTS
16. Scofield, J. H., J. Electron Spectrosc. 8, 129 (1976).
17. Koranne, M. M., Goodwin, J. G., and Marcelin, G., J. Catal. 148, 369
(1994).
18. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D., in “Hand-
book of X-ray Photoelectron Spectroscopy” (J. Chastain and R. C.
King Jr., Eds.). p. 241. Physical Electronics Inc., Eden Prairie, 1995.
19. Unger, W., Gross, Th., Bo¨se, O., Fritz, Th., and Gelius, U., Nachrichten
Chem. 48, 1108 (2000).
The authors thank Dr. U. Bentrup for performing the pyridine adsorp-
tion measurements and the DFG for financial support (Grant Br 1380/4-1).
REFERENCES
1. Blasco, T., and Lo´pez-Nieto, J. M., Appl. Catal. 157, 117 (1997).
2. Banares, M. A., Catal. Today 51, 319 (1999).
3. Mamedov, E. A., and Corte´s Cobera´n, V., Appl. Catal. A 127, 1 (1995).
4. Chaar, M., Patel, D., and Kung, H. H., J. Catal. 109, 463 (1988).
5. Michalakos, P. M., Kung, M. C., Jahan, I., and Kung, H. H., J. Catal.
140, 226 (1993).
20. Vedrine, J. C., in “Characterization of Heterogeneous Catalysts”
(F. Delannay, Ed.), p. 161. Dekker, New York, 1984.
21. Centi, G., Perathoner, S. and Trifiro´, F., J. Phys. Chem. 96, 2617
(1992).
22. Schraml-Marth, M., Wokaun, A., Pohl, M., and Krauss, H.-L., J. Chem.
Soc. Faraday Trans. 87, 2635 (1991).
6. Siew Hew Sam, D., Soenen, V., and Volta, J. C., J. Catal 123, 417 (1990).
7. Guerrero-Ruiz, A., Rodriguez-Ramos, I., Fierro, J. L. G., Soenen, V.,
Herrmann, J. M., and Volta, J. C., in “Novel Development in Selective
Oxidation byHeterogeneousCatalysis” (P. Ruizand B. Delmon, Eds.),
Vol. 72, p. 203. Elsevier, Amsterdam, 1992.
23. Morey, M., Davidson, A., Eckert, H., and Stucky, G., Chem. Mater. 8,
486 (1996).
24. Bru¨ckner, A., Appl. Catal. A 200, 287 (2000).
25. Oganowski, W., Hanuza, J., and Jezowska-Trzebiatowska, B., Bull. Pol.
Acad. Sci. 32, 355 (1984).
8. Kung, H. H., and Kung, M. C., Appl. Catal. A 157, 105 (1997).
9. Pantazidis, A., Aurox, A., Herrmann, J. M., and Mirodatos, C., Catal.
Today 32, 81 (1996).
10. Gao, X., Ruiz, P., Xin, Q., Guo, X., and Delmon, B., J. Catal. 148, 56
(1994).
26. Pantazidis, A., Burrows, A., Kiely, C. J., and Mirodatos, C., J. Catal.
177, 325 (1998).
27. Karge, H. G., Laniecki, M., Ziolek, M., Onyestyak, G., Kiss, A.,
Kleinschmit, P., and Siray, M., Zeolites: Facts Figures Future 1327
(1989).
11. Burrows, A., Kiely, C. J., Perregaard, J., Højlund-Nielsen, P. E.,
Vorbeck, G., Calvino, J. J., and Lo´pez-Cartes, C., Catal. Lett. 57, 121
(1999).
12. Niwa, M., and Murakami, Y., J. Catal. 76, 9 (1982).
13. Bru¨ckner, A., Kubias, B., Lu¨cke B., and R. Sto¨ßer, Colloids Surf. 115,
179 (1996).
28. “Handbook of Chemistry and Physics,” 75th ed. CRC Press, Boca
Raton, FL, 1994.
29. Buyevskaya, O. V., Bru¨ckner, A., Kondratenko, E. V., Wolf, D., and
Baerns, M., Catal. Today, in press.
30. Berndt, H., Martin, A., Bru¨ckner, A., Schreier, E., Mu¨ller, D., Kosslick,
H., Wolf, G.-U., and Lu¨cke, B., J. Catal. 191, 384 (2000).