Fundamental studies of butane oxidation over model-supported vanadium oxide catalysts: Molecular structure-reactivity relationships

Article abstract of DOI:10.1006/jcat.1997.1742

The oxidation of n-butane to maleic anhydride was investigated over a series of model-supported vanadia catalysts where the vanadia phase was present as a two-dimensional metal oxide overlayer on the different oxide supports (TiO₂, ZrO₂, CeO₂, Nb₂O₅, Al₂O₃, and SiO₂). No correlation was found between the properties of the terminal V=O bond and the butane oxidation turnover frequency (TOF) during in situ Raman spectroscopy study. Furthermore, neither the n-butane oxidation TOF nor maleic anhydride selectivity was related to the extent of reduction of the surface vanadia species. The n-butane oxidation TOF was essentially independent of the surface vanadia coverage, suggesting that the n-butane activation requires only one surface vanadia site. The maleic anhydride TOF, however, increased by a factor of 2-3 as the surface vanadia coverage was increased to monolayer coverage. The higher maleic anhydride TOF at near monolayer coverages suggests that a pair of adjacent vanadia sites may efficiently oxidize n-butane to maleic anhydride, but other factors may also play a contributing role (increase in surface Bronsted acidity and decrease in the number of exposed support cation sites). Varying the specific oxide support changed the n-butane oxidation TOF by ca. 50 (Ti > Ce > Zr ～ Nb > Al > Si) as well as the maleic anhydride selectivity. The maleic anhydride selectivity closely followed the Lewis acid strength of the oxide support cations, Al > Nb > Ti > Si > Zr > Ce. The addition of acidic surface metal oxides (W, Nb, and P) to the surface vanadia layer was found to have a beneficial effect on the n-butane oxidation TOF and the maleic anhydride selectivity. The creation of bridging V-O-P bonds had an especially positive effect on the maleic anhydride selectivity.

Full text of DOI:10.1006/jcat.1997.1742

JOURNAL OF CATALYSIS 170, 75–88 (1997)
ARTICLE NO. CA971742
Fundamental Studies of Butane Oxidation over Model-
Supported Vanadium Oxide Catalysts: Molecular
Structure-Reactivity Relationships
,1
,2
,3
ꢀ
ꢀ
ꢀ
ꢀ
Israel E. Wachs, Jih-Mirn Jehng, Goutam Deo, Bert M. Weckhuysen,
,4
V. V. Guliants,† J. B. Benziger,† and S. Sundaresan†
ꢀ
Zettlemoyer Center for Surface Studies, Department of Chemical Engineering, Lehigh University, Bethlehem, Pennsylvania 18015;
and †Department of Chemical Engineering, Princeton University, Princeton, New Jersey 08544
Received June 10, 1996; revised April 30, 1997; accepted May 5, 1997
INTRODUCTION
The oxidation of n-butane to maleic anhydride was investigated
over a series of model-supported vanadia catalysts where the vana-
dia phase was present as a two-dimensional metal oxide overlayer
on the different oxide supports (TiO2, ZrO2, CeO2, Nb2O5, Al2O3,
and SiO2). No correlation was found between the properties of the
Oxidation of n-butane to maleic anhydride over the het-
erogeneous vanadium–phosphorus–oxide (VPO) catalysts
is currently the only industrial catalytic process for partial
oxidation of an alkane. Crystalline vanadyl(IV) pyrophos-
phate, (VO)2P2O7, is present in the most selective cata-
lysts. However, various vanadium phosphates have been
detected in the VPO catalysts depending on the methods of
preparation and conditioning of the conventional catalyst
precursor, vanadyl(IV) hydrogen phosphate hemihydrate.
As a result, the current literature demonstrates the lack of
agreement on the nature of the active phase (1). Various
models of active sites in n-butane oxidation to maleic anhy-
dride have been proposed to date: (i) coherent interfaces
between various VOPO4 phases and (VO)2P2O7 (2), (ii) a
mixture of well-crystallized (VO)2P2O7 and an amorphous
VOPO4 phase involving corner-sharing octahedra (3, 4),
(iii) V5+ species in strong interaction with the VO(PO3)2
structure in the VPO catalysts (5), and (iv) the presence of
V5+ sites on the (200) planes of (VO)2P2O7 (6).
In many experimental studies, the limitations of the cur-
rently available bulk and surface characterization tech-
niques have been coupled with the complex solid-state
chemistry of vanadium phosphate. As a result, considerable
confusion and contradictions exist in the literature concern-
ing the identity of the active sites involved in different steps
of the catalytic reaction (1a, 6, 7). Experimental results sup-
port the conclusions that the best VPO catalysts preferen-
tially expose the (200) planes of (VO)2P2O7 and that the
bulk vanadyl pyrophosphate serves as a support for this
active surface (1a, 6–8). The recent models for n-butane
oxidation are based on the presence of vanadyl dimers in
the (200) plane of vanadyl pyrophosphate (6, 9, 10). The
following types of hypothetical active sites have been pro-
posed to exist on the surface (200) plane (8): (i) Brønsted
==
terminal V O bond and the butane oxidation turnover frequency
(TOF) during in situ Raman spectroscopy study. Furthermore, nei-
ther the n-butane oxidation TOF nor maleic anhydride selectivity
was related to the extent of reduction of the surface vanadia species.
The n-butane oxidation TOF was essentially independent of the
surface vanadia coverage, suggesting that the n-butane activation
requires only one surface vanadia site. The maleic anhydride TOF,
however, increased bya factor of2–3asthesurfacevanadia coverage
was increased to monolayer coverage. The higher maleic anhydride
TOF at near monolayer coverages suggests that a pair of adjacent
vanadia sites may efficiently oxidize n-butane to maleic anhydride,
but other factors may also play a contributing role (increase in
surface Brønsted acidity and decrease in the number of exposed
support cation sites). Varying the specific oxide support changed
the n-butane oxidation TOF by ca. 50 (Ti > Ce > Zr ꢁ Nb > Al > Si)
as well as the maleic anhydride selectivity. The maleic anhydride
selectivity closely followed the Lewis acid strength of the oxide
support cations, Al > Nb > Ti > Si > Zr > Ce. The addition of
acidic surface metal oxides (W, Nb, and P) to the surface vanadia
layer was found to have a beneficial effect on the n-butane
oxidation TOF and the maleic anhydride selectivity. The creation
– –
of bridging V O P bonds had an especially positive effect on the
maleic anhydride selectivity.
ꢂc 1997 Academic Press
1 Current address: Department of Chemical Engineering, National
Chung-Hsing University, Taichung 402, Taiwan, R.O.C.
2 Current address: Department of Chemical Engineering, Indian Insti-
tute of Technology, Kanpur, 208 016, India.
3 Current address: Centrum voor Oppervlaktechemie en Katalyse,
K. U. Leuven, Kardinaal Mercierlaan 92, B-3001 Heverlee, Belgium.
4 Current address: Praxair, Inc., 175 East Park Drive, P.O. Box 44,
Tonawanda, NY 14151.
–
acid sites: probably POH groups; (ii) Lewis acid sites:
75
0021-9517/97 $25.00
c
Copyright ꢂ 1997 by Academic Press
All rights of reproduction in any form reserved.

76
WACHS ET AL.
probably VIV and VV; (iii) one electron redox couple: tio employed during synthesis, which the authors related to
V
IV/VV, VIII/VIV; (iv) two electron redox couple: VIII/VV; the changes in acidity (17b). The studies of others on silica-,
– – – –
(v) bridging oxygen: V O V, V O P, or VO(P)V; (vi) ter- titania-, alumina-, and AlPO4-supported catalysts (17c–g)
minal oxygen: V O, V
V== IV==
O; and (vii) activated molec- demonstrated further that the activity of supported cata-
ular oxygen: peroxo and superoxo species. According to lysts depends on the nature of the support used and that
a recent model (9) the vanadyl dimers from clusters on the selectivity to maleic anhydride is affected by the aver-
the surface (200) planes, each composed of four vanadyl age valence state of vanadium. Overbeek (18) studied n-
dimers, which are isolated from other clusters by terminal butane oxidation to maleic anhydride on titania-supported
pyrophosphate groups. This so-called site isolation in VPO VPO catalysts. The catalystsshowed reasonable activityand
catalysts, which requires only the local ordering in surface selectivity in n-butane oxidation at temperatures ꢁ100 K
structure for the selectivity as well as the inertness of the lower than commercial catalysts. Overbeek concluded that
bulk of the catalyst in n-butane oxidation (11), is analogous the activity of titania-supported VPO catalysts was related
to supported metal oxide catalysts (12). In order to inves- to their reducibility based on the results of the tempera-
tigate the analogy between the bulk VPO and supported ture programmed reduction experiments, as well as to the
metal oxide catalysts we have performed a kinetic and average oxidation state of vanadium ions on the surface
physicochemical study of model supported vanadia cata- (18). Haber et al. (19) have recently studied oxidation of
lysts in the n-butane oxidation. Such a study is particularly n-butane to maleic anhydride on the VPO catalysts con-
valuable, as unlike the bulk VPO catalysts, detailed surface taining different concentrations of alkali and alkali earth
structural information on a molecular level can be obtained metal ion promoters. They found that the promoters af-
from model-supported vanadium oxide catalystscontaining fected the basicity of surface oxygen atoms and that the
two-dimentional vanadium oxide overlayers (13).
rate of n-butane activation was proportional to the basicity
Several earlier studies of alkane oxidation, including that of surface oxygen. On the other hand, selectivity to maleic
of n-butane, have been reported in the literature (14–17). anhydride correlated with the concentration of the acidic
Owens and Kung (14) studied n-butane dehydrogenation sites on the surface, passing through a maximum as the con-
on silica-supported vanadia catalysts. They employed an centration of acid sites increased. Haber et al. (19) believe
oxygen-lean feed gas (O2/butane ratio of 2 : 1) and a high that the surface acidic sites affect selectivity by controlling
reaction temperature (773–813 K) which resulted in pref- adsorption of reaction intermediates with acidic properties.
erential formation of dehydrogenation products, such as
Unlike the present study, the previous studies (14–19)
butenes, and only traces of maleic anhydride were de- were largely conducted on poorly characterized vanadia
tected. Mori et al. (15) studied oxidation of n-butane on and VPO catalysts, where it was difficult to establish the
titania-supported vanadia catalysts at very high tempera- relationship between the catalytic behavior and structure.
tures (ꢁ773 K) and high n-butane conversion. Such se- In previous studies, the experimental conditions were also
vere experimental conditions led to an overoxidation of frequently unfavorable for selective n-butane oxidation
n-butane to carbon oxides and, likely, to combustion of (e.g., very high reaction temperature, high butane conver-
the partial oxidation products, including maleic anhydride. sion, oxygen-lean environment). The present studyemploys
Busca et al. (16) employed titania-supported vanadia cata- well-defined model supported vanadium oxide catalysts,
lysts in n-butane oxidation and observed overoxidation of oxygen-18 labeling, and in situ Raman characterization dur-
the hydrocarbon on very active high surface area catalyst ing butane oxidation to provide structure reactivity rela-
(117 m2/g) at ꢁ723 K. When Busca et al. supported 10 wt.% tionships for this catalytic system.
V2O5 on the low surface area titania (17 m2/g), vanadia
EXPERIMENTAL
formed microcrystals which were detrimental to selective
oxidation and led to n-butane combustion. A number of
studies of supported VPO catalysts have been described in
the literature (17). Nakamura et al. (17a) studied oxidation
of 1-butene on alumina-supported catalysts and reported
reasonably high selectivities to maleic anhydride. The re-
sults of their study (17a) showed that supported VPO cata-
lysts can selectively oxidize C4 hydrocarbons to maleic an-
hydride. Zazhigalov et al. (17b) immobilized the VPO com-
ponent by adsorption of V(V) and P(V) sources onto the
silanol groups of silica support in nonaqueous medium. The
formation of the amorphous VPO clusters improved selec-
tivity to maleic anhydride. The n-butane conversion and
selectivity to maleic anhydride depended on the P/V ra-
a. Materials and Preparation
Supported vanadium oxide catalysts on TiO2 (Degussa,
ꢁ55 m2/g), ZrO2 (Degussa, ꢁ39 m2/g), Nb2O5 (Niobium
Products Company, ꢁ55 m2/g), CeO2 (ꢁ36 m2/g), Al2O3
(Harshaw, ꢁ180 m2/g), and SiO2 (Cab-O-Sil, EH 5, ꢁ380
m2/g) were prepared by the incipient-wetness impregnation
method. A vanadium isopropoxide (Alfa, 95–99% purity)
in a methanol solution (Fisher, 99.9% purity) was used to
impregnate the oxide support under a nitrogen environ-
ment. The samples were initially dried at room tempera-
ture for 2 hr to remove excess methanol and further dried
at 393 K for 16 hr under flowing N2. The samples were finally

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
77
calcined at 773 K for 1 hr under flowing N2; an additional Raman spectra were also recorded by decreasing of the
1 hr under flowing dry air for the TiO2, ZrO2, Nb2O5, and reactant gas flow rate from 100 to 50 cm3/min. The catalysts
CeO2 supports; and an additional 15 hr under flowing dry were subsequently reoxidized by flowing pure oxygen at
air for the Al2O3 and SiO2 supports to form the surface 623 K for 1 hr.
vanadium oxide species on oxide supports. Ammonium
The reactor aspects of the in situ Raman cell were also
metatungstate (Pfaltz & Bauer, 99.9% purity) in aqueous determined with an on-line mass spectrometer and differ-
solution was used to impregnate the 1% V2O5/TiO2 sample ent sample quantities. For the 4% V2O5/ZrO2 catalyst, the
(20), and the same drying-and-calcining procedure as above conversion of butane was less than 1% at 503 K and ca.
was followed to prepare the 6% WO3/1% V2O5/TiO2 sam- 3% at 573 K. The low conversions are due to bypassing
ple. A diluted phosphoric acid (Aldrich, 85% in water) solu- of the catalyst by the gaseous reactants and the presence
tion was preimpregnated on the TiO2 support and used as a of strong mass transfer limitations in the pressed catalyst
support for the preparation of the 1% V2O5/5% P2O5/TiO2 pellet. Consequently, only the exterior region of the cata-
sample (21).
lyst pellet participated in butane oxidation. Fortunately, for
such colored and dark samples the Raman signal primarily
originates from the sample exterior. Thus, the Raman signal
originates from the reaction zone.
b. Raman Spectroscopy
In situ Raman spectroscopy. The in situ Raman spec-
trometer system consists of a quartz cell and a sample
holder, a triple-grating spectrometer (Spex, Model 1877),
a photodiode array detector (EG&G, Princeton Applied
c. Kinetic Tests
During kinetic tests, ca. 1 g ofcatalyst was placed into a U-
Research, Model 1420), and an argon ion laser (Spectra- tube Pyrex glass reactor inside an aluminum split block. The
Physics, Model 165). The sample holder is made from a reactor was heated in the 1.2% n-butane flow at 20 cm3/min
metal alloy (Hastalloy C), and a 100- to 200-mg sample to 494 K, after which the kinetic data were collected for up
disc is held by the cap of the sample holder. The sample to 120 hr under the catalytic reaction conditions. All experi-
holder is mounted onto a ceramic shaft which is rotated by ments were carried out in once-through integral mode. The
a 115 V DC motor at a speed of 1000–2000 rpm. A cylindri- reaction was run under conditions where diffusional limita-
cal heating coil surrounding the quartz cell is used to heat tions could be neglected. CP grade n-butane from Mathe-
the sample and the temperature is measured by an internal son and dryhouse air were metered separatelyusingBrooks
thermocouple. The quartz cell is capable of operating up to Model 52-36A1 V Series mass flow controllers with Model
873 K, and flowing gas is introduced into the cell at a rate 5876 two-channel power supply box and were mixed in de-
of 50–300 cm3/min at atmospheric pressure. The 514.5-nm sired proportions. Only a small fraction of the total flow was
line of the Ar+ laser with 10–100 mW of power is focused metered to the reactor, with the rest being vented.
on the sample disc in a right-angle scattering geometry. An
The effluent stream was analyzed by on-line gas chro-
ellipsoid mirror collects and reflects the scattered light into matography. A side stream ran from the heated effluent
the spectrometer’s filter stage to reject the elastic scattering line to a HP 5790A Series gas chromatograph (GC), where
component. The resulting filtered light, consisting primar- partial oxidation products (mainly MA and traces of acetic
ily of the Raman component of the scattered light, is col- and acrylic acids) were separated on a 2-m-longPorapak QS
lected with an EG&G intensified photodiode arraydetector column. After the partial oxidation products were stripped
which is coupled to the spectrometer and is thermoelectri- from the effluent by passing through a water bubbler, the
cally cooled to ꢃ238 K. The photodiode array detector is effluent samples from a sample loop were injected into two
scanned with an EG&G OMA III optical multichannel an- GC columns in series: a 5-m-long 30% bis-2-ethoxyethyl se-
alyzer (Model 1463).
bacate column to separate CO2 and butane, and a 4-m-long
The in situ Raman spectra were obtained employing the molecular sieve 13X column to separate O2, N2, and CO.
following procedure. The samples were placed into the cell The lines running from the reactor to the HP 5790A gas
and heated to 773 K for 1 hr in a flow of pure oxygen gas chromatograph and the water bubbler were kept at 423 K
(Linde Specialty Grade, 99.99% purity). The dehydrated to prevent condensation of maleic anhydride. Calibration
Raman spectra were collected after the sample was cooled for maleic anhydride was achieved by periodic acid–base
to 503 K in a flow of pure oxygen gas for 30 min. After titrations of the bubbler solution using a phenolphthalein
the above treatment, a 1/4 ratio of oxygen and a gaseous indicator. The bubbler solution was sampled after passing
mixture 1.49% C4H10/He with a flow rate of 100 cm3/min the effluent through the deionized water in the bubbler at a
was introduced into the cell, and the Raman spectra under constant space velocity for 1 hr. Concentrations of carbon
reaction conditions were collected after reaching steady oxides, butane, nitrogen, and oxygen in the effluent were
state (ꢁ30 min at the 503 K reaction temperature). Raman determined using the calibration gas mixture (Airco) con-
spectra were also recorded at higher reaction temperatures taining certified concentrations of the above gases. Closure
of 573 and 623 K. At the 623 K reaction temperature, the on the overall carbon balances was ꢄ5% .

78
WACHS ET AL.
RESULTS
In Situ Raman Spectra during Butane Oxidation
7% V2O5/SiO2. The in situ Raman spectra of the 7%
V2O5/SiO2 catalyst at various stages of the butane oxida-
tion reaction are shown in Fig. 1. Prior to reaction, a sharp
and strong Raman band is present at 1038 cmꢃ1 which is
==
characteristic of the terminal V O bond of the dehydrated
surface vanadium oxide species (22). The 7% V2O5/SiO2
catalyst corresponds to approximately a quarter monolayer
coverage of the two-dimentional surface vanadate over-
layer. Monolayer coverages of metal oxides on SiO2 are
not achievable because of the weak interaction of oxides
with the silica surface (12a). During butane oxidation, the
Raman features of the surface vanadium oxide species are
not affected by the reaction environment. The Raman in-
tensities of silanol groups (Raman band at ꢁ977 cmꢃ1) de-
crease slightly during butane oxidation, which indicates ad-
sorption of some hydrocarbons on the SiO2 surface. Raman
features of adsorbed hydrocarbon functionalities in the
2000–3000cmꢃ1 region, however, could not be detected dur-
ing butane oxidation. No crystalline V2O5 (Raman bands
at ꢁ994, ꢁ702, ꢁ527, ꢁ404, ꢁ284, and ꢁ146 cmꢃ1) was
present in this series of experiments.
FIG. 2. In situ Raman spectra of the 17.5% V2O5 /Al2O3 catalyst dur-
ing n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He,
100 cm3/min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d)
C4H10 /O2/He, 100 cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K;
(f) O2, 623 K.
17.5% V2O5/Al2O3. The in situ Raman spectra of the
17.5% V2O5/Al2O3 catalyst during butane oxidation are
shown in Fig. 2. Monolayer coverage of surface vana-
dia species on Al2O3 corresponds to ꢁ20% V2O5/Al2O3.
==
Both dehydrated terminal V O bonds (Raman band
at ꢁ1028 cmꢃ1) and surface polyvanadate functionalities
(Raman band at ꢁ920, ꢁ800, ꢁ600, ꢁ550, ꢁ350, and ꢁ240
cmꢃ1) are present on the alumina surface. During butane
oxidation, the Raman intensities of these two surface vana-
dium oxide functionalities simultaneously decrease. The
surface polyvanadate functionality is more extensively re-
==
duced than the terminal V O bond, which is manifested in
the dramatically decreased Raman band at ꢁ550 cmꢃ1 char-
– –
acteristicofthe bendingmode ofthe bridgingV O V bond.
The Raman intensities of these two surface vanadia func-
tionalities further decrease at higher reaction temperatures.
Raman features of the surface vanadium oxide species on
alumina disappear with the absence of oxygen at 623 K.
The initial dehydrated Raman features of the surface vana-
dia species on alumina are obtained after reoxidation of the
catalyst at 623 K under flowing oxygen. No crystalline V2O5
Raman features were detected in this series of experiments.
FIG. 1. In situ Raman spectra of the 7% V2O5/SiO2 catalyst during n-
butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10/O2/He, 100 cm3/
min, 503 K; (c) C4H10/O2/He, 100 cm3/min, 573 K; (d) C4H10 /O2/He,
100 cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
79
surface vanadium oxide species. The crystalline V2O5 phase
is not affected by the butane oxidation environment.
3% V2O5/CeO2. The in situ Raman spectra of the 3%
V2O5/CeO2 catalyst during butane oxidation are shown in
Fig. 5. Monolayer coverage of surface vanadia species on
CeO2 corresponds to ꢁ4% V2O5/CeO2 (23). Dehydrated
==
surface vanadium oxide species with the terminal V
O
(Raman band at ꢁ1027 cmꢃ1) and surface polymeric vana-
date functionalities (Raman band at ꢁ920 cmꢃ1) coexist
on the ceria surface. During butane oxidation, the Raman
intensities of these two surface vanadium oxide functionali-
ties (ꢁ1027 and ꢁ920 cmꢃ1) decrease slightly. No crystalline
V2O5 Raman features were found in this series of experi-
ments.
1% V2O5/TiO2. The in situ Raman spectra of the
1% V2O5/TiO2 catalyst during butane oxidation are pre-
sented in Fig. 6. Monolayer coverage of vanadia on titania
corresponds to ꢁ6% V2O5/TiO2 (21). A sharp and strong
Raman band is present at 1026 cmꢃ1, characteristic of the
==
terminal V O group of the dehydrated surface vanadium
oxide species (20). During butane oxidation, the Raman in-
tensity of the surface vanadium oxide species (ꢁ1026 cmꢃ1
)
decreases somewhat at 503 K. At higher reaction tempera-
tures (573 and 623 K), the increased Raman intensity of the
FIG. 3. In situ Raman spectra of the 6% V2O5 /Nb2O5 catalyst during
n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He, 100 cm3/
min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d) C4H10 /O2/He, 100
cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.
6% V2O5/Nb2O5. The in situ Raman spectra of the 6%
V2O5/Nb2O5 catalyst duringbutane oxidation are presented
in Fig. 3. Monolayer coverage of surface vanadia species on
Nb2O5 corresponds to ꢁ7% V2O5/Nb2O5 (21). A sharp and
weak Raman band is present at ꢁ1032 cmꢃ1, characteristic
==
of the terminal V O bond of the dehydrated surface vana-
dium oxide species. A broad Raman band at ꢁ960 cmꢃ1
,
characteristic of the polymeric surface vanadate function-
alities on the niobia surface, is also present. During butane
oxidation, the Raman intensities of these two surface vana-
dium oxide functionalities decrease somewhat. No crys-
talline V2O5 Raman features were present in this series of
experiments.
4% V2O5/ZrO2. The in situ Raman spectra of the 4%
V2O5/ZrO2 catalyst during butane oxidation are shown in
Fig. 4. Monolayer coverage of the surface vanadia species
on ZrO2 corresponds to ꢁ4% V2O5/ZrO2 (21). Both de-
==
hydrated surface terminal V O groups (Raman band at
ꢁ1032 cmꢃ1) and surface polymeric vanadate functionali-
ties are present on the ZrO2 surface. A trace of crystalline
V2O5 (Raman band at ꢁ992 cmꢃ1) is also present on the
ZrO2 surface. During butane oxidation, the Raman inten-
FIG. 4. In situ Raman spectra of the 4% V2O5 /ZrO2 catalyst during
n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He, 100 cm3/
min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d) C4H10 /O2/He, 100
==
sities of the terminal V O and polymeric vanadate func-
tionalities decrease somewhat due to the reduction of the cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.

80
WACHS ET AL.
ence of the surface phosphate species (weak Raman band
at ꢁ1000 cmꢃ1) on the titania surface. Monolayer cover-
age of surface phosphate species on TiO2 corresponds to
ꢁ5% P2O5/TiO2 (25). There is no apparent change of the
Raman features of the surface vanadium oxide species dur-
ing butane oxidation at various temperatures. Thus, the sur-
face vanadium oxide species on P2O5/TiO2 do not appear
to be reduced during butane oxidation.
6% WO3/1% V2O5/TiO2. The in situ Raman spectra of
the 6% WO3/1% V2O5/TiO2 catalyst during butane oxida-
tion are presented in Fig. 9. The dehydrated surface vana-
dium oxide species (Raman band at ꢁ1028 cmꢃ1 due to
==
terminal V O bond) and surface tungsten oxide species
(Raman band at ꢁ1012 cmꢃ1 due to W O bond) coexist
==
on the titinia surface. Monolayer coverage of the surface
tungsten oxide speciescorrespondsto ꢁ8% WO3/TiO2 (20).
The Raman intensity of the dehydrated surface vanadium
oxide species decreases somewhat in the butane oxidation
environment, but the Raman intensity of the surface tung-
sten oxide species remains unaffected. The presence of
the Raman feature at ꢁ920 cmꢃ1 indicates that the addi-
tion of tungsten oxide to the 1% V2O5/TiO2 sample affects
the surface vanadia coverage and somewhat increases the
concentration of the surface polyvanadate functionalities
FIG. 5. In situ Raman spectra of the 3% V2O5 /CeO2 catalyst during
n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He, 100 cm3/
min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d) C4H10 /O2/He, 100
cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.
surface vanadium oxide species indicates that the presence
of water at the lower reaction temperature is primarily re-
sponsible for the reduced Raman intensity at 503 K (24).
There is no apparent reduction of the surface vanadium
oxide species during butane oxidation at 623 K (compare
Figs. 6e and 6f).
5% V2O5/TiO2. The in situ Raman spectra of the
5% V2O5/TiO2 catalyst during butane oxidation are shown
==
in Fig. 7. Both dehydrated surface terminal V O groups
(sharp Raman band at ꢁ1032 cmꢃ1) and surface polyvana-
date functionalities (broad Raman band at ꢁ940 cmꢃ1
)
coexist on the titania surface. During butane oxidation,
the Raman intensities of these two surface vanadium ox-
ide functionalities (ꢁ1032 and ꢁ940 cmꢃ1) decrease at all
temperatures. No crystalline V2O5 Raman features were
present in this series of experiments. Thus, the surface vana-
dium oxide species are partially reduced during butane ox-
idation for the 5% V2O5/TiO2 catalyst.
1% V2O5/5% P2O5/TiO2. The in situ Raman spectra of
the 1% V2O5/5% P2O5/TiO2 catalyst during butane oxida-
tion are presented in Fig. 8. The Raman band at ꢁ1034 cmꢃ1
is characteristic of the dehydrated surface vanadate species.
FIG. 6. In situ Raman spectra of the 1% V2O5/TiO2 catalyst during
n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He, 100 cm3/
min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d) C4H10 /O2/He, 100
The broadening of this Raman band is due to the pres- cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
81
==16
tally complete since some V
O still remained after the
oxygen-18 exchange process due to a small air leak in the
in situ cell. The 600 cmꢃ1 Raman band in Fig. 10 is due to
the ZrO2 support (see Ref. (24)). The butane oxidation was
carried out over this oxygen-exchanged catalyst at 573 K in
the in situ Raman cell in order to monitor the lifetime of
==18
the V
O bond during this reaction. As shown in Fig. 10,
==18
approximately half of the V
O groups was still present
after 17 min of reaction and after 25 min essentially no
==18
V
O could be detected. The conversion of butane under
these conditions was monitored with an on-line mass spec-
trometer and found to be ca. 3% (the selectivity could not
be determined because the lines to the mass spectrometer
were not sufficiently heated).
Kinetic Studies
The results of n-butane oxidation over a number of sup-
ported vanadia catalysts are summarized in Tables 1–3. As
shown in Tables 1–3, two important observations can be
made: (i) maleic anhydride was the dominant partial oxi-
dation product of n-butane oxidation over these supported
vanadia catalysts, and (ii) selective oxidation of n-butane
to maleic anhydride on supported vanadia catalysts took
place at a much lower temperature than in the case of
FIG. 7. In situ Raman spectra of the 5% V2O5 /TiO2 catalyst dur-
ing n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10 /O2/He,
100 cm3/min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K; (d) C4H10
/
O2/He, 100 cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K; (f) O2,
623 K.
(compare Figs. 6 and 9). Neither crystalline V2O5 nor WO3
Raman features (ꢁ805, ꢁ710, and ꢁ320 cmꢃ1) were present
in this series of experiments.
Oxygen-18 Labeled Studies
In situ Raman studies employing labeled oxygen-18 were
also performed to obtain additional insight into the nature
of the metal oxide functionalities participating in the bu-
tane oxidation reaction. In these oxygen-18-labeled stud-
==16
==18
ies, the V
O bonds were replaced with V
O bonds by
reduction in a butane/He stream and reoxidation with an
18O2/He mixture at 723 K. This oxygen exchange resulted
in shifting the Raman band associated with the terminal
==
V
O group from 1025 to 983 cmꢃ1 as shown in Fig. 10 for
– –
the 4% V2O5/ZrO2. The bridging oxygen in the V O V
functionality was also exchanged by this treatment. How-
ever, this exchange is not readily detected in Fig. 10 because
– –
of the broad V O V Raman features and short data collec-
tion timesemployed for the oxygen exchange duringbutane
oxidation. The V2O5/ZrO2 catalyst system was chosen for
these oxygen exchange experiments because preliminary
studies showed that this exchange can be readily achieved
FIG. 8. In situ Raman spectra of the 1% V2O5/5% P2O5/TiO2 catalyst
during n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b) C4H10/O2/He,
100 cm3/min, 503 K; (c) C4H10/O2/He, 100 cm3/min, 573 K; (d) C4H10/O2/
with this catalytic system. The oxygen exchange was not to- He, 100 cm3/min, 623 K; (e) C4H10/O2/He, 50 cm3/min, 623 K; (f) O2, 623 K.

82
WACHS ET AL.
FIG. 10. In situ Raman spectra of 18O-labeled 4% V2O5/ZrO2 catalyst
during n-butane oxidation at 494 K in 1.2 vol.% n-butane in air.
FIG. 9. In situ Raman spectra of the 6% WO3/1% V2O5/TiO2
catalyst during n-butane oxidation. (a) O2, 100 cm3/min, 503 K; (b)
C4H10 /O2/He, 100 cm3/min, 503 K; (c) C4H10 /O2/He, 100 cm3/min, 573 K;
(d) C4H10 /O2/He, 100 cm3/min, 623 K; (e) C4H10 /O2/He, 50 cm3/min, 623 K;
(f) O2, 623 K.
tem was the least active catalyst and the vanadia/titania sys-
tem was the most active catalyst for butane oxidation. The
general activity pattern was Ti > Ce > Zr ꢁ Nb > Al > Si.
The corresponding maleic anhydride TOF for these cata-
lysts was Ti > Nb > Ce > Zr ꢁ Al ꢅ Si. Thus, the vana-
dia/titania catalyst system was the most efficient catalyst
for the oxidation of butane to maleic anhydride over the
supported vanadia catalysts.
conventional VPO catalysts (494 K vs 650 K). For the cat-
alysts containing approximately a monolayer coverage of
surface vanadia (with the exception of vanadia/silica that
possessed only ca. a quarter monolayer coverage), the bu-
tane oxidation turnover frequency (TOF), the number of
n-butane molecules converted per vanadium per second,
was a strong function of the specific oxide support and var-
ied by a factor of ꢁ50 (see Table 1). The vanadia/silica sys-
The vanadia/titania catalysts were further investigated
because ofthe relativelyfavorable propertiesofthiscatalyst
TABLE 1
The Effect of the Metal Oxide Support on n-Butane Oxidation on Supported Vanadia Catalysts
at 494 K in 1.2 vol.% n-Butane in Air
Weight
(g)
Flow
(cm3/min)
Butane conversion
(mol.% )
MAa selectivity
(mol.% )
Butane TOF
MA TOF
(10ꢃ5 sꢃ1
)
Catalyst
(10ꢃ5 sꢃ1
)
b
7% V2O5/SiO2
0.577
0.625
0.885
0.794
0.622
0.566
7.4
17.8
13.9
11.3
14.2
25.5
2.8
7.2
17.3
16.0
10.6
27.8
91.8
44.5
36.7
9.3
12.6
30.5
0.4
0.9
3.6
4.5
6.3
0.3
0.4
1.3
0.4
0.8
6.0
17.5% V2O5/Al2O3
6% V2O5/Nb2O5
4% V2O5/ZrO2
3% V2O5/CeO2
5% V2O5/TiO2
19.6
a MA, maleic anhydride.
b
Considerable error is involved in determining the MA selectivity for this sample due to the extremely low
conversion.

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
83
TABLE 2
Performance of Titania Support in n-Butane Oxidation at 494 K in 1.2 vol.% n-Butane in Air
Weight
(g)
Flow
(cm3/min)
Butane conversion
(mol.% )
MA selectivity
(mol.% )
Butane TOF
MA TOF
(10ꢃ5 sꢃ1
)
Catalyst
(10ꢃ5 sꢃ1
)
1% V2O5/TiO2
2% V2O5/TiO2
3% V2O5/TiO2
4% V2O5/TiO2
5% V2O5/TiO2
7% V2O5/TiO2
0.890
0.612
0.603
0.700
0.566
0.572
8.7
17.6
19.9
30.7
25.5
21.9
16.0
24.2
27.0
30.7
27.8
18.5
22.8
9.9
17.0
29.9
30.5
24.7
12.4
27.7
23.4
25.5
19.6
8.0
2.8
2.7
4.0
6.6
6.0
2.0
system for the partial oxidation of butane to maleic anhy- species on the maleic TOF. The addition of surface niobia
dride. The TiO2 support was essentially inactive for butane species in the 6% Nb2O5/1% V2O5/TiO2 catalyst had a more
oxidation. The influence of surface vanadia coverage on ti- positive effect on both the butane oxidation and the maleic
tania upon the activity and selectivity of butane oxidation to anhydride TOFs. The surface phosphorus oxide additive,
maleic anhydride are presented in Table 2 and Fig. 11. The however, most significantly enhanced the maleic anhydride
butane oxidation TOF is relatively constant below mono- TOF because of the high maleic selectivity. Thus, the bu-
layer coverage (ꢁ6% V2O5/TiO2). Although the butane ox- tane oxidation and maleic TOFs are positively influenced
idation TOF for the V2O5/TiO2 catalystsincreasesslightlyat by the presence of acidic surface metal oxide additives. Mi-
low vanadia coverage, this increase is almost within the ex- crocrystalline metal oxide particles, WO3, appear to have a
perimental error. However, the butane oxidation TOF de- detrimental effect on this reaction.
creases some 50% at coverages above monolayer because
of the presence of microcrystalline V2O5 particles in this
sample. The maleic anhydride TOF depends somewhat on
the surface vanadia coverage and increases by a factor of 2
DISCUSSION
The molecular structure of the surface vanadia species
to 3 up to monolayer coverage. Above monolayer coverage, supported on metal oxides has been extensively charac-
the maleic anhydride TOF declines, reflecting the lower ac- terized with molecular spectroscopies (Raman, IR, solid
tivity of microcrystalline V2O5 particles for the oxidation state vanadium-51 NMR and EXAFS/XANES) in recent
of butane. Thus, both the butane oxidation and the maleic years (13). Under dehydrated conditions, the surface vana-
==
TOFs are influenced by the surface vanadia coverage on the dia species possess terminal V O groups and polymeric
titania support and the most efficient catalyst corresponds functionalities associated with isolated and polymerized
to monolayer coverage.
species on the oxide supports (20, 22, 26). The silica support
The influence of acidic promoters upon butane oxidation represents one exception where only isolated surface vana-
over the 1% V2O5/TiO2 catalyst is summarized in Table 3. dia species exist. The surface vanadia species contain only
==
All the acidic additives had a positive effect on both butane one termina0l V O bond, mono-oxo (27), and three bridg-
0
– –
and maleic TOFs. The addition of surface tungsten oxide ing V O M bonds, where M is either another V atom or
species in the case of the 6% WO3/1% V2O5/TiO2 catalyst the oxide support cation (22, 28). The in situ Raman spectra
had a modest effect on both the butane oxidation and the collected in flowing oxygen at 503 and 623 K (Figs. 1–9) cor-
maleic TOFs. However, the presence of microcrystalline respond to the dehydrated surface vanadia species on the
==
WO3 particles in the 9% WO3/1% V2O5/TiO2 catalyst can- different oxide supports. The terminal V O vibrations for
celled the positive influence of the surface tungsten oxide such dehydrated species occur at 1026–1038 cmꢃ1. These
TABLE 3
The Effect of Acidic Promoters on n-Butane Oxidation on the 1% V2O5/TiO2 Catalyst
at 494 K in 1.2 vol.% n-Butane in Air
Weight
(g)
Flow
(cm3/min)
Butane conversion
(mol.% )
MA selectivity
(mol.% )
Butane TOF
MA TOF
(10ꢃ5 sꢃ1
)
Catalyst
(10ꢃ5 sꢃ1
)
1% V2O5/5% P2O5/TiO2
6% WO3/1% V2O5/TiO2
9% WO3/1% V2O5/TiO2
6% Nb2O5/1% V2O5/TiO2
0.172
0.708
0.465
0.181
5.0
8.9
11.1
7.5
12.1
23.6
14.8
10.7
56.2
26.2
6.8
27.0
34.1
40.5
50.8
15.2
8.9
2.8
35.1
17.8

84
WACHS ET AL.
oxidation TOF. The extent of reduction of the surface vana-
dia species during butane oxidation over the supported
vanadia catalysts containing approximately monolayer cov-
erage followed the pattern Al > Ti > Zr ꢁ Nb ꢁ Ce ꢅ Si.
Furthermore, both the 1% and the 5% V2O5/TiO2 catalysts
were very active for butane oxidation, but the 1% catalyst
did not exhibit any reduction and the 5% catalyst exhib-
ited some reduction during butane oxidation. The maleic
anhydride selectivity did not appear to directly correlate
with the extent of reduction, since the selectivity pattern is
Al > Nb > Ti > Si > Zr ꢁ Ce. Furthermore, the most active
and selective catalyst, 1% V2O5/5% P2O5/TiO2, was not re-
duced during butane oxidation (see Fig. 8). Thus, the extent
of reduction of the surface vanadia species during butane
oxidation does not appear to relate to the butane oxidation
TOF and the maleic anhydride selectivity.
FIG. 11. Performance of titania support in n-butane oxidation on sup-
ported vanadia catalyst at 494 K in 1.2 vol.% n-butane in air.
The Raman band positions during butane oxidation sug-
gest that the terminal vanadyl oxygen is not involved in the
critical kinetic steps of this reaction (most likely the activa-
tion of butane). At near monolayer coverages, the Raman
dehydrated surface vanadia species can become hydrated
in the presence of moisture (24, 29). At temperatures be-
low ꢁ473 K, the amount of adsorbed moisture can be sig-
nificant, equivalent to several monolayers of water. This
adsorbed water can cause the hydrolysis of the bridging
==
band positions of the terminal V O stretch occur between
1027 and 1032 cmꢃ1 for the dehydrated catalysts and 1025
and 1028 cmꢃ1 during butane oxidation at 503 K, with the
exception of vanadia/silica catalyst which possesses approx-
imately a quarter monolayer of surface vanadia species and
has a Raman band at 1038 cm ꢃ1. The Raman band position
–
V O–support bonds with the resulting formation of hy-
drated vanadium oxide clusters. At temperatures of 473–
573 K, the amount of adsorbed moisture present on the
surface corresponds to submonolayer quantities and the
adsorbed water molecules coordinate to the oxygen func-
tionalities via hydrogen bonding. The in situ Raman spec-
tra during butane oxidation at 503 K reflect the presence of
submonolayer quantities of adsorbed moisture in the slight
==
==
of the terminal V O stretch is directly related to the V
O
bond strength (30). However, the Raman band positions of
==
the terminal V O group for the supported vanadia cata-
lysts are almost identical, while the butane oxidation TOF
simultaneously varies by more than an order of magnitude.
Thus, the absence of a relationship between the terminal
shift, ꢁ0–7 cmꢃ1, of the terminal V O Raman bands to
==
lower wavenumbers. At temperatures in excess of 573 K,
the amount of the adsorbed moisture present on the sup-
ported vanadia catalysts is further reduced and the influ-
ence of adsorbed moisture on the in situ Raman spectra
is almost undetectable. The in situ Raman spectra during
butane oxidation at 503–673 K reveal that the dehydrated
surface vanadia species are stable in the reaction environ-
ment and do not agglomerate to form microcrystalline V2O5
particles even in the presence of small amounts of adsorbed
moisture.
==
V
O bond strength and the butane oxidation TOF strongly
suggests that this functionality is not involved in the criti-
cal kinetic steps (most likely the activated adsorption of
butane) during butane oxidation.
In situ Raman experiments employing oxygen-18 ex-
change were also conducted in an attempt to obtain ad-
==
ditional insights into the role of the terminal V O bond
during butane oxidation. Unfortunately, the very short data
collection times employed during this exchange experiment
could not provide any meaningful Raman information as-
The in situ Raman spectra during butane oxidation at
573 K and higher temperatures also reveal that the surface
vanadia species can become partially reduced by the reac-
tion environment (reflected by a decrease in the Raman
intensity in Figs. 1–9). It also appears that the polymeric
functionalities, manifested in broad Raman bands below
1000 cmꢃ1, are somewhat more reducible than the termi-
– –
sociated with the bridging V O V functionalities, which
also underwent oxygen exchange. The characteristic time
required to exchange the 18O-labeled terminal vanadyl oxy-
gen during butane oxidation was tex ꢁ 35 min (when cor-
rected for the incomplete incorporation of oxygen-18 at the
beginning of the experiment). This long period of time can
not be due to bypassing in the in situ Raman cell since com-
plete exchange occurs in a H216O/16O2 stream in less than
7 min (24) (probably even much faster, but good quality
Raman signals cannot be obtained with the current OMA
detectors for very short collection times). The combined
effects of bypassing and strong mass transfer limitations
nal V O groups (Raman bands at 1020–1040 cmꢃ1). For
==
example, the reduction of the polymeric functionalities is
==
more extensive than the terminal V O groups during bu-
tane oxidation over the vanadia/alumina catalyst. However,
we did not observe a relationship between the extent of
reduction of the surface vanadia species and the butane

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
85
in the pressed catalyst wafer do not allow for an accurate basicity measured by the CO2 adsorption and the effective
determination of the TOF for butane oxidation or the trx, negative charge on surface oxygen atoms from the XPS ex-
the characteristic reaction time in the in situ Raman cell. periments. Therefore, we believe that additional characteri-
However, the Raman signal does measure the exterior of zation ofvanadia monolayerson the above supports, such as
the catalyst pellet, i.e., the reaction zone, where the cata- measurements of basicity, the binding energy of oxygen 1s-
lyst is directly exposed to the butane/oxygen stream and electrons, is necessary to both understand the fundamentals
mass transfer limitations are minimal. The characteristic of hydrocarbon activation by supported monolayer cata-
reaction time, trx, was estimated to be ꢁ100 s at 573 K by lysts and establish further mechanistic parallels between
extrapolating the measured TOF at 494 K and using the supported vanadia and bulk VPO catalysts for n-butane
experimental activation energy for butane oxidation over oxidation.
supported vanadia catalysts (31). Thus, to a first approxi-
The maleic anhydride selectivity trends reveal that the
mation, the ratio of the two characteristic reaction times electronegativity (33, 35, 36) properties of the bridging
==
–
is tex/trx ꢁ 20, which suggests that the terminal V O bond V O–support bond are not related to selectivity since the
is relatively stable during butane oxidation (surviving 20 selectivity trend is Al > Nb > Ti > Si > Zr > Ce. This selec-
reaction cycles). However, more accurate measurements tivity trend parallels the strength of the Lewis acidity of the
need to be performed with in situ Raman cells containing oxide supports since alumina has the strongest Lewis acid
packed beds of catalysts. Such in situ cell implementation sites followed by niobia; the other supports have very weak
will prevent bypassing and minimize mass transfer limita- (titania and zirconia) or no Lewis acid sites (silica) (37).
tions associated with the pressed catalyst wafers employed This observation is in agreement with the results of Haber
in the present study. Moreover, faster Raman detectors will et al. (19), who found that selectivity to maleic anhydride on
also aid in completely resolving the issue of the stability of promoted VPO catalysts correlated with the surface acid-
==
the V O bond during butane oxidation.
ity. The low selectivity of the vanadia/ceria catalyst may be
==
The lack of correlation of the terminal V O bond sta- related to the ability of ceria to directly supply oxygen to
bility with the butane oxidation TOFs shifts attention to reactants or products which may result in nonselective reac-
0
– –
– –
the bridging V O M . The bridging V O V bonds in the tions (38). The maleic anhydride TOF and selectivity, how-
polymeric surface vanadia species do not contribute to a ever, do appear to increase with increasing surface vanadia
significant change in the butane oxidation TOF. For ex- coverage on titania or decreasing average distance between
ample, the concentration of polymerized surface function- surface vanadia species (see Fig. 11 and Table 2). This trend
alities increases with vanadium oxide coverage on titania suggests that the efficiency for maleic anhydride formation
(compare 1% and 5% V2O5/TiO2 Raman spectra in Figs. 6 maybe related to the presence oftwo adjacent surface vana-
and 7), but the butane oxidation TOF does not significantly dia sites. This observation is reminiscent of the mechanism
change within experimental error. However, changing the of n-butane oxidation on bulk VPO catalysts proposed by
specific oxide support does have a significant effect on the Trifiro` et al. (6), who found that the rates of hydrocarbon ox-
butane oxidation TOF as shown in Table 1. This observa- idation on the VPO catalysts correlated with the process of
–
tion suggests that the bridging V O–support bonds may be simultaneousabstraction oftwo methylene hydrogen atoms
involved in the critical kinetic steps of butane oxidation. In and proposed an active site for butane oxidation based on
–
order to verify the hypothesis that the V O–support bonds vanadyl dimers present in the (100) plane of vanadyl py-
–
are involved in the C H bond activation, we resorted to the rophosphate, (VO)2P2O7 (6). The current butane oxidation
concepts of the density functional theory (32): the abso- kinetics on the supported vanadia catalysts also reveal that
lute hardness, h, the absolute electronegativity, c (33), and both isolated and adjacent surface vanadia species are able
the principle of hard and soft acids and bases, or HSAB to oxidize butane to maleic anhydride. For example, maleic
(34). However, the catalytic activity of supported vanadia anhydride is formed from vanadia/silica catalysts and the
–
catalysts in the heterolytic C H bond cleavage did not cor- 1% V2O5/TiO2 catalysts where the isolated surface vanadia
relate with the hardness of metal in the metal oxide sup- species is the only species or the dominant species present.
port and, therefore, with the higher electron density on the The mechanism by which butane is oxidized to maleic an-
bridging oxygen. Furthermore, the straightforward corre- hydride over the surface vanadia species is currently un-
lation between the electronegativity of the oxide support known. However, it appears that both isolated and adja-
metal and the activity of the supported vanadia catalysts in cent surface vanadia species can selectively activate butane.
n-butane oxidation is not observed when the Pauling (35) However, other factors may also contribute to the increase
and Sanderson (36) scales of electronegativity are em- in maleic anhydride TOF with increasing surface vanadia
ployed. Recently, Haber et al. have studied n-butane ox- coverage. The exposed titania sites present at lower sur-
idation on a number of promoted bulk VPO catalysts face vanadia coverages may also partially contribute to the
(19). They found that the activity of the promoted VPO lower maleic anhydride TOF since such sites are known to
catalysts in n-butane oxidation correlated with the surface decompose phthalic anhydride during o-xylene oxidation

86
WACHS ET AL.
over vanadia/titania catalysts (39). The surface concentra- (44). Furthermore, the addition of surface tungsten oxide
tion of Brønsted acid sites also increases with surface vana- and niobia species also did not influence the methanol ox-
dia coverage and may contribute to the efficient formation idation TOF and the formaldehyde selectivity (21). These
of maleic anhydride (40). In summary, the maleic anhydride observations suggest that this unimolecular oxidation re-
selectivity appears to be optimum near monolayer cover- action requires only one surface vanadia redox site and
ages at a high concentration of adjacent surface vanadia is insensitive to its immediate environment. In contrast to
species and in the presence of support cations that possess methanol oxidation, the selective catalytic oxidation of NO
strong Lewis acid sites.
with ammonia to N2 and water over the supported vana-
The enhancement in butane oxidation TOF and maleic dia catalysts strongly depended on the surface coverage
selectivity upon the introduction of acidic secondary metal and the presence of the surface tungsten oxide and nio-
oxide additives further confirms the positive role of acidity bia (an increase of about an order of magnitude in TOF).
in this reaction. The butane oxidation TOF was increased These observations suggest that the bimolecular DeNOx
by a factor of 2–3 when acidic surface metal oxide were reaction requires two adjacent surface sites: a surface vana-
introduced to the 1% V2O5/TiO2 catalysts (see Table 3). dia redox site and an adjacent surface acid site (41, 45). The
The surface tungsten oxide and niobia species coordinate oxidation of butane over the supported vanadia catalysts
to the titania support and only interact with the surface does not appear to significantly vary with surface vana-
vanadia species via lateral interactions in the monolayer dia coverage and the presence of adjacent acidic surface
(21). The surface niobia species possess very weak Brønsted tungsten oxide and niobia species (see Tables 2 and 3), but
acid sites (detectable with ammonia adsorption, but not these parameters appear to have a modest beneficial ef-
pyridine adsorption). The surface tungsten oxide species fect. These observations suggest that the oxidation of bu-
possess stronger Brønsted acid sites (detectable by both tane, most likely the activated adsorption step, essentially
ammonia and pyridine adsorption), and they are predom- requires only one surface vanadia redox site and that adja-
inantly present as surface Lewis acid sites (41). Thus, the cent acidic sites have only a second-order effect on butane
presence of these adjacent acidic metal oxide sites to the oxidation. The influence of surface phosphate on the re-
surface vanadia species has a beneficial effect on butane activity of the vanadia/titania catalyst system is extremely
oxidation TOF and maleic anhydride selectivity. The sur- informative. For methanol oxidation, the addition of the
face phosphate species coordinate to the titania support as surface phosphate additive decreases the TOF by an order
– –
tridentate hydrogenphosphate species which possess weak of magnitude due to the replacement of bridging V O Ti
– –
Brønsted acid sites (42). Unlike the surface tungsten ox- bonds with V O P bonds, and there is a modest decrease in
ide and niobia, the surface phosphate directly coordinates formaldehyde selectivity due to the acidic nature of the sur-
–
with the surface vanadia species via reaction of the P OH face phosphate group. For butane oxidation, however, the
– –
bonds (21). The formation of this V O P bond has a pos- addition of the surface phosphate group and the formation
– –
itive effect on the butane oxidation TOF and a significant of bridging V O P bonds has a positive effect on the TOF
enhancement of the maleic anhydride selectivity. This ob- and the maleic selectivity. The different responses of these
servation is consistent with the above observations that two oxidation reactions reflect their different requirements.
0
– –
bridging V O M bonds are critical in the oxidation of bu- The oxidation of methanol to formaldehyde requires only a
– – – –
tane to maleic anhydride and is also in agreement with re- redox site and the substitution of V O P bonds for V O Ti
– –
cent electron microscopy results revealing that the V O P bonds decreases the redox activity of the surface vanadia
bonds are selectively reduced during butane reduction of site. The more complex oxidation of butane to maleic an-
– –
bulk V P O catalysts (43). However, the conditions of the hydride requires both a redox site and some acidic charac-
electron microscopy experiment were very different from ter. Thus, the substitution of less reducible and more acidic
– –
– –
those of conventional VPO catalysis (6). Thus, the incorpo- V O P bonds for V O Ti bonds results in an increase in
ration of acidic secondary metal oxide additives to the sur- both the butane oxidation TOF and the maleic anhydride
face vanadia overlayer has a positive effect on the butane selectivity. The different requirements of these two oxida-
oxidation TOF and maleic anhydride selectivity. Further- tion reactions are also observed in the reaction selectivities
– –
more, the formation of bridging V O P bonds significantly when the specific oxide supports are varied. For methanol
enhances the maleic anhydride selectivity. oxidation, the selectivity towards redox products decreases
Comparison of the butane oxidation over supported with increasing Lewis acid strength of the oxide support
vanadia catalysts with other oxidation reactions over the cation (Ti ꢁ Zr ꢁ Ce > Nb > Al). In contrast, for butane ox-
same catalysts can provide additional insights into the idation the maleic anhydride selectivity increases with in-
fundamental aspects of this reaction. The oxidation of creasing Lewis acid strength of the oxide support cation
methanol to formaldehyde, as well as the formaldehyde (Al > Nb > Ti > Zr ꢁ Ce). For all three oxidation reactions
selectivity, over supported vanadia catalysts was found to the influence of the specific oxide support on the TOF
be essentially independent of the surface vanadia coverage follows a very similar trend (Ti ꢁ Zr ꢁ Ce > Nb > Al > Si)

BUTANE OXIDATION OVER MODEL-SUPPORTED VPO CATALYSTS
87
and is consistent with the conclusion that the bridging ported vanadia catalysts represent a suitable model sys-
– –
V O support bond is involved in the critical kinetic steps. tem capable of providing insights into the mechanism of
However, the range of TOFs is ꢁ104 for methanol oxidation n-butane oxidation on bulk VPO catalysts. The results of
and ꢁ102 for butane oxidation and DeNOx. The lower vari- this work also demonstrate that the goal of preparing the
ation of the TOFs for butane oxidation and DeNOx may active and selective supported catalysts for n-butane oxi-
reflect the additionalrole ofacidityin the controllingkinetic dation to maleic anhydride can be achieved by varying the
steps of these reactions. In summary, the oxidation of bu- properties of the metal oxide support.
tane to maleic anhydride depends on both the redox prop-
–
erties and acidic character of the bridging V O–support
ACKNOWLEDGMENTS
bond.
The work at Lehigh University was supported by the Division of
Basic Energy Sciences, Department of Energy under Grant DEFG02-
93ER14350. V.V.G., J.B.B., and S.S. wish to thank the AMOCO Chemi-
cal Corporation and National Science Foundation (Grant CTS-9100130)
for support. B.M.W. acknowledges the “Belgisch Nationaal Fonds voor
Wetenschappelijk Onderzoek” for a travel grant.
CONCLUSION
In the present study the supported vanadia catalysts
possessing the molecular structures reliably established by
Raman spectroscopy were employed in selective oxidation
of n-butane. The results of the present study demonstrate
that the nature of the metal oxide support plays a crucial
role in defining catalytic properties of vanadia monolayers
in n-butane oxidation to maleic anhydride. The butane ox-
idation TOF varied by ꢁ100 on changing the specific oxide
support. No evidence to suggest that the reducibility and
the average oxidation state of vanadium affect the catalytic
behavior has been found. The in situ Raman experiments
also demonstrated that the characteristics of the terminal
vanadyl oxygen do not correlate with the n-butane oxida-
tion TOFs and suggest the critical involvement of the bridg-
REFERENCES
1. (a) Catal. Today 16, 1 (1993) [Proceedings, Vanadyl Pyrophosphate
Catalysts (G. Centi, Ed.), Elsevier, Amsterdam] and references
therein; (b) Guliants, V. V., Benziger, J. B., Sundaresan, S., Yao, N.,
and Wachs, I. E., Catal. Lett. 32, 379 (1995); (c) Sananes, M. T.,
Hutchings, G. J., and Volta, J. C., J. Catal. 154, 253 (1995); Kiely,
C. J., Sajip, S., Ellison, I. J., Sananes, M. T., Hutchings, G. J., and Volta,
J. C., Catal. Lett. 33, 357 (1995).
2. Bordes, E., Catal. Today 1, 499 (1987), and references therein.
3. Bergeret, G., David, M., Broyer, J. P., Volta, J. C., and Hecquet, G.,
Catal. Today 1, 37 (1987), and references therein.
4. Harrouch Batis, N., Batis, H., Ghorbel, A., Vedrine, J. C., and Volta,
J. C., J. Catal. 128, 248 (1991).
5. Sananes, M. T., Tuel, A., and Volta, J. C., J. Catal. 145, 251 (1994).
6. Centi, G., Trifiro`, F., Ebner, J. R., and Franchetti, V. M., Chem. Rev.
88, 55 (1988), and references therein.
–
ing V O–support bond in n-butane oxidation (most likely
in the activated adsorption step).
Titania as a support resulted in favorably high activity
and selectivity to maleic anhydride in accordance with pre-
vious observations (18) and was, therefore, used to study
the effect of vanadia coverage on catalytic performance.
This study indicated that the specific activity per surface
vanadia species remained essentially unchanged with vana-
dia coverage until some microcrystalline vanadia detected
in the catalysts at high coverages resulted in reduced spe-
cific activity. The results of this study also suggested that
isolated surface vanadate species are capable of butane ox-
idation to maleic anhydride, although this oxidation was
more efficient when multiple vanadate sites were present
at higher coverages (two- to threefold increase in maleic
anhydride TOF). This correlation probably reflects the ac-
tivated adsorption of butane, since the complete reaction is
very complicated. Microcrystalline vanadia was found to be
detrimental for the process of maleic anhydride formation.
The negative effect of supported microcrystalline oxides
was also observed in the case of tungsten oxide promoter.
The selectivity of the supported catalysts to maleic anhy-
dride correlated with the Lewis acid strength of the metal
oxide promoters. Especially high selectivity to maleic an-
7. Catal. Today 1, 2 (1987) [Proceedings, 1st Symposium on New De-
velopments in Selective Oxidation (P. P. Ruiz and B. Delmon, Eds.),
Elsevier, Amsterdam] and references therein.
8. Cavani, F., and Trifiro`, F., Catalysis 11, 246 (1994), and references
therein.
9. Agaskar, P. A., DeCaul, L., and Grasselli, R. K., Catal. Lett. 23, 339
(1994).
10. Schiøtt, B., and Jørgensen, K. A., Catal. Today 16, 79 (1993), and ref-
erences therein.
11. Pepera, M. A., Callahan, J. L., Desmond, M. J., Milberger, E. C., Blum,
P. R., and Bremer, N. J., J. Am. Chem. Soc. 107, 4883 (1985).
12. (a) Wachs, I. E., Deo, G., Vuurman, M. A., Hu, H., Kim, D. S., and
Jehng, J.-M., J. Mol. Catal. 82, 443 (1993); (b) Wachs, I. E., Deo, G.,
Vuurman, M. A., Hu, H., Kim, D. S., and Jehng, J.-M., in “Proceedings,
10th International Congress on Catalysis, Budapest, 1992” (L. Guczi,
F. Solymosi and P. Tetenyi, Eds.), p. 543, Akade´miai Kiado´, Budapest,
1993.
13. Deo, G., Wachs, I. E., and Haber, J., Critical Rev. Surf. Chem. 4, 141
(1994).
14. Owens, L., and Kung, H. H., J. Catal. 144, 202 (1993).
15. Mori, K., Miyamoto, A., and Murakami, Y., J. Phys. Chem. 89, 4265
(1985).
16. Busca, G., Centi, G., and Trifiro`, F., Appl. Catal. 25, 265 (1986).
17. (a) Nakamura, M., Kawai, K., and Fujiwara, Y., J. Catal. 34, 345 (1974);
(b) Zazhigalov, V. A., Zaitsev, Y. P., Belousov, V. M., Parlitz, B., Hanke,
¨
W., and Ohlman, G., React. Kinet. Catal. Lett. 32, 209 (1989); (c)
– –
hydride was found when the V O P bonds formed after
Ramstetter, A., and Baerns, M., J. Catal. 109, 303 (1988); (d) Varma,
R. L., and Saraf, D. N., Ind. Chem. Eng. 20, 42 (1978); (e) Kuo, P. S.,
and Yang, B. L., J. Catal. 117, 301 (1989); (f) Martinez-Lara, M.,
addition of phosphorus oxide in accordance with previous
observations (17, 18). These findings indicate that the sup-

88
WACHS ET AL.
Moreno-Real, L., Pozas-Tormo, R., Jimenez-Lopez, A., Bruque, S., 29. Deo, G., and Wachs, I. E., J. Phys. Chem. 95, 5889 (1991).
Ruiz, P., and Poncelet, G., Can. J. Chem. 70, 5 (1992); (g) Derewin- 30. Hardcastle, F. D., and Wachs, I. E., J. Phys. Chem. 95, 5031 (1991).
ski, M., Haber, J., Kozlowski, R., Zazhigalov, V. A., Zajtsev, J. P., 31. Owens, L., and Kung, H. H., J. Catal. 144, 202 (1993).
Bacherikowa, I. W., and Belousov, W. M., Bull. Polish Acad. Sci. Chem. 32. Parr, R. G., and Yang, W., “Density-Functional Theory of Atoms and
39, 403 (1991).
18. Ruitenbeek, M., Overbeek, R. A., van Dillen, A. J., Koningsberger, 33. Pearson, R. G., Inorg. Chem. 27, 734 (1988).
D. C., and Geus, J. W., Recl. Trav. Chim. Pays-Bas 115, 519 (1996). 34. Pearson, R. G., J. Am. Chem. Soc. 85, 3533 (1963).
19. Zazhigalov, V. A., Haber, J., Stoch, J., Bacherikova, I. V., Komashko, 35. Pauling, L., “The Chemical Bond,” Cornell Univ. Press, Ithaca, NY,
G. A., and Pyatnitskaya, A. I., Appl. Catal. A 134, 225 (1996). 1967.
20. Vuurman, M. A., Hirt, A. M., and Wachs, I. E., J. Phys. Chem. 95, 9928 36. Sanderson, R. T., “Polar Covalence,” Academic Press, New York,
Molecules,” Chapters 5 and 10, Oxford Univ. Press, New York, 1989.
(1991).
1983.
21. Deo, G., and Wachs, I. E., J. Catal. 146, 335 (1994).
22. Das, N., Eckert, H., Hu, H., Wachs, I. E., Walzer, J., and Feher, F.,
J. Phys. Chem. 97, 8240 (1993).
23. Deo, G., and Wachs, I. E., unpublished manuscript.
24. Jehng, J.-M., Deo, G., Weckhuysen, B. M., and Wachs, I. E., J. Mol.
Catal. A: Chemical 110, 41 (1996).
37. Datka, J., Turek, A. M., Jehng, J.-M., and Wachs, I. E., J. Catal. 135,
186 (1992).
38. (a) Conatkes, S., and Wachs, I. E., unpublished manuscripts (b)
Kalakkad, D., Datye, A. K., and Robota, H., Appl. Catal. B 1, 191
(1992).
39. Wachs, I. E., Saleh, R. Y., Chan, S. S., and Chersich, C. C., Appl. Catal.
15, 339 (1985).
25. Hardcastle, F. D., and Wachs, I. E., unpublished manuscript.
26. (a) Chan, S. S., Wachs, I. E., Murrell, L. L., Wang, L., and Hall, W. K., J. 40. (a) Miyata, H., Fujii, K., and Ono, T., J. Chem. Soc. Faraday Trans. 84,
Phys. Chem. 88, 5831 (1984); (b) Went, G., Oyama, T., and Bell, A. T.,
J. Phys. Chem. 94, 4240 (1990); (c) Vuurman, M. A., and Wachs, I. E.,
J. Phys. Chem. 96, 5008 (1992).
3121 (1988); (b) Turek, A. M., Wachs, I. E., and DeCanio, E., J. Phys.
Chem. 96, 5000 (1992).
41. Wachs, I. E., Deo, G., Andreini, A., Vuurman, M. A., de Boer, M., and
Amiridis, M., J. Catal. 161, 211 (1996).
42. Ramis, G., Busca, G., and Lorenzelli, V., in “Structure and Reactivity
of Surface,” (C. Morterra, A. Zecchina, and G. Costa, Eds.), Elsevier,
Amsterdam, 1989.
27. G. Ramis, G., C. Cristiani, C., Forzatti, P., and Busca, G., J. Catal. 116,
574 (1990).
28. (a) Eckert, H., and Wachs, I. E., J. Phys. Chem. 93, 6796 (1989); (b)
Tanaka, T., Yamashita, H., Tsuchitani, R., Funabiki, T., and Yoshida,
S., J. Chem. Soc. Faraday Trans. 1 84, 2987 (1988); (c) Yoshida, S., 43. Gai, P. L., and Kourtakis, K., Science 267, 661 (1995).
Tanaka, T., Hanada, T., Hiraiwa, T., and Kanai, H., Catal. Lett. 12, 277
44. Deo, G., and Wachs, I. E., J. Catal. 146, 323 (1994).
(1992).
45. Topsoe, N. Y., Topsoe, H., and Dumesic, J. A., J. Catal. 151, 226 (1995).