1. Acid–base reactions on alumina-supported niobia
Mona A. Abdel-Rehim a
, Ana Carlota B. dos Santos b
,
Vera Lu´cia L. Camorim b
, Arnaldo da Costa Faro Jr.a,*
a
Departamento de Fı´sico-Quı´mica, Instituto de Quı´mica, UFRJ, Ilha do Funda˜o, CT,
Bloco A, Rio de Janeiro, RJ, CEP 21949-900, Brazil
b
PETROBRAS/CENPES/HTPE, Ilha do Funda˜o, Q 7, Rio de Janeiro, RJ, CEP 21949-900, Brazil
Received 1 October 2005; received in revised form 24 February 2006; accepted 5 March 2006
Available online 18 April 2006
Abstract
Catalysts containing from 8 to 28% wt. of niobia supported on g-alumina by impregnation with niobium pentaethoxide solutions were
characterized by infrared spectroscopy of adsorbed pyridine as a probe for acidic sites and CO2 adsorption as a probe for basic sites. The catalysts
had their activity measured in isopropanol dehydration at 453 K, 1-butene isomerization at 348 K and in cumene dealkylation at 728 K. The density
and strength of the Lewis acid sites and of the basic sites decreased with niobium content. On the other hand, the density of Brønsted acid sites
increased in the same direction. Different reactions responded differently to niobium addition. In isopropanol dehydration, the main factor
responsible for the observed decrease in catalytic activity was the decrease in concentration of basic sites. Activity for 1-butene isomerization also
decreased with niobium addition, but the main factor seemed to be the decrease in concentration of alumina-associated Lewis acidic sites. Evidence
from cis-/trans-2-butene ratio indicated that niobium addition modifies the properties of neighboring aluminum sites. Brønsted acidic sites created
by niobium addition are responsible for the development of activity in the cumene dealkylation reaction, but the sites associated with
tridimensional niobia species seem to be more effective in this reaction.
# 2006 Elsevier B.V. All rights reserved.
Keywords: Niobium oxide; Alumina; Isopropanol dehydration; 1-Butene isomerization; Cumene cracking
1. Introduction
Perhaps the most outstanding property of several niobium
compounds (oxide [1], sulfides [2] and phosphate [3]) is their
pronounced surface acidity. This has led to the application of
niobium oxide and niobic acid (hydrated niobium oxide), both
pure [2,4] and promoted with phosphate [5,6] or sulfate [6], and
niobium phosphate [7], in acid catalysed reactions, such as
alcohol dehydration [2], esterification [2], etherification [7] and
dealkylation of alkylbenzenes [6]. The subject of catalysis on
niobium materials has been reviewed recently [8,9]. However,
the practical application of niobium-based catalysts is severely
restricted by the elevated price of the raw-material and by the
difficulty in controlling their textural properties (surface area,
pore volume and pore size distribution).
A solution to these problems is to disperse the niobium-
containing phase as a thin layer on the surface of a suitable
support, such as silica or alumina, but considerable modifica-
tion of the properties of the supported oxide may occur, due to
interaction with the underlying support.
Alumina-supported niobia materials have been extensively
characterized by Wachs and coworkers [10–17] with respect to
the structure of the supported niobium species and surface
acid–base properties, but their catalytic properties were only
investigated in methanol conversion reactions.
In this paper we investigate the surface acid–base properties
of alumina-supported niobia, containing from 1/3 to 4/3 of the
theoretical monolayer, and their relation to catalytic activity
and selectivity in a series of model reactions (isopropanol
dehydration, 1-butene isomerization and cumene dealkylation).
2. Experimental
2.1. Catalysts preparation
The g-alumina used as the support was obtained by
calcination at 773 K for 3 h of a commercial boehmite, Pural
SB, provided by Condea Chemie.
www.elsevier.com/locate/apcata
Applied Catalysis A: General 305 (2006) 211–218
* Corresponding author. Tel.: +55 2125627821.
E-mail addresses: carlota@petrobras.com.br (A.C.B. dos Santos),
farojr@iq.ufrj.br (A. da Costa Faro Jr.).
0926-860X/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2006.03.023
2. The supported catalysts were prepared by multiple incipient
wetness impregnation steps with niobium pentaethoxide
dissolved in n-hexane. At each impregnation step, ca. 1/3 of
a theoretical niobia monolayer (6.3 niobium atoms/nm2
according to Asakura and Iwasawa [18]) was introduced.
After each step, the materials were dried at 343 K for 2 h under
a nitrogen stream, hydrolyzed for 4 h at 423 K under a nitrogen
stream saturated at room temperature with water vapor in order
to decompose the supported ethoxide and calcined at 773 K for
3 h. Since up to four successive impregnations were performed,
catalysts were prepared containing from 1/3 to 4/3 of the
theoretical monolayer.
Bulk niobium oxide was obtained by calcination of a niobic
acid (Nb2O5ÁxH2O) sample provided by CBMM under the same
conditions described for the catalysts.
2.2. Catalyst characterization
The chemical composition of the catalysts was determined
by mass spectrometry with an induced plasma source (ICP-
MS).
The presence of crystalline species in the catalysts was
investigated by powder X-ray diffraction (XRD) using a
Phillips DW-1710 diffractometer using the Cu Ka radiation and
graphite monochromator.
Surface areas were obtained from nitrogen adsorption
isotherms at 77 K using the BET method. The isotherms were
measured in a Micromeritics ASAP 2010C volumetric
apparatus. The samples were pre-treated at 623 K under
vacuum, until a degassing rate smaller than 2 Â 10À3
Torr minÀ1
was attained.
FTIR measurements were carried-out in a Nicolet Magna
760 spectrophotometer. Self-supported wafers of the samples,
containing 10 mg cmÀ2
, were evacuated at 10À5
Torr and
723 K for 16 h. After cooling to room temperature, the
spectrum was recorded. The samples were then exposed to
0.5 Torr pyridine vapors and, after evacuation at 423 K until the
vacuum returned to its base level, a second spectrum was
recorded.
Carbon dioxide adsorption isotherms were measured in the
same volumetric apparatus mentioned above. The samples were
pre-treated in situ beginning with a helium flow at 573 K
followed by evacuation at the same temperature. The samples
were then cooled to 308 K, where the adsorption measurements
were performed between 0.3 and 400 Torr. After evacuation at
308 K, a second CO2 adsorption isotherm was measured at the
same temperature and pressure range as the first. The amount of
irreversibly chemisorbed CO2 was obtained by subtraction of
the second isotherm from the first. The subtraction isotherm
was essentially a horizontal line in all cases.
Diffuse reflectance spectra (DRS) in the UV–vis region were
measured in a Varian Cary 500 spectrophotometer equipped
with a ‘‘praying mantis’’ DRS accessory, in the 180–800 nm
range. The samples were oven-dried at 393 K for 12 h and kept
in a dessicator until just before the analyses. The alumina
support was used as the reflectance standard for the supported
niobia catalysts and barium sulfate for the pure alumina and
niobia materials. The spectra are presented in terms of the
Schuster–Kubelka–Munk (SKM) function, F(R1).
2.3. Catalytic measurements
The catalytic measurements were performed in a batch all-
glass gas–solid reaction system with external re-circulation of
the gas phase using a non-contaminating bellows pump and
greaseless Kontes taps. In all cases, the catalysts were pre-
treated by evacuation at ca. 10À5
Torr and 728 K for 6 h. A
desired initial partial pressure of reactant was introduced into
the re-circulation loop and the pressure was brought to ca.
760 Torr by addition of helium containing a known amount of
nitrogen as an internal standard. After contacting the reaction
mixture with the catalyst, samples were periodically drawn
from the gas-phase and injected into a methylsiloxane capillary
column (100.0 m  250 mm  0.50 mm) installed in an Agi-
lent 6890 gas-chromatograph equipped with a model 5973
mass-selective detector (GC–MS system). Response factors for
quantitative analyses were measured by introducing known
amounts of the reactants and products in the re-circulation loop,
followed by injection into the GC–MS system. Initial reaction
rates were obtained from the slopes at time zero of reactant
conversion versus time functions.
For cumene dealkylation, isopropanol dehydration and 1-
butene isomerization, reaction temperatures were 728, 353 and
348 K, respectively, and initial reactant partial pressures were
0.59, 8.8 and 11.8 Torr, respectively.
3. Results
3.1. Chemical composition, textural properties and XRD
results
Chemical compositions and main textural properties of the
different catalysts are summarized in Table 1.
Addition of niobia only caused some significant decrease of
the surface area (>10%) for the catalyst containing more than
one niobia theoretical monolayer. This is not accompanied by a
corresponding increase in the average pore diameter that
would be expected if this decrease in surface area was caused
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218212
Table 1
Chemical compositions and textural properties of the catalysts
Sample Niobiaa
(wt.%)
TMb
(%)
SAc
(m2
gÀ1
)
PVd
(cm3
gÀ1
)
APDe
(nm)
Al2O3 – – 223 0.55 9.8
NA-8 8.2 28.6 214 0.53 9.8
NA-14 14.3 53.8 219 0.52 9.6
NA-19 18.6 73.6 200 0.47 9.5
NA-27 27.2 120 182 0.43 9.4
Nb2O5 100.0 – 40 – –
a
Determined by ICP-MS.
b
Percentage relative to the theoretical monolayer assumed to be 6.3 Nb/nm2
[REF].
c
BET surface area expressed per gram of support.
d
Total pore volume.
e
Average pore diameter obtained from 4 Â PV/SA.
3. by pore-mouth blockage. On the contrary, the average pore
diameter actually decreased slightly, which suggests that the
niobia was deposited as a layer on the pore walls.
In the X-ray diffraction profiles for the supported niobia
catalysts, only broad reflections characteristic of the low
crystallinity g-phase support were observed, which could be
recognized from the diffraction maxima at interplanar distances
of ca. 0.14 nm (440 plane), 0.15 nm (511 plane) and 0.23 nm
(222 plane) (Joint Committee on Powder Diffraction Standards,
files 10-0425 and 29-0063). This is in contrast with the pure
niobia support that, upon calcination at 773 K, gave rise to
sharp reflections at interplanar distances characteristic of the
TT form of niobium oxide. Therefore, the alumina support
prevents the growth of large niobia crystals even for a niobia
content higher than a theoretical monolayer.
3.2. UV–vis spectroscopy results
The DRS spectra of the supported catalysts and pure niobia
in the UV–vis region are shown in Fig. 1. In this region, intense
bands corresponding to ligand to metal charge transfer
transitions are observed. The bands are seen to increase in
intensity and to shift to higher wavelength as the niobium
content increases. In Fig. 2, the energy of the absorption edge,
obtained by extrapolation to zero absorption of [F(R1)hn]2
[19], is shown as a function of the niobia loading. The
absorption edge is seen to decrease from ca. 4.5 to 3.8 eVas the
niobium loading increases from 1.8 to 9.3 niobium atoms/nm2
.
The higher value is close to the one reported by Gao et al. [20]
for isolated NbO4 species in MCM41-supported niobium, while
the lower one is close to the one found here for niobium oxide
obtained by calcination of niobic acid at 773 K, shown in Fig. 2
as a dashed line.
3.3. FTIR results
The FTIR spectrum for the alumina in the OH bond
stretching region presented the typical bands described before
in the literature for this material. Three main absorption regions
were observed, centered around 3770, 3730 and 3685 cmÀ1
, as
well as a weak shoulder at the high wave-number side of the
band at ca. 3770 cmÀ1
(ca. 3785 cmÀ1
). According to the model
of the g-alumina surface proposed by Kno¨zinger and
Ratnasamy [21], the main absorption bands in this region
correspond, respectively, to isolated hydroxyl groups coordi-
nated to one (type I hydroxyls), two (type II hydroxyls) and
three (type III hydroxyls) Al3+
ions. From the peak areas, pure
niobia had a smaller degree of hydroxylation by about one order
of magnitude than g-alumina calcined at the same temperature
and hydroxyl bands were observed at 3700 and 3740 cmÀ1
.
Peak intensities showed that the added niobia titrates the
hydroxyl groups on the alumina surface, as already observed by
Wachs and coworkers [16]. With the two largest niobia
contents, only a broad band with maximum at ca. 3700 cmÀ1
remained on the surface.
In the FTIR spectra of adsorbed pyridine, intense absorption
was observed at about 1450, 1490 and 1540 cmÀ1
, as well as a
group of peaks in the region 1610–1620 cmÀ1
. Of these, the
peak at 1450 cmÀ1
(19b vibration) and the group at 1610–
1620 cmÀ1
(8a vibration) are characteristic of pyridine
coordinated to Lewis acid sites. The intensity of the 19b band
is proportional to the total number of Lewis sites, whereas the
positions of the peaks in the group at 1610–1620 cmÀ1
are
sensitive to the strength of these sites [22]. With g-alumina,
peak maxima occurred at ca. 1615 and 1620 cmÀ1
. With pure
niobia, only one peak was observed in this region at ca.
1610 cmÀ1
indicating that weaker Lewis sites are present at the
surface of this material [22,23]. All three peaks can be
discerned in the spectra of the NA materials and the relative
intensity of the peak at 1610 cmÀ1
clearly increased with
increasing niobium content.
A small peak at 1543 cmÀ1
, characteristic of protonated
pyridine clearly growed with increasing niobia content.
Fig. 3 summarizes the results for the density of Brønsted and
Lewis acid sites determined by integration of the bands with
absorption maxima at ca. 1540 and 1450 cmÀ1
, respectively
[24], of the FTIR spectrum of pyridine adsorbed on the different
materials. Acid site densities were calculated using the
integrated molar extinction coefficients published by Emeis
[25] (1.67 and 2.22 cm mmolÀ1
for Brønsted and Lewis sites,
respectively). It is clear that the density of Lewis acid sites
decreases approximately linearly with the niobia loading,
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218 213
Fig. 1. Diffuse reflectance spectra in the UV region.
Fig. 2. Band gap energy vs. niobium loading. Dashed line corresponds to bulk
niobium oxide calcined at 773 K.
4. whereas the density of Brønsted sites increases in a markedly
non-linear fashion: catalysts containing up to ca. 2 niobium
atoms/nm2
show very little Brønsted acidity and the largest
increase in this property occurs between ca. 2 and 6 Nb/nm2
.
Data relative to the distribution of Lewis acid sites
associated with surface aluminum and niobium were obtained
by deconvolution of the group of peaks around 1615 cmÀ1
in
the FTIR spectrum of adsorbed pyridine [22,23], using gaussian
functions. From Fig. 4, it is clear that this group is comprised of
four distinct bands: band A at 1610 cmÀ1
may be assigned to
niobium sites, since it increases with niobium content; bands B
and C are assigned to aluminum sites, since they are already
present on the surface of pure alumina; a broad band, D,
observed with approximately constant intensity with all
catalysts, may be assigned to weakly adsorbed pyridine.
The added areas of the peaks corresponding to bands B and
C and the one corresponding to band A, normalized with
respect to surface area, are plotted in Fig. 5 as a function of
niobium content. It is clear that, although there is large decrease
of bands B and C associated with aluminum Lewis sites, about
1/3 of these sites remain exposed at the surface even for a
niobium loading larger than one theoretical monolayer. This
indicates that some vertical growth of the niobia layer occurs
with increasing niobium content leaving some of the alumina
surface exposed. The band associated with niobium sites
increases approximately linearly with the niobium loading.
3.4. Carbon dioxide chemisorption
Fig. 6 shows the density of surface basic sites, as determined
from CO2 chemisorption, as a function of niobium loading. The
amount of adsorbed CO2 decreases strongly with niobium
addition and, for a niobium content corresponding to about 1/2
of the theoretical monolayer, it is only ca. 10% of the value
obtained with the alumina support. The chemisorption
measurements were carried-out at 308 K, at which temperature
CO2 is mainly adsorbed on surface hydroxyls, as bicarbonate
species [26]. The fact that the decrease in CO2 chemisorption
with niobium content was much more pronounced than the
degree of surface de-hydroxylation suggests that the supported
niobia is titrating the more basic fraction of the hydroxyl
groups, as proposed before by Wachs and coworkers [16].
3.5. Catalytic activity in isopropanol conversion
Table 2 shows the results for catalytic activity and selectivity
in isopropanol conversion. Only propene and di-isopropyl
ether, but no acetone, are observed as reaction products under
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218214
Fig. 3. Acid site densities from FTIR integrated peak areas for adsorbed
pyridine as a function of niobium loading. (*) Ca. 1450 cmÀ1
(Lewis sites);
(~) ca. 1540 cmÀ1
(Brønsted sites).
Fig. 4. Deconvolution of group of peaks corresponding to the 19b vibration
mode of adsorbed pyridine. (A) Niobium sites; (B, C) aluminium sites; (D)
weakly adsorbed pyridine.
Fig. 5. Integrated peak areas for the 19b vibration mode of adsorbed pyridine as
a function of niobium loading. (*) Aluminium sites; (~) niobium sites.
Fig. 6. Basic site density measured from carbon dioxide chemisorption at
308 K as a function of niobium loading.
5. the conditions used here. Selectivity for di-isopropyl ether
formation, measured at 10% conversion, was between ca. 11
and 24% for all alumina-supported catalysts, but with pure
niobia, only propene is observed among the reaction products.
Selectivity for di-isopropylether goes through a maximum as a
function of the niobia loading.
Fig. 7 shows that a linear correlation exists between catalytic
activity in isopropanol conversion and the density of basic sites,
as measured by carbon dioxide irreversible chemisorption. In
an attempt to correlate the activity with strong alumina-
associated Lewis acid sites, although there seemed to be a
general trend of increasing activity with increasing alumina-
associated Lewis acidity, the observed correlation was poor. An
even worse correlation was obtained when the activity was
plotted against the total Lewis acidity. This is obvious by
comparing Figs. 3 and 7: while the total Lewis acidity
decreased about 35% along the catalyst series relatively to the
alumina support, catalytic activity decreased by one order of
magnitude.
3.6. Catalytic activity in 1-butene isomerization
Table 3 summarizes the results for catalytic activity and
selectivity at 10% conversion in 1-butene isomerization to cis-
and trans-2-butene. Activity is approximately constant up to a
niobia content corresponding to ca. 50% of a monolayer and
then decreases pronouncedly as the niobium content increases.
The cis/trans ratio goes through a maximum at the same niobia
loading. The cis/trans ratio with the niobia catalyst is
considerably smaller than with alumina and the supported
niobia catalysts. The predicted thermodynamic conversion at
our reaction temperature is ca. 94% and the thermodynamic cis/
trans ratio is ca. 0.4.
Some correlation is found between the intrinsic rate con-
stant and the density of alumina-associated Lewis sites, as
shown in Fig. 8. The correlation with total Lewis acidity is
very poor.
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218 215
Fig. 7. Initial rate of isopropanol dehydration at 453 K as a function of basic
site density (measured from CO2 chemisorption). (*) niobia/aluminas; (~)
niobium oxide.
Table 2
Activity and selectivity in isopropanol dehydration at 453 K
Sample
Al2O3 NA-8 NA-14 NA-19 NA-27 Nb2O5
r0
a
(Â105
mol minÀ1
gÀ1
) 14.5 7.7 1.8 2.5 1.3 0.15
% Selectivity for i-propyletherb
11.3 20.4 23.5 16.7 13.0 0.0
a
Initial rate of reaction.
b
Measured at 10% conversion. The only other product identified was propene.
Fig. 8. Initial rate of 1-butene isomerization at 348 K as a function of the
density of alumina-associated Lewis acid sites (in arbitrary units/m2
). (*)
Niobia/aluminas; (~) niobium oxide.
Table 3
Activity and selectivity in 1-butene isomerization at 348 K
Sample
Al2O3 NA-8 NA-14 NA-19 NA-27 Nb2O5
r0
a
(Â105
mol minÀ1
gÀ1
) 4.5 3.4 4.1 1.8 1.1 0.17
cis-/trans-2-Butene ratiob
1.8 2.6 2.9 2.5 2.3 1.3
a
Initial rate of reaction.
b
Measured at 10% conversion.
6. 3.7. Catalytic activity in cumene dealkylation
The cumene dealkylation reaction was used as a probe for
Brønsted-type surface acidity, since this reaction is known to
occur on this type of site [27], as illustrated in Scheme 1. In our
work, only benzene and propene were observed as reaction
products.
A linear dependence between benzene gas-phase concen-
tration and reaction time was obtained, showing that, under the
present conditions, benzene dealkylation follows zero-order
kinetics. The intrinsic initial rate of dealkylation is shown in
Fig. 9 as a function of Brønsted acidity. At 728 K, the alumina
support has negligible activity, which is agreement with the
absence on its surface of Brønsted acid sites strong enough to
protonate pyridine. The activity is seen to increase with
Brønsted acidity, but in a non-linear fashion. It is pointed out
that the largest density of Brønsted acid sites in the graph
corresponds to pure niobia.
4. Discussion
4.1. Isopropanol dehydration
It is generally admitted that dehydration of alcohols on
amphoteric oxides such as alumina, to produce an alkene and
water, follows a bimolecular E2 type elimination mechanism,
involving acid–base pairs of sites [28–31]. Scheme 2
incorporates the main ideas proposed in the literature, where
the basic sites (BÀ
) may be hydroxyl or oxide anions. In the
present case, the main inhibiting effect of niobia on the i-
propanol dehydration activity of alumina seems to be mainly
related to the strong decrease in surface basic site density, as
revealed by the CO2 chemisorption measurements.
Selectivity for ether formation initially increases upon
niobia addition. However, total catalytic activity decreases by a
factor of nearly ten in moving from alumina to the NA-14
catalyst, whereas selectivity to di-isopropylether increases by a
factor of only ca. two in the same direction, so actually
dehydration to both ether and olefin are strongly inhibited by
addition of niobia. It has been proposed that ether formation
involves a SN2 type nucleophilic attack by an adsorbed
alkoxide species, such as the one shown in Scheme 2, on an
alcohol molecule coordinated to surface hydroxyl groups by
hydrogen bonding [31]. If the same alkoxide species is
involved, it is not surprising that both olefin and ether formation
are inhibited by niobia. The question remains as to why olefin
formation is more affected by niobia addition than ether
formation. The answer may be that, according to Scheme 1,
olefin formation requires the operation of two basic sites,
whereas ether production requires one basic site for alkoxide
formation and one more acidic hydroxyl group for bonding of
the alcohol molecule.
On pure niobia, only propene is observed as a reaction
product. According to the results (cf. Fig. 7), niobium oxide
does not irreversibly chemisorb carbon dioxide at or near room
temperature, so the mechanism of alcohol dehydration cannot
be the same as on alumina. Since Brønsted acid sites were
observed on the surface of our niobia, it is possible that an olefin
selective E1 type elimination mechanism [28,29] operates on
the niobia surface, as shown in Scheme 3. The increasing
contribution of this mechanism as coverage of the alumina
surface by niobia increases may be responsible for the decrease
in selectivity for ether with high niobia loadings.
4.2. 1-Butene isomerization
Despite its apparent simplicity, 1-butene double-bond
isomerization is actually rather complex and may involve
several different mechanisms. On Bro¨nsted acid catalysts, it is
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218216
Scheme 1.
Fig. 9. Initial rate of cumene dealkylation at 728 K as a function of the density
of Brønsted acid sites (in arbitrary units/m2
). (*) Niobia/aluminas; (~)
niobium oxide.
Scheme 2.
Scheme 3.
7. generally proposed that the mechanism involves protonation of
the olefin to produce a secondary carbenium species, followed
by de-protonation to produce cis- and trans-2-butene in an
approximately 1:1 ratio [32].
On oxides with surface basicity, such as magnesia or zirconia,
an allyl carbanion is thought to be formed by dissociation of an
allylic C–H bond on an acid–base pair [32,33]. Re-protonation of
the methylene carbon than leads to 2-butene formation with a
large cis/trans ratio (as much as seven).
The alumina surface is considered to have insufficient
Brønsted acidity for olefin protonation at low temperatures and
insufficient surface basicity for a carbanion mechanism to
operate. Trombetta et al. [34] have recently proposed that a s-
allyl species essentially covalently bonded to aluminum is
produced upon dissociation of the allylic C–H bond on an acid–
base pair at room temperature. Transformation of a 1-buten-3-
yl species into a 2-buten-1-yl species would be responsible for
the double-bond migration, as shown in Scheme 4. This is
closer to the carbanion mechanism than to the carbenium one,
and this may be the reason for the relatively large cis-/trans-2-
butene ratio (1.8) obtained under our conditions with alumina.
A trend of decreasing activity with decreasing density of
alumina-associated Lewis acidic sites is observed in Fig. 8,
however the range of activity values is much smaller than in
isopropanol dehydration. This results from the increasing
coverage of alumina-associated Lewis sites, leading to a
decrease in activity, together with increasing contribution of
niobia-associated Brønsted acid sites.
With pure niobia, the cis-/trans-2-butene ratio is 1.3, which
suggests the predominance of the carbenium mechanism with
this catalyst, as also found in isopropanol dehydration. With the
other catalysts, the cis/trans ratio goes trough a maximum as a
function of niobia content. The decreasing branch at high niobia
contents (>1/2 monolayer) may be credited to the contribution
of the carbenium mechanism, but the increase at low niobia
contents is more difficult to understand. Certainly it cannot be
attributed to the contribution of a carbanion mechanism, since
niobia addition actually decreases surface basicity. Possibly the
niobium species modify the properties of neighboring
aluminum cations leading to the observed effect. Perhaps the
fact that Nb5+
has a higher cation electronegativity than Al3+
[35] leads to an increase in Lewis acidic strength of neighboring
aluminum sites. This effect may also be responsible for the non-
linear behavior observed in Fig. 8.
4.3. Cumene dealkylation
Activity for cumene dealkylation increases with niobia
loading on the alumina-supported catalysts, certainly due to the
concomitant increase in density of Brønsted acid sites. Fig. 9
shows that the activity per unit surface area increases with
increasing density of Brønsted acid sites in a non-linear fashion.
This may reflect changes in the strength of the Brønsted sites as
the niobia loading increases.
It should be noticed first that the density of Brønsted sites
also increases in a non-linear fashion with the niobia loading, as
show in Fig. 5. Up to a loading of 2 Nb atoms/nm2
, very few
Brønsted sites are created due to niobia addition. The UV
spectroscopy data shown in Figs. 1 and 2 strongly suggest that
up to this loading isolated niobia species which have no
Brønsted acidity predominate on the alumina surface. As the
niobia loading increases, polymeric niobia species are formed,
responsible for the decrease in the band gap energy of the
materials. Concomitantly, the density of Brønsted acid sites
increases. For niobia loadings near to or higher than the
theoretical monolayer, the rate of increase of Brønsted acidity
with niobia loading decreases, which may be attributed to
vertical growth of the niobia domains, forming tridimensional
niobia particles. This is confirmed by the UV spectroscopy
results that show that band gap energy values approach the one
corresponding to bulk niobium oxide. If the Brønsted acid sites
associated with the tridimensional niobia particles are stronger
than the ones in the polymeric niobia species, the activity
pattern for cumene dealkylation may be understood.
5. Conclusion
Results for the characterization of the alumina-supported
niobium catalysts are generally consistent with the ones
reported before in the literature, with respect to the effect of
niobium content on the nature of the supported niobia species,
surface basicity, degree of hydroxylation and presence of
Brønsted acidic sites. Deconvolution of the group of peaks in
the FTIR spectrum of adsorbed pyridine corresponding to the
8a vibration mode additionally showed that niobium-associated
weak Lewis acidic sites are created by niobium addition to the
alumina surface, at the same time that stronger alumina-
associated Lewis sites are covered.
Different acid–base reactions respond differently to niobium
addition. In isopropanol dehydration, the main factor responsible
for the decrease in catalytic activity is the decrease in
concentration of basic sites, as measured by CO2 chemisorption.
Activity for 1-butene isomerization also decreases with niobium
addition, but the main factor here seems to be the decrease in
concentration of alumina-associated Lewis acidic sites. Evi-
dence from cis-/trans-2-butene ratio indicates that niobium
addition modifies the properties of neighboring aluminum sites.
Brønsted acidic sites created by niobium addition are
responsible for the development of activity in the cumene
dealkylation reaction, but the sites associated with tridimen-
sional niobia species seem to be more effective in this reaction.
Acknowledgements
We thank CAPES for the scholarship granted to M.A.A.R.,
to FINEP and PETROBRAS for financial support to this project
trough the CTPETRO fund and to INT for the DRS spectra.
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218 217
Scheme 4.
8. References
[1] T. Iizuka, K. Ogasawara, K. Tanabe, Bull. Chem. Soc. Jpn. 56 (1983)
2927.
[2] M. Danot, J.C. Afonso, M. Breysse, T. Courieres, Catal. Today 10 (1991)
629.
[3] A. Florentino, P. Cartraud, P. Magnoux, M. Guisnet, Appl. Catal. A 89
(1992) 143.
[4] Z.-H. Chen, T. Iizuka, K. Tanabe, Chem. Lett. (1984) 1085.
[5] S. Okazaki, N. Wada, Catal. Today 16 (1993) 349.
[6] A.C.B. dos Santos, W.B. Kover, A.C. Faro Jr., Appl. Catal. A 153 (1997)
83.
[7] C. Guo, Z. Qian, Catal. Today 16 (1993) 379.
[8] I. Nowak, M. Ziolek, Chem. Rev. 99 (1999) 3606.
[9] M. Ziolek, Catal. Today 78 (2003) 28.
[10] J.-M. Jheng, I.E. Wachs, Catal. Today 8 (1990) 37.
[11] J.-M. Jheng, I.E. Wachs, J. Phys. Chem. 95 (1991) 7373.
[12] J.-M. Jheng, I.E. Wachs, J. Mol. Catal. 67 (1991) 369.
[13] M.A. Vuurman, I.E. Wachs, J. Phys. Chem. 96 (1992) 5008.
[14] J.-M. Jheng, I.E. Wachs, Catal. Today 16 (1993) 417.
[15] I.E. Wachs, Catal. Today 27 (1996) 437.
[16] L.J. Burcham, J. Datka, I.E. Wachs, J. Phys. Chem. B 103 (1999) 6015.
[17] I.E. Wachs, Y. Chena, J.-M. Jehng, L.E. Briand, T. Tanaka, Catal. Today 78
(2003) 13.
[18] K. Asakura, Y. Iwasawa, Chem. Lett. 57 (1986) 859.
[19] A.G. Shikalgar, S.H. Pawar, Philos. Mag. B 40 (1979) 139.
[20] X. Gao, I.E. Wachs, M.S. Wong, J.Y. Ying, J. Catal. 203 (2001) 18.
[21] H. Kno¨zinger, P. Ratnasamy, Catal. Rev.-Sci. Eng. 17 (1978) 31.
[22] C. Lahousse, A. Aboulayt, F. Mauge´, J. Bachelier, J.C. Lavalley, J. Mol.
Catal. 84 (1993) 283.
[23] A.B.M. Saad, V.A. Ivanov, J.C. Lavalley, P. Nortier, F. Luck, Appl. Catal.
A: Gen. 94 (1993) 71.
[24] E.P. Parry, J. Catal. 2 (1963) 371.
[25] C.E. Emeis, J. Catal. 141 (1993) 347.
[26] C. Mortera, G. Magnaca, Catal. Today 27 (1996) 497.
[27] A. Corma, B.W. Wojciechowski, Catal. Rev.-Sci. Eng. 24 (1982) 1.
[28] H. Kno¨zinger, A. Scheglila, J. Catal. 17 (1970) 252.
[29] A. Gervasini, G. Belussi, J. Fenyvesi, A. Auroux, J. Phys. Chem. 99 (1995)
5117.
[30] B. Shi, H.A. Dabbagh, B.H. Davis, J. Mol. Catal. A: Chem. 141 (1999)
257.
[31] B.H. Davis, B. Shi, J. Catal. 157 (1995) 359.
[32] H. Hattori, Appl. Catal. A: Gen. 222 (2001) 247.
[33] Y. Nakano, T. Iizuka, H. Hattori, K. Tanabe, J. Catal. 57 (1979) 1.
[34] M. Trombetta, G. Busca, S.A. Rossini, V. Piccoli, U. Cornaro, J. Catal. 168
(1997) 334.
[35] C.L.T. da Silva, V.L.L. Camorim, J.L. Zotin, M.L.R.D. Pereira, A.C. Faro
Jr., Catal. Today 57 (2000) 209.
M.A. Abdel-Rehim et al. / Applied Catalysis A: General 305 (2006) 211–218218