A discussion of a mechanism for isomerization of n-butane on sulfated zirconia

Article abstract of DOI:10.1016/S1381-1169(03)00123-7

The mechanism of n-butane isomerization, considering experimental and theoretical works, was studied. MM2 and density functional theory simple calculations were conducted to clarify the nature of the supported active species in sulfated zirconia. A penta-coordinated species, monomeric or even dimeric, would be responsible for forming the most active sites at reaction conditions of n-butane isomerization. No superacidity was present before contact with n-butane. The sites formed at reaction conditions allowed the n-butane dehdyrogenation and its isomerization. At low reaction temperatures, not more than 250°C, additon of hydrogen as carrier gas decreased the conversion comparing it with N₂ as carrier gas. At higher temperatures, it increased the activity.

Full text of DOI:10.1016/S1381-1169(03)00123-7

Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
A discussion of a mechanism for isomerization of
n-butane on sulfated zirconia
b
a
E. Garcıa , M.A. Volpe , M.L. Ferreiraa,∗, E. Ruedab
´
a
PLAPIQUI (UNS-CONICET), Camino La Carrindanga km 7, CC 717, 8000 Bah´ıa Blanca, Argentina
b
UNS, Avda Alem 1253, 8000 Bah´ıa Blanca, Argentina
Received 19 December 2002; received in revised form 19 December 2002; accepted 16 February 2003
Abstract
Although a huge amount of experimental work has been done, there is no clear picture about the superacidity in sulfated
zirconia (SZ). Some authors think that sulfated zirconia is not superacid at all. Moreover, the structure of active site in
n-butane isomerization is still now under controversy. This work tries to give some light about the mechanism of n-butane
isomerization, taking into account experimental and theoretical works from others and from our own. MM2 and DFT simple
calculations were performed in order to clarify the nature of the supported active species in sulfated zirconia. It is clear that
a penta-coordinated species, monomeric or even dimeric, would be responsible for forming the most active sites at reaction
conditions of n-butane isomerization. This idea is the key of this article. Different studies demonstrated that no superacidity
is present before contact with n-butane. The sites formed at reaction conditions allow the n-butane dehydrogenation and its
isomerization. A bimolecular mechanism of n-butane isomerization on penta-coordinated sites is proposed to occur. Several
steps are discussed.
© 2003 Elsevier Science B.V. All rights reserved.
Keywords: n-Butane isomerization mechanism; Sulfated zirconia; Bisulfate species
1. Introduction
NMR, XPS, HRTEM and others) were applied to study
SZ. These studies were performed on SZ before the re-
action, without alkane, or after the reaction. It is very
difficult to study in situ the title reaction. In the open
literature, theoretical studies involving a whole picture
of the mechanism in n-butane isomerization are not
found. The structures of the surface species in SZ in-
volved in the intermediaries formation of the n-butane
isomerization reaction has been not clarified until now
[10–14]. When a H+ is presented no source is indi-
cated. When a hydride is depicted, no sink is analyzed.
It was generally accepted until recent times that a
stable tetragonal phase is necessary to obtain activ-
ity. Later, workers indicated that drastic sulfation to
overcome the inertness of the surface makes catalysts
out of the monoclinic oxide [15]. The route of drastic
Despite the huge amount of experimental work done
on sulfated zirconia (SZ), no clear picture about the
structure of active species at SZ surface for alkane
isomerization is presently available [1–3]. Moreover,
there is growing evidence that superacidity might not
be the source of the catalytic activity [3–9]. Redox and
acidic–basic properties could be alternative sources of
the activity of SZ in catalytic isomerization of hy-
drocarbons (specially n-butane). A variety of spectro-
scopic and surface sensitive techniques (infrared (IR),
∗
Corresponding author. Tel.: +54-291-4861700;
fax: +54-291-4861600.
E-mail address: mlferreira@plapiqui.edu.ar (M.L. Ferreira).
1381-1169/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S1381-1169(03)00123-7

264
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
sulfation is not widely used. Sulfation is needed to
improve acidic properties of ZrO2. Moreover, no
considerable activity was found with ZrO2 alone for
n-butane isomerization. Two competitive mechanisms
were generally proposed in the skeletal isomeriza-
tion of n-butane, mono- or bimolecular in nature.
The monomolecular mechanism involves the ener-
getically unfavorable formation of a primary carbe-
nium ion intermediate. In the bimolecular process, an
octyl cation (C8+) is obtained by alkylation of the
sec-butyl cation by butene before ␤-scission occurs.
In case of the monomolecular mechanism, isobutane
is the only product. With the bimolecular mechanism,
sub-products (pentanes and propane) arise [16–19].
The time dependence for the reaction at low tempera-
tures using SZ involves an induction period (where the
only product is isobutane) and later a second period
where subproducts evolve [20]. The induction period
is shorter at high temperatures. Davis et al. [11] re-
ported that in catalyst activity, the pre-treatment tem-
perature after calcination procedure at 600 ◦C is cru-
cial. The results showed that an optimum in the conver-
sion of n-butane was obtained when the catalyst was
heated between 300 and 400 ◦C prior to the reaction.
There are several points to take into account: source
of sulfate group (H2SO4 or (NH4)2SO4), the sulfa-
tion method, S final content, calcination procedure,
pre-treatment temperature and the presence of H2.
There is a lack of proposals of the steps and changes
that must take place at the SZ surface to explain the
observed experimental behavior in the open literature.
Being H2SO4 or (NH4)2SO4 the S source, bisulfate
is the main species present at the preparation aqueous
media. We think that the supported S is present in dif-
ferent ways on SZ surface. Several active supported
species have been proposed: dioxosulfate, monoxo-
sulfate with tricoordination with surface and bisulfate
group. These species can not explain some IR spec-
tra and NMR data. Knözinger and coworkers [21,22]
and White et al. [23] proposed monoxosulfate groups
with multiple coordination sites with SZ surface as the
main site. The presence of an OH bonding to S and to
Zr in SZ is a very important issue. This site gives rise
to Brönsted acidity. However, dimeric and polymeric
species claimed to play a role [24]. Penta-coordinated
S was proposed to be of paramount importance [23].
Lewis acidic–Lewis basic pair sites must be also con-
sidered in the analysis of the zirconia surface.
We performed a molecular mechanics and a simple
DFT calculation on a model of sulfated zirconia. The
aim was to clarify the structures of the SZ-most ac-
tive species in n-butane isomerization. We propose a
model that takes into account experimental and the-
oretical reported data. We also present experimental
data of our own results using different routes of prepa-
ration of SZ with different S content. Although other
workers have published results for the title reaction
with similar catalysts, in this work we carried out a
comparative study about catalytic activity from sam-
ples with a difference in the source of S and in the
preparation methods. Moreover, this article is not an
exhaustive theoretical or experimental study. This ar-
ticle presents a discussion of a possible mechanism
with different aspects analyzed with different theoreti-
cal and experimental tools. The objective of this work
is to present an alternative mechanism based on the
in situ generation of the active sites for n-butane iso-
merization. Hydrogen effects, pyridine adsorption and
experimental data are analyzed to correlate trends in
its behavior. Some key aspects related to butene gen-
eration, sec-butyl cation formation, isomerization re-
action at the proposed site and isobutane desorption
reaction are studied with simple DFT. MM2 calcula-
tion on pyridine adsorption is also presented. The aim
in this sense is to check qualitative trends in nature.
1.1. The methods
1.1.1. MM2
The used software was Chem 3D Ultra, version 5.0,
1999 (Cambridge Soft). Calculation using MM2 was
performed on clusters of different sizes.
1.1.2. The surface models
The models considered for sulfated zirconia were
monomeric and dimeric. The species I–IV are pre-
sented in Fig. 1(a). The dimeric species are claimed
to be derivated from species IV, because water releas-
ing, and they can be terminal or internal. Published
FT–IR data [23,25] showed that no dioxosulfate
groups are found at the prepared catalysts. From
all these species, species IV and dimeric-terminal
species from IV would show Brönsted acidity. From
here, these species are IV-I and IV-II. Depending on
the treatment temperature, the associated sites to site
IV-I—the closest Zr site—from zirconia, changes:

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
265
Fig. 1. (a) Proposed SO4 species in SZ; (b) formation of LA–LB pair.
from Brönsted acid (BA) Zr–OH to Lewis acid (LB)
Zr–O or Lewis acid Zr (LA) (see Fig. 1(a)). The pres-
ence of a LA site implies that a LB site was formed
close to it. Since the calcination temperature generally
used was 650 ◦C [11] the associated site is considered
to be a LA. This Zr (LA) is claimed to be highly
reactive with water even at room temperature and a
pre-activation procedure would be necessary to elim-
inate water before n-butane contacting. This LA has
surely a LB site associated at zirconia surface, pro-
vided by dehydration process (Fig. 1(b)). Dissociative
adsorption of water occurs over these LA–LB pair
sites. Fig. 2(a) and (b) show the site IV formation in
reaction conditions, involving n-butane and n-butene
generation. Clearly, Brönsted sites arise from zirconia
(Zr–OH–Zr) and from sulfated zirconia (S–OH–Zr).
Higher pre-activation temperatures than 400 ◦C in
He flow decrease the Brönsted acidity from S–OH–Zr.
This Brönsted acidic site would almost disappear
because of pretreatment at 600 ◦C, and inactivation
arises. Davis et al. [11] published that a very resis-
tant Brönsted acidic site remains at surface up to this
temperature. Temperatures higher than this destroy
activity.
The two main exposed planes of tetragonal zir-
conia surface are (1 0 1) and (0 0 1) planes. They
have Zr with coordination 6 or 7 to O, whereas O
has coordination 2 or 3. Using MM2, a cluster of
8–10 Zr and a relaxed structure was considered. A
structure with four Zr was also considered for com-
parison of dimeric and monomeric sites I and IV
species.
1.1.3. DFT
The reaction of formation sites was investigated by
means of DFT studies, using Gaussian’98 in a 512 Mb

266
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
RAM computer [7,26]. The calculations were done
with Becke’s exchange functional gradient-corrected
correlation function (BLYP) with sto-3g basis as an
initial approach. This basis set has demonstrated to
be useful to represent the formation reaction of the
active site structure [27]. The main structures con-
sidered were site I in Fig. 1(a) and that presented in
Fig. 2(b).
Fig. 2. (a) Formation of active site IV in reaction conditions, involving butane; (b) scheme of Brönsted “superacid” formation.

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
267
the bands energy in oxides is poorly represented by
Gaussian. Moreover, the exact change in the surface
electric field is very difficult to represent. So, there
are few successful trials to study ZrO2 using MM/QM
coupled methods because of these problems. In this
sense, the geometry of zirconia disposition at surface
seems to be of huge importance.
2. Experimental
The catalysts of sulfated zirconia were prepared us-
ing the method of impregnation and immersion. The
support, Zr(OH)4, was obtained by precipitation of a
aqueous solution of ZrOCl2 with NH4OH until pH 10
at ambient temperature. The obtained solid phase was
washed with bidistilled H2O until the complete elim-
ination of chloride ions.
Fig. 2. (Continued ).
Several stages were considered in the calculation:
(a) The formation of the active site, with loss of wa-
ter. In this stage, the stage of calcination or pre-
vious pretreatment to the reaction was simulated,
since it is known that the water is a poison for
these catalysts at high concentration on the surface
[8].
(b) The formation of the sec-butyl carbocation.
(c) The isomerization to isobutane.
(d) The isobutane desorption.
The sulfation of the samples was obtained by im-
pregnation or immersion with (NH4)2SO4 or H2SO4
at different initial concentration (see Table 1).The
immersion method requires 15 ml of solution per g
of catalyst during 45 min of contact. Other samples
were obtained by immersion method with different
volumes of H2SO4 0.5N followed by drying on a hot-
plate.
The samples were dried-off at 110 ◦C during 24 h
Some alternatives related to the bifuncional site are
also considered, presenting results with a model of
disulfate species. The main point here is that the re-
sults are used in qualitative trends, not quantitative.
With systems of few d electrons (involving Ti and Zr),
and later calcined 4 h to 600 ◦C. They were char-
2−
acterized by XRD and content of SO4 by LECO
method. The BET areas of several of them were in
the range from 90 up to 110 m2/g. Studies of pyridine
Table 1
Characterization of catalysts
2
Sample
Method
SO4 (g%)
FT–IR
XRDa (T%)
Conversionb (%)
Maximum
After 2 h
1
2
3
4
5
6
No sulfated
–
4.6
3.0
3.82
12.8
4.11
–
0
61
38
45
100
39
–
6.4
4
1.7
7.0
0.9
–
Immersion with (NH4)2SO4 6% nom.
Impregnation with H2SO4 2N
Immersion with H2SO4 6% nom.
Immersion with H2SO4 20% nom.
Impregnation with (NH4)2SO4 6% nom.
L ꢀ B pyridone
L ≤ B pyridonec
L ≤ B pyridoned
L = B-vw pyridonee
L < B pyridone
2.4
1.7
0.9
4.3
0.4
a Tetragonal:monoclinic ratio calculated as the peak height ratio between 2θ = 30.1◦ (tetragonal) and 2θ = 28.7◦ (monoclinic)—T%
calculated from the peak height ratio.
b Flow rate: 20 ml/min.
c Well-defined bands.
d Ill-defined bands.
e Pyridone with the highest absorbance (vw: very weak).

268
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Table 2
adsorption were performed to evaluate the acidity of
sulfated zirconia.
Results for steric energy in a cluster of 8–10 Zr
Species
Steric energy
(kcal/mol)
2.1. FT–IR studies of pyridine adsorption
Non-terminal bisulfate
Terminal bisulfate
276.7
306.2
313.9
363.9
428.7
452.6
462.4
560.3
267
Studies of adsorption of pyridine were done for all
the tested samples. Using low amounts of solid, not
more than 200 mg, a pretreatment of 2 h in N2 (near
10 ml/min) at 150 ◦C was done. After this, the sam-
ples were contacted with flowing inert for 1 h at the
same temperature in a pyridine reservoir. The sam-
ples remained 20 min in flowing N2 to achieve the
room temperature. At room temperature, we purged
these samples by another 20 min. The data are in-
cluded in Table 1. Using NaCl windows, we perform
the FT–IR spectra in a Nicolet 520, with a resolu-
tion of 4 cm−1 and 10 scans. We obtained signals of
Lewis and Brönsted sites and pyridone complexes on
these surfaces (at 1450, 1540 and 1635–1640 cm−1).
The spectrum of the bare windows was subtracted.
Because of the known criticism to this method in
the sense of over-consideration of LA sites [11],
only the relative presence of each kind of site is
included.
Sulfate at O bridge (reaction of 2OH)-1
Polysulfate (S3O9H)-BA
Sulfate by adsorption at LA–BA pair sites
Sulfate at O bridge (reaction of 2OH)-2
Sulfate at O of a Zr (both bridged)
Polysulfate (S3O8)
Penta-coordinated S-Brönsted acid
3. Results and discussion
3.1. Methods
3.1.1. MM2
The results for steric energies are presented in
Table 2 for different species tested at zirconia cluster
of 8–10 Zr. A minor cluster was also tested to check
penta versus tetrahedral coordination of S. The co-
ordination of Zr was completed with OH. It is clear
from Table 2 that penta-coordinated species like IV
are the most probable in steric terms. We concluded
that the most sterically probable species would be
named IV, monomeric and dimeric. The other species
have all problems: species I are not in agreement
with IR studies. Species II presents a +5 oxidation
state. Species III are claimed to be rather minor in
concentration at surface. Polymeric species are con-
sidered to be at high concentration with S content
higher than 2%. Species IV has S in +6 oxidation
state and it takes into account several reported data
about characterization. It is clear that Zr Lewis acidity
is improved because of the withdrawing of electrons
due to S=O presence in species IV (see Fig. 1). Brön-
sted acidity of site IV would be similar to zirconia
in case of Zr–OH–Zr because the electronegativities
of Zr and S are similar (2.3 versus 2.6). FT–IR data
of pyridine adsorption showed that Brönsted acidity
has almost no changes by sulfation, but Lewis acid-
ity increases [2,28–30]. However, pyridine results
must be considered with care. Davis et al. [31] found
equal amounts of Lewis and Brönsted acid sites over
sulfated zirconia after heating at 400 ◦C. Besides,
the Brönsted–Lewis acids site ratio was essentially
2.2. Reaction of n-butane isomerization
The catalytic test was carried out in a flow glass
reactor operating at atmospheric pressure. The reac-
tive mixture was 10% of butane in N2. Approximately
100 mg of sample were pretreated in the reactor at
350 ◦C during 2 h under chromatographic airflow in
order to activate the samples. Subsequently, the tem-
perature was lowered to 200 or 100 ◦C and the reac-
tive mixture was allowed to flow through the reactor
at 20 cm3/min. The reactor was connected to a gas
chromatograph (Shimadzu, GC-14B) provided with a
TCD detector and a Chrom PAW 40/60 mesh column,
working at 80 ◦C. The conversion of butane (X%), ex-
pressed as moles of butane converted percent, was
taken as a measurement of the activity of the samples.
The dependence of X%, at constant temperature and
space velocity, on time, was measured. We have also
calculated the selectivity to different products (S%).
The selectivity to isobutane was calculated as:
moles of isobutane
SI =
× 100·
total moles

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
269
zero after the sample was reactivated at 600 ◦C [31].
Previous studies showed that SZ was active for hy-
drocarbon conversions after activation at 650 ◦C [32].
These results are in agreement with the probability
that the true catalytic sites are generated during the
induction stage of the hydrocarbon conversion [33].
Several environments were considered, in agree-
ment with the data presented above. The results of
Table 2 can be explained taking into account the ad-
ditional stabilization because of the SOHZr bond for-
mation, resembling at surface tricoordination of O at
bulk. The conversion of this site IV to II can occur
by calcination at high temperatures because of water
evolution. A similar process is depicted in Fig. 1(b).
The S in site IV is penta-coordinated, as in SOF4, and
S=O bond remains. The second S=O is proposed to
react with a LA site of zirconia. The O− of bisulfate
reacts with a LA Zr. The remaining OH group reacts
with a OH group of zirconia with water loss and gen-
eration of basic O. The fifth coordination of S is to a
BA site of zirconia (see Table 3).
3.2. The mechanism
Any mechanism must explain the induction, the last
period, the evolution of methane [34] and H2 [13] at
Table 3
Results for steric energy (kcal/mol) in a cluster of four Zr no H depicted for clarity
Penta-coordinated S
302 (Zr 1 tetrahedral),
301.4 (Zr 1 trigonal plane)
Dioxosulfate
523.8
334.9
796.4
Dimer penta-coordinated S
Dimer tetrahedral S

270
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
the initial stage. The mechanisms are proposed con-
sidering monolayer coverage of sulfated sites. Some
of these sites are sites IV. This surface was selected
taking into account the optimum ratio of LA:BA of
near 2:1 found by FT–IR characterization of calcined
SZ [35].
Sulfated zirconia acts as an isomerization catalyst
at the induction period and as a cracking catalyst after
that, in the absence of H2 in the feed. This can be
related to the hydride concentration at surface and the
average lifetime of the carbonium ion. When it acts
as an isomerization catalyst, hydride concentration is
high and carbonium lifetime is short. As a cracking
catalyst, hydride concentration is low and carbonium
average lifetime is high [36].
metanoate anion versus metanoic acid (see Fig. (2b),
highlighted species). The resulting anion by H+ loss
would be a very weak base (see Fig. 2(a)). This species
allows to the negative charge to follow the positive
charge. The creation of a large potential energy for
charge delocalization is avoided. This is especially
true when C8+ is considered [34,39]. Recent published
work of Yamaguchi [40] based on isotopic analysis
of alkane isomerization, pointed out that bimolecu-
lar mechanism is preferred on SZ for butane even at
low temperatures, whereas it is not for n-pentane and
n-hexane.
The dehydrogenation step leaves butene as an ad-
ditional product, besides H2. These butenes can be
adsorbed on available acidic sites of zirconia. They
can also react with available H. Sites with different
Brönsted acidity and similar structure, strong and
weak, to take place at SZ surface. On the strong sites,
carbonium ions are formed upon butane adsorption.
On the weak sites, desorption of isobutane leaves
free the Zr(LA) site and the S–O–Zr bond with-
out H (see Figs. 2(a) and 3). Butene reaction with
3.2.1. The mechanism for the induction period
The mechanism for the induction period has some-
times reported to be monomolecular [19,20]. We pro-
pose that it could take place at monomeric sites IV-I
and IV-II. Upon butane adsorption, partial reduction
of one S6+ to S4+ takes place when sites IV-II are
analyzed. The efficiency for reconversion to dimeric
sulfate group is low. Deactivation arises because no
regeneration of S–O–S bond in the case of sites IV-II.
Conversion of n-butane shows a maximum with the
time on the stream because the activation of the surface
is necessary. Fresh species IV-II are active but they
are present at low concentration, especially with low
S content. The increase of temperature to 180–200 ◦C
can avoid the low initial activity step at lower temper-
atures. In the induction period, the sites IV-I dehydro-
genate n-butane at Zr(LA)–O(LB) pair sites, evolving
H2. Non-sulfated ZrO2 is able to dissociate H2 het-
erolytically [37]. H2 can be dissociated by SZ, giving
rise to H− associated to acidic Zr and H+ coordinated
to basic O from ZrO2. The site IV-I was developed
taking into account the results of Lunsford et al. [38].
At this active site IV-I, H2 evolution has two sources.
One of them is direct dehydrogenation of n-butane in
LA–LB pair sites (Hs’ from different carbons). An-
other one is the carbonium ion (Hs’ from the same car-
bon, penta-coordinated). Superacidity could be gen-
erated in reaction conditions, when butane is present
(see Fig. 2(a)). An increased stability of the surface
sulfate anion by charge delocalization could produce
species with higher Brönsted acidity than the initial
one, when no butane is present. This also occurs with
+
C4H9 adsorbed on S–O−–Zr leads to bimolecular
mechanism. The carbenium–carbonium mechanism
(C4H11+–C4H9+) takes place on S–OH–Zr. An al-
coxy bond could be formed on Zr–O–Zr nearest to
S–OH–Zr. The intermediary in isobutyl formation is
methylcyclopropyl cation. Methyl group can react
with hydride, generating methane. An allyl group re-
mains coordinated by Zr(LA) close to S (see Fig. 4).
Methane is not evolved from the protonated paraf-
fin. Initially, at the inductio+n step, no propane is
found with isobutane (C4H11 evolving CH4 leaves
C3H7+). Hong et al. [34] suggested that protonation
by the strongest acid sites of sulfated zirconia is not
the primary mode for initiation of butane isomer-
ization. Basicity of O in SOZr bond is similar to O
in Zr–O–Zr. Superacidity is lost because of hydride
transfer to isobutyl cation and isobutane release. Re-
generation or inactivation of S–OH–Zr is possible.
Deactivation can be reversible because of H2 disso-
ciation reaction (see Fig. 4). Irreversible deactivation
could take place, by reduction of sulfur (or even Zr)
to lower oxidation states. Further dehydrogenation of
adsorbed cationic species deactivates the site IV-I.
H+ at surface can avoid this process. Only in the
presence of a noble metal like Pt, the reactive H
concentration is enough to hydrogenate these species

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
271
Fig. 3. Monomolecular mechanism for n-butane isomerization on “strong” acidic sites IV.

272
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Fig. 4. Monomolecular mechanism for n-butane isomerization on “weak” acidic sites IV.

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
273
and avoid deactivation [31]. Hydrogen decreases the
butenes concentration and, therefore, this effect is
very dependent on its concentration.
Another termination reaction is the hydride abstrac-
tion from C4H10 to carbonium ions. This reaction is
energetically disfavored. No clear role of LA sites
arises in this case. Gates et al. [36] published that cat-
alytic activity for carbonium ion reaction is related to
LA site concentration.
H2 and n-butene formation would be possible on
dimeric species—site IV-II (see Fig. 5). The weak
point of this cycle is the regeneration of S–O–S bond.
These sites could suffer irreversible deactivation al-
though they could be the most active in monomolec-
ular isomerization, but the most easily deactivated.
S4+On forms a stronger base than S6+On because
S6+OH is a stronger acid than S4+OH in the case
shown in Fig. 5. Therefore, H is placed on S4+On and
inactive species are created.
+
In the monomolecular mechanism C4H9 has al-
ways a Zr–H close to it. Alcoxy groups adsorbed on
Zr–O–Zr provide steric hindrance in its surrounding.
+
It seems that the presence of Zr–H close to C4H9
is the+main point. In bimolecular mechanism, never a
C4H9 has a Zr–H close to it, because it reacts with
the octyl-cation to evolve products (see Fig. 6). There
is enough room for C4H8 coordination. A high cover-
age of the surface is needed and this is related to the
n-butane pressure.
The induction mechanism can only be seen with
the most active preparations, using low temperatures.
Otherwise, at temperatures of 200 ◦C or higher, it takes
Fig. 5. Monomolecular mechanism for n-butane isomerization on dimeric sites IV.

274
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Fig. 6. Bimolecular mechanism for n-butane isomerization on “weak” acidic sites.
place at the first 1–2 min of the reaction. Quickly, the
bimolecular mechanism begins to function.
provide the necessary H+ and H− to react with the
isobutyl cation. Final release of isobutane is achieved.
H+ from n-butane dehydrogenation can hydrogenate
allylic species at surface, giving propane (see Fig. 5).
3.2.2. Mechanism for the final period
+
This period presents evolution of by-products,
mainly pentanes and propane. Therefore, a bimolecu-
lar mechanism is strongly supported by experimental
evidence for n-butane isomerization. The bimolecular
mechanism could take place on sites IV-I. In this case,
n-butene concentration on the surface is high and the
Propane and pentanes are produced from C8 crack-
ing (see Fig. 6). Residual activity (activity at 2 h;
see Table 4) could be assigned to sites resistant to
reduction (polymeric). Besides, active sites after de-
activation should be resistant to partial reduction of S
and coke deposition.
The effect of butene (without H2) is easy to under-
stand: it triggers the n-butane conversion through a
bimolecular mechanism. The increase of butene con-
centration produces the catalyst deactivation by allyl
formation [41].
The switch from mono- to bi-molecular mechanism
can be related to the structure of the active site pro-
posed 2LA:1BA:1LB. If an alcoxy species is bonded
to a LB site, hydride will be+placed at Zr(LA) and a
true ion-pair is formed. C4H9 is now a proton donor
+
C8 formation would be favored. The initial site IV-I
has H- (or a Zr–H bond) and H+ near the monomeric
site (as OH). sec-Butyl cation is formed by reaction
for site IV-I with butene. Addition of a second butene
molecule produces the formation of octyl-cation. This
is followed by the isomerization and ␤-C scission
of C8+. Reaction with another molecule of n-butane
regenerates the active site SO–octyl-cation–Zr and
Zr–H. Only one butene molecule triggers a chain
reaction of n-butane isomerization. LB and LA sites

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
275
Table 4
Activity and turnover rate (TOR) for catalysts presented in this and other works
Sample Catalysts composition
(sulfate content (%))
Temperature (◦C)
Rate × 107 mol/(g catalyst s)
TOR (s−1) × 103a
Maximum FP
Carrierb
(reference)
Pretreatment Reaction Maximum FPc
2
3
4
5
6
–
–
–
–
–
–
(NH4)2SO4/ZrO2 (4.6)
H2SO4/ZrO2 (3)
H2SO4/ZrO2 (3.8)
H2SO4/ZrO2 (12.8)
(NH4)2SO4/ZrO2 (4.11)
(NH4)2S2O8/ZrO2 (4.7)
350
350
350
350
350
450
200
200
200
200
200
250
250
50
9.5
7.2
2.5
10.7
1.3
–
21.4
4.8
0.5
2.5
2.0
3.3
2.7
1.7
5.8
0.6
2.2
–
–
0.2
–
1.99
2.30
0.63
0.80
0.31
–
3.6
1.1
0.14
–
0.67
0.69
0.87
0.42
0.44
0.15
0.45
–
–
0.06
–
N2 (this work)
N2 (this work)
N2 (this work)
N2 (this work)
N2 (this work)
Air [44]
N2 [16]
N2 [12]
N2 [30]
He [43]
H2SO4/ZrO2 0.08N (5.75) 650
H2SO4/ZrO2 (4.2)
H2SO4/ZrO2 (3.33)
H2SO4/ZrO2
500
450
–
180
–
450
(NH4)2SO4/ZrO2 (2.9)
550
0.6
0.2
N2 [45]
a The mol of butane converted to per mol of sulfate.
b Recalculated values for our units and considering space velocity.
c Final period, after 2 h reaction.
(acid) because of the anionic character of O. This re-
action needs the coverage of the surface and activation
of the alcoxy bond. It was found that dehydrogena-
tion of butane takes place after an induction period
[41]. The ZrO2 is not active. Required acidity to trig-
ger the reaction can be related to the LA site near the
S–OH–Zr bond in sulfated zirconia. The LA sites of
ZrO2 are not enough acidic to abstract a hydride from
n-butane.
The formation of sec-butyl cation (with Zr–H for-
mation at the nearest Zr(LA)) was modeled on a cluster
of eight Zr. The steric energy was near 183 kcal/mol,
considering the formation of a carbocation and a neg-
ative O at surface. The bimolecular mechanism was
checked with the +same site, by reaction with a butene
molecule. The C8 formation implies 190.1 kcal/mol.
A difference of only 7.1 kcal/mol in steric energy with
the monomolecular mechanism was found.
cursors of the active sites. Another point to consider
is the relaxation of surface at this level.
Only 10% of the active sites in SZ are the most
active [34,42]. In this sense, active site is a minor
structure. With this idea in mind, we included an active
site with SOHZr bond. Clearfield et al. [10] suggested
that some bisulfate structure is stable until 600 ◦C.
Fig. 7 presents the results for the IV and I sites for-
mation. The value for dioxosulfate formation is similar
to that reported by Kanougi et al. [27], −55.4 versus
−59.2 kcal/mol. In the theoretical studies of this kind,
it is important to consider that H2SO4 and HSO4− con-
centrations change with solution concentration, pH,
sulfate agent used (H2SO4 or (NH4)SO4) and the pres-
ence of water in the neighborhood. Water is lost at the
calcination step and surface rearrangement arises. In
fact, no water is lost from site IV because of geometric
considerations. From site I, water is weakly adsorbed
and easily released. Taking into account these points,
we included a DFT study using Gaussian’98W in a
small model of the structure, resembling the claimed
penta-coordinated S (see Fig. 8(a) and (b)). We did not
3.2.3. DFT
Recent data were published about DFT studies on
SZ. Kanougi et al. [27] presented calculation on struc-
tural properties as well as electronic states of the sur-
face. This article was focused on H2SO4 reaction with
zirconia surface to afford water and SO3 in agree-
ment with Haase and Sauer [5]. These authors did not
take into account the experimental step of calcination,
where water is lost and subsequent reordering must be
considered. In this sense, refs. [5,27] present only pre-
+
include the C8 formation in this work because the
preliminary results indicate that our simple model can
not afford the complete interaction of the C8+ with the
sulfated zirconia surface. Several sites are involved in
the C8+ stabilization at surface. The main results of the
calculation were: endothermic character of butene for-
mation at first (+77.8 kcal/mol), exothermic reaction

276
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Fig. 7. DFT results for sites I and IV formation.
for sec-butyl formation (−28.1 kcal/mol), endothermic
isomerization to tert-butyl cation (32.4 kcal/mol) and
finally exothermic character for isobutane desorption
(−86.9 kcal/mol; see Fig. 8(b)). The initial structure is
not regenerated. The leaving site is an extremely weak
base. Zr is a strong Lewis acid and can extract hydro-
gen from butane. The sec-butyl cation formation from
this site is endothermic in 54.5 kcal/mol.
A simple calculation of sec-butyl cation forma-
tion on sites IV-II suggested that this reaction is
possible and endothermic by 115 kcal/mol (with par-
allel reduction of one S6+ to S4+). Isomerization to
tert-butyl cation is possible, with a change in energy
of −1.2 kcal/mol. The cycle involved in the regen-
eration of these sites is the weaker aspect of the
mechanism, because S4+ should be re-oxidized.
onal phase in higher amount (61 versus 39%). Also,
the amount of Lewis acidic sites in case of immersion
procedure is higher. In case of using H2SO4 (Samples
3 and 4) the structure in terms of FT–IR and XRD is
similar. The tetragonal phase is present at higher con-
centration in Sample 4. Initial conversion is clearly
different. This result is in line with those published by
Stichert et al. [45], where the authors proposed that
not only the acidity of sulfated zirconia determines
the catalytic performance, the surface structure is also
important. The bimolecular mechanism needs a favor-
able arrangement of surface groups. Tetragonal sys-
tem or drastically sulfated monoclinic structure in SZ
is especially suitable for this.
The presence of dimeric or polymeric sulfate
species-sites IV-II, with lower activity, could decrease
the turnover rate (TOR; see Table 4). Immersion pro-
cedure using high amounts of sulfate increases the
conversion to isobutane. The TOR in these conditions
is lower than the TOR of samples with lower amounts
of sulfate (comparing Table 4, Samples 3 and 5). The
TOR in Sample 4 is lower than Samples 2 and 3. In
Sample 4, the only crystalline phase is tetragonal. The
Samples 2 and 5 have the maximum rate while Sam-
ples 4 and 6 the lowest. The TOR follows the trend
3.2.4. The n-butane isomerization
Tables 1 and 4 show the catalyst characterization
and n-butane reaction results, respectively. Selectivity
to isobutane is always near 95%. Reactions at 100 ◦C
gave very low activity and results are not shown. It is
clear from Table 1 that the immersion method is better
than impregnation for (NH4)2SO4. The main differ-
ence between procedures is the presence of the tetrag-

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
277
found for maximum TOR. From comparison of our
results with those reported in the literature, we ob-
served that persulfate as precursor does not improve
activity at 2 h reaction.
signed to pyridone [21]. The pyridone band is not
present in bare ZrO2. The site IV-I gives rise to this
band because of steric hindrance for pyridinium for-
mation [11]. The other bands can be assigned to LA
sites with different environments (see Fig. 9(a)). On
the other hand, Morterra and Cerrato [46] showed that,
in the case of SZ the use of pyridine adsorption as a
titration method to obtain Lewis/Brönsted (L/B) acidic
sites ratio could lead to unreliable results [47]. In our
cases, we think that the Lewis acidic sites and pyridone
3.2.5. FT–IR data for pyridine adsorption
SZ presents the bands of adsorbed pyridine at 1540
(BA), 1445 (LA) and another band at 1640 cm−1
.
These bands arise when SZ is heated at 650 ◦C and
evacuated at 400–600 ◦C. Band at 1640 cm−1 is as-
Fig. 8. (a) DFT results for coordination of butene and isomerization to tert-butyl cation; (b) DFT results for desorption of isobutane and
coordination of a new butane.

278
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Fig. 8. (Continued ).
presence before reaction are important data to corre-
late with the activity of the sample. The existence of
sterically hindered hydrogen bonded to a S–O–Zr bond
is crucial. The best catalyst is found when Lewis acid
sites and pyridone are present together (see Table 1)
[48–50].
tion proceeds by monomolecular mechanism on Lewis
acid sites. On the other hand, surface alkenes have
claimed to be produced on the Brönsted acid sites. We
do not agree with this idea. Monomolecular mecha-
nism needs LA and BA sites. Bimolecular mechanism
needs BA sites [53].
Fig. 9(b) and (c) show the calculation results using
MM2 for pyridine adsorption on LA (35 kcal/mol) on
BA (152 kcal/mol) and on LA–BA pair (92 kcal/mol).
Lewis acid sites and LA–BA pairs are the preferred
sites for adsorption of pyridine when no butane is
present. Remember that the Brönsted acid in SZ at
this stage is not superacidic. It has almost the same
acidity as Brönsted acid in ZrO2. The Lewis acid site
(Zr) is stronger that the Brönsted one (S–OH–Zr) in
the site IV-I. A competition between both takes place
for pyridine adsorption. The pyridine is adsorbed first
on the LA site. After adsorption on the Lewis site,
when pyridone is found the C–O bond is formed from
reaction of C-2 of pyridine with OH from S–OH–Zr.
3.2.7. Effect of carrier–effect of H2
It is clear that adding H2 assures enough H− and
H+ concentration at surface, because of heterolytic
cleavage at LA–LB sites (strong, near monomeric ac-
tive sites, weaker, at zirconia LA–LB pairs). The H2
at the monomeric site cycle avoids allyl species accu-
mulation. It can also regenerate in situ superacidity.
At the dimeric site, it increases the concentration of
protonated species. The possibility of reaction of sur-
face H with migrating n-butenes at surface is higher
in the presence than in the absence of H2. All the de-
hydrogenation reactions are disfavored (see H2 effect
in Figs. 3–7). However, deactivation by reduction to
S4+ or reduction of Zr4+ to Zr3+ can not be avoided.
At the step of Zr–H formation, Zr4+ irreversible re-
duction to Zr3+ could be a problem. In the same way,
H2 regenerates the active site by a non-redox reaction
with dioxo-site I (see Fig. 5).
3.2.6. Lewis and Brönsted acid sites
Published data stressed the importance of the activa-
tion temperature before the reaction [20,51,52]. Arata
[51] and Matsuhashi et al. [20] reported that isomeriza-

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
279
Fig. 9. (a) Scheme of pyridine adsorption on different LA and BA sites; (b) pyridine adsorption on Zr(LA) in site IV vs. steric energy
and no H depicted for clarity; (c) pyridine adsorption on Zr(LA) and on Brönsted acid site vs. steric energy and no H depicted for clarity.

280
E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
Brenner et al. [53] published that no H2 is found
• Reported data about FT–IR spectroscopic charac-
terization of SZ before reaction and n-butane iso-
merization support this proposal.
when hydrocarbons are placed in contact with sul-
fated zirconia. Moreover, Hong et al. [34] found H2
evolution and these authors could correlate isobutene
production with H2 generation. Knözinger [37] has
published that no Raman signals of alkenic or aro-
matic compounds are observed after contact of SZ
with n-butane in a flow of H2 at 200 ◦C for 2 h. No
allylic species are found using H2–butane streams and
0.2% Pt-SZ. When H2 is present, hydrocarbon de-
posits seem to appear only at very low concentration.
The monomolecular mechanism has been proposed to
be preferred to bimolecular one [48,52]. At low re-
action temperatures, not more than 250 ◦C, addition
of hydrogen as carrier gas decreases the conversion
comparing it with N2 as carrier gas. At higher temper-
atures, it increases the activity. If the pretreatment is
done in oxidizing atmosphere-air, the catalyst is more
active than under inert atmosphere-N2. Under reduc-
ing atmosphere-H2, the SZ catalyst is almost inactive
[44]. This fact can be easily explained if the Zr3+ at
surface, no closely related to SOH, is oxidized to Zr4+
in air. The Zr4+ can dissociate hydrogen. The surface
concentration of hydride is crucial in the n-butane iso-
merization mechanism.
Acknowledgements
The authors want to acknowledge financial sup-
port from CONICET (Argentina) and UNS. Besides,
we also acknowledge the kind criticism and helpful
advices of Prof. Helmut Knözinger (Germany) who
helped and encouraged us to improve the article in the
way it is now presented.
References
[1] T. Jin, T. Yamaguchi, K. Tanabe, J. Phys. Chem. 90 (1986)
4794.
[2] T. Yamaguchi, Appl. Catal. 1 (1990) 61.
[3] M. Bensitel, O. Saur, J.C. Lavalley, B.A. Morrow, Mater.
Chem. Phys. 19 (1988) 147.
[4] L.M. Kustov, V.B. Kazansky, F. Figueras, D. Tichit, J. Catal.
150 (1994) 143.
[5] F. Haase, J. Sauer, J. Am. Chem. Soc. 120 (1998) 13503.
[6] B. Umansky, J. Engerlhardt, W.K. Hall, J. Catal. 127 (1991)
128.
[7] G. Yaluris, R.B. Larson, J.M. Kobe, M.R. González, K.B.
Fogash, J.A. Dumesic, J. Catal. 160 (1996) 290.
[8] M.R. González, J.M. Kobe, K.B. Fogash, J.A. Dumesic, J.
Catal. 33 (1997) 303.
4. Conclusions
[9] H. Armendariz, C. Sanchez Sierra, F. Figueras, B. Coq, C.
Mirodatos, F. Lefebvre, D.J. Tichit, J. Catal. 171 (1997) 85.
[10] A. Clearfield, G.P.D. Serrete, A.H. Khazi-Syed, Catal. Today
20 (1994) 295.
[11] B.H. Davis, R.A. Keogh, R. Sirinivasan, Catal. Today 20
(1994) 219.
[12] J.E. Tábora, R.J. Davis, J. Catal. 162 (1996) 125.
[13] F. Babou, B. Bigot, P. Sautel, J. Phys. Chem. 97 (1993) 11501.
[14] C.R. Vera, C.L. Pieck, K. Shimizu, C.A. Querini, J.M. Parera,
J. Catal. 187 (1999) 39.
[15] C. Morterra, G. Cerrato, F. Pinna, M. Signoretto, J. Catal.
157 (1995) 109.
[16] F.R. Chen, G. Coudurier, J.F. Joly, J.C. Vedrine, J. Catal. 143
(1993) 616.
[17] E.A. Garc´ıa, E.H. Rueda, A.J. Rouco, Appl. Catal. A 210
(2001) 363.
[18] A. Farcasiu, A. Ghenciu, J. Qi Li, J. Catal. 158 (1996) 116.
[19] K.J. Chao, H.C. Wu, L.J. Leu, J. Catal. 157 (1995) 289.
[20] H. Matsuhashi, H. Shibata, H. Nakamura, K. Arata, Appl.
Catal. Part A: Gen. 187 (1999) 99.
[21] M. Scheithauer, E. Bosch, U.A. Schubert, H. Knözinger, T.G.
Cheung, F.C. Jentoft, B.C. Gates, B. Tesche, J. Catal. 177
(1998) 137.
The main facts of this proposal are:
• Several redox cycles can be present (for dimeric
sites, for Zr4+/Zr3+).
• Superacidity (if any) would take place in the reac-
tion media. It is related to BA (SOHZr), LA (Zr)
and LB (Zr–O–Zr) sites. No superacidity is found
in studies with probe molecules before the reaction.
• Besides the sulfate content and the predominance of
the tetragonal crystalline phase, the surface structure
is very important in the formation of active sites and
intermediaries.
• Hydrogen and methane evolution can be explained.
• Brönsted and Lewis acidity can be taken into ac-
count.
• LA–LB pair sites of SZ are proposed to play a very
important role.
• The effects of H2, CO and butenes on n-butane iso-
merization in SZ can be explained.

E. Garc´ıa et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 263–281
281
[22] T. Riemer, D. Spielbauer, M. Hunger, G.A.H. Mekhemer, H.
[33] K. Ebitani, K. Noishi, H. Hattori, J. Catal. 130 (1991)
257.
[34] Z. Hong, K.B. Fogash, R.M. Watwe, B. Kim, B.L.
Masqueda-Jiménez, M.A. Natal-Santiago, J.M. Hill, J.A.
Dumesic, J. Catal. 178 (1998) 489.
[35] B. Li, R. Gonzalez, Catal. Today 46 (1998) 55.
[36] B.C. Gates, J.R. Katzer, G.C.A. Schuit, Chemistry of Catalytic
Processes, McGraw Hill, New York, 1980.
Knözinger, J. Chem. Soc., Chem. Commun. 1181 (1994).
[23] R. White, E. Sikabwe, M.A. Coelho, D. Resasco, J. Catal.
157 (1995) 758.
[24] S. Coman, V. Paˆrvulescu, P. Grange, V.I. Paˆrvulescu, Appl.
Catal. Part A: Gen. 176 (1999) 45.
[25] D. Spielbauer, G.A.H. Mekhemer, M.I. Zaki, H. Knözinger,
Catal. Lett. 40 (1996) 71.
[26] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A.
Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery
Jr., R.E. Stratmann, J.C. Burant, S. Dapprich, J.M. Millam,
A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi,
V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli,
C. Adamo, J. Clifford, S. Ochterski, G.A. Petersson, P.Y.
Ayala, Q. Cui, K. Morokuma, D.K. Malick, A.D. Rabuck,
K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Ortiz,
A.G. Baboul, B.B. Stefanov, G. Liu, A. Liashenko, P.
Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J. Fox,
T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M.
Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong,
J.L. Andres, C. Gonzalez, M. Head-Gordon, E.S. Replogle,
J.A. Pople, Gaussian, Gaussian’98, Revision A.9, Pittsburgh,
PA, 1998.
[27] T. Kanougi, T. Atoguchi, S. Yao, J. Mol. Catal. Part A: Chem.
177 (2002) 289.
[28] T. Yamaguchi, K. Tanabe, Y.C. Kung, Mater. Chem. Phys.
16 (1986) 67.
[29] V. Adeeva, G.D. Lei, W.M.H. Sachtler, Appl. Catal. Part A:
Gen. 118 (1994) 1.
[30] V. Adeeva, J.W. de Haan, J. Jänchen, G. De Lei, V.
Scünemann, L.J.M. van de Ven, W.M.H. Sachtler, R.A. van
Santen, J. Catal. 151 (1995) 364.
[37] H. Knozinger, Top. Catal. 6 (1998) 107.
[38] J.H. Lunsford, H. Sang, S.M. Campbell, C.-H. Liang, R.G.
Anthony, Catal. Lett. 27 (1994) 305.
[39] M. Boronat, P. Viruela, A. Corma, J. Phys. Chem. 100 (1996)
633.
[40] T. Yamaguchi, Catal. Today 20 (1994) 199.
[41] M.A. Cohelo, W.E. Alvarez, E.C. Sikabwe, R.L. White, D.E.
Resasco, Catal. Today 28 (1996) 415.
[42] Z. Hong, K.B. Fogash, R.M. Watwe, B. Kim, B.L.
Masqueda-Jiménez, M.A. Natal-Santiago, J.M. Hill, J.A.
Dumesic, Catal. Today 51 (1999) 269.
[43] H. Davis, Catal. Today 20 (1994) 219.
[44] H.K. Mishra, A.K. Dalai, K.M. Parida, S.K. Bej, Appl. Catal.
Part A: Gen. 217 (2001) 263.
[45] W. Stichert, F. Schüth, S. Kuba, H. Knözinger, J. Catal. 198
(2001) 277.
[46] C. Morterra, G. Cerrato, Chem. Phys. 1 (1999) 2825.
[47] N. Katada, J. Endo, K. Notsu, N. Yasunobu, N. Naito, M.
Niwa, J. Phys. Chem. B. 104 (2000) 10321.
[48] A. Sayari, Y. Yang, X. Song, J. Catal. 167 (1997) 346.
[49] G. Busca, Catal. Today 41 (1998) 191.
[50] http://webbook.nist.gov/cgi/cbook.cgi?ID=C142085&Units=
SI&Mask=281.
[51] K. Arata, Adv. Catal. 37 (1990) 165.
[31] B.H. Davis, R.A. Keogh, S. Alerasool, D.J. Zalwski, D.E.
Day, P.K. Doolin, J. Catal. 183 (1999) 45.
[52] J.A. Moreno, G. Poncelet, Appl. Catal. Part A: Gen. 210
(2001) 151.
[32] R. Keogh, D. Sparks, J. Hu, I. Wender, J.W. Tierney, W.
Wang, B.H. Davis, Energy Fuels 8 (1994) 755.
[53] A. Brenner, T. Schorodt, B. Shi, B.H. Davis, Catal. Today
44 (1998) 235.