Oxidative desulfurization (ODS) of organosulfur compounds catalyzed by peroxo-metallate complexes of WOx-ZrO2: Thermochemical, structural, and reactivity indexes analyses

Article abstract of DOI:10.1016/j.jcat.2011.06.010

An experimental and theoretical study on the relationships between oxidative reactivity, thermochemical viability, and structural requirement of the activity sites in oxidative desulfurization (ODS) process has been performed. A series of aromatic sulfur compounds and peroxo-metallate complexes of WOx-ZrO₂ with different structures have been studied. The models chosen for mimicking the catalyst correspond to surface densities of ～7 W nm^-2. The results indicate that the ODS takes place in two consecutive stages: (i) the formation of sulfoxide and (ii) the formation of sulfone. However, a detailed analysis suggests that these stages occur in two separated steps, (a) addition and (b) elimination, involving the formation of intermediate adducts and that the elimination of sulfoxide from the site surface is the rate-determining step. The results also reveal that the thermochemical feasibility of the studied reactions depends on both: the local structure of the WOx-ZrO₂ surface and on the nature of the aromatic sulfur compound. It was found that the reactions involving dibenzothiophene (DBT) and 4,6-dimethyldibenzothiophene (46DMDBT) are the most favored reactions, while the reaction of thiophene (Th) is the least favored. Therefore, highly substituted dibenzothiophenes are the most readily oxidized species, which is in agreement with experimental evidence. An explanation to the different reactivity shown by sulfur compounds, during ODS processes, is provided.

Full text of DOI:10.1016/j.jcat.2011.06.010

Journal of Catalysis 282 (2011) 201–208
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Oxidative desulfurization (ODS) of organosulfur compounds catalyzed
by peroxo-metallate complexes of WOx–ZrO : Thermochemical, structural,
2
and reactivity indexes analyses
E. Torres-García , A. Galano b, , G. Rodriguez-Gattorno c
a,
⇑
⇑
a
Instituto Mexicano del Petróleo, Eje Central Norte, Lázaro Cárdenas 152, San Bartolo Atepehuacan, 07730 México DF, Mexico
Departamento de Química, Universidad Autónoma Metropolitana-Iztapalapa, San Rafael Atlixco 186, Col. Vicentina, Iztapalapa, CP 09340 México DF, Mexico
Departamento de Física Aplicada, CINVESTAV-IPN, Mérida, Yucatán 97310, Mexico
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
An experimental and theoretical study on the relationships between oxidative reactivity, thermochemical
viability, and structural requirement of the activity sites in oxidative desulfurization (ODS) process has
Received 8 March 2011
Revised 27 May 2011
Accepted 13 June 2011
Available online 26 July 2011
been performed. A series of aromatic sulfur compounds and peroxo-metallate complexes of WOx–ZrO
with different structures have been studied. The models chosen for mimicking the catalyst correspond
2
ꢁ2
to surface densities of ꢀ7 W nm . The results indicate that the ODS takes place in two consecutive
stages: (i) the formation of sulfoxide and (ii) the formation of sulfone. However, a detailed analysis sug-
gests that these stages occur in two separated steps, (a) addition and (b) elimination, involving the for-
mation of intermediate adducts and that the elimination of sulfoxide from the site surface is the rate-
determining step. The results also reveal that the thermochemical feasibility of the studied reactions
Keywords:
Oxidative desulfurization
Tungstated zirconia
Sulfur compounds
Peroxo-metallate complexes
Density-functional theory (DFT)
2
depends on both: the local structure of the WOx–ZrO surface and on the nature of the aromatic sulfur
compound. It was found that the reactions involving dibenzothiophene (DBT) and 4,6-dimethyldibenzo-
thiophene (46DMDBT) are the most favored reactions, while the reaction of thiophene (Th) is the least
favored. Therefore, highly substituted dibenzothiophenes are the most readily oxidized species, which
is in agreement with experimental evidence. An explanation to the different reactivity shown by sulfur
compounds, during ODS processes, is provided.
Ó 2011 Elsevier Inc. All rights reserved.
1
. Introduction
Oxidative desulfurization (ODS) is considered a promising meth-
egy with great potential to be complementary to traditional HDS for
producing deeply desulfurized light oil and for reaching the environ-
mental regulations.
od for ultra-deep desulfurization of fuel oils. The interest on such
process is well justified since the removal of sulfur compounds from
fuels is necessary for industrial and environmental reasons. Organic
sulfur compounds present in diesel widely vary in structure and
therefore in reactivity during catalytic hydrodesulfurization (HDS)
In the ODS process, the organosulfur compounds are oxidized to
their corresponding sulfoxides or sulfones. The process is carried
out under mild conditions in the presence of a catalyst and an oxi-
dant agent, and the oxidized sulfur compounds are subsequently
removed by extraction, adsorption, distillation, or decomposition
[5–8]. The main advantage of ODS is that sulfur compounds can
be removed without using a hydrogen atmosphere, at relatively
low temperatures and at atmospheric pressure. However, from
technologic points of view, the development and implementation
of the ODS alternative for ultra-low-sulfur processes are still a
challenge for both academia and industry.
An extremely important point relevant to the ODS process is
related to the catalytic material, particularly to the structural
requirements of the active sites and its relationships with the cat-
alytic efficiency. Consequently, a large variety of studies have been
carried out on oxidizing sulfur compounds with different catalysts
and under different conditions [2,3,9–17]. Regardless of the
numerous works on ODS, considering both heterogeneous and
[
1,2]. In HDS, the reactivity of dibenzothiophenes dramatically de-
creases with the increase of methyl substitutes at the sterically hin-
dered positions 4 and 6 [3,4]. Consequently, in deep HDS, the
conversion of these substituted dibenzothiophenes strongly depend
on the experimental conditions. Therefore, an important feature for
application of the ODS in ultra-low-sulfur process is that its relative
efficiency is different from that of HDS: DBT > 4-MDBT > 46-
DMDBT > BT [4–6]. As a consequence, refractory sulfur compounds
in HDS are easily removed by ODS. This makes ODS a chemical strat-
⇑
Corresponding authors.
E-mail addresses: etorresg@imp.mx (E. Torres-García), agalano@prodigy.net.mx
(
A. Galano).
0
021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.06.010

2
02
E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
homogeneous systems, little attention has been paid to systematic
studies on the influence of the structural characteristics of the per-
oxo-complexes surface on the thermochemical feasibility and the
relative reactivity of organosulfur compounds during ODS process.
The researches in this field are difficult, in part due to the fact that
it is very complicated to separate the different factors and also
because they presume establishing the influence of each compo-
nent during the process. Another reason is the absence of system-
atic studies on the chemical nature of sulfur compounds and on
possible relationships between such nature and the characteristics
of the reactive sites on the catalyst surface.
2 2
H O /S ratios 10:1. A Robinson–Mahoney reactor was used in order
to perform the chemical reactions. The oxidation was carried out at
333 K, and the samples were collected from the reaction mixture
during the reaction at time intervals. Both oil and polar phase with
sulfides were analyzed by GC-HP6890 equipped with a HP-5 capil-
lary column (30 m ꢂ 0.25 mm, 0.32
lm film thickness) and FPD/
FID detector. The hydrogen peroxide, presented during the reac-
tion, was measured by standard iodometric titration.
2.3. Conversion of sulfur
In correspondence, the purpose of this work is to provide an anal-
ysis of the relationships between the structural requirements of the
active sites on the catalyst’s surface, the nature of sulfur compounds,
and the thermodynamic feasibility of the key steps in the ODS pro-
cess. To that purpose, a combined experimental and theoretical
study on different sulfur compounds has been realized. For the
experimental study, four organosulfur compounds have been used.
They are thiophene (Th), benzothiophene (BT), dibenzothiophene
The data gathered from the experiments were used to calculate
the removal fraction (S ) of sulfides as follows:
t
ꢀC
ꢁ
0
ꢁ C
t
S
t
¼
C
0
where C is the initial concentration of sulfides in the heptane solu-
0
tion and C is the sulfides concentration of the oil phase after reac-
t
tion time (t).
(
DBT), and 4,6-dimethyldibenzothiophene (46DMDBT) and were se-
lected because they are the most frequents in fuels. For the theoret-
ical study, a larger series of sulfur compound have been chosen, in
order to search for generalizations regarding structure–reactivity
relationships. They are thiophene (Th), benzothiophene (BT), diben-
zothiophene (DBT), 2,5-dimethylthiophene (25DMT), 2-meth-
ylbenzothiophene (2MBT), 4-methylbenzothiophene (4MBT), 2,4-
dimethylbenzothiophene (24DMBT), and 4,6-dimethyldibenzothi-
ophene (46DMDBT). Various reactivity indexes have been analyzed
2.4. Oxidation of different organosulfur compounds
The oxidative reaction of each sulfur compound follows first-
order kinetics, according to previous reports obtained with aro-
matic sulfur compounds on solid catalysts [5,21–24]. Due to the
large excess of hydrogen peroxide, the concentration of this com-
pound remains practically constant. Therefore, the following
equation applies: ln(C /C ) = ꢁkt, where k = A exp(ꢁE/RT), A is the
t
0
to that purpose: hardness (
g
), electrophilicity (
x
), electroaccepting
pre-exponential factor, E the apparent activation energy, R and T
are the gas constant and reaction temperature, respectively.
+
ꢁ
power (
-
), and electrodonating power (
-
). They have been used
to provide a viable explanation to the experimental reactivity of dif-
ferent organosulfur compounds during the ODS process.
2.5. Computational details
The calculations were performed using Gaussian 03 [25] package
of programs. Geometry optimizations have been performed using
the generalized gradient approximation (GGA), without any sym-
metry constraints. The Perdew and Wang’s 1991 functional [26]
(PW91) was used for exchange and correlation potentials. We have
used the Dunning/Huzinaga valence double-zeta D95 V [27] basis
set for O, C, N, and H atoms and the Stuttgart/Dresden [28–30] rela-
tivistic effective core potential (ECP) for W and Zr. Frequency calcu-
lations were carried out for all the studied systems at the same level
of theory. Thermodynamic corrections at 333 K were included in the
calculation of the relative energies. The stationary points were first
modeled in gas phase (vacuum), and solvent effects were included
a posteriori by single-point calculationsusing polarizable continuum
model, specifically the integral-equation-formalism (IEF-PCM) [31–
34] at the same level of theory, with acetonitrile as solvent to mimic
the experimental conditions. For all the modeled paths, the solute
cavity was computed using atomic radii from the universal force
field (RADII = UFF), which assigns individual spheres to H atoms (ex-
plicit hydrogens). Different levels of theory were tested for the cal-
culation of reactivity indexes of the sulfur compounds, and the
reported values are those obtained at B3LYP/6-311++G(d,p) level
of theory.
2
. Experimental
.1. Catalyst preparation
High-surface-area ZrO2ꢁx(OH)2x (320 m /g) was prepared by
2
2
precipitation from zirconium oxychloride solutions with ammonia
as previously described [18,19]. The precipitate was filtrated,
washed repeatedly by re-dispersion with a NH
4
OH solution (pH
ꢁ
1
0) until the elimination of Cl , and then dried at 383 K for 24 h.
Subsequently, ZrO2ꢁx(OH)2x was impregnated with an ammonium
metatungstate solution, adjusted to pH 10 with ammonia, as
reported in Ref. [20]. After drying at 383 K for 24 h, the samples
were treated in flowing dry air for 3 h at 1073 K. The stabilized spe-
cific surface area of WOx–ZrO
m /g and pore size of about 6.2 nm. WOx surface densities used
2
in calcined samples was about 50
2
ꢁ2
in this study was of ꢀ7 W nm
.
2.2. Catalytic test
A
synthetic diesel was prepared with heptane (C H ,
7 16
9
1
9.0 mol.%) as solvent, and 140 S ppmw of thiophene (Th, 99+%),
86 S ppmw of benzothiophene (BT, 98%), 222 S ppmw of dibenzo-
The rate constants (k) were calculated using Conventional
Transition State Theory (TST) [35–37] and 1 M standard state as:
thiophene (DBT, 98%), and 173 S ppmw of 4,6-dimethyldiebenzo-
thiophene (46DMDBT, 97%) were used as the sulfides. Hydrogen
peroxide (aqueous solution, 30 wt.%) and acetonitrile (CH
k
h
B
T
G–Þ=RT
ꢁ
ð
D
3
CN,
k ¼
e
ð1Þ
9
9.93 mol.%) were used as oxidizing agent and as extraction sol-
–
vent, respectively. All the reagents used in this study are commer-
cially available (Sigma/Aldrich).
where k
B
and h are the Boltzmann and Planck constants and DG is
the Gibbs free energy of activation.
In an typical run, the solid catalyst, 0.1 g of WOx–ZrO
2
catalyst
Some of the calculated rate constants (k) are in the diffusion-
limit. Accordingly, the apparent rate constant (kapp) cannot be
directly obtained from TST calculations. In the present work, we
have used the Collins–Kimball theory for that purpose [38]:
was suspended under vigorous stirring (1200 rpm) in a mixture
containing 50 mL synthetic diesel, solvent (model fuel/solvent ra-
tios 1:1, v/v), and oxidizing hydrogen peroxide was added in

E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
203
k
D
k
act
this analysis showed that to obtain an adequate balance between lo-
cal coordination, polymerization degree of the WOx domains, and
high-density surface of Brönsted acid sites, the WOx–ZrO system
2
requires an exact control of surface acid properties.
k
app ¼ k
ð2Þ
is
D
þ kact
where kact is the thermal rate constant, obtained from TST, and k
the steady-state Smoluchowski [39] rate constant for an irreversible
bimolecular diffusion-controlled reaction:
D
Accordingly, our catalytic study focuses on the zone where the
oxidative efficiency of the reaction reaches its maximum (WOx sur-
ꢁ2
face density ꢀ7 W nm ). Oxidative removals for different organic
k
D
¼ 4
pRDAB
N
A
ð3Þ
sulfur compounds as a function of reaction time over WOx–ZrO at
2
where R denotes the reaction distance, N
A
is the Avogadro number,
333 K are shown in Fig. 2A. The maximum conversion is obtained
for DBTs and BT after 60 min of reaction. By contrast, the reaction
of Th is the least favored. The oxidative reactivity, evaluated as oxi-
dation rate constants (k) (Fig. 2B), decreased according to the follow-
ing order: DBT > 46DMDBT > BT ꢃ Th, in good agreement with
theoretical and experimental previous reports [5,10,11,43,44]. The
oxidized products of DBTs and BT are the corresponding sulfones,
while sulfoxides were not detected as products. In the case of thio-
and DAB is the mutual diffusion coefficient of the reactants. How-
ever, in this study, DAB has been assumed equal to D (where A rep-
resents the sulfur compound) since the catalyst is not expected to
diffuse. D
40]:
A
A
has been estimated from the Stokes–Einstein approach
[
B
k T
pga
D
A
¼ 6
ð4Þ
2 4
phene, the reaction yielded H SO and thiophene sulfoxide, but thi-
ophene sulfone was not detected. This can be explained based on the
results of our calculations. Among all the studied sulfur compounds,
only in the case of thiophene, the structure of the addition product
formedin stage 2 (a) showsa noticeableweakeningof the C–Sbonds.
Moreover, this feature is also present in the thiophene sulfone. This
structural characteristic supports the possibility of the sulfone
where k
B
is the Boltzmann constant, T is the temperature,
g
denotes
the viscosity of the solvent, in our case acetonitrile (
g
= 4.41 ꢂ
ꢁ4
1
0
Pa s), and a the radius of the solute (aDBT = 8.15 Å).
3
. Results and discussion
decomposition yielding SO
phene. Under oxidation conditions, it can evolve into SO
2
, when the ODS process involves thio-
. Addition-
3.1. Catalytic reaction
3
ally, in the presence of water, these sulfur oxides can be converted to
the corresponding acids. These results suggest that the products
arising from oxidation reactions would be dependent on the nature
of the reacting sulfur compound.
The results clearly indicate that DBT and 46DMDBT are the most
readily oxidized species, suggesting that under the experimental
condition used in this study, the order of reactivity can be related
to the electron density on the sulfur atoms in these compounds
On the basis of previous works [18,41] and in order to establish
the relationships between the structural characteristics of the reac-
tive sites on the catalyst surface and the thermochemical feasibility
of the reactions of different organosulfur compounds in oxidative
2
desulfurization, a model system with WOx–ZrO as catalyst, aceto-
nitrile as extraction solvent, and hydrogen peroxide as oxidant has
been chosen. Recently, we reported that oxidative efficiency (ex-
ꢁ2
ꢁ1
pressed as % sulfur removed/W atom nm min ) of the WOx–
ZrO system (per W atom) for ODS of DBT increases with increasing
[
5,43,44]. Apparently, the number of rings and the number of methyl
2
groups both increase the oxidative reactivity of these compounds.
This is probably because the methyl groups in the aromatic rings
act as electron donors. This behavior enhances the electron density
on the sulfur atom, promoting the catalytic reaction.
Therefore, to get deeper insight, a detailed study of the reaction
steps is crucial to rationalize the relative oxidative reactivity of the
series of organosulfur compounds found in fuels. To that purpose,
a theoretical systematic study was accomplished focusing on the
thermochemical viability of such process through estimations of
spontaneity. Consequently, the relevant points would be those cor-
responding to the initial and final states. This approach is supported
by the Hammond postulate and the Bell–Evans–Polanyi principle.
ꢁ2
WOx surface density up to a maximum of ꢀ7 W nm (see Fig. 1).
This value represents the point where the local coordination and
polymerization degree of the WOx domains, in mutual cooperation,
produce a maximum number of Brönsted sites per surface unit, pro-
moting the addition of OOH groups. The study also revealed that the
combined presence of Lewis and Brönsted sites energetically fa-
vored the formation of peroxo-metallate and that dimeric species
are adequate to describe the properties of the peroxo-complexes
surface from chemical and catalytic points of view [19,42]. Clearly,
3
3
.2. Theoretical considerations
.2.1. Thermochemical viability of the ODS reactions
According to the experimental results shown in the Fig. 1, all
models in this study correspond to the zone where catalytic reaction
ꢁ2
reaches its maximum (ꢀ7 W nm ). Six different arrangements
have been chosen to mimic the local surface of the WOx–ZrO cata-
2
lyst. The structures used to modelthe peroxo-complexes correspond
to those previously proposed as the most likely products arising
from peroxidation of WOx–ZrO
Supporting Information (Fig. 1S). The original WOx–ZrO
2
surfaces [42] and are provided as
structures
2
correspond to tetrahedral W, with different degrees of surface
hydroxylation [19,42]. Their ODS reactions with eight aromatic sul-
fur compounds have been modeled. The studied sulfur compounds
are thiophene (Th), benzothiophene (BT), dibenzothiophene
(
(
DBT), 2,5-dimethylthiophene (25DMT), 2-methylbenzothiophene
2MBT), 4-methylbenzothiophene (4MBT), 2,4-dimethylbenzothi-
Fig. 1. Catalytic efficiency as function of WOx surface density (W nmꢁ2) [42].
ophene (24DMBT), and 4,6-dimethyldibenzothiophene(46DMDBT).

2
04
E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
Fig. 2. (A and B) ODS of different sulfur compounds over WOx–ZrO
2
catalyst as a function reaction time. (B) Rate constants (k) of different sulfur compounds, evaluated as
/S = 10:1; model fuel/acetonitrile ratios 1:1 v/v; 333 K and 101 kPa.
oxidation rate constants (k). Reaction conditions: WOx–ZrO , 0.1 g; H O
2
2 2
The ODS is proposed to take place in two consecutive stages, the
first one leading to the formation of sulfoxide and the second one
yielding sulfone:
sulfur compound, that is DBT and 46DMDBT were found to be the
ones involved in the most thermochemically favored reactions. Th
and 25DMT, on the other hand, were found to be the sulfur com-
pounds leading to the least thermochemically favored first stages
of the ODS reactions. These considerations are agreement with
our experimental result.
Stage 1: POO + XS ? PO + XSO
Stage 2: POO + XSO ? PO + XSO
2
where POO represents the peroxo-metallate, XS the aromatic
sulfur compound, XSO the sulfoxide, and XSO the sulfone. The di-
rect mechanism yielding the XSO compounds was found to be lar-
gely endergonic (by more than 100 kcal/mol), and therefore, it was
ruled out.
The enthalpies and Gibbs free energies of reactions of the ODS
stages are reported in Tables 1 and 2. Peroxo-complexes with
structures B1 and Eb were found to be those leading to the most
thermochemically favored reactions, for all the modeled sulfur
compounds. Their structures are shown in Fig. 3. In general, the
thermochemical feasibility of the second stage of reaction was
found to be larger than that of the first one. Therefore, the latter
seems to be the key stage for the success of the ODS process.
The first stage of reactions involving structures B2, Ea, and Fa
3
.2.2. Kinetics analysis and mechanism of reaction
So far, stages 1 and 2 have been considered to take place as ele-
2
2
mentary reactions. However, another alternative is also possible
that each stage actually takes place in two steps, involving the for-
mation of intermediate adducts (Fig. 4). According to this hypoth-
esis, the ODS mechanism would become:
Stage 1:
Step 1a: POO + XS ? POO-SX
Step 1b: POO-SX ? PO + XSO
Stage 2:
Step 2a: POO + XSO ? POO-S(O)X
Step 2b: POO-S(O)X ? PO + XSO2
(
Figure 1S) was found to be endergonic for all the studied sulfur
Within this scheme, step 1a represents the addition of the sulfur
compound to the peroxo-metallate, step 1b is the elimination of the
sulfoxide, step 2a the addition of the sulfoxide to another peroxo
site, and step 2b the elimination of the sulfone.
The four steps scheme has been modeled for all the studied sul-
fur compounds and cluster B1. The corresponding Gibbs free ener-
gies of reaction are reported in Table 3. As these values show the
compounds. This suggests that such kinds of sites are unlike to par-
ticipate in the ODS reactions. When the peroxo-complex corre-
sponds to structure D (Figure 1S), most of the first stages of
reaction are predicted to be endergonic, with the exceptions of
the reactions involving DBT and 46DMDBT. In fact, a general trend
was found for the first stage of reaction with respect to the reacting
Table 1
Enthalpies of reaction for the OOS process, in kcal/mol at 333 K.
Th
25DMT
BT
2MBT
4MBT
DBT
24DMBT
46DMDBT
Stage 1
B1
B2
D
Ea
171
1509
9.18
1421
ꢁ0.69
1673
ꢁ0.60
9.50
3.58
ꢁ1.29
ꢁ1.42
8.68
2.76
5.39
4.79
ꢁ1.83
8.26
2.34
ꢁ5.26
ꢁ2.05
8.04
2.12
ꢁ8.68
4.70
8.80
2.89
4.83
ꢁ1.09
1.54
ꢁ1.20
621
5.51
4.66
10.42
4.97
4.75
3.83
Eb
Fa
ꢁ3.97
ꢁ5.20
ꢁ8.64
ꢁ5.42
ꢁ11.08
6.34
11 12
10.30
9.88
6.45
9.66
Stage 2
B1
B2
D
Ea
ꢁ1207
131
4.60
ꢁ11.80
ꢁ1.71
ꢁ7.62
ꢁ4.99
ꢁ15.17
ꢁ0.08
ꢁ9.62
0.47
ꢁ5.44
ꢁ2.82
ꢁ12.99
2.09
ꢁ10.22
ꢁ0.13
ꢁ6.04
ꢁ3.42
ꢁ13.59
1.50
ꢁ9.91
0.18
ꢁ5.74
ꢁ3.11
ꢁ13.28
1.80
ꢁ9.78
0.31
ꢁ5.60
ꢁ2.98
ꢁ13.15
1.94
ꢁ10.35
ꢁ0.26
ꢁ6.17
ꢁ3.55
ꢁ13.72
1.36
14.14
ꢁ0.77
ꢁ6.67
ꢁ1.64
16.55
0.87
043
Eb
Fa
ꢁ1448
294

E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
205
Table 2
Gibbs free energies of reaction for the ODS process, in kcal/mol at 333 K.
Th
25DMT
BT
2MBT
4MBT
DBT
24DMBT
46DMDBT
Stage 1
Bl
B2
D
Ea
3.59
15.19
8.48
11.31
2.86
15.69
ꢁ2.05
9.56
2.85
ꢁ2.84
ꢁ2.77
8.83
2.12
ꢁ3.27
8.34
1.63
ꢁ6.80
ꢁ3.43
8.18
1.47
ꢁ6.60
5.01
8.77
2.06
4.80
ꢁ1.91
0.91
ꢁ1.70
5.67
4.88
4.94
4.45
4.29
1.12
Eb
Fa
ꢁ2.77
ꢁ3.56
ꢁ3.50
ꢁ3.99
ꢁ7.53
ꢁ4.15
ꢁ7.72
5.50
10.05
9.26
9.33
8.83
5.29
8.67
Stage 2
Bl
B2
D
Ea
ꢁ9.18
2.43
ꢁ4.28
ꢁ1.46
ꢁ9.90
2.92
ꢁ12.57
ꢁ0.96
ꢁ7.67
ꢁ4.85
ꢁ13.29
ꢁ0.47
ꢁ10.14
1.46
2.5
ꢁ2.42
ꢁ10.87
1.96
ꢁ10.99
0.62
ꢁ6.09
ꢁ3.27
ꢁ11.71
1.11
ꢁ10.79
0.82
ꢁ5.89
ꢁ3.07
ꢁ11.51
1.31
ꢁ10.27
1.34
ꢁ5.37
ꢁ2.55
ꢁ10.99
1.83
ꢁ11.72
0.11
ꢁ6.82
ꢁ400
ꢁ1244
0.38
ꢁ11.67
ꢁ0.06
ꢁ6.77
ꢁ3.95
ꢁ12.39
0.43
Eb
Fa
to be endergonic for all the studied systems. This means that this
is a reversible step, for which the direct reaction is not thermo-
chemically favored. However, the products from step 1b can still
react further if the following step (2a) is sufficiently exergonic to
provide a driving force and if its barrier is low. In such cases, the
small amounts of sulfoxide formed from step 1b would be con-
sumed though step 2a causing the equilibrium to evolve toward
the formation of more sulfoxide, and the overall process would
proceed slowly. Accordingly, step 1b (the elimination of sulfoxide
from the catalyst surface) is proposed to be the rate-determining
step.
To test this hypothesis, kinetic calculations have been performed
for theODS process of DBT, using the B1cluster to model the catalyst.
The values of the Gibbs free energy of activation and rate constants
for the four steps are provided in Table 4. These values show that the
barriers of the addition steps (1a and 2a) are significantly lower than
those of the eliminationsteps (1b and 2b). Moreover, the highest bar-
rier corresponds to Step 1b (Fig. 5). According to the calculated rate
constants, the addition steps are predicted to be diffusion controlled.
The smallest rate constant corresponds to Step 1b, supporting the
Fig. 3. Structures of the peroxo-metallates B1 and Eb. The boxes highlight the
peroxo-sites. (red = O, dark blue = W, light blue = Zr, white = H). (For interpretation
of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
addition steps are systematically more favored than the elimina-
tion steps. Moreover, the elimination of the sulfoxide was found
Fig. 4. Reaction steps of the ODS process, exemplified for the B1 + DBT. (red = O, dark blue = W, light blue = Zr, white = H, gray = C, yellow = S). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

2
06
E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
Table 3
Gibbs free energies of reaction for the four steps scheme, in kcal/mol at 333 K.
Step 1a
Step 1b
Step 2b
Step 2b
Th
ꢁ5.59
ꢁ10.16
ꢁ10.00
ꢁ10.25
ꢁ10.60
ꢁ12.27
ꢁ11.71
ꢁ12.50
9.18
8.11
7.17
7.47
7.33
5.46
8.28
5.90
ꢁ15.93
ꢁ13.23
ꢁ9.33
6.75
0.66
2
5DMT
BT
ꢁ0.82
ꢁ0.94
ꢁ1.96
ꢁ3.06
ꢁ1.66
ꢁ3.84
2
4
MBT
MBT
ꢁ10.04
ꢁ8.83
DBT
ꢁ7.21
2
4
4DMBT
6DMDBT
ꢁ10.06
ꢁ7.83
Table 4
Gibbs free energy of activation (kcal/mol) and rate constants for the different steps of
the ODS process involving B1 and DBT at 333 K.
–
D
G
k
ꢁ
l
l
ꢁ1
ꢁ1
Step 1a
Step 2b
Step 2a
Step 2b
4.40
23.64
6.63
1.12E+09
2.38Eꢁ03
1.00E+09
1.30E+04
M
s
M
s
s
s
ꢁ
1
Fig. 6. Correlation between the Gibbs free energy of reaction for the step 1b
(highest energy barrier) and the experimental reaction constants. Note that the
parameters are expressed as ln (kexperimental) and the negative value of Gibbs energy
of reaction.
ꢁ
ꢁ
1
13.37
between the experimental k and the calculated
DG1b strongly sup-
ports the proposed mechanism and the identification of the rate-
determining step.
3.2.3. Reactivity Indexes analyses
The theoretical analysis was extended to global reactivity in-
dexes due to their potential impact in estimating the physico-
chemical behavior of these model molecules.
They were calculated from vertical ionization energies (I) and
vertical electron affinities (A), which were computed according to:
X
nꢁ1
X
n
I ¼ E ðg Þ ꢁ E ðg Þ
ð6Þ
n
n
X
n
X
nþ1
A ¼ E ðg Þ ꢁ E ðgnÞ
ð7Þ
n
X
where E ðg Þ is the energy of the n-electron system calculated at the
n
n
X
X
optimized geometry g
n
and E ðg Þ and E ðg Þ are the energies of
nꢁ1 nþ1
n
n
Fig. 5. Reaction profile for the ODS process involving B1 and DBT.
the (n ꢁ 1)-electron and (n + 1)-electron species calculated at the g
n
geometry. Since the reactivity indexes are obtained from I and A
values, their accuracy would be determined by those of I and A.
Therefore, a comparison between calculated and experimental val-
ues of I and A has been performed in order to test the reliability of
the present calculations (Table 1S).
hypothesis that this is the rate-determining step of the ODS process.
The calculated rate constant for this step was found to be
ꢁ
3 ꢁ1
2
.38 ꢂ 10
mental value 7.3 ꢂ 10
.2 times larger than that obtained from the experiments. This
s , which is in excellent agreement with the experi-
ꢁ
4 ꢁ1
s
. In fact, the calculated constant is only
3
The analyzed reactivity indexes are hardness (
g
), electrophilic-
+
agreement validates not only the above-proposed hypothesis but
also the reliability of the performed calculations.
It is also notable to observe the correlation obtained between
the Gibbs free energy of reaction step 1b and the experimental rate
constants. In Fig. 6, plots of both parameters are shown simulta-
neous for each model molecule.
Even though the Eyring equation (1) establishes a direct rela-
tionship between the Gibbs free energy of activation and the rate
constant, the Bell–Evans–Polanyi principle also establishes a linear
relation between the activation energy (Ea) and enthalpy of reac-
ity (
x
), electroaccepting power (
-
), and electrodonating power
ꢁ
(-
). The absolute hardness plays an important role in Pearson’s
hard and soft acids and bases (HSAB) [45–47] and maximum hard-
ness principles (MHP) [47,48]. It is defined as the second derivative
of the electronic energy of the system with respect to the number
of electrons, at a constant external potential [49]:
!
2
1
2
@ E
g
¼
ð8Þ
2
@N
mðrÞ
It can be evaluated, based on the commonly used finite difference
approximation, as:
tion (D
Hr) within a series of closely related chemical processes:
Ea ¼ A þ B
D
Hr
ð5Þ
g
¼ I ꢁ A
ð9Þ
In the studied case, the rate-determining step is of similar nature for
all sulfur compounds, i.e., an elimination step. Therefore, the entro-
py change associated with this step is expected to be similar for all
of them, and the Bell–Evans–Polanyi principle is fulfilled. The
simultaneous manifestation of this principle and the Eyring
equation justifies the correlation observed in Fig. 6. This correlation
and represents the resistance of a chemical species to change its
number of electrons or the shape of its electron cloud.
Electrophilicity is also a very important global reactivity
descriptor since its value would determine the chemical behavior
of the reacting species, i.e., when comparing different species that
can act as electrophile, as it is expected to be the case of the sulfur

E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
207
compounds in the reactions studied in this work, the one with the
highest electrophilicity is expected to be the most reactive one. In
Since, as discussed above, the first stage of the ODS process
seems to be the most difficult to accomplish from a thermochem-
ical point of view, it is proposed that structures resembling clusters
B1 and Eb would favored the desired reactions. When these struc-
tures are involved, both stages of reaction were found to be exer-
gonic for all the studied sulfur compounds but Th. For the latter,
the first stage of the ODS was found to be endergonic, regardless
of the structure of the active site in the catalyst. However, even
for thiophene, the sum of the Gibbs free energies of the two stages
leads to a negative overall Gibbs free energy. Regarding the sulfur
compounds, it is predicted that the reactions involving DBT and
this work,
x
has been calculated as:
2
ðI þ AÞ
x
¼
ð10Þ
8
ðI ꢁ AÞ
as proposed by Parr et al. [50] for the ground-state parabola model.
Electroaccepting and electrodonating powers, which have been
recently presented by Gazquez et al. [51], are ideal for describing
the propensity of a given chemical species to donate or accept frac-
tional amounts of charge. They are expected to show a similar
behavior to that of the first ionization potential and the electron
affinity, respectively. However, while I and A measure the capability
4
6DMDBT would lead to the largest energy release. However, since
both stages of the ODS process (and the overall Gibbs free energy of
reaction) were found to be significant exergonic for all the studied
sulfur compounds, when reacting with peroxo-complex B1 and Eb
sites, such structures are predicted to efficiently accomplish oxida-
tive desulfurization. Additionally, our results show that not all the
reaction sites on the surface of peroxo-complexes domains are
equally actives and predict that thermochemical feasibility
strongly depend on both: chemical nature of the active sites on
the catalyst and nature of the aromatic sulfur compound.
+
ꢁ
of a chemical system to donate or accept one electron,
- and -
measure the capability of a chemical system to donate or to accept
a small fractional amount of charge [52]. They have been calculated
as:
2
ðI þ 3AÞ
-þ
-ꢁ
¼
ð11Þ
1
6ðI ꢁ AÞ
2
ð3I þ AÞ
¼
ð12Þ
1
6ðI ꢁ AÞ
4
. Conclusions
The values of the ionization energies, electron affinities, hardness,
electrophilicity, electrodonating power, and electroaccepting power
for the studied sulfur compounds are reported in Table 5. As these
values show the trends described above, concerning the thermo-
chemical feasibility of the first stage of the ODS process correlates
The experimental and theoretical study of the relationships be-
tween oxidative reactivity, thermochemical viability, and surface
structural requirement of the active sites in oxidative desulfuriza-
tion (ODS) process allows establishing the following conclusions:
The results also reveal that the ODS takes place in two stages,
the formation of sulfoxide and the formation of sulfone and that
each stage occur in two independent steps, addition and elimina-
tion, involving the formation of intermediate adducts. So the val-
ues of the Gibbs free energy of activation show that the barriers
of the addition steps (1a and 2a) are significantly lower than those
of the elimination steps (1b and 2b), supporting the hypothesis that
the elimination is the rate-determining step of the ODS process.
The experimental oxidative reactivity evaluated as oxidation
rate constants (k) decreased according to the order:
DBT > 46DMDBT > BT ꢃ Th, which is related to the electron density
on the sulfur atoms.
ꢁ
+
well with the values of g, - , and - . The compounds with the low-
er hardness (DBT and 46DMDBT) are those leading to the most neg-
ative values of Gibbs free energies of reaction, while those with the
highest hardness (Th and 25DMT) correspond to the least thermo-
dynamically favored reactions.
At this point, it seems important to keep in mind that since
charge acceptance processes stabilize the system, larger values of
+
-
imply a larger capability to accept electrons. Charge donating
processes, on the other hand, destabilize the system, and therefore,
ꢁ
smaller values of
-
indicate a larger capability to donate elec-
ꢁ
+
trons [52] Therefore, and according to the values of
- and -
reported in Table 5, DBT and 46DMDBT have the largest capability
to accept electrons and the lowest capability to donate electrons.
The reactivity indexes of the studied sulfur compounds provide
an explanation to the thermochemical feasibility of the ODS pro-
cess and confirm that the thermochemistry of the reaction depends
on both: the structural characteristic of the surface sites and the
nature of the aromatic sulfur compounds.
The oxidized products of DBTs and BT were experimentally
identified as the corresponding sulfones, while sulfoxides were
not detected as products. In the case of thiophene, the reaction
2
5DMT and Th, on the other hand, have the lowest capability to
accept electrons and the largest capability to donate electrons.
Therefore, it can be inferred that in the first stage of the ODS pro-
cess, the charge transfer takes place from the peroxo-complex to-
ward the sulfur compound. This seems coherent with the fact
that the oxygen atoms in the peroxo-group of the catalyst are neg-
atively charged, while the S atoms in the sulfur compounds are
positively charged. Therefore, the computed reactivity indexes pro-
vide an explanation to the thermochemical feasibility of the differ-
ent modeled reactions.
2 4
yielded H SO and thiophene sulfoxide, while thiophene sulfone
was not detected. This has been explained based on the structural
features of the thiophene sulfone, obtained from calculations. All
this supports that the products arising from oxidation reactions
would be dependent on the nature of the reacting sulfur
compound.
The thermochemical study indicates that the sulfone formation
(stage 2) is most favored than the sulfoxide formation (stage 1) and
supports the experimental evidence that ODS reactions involving
DBT and 46DMDBT are particularly favored, compared to those
involving other sulfur compounds.
Table 5
Ionization energies (I), electron affinities (A), hardness
(
-
g
), electrophilicity
(
x
),
ꢁ
+
electrodonating power (
-
), and electroaccepting power (
), in eV, for the studied
sulfur compounds.
ꢁ
+
I
A
g
x
-
-
Th
6.55
0.99
0.90
1.34
1.30
1.31
1.66
1.28
1.56
5.57
5.13
4.77
4.73
4.74
4.38
4.69
4.41
1.28
1.17
1.45
1.42
1.43
1.70
1.40
1.61
4.78
4.39
5.07
4.97
4.99
5.59
4.91
5.37
1.02
0.92
1.34
1.31
1.31
1.74
1.28
1.61
2
5DMT
6.03
6.11
6.03
6.05
6.05
5.97
5.97
Thermochemical feasibility and global reactivity indexes
computed for different sulfur compounds indicates that the ODS
reactivity decreases in the following order: DBT > 46DMDBT >
BT
2
4
MBT
MBT
DBT
2
4DMBT > 4MBT ꢀ BT ꢀ 2MBT ꢀ 25DMT ꢃ Th. This trend con-
2
4
4DMBT
6DMDBT
firms that the reactivity of the sulfur compounds correlate well
with their electronic properties, such as: g, - , and - and that
ꢁ
+

2
08
E. Torres-García et al. / Journal of Catalysis 282 (2011) 201–208
dibenzothiophene (DBT), alkyl-substituted sulfur compounds, espe-
cially 4,6-dimentyldibenzothiophene (46-DMDBT), are particularly
reactive in the oxidation reactions.
[22] L. Cedeño-Caero, E. Hernandez, F. Pedraza, F. Murrieta, Catal. Today 107–108
2005) 564.
(
[
[
23] V. Hulea, F. Fajula, J. Bousquet, J. Catal. 198 (2001) 179.
24] A. Ishihara, D. Wang, F. Dumeignil, H. Amano, E.W. Qian, T. Kabe, Appl. Catal. A:
Gen. 279 (2005).
[
25] Gaussian 03, Revision E.01, M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria,
M.A. Robb, J.R. Cheeseman Jr., J.A. Montgomery, T. Vreven, K.N. Kudin, J.C.
Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G.
Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R.
Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M.
Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, V. Bakken, C. Adamo, J.
Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C.
Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J.
Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas,
D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G.
Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I.
Komaromi, 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,
C. Gonzalez, J.A. Pople, Gaussian, Inc., Wallingford CT, 2004.
Acknowledgments
A. G. thanks Laboratorio de Visualización y Cómputo Paralelo at
UAM – Iztapalapa for the access to its computer facilities.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.06.010.
[
[
26] J.P. Perdew, K. Burke, Y. Wang, Phys. Rev. B 54 (1996) 16533. and references
therein.
27] T.H. Dunning Jr., P.J. Hay, Modern Theoretical Chemistry, in: H.F. Schaefer III
(Ed.), vol. 3, Plenum, New York, 1976.
References
[
1] M. Houalla, D.H. Broderick, A.V. Sapre, N.K. Nag, V.H.J. De Beer, B.C. Gates, H.
Kwart, J. Catal. 61 (1980) 523.
[28] L.v. Szentpaly, P. Fuentealba, H. Preuss, H. Stoll, Chem. Phys. Lett. 93 (1982)
555.
[29] H. Stoll, P. Fuentealba, P. Schwerdtfeger, J. Flad, L.v. Szentpaly, H. Preuss, J.
Chem. Phys. 81 (1984) 2732.
[30] P. Fuentealba, H. Preuss, H. Stoll, L.v. Szentpaly, Chem. Phys. Lett. 89 (1989)
418.
[
[
[
[
2] C. Song, X. Ma, Appl. Catal. B: Environ. 41 (2003) 207.
3] F.M. Collins, A.R. Lucy, C. Sharp, J. Mol. Catal. A: Chem. 117 (1997) 397.
4] M. Te, C. Fairbridge, Z. Ring, Appl. Catal. A: Gen. 219 (2001) 267.
5] S. Otsuki, T. Nonaka, N. Takashima, W. Qian, A. Ishihara, T. Imai, T. Kabe, Energy
Fuels 14 (2000) 1232.
[
[
6] J.M. Campos-Martin, M.C. Capel-Sanchez, P. Perez-Presas, J.L.G. Fierro, J. Chem.
Technol. Biotechnol. 85 (2010) 879.
7] C. Li, Z. Jiang, J. Gao, Y. Yang, S. Wang, F. Tian, F. Sun, X. Sun, P. Ying, C. Han,
Chem. Eur. J. 10 (2004) 2277.
8] J.M. Fraile, J.I. García, B. Lázaro, J.A. Mayoral, Chem. Commun. (1998) 1807.
9] J.B. Gao, S.G. Wang, Z.X. Jiang, H.Y. Lu, Y.X. Yang, F. Jing, C. Li, J. Mol. Catal. A:
Chem. 258 (2006) 261.
[31] M.T. Cances, B. Mennucci, J. Tomasi, J. Chem. Phys. 107 (1997) 3032.
[32] B. Mennucci, J. Tomasi, J. Chem. Phys. 106 (1997) 5151.
[33] B. Mennucci, E. Cances, J. Tomasi, J. Phys. Chem. B 101 (1997) 10506.
[34] J. Tomasi, B. Mennucci, E. Cances, THEOCHEM 464 (1999) 211.
[35] H. Eyring, J. Chem. Phys. 3 (1935) 107.
[36] M.G. Evans, M. Polanyi, Trans. Faraday Soc. 31 (1935) 875.
[37] D.G. Truhlar, W.L. Hase, J.T. Hynes, J. Phys. Chem. 87 (1983) 2664.
[38] F.C. Collins, G.E. Kimball, J. Colloid Sci. 4 (1949) 425.
[39] M. Smoluchowski, Z. Phys. Chem. 92 (1917) 129.
[40] (a) A. Einstein, Ann. Phys. (Leipzig) 17 (1905) 549;
(b) G.G. Stokes, Mathematical and Physical Papers, vol. 3, Cambridge
University Press, Cambridge, 1903. esp. Sect. IV, p. 55.
[41] L.F. Ramirez-Verduzco, E. Torres-Garcia, R. Gomez-Quintana, V. Gonzalez-
Pena, F. Murrieta-Guevara, Catal. Today 98 (2004) 289.
[42] G. Rodriguez-Gattorno, A. Galano, E. Torres-Garcia, Appl. Catal. B: Environ. 92
(2009) 1.
[
[
[
[
[
10] H. Lü, J. Gao, Z. Jiang, F. Jing, Y. Yang, G. Wang, C. Li, J. Catal. 239 (2006) 369.
11] H. Lü, Y. Zhang, Z. Jiang, C. Li, Green Chem. 12 (2010) 1954.
12] F. Figueras, J. Palomeque, S. Loridant, C. Fèche, N. Essayem, G. Gelbard, J. Catal.
26 (2004) 25.
13] D. Huang, Z. Zhai, Y.C. Lu, L.M. Yang, G.S. Luo, Ind. Eng. Chem. Res. 46 (2007)
447.
14] F. Al-Shahrani, T.C. Xiao, S.A. Llewellyn, S. Barri, Z. Jiang, H.H. Shi, G. Martinie,
M.L.H. Green, Appl. Catal. B: Environ. 73 (2007) 311.
15] H. Li, L. He, J. Lu, W. Zhu, X. Jiang, Y. Wang, Y. Yan, Energy Fuels 23 (2009) 1354.
16] J.L. Garcia-Gutierrez, G.A. Fuentes, M.E. Hernandez-Teran, P. Garcia, F.
Murrieta-Guevara, F. Jimenez-Cruz, Appl. Catal. A: Gen. 334 (2008) 366.
17] H. Gómez-Bernal, L. Cedeño-Caero, A. Gutiérrez-Alejandre, Catal. Today 142
2
[
[
1
[
[
[43] A. Milenkovic, E. Schulz, V. Meille, D. Loffreda, M. Forissier, M. Vrinat, P. Sautet,
M. Lemaire, Energy Fuels 13 (1999) 881.
[44] Y. Shiraishi, K. Tachibana, T. Hirai, I. Komasawa, Ind. Eng. Chem. Res. 41 (2002)
4362.
[
[
[
[
[
(
2009) 227.
18] E. Torres-Garcia, G. Canizal, S. Velumani, L.F. Ramirez-Verdusco, F. Murrieta-
Guevara, J.A. Ascencio, Appl. Phys. A 79 (2004) 2037.
19] A. Galano, G. Rodriguez-Gattorno, E. Torres-Garcia, Phys. Chem. Chem. Phys.
[45] R.G. Pearson, J. Am. Chem. Soc. 85 (1963) 3533.
[46] R.G. Pearson, Science 151 (1966) 172.
[47] R.G. Pearson, Chemical Hardness, John Wiley and Sons, New York, 1997.
[48] R.G. Pearson, Acc. Chem. Res. 26 (1993) 250.
[49] R.G. Parr, R.G. Pearson, J. Am. Chem. Soc. 105 (1983) 7512.
[50] R.G. Parr, L.v. Szentpaly, S. Liu, J. Am. Chem. Soc. 121 (1999) 1922.
[51] J.L. Gazquez, A. Cedillo, A. Vela, J. Phys. Chem. A 111 (2007) 1966.
[52] J.L. Gázquez, J. Mex. Chem. Soc. 52 (1) (2008) 3.
1
0 (2008) 4181.
20] A. Iribarren, G. Rodriguez-Gattorno, J.A. Ascencio, A. Medina, E. Torres-Garcia,
Chem. Mater. 18 (2006) 5446.
21] P. Moreau, V. Hulea, S. Gomez, D. Brunel, F. Di Renzo, Appl. Catal. A: Gen. 155
(
1997) 253.