G Model

CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS

Catalysis Today xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Influence of active phase structure of CoMo/Al2 O3 catalyst on the

selectivity of hydrodesulfurization and hydrodearomatization
W. Chen ∗ , X. Long, M. Li, H. Nie, D. Li
Research Institute of Petroleum Processing, SINOPEC, P.O. Box 914, Beijing 100083, PR China

a r t i c l e i n f o a b s t r a c t

Article history: The relationship between the morphology and the selectivity of the sulfide phase in the hydrodesulfur-
Received 2 June 2016 ization (HDS) of 4,6-dimethyldibenzothiophene (4,6-DMDBT) and hydrodearomatization (HDA) reaction
Received in revised form 13 August 2016 of 1-methylnaphthalene (1-MN) was studied. The morphology of the CoMo catalyst was tuned by increas-
Accepted 18 September 2016
ing the metal content and adding organic compounds. The concentration of CoMoS sites at the corners
Available online xxx
(C(CoMoSC ) and edges C(CoMoSE )) of the sulfide slabs were measured by X-ray photoelectron spec-
troscopy and high resolution transmission electron microscopy techniques. The corner and edge sites of
Keywords:
the sulfide phase account for the direct desulfurization (DDS) and hydrogenation desulfurization (HYDS)
Hydrodesulfurization
Hydrodearomatization routes in the HDS of 4,6-DMDBT, respectively. The hydrogen consumption analysis showed that the HYDS
Selectivity of 4,6-DMDBT and HDA of 1-MN both occur at the edge sites of the MoS2 slabs. The HYDS route in 4,6-
Morphology DMDBT HDS was more promoted than the 1-MN HDA reaction when the stacking was increased. The
4,6- DMDBT turnover frequency of the corner sites for the DDS reaction and edge sites for the HYDS and HDA reac-
Hydrogen consumption tions were correlated with the length and stacking of the sulfide phase. The results revealed how the
selectivities of the DDS, HYDS and HDA reactions as well as the hydrogen consumption can be tuned by
modifying the morphology of the sulfide phase.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Hydrodesulfurization (HDS) plays an important role in upgrad-
ing diesel fuel to an ultra-low sulfur level [1–3]. In the process, the
hetero atoms (S, N) are removed over the HDS catalyst in the pres-
ence of high pressure hydrogen and a large quantity of hydrogen
is continuously consumed by hydrodesulfurization (HDS), hydro-
denitrogenation (HDN) and hydrodearomatization (HDA) reactions
[4]. Therefore, it is essential to reduce the hydrogen consump-
tion in order to save cost and enhance the profits of the refinery.
Dibenzothiophene (DBT) and 4,6-dimethyldibenzothiophene (4,6-
DMDBT) are regarded as the most refractory molecules in the
ultra-deep HDS of diesel [5]. The sulfur atoms are mainly removed
through two pathways, named the direct desulfurization (DDS) and
hydrogenation desulfurization routes (HYDS). In the DDS route, the
sulfur is eliminated by direct scission of the C S bonds whereas the
prehydrogenation of the aromatic ring in the sulfur compounds is
required in the HYDS route [1]. Obviously, the HYDS route con-
sumes a higher amount of hydrogen than the DDS route. Besides
∗ Corresponding author at: Research Institute of Petroleum Processing, SINOPEC,
P.O. Box 914, 15th Department, Beijing 100083, PR China.
E-mail address: chenwb.ripp@sinopec.com (W. Chen).
HDS reactions, the HDN and aromatics saturation reactions also
consume hydrogen. Due to the difference in content, hydrogen con-
sumption in the HDS of diesel is mainly dependent on the HDS and
HDA reactions. In particular, the portion of refractory sulfur com-
pounds and aromatics in the diesel feedstock exhibits an increasing
trend since more fluid catalytic cracking (FCC) diesels fractions
have to be blended and upgraded [6]. Therefore, tuning the selec-
tivities of the catalyst in the HDS of refractory sulfur-containing
compounds and HDA reactions is important for reducing hydrogen
consumption.
The transformation of refractory sulfur compounds over HDS
catalyst has been extensively studied in the literature [7–18]. In
the removal of DBT, the DDS pathway is much more reactive than
the HYDS route [19–26]. The ratio of the DDS and HYDS selectiv-
ities (SDDS /SHYDS ) may vary in the range of 4:1–9:1 depending on
the metal type and reaction conditions. When the methyl substitu-
tions in the aromatic rings are located adjacent to the sulfur atom
of DBT, the DDS route is remarkably inhibited due to steric hin-
drance whereas the HYDS route is only slightly affected [19–25].
Consequently, the HYDS route turns out to be important in the
HDS of 4,6-DMDBT and the SDDS /SHYDS ratio can be decreased
to 1.3:1–0.5:1. Regarding the reaction mechanism, it is generally
accepted that the DDS reaction involves ␴ adsorption of the DBT
molecule via the sulfur atom, whereas HYDS proceeds through ␲

http://dx.doi.org/10.1016/j.cattod.2016.09.029
0920-5861/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
2 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx
adsorption of the reactant via the aromatic ring [23]. In the real
feedstock, the presence of H2 S and aromatics affects the selectiv-
ity of DDS and HYDS due to their competitive adsorptions on the
active sites [23,26–28]. The rate of the DDS route in the HDS of
DBT and 4,6-DMDBT can be markedly decreased by H2 S whereas
the HYDS pathway is less blocked [29]. Regarding the inhibition by
aromatics, Kim et al. reported that 1- methylnaphthalene (1-MN)
produced a more negative effect on the hydrogenation pathway
than on the hydrogenolysis route in the HDS of 4,6-DMDBT [30].
Farag et al. reported that naphthalene had more impact on the HYDS
than on the DDS route in the HDS of 4,6-DMDBT, leading to a slight
increase of the SDDS /SHYDS selectivity [31]. Even though the impact
of aromatics has been frequently studied, the selectivity ratio of
HDS and HDA (SHDS /SHDA ) has been less studied, particularly when
the content of aromatics was high.
To better tune the SDDS /SHYDS and SHDS /SHDA selectivities in the
HDS process, the structure and location of the active sites for spe-
cific reactions should be taken into consideration. The active phase
in sulfide catalysts is present as small hexagonal MoS2 or WS2 slabs
with the edges and corners decorated by the promoter atoms (Co
or Ni) [1,8,16]. The real structure of the catalytic sites involved in
hydrogenolysis and hydrogenation reactions has been heavily dis-
cussed for many years [1,7,8,16,32,33]. Topsøe et al. estimated the
fraction of edge sites in the catalyst and pointed out that an increas-
ing fraction of the atoms along the edges would be “corner atoms”
as the MoS2 particle size is decreased. The edges sites are more
active for HDS than the corner atoms [34,35]. Massoth found that
the sites at edges and corners may have different reactivity and
suggested that the HDS reaction occurred at vacancies on corner
sites and that the hydrogenation reaction took place at vacancies
on edge sites [36]. In the “rim-edge” model developed by Daage
and Chianelli, the unsupported MoS2 particle was regarded as a
stack of several disks in which the rim sites were located at the
top and bottom disks and the edge sites was associated with the
disks “sandwiched” between the top and bottom disks [37]. The
model indicated that the sulfur hydrogenolysis occurred on both
the rim and edge sites, whereas DBT hydrogenation was believed to
occur exclusively on the rim sites. In recent years, the Topsøe group
viewed a kind of metallic- like site (named BRIM site) in sulfide clus-
ter in scanning tunneling microscopy (STM) images [32,38,39]. The
BRIM sites are located adjacent to the edges and bind the sulfur-
containing compounds and catalyze the hydrogenation reaction
with the hydrogen activated at the neighboring edge sites on the
MoS2 and CoMoS structure [40,41]. The direct image of the adsorp-
tion of 4,6-DMDBT with STM at low temperatures showed that the
adsorption onto sulfur vacancies on MoS2 edges in the DDS route
was strongly sterical inhibited and certain corner sites in the MoS2
slab and the promoted CoMoS structures may directly facilitate the
adsorption [41].
According to these arguments, different reactions may take
place on specific position of the MoS2 slabs. Therefore, the vari-
ation of the morphology of the MoS2 slabs may alter the selectivity
of the catalyst. Silva et al. tested the SHDS /SHDA selectivity in the
HDS of DBT and hydrogenation of 1- MN (20 wt%) over a series
of Mo/Al2 O3 catalysts in the presence of isoquinoline [42]. It was
found that the HDA activity was gradually increased whereas the
HDS activity remained constant as the Mo loading in the catalyst
increased. The HDA activity was related to the number of edge or
edge + corner sites whereas the HDS was not related. Hensen et al.
indicated that the selectivities for HDS and hydrogenation can be
tuned by the morphology of MoS2 phase and the choice of sup-
port. The high hydrogenation rate constant for a Mo/C catalyst was
ascribed to the large number of corner sites [43]. Liu et al. [44] found
the hydrogenation selectivity of a NiMo catalyst increased linearly
with the stacking number. The high stacking degree of MoS2 slabs
was beneficial for the flat adsorption and reaction of DBT. Li et al.
Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
reported that the morphology of MoS2 slabs can be modified on
different supports. The HDS selectivity in the hydrogenation of thio-
phene and 1-hexene correlated linearly with the length of the slab,
indicating that the corner sites rather than the edge sites may favor
hydrogenation [45].
These results indicate that the selectivity of hydrogenolysis and
hydrogenation correlates with the morphology of the active sites,
but the corresponding sites are still under debate. This inevitably
causes ambiguity for tuning the SDDS /SHYDS and SHDS /SHDA selec-
tivity in diesel upgrading. In order to clarify the effect of the
morphology structure on the selectivity of HDS catalyst in HDS
and HDA reactions, two solutions were used to in the present work
modify the morphology of the sulfide phase including varying the
metal contents [42,44,46–48] and introducing ethylene glycol (GL)
[49] and citric acid (CA) [50–56]. The SDDS /SHYDS and SHDS /SHDA
selectivities were evaluated in the simultaneous HDS of 4,6-DMDBT
and HDA of 1-MN. The relationship between the structure and the
catalytic functions was studied.
2. Experimental
2.1. Catalyst preparation
A series of CoMo catalysts with different metal loadings was
prepared by the classical incipient wetness impregnation method.
The basic cobalt carbonate and molybdenum oxide precursors
were sequentially added to the proper amount of phosphoric acid
solution in the presence of ethylene glycol (abbreviated as GL).
The solution was heated under stirring until the precursors were
completely dissolved. Afterwards, the solution was mixed with
extruded alumina under quick shaking. The impregnated catalyst
was dried at 120 ◦ C for 3 h. The obtained oxidic catalysts with dif-
ferent metal contents were nominated as CoMoGL-1, CoMoGL-2,
CoMoGL-3 and CoMoGL-4, respectively. Following the same prepa-
ration procedure, catalysts without addition of chelating agent and
containing citric acid (CA) were prepared and named CoMoP and
CoMoCA, respectively.
The contents of the metal components in the catalysts are dis-
played in Table 1. In order to calculate the volume concentration of
the metals in the catalyst, around 5 mL of support with 40–60 mesh
particle size was packed in a 5 mL cylinder and the volume of the
support was determined after sufficient densification of the parti-
cles by knocking the cylinder. The packing density of the support
was calculated by dividing its weight by the measured volume.
The packing density of the support (dAl2 O3 , g/mL) was the average
value of three measurements. The densities of all catalysts were
also measured by the same method.
The volume molar concentration of Mo (C(Mo), mol/L) and Co
(C(Co), mol/L) atoms was determined by the equation:
[MoO3 ] [Al2 O3 ] [CoO] [Al2 O3 ]
C (Mo) = ÷ × 1000, C (Co) = ÷ × 1000 (1)
MMoO3 dAl2 O3 MCoO dAl2 O3
where [MoO3 ], [CoO] and [Al2 O3 ] are the weight contents of
MoO3 , CoO and Al2 O3 in the catalysts, MMoO3 and MCoO are the
molar mass of MoO3 and CoO. The detailed data are listed in Table 1.
2.2. Catalyst characterization
2.2.1. Preparation of sulfide catalyst
Prior to the XPS analysis, the sulfide catalyst was prepared under
H2 S/H2 atmosphere. The oxidic catalyst (40–60 mesh) was installed
in the homemade sulfidation equipment. The sulfidation was car-
ried out at 400 ◦ C for 3 h (ramp rate: 10 ◦ C/min) under H2 S (2%)/H2
gas. N2 gas was switched to flush the sulfide sample for 30 min
when the temperature was decreased to 120 ◦ C. After the temper-
ature was decreased to room temperature, the sample was sealed
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 3

Table 1
Composition and density of prepared catalysts and support.
a a
Sample MoO3 CoO P2 O5 Co/Mo GL(CA)/Co C(Mo) C(Co) density
% % % mol/L mol/L g/mL

CoMoGl-1 16.0 3.36 3 0.403 1.0 0.794 0.320 0.755

CoMoGL-2 18.0 3.78 3 0.403 1.0 0.922 0.372 0.804
CoMoGL-3 20.0 4.2 3 0.403 1.0 1.058 0.427 0.812
CoMoGL-4 22.0 4.62 3 0.403 1.0 1.204 0.486 0.834
CoMoP 18.0 3.78 3 0.403 1.0 0.922 0.372 0.775
CoMoCA 18.0 3.78 3 0.403 1.0 0.922 0.372 0.843
Al2 O3 b – – – – – 0.554
a
Atomic ratio.
b
For the support, specific surface area = 240 m2 /g, specific pore volume = 0.66 mL/g, average pore diameter = 11 nm.

in the reactor and transported to the glove box in which the O2 where AMo4+ , AMo5+ , and AMo6+ are the area of each species fitted
and H2 O contents were below 0.1 ppm. The sulfide sample was from the Mo3d XPS envelope. [Mo5+ ] and [Mo6+ ] can be calculated
put under into the cyclohexane for the analysis of structure and by the same way.
morphology. The volume molar concentrations of the MoS2 species (C(MoS2 ))
were determined by the equation:
2.2.2. X-ray photoelectron spectroscopy (XPS)
The sulfide sample was transferred to the chamber of a Thermo C(MoS2 ) = C(Mo) × [Mo4+ ]/100 (5)
Fischer-VG ESCALAB 250L XPS spectrometer under Ar atmosphere.
After the sample was mounted, the chamber was evacuated. The where C(Mo) is the mole concentration of Mo atom in the oxidic
final pressure for analysis was about 1 × 10−7 Pa. The XPS spectra catalyst (mol/L).
were acquired with a 150 W monochromatized Al K␣ source with The relative fractions of CoMoS, Co9 S8 and Co2+ in the total Co
0.01 eV step−1 . The spectra of Al2p, Mo3d, Co2p, S2s, S2p, O1s, C1s species ([CoMoS], [C9 S8 ] and [Co2+ ]) were calculated by the equa-
and P2p were recorded. The obtained Mo3d, Co2p and S2p spectra tion:
were referenced to the Al2p line at 74.7 eV. The envelopes of various ACoMoS
elements were decomposed using an iterative least-squares com- [CoMoS](%) = × 100 (6)
ACoMoS + ACo9 S8 + ACo2+
puter program XPSPEAK41. The Shirley type baseline was applied
for all envelopes. Firstly, the S2p spectra were decomposed into where ACoMoS , ACo9 S8 and ACo2+ correspond to their fitted area in the
double bands at 161.9 and 163.2 eV corresponding to S2− (and a
Co2p3/2 spectra. [C9 S8 ] and [Co2+ ] can be calculated by the same
small part of terminal S2 2− ) and oxysulfide (S-O) species [57]. The
way.
binding energies and peak areas of S2s were constrained by the
The volume molar concentration of CoMoS species (C(CoMoS))
following equations:
were determined by the equation:
BE(S2s) = BE(S2p3/2 ) + 64.3, Area(S2s) = Area(S2p) × 0.66 (2)
C(CoMoS) = C(Co) × [CoMoS]/100 (7)
The spin-orbit splittings of Mo4+ was fixed as 3.15 eV and Mo5+
and Mo6+ fixed as 3.2 eV. Their area ratios were kept to theoretical where C(Co) is the volume molar concentration of Co in the oxidic
values of 1.5 [57]. The full widths at half maximum (FWHM) of the catalyst (mol/L).
Mo(3d5/2 ) and Mo(3d3/2 ) peaks were assumed to be equal and the The mole ratio of CoMoS and MoS2 species (fCoMoS/MoS2 ) in the
FWHM ratios for the Mo4+ , Mo5+ , Mo6+ peaks was restricted as 1: sulfide catalysts was calculated by the equation:
1.2: 1.4. The binding energies for Mo5+ (3d5/2 ) and Mo6+ (3d5/2 ) were
C(CoMoS)
fixed at 230.6 and 232.8 eV [58–60]. For the Co2p3/2 envelopes, fCoMoS/MoS2 = (8)
C(MoS2 )
the species of CoMoS, Co9 S8 , Co2+ and shake-up peaks were fit-
ted. The binding energies for Co9 S8 , Co2+ and shake-up peaks were
restricted at 778.2, 781.8 and 785 eV, respectively. The FWHM 2.2.3. High resolution transmission electron microscopy (HRTEM)
ratios for the CoMoS, Co9 S8 , Co2+ and shake-up peaks were fixed as The morphology of the sulfide phase was determined with a FEI
1:0.9:1.2:1.5 [57]. All curves were taken as 70% Gaussian and 30% TECNAI G2 F20S STWIN HRTEM apparatus. The sulfided sample was
Lorentzian. The detailed decomposition specifications are given in deposited on a carbon-coated copper grid. Around 10 typical micro-
the reference [57]. graphs for each sample were made in high-resolution mode. The
The surface atomic concentration for each species was calcu- length and stacking of the MoS2 slabs were determined by measur-
lated according to the equation: ing 700–800 slabs. The length measurement was performed with
the software Image Pro Plus 7.0 program and the stacking was man-
Ai /SFi
C(i)S =  × 100 (3) ually counted. The average slab length (L) and stacking number (N)
A /SFi
i=1...n i were calculated
 according to thefirst moment of the distribution
n n
ni Li ni Ni
where C(i)S is the surface atomic content of species i, Ai is the mea- given by L = i=1
n and N = i=1
n [61].
sured spectra area of species i, SFi is the sensitivity factor of element i=1
ni n
i=1 i
i (provide by the XPS apparatus manufacture). Based on the method developed by Kastzelan [48], the number
The relative percentages of the Mo4+ , Mo5+ and Mo6+ species of edge sites and corner sites in the slab can be calculated [45,62,63].
([Mo4+ ], [Mo5+ ] and [Mo6+ ]) in the total Mo component were cal- Assuming that all slabs have perfect hexagonal shape, the number
culated by the equation: of Mo in one side of a slab (Moside ) can be calculated using the
equation:
AMo4+
[Mo4+ ](%) = × 100 (4)
AMo4+ + AMo5+ + AMo6+ Moside = 10 × L̄/6.4 + 0.5 (9)

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
4 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx

The number of Mo sites in the edge (MoE ), in the corner (MoC ) The rate constants of HDS (kHDS ) of 4,6-DMDBT, direct desul-
and total Mo site (MoT ) of one slab were determined by the equa- furization (kDDS ) and hydrogenation desulfurization (kHYDS ) routes
tions: were calculated based on pseudo-first-order reaction by the equa-
tions:
MoE = (6 × Moside − 12) × N̄ (10)
F4,6-DMDBT
MoC = 6 × N̄ (11) kHDS = − × ln(1 − x(4, 6-DMDBT)) (17)
Vcat
2
MoT = (3 × Moside − 3 × Moside + 1) × N̄ (12)
kDDS = kHDS × SDDS (18)
Mo
Therefore, the fraction of Mo sites in the edge ( MoE ), the fraction
T
Mo
of Mo sites in the corner ( MoC ) and the ratio of Mo sites in the corner kHYDS = kHDS × SHYDS (19)
T
Mo
and in the edge ( MoC ) can be calculated.
E
where kHDS is the rate constant of 4,6-DMDBT HDS (mol h−1 L−1 ),
The volume molar concentration of edge sites (C (MoE )) and cor- F4,6-DMDBT is the reactant molar flow of 4,6-DMDBT (mol h−1 ), Vcat
ner sites (C(MoC )) in the catalyst was determined by the equation: (L) is the volume of the catalyst (condensed 40–60 mesh particles),
MoE MoC x(4,6-DMDBT) is the conversion of 4,6-DMDBT, kDDS and kHYD are
C (MoE ) = × C (MoS2 ) , C (MoC ) = × C (MoS2 ) (13) the rate constants of 4,6-DMDBT converted through DDS and HYDS
MoT MoT
pathways, respectively.
The promotion factor in the slab was calculated by the equation: In order to calculate the amount of sulfur and hydrogen con-
C(CoMoS) sumption in the reaction, the mass balance of decane before and
fprom = (14) after the reaction was considered and expressed as the equation:
C(MoE ) + C(MoC )

2.3. Evaluation of the catalytic activity and selectivity mf × wf (dec) = mp × wp (dec) (20)

where mf and mp are the mass of feed and product, wf (dec) and
The HDS and hydrogenation activity for all catalysts were eval-
wp (dec) are the weight content of decane in the feed and product.
uated in a fixed bed high-pressure flow micro-reactor under the
Therefore, the amount of removed sulfur per mass of feed in the
same condition. The proper amount of oxidic catalyst (40–60 mesh)
reaction can be determined by the equations:
was installed in the constant temperature zone of the reactor.
Before the catalytic test, the catalyst was sulfided with a solution of  wp (i)   wp (i) 
mp × ( M(i)
) wf (dec)
i=1...n
5 wt% CS2 in cyclohexane (flow rate: 0.4 mL/min) under 6.4 MPa of Srem (HDS) = = × (21)
mf wp (dec) M (i)
hydrogen (flow rate: 400 mL/min). The temperature was increased i=1...n
from room temperature to 360 ◦ C (5 ◦ C/min) and maintained for 3 h.
After the sulfidation, the reaction feed was switched to the reactor Srem (DDS) = Srem (HDS) × SDDS (22)
for the catalytic test. The reactant feed consisted of 0.45 wt% 4,6-
Srem (HYDS) = Srem (HDS) × SHYDS (23)
DMDBT, 40 wt% 1- methylnaphthalene (1-MN), 1.2 wt% CS2 and a
balanced amount of decane. The reaction conditions were fixed at where Srem (HDS), Srem (DDS) and Srem (HYDS) are the amount
330 ◦ C, 6.4 MPa, 12 h−1 LHSV and 500 H2 -to-oil ratio. After 3 h time- of total removed sulfur, sulfur removed by DDS route and removed
on-stream, stable activity was obtained and the HDS products were by HYDS route (mol/g), respectively; SDDS and SHYDS are DDS and
condensed and collected. The mass composition of the products HYDS selectivity.
were analyzed by the area correction normalization method on an The amount of consumed hydrogen per mass of feed in the 4,6-
Agilent 7890 gas chromatograph (GC) equipped with a capillary col- DMDBT HDS reaction can be calculated by the equation:
umn of HP-1 methyl siloxane and a flame ionization detector (FID).

The products of the HDS of 4,6-DMDBT and HDA of 1-MN were wf (dec)  wp (i )  
identified by gas chromatography, mass spectrometry (MS) anal- H2 (HDS) = × × H2 i (24)
wp (dec) i =1...n M (i )
ysis and compared with authentic samples. The reproducibility of
the catalytic tests was better than 3%. where H2 (HDS) (mol/g) is the amount of consumed hydrogen per
The conversion of 4,6-DMDBT (x(4,6-DMDBT)) and selectivity of mass of feed by transforming 4,6-DMDBT, wp (i’) is the mass con-
product i from 4,6-DMDBT (Si ) were calculated by equations: tent of the products from 4,6-DMDBT including the hydrogenated
   but undesulfurized products, H2 (i’) is the number of H2 molecules
wp (i)
i=1...n Mi consumed in the transformation of one molecule of 4,6-DMDBT to
x (4, 6-DMDBT) =    × 100 (15) product i’. No undesulfurized products were found in the present
wp (4,6-DMDBT) wp (i)
+
M 4,6-DMDBT i=1...n Mi work.
wp (i)
H2 (DDS) (mol/g) and H2 (HYDS) (mol/g), the amount of con-
M(i) sumed hydrogen by the DDS and HYDS routes, can be calculated
Si =  wp (i)
× 100 (16)
in similar manner as Eq. (24) by considering the product of each
i=1...n
( M(i)
)
route.
where wp (4,6-DMDBT) is the mass content of 4,6-DMDBT in the The products of the 1-MN hydrogenation consisted of various
product, wp (i) is the mass content of the products from the 4,6- kinds of hydrogenated compounds, ring opening and a very small
DMDBT conversion, M4,6-DMDBT and M(i) are the molar mass of 4,6- amount of demethylated and cracked compounds in agreement
DMDBT and products i from the HDS of 4,6-DMDBT. with other reports [64,65]. The main hydrogenated compounds
In the HDS products, 3,3 -dimethlybiphenyl (DMBiPh) were 1-methyltetralin, 5-methyltetralin, and methyldecalins. The
was transformed from the DDS pathway, whereas alkylbenzenes compounds were probably formed by the ring open-
3-(3 -methylcyclohexyl) – toluene(DMCHT) and 3,3 - ing of methyltetralin after isomerization [64,66]. In order to identify
dimethyldicyclohexane (DMDCH) were produced via the HYDS the hydrogenation activity of the catalysts, the products were clas-
route. The selectivity of the DDS (SDDS ) and HYDS (SHYDS ) path- sified into four groups: methyltetralin and their isomers (MT),
ways were calculated by summing the total selectivities of the methyldecalin and their isomers (MD) and ring-opening products
corresponding products. (OP) as well as demethylated and cracked products (CR).

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 5

The conversion of 1-MN (x(1-MN)) and selectivity of product j Mo4+

from 1-MN were calculated by the equations: Mo3d Co2p3/2
  
wp (j)
j=1...n M(i) S2s CoMoS
x (1-MN) =    × 100 (25)
wp (1-MN) wp (j)
M1-MN
+ j=1...n M(j)

wp (j) CoMoGL-1
Sj = 
M(j)
  × 100 (26)
wp (j)
j=1...n M(j)
CoMoGL-2

where wp (1-MN) is the mass content of 1-MN in the product, wp (j)

is the mass content of the products converted from 1-MN, M1-MN CoMoGL-3
and M(j) are the molar mass of 1-MN and products j from 1-MN
conversion.
The conversion of 1-MN obeys first-order kinetics [67] and the CoMoGL-4
rate constant of 1-MN hydrogenation were determined by the
equation:
CoMoP
F1-MN
kHDA =− × ln(1 − x(1-MN)) (27)
Vcat
CoMoCA
where kHDA is the rate constant of the pseudo-first-order of the 1-
MN HDA reaction and F1-MN is the reactant molar flow of 1-MN 790 780 770
240 234 228 222
(mol h−1 ).
Bingding Energy (eV) Bind ing Energy (eV)
The rate constants for the products of MT (kHDA (MT)), MD
(kHDA (MD)) OP (kHDA (OP)) and CR (kHDA (CR)) were calculated by
Fig. 1. XPS spectra of Mo3d and Co2p3/2 in various catalysts.
kHDA multiplying the corresponding selectivities. An example is
given in the equation:

kHDA (MT) = kHDA × SHDA (MT)
where SHDA (MT), SHDA (MD), SHDA (OP) and SHDA (CR) are the selec-
tivites of MT, MD, OP and CR products, respectively. The selectivities
of every group were calculated by summing the selectivities of the
corresponding the products.
Following the methodology of Eq. (24), the consumed hydro-
gen for the hydrogenation reaction in the HDA of 1-MN can be
determines by the equation:

wf (dec)  wp (j )  
H2 (HDA) = × × H2 j
wp (dec) j=1...n M (j )
where H2 (HDA) is the total consumed hydrogen per mass of feed
for saturation of 1-MN and subsequently hydrogenation of its prod-
ucts, wp (j’) is the mass content of product j’, H2 (j’) is the number of
hydrogen molecules needed for the transformation of one molecule
1-MN towards product j’, which only include MT, MD and OP prod-
ucts.
3. Results
3.1. XPS analysis
The properties and concentrations of various surface species
in the sulfide catalysts were determined by XPS. S2p, Mo3d and
Co2p3/2 profiles of all catalysts are presented in Fig. 1. In the Mo3d
envelope, the strong doublets located at 229.0 and 232.4 eV are
characteristics of Mo4+ in the MoS2 phase. The overlapped band at
226.2 eV is attributed to S2s core-levels. In order to precisely deter-
mine the band intensities of the Mo species, the band locations
and intensities of S2s were measured from S2p spectra based on
their relationships shown in the experimental part [57]. The Mo3d
spectra were decomposed into Mo3d (Mo4+ ), Mo3d (Mo5+ ), Mo3d
(Mo6+ ) and S2s species, which correspond to sulfide Mo, oxysulfide
Mo, oxidic Mo and sulfur species. The example of fitting S2p and
Mo3d profiles of the sulfide CoMoGL-2 catalyst is shown in Fig. 2
and the binding energies and FWHM of the various species are listed
in Table 2. The binding energies for all species are in line with other
(28) reports [57–60,68]. The binding energies of various species were
similar for all catalysts suggesting that the chemical surroundings
of the sulfide MoS2 phase remained constant with increasing metal
loading and in the presence of GL and CA.
The relative amounts of the Mo4+ , Mo5+ and Mo6+ components
were calculated according to the spectra decomposition, as shown
in Table 3. From CoMoGL-1 to CoMoGL-4 catalysts, the relative
amount of Mo4+ was around 90% and only small amount of Mo5+
and Mo6+ were observed. Apparently, the MoO3 phase was largely
transformed into sulfide form during the sulfidation. As shown in
(29) Table 3, the percentage of Mo4+ and S/(Co + Mo) ratio indicate that
the sulfidation degree of the Mo species did not change suggesting
that the metal-support interactions remained constant because of
the same GL/Co atomic ratio present in the catalysts. For the pure
CoMoP catalyst, the fraction of Mo4+ was slightly lower than in the
CoMoGL-2 catalyst and suggested that the metal-support interac-
tion was slightly stronger in the absence of GL. In the presence of CA,
the fraction of Mo4+ and Mo5+ appears slightly higher than in the
CoMoP catalyst. However, the variations of the sulfidation degree
in the presence of GL and CA were very limited, in line with the
similar S/(Mo + Co) ratio observed for all catalysts.
As seen in Fig. 2, the envelopes of the Co species show an appar-
ent peak at 779.2 eV, which is attributed to the formation of CoMoS
structure [57–60]. The decomposition of the Co2p3/2 spectra of the
CoMoGL-2 catalyst is displayed as an example in Fig. 2. The bind-
ing energies and FWHM as well as the relative percentage of every
component are listed in Tables 2 and 3. Table 3 shows that the
relative amount of CoMoS phase did not change with increasing
metal content in agreement with the sulfidation. Therefore, the use
of the same GL/Co ratio had a similar effect on the formation of
the CoMoS phase. Around 71.5% of the Co atoms are presented as
CoMoS species and only a small part of Co are in form of Co9 S8 . The
high portion of Co atoms that formed CoMoS structures is likely
due to two reasons. On the one hand, the addition of GL provides
very good protection to the Co atoms and enhances the formation
of the CoMoS structure. On the other hand, the Co/(Co + Mo) ratio
(0.287) in these catalysts is slightly lower so that the trend towards
forming Co9 S8 species may decrease [1]. Compared to the CoMoGL-

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
6 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx

S2p Mo3d Co2p

Mo4+ CoMoS
S2- and terminal S22-

Co2+

Mo5+ shakeup
Bridge S22- and Co9S8
oxysulfide Mo6+ S2s

172 170 168 166 164 162 160 158 240 234 228 222 790 785 780 775 770
Bing Energy (eV) Binding Energy (eV) Binding Energy (eV)

Fig. 2. Decomposition of S2p, Mo3d and Co2p3/2 XPS spectra.

Table 2
Binding energies of various species in sulfide catalysts.

Sample Binding energies (eV) FWHM

4+ 2− 2−
Mo3d5/2 (Mo ) Co2p3/2 (CoMoS) S2s (S ) S2p3/2 (S ) P2p Mo3d5/2 (Mo4+ ) Co2p3/2 (CoMoS) S2s S2p3/2 (S2− ) P2p

CoMoGL-1 229.0 779.2 226.1 161.8 134.4 1.16 1.74 2.24 1.13 1.73
CoMoGL-2 229.1 779.3 226.2 161.9 134.2 1.15 1.83 2.42 1.19 1.52
CoMoGL-3 229.1 779.0 226.2 161.9 134.3 1.16 1.98 2.24 1.18 1.71
CoMoGL-4 229.0 779.0 226.2 161.9 134.5 1.04 1.76 2.08 1.01 1.08
CoMoP 229.1 779.2 226.2 161.9 134.3 1.10 1.99 2.31 1.12 1.04
CoMoCA 229.0 779.2 226.1 161.8 134.4 1.07 1.79 2.17 1.04 1.40

Table 3
Surface atomic ratio, relative amount and concentration of various species detected by XPS.

Sample Atomic ratio Relative amount (%) Concentration (mol/L)a fCoMoS/MoS2

C(Mo)s C(Co)s C(S)s
C(Al)S C(Al)S C(Mo)s +C(Co)s
[Mo4+ ] [Mo5+ ] [Mo6+ ] [CoMoS] [Co9 S8 ] [Co2+ ] C(MoS2 ) C(CoMoS)

CoMoGL-1 0.048 0.016 2.2 89.8 5.9 4.3 71.5 5.9 22.6 0.713 0.229 0.321
CoMoGL-2 0.052 0.021 2.4 91.2 5.9 2.9 71.5 6.4 22.1 0.840 0.266 0.316
CoMoGL-3 0.072 0.026 2.2 88.6 6.7 4.7 70.0 7.8 22.2 0.937 0.299 0.319
CoMoGL-4 0.089 0.030 2.1 91.9 3.8 4.3 71.6 5.3 23.1 1.106 0.348 0.314
CoMoP 0.052 0.020 2.1 87.3 7.2 5.5 64.0 8.5 27.5 0.805 0.238 0.296
CoMoCA 0.066 0.029 2.0 88.8 5.9 5.3 73.5 1.5 25.0 0.818 0.273 0.334
a
The concentration (mol/L) means the mole of MoS2 and CoMoS species in one liter of sufficiently packed catalysts with 40–60 mesh particle size.

2 catalyst, the percentage of CoMoS phase was around 7.5% lower
and the formation of Co9 S8 slightly higher without the presence of
GL (Table 1). This reflects the beneficial role of GL on the formation
of the CoMoS active structure. In the presence of CA, the formation
of CoMoS atoms was also very high and the amount of Co9 S8 species
was strongly reduced. Furthermore, the effect of CA was stronger
than of GL for transforming Co to CoMoS structures.
The deconvolution of the XPS spectra allows to calculate the
concentration of MoS2 and CoMoS species when the total Mo and
Co concentrations are known. The results are shown in Table 3. The
concentrations of the MoS2 and CoMoS phase gradually increase
with metal loading. The presence of GL and CA does not obviously
modify the amount of MoS2 species but increases the concentration
of CoMoS phase in the sulfide catalysts. The mole ratios of CoMoS
and MoS2 (fCoMoS/MoS2 ) over the series of catalysts containing GL are
about 0.320, which are lower than the real ratio of Co and Mo atoms
(0.403) in the oxidic catalysts. This is because the percentage of Co
atoms transformed to CoMoS species is lower than the portion of
Mo atoms sulfided to MoS2 phase. The ratio of fCoMoS/MoS2 with the
presence of GL and CA ranks as CoMoP < CoMoGL–2 < CoMoCA.
Table 3 shows that the Mo/Al and Co/Al surface atom ratios (from
CoMoGL-1 to CoMoGL-4) increase linearly with the metal content,
suggesting that the dispersion of the Mo species remains similar
due to the same GL/Co ratio. The Mo/Al and Co/Al atom ratios in
the CoMoP catalyst are similar to those in the CoMoGL-2 catalyst.
It suggests that the dispersion of the Mo and Co species did not
change with GL addition. The Mo/Al and Co/Al ratio of the CoMoCA
catalyst are both higher than that of the CoMoP catalyst. Therefore,
CA addition promoted the dispersion of Mo and Co phase onto the
surface of catalyst.
3.2. HRTEM analysis
The morphology of the sulfided catalysts was studied by HRTEM
measurements [48,61,62]. Fig. 3 shows representative micrographs
of all catalysts and Fig. 4 displays the length and slab distribution of
MoS2 slabs. The length of the majority of the MoS2 particles varied
in the range 2–6 nm and the slabs with the highest frequency were
distributed at 3–4 nm. With increasing metal loading in the catalyst,
the length distribution gradually shifted to higher length, indicating
that larger MoS2 slabs were formed. Regarding the CoMoP cata-
lyst, the length of MoS2 slabs with highest frequency was longer
than that of the CoMoGL-2 catalyst. Apparently, the presence of
GL favors the formation of shorter slabs, likely because GL reduces
the metal-support interaction [49]. In a similar manner, the profile
of the length distribution of the CoMoCA catalyst also shifted to
smaller length due to the change of metal-support interaction with
the addition of CA. Regarding the stacking, 1–4 layers of MoS2 were
mainly observed in all catalysts (Fig. 4). In general, the stacking
number did not change markedly for the series of catalyst in which

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 7

Fig. 3. HRTEM micrographs of the sulfided catalysts.

25 CoMoG L-1 50 CoMoGL-1

CoMoG L-2 CoMoGL-2
20 CoMoG L-3 40 CoMoGL-3
CoMoG L-4 CoMoGL-4
Frequency (%)

Frequency (%)

15 CoMoP 30 CoMoP
CoMoCA CoMoCA
10 20

5 10

0 0
0 2 4 6 8 1 2 3 4 5 6
Length (nm) Stacking

Fig. 4. Length and stacking distribution of the MoS2 slabs in various catalysts.

GL was present. The MoS2 slabs of the CoMoP catalyst exhibited a
higher frequency of one and two layers than the other catalysts. The
highest frequency of MoS2 slabs with 3 and 4 layers was observed
in the CoMoCA catalysts. This indicates that the stacking of MoS2
slabs is increased with the presence of GL and CA. The variation of
MoS2 slabs and stacking is in line with other reports [50,51,54].
The average length and stacking of MoS2 slabs in the sulfide
catalysts are compiled in Table 4. The average slab length grad-
ually increased from 3.2 to 4.3 nm with increasing metal loading,
whereas the average stacking variation was very limited (2.2 ± 0.1).
The size and layers of the MoS2 particles changed with the metal
loading in line with other work [42]. Regarding the catalysts with
the same metal loading, the average length of the MoS2 slabs ranked
as CoMoP > CoMoGL–2 > CoMoCA, whereas the average stacking
morphology properties are tuned by the increasing metal loading
and the introduction of GL and CA in the catalysts in agreement
with the expectation.
The calculated numbers of Mo atoms in the corner and edge sites
Mo
are listed in Table 4. The fractions of edge and corner sites ( MoE and
T
MoC
)
varied as a function of the average length of the slab. Longer
MoT
slabs have a lower fraction of edge and corner sites. The ratio of cor-
Mo
ner and edge sites ( MoC ) also monotonously decreased with the slab
E
length. Therefore, the increase of metal loading and introduction GL
and CA modified the number of corner and edge sites as well as their
ratio. Since the concentration of MoS2 has been detected by XPS
(Table 3), the total concentration of corner and edge sites as well as
Mo Mo
their ratio can be calculated from MoE and MoC . Table 4 shows that
T T

showed the reverse order. From the above, it can be seen that the the concentration of edge sites (C(MoE )) increased with increasing

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
8 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx

Table 4
Morphology properties of the active phase in sulfided catalysts.
MoE MoC MoC
Sample L̄(nm) N̄ Moside MoE MoC MoT MoT MoT MoE
C(MoE ) (mol/L) C(MoC ) (mol/L) fprom C(CoMoSE ) C(CoMoSC )
(mol/L) (mol/L)

CoMoGL-1 3.2 2.2 5.5 46.2 13.2 165.6 0.279 0.080 0.286 0.199 0.057 0.90 0.178 0.051
CoMoGL-2 3.6 2.1 6.1 52.0 12.6 199.9 0.260 0.063 0.242 0.219 0.053 0.98 0.214 0.052
CoMoGL-3 3.9 2.3 6.6 63.4 13.8 256.8 0.247 0.054 0.218 0.231 0.050 1.06 0.238a 0.052a
CoMoGL-4 4.3 2.3 7.2 72.0 13.8 312.1 0.231 0.044 0.192 0.255 0.049 1.14 0.274a 0.052 a
CoMoP 3.9 1.8 6.6 49.6 10.8 201.0 0.247 0.054 0.218 0.199 0.043 0.98 0.195 0.043
CoMoCA 3.5 2.6 6.0 61.9 15.6 233.9 0.265 0.067 0.252 0.217 0.055 1.01 0.217 0.055

L̄: average slab length;N̄: average stacking number; Moside : number of Mo in one side of a slab; MoE : number of Mo sites in the edge; MoC : number of Mo sites in the corner
site; MoT : total Mo atoms;
MoE Mo Mo
Mo
: fraction of Mo sites in the edge; MoC : fraction of Mo sites in the corner; MoC : ratio of Mo sites in the corner and edge; C(MoE ): volume molar concentration of edge sites
T T E
in the sulfide catalyst; C(MoC ): volume molar concentration of corner sites in the sulfide catalyst; fprom : promotion factor in the slab; C (CoMoSE ): volume molar concentration
of CoMoS sites in the edge; C (CoMoSC ): volume molar concentration of CoMoS sites in the corner.
a
Corrected concentration: calculated by (C (MoC ) + C(CoMoSC ))/2 and (C(MoE ) + C (CoMoSE ))/2.

the metal content, even though the fraction of edge sites in the slab
is decreasing. However, the concentration of corner sites shows a
reverse relationship with the metal loading. This indicates that the
increase in the MoS2 concentration is higher than the decrease of
the edge sites concentration but lower than that of the corner sites.
The increase of the MoS2 concentration by 55% from CoMoGL-1 to
CoMoGL-4 leads to an enhancement of total number of edge sites
by 28% and of the total number of corner sites by −14%. Table 4
shows that the concentration of total edge sites and corner sites
of the CoMoP catalyst are enhanced by ∼10% and ∼25% with the
addition of GL and CA, respectively.
It is widely accepted that the Co atoms are located at the periph-
ery of the MoS2 slab and form the Co-Mo-S structure in the sulfide
CoMo catalyst, which is very active in HDS reaction [1]. Therefore,
the fraction of promotion (fprom ) was calculated based on the ratio
of the concentration of CoMoS species and the total number of edge
and corner sites. As shown in Table 4, the promotion factor linearly
improved from 0.90 to 1.14 when the metal loading increased from
CoMoGL-1 to CoMoGL-4. It suggests that the extent of Co occupy-
ing the periphery of MoS2 slabs varied from partial promotion until
full promotion.
3.3. HDS and HDA activity
To simulate the HDS of diesel, the feed for the HDS reaction
used in the present work contained 4,6-DMDBT and a large amount
of 1-MN (40 wt%) as well as 1.2 wt% CS2 . During reaction, CS2
rapidly decomposes and releases H2 S to generate a similar reac-
tion atmosphere as in the HDS in a real diesel feedstock [69–71].
The experimental conditions were chosen to reach a high HDS con-
version, to reach near complete removal of sulfur. The conversion
and product distributions of the HDS of 4,6-DMDBT and the HDA
of 1-MN were shown in Table 5.
Fig. 5 displays the activity and selectivity for HDS of 4,6-DMDBT
over the catalysts containing GL. It shows that the HDS rate con-
stant increases linearly with MoO3 content and that the maximum
enhancement is around ∼13%. However, the activities of the DDS
and HYDS routes varied in a different way. The rate constant of
the HYDS pathway monotonously enhanced up to ∼18% whereas
the rate of the DDS route slightly decreased with increasing metal
loading. As a result, the selectivity of the HYDS pathway (SHYDS )
increased by 5% whereas the DDS selectivity reduced by 26% and
the ratio of the DDS and HYDS selectivities (SDDS /SHYDS ) gradually
decreased with increasing metal concentration (Fig. 5). The DDS
selectivity was lower than that reported in the HDS of pure 4,6-
DMDBT [23,72], mostly due to the presence of H2 S that acted as an
inhibitor in the DDS route [23,26].
The CoMoP catalyst exhibited a lower HDS activity than the
CoMoGL and CoMoCA catalysts (Fig. 6). The presence of CA signifi-
cantly enhanced the HYD (+126%), DDS (+150%) and HYDS (+122%)
activities relative to that of the CoMoP catalyst. The beneficial effect
of GL is lower than of CA. The selectivity of SHYDS slightly deceased
in the presence of GL and CA, while the DDS selectivity showed
the reverse trend. The variation led to a gradual increase of the
SDDS /SHYDS selectivity with the addition of GL and CA.
The conversion of 1-MN yielded a large number of products
from hydrogenation, ring-opening, transalkylation, dealkylation
and cracking. The reaction network of the 1- MN hydrogenation is
described in the literature [64,73–75]. The activity and selectivity
of 1-MN HDA over the GL modified catalysts are shown in Fig. 7.
More than 90% conversion (from 93.7% to 96.2%) was obtained
on these catalysts. The rate constant of 1-MN transformation (kHDA )
linearly increased up to a maximum enhancement of 18% with
increasing metal loading. The rates of kHDA (MT), kHDA (MD) and
kHDA (OP) rate constants also gradually increased with metal con-
tent, while the kHDA (OP) remained almost constant. The selectivity
of MD (SHDA (MD)) exhibited linear increase at the expense of
SHDA (MT), suggesting that more MT compounds were further
hydrogenated when the metal loading was increased. The selec-
tivity of the OP products was around 5% and remained constant.
Fig. 8 exhibits the rate constants and selectivities in the 1-MN
HDA on CoMoP and CoMoCA catalysts. With CA addition, the
kHDA , kHDA (MT), kHDA (MD) and kHDA (OP) rate constants markedly
improved by 24%, 20%, 132% and 28%, respectively. The beneficial
effect of CA on kHDA , kHDA (MT), kHDA (MD) and kHDA (OP) was gener-
ally higher than of GL. Regarding the selectivity, SHDA (MT) slightly
decreased whereas SHDA (MD) increased due to CA addition. The
variation of SHDA (OP) and SHDA (CR) were limited.
The removed sulfur and consumed hydrogen in each pathway
of HDS and HDA reactions were calculated and are presented in
Table 5. The amount of total sulfur removed and sulfur removed in
each pathway are related to the HDS conversion and selectivity of
the catalyst. The hydrogen consumption for every route in HDS and
HDA can reflect the depth of aromatics saturation.
4. Discussion
4.1. Structure of the active site
The sulfide catalysts were characterized by XPS and HRTEM
methods. The Mo/Al and Co/Al ratio increased with the presence of
CA, indicating that the dispersion of Mo and Co was improved. The
organic compounds may interact with the most basic OH groups so
that the interaction between metal and support is reduced and the
dispersion is improved [51,55]. Moreover, Mo species in GL contain-
ing catalysts exhibited similar sulfidation degree (90%) and fraction
of CoMoS species (71%). It suggested that the metal-support inter-
action was similar when the same GL/Co ratio was used in the

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 9

Table 5
The amount of removed sulfur and hydrogen consumption in 4,6-DMDBT HDS and 1-MN HDA.

Sample x(4.6-DMDBT) SDMBiPh SDMCHT SDMDCH x(1-MN) Amount of removed sulfur (×10−6 mol/g) Hydrogen consumption (×10−6 mol/g)

% % % % % Srem (HDS) Srem (DDS) Srem (HYDS) H2 (HDS) H2 (DDS) H2 (HYDS) H2 (HDA)

CoMoGL-1 98.28 15.98 53.83 30.19 93.73 21.18 3.30 17.87 122.11 6.61 115.51 6143.41
CoMoGL-2 98.35 12.56 50.40 37.04 95.58 22.94 2.88 20.06 140.06 5.76 134.30 6315.23
CoMoGL-3 98.71 11.94 47.85 40.21 95.87 23.35 2.79 20.57 145.97 5.58 140.40 6350.75
CoMoGL-4 98.97 11.50 45.18 43.32 96.19 23.70 2.72 20.97 151.38 5.45 145.93 6429.85
CoMoP 95.31 12.25 57.98 29.77 93.55 21.35 2.62 18.74 124.33 5.23 119.10 6033.56
CoMoCA 99.90 13.57 43.65 42.78 96.70 24.42 3.31 21.10 153.92 6.63 147.29 6564.53

1.2 100% 0.4

SHYDS
kHDS
Rate constant of HDS (mol/h/L)

80%

S DDS and S HYDS selectivity

0.3
0.8 kHYDS

S DDS/S HYDS
60%
SDDS/SHYDS 0.2
40%
0.4
0.1
kDDS 20%

SDDS
0.0 0% 0.0
15 18 21 24 15 18 21 24
MoO3 content (wt%) MoO3 content (wt%)

Fig. 5. Activity and selectivity in 4,6-DMDBT HDS over the catalysts containing GL.

1.6 100% 0.4

Rate constant of HDS (mol/h/L)

SHYDS
80%
S DDS and S HYDS selectivity

1.2 0.3
kHDS

S DDS/S HYDS
60%
0.8 0.2
40% SDDS/SHYDS
kHYDS

0.4 SDDS 0.1

kDDS 20%

0.0 0% 0.0
CoMoP CoMoGL-2 CoMoCA CoMoP CoMoGL-2 CoMoCA

Fig. 6. Activity and selectivity in 4,6-DMDBT HDS over CoMoP, CoMoGL-2 and CoMoCA catalysts.

100 100%
Rate constant of 1-MN HDA (mol/h/L)

kHDA

80 80%
Selectivity of 1-MN HDA

SHDA(MT)
kHDA(MT)

60 60%

kHDA(MD) SHDA(MD)
40
10 40%
10%
kHDA(OP) SHDA(OP)
20
5 20%
5%
kHDA(CR) SHDA(CR)

0 0%
15 18 21 24 15 18 21 24
MoO3 content (wt%) MoO3 content (wt%)

Fig. 7. Activity and selectivity in 1 -MN HDA reaction over the catalysts containing GL.

catalysts. Therefore, the concentration of MoS2 and CoMoS species and CoMoCA. Nevertheless, the sulfidation of Mo did not signif-
increased linearly with metal loading as shown in Table 3. Com- icantly change with GL and CA addition and already reached a
pared to the CoMoP catalyst, the addition of GL and CA did not very high level. In contrast, the effect of GL and CA on the forma-
modify the concentration of MoS2 in a marked way. The Mo5+ /Mo tion of CoMoS phase was more marked. The CoMoS concentration
ratio in the CoMoP catalyst was slightly higher than in CoMoGL-2 ([CoMoS]) improved significantly with GL and CA addition. Table 3

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
10 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx

100 100%

Rate constant of 1-MN HDA (mol/h/L)

kHDA
80 80%

Selectivity of 1-MN HDA

kHDA(MT) SHDA(MT)

60 60%

kHDA(MD) SHDA(MD)
40
10 40%
10%

205 kHDA(OP) 20%

5% SHDA(OP)
kHDA(CR) SHDA(CR)
0 0%
CoMoP CoMoGL-2 CoMoCA CoMoP CoMoGL-2 CoMoCA

Fig. 8. Activity and selectivity in 1 -MN HDA over the CoMoP, CoMoGL-2 and CoMoCA catalysts.

shows that the effect of CA was slightly higher than that of GL. The of kDDS decreased simultaneously. This suggests that the DDS and
fraction of Co9 S8 species also decreased by CA addition. As a result, HYDS reactions do not take place at the same active sites. Assuming
the concentration of the CoMoS phase ranked as CoMoP < CoMoGL- that the DDS and HYDS reactions can happen on every CoMoS site,
2< CoMoCA. The beneficial effect of GL and CA is due to the “known” the turnover frequency of CoMoS sites for DDS and HYDS routes,
chelating effect of organic compounds that may favor the forma- TOFDDS (CoMoS) and TOFHYDS (CoMoS), can be calculated. As shown
tion of promoted sites [49–51]. The beneficial effect of CA on CoMoS in Table 6, TOFDDS (CoMoS) and TOFHYDS (CoMoS) showed reverse
sites has been detected by CO and NO adsorption followed by IR trends with increasing metal content. The structure of the CoMoS
spectroscopy [51,56] in agreement with the present XPS results. sites changes and less DDS and more HYDS. Examining the mor-
As shown in Table 4, the average length of the MoS2 slab phology of the MoS2 phase, the concentration of corner sites of
increased from 3.2 to 4.3 nm with increasing Mo loading but the MoS2 (C (MoC )) tended to decrease whereas the edge sites (C(MoE ))
stacking increased only slightly. In previous work, Silva et al. changed in a reverse way. The variation of the concentration of
observed that the average slab length varied from 2.8 to 4.0 nm and corner and edge sites agree very well with the change of the DDS
the degree of stacking increased from 1.1 to 1.4 when the MoO3 con- and HYDS functions of the CoMoS sites. Therefore, it is proposed
tent increased from 9.8 wt% to 22.5 wt% [42]. Payen et al. reported that corner and edge sites account for the DDS and HYDS functions,
that the increase of the Mo loading from 4 wt% to 14 wt% corre- respectively.
sponded to an increase of the length of the MoS2 crystallite from In CoMo catalysts, the promoted sites are much more active
2.0 to 3.5 nm [46]. Pratt et al. found that the fraction of longer MoS2 than the unpromoted sites. Therefore, the concentration of CoMoS
slabs increased with MoO3 content (4–40 wt%) [76]. Furthermore, sites located at the corner and edge were evaluated from the XPS
Dhar et al. indicated that the crystallite size of MoS2 supported on and HRTEM results. Assuming that the Co atoms are homoge-
TiO2 - Al2 O3 calculated from oxygen uptakes was more or less con- nously dispersed on the corner and edge positions of the slab, the
stant up to 8 wt% MoO3 and increased rapidly with further increase number of CoMoS sites at the corner and edge (C(CoMoSC )) and
of the molybdenum loading (8–14 wt%) [77]. In sulfided NiMo/Zr- (C(CoMoSE )) can be calculated from the corner and edge ratio and
SBA-15 catalysts, the size of the Mo species also showed a progres- the total concentration of CoMoS species. The calculation equations
sive increase with metal loading [78]. Therefore, the increase of the for C(CoMoSC ) and C(CoMoSE ) are:
particle size of MoS2 with MoO3 loading observed in the present
C(MoC )
work is in agreement with the literature. C(CoMoSC ) = C(CoMoS) × (30)
Regarding the effect of GL, Escobar et al. found that the presence C(MoC ) + C(MoE )
of GL (GL/Ni = 1) in NiMo/Al2 O3 catalysts decreased the slab length C(MoE )
from 6.5 to 4.8 nm whereas it promoted the stacking height from C(CoMoSE ) = C(CoMoS) × (31)
C(MoC ) + C(MoE )
3.4 to 3.7 [49]. The slab length and stacking of the MoS2 phase were
quite higher than those observed in the present work, likely due to The data of C(CoMoSC ) and C (CoMoSE ) in Table 4 show that
the difference of the supports. Nevertheless, the effect of GL on the the promotion degree (fprom ) gradually increased from 0.90 to 1.14
length and stacking are similar. In many works, CA was selected to with increasing metal loading. A promotion degree higher than one
tune the properties of the sulfide catalyst. One important effect of (fprom > 1) was observed on the CoMoGL-3 and CoMoGL-4 catalysts.
CA appears to be the ability to modify the morphology of the sulfide The reason for the “over-promotion” may be due to the underesti-
phase. Our previous works on the NiW/Al2 O3 catalyst showed that mation of the amount of MoS2 species. Part of sulfide species may
in the presence of CA the slab became shorter and higher [52,53]. exist as Mo5+ species, which may partially link with the support.
In agreement with this, the stacking marked increased whereas This part of the Mo atoms was not taken into account in the con-
the length decreased when CA was added to the CoMoP catalysts centration of MoS2 . Therefore, the concentration of corner and edge
[79,80]. In the present work, the slab length of MoS2 decreased with sites (C(MoC ) and C(MoE )) may be underestimated, which led to
CA addition, in agreement with XPS results that indicated higher a higher evaluation of the promotion degree. Therefore, the cor-
dispersion. rected concentration of CoMoS sites in the corner and edge on
CoMoGL-3 and CoMoGL-4 catalysts were calculated as (C(MoC ) + C
(CoMoSC ))/2 and (C(MoE ) + C(CoMoSE ))/2. The results are listed in
4.2. Structure and selectivity relationship Table 4. The ratios of C(CoMoSC )/C (CoMoSE ) in various catalysts are
displayed as a function of kDDS and kHDYS ratio in Fig. 9. A direct lin-
It is generally agreed that CoMoS species account for the active ear relationship can be observed, which supports the assignment
sites in the HDS reaction. Co atoms decorate the edge positions of of corner sites to DDS reaction and edge sites to HYDS function. The
MoS2 and promote the HDS catalyst. The linear increase of CoMoS vertical adsorption of the S atom in 4,6-DMDBT on the active site
sites gradually improved the rate constant of kHYDS but the rate is highly inhibited by the neighboring methyl groups. The corner

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 11

Table 6
Turnover frequency of various active sites in 4,6-DMDBT HDS and 1-MN HDA reaction.

Sample TOF of CoMoS sites (h−1 ) TOF of corner and edge sites (h−1 )

TOFDDS (CoMoS) TOFHYDS (CoMoS) TOFHDA (CoMoS) TOFC (DDS) TOFE (HYDS) TOFE (HDA)

CoMoGL-1 0.60 3.23 340.08 2.69 4.16 437.25

CoMoGL-2 0.42 2.91 328.02 2.14 3.62 407.54
CoMoGL-3 0.37 2.77 299.10 2.16 3.47 374.86
CoMoGL-4 0.33 2.51 263.45 2.17 3.19 334.86
CoMo 0.34 2.43 320.77 1.90 2.96 390.60
CoMoCA 0.74 4.71 352.32 3.71 5.95 444.48

0.20 than the rate constants since the flat adsorption of 4,6-DMDBT may
occur on several sites. The ratio of H2 (DDS) and H2 (HDA) also
linearly increased with C (CoMoC )/C(CoMoSE ), indicating that the
0.18 HYDS route and HDA reaction may take place at the same active
k DDS/kHYDS

center. This is in agreement with the adsorption of 4,6-DMDBT and

0.16
0.14
0.12
0.10
0.16 0.2 0.24 0.28 0.32
C(CoMoSC)/C(CoMoSE )
Fig. 9. Relationship between C (CoMoSC )/C(CoMoSE ) and kDDS /kHYDS on various cat-
alysts.
Fig. 10. Relationship between C (CoMoSC )/C(CoMoSE ) and hydrogen consumption
ratios in the HDS and HDA.
sites should be the most “open” positions on the slab. The vertical
adsorption of S in 4,6-DMDBT would preferentially occur on the
corner sites where the DDS reaction takes place. Therefore, it is
reasonable to assign the DDS function to the corner sites.
Regarding the hydrogen consumption in HDS and HDA, the ratio
of H2 (DDS) and H2 (HYDS) showed linearly relationship with
the C (CoMoC )/C(CoMoSE ) ratio (Fig. 10). H2 (DDS)/H2 (HYDS)
stands for the total hydrogen consumption ratio of the DDS and
HYDS routes. The hydrogen consumption reflects the extent of
hydrogenation of 4,6-DMDBT. For instance, the same selectivity
to DMCHT and DMDCH does not mean the same extent of hydro-
genation. The hydrogen consumption in each route is related the
amount of hydrogen activated in the reaction, which may depend
on the concentration of active sites. Therefore, the hydrogen con-
sumption may be more linearly related to the number of active sites
1-MN on the active center for HYDS and HDA, which both are via
flat adsorption [1,8,29].
From the C(CoMoC ), C (CoMoSE ) and the rate constants in the
HDS and HDA reactions, the turnover frequency (TOF) of corner
sits for the DDS route (TOFC (DDS)) and edge sites for the HYDS
and HDA routes (TOFE (HYDS) and TOFE (HDA)) were calculated.
The results are shown in Table 6. For the GL containing catalysts,
TOFC (DDS) did not show substantial change whereas TOFE (HYDS)
and TOFE (HDA) decreased gradually with increasing metal load-
ing. The reason might be that longer slabs have a higher ratio of
Mo edge and sulfur edge sites [81]. Recently, Chen and Maugé
et al. proved that the intrinsic activity of the sulfur edge site was
60% more higher than that of the Mo edge site in thiophene HDS
[56]. This may explain why lower TOFE (HYDS) and TOFE (HDA) were
observed on the slab with higher length. Regarding the effect of GL
and CA, TOFC (DDS), TOFE (HYDS) and TOFE (HDA) increased linearly
with the stacking of MoS2 . It suggests that the intrinsic activities
of the active sites having higher stacking were higher on stacked
catalysts, in line with another report in the literature [43]. From the
calculated TOFs, it can be deduced that TOFC (DDS) mainly depends
on the stacking, whereas TOFE (HYDS) and TOFE (HDA) are related to
the length and stacking of the sulfide phase. Therefore, the intrinsic
activities of corner and edge sites can be roughly fitted. TOFC (DDS)
and TOFE (HYDS) are expressed as: TOFC (DDS) = 2.31 × N − 2.38,
TOFE (HYDS) = −1.33 × L + 2.93 × N + 2.35.
As generally reported, 4,6-DMDBT and aromatics may adsorb on
the same active sites since the HYDS route was mainly inhibited
by aromatics, which is supported by Fig. 10. Does the selectiv-
ity of HYDS and HDA change with the morphology of the sulfide
phase? In the literature, different relationships have been estab-
lished between the morphology of the sulfide phase and the
HDS/hydrogenation selectivity [37,45,63]. Therefore, we checked
the relationship between HYDS/HDA selectivity. The ratio of hydro-
gen consumed in HYDS and in HDA reactions is plotted as a function
of C (CoMoC )/C(CoMoSE ) in Fig. 11. H2 (HYDS)/H2 (HDA)
increased linearly with C(CoMoC )/C(CoMoSE ) for the GL contain-
ing catalysts but the CoMoP and CoMoCA catalysts showed the
revise trend. This contradiction suggests that the corner and edge
sites ratio may not control the hydrogenation ratio in HYDS and
HDA. This can be understood by realizing that the corner sites do
not account for the HYDS reaction. When H2 (HYDS)/H2 (HDA)
is plotted as a function of the stacking of MoS2 , a more rational
correlation is observed (Fig. 11). It indicates that higher stack-
ing tends to be more selective for 4,6-DMBT HYDS than for 1-MN
HDN. This may be due to the fact that higher stacking can pro-
vide more accessibility for 4,6-DMDBT than for 1-MN molecule as
the size of 4,6-DMDBT is larger. When the stacking of MoS2 was
increased (CoMoP vs CoMoCA), the TOFE (HYDS) was doubled while
the TOFE (HDA) was only improved by 14%. This also indicates that

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
12 W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx
Fig. 11. Relationship between the hydrogen consumption and morphology of the sulfide phase.
the edge sites on higher layer of MoS2 slab are more beneficial for
4,6-DMDBT HDS than for 1-MN HDA. The above results provide
indications for tuning SDDS /SHYDS and SHDS /SHDA selectivities and
hydrogen consumptions in HDS and HDA reactions.
The present work shows that the DDS and HYDS functions in
the HDS of 4,6-DMDBT are located on the corner and edge sites,
respectively. The assignment of the hydrogenation sites to the edge
sites is in agreement with the work reported by Silva et al. [42]. It
should be noted that the BRIM sites, reported by the Topsøe group,
can capture the 4,6-DMDBT molecules and catalyze the HYDS reac-
tion [32,41]. Considering this point, the concentration of active sites
for the HYDS route may be underestimated in this work. Since the
positions of the BRIM sites are parallel adjacent to the edge sites,
the concentration of BIRM sites would be proportional to the num-
ber of edge sites (C (CoMoSE )). Supposing the concentration ratio
of the BRIM and edge sites is n, the total active sites for the HYDS
is (1 + n)*C(CoMoSE ). The kDDS /kHYDS would still linearly relate to
1/(1 + n)*C (CoMoSC )/C(CoMoSE ). Therefore, the assignment of the
DDS and HYDS functions to the corner and edge sites can also be
established. In this case, the intrinsic activity of the DDS route is
not affected, whereas the intrinsic activity of the HYDS would be
the current TOFE (HYDS) multiplying (1/(1 + n)).
In the “rim-edge” model, the active sites for the hydrogenoly-
sis of DBT were assigned to both the rim and edge sites, whereas
the hydrogenation function was exclusively attributed to the rim
sites [37]. This model showed that the ratio of the rate constants
of hydrogenolysis and hydrogenation linearly increased with the
stacking number. In the present work, the kDDS /kHYDS did not show
clear relationship with the stacking height of the sulfide phase. It
indicates that the stacking number did not likely determine the
kDDS /kHYDS selectivity in the HDS of 4,6-DMDBT. This is likely due
to the steric effect in the structure of 4,6-DMDBT, which prevents
its vertical adsorption on the edge and rim sites. For the HDS and
hydrogenation of even smaller molecules such as thiophene and
hexene, many works showed the edge sites of MoS2 are preferable
for the selectivity of thiophene, whereas the corner sites account
for the hydrogenation function [45,62]. It reflects that the positions
of different catalytic functions also relate to the structure of the
reactants.
5. Conclusions
The activity, selectivity and hydrogen consumption in 4,6-
DMDBT HDS and 1-MN HDA were evaluated for a series of CoMo
catalysts with different morphology of the sulfide phase tuned by
increasing the metal content and by adding organic compounds
(GL and CA). The results showed that the length of MoS2 slabs pro-
gressively increased with metal loading whereas stacking hardly
changed. The introduction of GL and CA led to a reduction of slab
Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029
length and an enhancement of the stacking due to the variation
of metal-support interaction. Based on XPS and HRTEM measure-
ments, the concentration of the CoMoS species at corner and edge
sites were calculated. The ratio of the rate constants for the DDS
and HYDS routes varied linearly with the ratio of the CoMoS con-
centrations at the corner and edge sites, indicating that the corner
and edge sites account for DDS and HYDS reactions in 4,6-DMDBT
HDS, respectively. The comparison of hydrogen consumption in
DDS, HYDS and HDA suggested that 4,6-DMDBT HYDS and 1-MN
HDA reactions occur on the edge site of the sulfide slab. On the
other hand, it was found that the ratio of hydrogen consumption
in 4,6-DMDBT HYDS and 1-MN HDA was enhanced with increas-
ing stacking of the MoS2 slabs. The results indicated that the flat
adsorption of 4,6-DMDBT was more favored on higher layer than
the adsorption of 1-MN. Furthermore, it was found that the intrinsic
activity of corner sites for DDS was mainly related to the stacking,
whereas the TOF of edges sites for HYDS and HDA reactions were
dependent on the length and stacking of MoS2 slab.
Acknowledgements
The authors gratefully acknowledge the funding of the State
Key Project (Grant 2012CB224802) and the funding from Sinopec
project (115006). They thank Mr. Wang Yifan, Dr. Qiu Limei and
Dr. Xiang Yanjuan for activity evaluation, XPS and HRTEM analysis,
respectively.
References
[1] H. Topsøe, B.S. Clausen, F.E. Massoth, Catalysis: Science and Technology,
Springer, 1996.
[2] R.G. Leliveld, S.E. Eijsbouts, Catal. Today 130 (2008) 183.
[3] D. Li, Chin. J. Catal. 34 (2013) 48.
[4] L.C. Castañeda, J.A.D. Muñoz, J. Ancheyta, Fuel 90 (2011) 3593.
[5] S.S. Shih, S. Mizrahi, L.A. Green, M.S. Sarli, Ind. Eng. Chem. Res. 31 (1992) 1232.
[6] A. Stanislaus, A. Marafi, M.S. Rana, Catal. Today 153 (2010) 1.
[7] A. Stanislaus, B.H. Cooper, Catal. Rev. Sci. Eng. 36 (1994) 75.
[8] D.D. Whitehurst, T. Isoda, I. Mochida, Adv. Catal. 42 (1998) 345.
[9] C. Song, Catal. Today 86 (2003) 211.
[10] M. Vrinat, R. Bacaud, D. Laurenti, M. Cattenot, N. Escalona, S. Gamez, Catal.
Today 107–108 (2005) 570.
[11] R. Prins, M. Egorova, A. Röthlisberger, Y. Zhao, N. Sivasankar, P. Kukula, Catal.
Today 111 (2006) 84.
[12] G. Pérot, Catal. Today 86 (2003) 111.
[13] S.K. Bej, S.K. Maity, U.T. Turaga, Energy Fuels 18 (2004) 1227.
[14] R. Shafi, G.J. Hutchings, Catal. Today 59 (2000) 423.
[15] M. Breysse, G. Djega-Mariadassou, S. Pessayre, C. Geantet, M. Vrinat, G. Perot,
M. Lemaire, Catal. Today 84 (2003) 129.
[16] S. Eijsbouts, S.W. Mayo, K. Fujita, Appl. Catal. A 322 (2007) 58.
[17] R.R. Chianelli, G. Berhault, B. Torres, Catal. Today 147 (2009) 275.
[18] S.A. Hanafi, M.S. Mohamed, Energy Sources A 33 (2011) 495.
[19] D.R. Kilanowski, H. Teeuwen, V.H.J. de Beer, B.C. Gates, G.C.A. Schuit, H. Kwart,
J. Catal. 55 (1978) 129.
[20] T. Kabe, A. Ishihara, Q. Zhang, Appl. Catal. A 97 (1993) L1.
[21] V. Meille, E. Schulz, M. Lemaire, M. Vrinat, J. Catal. 170 (1997) 29.
G Model
CATTOD-10396; No. of Pages 13 ARTICLE IN PRESS
W. Chen et al. / Catalysis Today xxx (2016) xxx–xxx 13

[22] P. Michaud, J.L. Lemberton, G. Pérot, Appl. Catal. A 169 (1998) 343. [52] H. Nie, X. Long, Q. Liu, D. Li, Acta Petrol. Sin. (Petrol. Proc. Sect.) 26 (2010) 329.
[23] M. Egorova, R. Prins, J. Catal. 225 (2004) 417. [53] H. Li, M. Li, Y. Chu, F. Liu, H. Nie, Appl. Catal. A 403 (2011) 75.
[24] E. Rodríguez-Castellón, A. Jiménez-López, D. Eliche-Quesada, Fuel 87 (2008) [54] P. Castillo-Villalón, J. Ramirez, J.A. Vargas-Luciano, J. Catal. 320 (2014) 127.
1195. [55] D. Valencia, T. Klimova, Appl. Catal. B 129 (2013) 137.
[25] M.V. Landau, D. Berger, M. Herskowitz, J. Catal. 159 (1996) 236. [56] J. Chen, F. Maugé, J. El Fallah, L. Oliviero, J. Catal. 320 (2014) 170.
[26] F. Bataille, J.L. Lemberton, G. Pérot, P. Leyrit, T. Cseri, N. Marchal, S. Kasztelan, [57] L. Qiu, G. Xu, Appl. Surf. Sci. 256 (2010) 3413.
Appl. Catal. A 220 (2001) 191. [58] A.D. Gandubert, E. Krebs, C. Legens, D. Costa, D. Guillaume, P. Raybaud, Catal.
[27] L.E. Kallinikos, A. Jess, N.G. Papayannakos, J. Catal. 269 (2010) 169. Today 130 (2008) 149.
[28] T. Kabe, K. Akamatsu, A. Ishihara, S. Otsuki, M. Godo, Q. Zhang, W. Qian, Ind. [59] J.C. Dupin, D. Gonbeau, I. Martin-Litas, P. Vinatier, A. Levasseur, Appl. Surf. Sci.
Eng. Chem. Res. 36 (1997) 5146. 173 (2001) 140.
[29] F. Bataille, J.-L. Lemberton, P. Michaud, G. Perot, M. Vrinat, M. Lemaire, E. [60] L. Portela, P. Grange, B. Delmon, J. Catal. 156 (1995) 243.
Schulz, M. Breysse, S. Kasztelan, J. Catal. 191 (2000) 409. [61] D.H. Zuo, M. Vrinat, H. Nie, F. Maugé, Y.H. Shi, M. Lacroix, D.D. Li, Catal. Today
[30] J.H. Kim, X. Ma, C. Song, Prepr. Pap. Am. Chem. Soc. Div. Fuel Chem. 48 (2003) 93–95 (2004) 751.
553. [62] P.A. Nikulshin, D.I. Ishutenko, A.A. Mozhaev, K.I. Maslakov, A.A. Pimerzin, J.
[31] H. Farag, K. Sakanishi, I. Mochida, D.D. Whitehurst, Energy Fuels 13 (1999) Catal. 312 (2014) 152.
449. [63] P.A. Nikulshin, V.A. Salnikov, A.V. Mozhaev, P.P. Minaev, V.M. Kogan, A.A.
[32] H. Topsøe, Appl. Catal. A 322 (2007) 3. Pimerzin, J. Catal. 309 (2014) 386.
[33] P.J. Owens, C.H. Amberg, Adv. Chem. 33 (1961) 182. [64] X. Ju, Y. Zhang, L. Wang, D. Li, Acta Petrol. Sin. (Petrol. Proc. Sect.) 28 (2012)
[34] H. Topsøe, B.S. Clausen, N.Y. Topsøe, Ind. Eng. Chem. Fund. 25 (1986) 25. 538.
[35] H. Topsøe, R. Candia, N.-Y. Topsøe, B.S. Clausen, H. Topsøe, Bull. Soc. Chim. [65] P. Ge, X. Gao, L. Ren, Pet. Process Petroche 45 (2014) 36.
Belg. 93 (1984) 783. [66] H. González, O.S. Castillo, J.L. Rico, A. Gutiérrez-Alejandre, J. Ramírez, Ind. Eng.
[36] F.E. Massoth, G. Muralidhar, J. Shabtai, J. Catal. 85 (1984) 53. Chem. Res. 52 (2013) 2510.
[37] M. Daage, R.R. Chianelli, J. Catal. 149 (1994) 414. [67] J.F. Patzer, R.J. Farrauto, A.A. Montagna, Ind. Eng. Chem. Proc. Des. Dev. 18
[38] M. Brorson, A. Carlsson, H. Topsøe, Catal. Today 123 (2007) 31. (1979) 625.
[39] J.V. Lauritsen, J. Kibsgaard, G.H. Olesen, P.G. Moses, B. Hinnemann, S. Helveg, [68] R. Huirache-Acuña, B. Pawelec, E. Rivera-Muñoz, R. Nava, J. Espino, J.L.G.
J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard, F. Besenbacher, J. Catal. Fierro, Appl. Catal. B 92 (2009) 168.
249 (2007) 220. [69] C.N. Satterfield, S. Gultekin, Ind. Eng. Chem. Proc. Res. Dev. 20 (1981) 62.
[40] J.V. Lauritsen, M. Nyberg, J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard, F. [70] S. Gultekin, M. Khaleeq, M.A. Al-Saleh, Ind. Eng. Chem. Res. 28 (1989) 729.
Besenbacher, J. Catal. 224 (2004) 94. [71] V. Rabarihoela-Rakotovao, S. Brunet, G. Perot, F. Diehl, Appl. Catal. A 306
[41] J.V. Lauritsen, F. Besenbacher, J. Catal. 328 (2015) 49. (2006) 34.
[42] P.D. Silva, N. Marchal, S. Kasztelan, Stud. Surf. Sci. Catal. 106 (1997) 353. [72] A. Hu, H. Nie, W. Chen, X. Long, Pet. Process Petroche 46 (2015) 1.
[43] E.J.M. Hensen, P.J. Kooyman, Y. van der Meer, A.M. van der Kraan, V.H.J. de [73] Y. Miki, Y. Sugimoto, Fuel Process. Technol. 43 (1995) 137.
Beer, J.A.R. van Veen, R.A. van Santen, J. Catal. 199 (2001) 224. [74] K. Jaroszewska, A. Masalska, J.R. Grzechowiak, J. Grams, Appl. Catal. A 505
[44] L.- h. Liu, D. Liu, B. Liu, G.-c. Li, Y.-q. Liu, C.-g. Liu, J. Fuel Chem. Technol. 39 (2015) 116.
(2011) 838. [75] C. Petitto, G. Giordano, F. Fajula, C. Moreau, Catal. Commun. 3 (2002) 15.
[45] M. Li, H. Li, F. Jiang, Y. Chu, H. Nie, Catal. Today 149 (2010) 35. [76] K.C. Pratt, J.V. Sanders, V. Christov, J. Catal. 124 (1990) 416.
[46] E. Payen, S. Kasztelan, S. Houssenbay, R. Szymanski, J. Grimblot, J. Phys. Chem. [77] G.M. Dhar, B.N. Srinivas, M.S. Rana, M. Kumar, S.K. Maity, Catal. Today 86
93 (1989) 6501. (2003) 45.
[47] E. Payen, R. Hubaut, S. Kasztelan, O. Poulet, J. Grimblot, J. Catal. 147 (1994) 123. [78] O.Y. Gutiérrez, F. Pérez, G.A. Fuentes, X. Bokhimi, T. Klimova, Catal. Today 130
[48] S. Kasztelan, H. Toulhoat, J. Grimblot, J.P. Bonnelle, Appl. Catal. 13 (1984) 127. (2008) 292.
[49] J. Escobar, M.C. Barrera, J.A. Toledo, M.A. Cortés-Jácome, C. Angeles-Chávez, S. [79] T. Fujikawa, H. Kimura, K. Kiriyama, K. Hagiwara, Catal. Today 111 (2006) 188.
Núñez, V. Santes, E. Gómez, L. Díaz, E. Romero, J.G. Pacheco, Appl. Catal. B 88 [80] T. Fujikawa, M. Kato, T. Ebihara, K. Hagiwara, T. Kubota, Y. Okamoto, J. Jpn.
(2009) 564. Petrol. Inst. 48 (2005) 114.
[50] N. Rinaldi, T. Kubota, Y. Okamoto, Ind. Eng. Chem. Res. 48 (2009) 10414. [81] H. Schweiger, P. Raybaud, G. Kresse, H. Toulhoat, J. Catal. 207 (2002) 76.
[51] N. Rinaldi, Usman, K. Al-Dalama, T. Kubota, Y. Okamoto, Appl. Catal. A 360
(2009) 130.

Please cite this article in press as: W. Chen, et al., Influence of active phase structure of CoMo/Al2 O3 catalyst on the selectivity of
hydrodesulfurization and hydrodearomatization, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.09.029