Effect of sulfidation atmosphere on the hydrodesulfurization activity of SiO2-supported Co-Mo sulfide catalysts: Local structure and intrinsic activity of the active sites

Article abstract of DOI:10.1016/j.jcat.2009.08.017

The effects of the sulfidation atmosphere on the local structure and intrinsic activity of the active sites, Co-Mo-S, are studied with SiO₂-supported Co-Mo sulfide catalysts prepared by a chemical vapor deposition method using Co(CO)₃

Full text of DOI:10.1016/j.jcat.2009.08.017

Journal of Catalysis 268 (2009) 49–59
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Effect of sulfidation atmosphere on the hydrodesulfurization activity
of SiO2-supported Co–Mo sulfide catalysts: Local structure and
intrinsic activity of the active sites
a
a
b
Yasuaki Okamoto a, , Kazuya Hioka , Kenichi Arakawa , Takashi Fujikawa ,
*
Takeshi Ebihara b, Takeshi Kubota a
a Department of Material Science, Shimane University, Matsue 690-8504, Japan
b Research and Development Center, Cosmo Oil Co. Ltd., Satte 340-0193, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
The effects of the sulfidation atmosphere on the local structure and intrinsic activity of the active sites,
Co–Mo–S, are studied with SiO2-supported Co–Mo sulfide catalysts prepared by a chemical vapor depo-
sition method using Co(CO)3NO. The Co–Mo sulfide catalysts were sulfided in a stream of 10% H2S/H2 or
10% H2S/He or their combinations. The intrinsic activity of Co–Mo/SiO2 presulfided in H2S/He is 1.6 times
as high as that of Co–Mo–S Type II presulfided in H2S/H2 for the hydrodesulfurization (HDS) of thiophene,
demonstrating that the location and/or local structure of Co–Mo–S are varied by the sulfidation atmo-
sphere as well as by the MoS2-support interaction. The catalysts were characterized by X-ray absorption
fine structure and transmission electron microscope. It is suggested that the intrinsic activities of Co–Mo–
S on the S-edge and Mo-edge of MoS2 particles are distinctly different for HDS. Extended X-ray absorption
fine structure analysis shows a good correlation between the coordination number of Co–S and the intrin-
sic activity of ‘‘active Co–Mo–S”.
Received 13 June 2009
Revised 30 August 2009
Accepted 31 August 2009
Available online 26 September 2009
Keywords:
Hydrodesulfurization
Co–Mo sulfide catalysts
Active site structure
Intrinsic activity
Sulfidation atmosphere
EXAFS
Ó 2009 Elsevier Inc. All rights reserved.
XANES
Chemical vapor deposition
1. Introduction
and the support, resulting in the formation of Co–Mo–S Type II.
Van Veen et al. [12] showed, however, that Co–Mo–S Type II sup-
Improving the performance of HDS catalysts has been one of the
principal issues in the petroleum industry to protect the environ-
ment [1–3]. Supported Mo or W sulfide catalysts promoted by Co
or Ni have been widely used in industry for the HDS reaction [3–
5]. Detailed knowledge of the surface structure of HDS catalysts
and the number, intrinsic activity, and local structure of the active
sites is required to improve further the catalyst performance to
meet the increasingly stricter demands for cleaner fuels.
It is generally accepted that with Co–Mo HDS catalysts the so
called Co–Mo–S phase (structure), in which Co atoms are located
on the edge of MoS2 particles, is the active phase [3–5]. The forma-
tion of Co–Mo–S explains the generation of strong catalytic syner-
gies between Co and Mo sulfides. In addition, Co–Mo–S is at least
classified into two types depending on the intrinsic activity, Type
I and Type II, the latter showing activity that is about twice higher
than that shown by the former [5–7]. Several groups [7–11] re-
vealed that complete sulfidation of Mo precursors to MoS2 parti-
cles removes the chemical interactions between MoS2 particles
ported on activated carbon exhibits about twice as high an intrinsic
activity as Co–Mo–S Type II prepared on SiO2 or Al2O3. This was la-
ter confirmed by Bouwens et al. [10]. These findings may suggest
that the intrinsic activity of Co–Mo–S is not determined simply
by the interaction between MoS2 particles and the support.
The location and structure of Co–Mo–S have been studied using
a variety of physicochemical techniques [4,5,10,11,13–17]. How-
ever, the conclusions are still under debate. Recently, density func-
tional theory (DFT) calculations have provided important insight
into the detailed structure of the active sites of Co–Mo catalysts
[18,19]. Byskov et al. [20] suggested that Co atoms on the
ꢀ
(1 0 1 0) edge (S-edge) of MoS2 particles are energetically more
ꢀ
favorable than those on the (1 0 1 0) edge (Mo-edge). Schweiger
et al. [21] predicted for MoS2 edges fully promoted by Co atoms
that at strongly sulfiding conditions (high chemical potential of
sulfur) Co atoms are situated on both edges with 100% sulfur cov-
erage, while at usual HDS conditions Co atoms are present on the
S-edge with 50% sulfur coverage and the Co atoms on the Mo-edge
are energetically less favorable. In agreement with their predic-
tions, scanning tunneling microscope (STM) studies by Lauritsen
et al. [22–24] showed that Co atoms are preferentially located on
* Corresponding author. Fax: +81 852 32 6466.
E-mail address: yokamoto@riko.shimane-u.ac.jp (Y. Okamoto).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.08.017

50
Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
the S-edge of MoS2 particles with Co–MoS2/Au(111) prepared at
temperature, first in air and finally in vacuum, to avoid the oxida-
tion of the thiomolybdate. The dried material was treated in a 10%
H2S/H2 or 10% H2S/He stream at 673 K for 2 h. Mo/Si thus prepared
is designated as Mo/Si(ATM) hereafter. Co/Mo/Si(ATM) was pre-
pared by the CVD method as described above.
PH S=PH ¼ ca: 500.
2
2
With the effect of the coverage of Co atoms on the edge, Sun
et al. [25] suggested by use of DFT calculations that the surface en-
ergy of MoS2 edges with 50% Co coverage is lower than that with
100% Co coverage. More recently, Krebs et al. [26] predicted
through DFT calculations that under usual HDS conditions the S-
edge of MoS2 particles is fully promoted by Co, while 50% of the
Mo atoms are substituted by Co on the Mo-edge. Thus, it is ex-
pected that the location and local structure of Co–Mo–S depend
on the sulfidation and reaction conditions. It is of paramount
importance to study the intrinsic activity of these possible Co–
Mo–S structures for better understanding of HDS catalysts. How-
ever, experimental approaches have rarely been reported except
for a recent study by Gandubert et al. [27] using X-ray photoelec-
tron spectroscopy (XPS) in conjunction with DFT, suggesting that
Co–Mo-mixed sites on the Mo-edge are active for the hydrogena-
tion of toluene, while the S-edge fully promoted by Co shows a
lower activity.
In a previous study [11], we investigated the effect of the pre-
sulfidation temperature on the catalytic properties of SiO2-sup-
ported Co–Mo sulfide catalysts. X-ray absorption fine structure
(XAFS) and magnetic measurements showed that the MoS2-sup-
port interactions strongly modify the local structure and intrinsic
activity of Co–Mo–S for the HDS of thiophene. In the present study,
we investigate the effect of the sulfidation atmosphere (10% H2S/H2
and 10% H2S/He) on the HDS activity of SiO2-supported Co–Mo
sulfide catalysts and on the local structure and intrinsic activity
of Co–Mo–S to get deeper insight into the structure–reactivity rela-
tionships. We suggest that the catalytic properties of Co–Mo–S on
the S-edge and Mo-edge of MoS2 particles are distinctly different
for HDS.
CoO–MoO3/SiO2 (6.7 wt% Mo and 1.5 wt% Co) was prepared by a
double impregnation technique (Mo first and then Co) using AHM
and Co(NO3)2ꢁ6H2O. The catalyst was calcined at 773 K for 5 h after
each impregnation step. The sulfided catalyst is denoted as Co–Mo/
Si(imp).
MoO3/Al2O3 (8.7 wt% Mo, BET surface area of c
-Al2O3: 180 m2/
g) was prepared by an impregnation technique using AHM as a
precursor. MoO3/Al2O3 was calcined at 773 K for 5 h in air. The sul-
fided catalysts are denoted as Mo/Al and Co/Mo/Al.
2.2. Reaction procedure
The initial activity of the freshly prepared catalyst for the HDS
of thiophene was evaluated at 623 K using a circulation reaction
system made of glass under mild reaction conditions (initial H2
pressure, 20 kPa) [11]. The reaction products were mainly H2S, bu-
tenes, and n-butane, accompanying a trace amount of propylene.
The HDS activity was calculated on the basis of the accumulated
amount of H2S. The detailed reaction procedures have been re-
ported previously [28,29].
The HDS of dibenzothiophene (DBT) was carried out in a liquid
phase using an autoclave (200 mL) at 1.4 MPa of H2 pressure and
603 K. Sulfided Co/Mo/Si (0.1 g) was dispersed in 50 mL of decalin
containing 1 wt% of DBT in a glove bag filled with N2 to avoid con-
tact with air. After setting the reactant solution and the catalyst in
the autoclave vessel, the air was replaced first by N2 and then H2
five times each at room temperature. The H2 pressure was in-
creased and then the temperature was raised to 603 K (12 K/min)
while stirring (300 rpm) the reaction mixture. After 3–5 h of reac-
tion at 603 K, we analyzed the reaction mixture by GLC furnished
with a flame ionization detector (Shimadzu GC-14B).
2. Experimental
2.1. Catalyst preparation
2.3. Characterization
MoO3/SiO2 (6.7 wt% Mo) (SiO2, 370 m2 gꢀ1) was prepared in a
way analogous to the previous study by an impregnation method
using (NH4)6Mo7O24ꢁ4H2O (AHM) as a precursor [11]. MoO3/SiO2
was calcined at 773 K for 5 h in air. The MoO3/SiO2 catalyst thus
prepared (0.1 g) was presulfided at 673 K for 2 h in a 10% H2S/H2
or 10% H2S/He flow (100 mL/min) at atmospheric pressure. The
Mo sulfide catalyst is denoted as Mo/Si followed by the sulfidation
atmosphere, when necessary.
Cobalt was introduced by the chemical vapor deposition (CVD)
technique to Mo/Si [11,28–31]. Briefly, Mo/Si was first evacuated at
673 K for 1 h and subsequently exposed for 5 min at room temper-
ature to a vapor of Co(CO)3NO kept at 273 K. After evacuation at
room temperature for 10 min, the sample was sulfided again at
673 K in a 10% H2S/H2 or 10% H2S/He flow for 1.5 h to prepare
Co/Mo/Si. When the first and second sulfidation streams were
H2S/H2, Co/Mo/Si is denoted as Co/Mo/Si (H2S/H2). Similarly, Co/
Mo/Si sulfided twice in H2S/He is designated as Co/Mo/Si (H2S).
When the first and the second sulfidation flows were changed,
the catalyst is denoted, for instance, as Co/Mo/Si (H2S/H2 ? H2S),
in which Mo/Si was sulfided in H2S/H2 and then in H2S/He after
the Co addition by the CVD, and vise versa. The amount of Co incor-
porated by the CVD technique was determined for the catalysts
after the HDS reaction by means of X-ray fluorescence analysis
(XRF) (Shimadzu EDX-700HS) within an accuracy of 5%.
TEM images of Mo/Si were taken on an electron microscope
JEM-2010 with an accelerating voltage of 200 keV, as reported pre-
viously [11]. The distributions of MoS2 particle size and stacking
number were calculated over 150 particles in several arbitrary cho-
sen areas.
The Co K-edge XAFS spectra of the catalysts and reference com-
pounds were measured at room temperature at BL-9C (proposal
2006G331) of the Institute of Material Structure Science, High
Energy Accelerator Research Organization (KEK-IMSS-PF) with
2.5 GeV ring energy and 350–280 mA stored current in a fluores-
cence mode by using a Lytle type detector [11]. The higher har-
monics were eliminated by detuning the incident X-ray intensity
to 60%. The scan step was 0.5 eV. The synchrotron radiation was
monochromatized by a Si (1 1 1) double crystal monochromator.
The catalyst was evacuated at room temperature after sulfidation
and then transferred to an in situ XAFS cell with Kapton windows
without exposure to air.
The EXAFS data were analyzed with the same procedure as de-
scribed previously [11]. The analysis involved pre-edge extrapola-
tion and back ground removal by a cubic smoothing method to
obtain EXAFS oscillations, which were Fourier transformed from
k-space to R-space. Curve fittings were performed on the k-range
between 30 and 145 nmꢀ1 with k3-weighted
v(k) and an R-win-
Another Mo/Si (6.7 wt% Mo) was prepared by impregnation of
the SiO2 support with (NH4)2MoS4 (ATM, Kanto Chemicals) dis-
solved in N,N-dimethylformamide [11]. After one-night impregna-
tion, the solvent was removed by evaporation at room
dow of 0.10–0.33 nm. The data were analyzed by a curve fitting
procedure to obtain the coordination number N, atomic distance
R, and Debye–Waller factor
r, inner potential E0 using Eqs. (1)
and (2)

Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
51
k
3vðkÞ ¼ k2NFðkÞexpðꢀ2
r
2k2Þsinð2kR þ /ðkÞÞ=R2
ð1Þ
ð2Þ
presulfided in H2S/He compared to H2S/H2 irrespective of the sup-
port and preparation method.
2
1=2
k ¼ ð2m=hꢀ ðE ꢀ E0ÞÞ
In the present study, we employed a mild reaction condition to
intend to conserve the original surface states, prepared by the pre-
sulfidation, during the reaction as much as possible to establish a
structure–reactivity relationship. The stability of the catalytic
activity was studied first by repeating the 1 h-reaction, intermedi-
ated by 10 min-evacuation at the reaction temperature and subse-
quent admission of the fresh reactants (thiophene and hydrogen).
Fig. 1 shows the activity of the Co/Mo/Si catalysts presulfided in
H2S/H2 and H2S/He for the HDS of thiophene as a function of the
reaction time. The activities of both catalysts decrease gradually
during the reaction. However, the activity ratio of the catalysts is
constant regardless of the reaction time: Co/Mo/Si (H2S) shows a
1.4 times higher activity than the catalyst presulfided in H2S/H2,
indicating that Co/Mo/Si (H2S) is as stable as Co/Mo/Si (H2S/H2) un-
der the present reaction conditions. The activity decrease may be
due to the carbon deposition during the reaction rather than to
the degradation of active sites.
where m is the mass of an electron, and E and E0 are X-ray photon
energy and threshold energy for photoemission, respectively.
The parameters (back scattering amplitude F(k) and phase shift
u(k)) for the curve fittings were obtained by using FEFF calcula-
tions (FEFF codes: 8.2) [32,33] for the three scattering paths (Co–
Co, Co–S, and Co–Mo). A S20 value of 0.95 and a Debye–Waller factor
value of 0.006 nm were employed in the FEFF calculations. The de-
gree of error bars in the present analysis is estimated to be 0.5 for
N, 0.001 nm for R, 3 eV for
using the diagonal elements of the covariance matrix.
The normalized XANES spectra were obtained by subtracting
the pre-edge background from the raw data assuming a straight
line. Normalization was carried out by the edge height.
DE0, and 0.0003 nm for r, respectively,
3. Results
We have shown previously that Co–Mo–S is selectively and
fully prepared by the present CVD technique [28–31]. Besides,
the amount of Co deposited on SiO2 surface has been marginally
small [28]. As a consequence, the intrinsic activity of the catalyst,
as expressed by the turnover frequency (TOF) of the thiophene
HDS on Co–Mo–S, is evaluated on the basis of the amount of Co an-
chored by the CVD technique: TOF (hꢀ1) = (activity (mol/g h))/
(amount of Co (mol/g)). The reproducibility of TOF thus calculated
was better than 10%.
The TOF is given in Table 1 together with the amount of Co
(wt%), which is also expressed by the Co/Mo atomic ratio. As dem-
onstrated previously with Co/Mo/Si [11], the TOF attained by the
sulfidation at 873 K is higher than that at 673 K and corresponds
to the TOF for ‘‘Co–Mo–S Type II”, which is defined here, like other
researchers [3–5], as Co–Mo–S formed at usual H2S/H2 conditions
and without any strong interactions with the support. Neverthe-
less, when Co/Mo/Si is sulfided in H2S/He, the TOF is significantly
increased and 1.6 times greater even than that for Co–Mo–S Type
II. In agreement with our previous study [11], the TOF of Co/Mo/
Si(ATM) (H2S/H2) reveals the formation of Co–Mo–S Type II at
673 K. The presulfidation of Co/Mo/Si(ATM) in H2S/He greatly in-
creases the TOF to 28.5 hꢀ1, 1.8 times of Co–Mo–S Type II. Obvi-
ously, the sulfidation of Co/Mo/Si in H2S/He leads to the
formation of extraordinary active Co–Mo–S on SiO2 regardless of
the precursor of Mo. We call this highly active Co–Mo–S as
‘‘Co–Mo–S Type III” hereafter just for convenience. Fig. 1 shows
3.1. Catalytic activity of Co/Mo/Si
Table 1 summarizes the catalytic activity of Co/Mo/Si presulfid-
ed in a variety of sulfidation atmospheres for the HDS of thiophene.
In agreement with the previous results [11], the HDS activity of Co/
Mo/Si (H2S/H2) is varied little by the sulfidation at 873 K. A great
activity increase, however, is observed when Co/Mo/Si is presulfid-
ed at 673 K in 10% H2S/He (without H2). The amounts of H2S de-
tected in blank experiments, in which 20 kPa of H2 has been
circulated over Co/Mo/Si (H2S) at 623 K without thiophene, were
less than 0.1 mmol/h g, confirming that the activity increase ob-
served for Co/Mo/Si (H2S) is ascribed neither to the desorption of
H2S from the catalyst adsorbed during the presulfidation nor to
the H2S formation due to the hydrogenation of excess sulfur in
the catalyst.
Table 1 shows that Co/Mo/Si(ATM) (H2S) also exhibits a consid-
erably higher activity than Co/Mo/Si(ATM) (H2S/H2). The HDS
activity of the Co–Mo/Si impregnation catalyst is almost doubled
by the presulfidation in H2S/He compared to the sulfidation in
H2S/H2. A considerably higher activity is also observed for Co/
Mo/Al (H2S) than for Co/Mo/Al (H2S/H2). Thus, it is evident that
the HDS activity of the catalyst is significantly increased when
Table 1
Effect of sulfidation atmosphere on the thiophene HDS activity, Co content, and TOF
for Co/Mo/Si, Co/Mo/Si (ATM), Co/Mo/Si (imp), and Co/Mo/Al.
Catalyst
Sulfidation
conditionsa
HDS activity
Co content
TOF (hꢀ1
)
(mmol gꢀ1 hꢀ1
)
Wt% Co/Mo
atomic
ratio
Co/Mo/Si
H2S/H2
H2S/H2
H2S/He
H2S/H2 ? H2S/He 2.74
H2S/He ? H2S/H2 2.52
2.20
2.24
3.10
0.95 0.23
0.81 0.20
0.71 0.17
0.90 0.22
0.74 0.18
11.9
16.2
26.2
18.0
20.2
b
Co/Mo/Si
(ATM) H2S/He
H2S/H2
2.67
3.25
1.1
0.67 0.16
0.24
16.0
28.5
Co–Mo/Si H2S/H2
1.14
2.23
1.5
1.5
0.36
0.36
–
–
(imp)
H2S/He
Co/Mo/Al H2S/H2
H2S/He
3.14
3.83
2.5
2.1
0.46
0.39
7.6
14.2
a
The Mo catalyst was first presulfided for 2 h, followed by the introduction of Co
by the CVD method and subsequent second sulfidation for 1.5 h. The first and the
second sulfidation streams (10% H2S/H2 or 10% H2S/He) were the same. When they
were changed, the sequence is indicated by an arrow. The sulfidation temperature
was 673 K unless otherwise noted.
Fig. 1. Catalytic activity of Co/Mo/Si catalysts for the HDS of thiophene at 623 K as a
function of reaction time. d, Co/Mo/Si (H2S) and N, Co/Mo/Si (H2S/H2).
b
The first sulfidation was carried out at 873 K and the second one at 673 K.

52
Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
that Co–Mo–S Type III is stable under the present reaction condi-
tions. It is noteworthy in Table 1 that the amount of Co accommo-
dated in Mo/Si presulfided in H2S/He is smaller than that in H2S/H2.
In order to study the formation and stability of Co–Mo–S Type
III, the combinations of the first and second presulfidation steps
were changed. Mo/Si was first sulfided at 673 K for 2 h in H2S/H2
and subsequently Co was introduced by the CVD method, followed
by the second sulfidation at 673 K for 1.5 h in H2S/He. The activity
of Co/Mo/Si (H2S/H2 ? H2S) is presented in Table 1. The HDS activ-
ity and TOF are increased by the second sulfidation in H2S/He com-
pared with those for Co/Mo/Si (H2S/H2), but are smaller than those
for Co/Mo/Si (H2S). The reversed combination of the sulfidation
atmosphere forms Co/Mo/Si (H2S ? H2S/H2) with intermediate
HDS activity and TOF. It should be noted that the TOF values of
these Co/Mo/Si catalysts are higher than that for Co–Mo–S Type
II. Table 1 clearly shows that the amount of Co anchored in Co/
Mo/Si is determined by the sulfidation atmosphere of Mo/Si.
The catalytic activity of Co/Mo/Si (H2S) was measured for the
HDS of DBT at a H2 pressure of 1.4 MPa to examine the stability
of the catalyst. The HDS of DBT has been found to be first order
with respect to the concentration of DBT. The HDS product was
predominantly biphenyl with a small amount of hexylbenzene
(1–2%). Table 2 compares the rate constant and TOF, which is cal-
culated by dividing the rate constant by the amount of Co. Co/Mo/
Si (H2S) shows a 1.2 times higher HDS activity than Co/Mo/Si sul-
fided at 873 K (Type II catalyst), compared to the 1.4 times activity
increase shown in Table 1 for the HDS of thiophene under 20 kPa
of H2. The relative intrinsic activity of Co–Mo–S is increased by
40% for the HDS of DBT in comparison with 60% increase for the
HDS of thiophene. These results suggest that Co–Mo–S Type III
is fairly stable even during the HDS of DBT at 603 K and
1.4 MPa of H2.
3.2. Morphology
A representative TEM image of Mo/Si (H2S) is presented in
Fig. 2. As reported previously [11], in the case of Mo/Si (H2S/H2)
sulfided at 673 K, nanosized entities ascribable to amorphous
MoS2 particles and/or Mo (oxy)sulfides have been observed by
TEM along with MoS2 particles. However, no such entities are de-
tected in the TEM image shown in Fig. 2A, suggesting complete
sulfidation of Mo oxides in H2S/He at 673 K in contrast to Mo/Si
(H2S/H2). The average size and stacking number of MoS2 particles
calculated from the TEM images are summarized in Table 3. The
dispersion of MoS2 particles is lower and the stacking is more
extensive for Mo/Si (H2S/H2) than those for the previous catalyst
[11]. However, the TOF value presented in Table 1 is the same as
the previous one within the accuracy.
The TEM analysis of Mo/Si(ATM) (H2S/H2) and (H2S) was simi-
larly conducted. A typical TEM image for Mo/Si(ATM) (H2S/H2) is
shown in Fig. 2B. In contrast to the TEM image for Mo/Si (H2S/
H2) [11], no nanosized entities are detected in both catalysts pre-
pared using ATM. The morphological parameters are listed in Table
3. The dispersion of MoS2 particles is not varied very much by the
Mo precursors.
3.3. XAFS analysis
Fig. 3 shows the Co K-edge XANES spectra for Co/Mo/Si presul-
fided in various atmospheres and reference compounds (Co9S8 and
CoS2) for comparison. The spectra for Co/Mo/Si (H2S/H2) and Co/
Mo/Si(ATM) (H2S/H2) are taken from our previous study [11]. The
1s–3d pre-edge peak around 7709 eV is compared in Fig. 3B for
clarity. It is apparent that the pre-edge peak intensity for the Co/
Table 2
Table 3
Effect of sulfidation atmosphere on the activity of Co/Mo/Si for the HDS of
TEM analysis of Mo/Si presulfided at 673 K for 2 h.
dibenzothiophene.a
Catalyst
Sulfidation
atmosphere size (nm) stacking
number
Average
Average Co/Mo atomic ratio
Sulfidation
condition
Rate constant
Co contentc
(mmol gꢀ1
TOFd (hꢀ1 mol-
Coꢀ1
Estimated
from TEMa
Observed
(hꢀ1 gꢀ1
)
)
)
H2S/H2 (873 K)b
H2S/He (673 K)
0.55
0.66
0.14 (0.20)
0.12 (0.17)
3.9 ꢂ 103
Mo/Si
H2S/H2
H2S/He
4.0
3.2
3.7
4.4
0.24
0.28
0.23
0.18
5.5 ꢂ 103
a
H2 pressure: 1.4 MPa, reaction temperature: 603 K.
The second sulfidation was carried out at 673 K after the addition of Co by the
Mo/Si (ATM) H2S/H2
H2S/He
3.9
4.1
3.7
4.3
0.24
0.24
0.24
0.16
b
CVD.
c
a
The Co/Mo atomic ratio is given in the parentheses.
Assuming a regular hexagon, according to a model proposed by Kastzelan et al.
[34].
d
TOF was evaluated by dividing the rate constant by the amount of Co (mol gꢀ1).
Fig. 2. Representative TEM images of Mo/Si. (A) Mo/Si (H2S) and (B) Mo/Si(ATM) (H2S/H2).

Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
53
Mo/Si catalysts treated in H2S/He is considerably lower than that
for Co/Mo/Si (H2S/H2) and even than that for Co/Mo/Si(ATM)
(H2S/H2) and is close to the intensity for CoS2, in which Co is in
an octahedral symmetry. Fig. 3B thus indicates that the Co atoms
of Co–Mo–S in the Co/Mo/Si catalysts sulfided in H2S/He are in a
higher symmetry than in a square pyramidal configuration. In
our previous study [11], we have pointed out that a shoulder peak
around 7720 eV grows as Co–Mo–S shifts from Type I to Type II.
Fig. 3A shows that the shoulder peak for Co/Mo/Si once treated
in H2S/He is as prominent as that for Co/Mo/Si(ATM) (H2S/H2)
(Type II catalyst). Co9S8 and CoS2 show a corresponding peak but
the energy is several eVs lower than that for Co–Mo–S, suggesting
that the local structure and/or electronic state of Co–Mo–S are dif-
ferent from those of the bulk sulfides. With Co/Mo/Al (H2S), the
XANES features are close to those for Co/Mo/Si (H2S/H2), but appar-
ently different from those for Co/Mo/Al (H2S/H2) (Type I catalyst)
presented in [11]. The slight but distinct differences in the XANES
spectra clearly suggest the modifications of the local structure and/
or electronic state of Co–Mo–S by the presulfidation atmosphere.
Fig. 4 presents the Co K-edge k3v(k) EXAFS oscillations. Fig. 5
shows the Fourier transforms (FT) for Co/Mo/Si, Co/Mo/Si(ATM),
and Co/Mo/Al sulfided only in H2S/He. Fig. 6 compares the FTs of
Co/Mo/Si sulfided at a various conditions at 673 K. The EXAFS re-
sults for Co/Mo/Si (H2S/H2) sulfided at 673 K and 873 K and for
Co/Mo/Al (H2S/H2) have already been reported previously [11].
The FTs for Co/Mo/Si in Figs. 5 and 6 show well resolved three
peaks around 0.18, 0.24, and 0.28 nm (phase shifts: uncorrected)
irrespective of the sulfidation atmosphere and support, as reported
previously for Co/Mo/Si (H2S/H2) [11]. The FTs were analyzed by
the use of FEFF parameters, assuming Co–S, Co–Mo, and Co–Co
atomic pairs in order of increasing distance. The structural param-
eters are summarized in Table 4. The accuracy of the coordination
number (CN) of Co–S was estimated to be 0.5 using the diagonal
elements of the covariance matrix. We confirmed, however, with a
few catalysts that the CN of Co–S was reproducible within 0.1 as
long as we used the same sulfidation conditions and EXAFS analy-
Fig. 4. Co K-edge
k
3v(k) EXAFS oscillations for Co/Mo/Si (H2S), Co/Mo/Si
(H2S ? H2S/H2), Co/Mo/Si (H2S/H2 ? H2S), Co/Mo/Si(ATM) (H2S), and Co/Mo/Al
(H2S).
sis procedures. In the present study, we discuss mainly the relative
values of CN of Co–S to compare the local structure of Co–Mo–S as
a function of the sulfidation atmosphere. A comparison between
Fig. 3. (A) Co K-edge XANES spectra for Co/Mo/Si (H2S), Co/Mo/Si (H2S ? H2S/H2), Co/Mo/Si (H2S/H2 ? H2S), Co/Mo/Si(ATM) (H2S), Co/Mo/Al (H2S), and reference compounds
(Co9S8 and CoS2). The XANES spectra for Co/Mo/Si (H2S/H2) and Co/Mo/Si(ATM) (H2S/H2) are taken from [11] for comparison. (B) The pre-edge region of Co K-edge spectra is
shown for clarity.

54
Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
intensity decreases as the CN of Co–S increases, implying that
the relative CN of Co–S is a reliable parameter to describe the
change in the local structure of Co–Mo–S in spite of the estimated
error bar of 0.5 for the absolute values.
With Co/Mo/Si (H2S), the Co–S, Co–Mo, and Co–Co atomic dis-
tances are 0.220, 0.282, and 0.315 nm and close to those for Co/
Mo/Si (H2S/H2) sulfided at 873 K or the Co–Mo–S Type II catalyst
in [11]. The coordination number (CN) of the Co–S bonds is, how-
ever, larger than that for the Type II catalyst (4.5–4.9) [11], while
the CN of Co–Mo is not changed (1.5 0.5). Essentially identical
structural parameters are obtained for Co/Mo/Si(ATM) (H2S) and
Co/Mo/Si (H2S/H2 ? H2S). On the other hand, Co/Mo/Si
(H2S ? H2S/H2) shows a smaller CN of Co–S (4.8) compared with
Co/Mo/Si (H2S) (5.3). With Co/Mo/Al (H2S), the atomic distances
of Co–Mo and Co–Co are slightly elongated compared with those
for Co/Mo/Al (H2S/H2) (Co–Mo–S Type I by definition [5–7]) in
[11]. However, the CN of Co–S is not significantly varied by the
sulfidation in H2S/He (from 4.2 to 4.3) in contrast to Co/Mo/Si
(from 4.5 to 5.3).
We have previously noted a tendency of an increasing TOF with
the increasing CN of Co–S [11]. The significantly high intrinsic
activity of Co–Mo–S Type III may be correlated to the relatively
high CN of Co–S. Fig. 7 presents the TOF as a function of the CN
of Co–S for Co/Mo/Si and Co/Mo/Al prepared in H2S/H2 and H2S/
He, including the previous results [11]. Table 5 lists the CN of
Co–S for the catalysts relevant in the present study. Fig. 7 shows
a clear tendency that the TOF increases as the CN of Co–S increases
from Co–Mo–S Type I for Co/Mo/Al (H2S/H2) through Co–Mo–S
Type II for Co/Mo/Si (H2S/H2) presulfided at 873 K to Co–Mo–S
Type III for Co/Mo/Si (H2S) despite of some scattering in the corre-
lation. Bouwens et al. [10] reported a similar correlation including
Al2O3-, SiO2-, and activated carbon-supported Co–Mo sulfide
catalysts.
On the basis of the structural parameters shown in Table 4, we
tentatively propose a structure model of Co–Mo–S Type III in Fig. 8.
The structure of Co–Mo–S Type III may be described as a dinuclear
Co sulfide clusters as proposed for Co–Mo–S Type I and Co–Mo–S
Type II [11,16]. Since we have assumed a single Co–S peak for
five-six Co–S bonds in the analysis of the EXAFS data, it is conjec-
tured that the CN of Co–S of 5.3-5.4 presented in Table 4 is compat-
ible with the model within the accuracy ( 0.5). It seems that the
CN of Co–Co is not incompatible with the model within the accu-
racy ( 0.5). The weak 1s–3d pre-edge peak in Fig. 3 for Co/Mo/Si
(H2S) is consistent with a preferential formation of Co in a distorted
octahedral configuration. We tentatively propose that Co–Mo–S
Type III has a S-dimer between the two Co atoms as illustrated in
Fig. 8.
Fig. 5. Co K-edge k3-weighted Fourier transforms for Co/Mo/Si (H2S), Co/Mo/
Si(ATM) (H2S), and Co/Mo/Al (H2S). Dots: observed Fourier transforms, and solid
line: best curve-fittings.
the intensity of the pre-edge peak shown in Fig. 3B and the CN of
Co–S shows a small but clear tendency that the pre-edge peak
4. Discussion
4.1. Location of Co–Mo–S
In our previous study [11], we have estimated the amount of Co
required to fully decorate the edges of MoS2 particles on the basis
of the average size of MoS2 particles evaluated from the TEM anal-
ysis and the structural model proposed by Kasztelan et al. [34],
assuming a regular hexagon and a 1:1 substitution of Mo by Co.
An excellent agreement between the calculated and observed
amounts of Co for Co/Mo/Si prepared by the CVD technique sug-
gested that Co atoms fully occupy the substitutional sites of Mo
on both Mo-edge and S-edge of MoS2 particles [11]. We estimated
the Co/Mo ratios on the basis of the TEM results presented in Table
3 for Mo/Si (H2S/H2) and Mo/Si (H2S) and for the Mo/Si(ATM) coun-
terparts. They are compared in Table 3 with the observed Co/Mo
ratios. With Mo/Si (H2S/H2) and Mo/Si(ATM) (H2S/H2), the calcu-
lated ratios are in excellent agreement with the observed ratios,
Fig. 6. Co K-edge k3-weighted Fourier transforms for Co/Mo/Si (H2S), Co/Mo/Si
(H2S/H2 ? H2S), Co/Mo/Si (H2S ? H2S/H2), and Co/Mo/Si (H2S/H2). The Fourier
transform for Co/Mo/Si (H2S/H2) is taken from [11] for comparison.

Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
55
Table 4
Structural parametersa derived from Co K-edge EXAFS analysis of Co/Mo/Si and Co/Mo/Al catalysts.
Catalystb (presulfidation atmosphere)
Atomic pair
CN
R (nm)
D
E0 (eV)
r
(10ꢀ3 nm)
Rf (%)
Co/Mo/Si (H2S/H2 ? H2S/He)
Co–S
Co–Mo
Co–Co
5.3
1.4
0.9
0.2197
0.2810
0.3151
2.0
3.8
ꢀ7.8
8.2
9.4
10.8
2.5
Co/Mo/Si (H2S/He ? H2S/H2)
Co/Mo/Si (H2S/He)
Co–S
Co–Mo
Co–Co
4.8
1.4
0.5
0.2209
0.2799
0.3154
2.0
4.3
ꢀ7.8
3.1
5.6
ꢀ8.7
2.9
4.3
ꢀ8.9
1.2
4.1
ꢀ8.5
8.1
10.0
11.5
1.0
1.3
1.3
1.1
Co–S
Co–Mo
Co–Co
5.3
1.4
0.8
0.2198
0.2819
0.3150
8.2
9.8
11.2
Co/Mo/Si (ATM) (H2S/He)
Co/Mo/Al (H2S/He)
Co–S
Co–Mo
Co–Co
5.4
1.6
1.0
0.2205
0.2820
0.3151
8.0
9.2
10.8
Co–S
Co–Mo
Co–Co
4.3
1.5
1.0
0.2205
0.2831
0.3153
8.6
9.5
10.7
P
P
a
CN, coordination number; R, distance; E0, inner potential;
The Mo catalyst was first presulfided for 2 h, followed by the introduction of Co by the CVD method and subsequent second sulfidation for 1.5 h. The first and the second
sulfidation streams (10% H2S/H2 or 10% H2S/He) were the same. When they were changed, the sequence is indicated by an arrow. The sulfidation temperature was 673 K.
r
, Debye–Waller-like factor; Rf, R factor defined as Rf = {
[v
obs(k) ꢀ vcal (k)]2/ vobs (k)2}1/2
.
b
Fig. 7. Correlation between the TOF on Co–Mo–S for the HDS of thiophene and the
coordination number of Co–S bonds. Closed circles, Co/Mo/Si (H2S/H2) (673 K and
873 K) and Co/Mo/Si(ATM) (H2S/H2); triangles, Co/Mo/Al (H2S/H2) and (H2S); and
open circles, Co/Mo/Si (H2S), Co/Mo/Si (H2S ? H2S/H2), Co/Mo/Si (H2S/H2 ? H2S),
and Co/Mo/Si(ATM) (H2S). Refer to Table 5 for the CN of Co–S.
Table 5
Effect of sulfidation atmosphere on the coordination number of Co–S bonds and the
TOF of ‘‘active Co–Mo–S”.
Catalyst
Sulfidation
conditionsa
CN of Co–S
TOF on ‘‘active
Co–Mo–S” (hꢀ1
Fig. 8. Structural models of Co–Mo–S in Co/Mo/Si sulfided in H2S/He at 673 K. Large
light ball (yellow), sulfur; small light ball (blue), cobalt; and small dark ball (black),
molybdenum. (For interpretation of the references in color in this figure legend, the
reader is referred to the web version of this article.)
)
Co/Mo/Si
H2S/H2
H2S/H2
4.5c
4.9c
5.3
5.3
4.8
18.3
24.9
26.2
27.7
20.2
b
H2S/He
H2S/H2 ? H2S/He
H2S/He ? H2S/H2
confirming the previous conclusion that Co–Mo–S is formed on
both edges of MoS2 particles [11].
Co/Mo/Si (ATM)
Co/Mo/Al
H2S/H2
H2S/He
4.7c
5.4
4.2c
4.3
24.6
28.5
In sharp contrast to the excellent agreement between the ob-
served and calculated Co/Mo ratios for Co/Mo/Si presulfided in
H2S/H2, the experimental Co/Mo ratios are considerably smaller
for Co/Mo/Si (H2S) and Co/Mo/Si(ATM) (H2S) than the calculated
ones (Table 3). The disagreement suggests that only 65% (0.18/
0.28 = 0.64 and 0.16/0.24 = 0.67) of the edges of MoS2 particles in
Mo/Si presulfided in H2S/He is decorated with Co after the CVD.
According to the DFT calculations [20,25,35–39], both Mo-edge
and S-edge of MoS2 particles are expected to be fully covered by
H2S/H2
H2S/He
11.7
14.2
a
The Mo catalyst was first presulfided for 2 h, followed by the introduction of Co
by the CVD method and subsequent second sulfidation for 1.5 h. The first and the
second sulfidation streams (10% H2S/H2 or 10% H2S/He) were the same. When they
were changed, the sequence is indicated by an arrow. The sulfidation temperature
was 673 K unless otherwise noted.
b
The first sulfidation was carried out at 873 K and the second one at 673 K.
Taken from [11].
c

56
Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
sulfur (100% sulfur coverage), when Mo/Si or Mo/Si(ATM) is sulfid-
ed in H2S/He (without H2). Schweiger et al. [36] predicted by DFT
calculations coupled with thermodynamics that the surface energy
of the Mo-edge with 100% sulfur coverage is considerably lower
than that of the corresponding S-edge and, as a consequence, that
the shape of MoS2 particles is almost triangular consisting of the
Mo-edge (ca. 90%). The STM study on MoS2/Au(1 1 1) substantiated
the calculations [22]. Although the Mo sulfide-support interactions
and the chemical processes during the sulfidation may modify the
morphology of the resultant MoS2 particles, it is expected that after
the sulfidation in H2S/He, the Mo-edge prevails over the S-edge in
the fraction, both of which have 100% sulfur coverage: the Mo-edge
is covered by S-dimers and the sulfur atoms coordinate to only a
single Mo atom, while the S-edge is covered by relatively coordin-
atively saturated sulfur atoms [20,36,39]. In the present prepara-
tion of Co–Mo sulfide catalysts, Co(CO)3NO molecules are
allowed to contact the edges of MoS2 particles and subsequently
react with the sulfur atoms on the edges to form Co–Mo–S if the
sulfur atoms are reactive enough. Taking into consideration the
configuration and fraction of the Mo-edge and S-edge, we propose
that with Co/Mo/Si (H2S) Co–Mo–S is formed selectively on one of
the edges (65% of the total edges) of MoS2 particles, presumably on
the Mo-edge, while the other edge, possibly the S-edge, remains in-
tact. Under strongly sulfiding conditions (high chemical potential
of sulfur) such as those in H2S/He, Schweiger et al. [36] and Krebs
et al. [26] suggested that Co atoms are stabilized on both Mo-edge
and S-edge from the point of edge energy. It is considered that the
preferential location of Co on one of the edges (the Mo-edge) in the
present catalyst preparation is not thermodynamics-controlled,
but is kinetics-controlled due to a higher reactivity of this edge
(the Mo-edge) toward Co(CO)3(NO) molecules.
Schweiger et al. [36] predicted for MoS2 particles that under
typical sulfidation conditions the 50% sulfur coverage is stabilized
on both Mo-edge and S-edge involving coordinatively unsaturated
sulfur atoms. It is conceived that due to coordinative unsaturation
these edges are highly reactive toward Co(CO)3NO molecules, facil-
itating the formation of Co–Mo–S on both edges of MoS2 particles
for Mo/Si sulfided in H2S/H2. According to the theoretical calcula-
tions by Byskov et al. [20] and Schweiger et al. [21], the full substi-
tution of Mo atoms by Co on the Mo-edge is energetically more
expensive than that on the S-edge, suggesting the preferential for-
mation of Co–Mo–S on the S-edge under usual sulfidation or reac-
tion conditions such as those employed in the present study. Sun
et al. [25] and Krebs et al. [26], however, suggested under these
conditions that Co atoms are stabilized even on the Mo-edge when
the Mo atoms are partially substituted by the Co atoms (50%). In-
stead, we tentatively propose from Table 3 that both edges of
MoS2 particles are fully covered by Co in the present preparations
of Co/Mo/Si and Co/Mo/Si(ATM) presulfided in H2S/H2, since we
need to assume a 60–70% larger (Moedge/Mototal) ratio than that ex-
pected from the model proposed by Kasztelan et al. [34] when the
Mo-edge is only 50% covered by Co [25,26].
strongly depends on the sulfidation atmosphere. Co–Mo–S Type III
that is formed by the sulfidation in H2S/He at 673 K is 1.6–1.8
times as active as Co–Mo–S Type II, which is about twice as active
as Co–Mo–S Type I, for the HDS of thiophene. Co–Mo–S Type III is
as stable as Co–Mo–S Type I/II during the present mild reaction
conditions (Fig. 1). When Co/Mo/Si (H2S) is used for the HDS of
DBT at 1.4 MPa of H2 pressure, the differences in the activity and
TOF between Co–Mo–S Type III and Type II become smaller than
those observed for the HDS of thiophene. This may be accounted
for by a partial transformation of Co–Mo–S Type III to Co–Mo–S
Type II during the reaction at a higher H2 pressure, suggesting lim-
ited stability of Co–Mo–S Type III in practical HDS conditions. It is
considered that the degradation of Type III to Type II accompanies
the reduction of a S-dimer to a bridging sulfur (Co–S–Co [11,16]).
Before discussing the nature of Co–Mo–S Type III, it is noted
that the intrinsic activity of Co–Mo–S is determined mainly by
the interaction of MoS2 particles with the support surface [11],
the location of Co–Mo–S, and the local structure of Co–Mo–S. It
is conceived that these three determining factors are strongly con-
trolled by the sulfidation atmosphere and reaction conditions. In
the present study, we discuss the change in the intrinsic activity
of Co–Mo–S with the sulfidation atmosphere in terms of these
three points.
Table 1 shows that Co/Mo/Si(ATM) (H2S/H2) exhibits a TOF
characteristic of Co–Mo–S Type II because of complete sulfidation
of ATM to MoS2 in H2S/H2 at 673 K, in contrast to Co/Mo/Si (H2S/
H2) which needs a sulfidation temperature of >873 K to form Co–
Mo–S Type II, as revealed previously [11]. The temperature pro-
gramed sulfidation profile of MoO3/SiO2 in H2S/H2 has shown that
the sulfidation of Mo oxides proceeds through O–S exchange to
form MoO3ꢀxSy (<450 K), followed by H2-reduction to MoS2ꢀxOy
at 480 K and subsequent sulfidation to MoS2 (>500 K) [11,40,41].
Crystalline MoO3 present in MoO3/SiO2 is in part sulfided via
MoO2 because of diffusion limitations. It has been shown that
MoS2–O–SiO2 and/or MoS2–O–MoO2 interactions are completely
eliminated at >873 K to form Co–Mo–S Type II. On the other hand,
it is considered that such detrimental interactions are minimized
by using ATM as a precursor and thus Co–Mo–S Type II is formed
in H2S/H2 even at 673 K. Accordingly, it is rational to assume com-
plete sulfidation of Mo to MoS2 in Mo/Si(ATM) in H2S/H2 at 673 K.
In agreement with the assumption, the TEM image of Mo/Si(ATM)
(H2S/H2) presented in Fig. 2B shows no nanosized particles, which
are ascribed to Mo (oxy)sulfides [42,43].
The TEM image of Mo/Si (H2S) presented in Fig. 2A shows the
absence of nanosized Mo (oxy)sulfide particles that have been ob-
served for Mo/Si (H2S/H2) [11]. Besides, the Co K-edge XANES spec-
tra (Fig. 3A) for Co/Mo/Si once presulfided in H2S/He show a
distinct shoulder peak at 7720 eV characteristic of Co–Mo–S Type
III and Type II. Furthermore, it is evident from Table 1 that the
TOF of Co/Mo/Si (H2S) is the same as that of Co/Mo/Si(ATM)
(H2S) within the accuracy. All these findings indicate that the sulf-
idation of Mo oxides supported on SiO2 is completed in H2S/He at
as low as 673 K and thereby strong interactions such as MoS2-O-
support and/or MoS2-O-MoO2 are absent, since it is considered that
the sulfidation of MoO3 to MoS2 proceeds through the formation of
a MoS3 intermediate and its subsequent decomposition to MoS2 in
H2S/He without possible formation of MoO2, which is estimated to
be the origin of MoS2-O-MoO2 detrimental interactions. Similarly,
it is very likely that the sulfidation of Mo oxides is completed when
Mo/Si is once sulfided in a stream of H2S/He at 673 K for >1.5 h. In
the case of Co/Mo/Al (H2S), however, the TOF is still lower than that
of the Type II catalysts, suggesting incomplete sulfidation of Mo
oxides even in H2S/He due to strong Mo–O–Al linkages [4,5,44].
It is concluded that both Co–Mo–S Type III and Co–Mo–S Type II
are formed only when Mo precursors are completely sulfided to
MoS2 particles without any strong interactions with the support.
When the Co atoms located on the Mo-edge of Mo/Si presulfid-
ed in H2S/He are exposed to H2S/H2, the EXAFS analysis of Co/Mo/Si
(H2S ? H2S/H2) presented in Table 4 and the XANES spectrum pre-
sented in Fig. 3 show no sign of the degradation of Co–Mo–S to Co
sulfide clusters (vide infra), suggesting that Co–Mo–S, presumably,
on the Mo-edge is stable even in H2S/H2 once it is formed. In line
with the DFT calculations [20,21,26], the EXAFS and XANES results
of Co/Mo/Si (H2S/H2 ? H2S) indicate the stability of Co–Mo–S on
both edges at highly sulfiding conditions (H2S/He).
4.2. Formation of Co–Mo–S Type III
In the present study on SiO2-supported Co–Mo sulfide catalysts,
it is distinctly demonstrated that the intrinsic activity of Co–Mo–S

Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
57
Nevertheless, Type III is a type of Co–Mo–S formed only at strongly
sulfiding conditions (H2S/He in the present study), whereas Type II
is the one produced at usual sulfidation (H2S/H2 here) or reaction
conditions. It should be noted again that ‘‘Co–Mo–S Type II” is de-
fined in the present study and tacitly in most of the studies [4,5] as
Co–Mo–S formed at usual H2S/H2 conditions and without any strong
interactions with the support.
we assume, for simplicity, that the morphology of MoS2 particles
is not significantly varied in the H2S/H2 and H2S/He atmospheres
(the fraction of the Mo-edge for ‘‘active Co–Mo–S”; 65%) and that
‘‘less active Co–Mo–S” shows no activity, we can estimate the
TOF on ‘‘active Co–Mo–S” by simply dividing the observed TOF
by 0.65 for the Co/Mo catalysts presulfided in H2S/H2 before the
addition of Co. The first assumption may be verified by the DFT cal-
culations by Schweiger et al. [36] for MoS2 particles, predicting 70%
Mo-edge at typical sulfidation and reaction conditions vs. 90% at
strongly sulfiding conditions. The fraction of the Mo-edge, 65%, in
the present study is compatible with these DFT predictions. The
TOF on ‘‘active Co–Mo–S” thus estimated is summarized in Table
5 together with the CN of Co–S bonds. Fig. 9 shows a correlation
between the TOF on ‘‘active Co–Mo–S” and ‘‘the average CN of
Co–S bonds” over ‘‘active” and ‘‘less active” Co–Mo–S for all the
Co/Mo/Si and Co/Mo/Al catalysts examined in the present and pre-
vious [11] studies. A considerably improved correlation is obtained
in Fig. 9, which includes Co–Mo–S Type I, Type II, and Type III cat-
alysts, compared with a rather scattered one shown in Fig. 7. The
satisfactory correlation in Fig. 9 may substantiate our propositions:
one of the edges of MoS2 particles is selectively decorated by Co–
Mo–S in the Co/Mo catalysts presulfided in H2S/He before the
admission of Co and only the Co–Mo–S structure located on this
edge of MoS2 particles is catalytically active for HDS, while the
Co–Mo–S structure on the other edge is much less active. As dis-
cussed in Section 4.1., it is suggested that ‘‘active Co–Mo–S” is pre-
sumably situated on the Mo-edge rather than on the S-edge. Very
recently, Gandubert et al. [27] suggested that Co–Mo-mixed sites
on the Mo-edge are active for the hydrogenation of toluene, while
the S-edge fully promoted by Co is much less active. The present
and their results suggest that the catalytic properties of Co–Mo–S
on the Mo-edge and S-edge are considerably different.
4.3. Local structure and intrinsic activity of Co–Mo–S
Table 4 shows that the structural parameters of Co–Mo–S on
Co/Mo/Si (H2S/H2 ? H2S) are very similar to those for Co/Mo/Si
(H2S). Furthermore, the parameters for Co/Mo/Si (H2S ? H2S/H2)
are almost identical with those reported previously [11] for Co/
Mo/Si (H2S/H2) sulfided at 873 K. These findings clearly demon-
strate that the local structure of Co–Mo–S is determined by or, pos-
sibly, equilibrated with the second sulfidation atmosphere, so long
as there are no strong interactions between MoS2 particles and the
support or MoO2. On the other hand, the local structure of Co–Mo–
S, in particular, the CN of Co–S for Co/Mo/Al (H2S) is considerably
smaller than that of the SiO2-supported counterparts. In addition,
the CN of Co–S for Co/Mo/Si (H2S/H2) was 4.5 [11] and slightly
smaller than that for Co/Mo/Si (H2S ? H2S/H2) or Co/Mo/Si(ATM)
(H2S/H2). It is concluded that the MoS2-support (or MoO2) interac-
tions strongly modify the edge structure and thereby the intrinsic
activity of Co–Mo–S, possibly, due to increased metal-sulfur bond
energies [45]. Our previous magnetic study on Co–Mo–S [11] has
shown that the local structure and intrinsic activity of Co–Mo–S
are affected by such strong interactions through the electronic
modifications of MoS2 particles and that the difference between
Co–Mo–S Type I and Co–Mo–S Type II is caused by the MoS2-sup-
port interactions.
In our previous study [11], we have pointed out that the Co–S-
Co angle is slightly increased by the transformation of Co–Mo–S
Type I to Co–Mo–S Type II. We estimated the Co–S-Co angle from
the structural parameters given in Table 4 to be 91.5 0.1° for
Co–Mo–S Type III, which is identical with the angle 91.4 0.1°
for Type II and is larger than the angle 90.3° for Type I.
In order to examine the nature of ‘‘less active Co–Mo–S”, we
carefully analyzed the XANES spectra presented in Fig. 3. It is well
established that separate Co sulfide clusters show much lower
activity for HDS than does Co–Mo–S [4,5,46]. Accordingly, it is con-
sidered that Co sulfide clusters might be the strongest candidate as
‘‘less active Co–Mo–S”. Hence, we tried to examine whether the
XANES spectrum for Co/Mo/Si (H2S/H2 ? H2S), which is proposed
to contain 35% of Co as ‘‘less active Co–Mo–S” and 65% as ‘‘active
Co–Mo–S”, can be simulated using the XANES spectrum for
Co/Mo/Si (H2S) shown in Fig. 3, which has exclusively ‘‘active
As presented in Table 1 and illustrated in Fig. 7, the TOF values
of Co/Mo/Si (H2S) and Co/Mo/Si(ATM) (H2S) are significantly differ-
ent from that of Co/Mo/Si (H2S/H2 ? H2S), despite of their very
similar local structures of Co–Mo–S. Co/Mo/Si(ATM) (H2S/H2) and
Co/Mo/Si (H2S ? H2S/H2) provide another such pair. With all these
catalysts, the effect of support can be neglected due to the com-
plete sulfidation of Mo as discussed above. In order to solve these
discrepancies, we propose in the present study that the intrinsic
activity of Co–Mo–S strongly depends on which edge of MoS2 par-
ticles it is located, that is, the intrinsic activities of Co–Mo–S on the
Mo-edge and S-edge are significantly different for HDS. As dis-
cussed in Section 4.1., we propose that Co–Mo–S is fully situated
on only one of the edges of MoS2 particles for Co/Mo/Si which
has been sulfided in H2S/He before the Co addition. On the other
hand, Co–Mo–S is fully formed on both edges of MoS2 particles
and, accordingly, the TOF is averaged over the TOF values of Co–
Mo–S on Mo-edge and S-edge for Co/Mo/Si which has been sulfid-
ed in H2S/H2 before the exposure to Co(CO)3NO vapor. The findings
in Table 1 that Co/Mo/Si (H2S) and Co/Mo/Si(ATM) (H2S) show sig-
nificantly higher TOF than Co/Mo/Si (H2S/H2) sulfided at 873 K or
Co/Mo/Si(ATM) (H2S/H2) (Type II catalysts) prompt us to propose
that Co–Mo–S situated on one of the edges of MoS2 particles, pos-
sibly on the Mo-edge, exhibits a remarkably higher intrinsic activ-
ity for the HDS of thiophene and DBT than does Co–Mo–S on the
other edge, possibly the S-edge. Such an active Co–Mo–S structure
is denoted simply as ‘‘active Co–Mo–S” hereafter. With Co/Mo/Si
sulfided first in H2S/H2, the TOF given in Table 1 is an average
TOF over ‘‘active” and ‘‘less active” Co–Mo–S structures. When
Fig. 9. Correlation between the TOF on ‘‘active Co–Mo–S” for the HDS of thiophene
and the average coordination number of Co–S bonds over ‘‘active” and ‘‘less active”
Co–Mo–S. Closed circles, Co/Mo/Si (H2S/H2) (673 K and 873 K) and Co/Mo/Si(ATM)
(H2S/H2); triangles, Co/Mo/Al (H2S/H2) and (H2S); and open circles, Co/Mo/Si (H2S),
Co/Mo/Si (H2S ? H2S/H2), Co/Mo/Si (H2S/H2 ? H2S), and Co/Mo/Si(ATM) (H2S).
Refer to Table 5 for the CN of Co–S and the TOF on ‘‘active Co–Mo–S”.

58
Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
Co–Mo–S”, and the spectrum for Co/Al (H2S/H2) prepared by the
CVD [11,46], which has only highly dispersed Co sulfide clusters.
We synthesized a XANES spectrum by adding 65% of the spectral
intensity for Co/Mo/Si (H2S) and 35% of that for Co/Al (H2S/H2).
The convoluted spectrum is compared in Fig. 10 with that for Co/
Mo/Si (H2S/H2 ? H2S). The difference spectrum is also shown in
the inset for clarity. The simulated spectrum is slightly but dis-
tinctly different from the XANES spectrum for Co/Mo/Si (H2S/
H2 ? H2S), showing that ‘‘less active Co–Mo–S” is not ascribed to
Co sulfide clusters, but to Co–Mo–S. The analysis of the difference
spectrum between Co/Mo/Si (H2S/H2 ? H2S) and Co/Mo/Si (H2S) in
the inset of Fig. 10 suggested that the proportion of Co sulfide clus-
ters in the former catalyst is, if any, much less than 10% of the total
amount of Co atoms incorporated by the CVD. Besides, a similar
simulation of the XANES spectrum for Co/Mo/Si (H2S ? H2S/H2)
using the spectra for Co/Mo/Si (H2S) and Co/Al (H2S/H2) clearly
indicated that ‘‘active Co–Mo–S” once formed in H2S/He is not de-
graded to Co sulfide clusters during the successive sulfidation in
H2S/H2 at 673 K. With the Al2O3-supported catalysts, it was diffi-
cult from a similar XANES analysis to estimate the possible fraction
of Co sulfide clusters because of broadened nature of the XANES
spectra. However, it cannot be ruled out that a small fraction of
Co sulfide clusters is present on Co/Mo/Al and thus somewhat re-
duces the TOF of Co/Mo/Al.
We analyzed EXAFS Fourier transforms assuming a single Co–S
peak and hence the CN of Co–S is modified by the presence of the
five-six Co–S bonds with close distances for Co–Mo–S on the Mo-
edge and S-edge (Fig. 8). Hence, the absolute CN of Co–S of ‘‘active
Co–Mo–S” may be subject to small changes ( 0.5 from the error bar
estimation). In addition, the local structures of Co–Mo–S of ‘‘active”
and ‘‘less active” Co–Mo–S cannot be discriminated in the present
EXAFS analysis (‘‘average CN of Co–S” in Fig. 9). However, the rel-
ative CN obtained in the same series of the EXAFS analysis is fairly
reliable and consistent with the pre-edge intensity of XANES
(Fig. 3B). As a consequence, we believe that the correlation in
Fig. 9 shows a distinct structure–reactivity relationship with the
Co–Mo–S structure for HDS.
In our previous study [11] on the effect of the sulfidation tem-
perature on the intrinsic activity of the Co/Mo catalysts sulfided in
a conventional H2S/H2 stream, we have proposed that the local
structures of Co–Mo–S Type I and Co–Mo–S Type II are essentially
identical and that the higher intrinsic activity of Co–Mo–S Type II
is ascribed to the absence of the interactions between MoS2 par-
ticles and the support (or MoO2) and thereby to the change in
the electronic state of dinuclear Co sulfide clusters on the edge
of MoS2 particles. The electronic modifications of Co–Mo–S by
the support have been substantiated by the magnetic property
of Co. The effect of the support has been discussed in the same
framework [11]. It is considered that the increase of the CN of
Co–S by the shift from Co–Mo–S Type I to Co–Mo–S Type II (Figs.
7 and 9) reflects the change of the adsorptive properties of Co–
Mo–S toward sulfur to form S-dimers bonded to Co–Mo–S, the
change which is induced by the electronic modifications of Co
due to the interactions with the support. As discussed above, both
Co–Mo–S Type III and Co–Mo–S Type II have no strong interac-
tions with the support. However, the number of S-dimers on the
Type II catalysts (Co/Mo/Si (H2S/H2) sulfided at 873 K and Co/
Mo/Si(ATM) (H2S/H2)) is considerably limited because of the ther-
modynamic equilibration with the gas phase (10% H2S/H2 at 873 K
and 673 K, respectively), while much more abundant S-dimers are
thermodynamically allowed to form on Co–Mo–S Type III. In other
words, the concept of ‘‘Co–Mo–S Type II”, which has been tacitly
defined as Co–Mo–S formed at usual H2S/H2 conditions and with-
out any strong interactions with the support, can be extended to
include ‘‘Co–Mo–S Type III” formed abundantly at strongly sulfid-
ing conditions. It is concluded in the present study that highly ac-
tive Co–Mo–S Type III is characterized by a dinuclear Co sulfide
cluster [11,16] with a S-dimer between two Co atoms and situated
on the Mo-edge of MoS2 particles, as illustrated in Fig. 8. Recently,
Krebs et al. [26] suggested a Co–S–S–Co dinuclear structure for
100% promoted Co–Mo–S on the Mo-edge with a 50% coverage
of S at a high PH S=PH
.
2
2
The formation of the Co–(S2)–Co structure illustrated in Fig. 8 is
thermodynamically favored only under the strongly sulfiding
atmosphere (high chemical potential of sulfur). According to DFT
calculations [20,36,39], the sulfur atoms are stabilized as dimers
on the fully sulfided edge of MoS2 (100% S coverage). Besides, it
is proposed by use of DFT that H2 molecules are easily activated
on S-dimer species [38,39]. It is concluded that the significantly
high intrinsic activity of Co–Mo–S Type III is correlated to the ab-
sence of interactions between MoS2 particles and the support
and to the presence of S-dimers with a high probability in the
Co–Mo–S structure for hydrogen activation. The structure–reactiv-
ity relationship presented in Fig. 9 verifies the second conclusion.
5. Conclusions
In the present study, we investigated the effect of the sulfida-
tion atmosphere on the local structure and intrinsic HDS activity
of Co–Mo–S supported on SiO2. We employed the sulfidation pro-
cedures of the Co–Mo catalysts at 673 K in 10% H2S/H2, 10% H2S/
He, and their combinations to vary the sulfidation atmosphere.
We characterized the sulfide catalysts by TEM and XAFS. The sali-
ent findings in the present study are as follows:
Fig. 10. A comparison of the Co K-edge XANES spectrum (solid line) for Co/Mo/Si
(H2S/H2 ? H2S) with a spectrum (dotted line) simulated using the XANES spectra
for Co/Al (H2S/H2) (35%) and Co/Mo/Si (H2S) (65%). The latter two spectra are shown
for clarity. The inset shows the difference XANES spectra between Co/Mo/Si (H2S/
ꢃ The intrinsic activity of Co–Mo–S is strongly affected by the sulf-
idation atmosphere. The presulfidation of Co–Mo–S in a strongly
sulfiding atmosphere (H2S/He) forms the Co–Mo–S structure
with a high intrinsic activity for the HDS of thiophene and DBT.
H2 ? H2S) and simulated one (dotted line) and between Co/Mo/Si (H2S/H2 ? H2S)
and Co/Mo/Si (H2S) (solid line).

Y. Okamoto et al. / Journal of Catalysis 268 (2009) 49–59
59
[14] S.M.A.M. Bouwens, J.A.R. van Veen, D.C. Koningsberger, V.H.J. de Beer, R. Prins,
J. Phys. Chem. 95 (1991) 123.
[15] S.P.A. Louwers, R. Prins, J. Catal. 133 (1992) 94.
[16] Y. Okamoto, M. Kawano, T. Kawabata, T. Kubota, I. Hiromitsu, J. Phys. Chem. B
109 (2005) 288.
[17] H. Topsøe, Appl. Catal. A 322 (2007) 3.
[18] P. Raybaud, Appl. Catal. A 322 (2007) 76.
[19] J.-F. Paul, S. Cristol, E. Payen, Catal. Today 130 (2008) 139.
[20] L.S. Byskov, J.K. Nørskov, B.S. Clausen, H. Topsøe, J. Catal. 187 (1999) 109.
[21] H. Schweiger, P. Raybaud, H. Toulhoat, J. Catal. 212 (2002) 33.
ꢃ It is proposed that Co–Mo–S on one of the edges of MoS2 parti-
cles, possibly on the Mo-edge, is highly active for the HDS, while
Co–Mo–S on the other edge, possibly on the S-edge, is less
active.
ꢃ The local structure of the highly active Co–Mo–S formed in H2S/
He is described as a dinuclear Co sulfide cluster with a sulfur-
dimer between the two Co atoms.
ꢃ The intrinsic activity of the ‘‘active Co–Mo–S” increases as the
coordination number of Co–S bond increases. A structure–reac-
tivity relationship is proposed with the Co–Mo–S structure for
HDS, accounting for the effects of the sulfidation temperature
and atmosphere as well as the effect of the support on Co–Mo
sulfide catalysts.
ꢃ We demonstrate in the present study that the intrinsic HDS
activity of Co–Mo–S is strongly affected by the location and
the local structure, which is equilibrated with the sulfidation
atmosphere or reaction conditions, as well as by the MoS2-sup-
port interactions.
[22] J.V. Lauritsen, S. Helveg, E. L
ꢁgsgaad, I. Stengaad, B.S. Clausen, H. Topsøe, F.
Besenbacher, J. Catal. 197 (2001) 1.
[23] J.V. Lauritsen, R.T. Vang, F. Besenbacher, Catal. Today 111 (2006) 34.
[24] J.V. Lauritsen, J. Kibsgaad, G.H. Olesen, P.G. Moses, B. Hinnemann, S. Helveg, J.K.
Nørskov, B.S. Clausen, H. Topsøe, E. L
(2007) 220.
ꢁgsgaad, F. Besenbacher, J. Catal. 249
[25] M. Sun, A.E. Nelson, J. Adjaye, J. Catal. 226 (2004) 32.
[26] E. Krebs, B. Silvi, P. Raybaud, Catal. Today 130 (2008) 160.
[27] A.D. Gandubert, E. Krebs, C. Legens, D. Costa, G. Guillaume, P. Raybaud, Catal.
Today 130 (2008) 149.
[28] Y. Okamoto, K. Ochiai, M. Kawano, K. Kobayashi, T. Kubota, Appl. Catal. A 226
(2002) 115.
[29] Y. Okamoto, S. Ishihara, M. Kawano, M. Satoh, T. Kubota, J. Catal. 217 (2003) 12.
[30] Y. Okamoto, T. Kubota, Catal. Today 86 (2003) 31.
[31] Y. Okamoto, Catal. Today 132 (2008) 9.
[32] <http://leonardo.phys.washington.edu/feff/>.
[33] J.J. Rehr, J.M. de Leon, S.I. Zabinsky, R.C. Albers, J. Am. Chem. Soc. 113 (1991)
5135.
References
[34] S. Kasztelan, H. Toulhoat, J. Grimblot, J.P. Bonnelle, Appl. Catal. 13
(1984) 127.
[35] P. Raybaud, J. Hafner, G. Kresse, S. Kasztelan, H. Toulhoat, J. Catal. 189 (2000)
129.
[36] H. Schweiger, P. Raybaud, G. Kresse, H. Toulhoat, J. Catal. 207 (2002) 76.
[37] S. Cristol, J.F. Paul, E. Payen, D. Bougeard, S. Clemendot, F. Hutsehka, J. Phys.
Chem. B 104 (2000) 11220.
[38] S. Cristol, J.F. Paul, E. Payen, D. Bougeard, S. Clemendot, F. Hutsehka, J. Phys.
Chem. B 106 (2002) 5659.
[39] M.V. Bollinger, K.W. Jacobsen, J.K. Nørskov, Phys. Rev. B 67 (2003) 085410.
[40] B. Scheffer, P. Arnoldy, J.A. Moulijn, J. Catal. 112 (1998) 516.
[41] P. Arnoldy, J.A.M. van den Heijkant, G.D. de Bok, J.A. Moulijn, J. Catal. 92 (1985)
35.
[42] H.R. Reinhoudt, A.D. van Lahgeveld, P.J. Kooyman, R.M. Stockmann, R. Prins,
H.W. Zandbergen, J.A. Moulijn, J. Catal. 179 (1998) 443.
[43] P.J. Kooyman, E.J.M. Hensen, A.M. de Jong, J.W. Niemantsverdriet, J.A.R. van
Veen, Catal. Lett. 74 (2001) 49.
[1] K.G. Knudsen, B.C. Cooper, H. Topsøe, Appl. Catal. A 189 (1999) 205.
[2] C. Song, Catal. Today 86 (2003) 211.
[3] D. Whitehurst, T. Isoda, I. Mochida, Adv. Catal. 42 (1998) 345.
[4] S. Eijsbouts, Appl. Catal. A 158 (1997) 53.
[5] H. Topsøe, B.S. Clausen, F.E. Massoth, in: J.R. Anderson, M. Boudard (Eds.),
Catalysis Science and Technology, vol. 11, Springer, Berlin, 1996, p. 1.
[6] H. Topsøe, B.S. Clausen, Catal. Rev. – Sci. Eng. 26 (1984) 395.
[7] H. Topsøe, B.S. Clausen, N. Topsøe, E. Pedersen, Ind. Eng. Chem. Fundam. 25
(1986) 25.
[8] R. Candia, O. Sørensen, J. Villadsen, N. Topsøe, B.S. Clausen, H. Topsøe, Bull. Soc.
Chim. Belg. 93 (1984) 763.
[9] J.A.R. van Veen, H.A. Colijn, P.A.J.M. Hendriks, A.J. van Welsenes, Fuel Proc.
Technol. 35 (1993) 137.
[10] S.M.A. Bouwens, F.B.M. van Zon, M.P. van Dijk, A.M. van der Kraan, V.H.J. de
Beer, J.A.R. van Veen, D.C. Koningsberger, J. Catal. 146 (1994) 374.
[11] Y. Okamoto, A. Kato, Usman, N. Rinaldi, T. Fujikawa, H. Koshika, I. Hiromitsu, T.
Kubota, J. Catal. 265 (2009) 216.
[12] J.A.R. van Veen, E. Gerkema, A.M. van der Kraan, A. Knoester, J. Chem. Soc.,
Chem. Commun. (1987) 1684.
[44] Y. Okamoto, T. Imanaka, J. Phys. Chem. 92 (1988) 7102.
[45] B. Hinnemann, J.K. Nørskov, H. Topsøe, J. Phys. Chem. B 109 (2005) 2245.
[46] Y. Okamoto, K. Ochiai, M. Kawano, T. Kubota, J. Catal. 222 (2004) 143.
[13] W. Niemann, B.S. Clausen, H. Topsøe, Catal. Lett. 4 (1990) 355.