Elucidation of sulfidation state and hydrodesulfurization mechanism on ruthenium-cesium sulfide catalysts using 35S radioisotope tracer methods

Article abstract of DOI:10.1016/S0021-9517(03)00031-9

A number of approaches aiming at performing deep HDS of light gas oils have been studied. For deep HDS reactions, the development of very efficient new HDS catalysts, with knowledge of the active phase and the HDS reaction mechanism, is required. Alumina-supported Ru-Cs catalysts were persulfided using [³⁵S]H₂S pulse tracer method to evaluate their sulfidation state. HDS reactions of dibenzothiophene were performed and the mobility of ³⁵S introduced during the presulfidation state was studied using these previously ³⁵S-labeled catalysts. The amount of labile sulfur was much smaller than the total amount of sulfur accommodated on the catalyst (S_total). The labile sulfur was not formed on RuCs catalysts until the HDS reaction proceeded. S_total increased with increasing Cs content. Labile sulfur was difficult to form over the Ru-Cs/Al₂O₃ catalysts during the presulfidation stage.

Full text of DOI:10.1016/S0021-9517(03)00031-9

Journal of Catalysis 217 (2003) 59–68
www.elsevier.com/locate/jcat
Elucidation of sulfidation state and hydrodesulfurization mechanism on
3
5
ruthenium–cesium sulfide catalysts using S radioisotope tracer methods
∗
Atsushi Ishihara, Jeayoung Lee, Franck Dumeignil, Ryuichirou Higashi, Anjie Wang,
Eika W. Qian, and Toshiaki Kabe
Department of Chemical Engineering, Tokyo University of Agriculture & Technology, Nakacho, Koganei, Tokyo 184-8588, Japan
Received 12 August 2002; revised 26 December 2002; accepted 23 January 2003
Abstract
3
5
Alumina-supported ruthenium–cesium catalysts were presulfided using [ S]H S pulse tracer method to evaluate their sulfidation state.
2
3
5
Subsequently, using these previously S-labeled catalysts, HDS reactions of dibenzothiophene (DBT) were performed and the mobility of
S introduced during the presulfidation stage was investigated. The results showed that the amount of labile sulfur (S ) was much smaller
3
5
0
than the total amount of sulfur accommodated on the catalyst (S
). DBT conversion and S
increased linearly with Ru content. In a
total
total
second part, labile sulfur amount was also determined under the catalyst working conditions and different results were obtained. Indeed,
3
5
35
when the catalysts were marked with [ S] and with [ S]DBT under HDS reaction conditions, the obtained labile sulfur quantities (S
)
0
A
were significantly higher than the ones measured during the presulfidation stage (S ). These results showed that the labile sulfur is not formed
0
on RuCs catalysts until the HDS reaction proceeds, which is quite different from that reported before for Mo, Pt, or Pd systems.

2003 Elsevier Science (USA). All rights reserved.
35
Keywords: Hydrodesulfurization; Dibenzothiophene; S; Radioisotope; Tracer; Ruthenium; Cesium; Sulfide; Alumina
1
. Introduction
and the catalysts derived from supported metal carbonyls
showed a catalytic activity higher than that of convention-
ally prepared catalysts [21,22], which was also observed by
some authors of the present paper [10,23]. Furthermore, ce-
sium hydroxide addition during the preparation ruthenium
carbonyl-based catalysts showed a remarkable increase in
catalytic activity [24–27].
In big city areas the problem of air pollution due to
diesel engine fumes has become more and more serious.
Therefore, a number of approaches aiming at performing
deep hydrodesulfurization (HDS) of light gas oils have been
investigated. For deep HDS reactions, the development of
very efficient new HDS catalysts, with knowledge of the
active phase as well as the HDS reaction mechanism [1] is
needed.
Among a large number of attempts, much attention has
been focused on unsupported ruthenium sulfides, which
were found to be the most active in the hydrodesulfur-
ization of thiophenes using transition metal sulfides [2–6].
Some research groups have performed HDS reactions cat-
alyzed by ruthenium sulfide supported on alumina [7–10]
carbon [11–13], zeolites, [14–18], MgF2, [19], and amor-
phous silica–alumina [20]. However, supported ruthenium
sulfide activity was not always so high [8]. Preparation con-
ditions were found to play an important role in the activity,
For more conventional catalysts, the necessity and/or in-
fluence of presulfiding parameters has been extensively re-
ported [28,29]. Using conventional investigation methods,
oxidation states before and after presulfiding can be ob-
tained. However, catalyst sulfidation state is very unstable,
and it appears necessary to investigate the working cata-
lyst in situ to obtain reliable structural information. There-
fore, in order to clarify the sulfur behavior on the work-
35
ing catalysts during practical HDS conditions, S radioiso-
tope tracer methods were developed, giving access to quan-
35
titative data [30–53]. In the present paper, S radioisotope
tracer methods were used to study the particular case of
Ru-based HDS catalysts. Indeed, under a H2 atmosphere
it is known that sulfided Ru species can be reduced (giv-
*
ing H S release), and a decrease in HDS activity is ob-
Corresponding author.
2
E-mail address: atsushii@cc.tuat.ac.jp (A. Ishihara).
served. Ru-based HDS catalyst sulfidation state is therefore
0021-9517/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9517(03)00031-9

60
A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
a sensitive parameter which can vary under different ex-
perimental conditions. We decided therefore to check the
difference between the sulfidation state after the presulfi-
dation stage and the sulfidation state of the working cat-
alysts. For this purpose two different experimental meth-
ods were carried out. In the first one, catalysts were pre-
Subsequently, the crucible was charged into an autoclave
(Nittoh Kouatsu Co., Ltd., 367 mL). After H purge, the
2
2
autoclave was pressurized with H to 50 kg/cm and heated
2
◦
at 430 C, and this temperature was kept for 1 h.
After cooling to room temperature, the autoclave was
3
5
further cooled into liquid nitrogen to solidify [ S]H S.
2
3
5
sulfided using [ S]H2S and sulfidation state was estimated
quantitatively (quantity of accommodated sulfur). Then, us-
ing the obtained 35S-labeled catalysts, the HDS reaction of
DBT was performed to estimate the amount of labile sul-
fur generated under the presulfidation conditions (i.e., de-
Unreacted hydrogen was then completely removed using a
3
5
vacuum pump and 100% gaseous [ S]H S was collected
2
by returning the autoclave to room temperature. Then, by
adding hydrogen in a proper quantity, the mixed gas used
for the presulfiding experiments (about 30% of H S in H )
2
2
3
5
termination of released [ S]H2S quantity during the DBT
HDS reaction). In the second experimental procedure, DBT
HDS reactions of 3 S-labeled DBT (equivalent to a label-
was prepared. Accurate gas concentration was measured by
using a GC-TCD apparatus (Hitachi 263) equipped with
a Porapack Q column (length, 3 m; id, 3 mm). Column,
injection, and detector temperatures were set to 180, 200,
5
3
5
ing of the active sites directly with [ S]DBT) were carried
out on the catalysts and the amount of labile sulfur was ob-
◦
and 200 C, respectively.
3
5
tained by quantifying the released [ S]H2S quantity when
3
2
feed was changed to [ S]DBT. Labile sulfur quantities ob-
tained by both methods were compared, and the obtained
differences were discussed. Further, results were compared
with those previously obtained for Mo, Pt, and Pd-based cat-
alysts [52,54].
2
.3. Measurement of surface area, pore volume, and pore
distribution
0
.1 g of presulfided sample was weighted and then
charged into the BET apparatus (Coulter Co. SA3100).
◦
After outgassing at 200 C for 2 h, a 10-point BET method
measurement was performed. The pore distribution was
obtained using the BJH method [55].
2
. Experimental section
Table 1 shows the surface area and the pore volume of the
catalysts. For the catalysts with Cs/Ru ratio equal to 1, when
the Ru content was gradually increased from 4 to 20 wt%,
2
.1. Preparation of catalysts
Ruthenium carbonyl and cesium hydroxide were pur-
2
specific surface area decreased from 246 to 144 m /g. Pore
chased from Kishida Chemicals Co., Ltd., and γ -Al2O3
was supplied from Nippon Ketjen Co., Ltd. γ -Al2O3 was
crushed to 20–30 mesh prior to use. An anionic ruthenium
carbonyl complex was prepared by reaction of Ru3(CO)12
with an excess of CsOH in methanol using Cs/Ru = 1–3. It
is well known that Ru3(CO)12 reacts with alkali metal hy-
droxides to give anionic ruthenium hydride complexes [25]
M[HRu3(CO)11] with M = Cs in the present paper. Then,
alumina was added into the prepared {Cs[HRu3(CO)11] +
unreacted CsOH} solution. After the mixture was stirred
for 2 h, the solvent was removed in vacuo. Various cat-
alysts were prepared by variation of the metal loading.
First, Ru content was varied from 4 to 20 wt% keeping the
Ru + alumina weight equal to 1 g, and Cs/Ru = 1. A sec-
ond series of catalysts was prepared by fixing the Ru load-
ing at 8 wt% and varying the Cs/Ru ratio from 0 to 3. Be-
cause the prepared catalysts are unstable under atmospheric
conditions, they were prepared and stored under argon at-
mosphere.
volume was quite constant for 4–8 Ru wt% but decreased
for higher Ru loading. Fig. 1 shows pore distribution curves.
Mean pore radius was found to be around 40 Å for all
the catalysts. Therefore, pore radius variation was not a
parameter that could explain the variations of DBT HDS
activity over the catalysts.
Catalysts with various Cs/Ru ratios exhibited a decrease
in pore volume and specific surface area (see the bottom
of Table 1). Fig. 2 shows the pore distribution curves for
the catalysts with Cs/Ru = 0–3. Mean pore radius was also
found to be about 40 Å for all the catalysts.
Table 1
BET surface area and pore volume of the catalysts
Ruthenium content
wt%)
(alumina)
Cs/Ru ratio
(mol/mol)
Surface area
(m /g)
Pore volume
(cm /g)
2
3
(
0
4
6
8
2
6
–
1
1
1
1
1
1
0
2
3
256
246
245
210
188
158
144
250
169
146
0.647
0.494
0.549
0.489
0.400
0.358
0.279
0.609
0.430
0.247
3
5
2
.2. Synthesis of [ S]H2S
1
1
20
8
8
8
3
5
35
S-labeled hydrogen sulfide ([ S]H2S) was prepared as
follows.
First, 32S and 35S were added into a crucible, to obtain
a total radioactivity of 4 MBq and total weight of 7 g.

A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
61
Fig. 2. Pore distribution of Ru–Cs/Al O catalysts.
2
3
Fig. 1. Pore distribution of Ru–Cs/Al O catalysts.
2
3
became equal to the introduced one (steady state), i.e., when
S uptake on the studied catalyst reached saturation. After the
presulfiding experiment, a liquid-gas separator was installed
under the reactor tube, and the DBT HDS reaction was car-
2
.4. Determination of (pre)sulfidation state of Ru–Cs
3
5
catalysts using [ S]H2S radioisotope pulse tracer method
A pressurized fixed-bed flow reactor was used for pre-
sulfiding experiments. Fig. 3 shows the reaction appara-
◦
ried out using the following parameters: 300 C, total pres-
2
−1
3
5
sure = 50 kg/cm , WHSV = 28 h , hydrogen flow rate =
2.5 L/h, with a 1 wt% DBT solution in decalin. Reaction
products in the liquid phase were analyzed by GC-FID.
tus used for the [ S]H2S pulse tracer method. 0.5 g of
catalyst was packed into the reactor. The mixed gas (30%
1
3
5
3
[
S]H2S/H2) was charged into a 5.04 cm loop line. By
3
5
using a 6-way valve, [ S]H2S/H2 mixed gas was period-
ically introduced into the reactor kept under a flow of N2
carrier gas (50 kg/cm , 5 L/h) at 100, 200, and 300 C. Af-
ter each pulse, the unreacted [ S]H2S/H2 was collected in a
trap placed at the outlet of the reactor (absorption into 10 mL
of a basic Carbosorb scintillation solution). The amount of
collected radioactivity after each pulse was then measured
by using a liquid scintillation counter. This operation was re-
peated until the amount of radioactivity in a recovered pulse
3
5
2
.5. Hydrodesulfurization of [ S]DBT on Ru–Cs catalysts
2
◦
3
5
35
The [ S]DBT HDS experimental procedure used in this
paper has been already described elsewhere [49] but we
will give a brief description of it in order to make the
experimental results easier to interpret for the reader. HDS of
3
5
[ S]DBT was carried out in a pressurized flow reactor, and
◦
typical reaction conditions were fixed as follows: 300 C,
35
Fig. 3. Experimental apparatus of S pulse tracer method.

62
A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
2
−1
35
total pressure = 50 kg/cm , WHSV = 28 h , hydrogen
flow rate = 25 L/h, with a 1 wt% DBT solution in decalin.
The catalysts were presulfided with a 3 vol% H2S/H2 gas
Catalysts presulfided with [ S]H2S by the method de-
scribed above were purged with decalin (further used as
a solvent for the reactant feed) before introduction of the
1 wt% DBT solution. The amount of radioactivity released
from the catalyst (in the liquid and the solid phase) was fol-
lowed during the HDS reaction. The radioactivity of S de-
tected during the purge with decalin was attributed to the
◦
mixture at 300 C during 3 h. In a first step, the HDS
3
2
reaction of DBT was carried out using a 1 wt% [ S]DBT
solution in decalin during 3 h. Then, when the conversion
3
5
3
2
of DBT became constant, [ S]DBT was replaced by 1 wt%
3
5
35
[
S]DBT and the reaction was pursued for 4–7 h (HDS sites
labeling with [ S]). Then, decalin solution was introduced
into the reactor for 3 h and the HDS reaction using [ S]DBT
was performed again for 4–7 h. [ S]H2S produced during
this last reaction was absorbed in a basic scintillator solution
and liquid products were analyzed in the same way as
described in [54]. The radioactivities of both unreacted
evacuation of physisorbed [ S]H2S species which dissolved
3
5
in the solvent. The physisorbed H2S quantity was found to be
very small when compared to the accommodated sulfur total
quantity. However, we decided to subtract it from the calcu-
lated accommodated sulfur quantity to calculate accurately
Stotal, which represented therefore the total chemisorbed S
quantity present on the catalyst surface after presulfidation.
Then, when DBT solution was introduced under a pressur-
ized hydrogen atmosphere, the HDS reaction was followed
by S detection in both liquid and gas phases (Fig. 5). Re-
leased radioactivity was regarded as the labile sulfur created
during the presulfidation step (S0). Obtained S0 values were
surprisingly much smaller than the total amount of sulfur
incorporated during the presulfiding step. That means that
only a very little amount of labile sulfur was created during
the presulfidation step. The global results are given in Figs. 6
and 7 and in Table 2. The conversion of DBT and the amount
of sulfur accommodated on the catalyst (Stotal) as well as S0
increased linearly with increasing Ru content (Fig. 6). As a
remark, it is well-known that Ru can be sulfided into RuS2,
that means that Ru on the catalysts was only sulfided up to
about 50% during the presulfidation stage.
3
2
3
5
3
5
35
[
S]DBT and formed [ S]H2S were measured with a
liquid scintillation counter (Beckman LS-6500, Beckman
Co., Ltd.) after a proper scintillation solvent (Permafluor
or Instafluor, Packard Japan Co., Ltd.) was added to each
sampled fraction. The liquid scintillation counting method
was explained in another paper [56] and detailed calculation
methods for accommodated sulfur quantity and labile sulfur
quantity are also given elsewhere [49].
3
5
3
. Results and discussion
3
.1. Sulfur quantitative analysis of the presulfided Ru–Cs
catalysts by a S radioisotope pulse tracer method
3
5
The effect of Cs on Ru sulfidation state was checked.
In a first part, we decided to check Cs–alumina solids. We
prepared a series of Cs/Al2O3 catalysts and presulfiding
Fig. 4 shows the changes in recovered radioactivity after
3
5
each [ S]H2S pulse introduction. Each box on the graph
3
5
represents the radioactivity of [ S]H2S released from the
reactor after each pulse. The shaded part represents therefore
the radioactivity of sulfur incorporated into the catalyst.
Then, the total amount of sulfur (Stotal) accommodated at
each temperature can be easily deduced.
3
5
[
S]H2S experiments were performed as described above.
Fig. 8 shows the variation of the total amount of sulfur
accommodated on the catalysts with Cs content. With
increasing Cs content, the total amount of sulfur (Stotal)
incorporated into Cs/Al2O3 catalysts increased linearly in
35
35
Fig. 4. Sulfidation of 8 wt% Ru–Cs/Al O with [ S]H S pulses (Ru:Cs =
Fig. 5. Changes in radioactivity of S with reaction time (8 wt% Ru–Cs/
Al2O3 (Ru:Cs = 1:3)).
2
3
2
1:3).

A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
63
Fig. 6. Effect of Ru content on DBT conversion and S
.
total
Fig. 8. Change in total sulfur with cesium content (Ru–Cs/Al O ,
2
3
Cs/Al O ).
2
3
3
.2. Effect of Ru/Cs ratio on the Ru–Cs catalyst sulfidation
state
We have already reported that a significant increase in
the catalytic activity is observed when the Cs/Ru ratio
increases [25]. We decided therefore to perform tracer
experiments on solids prepared with various Cs/Ru ratios.
Results are summarized in Figs. 9 and 10, and Table 3. The
DBT conversion and Stotal changed only a little when the
Cs/Ru ratio increased from 0 to 1. However, a further Cs/Ru
ratio increase led to an increase of Stotal (Fig. 9). Using FTIR
experiments, we have shown in a previous paper [25] that for
Cs/Ru = 1, created Cs[HRu3(CO)11] species are destroyed
during the impregnation stage. Indeed, HRu3(CO)12 species
were detected on the surface of the catalysts, indicating that
Cs reacted with the acid sites of the alumina. Therefore, in
such a system, Ru-containing particles were isolated from
the Cs-containing ones before the presulfidation stage and
a synergetic effect was not observed. Indeed, DBT HDS
activity of the solid prepared with Cs/Ru = 1 was almost
the same as the one observed over the corresponding Cs-free
catalyst. Results were different for higher loading of Cs
Fig. 7. Change in amount of sulfur with Ru content in Ru–Cs/Al O
catalysts (Ru/Cs = 1).
2
3
parallel with the one observed for Ru–Cs/Al2O3 catalysts.
It was observed that approximately 60–82% of the total
amount of sulfur accommodated on Ru–Cs/Al2O3 catalysts
was equivalent to Stotal of Cs/Al2O3 catalysts (for the same
Cs loading). That means that a considerable amount of Stotal
in the Ru–Cs/Al2O3 catalysts was linked to the presence of
Cs in the system.
Table 2
Kinetic parameters on Ru–Cs/Al O
2
3
Ru
wt%)
Conversion
(%)
S
S
S
0A
(mg/g-cat)
total
0
(
(mg/g-cat)
(mg/g-cat)
4
8
2
6
0
28
54
91
94
89
12 (RuS
28 (RnS
41 (RuS
46 (RuS
66 (RuS
)
)
)
)
)
1.0
2.8
3.3
4.3
3.7
7.3 (RuS
12.9 (RuS
20.8 (RuS
37.9 (RuS
45.0 (RuS
)
0.94
1.11
1.08
0.91
1.04
057
)
)
)
)
0.51
0.55
0.75
0.71
1
1
2
(Ru/Cs = 1).
Fig. 9. Change in conversion and S
with Cs/Ru (8 wt% Ru–Cs systems).
total

64
A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
Fig. 10. Change in S
and S with Cs content in Ru–Cs/Al O catalysts
0 2 3
total
(Ru 8 wt%).
Fig. 11. Change in S
with Cs content [Ru–Cs/Al O (Ru 8 wt%,
2 3
total
Cs/Ru = 0–3), Cs/Al O ].
2
3
Table 3
Kinetic parameters on Ru–Cs/Al O
2
3
both Ru–Cs/Al O and Cs/Al O increased approximately
2
3
2
3
linearly with increasing Cs content. It was confirmed that a
considerable amount of Stotal in Ru–Cs/Al O catalysts was
related to the presence of Cs.
Ru:Cs
Conversion
%)
S
S
S
0A
(mg/g-cat)
9.2 (RuS
0.36
18.3 (RuS
38.5 (RuS
39.3 (RuS
total
0
(
(mg/g-cat)
(mg/g-cat)
2
3
1:0
1:1
1:2
1:3
50
54
94
97
31.4 (RuS
)
)
)
)
2.2
2.8
14.7
22.0
)
)
)
)
1
1
1
2
.27
27.5 (RuS
40.4 (RuS
75.1 (RuS
In brief, a direct interaction between Cs and Ru was
.11
.63
.72
0.72
1.52
1.55
responsible for an increase in S , inducing an increase
0
in the observed HDS activity, and Cs presence induced
an increase in Stotal, supposedly by the creation of Cs–S
Ru 8 wt%.
bond-containing species.
for which Cs–Ru interaction was preserved. Indeed, FTIR
−
results show that the [HRu3(CO)11] structure was present
3.3. Comparison of the amount of labile sulfur on the
presulfided catalysts and on the working catalysts
([ S]DBT HDS reaction)
for Cs/Ru = 2, indicating that this ratio is enough to
saturate alumina acid sites with Cs atoms, leaving a part
of the prepared complexes intact on the alumina surface.
While during the presulfidation stage the complex structure
was broken, it is likely that Ru to Cs atom proximity
was preserved. It is therefore reasonable to postulate that
Ru–Cs was responsible for the increase in HDS activity and
Stotal observed in Fig. 9. For Cs/Ru = 3, FTIR analysis
showed that anionic species were also maintained on the
3
5
In part 1 labile sulfur quantity created during the presul-
fidation stage was found to be very low, which could not
explain the high activity of the catalytic system. Therefore,
the amount of labile sulfur formed under working HDS con-
ditions was checked. This amount (S0A) evaluated in HDS
3
5
of [ S]DBT [49] was found to be very high when compared
−
35
surface of the catalysts, but not in the [HRu3(CO)11] form.
to S observed after presulfiding reaction with [ S]H S.
0
2
Nevertheless, Cs–Ru interaction was preserved, which is
linked with further increase in Stotal observed in Fig. 9.
Moreover, for Cs/Ru = 3, the observed value of Stotal
suggested that there was a lot of sulfur existing not only as
RuS2 but also as CsSH.
In addition, the amount of labile sulfur increased drasti-
cally for Ru/Cs > 1 (Fig. 10), i.e., when FTIR data suggest
that the direct interaction between Cs and Ru species in the
Cs[HRu3(CO)11] complex is preserved after impregnation.
This suggests that Cs strengthens the Ru–S bond of the Ru
sulfide species, which leads to an increase in S0 and there-
fore to an increase in activity.
Further, the total sulfur amount accommodated on Cs/
Al2O3 catalysts was compared with the one accommodated
on Ru–Cs catalysts with various Cs/Ru ratios. Fig. 11
shows the variation of (Stotal) for Ru–Cs and Cs catalysts.
Except for Ru/Al2O3 catalyst (i.e., Cs free), Stotal of
Fig. 12 shows the change in the radioactivities of pro-
3
5
35
duced [ S]H S and unreacted [ S]DBT during the HDS
2
3
5
32
reaction of [ S]DBT. After the HDS of [ S]DBT reached
the steady state, the reaction solution of [ S]DBT was re-
placed by that of [ S]DBT. The radioactivity of unreacted
3
2
3
5
3
5
[ S]DBT in the outlet of the reactor increased immediately
and reached a steady state. However, the radioactivity of pro-
3
5
duced [ S]H S showed a delay to reach the steady state.
2
This result indicates that the sulfur in DBT is not imme-
diately released as hydrogen sulfide, but accommodated on
the catalysts, as observed on conventional catalysts [42,59].
In order to determine the amount of labile sulfur, after
3
5
the radioactivity of [ S]H S reached the steady state, de-
2
calin only was fed into the reactor. The radioactivity of
[ S]DBT decreased immediately, while the radioactivity of
3
5
3
5
[ S]H2S gradually decreased. After the radioactivity of re-
3
5
32
leased [ S]H2S reached a value of 0 dpm, a [ S]DBT solu-

A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
65
Fig. 13. Effect of Ru content on DBT conversion and S
.
total
experimental results and models previously developed for
other catalysts, we proposed a structure for Ru–Cs catalysts
in Fig. 14. For the catalysts with Cs/Ru ratios equal to 2
or 3 (Fig. 14a), Cs and Ru interaction (preserved as shown
by FTIR experiments) strengthens the Ru–S bond. Since
the dispersion of Ru was expected to be very high (linear
increase of the HDS activity up to 16 wt% Ru), a MoS2-
like single-layer structure was applied [57,58]. Four kinds
3
5
35
Fig. 12. Change in radioactivity of [ S]DBT and [ S]H S.
2
tion was fed into the reactor. The 35S accommodated on the
catalyst during [ S]DBT HDS reaction was exchanged with
S released by [ S]DBT HDS and [ S]H2S was detected
3
5
3
2
32
35
3
2
again. When the [ S]DBT solution was introduced, the ac-
tivities of the catalysts with high Ru contents decreased to
about 70% of the original activity. This reflects the fact that
since Ru sulfide is unstable under high H2 pressure, sulfur
in the catalyst reacted with hydrogen during the decalin in-
troduction step (no sulfur was present in the feed) and was
released as hydrogen sulfide (Ru reduction). Therefore, S0A
were evaluated by the addition of S0B (sulfur released by re-
duction when no sulfur is fed into the reactor) and S0C (sul-
fur exchanged by [ S]DBT reaction) (see Fig. 12). Fig. 13
shows the changes in DBT conversion and S0A with Ru con-
tent for catalysts with Cs/Ru = 1. Both values increased
with increasing Ru content but conversion exhibited a slight
decrease for Ru > 16%. From these results, it was suggested
that Ru species are uniformly dispersed on the alumina at
least up to 16 wt% Ru. S0A variation was also determined
for catalysts with different Cs/Ru ratios. S0A of the latter are
shown in Table 3, in which we reported also the correspond-
ing S0. While for Ru–Cs/Al2O3 catalysts with Cs/Ru = 1,
S0A given as the number of labile sulfur for 1 Ru atom was
a
b
c
d
of sulfur, S , S , S , and S can exist in the proposed
a
structure. The S emerging from the catalyst surface might
b
be easy to remove while the S in contact with alumina
and covered with Ru might not be easy to be involved
c
into a reaction as they are difficult to approach. S and
d
a
b
S can be considered to be of a type between S and S .
Therefore, in the best case 75% of the accommodated sulfur
can move, which is consistent with the experimental results.
For the catalysts with Cs/Ru = 1, impregnated complex is
destroyed (FTIR results), and the structure of the catalysts
is very likely to be as the one presented in Fig. 14b. Indeed,
all the introduced Cs is consumed to neutralize the alumina
3
2
b
c
acid sites. Moreover, S and S cannot move because of
the steric hindrance but also because the Ru–Cs bond is
not destabilized by the direct interaction with Cs species.
Table 4
35
Evaluation of sulfidation state on various catalysts using S pulse tracer
method
0
0
.51–0.75, it was 0.36–1.55 for the catalysts with Cs/Ru =
∗
Catalyst
Sulfidation state
of metal sulfide
S , S /S
0
0A total
–3 and 8 wt% Ru (Table 3).
We have shown before that S0 (labile sulfur quantity on
a
b
c
d
Ru–Cs/Al O
Ru–Cs/Al O
Ru–Cs/Al O
Ru–Cs/Al O
RuS
RuS
RuS
RuS
CsS
6–10%
34–48%
7–36%
23–59%
–
< 75% in MoS
< 60% in MoS
< 40% in MoS
100%
2
3
3
3
3
0.91–1.11
1.43–1.69
1.11–2.72
1.54–3.40
the presulfided catalyst) was low on the catalysts, especially
in the case of Cs/Ru = 1, meaning that for low Cs/Ru
ratios, labile sulfur was hardly formed before the HDS
reaction began. To compare the catalytic system presented
in this study with other systems, we also listed in Table 4
presulfidation results reported by our group for Mo, Pt, and
Pd systems. It is remarkable that for more conventional
catalysts, the number of labile sulfur is high after the
presulfidation (high value of S0), while for Ru–Cs catalysts
this number is low and labile sulfur is created only under
the working conditions. Finally, taking into account the
∗
∗
2
2
2
Cs/Al O
2
3
0.62–0.74
Mo/Al O
MoS
2
3
2
2
2
2
CoMo/Al O
NiMo/Al O
Pt–Pd/Al O
CoS MoS
2
3
2
NiS MoS
2
2
3
PtS
PdS
0.25 0.25
2
3
a
After presulfiding (Ru 4–20 wt%, Cs/Ru = 1).
During HDS (Ru 4–20 wt%, Cs/Ru = 1).
After presulfiding (Ru 8 wt%, Cs/Ru = 0–3).
During HDS (Ru 8 wt%, Cs/Ru = 0–3).
b
c
d

66
A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
and NiS forms, respectively [56,59]. Therefore, if we replace
Ru by Mo in Fig. 15, steady state is C, where vacancies are
not present. On the other hand, Pt and Pd catalysts were
found to be in intermediate sulfidation states (PtS0.25 and
PdS0.25) under the operating condition [52]. When Ru is
replaced by Pt or Pd in Fig. 15, the structure of the Pt or Pd
catalyst may be close to that of A or E. Under the reaction
conditions Pt or Pd catalysts can exchange a lot of sulfur,
as the amount of labile sulfur was found to be 100% (see
Table 4). From these results, it is suggested that Ru catalysts
have a medium sulfidation state between the one of Mo
catalysts and the one of Pt or Pd catalysts.
4
. Conclusions
3
5
1
. During [ S]DBT HDS reaction over Ru–Cs/Al2O3
catalysts, it was observed that the DBT conversion and
labile sulfur amount (S0A) increased with increasing Ru
content. S0A corresponded to 0.51–0.75 atom of S for 1
atom of Ru.
Fig. 14. Sulfided Ru–Cs/Al O catalysts structure. (a) Proposed structure
for Cs/Ru = 2–3. (b) Proposed structure for Cs/Ru = 1.
2
3
a
Only S might be easy to remove and that is the explanation
3
5
2
. After Ru–Cs/Al2O3 catalysts presulfidation using [ S]
H2S, the conversion of DBT and the amount of total
sulfur (Stotal) increased with increasing Ru content.
Stotal corresponded to 0.91–1.11 atom of S for 1 atom
of Ru.
why the observed numbers of labile sulfur as well as the
total number of accommodated sulfur were low. Fig. 15
shows the DBT hydrodesulfurization reaction mechanism
over proposed catalyst structure in Fig. 14. First, DBT is
adsorbed on the anion vacancy of the A structure and B
state is formed. Then, DBT is desulfurized, one S atom
remains on the active phase, and the structure C is formed.
Furthermore, C reacts with H2 to form E through D (H2S
desorption). For high Cs contents, the Ru–S bond is rather
strong and therefore the catalyst steady state is likely to be B.
Sulfur exchange proceeds in a manner like the one presented
3
5
35
3. [ S]H2S Cs/Al2O3 catalysts [ S]H2S presulfidation
experiments showed that the amount of total sulfur
(Stotal) increased with increasing Cs content. Stotal cor-
responded to a CsS0.62–0.74 formulation, and incorpo-
rated S quantity represented 70–80% of that incorpo-
rated on the corresponding Ru–Cs catalysts.
on Fig. 15 and when all of the labile sulfur is labeled by 35
S
4. The amount of labile sulfur (S0) formed after presulfi-
3
5
the quantitative analysis of labile sulfur becomes possible
under the operating conditions.
In the case of Mo-based catalysts, Mo is sulfided up to
MoS2, and Co or Ni promoters are approximately in the CoS
dation of Ru–Cs/Al2O3 catalysts using [ S]H2S was
much smaller than Stotal or S0A, indicating that the la-
bile sulfur is difficult to form over the Ru–Cs/Al2O3
catalysts during the presulfidation stage.
Fig. 15. DBT HDS reaction mechanism over sulfided Ru–Cs/Al O catalysts.
2
3

A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
67
5
6
7
. The amount of labile sulfur (S0A) observed during
[16] J.L. Zotin, M. Cattenot, J.L. Portefaix, M. Breysse, Bull. Soc. Chim.
Belg. 104 (1995) 213.
3
5
[
S]DBT HDS as well as the conversion remarkably
[17] M. Breysse, M. Cattenot, V. Kougionas, J.C. Lavalley, F. Mauge, J.L.
Portefaix, J.L. Zotin, J. Catal. 168 (1997) 143.
increased with increasing Cs/Ru (keeping Ru loading
as a constant). S0A corresponded to 0.36–1.55 atom of S
for 1 atom of Ru.
[18] S.-J. Liaw, R. Lin, A. Raje, B.H. Davis, Am. Chem. Soc. Dir. Petrol.
Chem. 39 (1994) 636.
3
5
. Ru–Cs/Al2O3 catalyst [ S]H2S presulfidation experi-
ments showed that the DBT conversion and the amount
of total sulfur (Stotal) remarkably increased with in-
creasing Cs/Ru ratio. After presulfidation, Stotal corre-
sponded to 1.11–2.72 atom of S for 1 atom of Ru.
. Taking into account various experimental results, a mo-
del proposing two Ru–Cs catalysts structures corre-
sponding to two Cs/Ru domains was proposed. The
DBT HDS mechanism of this model was discussed and
compared with the one developed for Mo, Pd, and Pt
catalysts.
[19] M. Wojciechowska, M. Pietrowski, S. Lomnicki, Chem. Commun.
1999) 463.
(
[20] R. Navarro, B. Pawelec, J.L.G. Fierro, P.T. Vasudevan, Appl. Catal.
A 148 (1996) 23.
[21] A. Maezawa, M. Kitamura, K. Wakamoto, Y. Okamoto, T. Imanaka,
Chem. Expression 3 (1988) 1.
[22] Y. Okamoto, M. Odawara, H. Onimatsu, T. Imanaka, Ind. Eng. Chem.
Res. 34 (1995) 3703.
[23] A. Ishihara, M. Nomura, T. Kabe, J. Jpn. Petrol. Inst. 37 (3) (1994)
300.
[
[
24] A. Ishihara, M. Nomura, T. Kabe, Chem. Lett. (1992) 2285.
25] A. Ishihara, M. Nomura, N. Takahama, K. Hamaguchi, T. Kabe, J. Jpn.
Petrol. Inst. 39 (1996) 211.
[26] A. Ishihara, M. Nomura, K. Shirouchi, T. Kabe, J. Jpn. Petrol. Inst. 39
(
1996) 403.
The results showed that the sulfidation state and the mo-
[
[
27] A. Ishihara, K. Hamaguchi, T. Kabe, J. Jpn. Petrol. Inst. 40 (1997) 516.
28] S. Yamada, W. Qian, A. Ishihara, G. Wang, L. Li, T. Kabe, J. Jpn.
Petrol. Inst. 44 (2001) 217.
[29] W. Qian, S. Yamada, A. Ishihara, M. Ichinoseki, T. Kabe, J. Jpn. Petrol.
Inst. 44 (2001) 225.
bility of sulfur over Ru catalysts greatly changes depending
on the reaction conditions and suggested that the Ru cat-
alysts have unique properties, different from the ones ob-
served over Mo, Pt, or Pd catalysts.
[
[
[
30] C.G. Gachet, E. Dhainaut, L. De Mourgues, J.P. Candy, P. Fouilloux,
Bull. Soc. Chim. Belg. 90 (1981) 1279.
31] G.V. Isagulyants, A.A. Greish, V.M. Kogan, G.M. V’unova, G.V. An-
toshin, Kinet. Catal. 28 (1987) 550.
32] G.V. Isagulyants, A.A. Greish, V.M. Kogan, Kinet. Catal. 28 (1987)
220.
Acknowledgments
This work has been carried out as a research project of
The Japan Petroleum Institute commissioned by the Japan
Cooperation Center, Petroleum, with subsidy of the Ministry
of Economy, Trade, and Industry. The authors are grateful to
Mr. H. Godo and Mrs. R. Kanamori for their kind help with
the research.
[33] G.V. Isagulyants, A.A. Greish, V.M. Kogan, Kinet. Catal. 28 (1987)
55.
34] G.V. Isagulyants, A.A. Greish, V.M. Kogan, in: M.J. Pjilips, M. Ternan
Eds.), Proceedings of the 9th International Congress on Catalysis,
Calgary, Ottawa, Canada, Vol. 1, 1988, p. 35.
[35] G.V. Isagulyants, A.A. Greish, V.M. Kogan, Catal. Lett. 6 (1990) 157.
5
[
(
[36] M. Dobrovolszky, K. Matusek, Z. Paal, P. Tetenyi, J. Chem. Soc.,
Faraday Trans. 89 (1993) 3137.
[
37] V.M. Kogan, N.T. Dung, V.I. Yakerson, Bull. Soc. Chim. Belg. 104
References
(1995) 303.
[
[
38] M. Dobrovolszky, Z. Paal, P. Tetenyi, Appl. Catal. 142 (1996) 159.
39] V.M. Kogan, N.M. Parfenova, in: G.F. Froment, B. Delmon, P. Grange
[
1] T. Kabe, A. Ishihara, W. Qian, Hydrodesulfurization and Hydrodeni-
trogenation, Kodansha Scientific, Wiley–VCH, Tokyo, 1999.
2] M. Vrinat, M. Lacroix, M. Breysse, L. Mosoni, M. Roubin, Catal.
Lett. 3 (1989) 405.
(Eds.), Hydrotreating and Hydrocracking of Oil Fractions, Elsevier,
[
Amsterdam, 1997, p. 449.
[
40] V.M. Kogan, in: T. Weber, R. Prins, R. van Santen (Eds.), Transition
Metal Sulfides: Chemistry and Catalysis, High Technology, in: NATO
ASI Series 3, Vol. 60, Kluwer Academic, Dordrecht, 1998, p. 235.
41] V.M. Kogan, N.N. Rozhdestvenskaya, I.K. Korshevets, Appl. Catal. A:
Gen. 234 (2002) 207.
[
[
3] A.P. Raje, S.-J. Liaw, B.H. Davis, Appl. Catal. 150 (1997) 297.
4] A.P. Raje, S.-J. Liaw, R. Srinivasan, B.H. Davis, Appl. Catal. 150
(1997) 319.
[
[
[
5] T.A. Pecoraro, R.R. Chianelli, J. Catal. 67 (1981) 430.
6] M. Lacroix, N. Boutarfa, C. Guillard, M. Vrinat, M. Breysse, J. Ca-
tal. 120 (1989) 473.
[
[
42] T. Kabe, W. Qian, A. Ishihara, J. Phys. Chem. 98 (1994) 3912.
43] A. Ishihara, M. Yamaguchi, H. Godo, W. Qian, T. Kabe, Chem. Lett.
[
[
7] J. Shabtai, N.K. Nag, F.E. Massoth, J. Catal. 104 (1987) 413.
8] J.A. De Los Reyes, M. Vrinat, C. Geantet, M. Breysse, Catal. Today 10
(1996) 743.
[44] W. Qian, A. Ishihara, Y. Okoshi, W. Nakagami, M. Godo, T. Kabe,
J. Chem. Soc., Faraday Trans. 93 (1997) 4395.
(1991) 645.
[9] K. Lu, Y.J. Kuo, B.J. Tatarchuk, J. Catal. 116 (1989) 373.
[45] A. Ishihara, M. Yamaguchi, H. Godo, W. Qian, M. Godo, T. Kabe,
J. Jpn. Petrol. Inst. 41 (1998) 51.
[
10] A. Ishihara, M. Nomura, T. Kabe, J. Catal. 150 (1994) 212.
11] M.J. Ledoux, O. Michaux, G. Agostini, P. Panissod, J. Catal. 102
[
(
1986) 275.
[46] T. Kabe, A. Ishihara, W. Qian, M. Godo, Catal. Today. 45 (1998) 285.
[47] W. Qian, Y. Yoda, Y. Hirai, A. Ishihara, T. Kabe, Appl. Catal. A:
Gen. 184 (1999) 81.
[12] J.P.R. Vissers, C.K. Groot, E.M. Van Oers, V.H.J. de Beer, R. Prins,
Bull. Soc. Chim. Belg. 93 (1984) 813.
[13] S. Eijsbouts, V.H.J. de Beer, R. Prins, J. Catal. 109 (1991) 217.
[14] T.G. Harvey, T.W. Matheson, J. Catal. 101 (1986) 253.
[15] J.L. Lamberton, M. Cattenot, V. Kougionas, M. Mhaouer, J.L. Porte-
faix, M. Breysse, G. Pérot, in: G.F. Froment, B. Delmon, P. Grange
[48] T. Kabe, W. Qian, A. Funato, Y. Okoshi, A. Ishihara, Phys. Chem.
Chem. Phys. 1 (1999) 921.
[49] A. Ishihara, H. Godo, R. Kanamori, W. Qian, T. Kabe, Appl. Catal. A:
Gen. 182 (1999) 345.
(
Eds.), Hydrotreatment and Hydrocracking Oil Fractions, Elsevier,
[50] T. Kabe, A. Ishihara, W. Qian, A. Funato, T. Kawano, React. Kinet.
Catal. Lett. 68 (1999) 69.
Amsterdam, 1997, p. 529.

68
A. Ishihara et al. / Journal of Catalysis 217 (2003) 59–68
[
51] W. Qian, A. Ishihara, Y. Aoyama, T. Kabe, Appl. Catal. A: Gen. 196
[55] E.P. Barrett, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951)
373.
(2000) 103.
[52] T. Kabe, W. Qian, Y. Hirai, L. Li, A. Ishihara, J. Catal. 190 (2000) 191.
[53] A. Ishihara, W. Qian, T. Kabe, J. Jpn. Petrol. Inst. 44 (2001) 80.
[54] T. Kabe, W. Qian, K. Tanihata, A. Ishihara, M. Godo, J. Chem. Soc.,
Faraday Trans. 93 (1997) 3709.
[56] T. Kabe, W. Qian, S. Ogawa, A. Ishihara, J. Catal. 143 (1993) 239.
[57] N.-Y. Topsøe, H. Topsøe, J. Catal. 139 (1993) 631.
[58] N.-Y. Topsøe, H. Topsøe, J. Catal. 139 (1993) 641.
[59] T. Kabe, W. Qian, A. Ishihara, J. Catal. 149 (1994) 171.